Response to Comments for the Federal Register Notice for the Texas and
Oklahoma Regional Haze State Implementation Plans; Interstate Visibility
Transport State Implementation Plan to Address Pollution Affecting Visibility
and Regional Haze; and Federal Implementation Plan for Regional Haze
Docket No. EPA-R06-OAR-2014-0754
12/9/2015
-------
Table of Contents
List of Public Comments
1. General and Miscellaneous
2. State and Federal Roles in the Regional Haze Program
3. Our Clarified Interpretation of the Reasonable Progress and Long-Term Strategy
Requirements
4. Consideration of Visibility in the Reasonable Progress Analysis
5. Consultation between Oklahoma and Texas 1
6. Source Category and Individual Source Modeling 1
7. Constitutional Law 1
8. Stay of Effective Date, Consolidated Appropriations Act, and Executive Orders 1
9. Controls in Addition to CAIR/CSAPR, and CSAPR Better than BART 1
10. Installation of Controls Beyond the First Planning Period 1
11. Cost 1
11 .a. Use of Confidential Business Information 2
11 .b. General cost comments 2
I I.e. Individual Cost Criticisms (AFUDC, lifetime, escalation, etc.) 2
II .d. Control Level 2
11 .e. Use of IPM cost algorithms 3
ll.f. $/deciview 3
12. Cost versus visibility benefit 3
13. Natural Conditions 4
14. Consistency with Our Other Regional Haze Actions 4
15. Modeling 5
15.a. Modeling Protocol 5
15.b Model Selection/Model Performance/Model Uncertainty/Model Inputs 5
15.c Modeling Emission Inventory 5
15.d Monitor Data and Model Performance 5
15.e Camx Not Approved Model for LRT 5
15.f National Academy of Science 5
15.g Appendix W requires Modeling a 3-yr Period 5
15.h ETEX and CAPTEX Tracer Studies and CAMx Model Uncertainty 5
15.i Back Trajectory analysis 6
15.j CALPUFF Modeling for Coleto Creek 6
15.k Consideration of Maximum Impacted Day or Other Days 6
ii
-------
15.1 Context for Evaluation of Modeling Results (CALPUFF vs. CAMx) 641
15.m Consideration Of Additional Controls And Impacts At Other Class I Areas 646
15.n Impacts At Other Class I Areas and Consideration of Additional Controls 646
Harrington 660
Oklaunion 663
H.W. Pirkey 666
Twin Oaks 668
Works No. 4 672
Sommers-Deely-Spruce 673
Big Spring Carbon 675
Parish and Welsh 678
15.o Consideration of NOx Controls 694
15.p Oil and Gas Sources 696
15.q Identification of Sources using Q/d and Photochemical Modeling 702
Q/d Analysis 702
0.3% Threshold 719
15.r Legality of Visibility Metrics Used 744
15.s Adjustment of CAMx Results: Linear Extrapolation and Nox Vs. SO2, Natural
Background, Recent Actual Emissions 754
15.t Linear Extrapolation 754
15.u Natural Background 761
15.v EPA Miscalculated The Rpgs, Rpgs Do Not Account for Reductions Outside of Texas
(E.G. OK FIP) 780
15.w General Four-Factor Analysis Comments 788
16. Interstate Visibility Transport 788
17. Disapproval of the Oklahoma and Texas Reasonable Progress Goals 797
18. International Emissions 812
19. Grid Reliability 833
20. Determination of Nationwide Scope and Effect 837
21. Reasonable Progress Determination for the Texas Class I Areas 844
22. Deference 860
23. Reasonable Progress Goal Progress 865
iii
-------
List of Public Comments
Docket ID
Document
Commenter Affiliation
Number
title number
(EPA-R06-
OAR-2014-
(TX166.xxx)
0754-xxxx)
0043
TX166.043
The Honorable Marc A. Veasey,
Member of Congress, 33rd District,
Texas
0044
TX166.044
Anonymous comment
0046
TX166.046
Anonymous comment
0047
TX166.047
Anonymous comment
0048
TX166.048
Citizen comment
0049
TX166.049
Arlington Urban Ministries
0050
TX166.050
Citizen comment
0051
TX166.051
The Honorable Beto O'Rourke, Member
of Congress, 16th District, Texas
0052
TX166.052
U.S. Fish and Wildlife Service
0053
TX166.053
Transcript of Public Hearing 1 in
Austin, Texas, January 13, 2015
0054
TX166.054
Transcript of Public Hearing 2 in
Oklahoma City, Oklahoma, January 15,
2015
0055
TX166.055
American Electric Power Company,
Inc. (AEP) on behalf of Southwestern
Electric Power Company (SWEPCO)
and Public Service Company of
Oklahoma (PSO)
0056
TX166.056
Public Utility Commission of Texas
(PUCT) and Texas Commission on
Environmental Quality (TCEQ) -
Includes ERCOT attachment
0057
TX166.057
Oklahoma Gas and Electric Company
(OG&E)
0058
TX166.058
Nucor Steel - Arkansas and Nucor-
Yamato Steel Company (Nucor Steel)
iv
-------
Docket ID
Document
Commenter Affiliation
Number
title number
(EPA-R06-
OAR-2014-
(TX166.xxx)
0754-xxxx)
0059
TX166.059
American Chemistry Council,
American Coalition for Clean Coal
Electricity, American Coke and Coal
Chemicals Institute, American Forest &
Paper Association, American Fuel &
Petrochemical Manufacturers,
American Iron and Steel Institute,
American Petroleum Institute,
American Wood Council, Brick
Industry Association, Council of
Industrial Boiler Owners, Electricity
Consumers Resource Council,
Independent Petroleum Association of
America, Industrial Energy Consumers
of America, International Liquid
Terminals Association, National
Association of Manufacturers, National
Lime Association, National Mining
Association, National Oilseed
Processors Association, Portland
Cement Association, Texas Cotton
Ginners' Association, Aluminum
Association, and U.S. Chamber of
Commerce (collectively, the
"Associations")
0060
TX166.060
San Miguel Electric Cooperative (San
Miguel)
0061
TX166.061
Luminant Generation Company.
Includes reports prepared for Luminant
by:
-Sargent & Lundy (S&L) (App A)
-AECOM (App B)
-NERA Economic Consulting (App C)
0062
TX166.062
AECOM: Spreadsheet attachments to
AECOM report prepared for Luminant
Generation Company (Attachments to
App B to 0061)
0063
TX166.063
Gulf Coast Lignite Coalition (GCLC)
0064
TX166.064
Southwestern Public Service Company
(SPS) doing business as Xcel Energy
0065
TX166.065
Utility Air Regulatory Group (UARG)
V
-------
Docket ID
Document
Commenter Affiliation
Number
title number
(EPA-R06-
OAR-2014-
(TX166.xxx)
0754-xxxx)
0066
TX166.066
The Honorable Greg Abbott, Governor
of Texas (Texas Governor)
0067
TX166.067
Earthjustice, National Parks
Conservation Association, and Sierra
Club (Earthjustice at al.)
0068
TX166.068
Report of Vicki Stamper, along with
Exhibits 1-69 (in 32 separate files),
which together accompany the
comments of Earthjustice, National
Parks Conservation Association, and
Sierra Club
0070
TX166.070
Report of Dr. H Andrew Gray, along
with Exhibit 1, which together
accompany the comments of
Earthjustice, National Parks
Conservation Association, and Sierra
Club
0071
TX166.071
Report of Dr. George Thurston (with
Exhibits T1-T3), which accompanies
the comments of Earthjustice, National
Parks Conservation Association, and
Sierra Club
0072
TX166.072
Sierra Club (4,500 form letter emails)
0073
TX166.073
Forty individual e-mails and ten late
form-letter e-mails from members of
Sierra Club
0074
TX166.074
Association of Electric Companies of
Texas (AECT)
0075
TX166.075
Coleto Creek Power LP
--AECOM (nearly identical attachment
as Luminant 0061 App B)
--AECOM Analysis of Visibility
Impacts from Coleto Creek Unit 1
Using CALPUFF
0076
TX166.076
Edison Electric Institute (EEI)
0077
TX166.077
National Park Service (NPS)
0078
TX166.078
NRG Texas Power LLC
—Alpine Geophysics attachment
0079
TX166.079
Oklahoma Department of
Environmental Quality (ODEQ)
vi
-------
Docket ID
Document
Commenter Affiliation
Number
title number
(EPA-R06-
OAR-2014-
(TX166.xxx)
0754-xxxx)
0080
TX166.080
The Honorable Jeannie McDaniel,
Representative, Oklahoma House of
Representatives, 78th District
0081
TX166.081
Citizen comments
0082
TX166.082
Citizen comments
0083
TX166.083
USDA Forest Service
Notes:
1. The following public submissions were requests for extensions of the public comment
period that are not summarized in this document: EPA-R06-OAR-2014-0754-0037,
0038, 0039, 0040, 0041, and 0042.
2. Docket item 0071 is a corrected version that supersedes item 0069.
vii
-------
1. General and Miscellaneous
Comment: We received over 4,500 comments via email communications in support of our
rulemaking that were similar in content and format. The Commenters noted that reducing haze
causing pollution from Texas power plants will not only preserve and protect our national parks
and wilderness areas for future generations, but will directly benefit public health by requiring a
more than 60% statewide reduction in harmful sulfur dioxide emissions.
Response: We acknowledge the commenters for their support of this action. While the purpose
of the action is to reduce regional haze, we agree that reductions in air pollution can have other,
ancillary benefits, including improving public health.
Comment: We received many individual comments in support of our rulemaking, specifically
regarding the requirements that Texas coal plants reduce SO2 emissions. These comments were
from members representing various organizations, members of Congress and other government
agencies, and members of the general public. At the public hearings in Austin, Texas and
Oklahoma City, Oklahoma, over 100 people expressed general support for the plan. The
speakers at the public hearings included members of various organizations and members of the
general public.
Response: We acknowledge these commenters for their support of our proposal, including the
proposed requirements for SO2 emission reductions within Texas. We thank those who attended
and participated at our public hearings.
Comment: Some of the above individual commenters, e-mail commenters and several other
commenters who generally supported our action had also requested that we also consider the
addition of controls on several large coal plants in Texas (Welsh, Pirkey, and NRG Parish) and
include the reduction of NOx pollution from all Texas coal plants in the plan, many referring to
NOx BART reductions in New Mexico.
Response: We respond to specific comments concerning additional controls on Welsh, Pirkey,
and Parish, as well as comments on NOx reductions elsewhere in this document. Please refer to
the sections of this document where we discuss cost, cost versus visibility, and modeling.
Comment: We received three letters from Federal Land Managers1 in support of this
rulemaking. The National Park Service stated its support for the proposed reductions of SO2
emissions, which will improve visibility at Big Bend, Carlsbad Caverns, and Guadalupe
Mountains National Parks. The U.S. Department of Agriculture Forest Service stated its support
of the FIP and the additional controls on fifteen Texas sources which will benefit visibility in the
Class I areas of Upper Buffalo and Caney Creek in Arkansas, which it manages. The U.S. Fish
and Wildlife Service stated its support for our effort to review SO2 pollution controls and to
1 The Clean Air Act at 42 U.S. Code Section 7602 (i) states that he term "Federal land manager" means, with
respect to any lands in the United States, the Secretary of the department with authority over such lands.
-------
determine reasonable levels of air quality emissions that ensure reasonable progress toward our
nation's visibility goal at the Wichita Mountains.
Response: We acknowledge these Federal Land Managers for their support of this action.
Comment: The TCEQ supported the EPA's proposal to approve TCEQ's BART determinations.
Response: We acknowledge the TCEQ for its support for this component of our proposed
action. We discuss issues regarding BART requirements in Texas in greater detail in our
responses to specific comments elsewhere in this document.
Comment: One commenter at the public hearing in Austin, Texas, stated general support for the
EPA's proposal and expressed concern about mercury exposure from power plants.
Response: We acknowledge the commenters support. Our action on the Texas SIP and our
promulgation of the FIP is in accordance with the Clean Air Act requirements for addressing the
problem of regional haze. The Act and the regional haze rule prohibits us from addressing
mercury emissions in the review of a regional haze SIP submittal.
Comment: One commenter at the public hearing in Austin, Texas, stated general support of the
EPA's proposal and wondered why Arkansas was left out of the rule.
Response: We note that the EPA previously acted on the Arkansas Regional Haze SIP with a
partial approval and partial disapproval of the State plan (77 FR 14604, March 12, 2012). For
those components of the State plan that we disapproved, we proposed a rule to establish a FIP to
address regional haze and visibility transport requirements that was published on April 8, 2015
(80 FR 18944). The comment period for the proposal ended on July 15, 2015, and was
subsequently reopened for 15 days, ending on August 7, 2015. You may view the docket for this
proposal at the Federal e-Rulemaking Portal: http://www.regulations.gov; Docket No. EPA-
R06-OAR-2015-0189.
Comment: One commenter stated that she was part of a contingent that met with senior EPA
officials in October 2011, and later addressed a letter to the Administrator in December 2011,
that objected to the Texas SIP not reducing pollution or imposing BART on 117 sources and
asserted a need for agency resources to address the problem. The commenter received a
response letter affirming EPA's commitment to act on the SIP according to earlier applicable
consent decree deadlines. The Commenter asked EPA to remember this past representation
when considering the perpetual lack of action by the TCEQ on SO2 and NOx.
Response: Our consent decree was later extended to December 9, 2015. Except for Texas EGU
BART, we are completing our consent decree obligation for final action on the Texas Regional
Haze SIP in accordance with that deadline. Due to the CSAPR remand, we are unable to act on
2
-------
Texas EGU BART, and will complete that part of our obligation in a future action. To the extent
the comment suggests that changed deadlines and related assurances would change our
substantive review obligations, we disagree. Our review is based on whether the submitted SIP
meets the applicable requirements of the Clean Air Act.
Comment. We received five letters and e-mails which stated general opposition to the EPA's
proposed rulemaking from citizens and a representative of one organization. The Commenters
expressed concerns that less Federal regulation of the power industry is better and, that in
general, the EPA's proposal would cause problems with electric power grid reliability and
electricity affordability.
Response: The power industry, and individual facilities that are part of that industry, may be
subject to requirements of the Clean Air Act to address regional haze, as specified by Congress.
Elsewhere in our response to comments, we provide substantial explanations and reasons for our
authority to regulate these sources, disapproving elements of the Texas SIP, and finalizing our
FIP. We note that we received specific comments on aspects of electrical generation, including a
report on alleged grid reliability issues from ERCOT. We discuss this issue in greater detail in
the sections which deal with electric reliability in this document.
Comment: AECT stated that the EPA has not supported its proposal. Thus, AECT requested
that the EPA withdraw its proposal and re-evaluate the Texas Regional Haze SIP using the
corrected process, criteria, and information discussed by AECT and AECT member companies
in their comments.
Response: We direct AECT to the detailed responses to specific comments provided elsewhere
in this document for the reasons we are disapproving elements of the Texas SIP and finalizing
our FIP.
Comment: AEP supported the EPA's approval of the TCEQ's BART rules; however AEP urges
and requests that the EPA withdraw its proposed disapproval of the Texas and Oklahoma SIP
provisions and fully approve the Texas and Oklahoma regional haze SIP provisions. AEP states
that the EPA must withdraw its proposed FIP because the proposal departs from past practices
and precedent; is outside the scope of the requirements of the CAA, Regional Haze Rule, and
reasonable progress guidance; is arbitrary and capricious; and is an abuse of discretion.
Response: We appreciate the Commenter's support of the EPA's approval of the TCEQ's BART
rules, and direct AEP to the detailed responses to specific comments provided elsewhere in this
document for the reasons we are disapproving elements of the Texas and Oklahoma SIPs and
finalizing our FIPs.
Comment: NRG stated that it supports the EPA's proposed findings that Texas has met
requirements for identifying the baseline conditions at Big Bend and the Guadalupe Mountains,
3
-------
addressing "reasonably attributable visibility impairment," mitigating the impacts of construction
activities on visibility impairment, considering source retirement and replacement schedules as
part of the State's long-term visibility strategy, implementing smoke management techniques,
providing enforcement authority, quantifying visibility changes that may result from emission
changes over the term of the long-term strategy, and implementing a regional haze monitoring
strategy, emissions inventory, and appropriate federal land manager consultations.
NRG also stated that it does not support the EPA's proposed disapproval of various plan
elements or promulgation of a FIP, and that the EPA should also approve all other aspects of the
Texas regional haze plan, and not implement a FIP.
Response: We appreciate NRG's support of our findings and direct it to the detailed responses to
specific comments provided elsewhere in this document for the reasons we are disapproving
elements of the Texas SIP and finalizing our FIP.
Comment: UARG stated that because EPA has no basis for disapproving the Texas and
Oklahoma regional haze SIPs, its proposed FIPs are by definition unlawful. But even assuming
for the sake of argument that EPA's SIP disapprovals were valid, the proposed FIPs would
violate the CAA and the regional haze rule. The FIPs are based on a flawed and incomplete
reasonable progress analysis and would require emission reductions that are beyond EPA's
authority to impose in this rulemaking.
Response: We direct UARG to the detailed responses to specific comments provided elsewhere
in this document for the reasons we are disapproving elements of the Texas and Oklahoma SIPs
and finalizing our FIPs.
Comment: Commenter 0053-60 questioned if there is a way to institute the regulation and at the
same time prevent the energy companies from passing the cost onto their customers. The
commenter suggested that companies be fined instead of having customers pay for their
problems.
Response: It is an unfortunate consequence that the costs associated with controlling the
emissions from power plants often result in increasing the cost of electricity. Because of the
competitive nature of the ERCOT grid, we cannot speculate on how much an average person's
electric bill will increase. However, we are very sensitive to the ramifications of our actions and
we seek to select the most cost-effective option when we propose and finalize these controls.
Comment: The Regional Haze Process Must Be Implemented Reasonably Going Forward.
[EEI (0076) p. 2-4]
According to EEI, the Clean Air Act (CAA) regional haze program tasks states with determining
what is reasonable progress toward elimination of man-made visibility impairment, for which
EPA has set a goal of 2064, along with specific progress milestones (10-year planning and SIP
4
-------
revisions, with program reviews in the middle of the 10-year planning periods).2 The regional
haze program contemplates gradual visibility improvements along a "glide path" that considers
the 2064 goal, and does not require immediate reductions that exceed making "reasonable
progress", as determined by the state based on four statutory factors,3 in the first planning period
through 2018 or in any subsequent planning period. Thus, it neither requires nor authorizes the
frontloading of extensive control requirements. Instead, the regional haze program should be
implemented in a manner that allows states, through state environmental and electric utility
regulators and in conjunction with power companies, to plan the optimal timing of emission
control projects. This planning process is vital in order to minimize impacts on the cost and
reliable provision of electric power and to allow investment decisions to be made over suitable
planning horizons. It is additionally justified given the, at best, minimal visibility benefits EPA
claims would be achieved in the proposed rulemaking.
Thus, as EPA and the states begin to implement the next rounds of the regional haze program to
continue reasonable progress, EEI suggested that EPA should allow states to consider the timing
and scope of additional control activities, consistent with effective long-term utility planning.
EPA also must take into consideration the progress already made through the installation of
controls to satisfy Best Achievable Retrofit Technology (BART) requirements and BART-
equivalent measures such as the Cross-State Air Pollution Rule (CSAPR) and through other
CAA regulations that can result in reduced emissions that may contribute to visibility
impairment.
Footnotes:
2 The first 10-year planning period began in 2009 and ends in 2018. The next runs from 2019-2028, with SIPs due
in 2018.
3 Cost of compliance, time necessary for compliance, energy and non-air quality impacts of compliance, and
remaining useful life of any existing source subject to such requirements.
Response: We agree with EEI that the regional haze program contemplates gradual visibility
improvements. We do not consider the controls we have proposed and those we finalize in this
action as being extensive or frontloading. We believe the regional haze process set up by us
through the regional planning organizations provided states with an opportunity to implement the
program "in a manner that allows states, through state environmental and electric utility
regulators and in conjunction with power companies, to plan the optimal timing of emission
control projects." We believe that Texas had an opportunity to "minimize impacts on the cost
and reliable provision of electric power and to allow investment decisions to be made over
suitable planning horizons," while satisfying our regulations. However, as we have outlined in
our proposal and our final rulemaking, Texas failed to comply with certain aspects of the
Regional Haze Rule and thus portions of its regional haze plan are not approvable. We disagree
with EEI that the visibility benefits that we proposed are minimal. As we have done in the first
round of regional haze SIPs, we will allow states to consider the timing and scope of additional
control activities, consistent with effective long-term utility planning, as the states begin to
develop future SIPs.
Comment: [CCP (0075) p. 2] CCP incorporated the comments of the Association of Electric
Companies of Texas (AECT) (comment 0074) by reference.
5
-------
Response: The EPA acknowledges CCP's support of the comments submitted by AEC.
Comment: [AEP (0055) p. 2] AEP stated that they support and incorporate by reference the
industry member association comments of the Edison Electric Institute (EEI) (comment 0076),
Utility Air Resources Group (UARG) (comment 0065), Association of Electric Companies of
Texas (AECT) (comment 0074) and Gulf Coast Lignite Coalition (GCLC) (comment 0063) all
of which AEP is a member.
Response: The EPA acknowledges AEP's support of the comments submitted by EEI, UARG,
AECT, and GCLC.
Comment: [EEI (0076) p. 11] EEI urged EPA to consider the technical comments of both
Southwestern Public Service (0064) and Luminant (0061) regarding unit-specific and other
concerns.
Response: The EPA acknowledges EEI's support of the comments submitted by SPS (Xcel
Energy) and Luminant.
Comment: [San Miguel (0060) p. 5] San Miguel is an active member of the Gulf Coast Lignite
Coalition (GCLC). San Miguel refers to, and fully supports the more expansive comments
submitted by GCLC in this rulemaking. San Miguel stated that GCLC comment letter explains
why the EPA's proposed disapproval of key components of Texas' SIP, and its proposal of a FIP
is without basis, is without prior precedent, and unfairly targets and burdens Texas sources.
Specific topics covered in the GCLC comments are:
• EPA may not supplant Texas' SIP with what EPA believes is a more reasonable FIP
• EPA is unlawfully attempting to double-burden sources already complying with BART
requirements and attempts to apply beyond-BART requirements to sources that are
explicitly exempted from single-source BART requirements
• Texas' reasonable progress analysis and associated SIP submission complies with all
CAA requirements and must be approved
• Texas' long-term strategy and associated SIP submission complies with all CAA
requirements and must be approved
• Every factor of EPA's proposed reasonable progress analysis fails
• EPA has violated its regional consistency regulations by applying different and more
stringent standards on Texas units compared to other states and regions
• EPA may not issue this FIP prior to providing Texas the opportunity to submit a SIP
responsive to EPA's determination that Texas' 2009 SIP submission was inadequate
• EPA's Regional Haze FIP is not a rule of "nationwide scope and effect."
Response: The EPA acknowledges San Miguel's support of the comments submitted by the
GCLC. Our responses to GCLC's specific comments are provided throughout this document.
6
-------
Comment: [Earthjustice (0067) p. 1, 61] Earthjustice et al., incorporated by reference and
attached the comments submitted by Earthjustice, National Parks Conservation Association, and
the Sierra Club regarding prior actions taken in the development of the Texas regional haze plan.
Earthjustice et al., also attached several references cited in their comment. These attachments
include the following:
• ODEQ Aug. 3, 2007 Letter to TCEQ; TCEQ Oct. 15, 2007 Letter to ODEQ; TCEQ
Mar. 25, 2008 letter to ODEQ; ODEQ May 12, 2008 Letter to TCEQ.
• Letter from McCrystie Adams & Michael Hiatt, Earthjustice, to EPA (Feb. 28, 2012),
Docket ID No. EPA-HQ-OAR-2011-0729 Re: Proposed Partial Regional Haze FIP for
Texas Exempting BART-Eligible Texas EGUs from BART Based on EPA's
Determination that the Cross-State Air Pollution Rule is "Better than BART."
• Conservation Organizations' October 26, 2011 comments to EPA Re: Texas Regional
Haze Plan - Efficacy in the Face of the Anticipated "Better than BART" Rulemaking for
the Cross-State Air Pollution Rule.
• Letter from McCrystie Adams & Michael Hiatt, Earthjustice, to Margaret Earnest, Texas
Commission on Environmental Quality (Oct. 1, 2013), Re: 2014 Five-Year Regional
Haze SIP Revision—Project No. 2013-013-SIP-NR.
• U.S. Fish and Wildlife Service and National Park Service Comments Regarding Texas
Proposed Regional Haze Rule State Implementation Plan.
• Benchmarking Air Emissions of the 100 Largest Electric Power Producers in the United
States (May 2014).
• Daniel Cohan, Addressing pollution from legacy coal power plants in Texas (June 2013).
• Summary of Oil and Gas Sector TSD: Significant Stationary Source of NOx Emissions,
October 2012.
• Conservation Organizations' August 2, 2012 comments to EPA Re: Docket ID No. EPA-
R08-OAR-2012-0026, comments on EPA Proposed Approval, Disapproval, and 2
Promulgation of Implementation Plans; State of Wyoming; Regional Haze State
Implementation Plan; Federal Implementation Plan for Regional Haze.
• Texas 2014 Five-Year Regional Haze State Implementation Plan Revision, Proposal,
June 18, 2013.
• Texas 2014 Five-Year Regional Haze State Implementation Plan Revision, February 26,
2014.
• 2011 Oil and Gas Emission Inventory Enhancement Project for CenSARA States.
• Conservation Organization's August 26, 2013 comments to EPA Re: Docket ID No.
EPA-R08-OAR-2012-0026, Comments on EPA's Re-Proposed Approval, Disapproval,
and Promulgation of Implementation Plans; State of Wyoming; Regional Haze State
Implementation Plan; Federal Implementation Plan for Regional Haze, 78 Fed. Reg.
34,738 (June 10, 2013).
• Environmental Commenters' August 20, 2014 comments to EPA Re: Comments of Clean
Air Task Force, National Parks Conservation Association, Earthjustice, WildEarth
Guardians, and Southern Utah Wilderness Alliance on Managing Emissions From Oil
and Natural Gas Production in Indian Country: Advanced Notice of Proposed
Rulemaking, 79 Fed. Reg. 32,502 (June 5, 2014).
7
-------
Response: We acknowledge and appreciate receipt of the supplemental documents submitted by
Earthjustice et al. We take no position in this action regarding these documents beyond our
responses to any comments specifically referencing the documents.
Comment: [Stamper (0068) p.57] Ms. Stamper submitted 69 files as exhibits attached to her
comments. These files were used as references in her review of the proposed rule and analyses.
List of Exhibits submitted with Stamper (0068)
Exhibit
Number
Title/Description
1
November 3, 2010 letter from David C. Foerter, ICAC to Senator Carper.
2
U.S. EPA, An Assessment of the Feasibility of Retrofits for the Toxics Rule, March
9, 2011.
3
Direct Testimony of Mr. Chad Teply, PacifiCorp, before the Wyoming Public
Service Commission.
4A
August 3, 2011 "B&W gets contract for dry scrubber project at Karn coal plant."
4B
December 17, 2014 Extension Request for Consumers Energy Company's D.E.
Karn Plant (SRN B2840) Units 1 & 2 for Compliance with the Mercury and Air
Toxics Standard (40 CFR 63 Subpart UUUUU) and the Michigan Mercury Rule
(R336.2501)
5
July 9, 2014 TVA - Gallatin Fossil Plant (GAF) - Request for Compliance
Extension - Mercury and Air Toxics (MATS).
6
November 5, 2013 Request for One-Year Extension of the Compliance Deadline
for the Mercury and Air Toxics Standards and of the Expiration Date of the Plan
Approval for the Installation of Flue Gas Desulfurization Units
7
October 4, 2012 Construction Extension for Consumers Energy Company's JH
Campbell Facility Pursuant to the Mercury and Air Toxics Standard (40 CFR 63
Subpart UUUUU, also known as MATS) as well as the Michigan Mercury Rule
(R336.2501, et seq)
8A
Hitachi Power Systems America Awarded Contract to Supply Pollution Controls
Equipment for KCP&L."
8B
June 22, 2012 Request for Extension of the Mercury and Air Toxics Standards
(MATS) Compliance Deadline KCP&L La Cygne, Source ID No. 1070005
9
Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC
Technologies, Wet FGD Cost Development Methodology, Final March 2013.
10A
EPA's Cost Effectiveness Spreadsheet for SDA at the Corette Power Plant.
10B
EPA's Cost Effectiveness Spreadsheet for wet FGD at the Corette Power Plant.
10C
EPA's Cost Effectiveness Spreadsheet for DSI at the Corette Power Plant.
11
January 30, 2013 NIPSCO - Michigan City and R.M. Schahfer Generation
Stations Request for Extension of Time to Comply with the Utility MATS
NESHAP
12
EPA's NOx Cost Effectiveness Analyses for Jim Bridger Power Plant, EPA-R08-
OAR-2012-0026-0085.
13
OMB Circular A-94.
14
Black & Veatch vendor brochure on CT-121.
8
-------
15
Yasuhiko Shimogama, Hirokazu Yasuda, Naohiro Kaji, Fumiaki Tanaka, and
David K. Harris, Commercial Experience of the CT-121 FGD Plant for 700 MW
Shinko-Kobe Electric Power Plant, Paper No. 27, presented at MEGA Symposium,
Air & Waste Management Association, May 19-22, 2003.
16
CT-121 FGD Process - Jet Bubbling Reactor.
17
Jonas S. Klingspor, Kiyoshi Okazoe, Tetsu Ushiku, and George Munson, High
Efficiency Double Contact Flow Scrubber for the U.S. FGD Market, Paper No.
135 presented at MEGA Symposium, Air & Waste Management Association, May
19-22, 2003.
18
Yoshio Nakayama, Tetsu Ushiku, and Takeo Shinoda, Commercial Experience and
Actual-PI ant-Scale Test Facility ofMHI Single Tower FGD.
19
Mitsubishi High SO2 Removal Experience.
20
White Bluff Station Units 1 and 2 Evaluation of Wet vs Dry FGD Technologies,
Rev. 3, October 28, 2008, prepared by Sargent & Lundy.
21
Wet FGD Actual SO2 Emission Rates from CAMD data.
22
SDA Actual SO2 Emission Rates from CAMD data.
23
Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC
Technologies, Dry Sorbent Injection for SO2 Control Cost Development
Methodology, March 2013.
24
Fischer, Diane and Preston Tempero, Black&Veatch, Early Lessons Learned from
Implementation of Dry Sorbent Injection Systems, 2012.
25
Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC
Technologies, SDA FGD Cost Development Methodology, Final March 2013.
26
Lawrence Gatton, Alstom Power, Next Generation NID for PC Market, Coal-Gen,
August 17-19, 2011.
27
Black & Veatch, LG&E/KU - Mill Creek Station, Phase II Air Quality Control
Study, Air Quality Control Validation Report, March 4, 2011, Revision D - Issued
for Project Use.
28
Alstom Brochure, NID™ Flue Gas Desulfurization System for the Power Industry.
29
February 8, 2012 Direct Testimony of Christian T. Beam on behalf of
Southwestern Electric Power Company, In the Matter of Southwestern Electric
Power Company's Petition for a Declaratory Order Finding that Installation of
Environmental Controls at the Flint Creek Power Plant is in the Public Interest,
Before the Arkansas Public Utilities Commission, Docket 12-008-U.
30
Technical Support Document to Comments of Conservation Organizations,
Proposed Montana Regional Haze FIP - June 15, 2012.
31
Sargent & Lundy, Big Sandy Plant Unit 2, Order-of-Magnitude FGD Cost
Estimate, Volume 1 - Summary Report, September 29, 2010.
32
Spreadsheet with data on EGUs with wet scrubbers in arid areas.
33
Xcel Energy, We are energized, Texas and New Mexico.
34
Wet FGD Cost IPM TX Sources Revised VS Mar 27 2015.xlsx.
35
NID CDS Cost IPM TX Sources VS Mar 27 2015.xlsx.
36
SDA Cost IPM TX Sources Revised VS Mar 27 2015.xlsx.
37
DSI Cost IPM TX Sources Revised VS Mar 27 2015.xlsx.
38
Tolk Costs for Water Rights Purchase for SO2 Scrubbers.xlsx.
9
-------
39
March 2011 National Park Service spreadsheet "EGUs with Proposed BART
Controls."
40
May 28, 2009 Wyoming Department of Environmental Quality BART Application
Analysis, Dave Johnston Plant.
41
May 28, 2009 Wyoming Department of Environmental Quality BART Application
Analysis, Naughton Plant.
42
Letter from Arizona Department of Environmental Quality to Steve Fry, EPA
Region IX, Re: Consultation Regarding Best Available Retrofit Technology
Analyses for the Four Corners Power Plant and Navajo Generating Station, May
12, 2008.
43
URS, Assessment of Technology Options Available to Achieve Reductions of
Hazardous Air Pollutants, 4/5/11.
44
Babcock Power Environmental, Wet Flue Gas Desulfurization Scrubber Upgrades,
2009.
45
Moretti, Albert L., State-of-the-Art Upgrades to Existing Wet FGD Systems to
Improve SO2 Removal, Reduce Operating Costs and Improve Reliability, Presented
to Power-Gen Europe, Cologne, Germany, June 3-5, 2014.
46
Frazer, C., A. Jayaprakash, S.M. Katzberger, Y.J. Lee, B.R. Tielsch, presented to
EPRI Power Plant Air Pollutant Control Mega Symposium, August 30 - September
2, 2010, Baltimore, MD.
47
February 2011, Babcock Power, LG&E Services Company Contract No. 501654,
Mill Creek FGD Performance Upgrade Study.
48
SO2 Scrubber Upgrade Costs Data.
49
May 28, 2009 Wyoming Department of Environmental Quality BART Application
Analysis for Jim Bridger Power Plant.
50
December 12, 2007 Coal Creek Station Units 1 and 2 Best Available Retrofit
Technology Analysis.
51
January 2008 BART Analysis for Cholla Unit 4.
52
Colorado Department of Public Health and Environment, Best Available Retrofit
Technology (BART) Analysis of Control Options for Public Service Company-
Hayden Station.
53
Hayden BART Cost Analysis.
54
EPA's Colstrip Unit 1 SO2 Emissions and Costs Summary.
55
EPA's Colstrip Unit 2 SO2 Emissions and Costs Summary.
56
See Permit Amendment, Source Analysis & Technical Review, Public Service
Company of Oklahoma, Oklaunion Power Station, Permit Number
9015/PSDTX325M2.
57
Emission Sources - Maximum Allowable Emission Rates, Permit Numbers 9015
and PSDTX325M2, dated February 3, 2012.
58
Worksheet entitled "Pirkey and Oklaunion Coal Info."
59
Worksheet entitled Oklaunion and HW Pirkey CAMD 2000 to 2014.
60
Federal Operation Permit, H.W. Pirkey Power Plant, November 22, 2010.
61
TCEQ, Construction Permit Amendment, Review Analysis & Technical Review,
Big Spring Carbon Black Manufacturing Plant, Permit No. 6580.
62
September 25, 2012 Federal Operating Permit for Sid Richardson Carbon Big
Spring Facility.
10
-------
63
Statement of Basis of the Federal Operating Permit for Sid Richardson Carbon
Company Big Spring Facility.
64
Clean Air Markets Database Emission Data for Twin Oaks (TNP One Steam
Electric Station).
65
Statement of Basis of the Federal Operating Permit, Optim Energy Twin Oaks, LP.
66
Federal Operating Permit for Optim Energy Twin Oaks, LP, April 4, 2011.
67
Burns & McDonnell, Utility FGD Design Trends, which provides, among other
things, the year each FGD system at an EGU began operation.
68
Jianmin Wang, et. al., Leaching Behavior of Coal Combustion Products and the
Environmental Implication in Road Construction, A National University
Transportation Center at Missouri University of Science and Technology,
NUTCR214, April 2011.
69
Spreadsheet with TX EGUs 2012 to 2014 CAMD Data Ranked for SO2.
Response: We acknowledge and appreciate receipt of the supplemental reference documents
submitted by Ms. Stamper. We take no position in this action regarding these documents
beyond our responses to any comments specifically referencing the documents.
Comment: Commenter 0053-37 supported the proposed rule, and referred to a report produced
by Dr. Scott Ny strom of Regional Economic Modeling Incorporated (REMI) on behalf of
Citizens Climate Lobby. The report modeled carbon pricing legislation and has been used by
energy producers in the natural gas and nuclear arenas. It was beneficial in every way in terms
of economic growth, in terms of job growth, and had all the ancillary benefits of haze reduction
and the reduction of other pollutants in the environment. The commenter asked that state and
federal legislatures to take an opportunity to look at the report.
Response: Because the report was not attached to the comment letter, it was not loaded in the
docket. We believe the referenced report is available here:
https://citizensclimatelobbv.Org/remi-report/1
The REMI report examined the impact of a steadily-rising fee on carbon-based fuels with
revenue from that fee returned to households in equal shares. While we appreciate the
commenter bringing this report to our attention, we note that carbon pricing is outside the scope
of our proposed action with respect to regional haze implementation plans.
Comment: A commenter suggested that 2064 is too long for Texas polluters to improve air
quality to natural conditions.
Response: Our federal regulations require that states attempt to meet the established national
goal of natural visibility conditions by 2064, or demonstrate why that goal cannot be met. See
40 CFR 51.308(d)(1).
11
-------
Comment: [TCEQ/PUCT (0056) p. 16] The TCEQ noted the statute requires the regulating
entity to consider "the energy and non-air quality environmental impacts of compliance" when
developing the RPG. Nowhere in the EPA's proposal is this factor further defined. The EPA
provides guidance to states on how to consider this factor, but ignores a crucial part of the term.
The EPA cites only one element of its BART guidance as the basis of its analysis of this factor,
but ignores another more important element: the impact to energy reliability and costs due to
compliance with the RPG controls in the proposed FIP that are developed for a large segment of
the electric energy production in Texas.
Response: The TCEQ/PUCT is incorrect in its assertion that we did not consider the energy and
non-air quality environmental impacts of compliance. This factor is specifically evaluated in our
FIP TSD:2
Regarding the analysis of energy impacts, the BART Guidelines advise, "You
should examine the energy requirements of the control technology and determine
whether the use of that technology results in energy penalties or benefits."3 As
discussed below in our cost analyses for Dry Sorbent Injection (DSI) and Spray
Dryer Absorber (SDA) SO2 scrubbers, our cost model allows for the inclusion or
exclusion of the cost of the additional auxiliary power required for the pollution
controls we considered to be included in the variable operating costs. We chose
to include this additional auxiliary power in all cases. Consequently, we believe
that any energy impacts of compliance have been adequately considered in our
analyses.
Regarding the analysis of non-air quality environmental impacts, the BART
Guidelines advise:4
Such environmental impacts include solid or hazardous waste
generation and discharges of polluted water from a control device.
You should identify any significant or unusual environmental
impacts associated with a control alternative that have the potential
to affect the selection or elimination of a control alternative. Some
control technologies may have potentially significant secondary
environmental impacts. Scrubber effluent, for example, may affect
water quality and land use. Alternatively, water availability may
affect the feasibility and costs of wet scrubbers. Other examples of
secondary environmental impacts could include hazardous waste
discharges, such as spent catalysts or contaminated carbon.
Generally, these types of environmental concerns become
important when sensitive site-specific receptors exist or when the
incremental emissions reductions potential of the more stringent
control is only marginally greater than the next most-effective
option. However, the fact that a control device creates liquid and
2 See discussion beginning on page 6 of our FIP TSD.
3 70 FR 39168 (July 6, 2005).
4 70 FR 39169 (July 6, 2005).
12
-------
solid waste that must be disposed of does not necessarily argue
against selection of that technology as BART, particularly if the
control device has been applied to similar facilities elsewhere and
the solid or liquid waste is similar to those other applications. On
the other hand, where you or the source owner can show that
unusual circumstances at the proposed facility create greater
problems than experienced elsewhere, this may provide a basis for
the elimination of that control alternative as BART.
The SO2 control technologies we considered in our analysis - DSI and scrubbers
- are in wide use in the coal-fired electricity generation industry. Both
technologies add spent reagent to the waste stream already generated by the
facilities we analyzed, but do not present any unusual environmental impacts. As
discussed below in our cost analyses for DSI and SDA SO2 scrubbers, our cost
model includes waste disposal costs in the variable operating costs.
Consequently, we believe that with one possible exception, any non-air quality
environmental impacts have been adequately considered in our analyses. An
examination of the aerial photo of the Tolk facility, which we present in section
5.4, does not reveal any obvious source of surface water. We therefore assume
that well water is used. In light of this and its potential relationship to the energy
and non-air quality environmental impacts of compliance, we limit our SO2
control analysis for Tolk to DSI and dry scrubbers.
As can be seen from the above discussion regarding this factor, the TCEQ/PUCT is also
incorrect that this factor should also involve a consideration of grid reliability. The "energy" part
of this factor involves considerations of potential energy penalties due to the control technology
at the facility in question, not on the grid. Nevertheless, we did evaluate the potential impact our
proposed controls would have on grid reliability in a response to another comment.
Comment: Earthjustice provided background on the Clean Air Act's Regional Haze
Program. [Earthjustice (0067) p.5]
Earthjustice5 explained that since the nation's founding, the United States has valued its diverse
and stunning natural scenery. See, e.g., John Copeland Nagle, The Scenic Protections of the
Clean Air Act, 87 N.D. L. Rev. 571, 576 (2011). In what has been lauded as "America's best
idea," Congress first set aside national parks in the 19th century to preserve and celebrate some
of the nation's most spectacular scenery. Id. With the nation's rapid industrialization, however,
these remarkable scenic views have become increasingly marred by air pollution. See Id. at 573
Today, air pollution is "perhaps the greatest threat to national parks," and pollution all too often
degrades visibility in these iconic scenic areas. Id.
5 When we refer to Earthjustice, we also mean the National Parks Conservation Association
(NPCA) and the Sierra Club as these groups collectively submitted comments. These groups
also contracted with independent technical experts including Ms. Victoria Stamper, Dr. H.
Andrew Gray, and Dr. George D. Thurston.
13
-------
Recognizing the "intrinsic beauty and historical and archaeological treasures" of the national
parks and wilderness areas,1 Congress established "as a national goal the prevention of any
future, and the remedying of any existing, impairment of visibility in mandatory Class I Federal
areas which impairment results from manmade air pollution." 42 U.S.C. § 7491(a)(1). In 1990,
after finding that the U.S. Environmental Protection Agency ("EPA") and the States had not
made adequate progress toward reducing visibility impairment in the nation's Class I areas,2
Congress amended the Clean Air Act to curb emissions that may reasonably be anticipated to
cause or contribute to visibility impairment at national parks and wilderness areas. Id. § 7492.
Earthjustice explained that Congress delegated implementation of the Clean Air Act's visibility
program to EPA. In 1999, EPA promulgated the Regional Haze Rule, which requires the States
(or EPA where a State fails to act) to make incremental, "reasonable progress" toward
eliminating human-caused visibility impairment at each Class I area by 2064. 40 C.F.R. §
51.308(d)(1), (d)(3). In the 1999 regulations, EPA recognized that visibility impairing pollution
was a regional problem that required regional solutions; the regulations create the necessary
region-wide scheme to restore Class I areas to natural conditions. Furthermore, the regional haze
regulations require evaluation of all sources of visibility impairment.
Earthjustice et al., noted that in order to achieve the goal of natural visibility in Class I areas,
implementation plans must contain "emission limits, schedules of compliance and other
measures as may be necessary to make reasonable progress toward the national goal." 42 U.S.C.
§ 7491(b)(2). The Regional Haze Rule includes several interlocking measures designed to make
"reasonable progress" towards achieving the 2064 natural visibility goal. These measures
include requirements to (1) develop reasonable progress goals based on the evaluation of any and
all sources contributing to visibility impairment; (2) determine baseline and natural visibility
conditions; (3) create a long- term strategy for making reasonable progress; and (4) implement
the best available retrofit technology (BART) for some of the oldest and dirtiest sources of haze-
causing pollutants. Id.; 40 C.F.R. § 51.308(d), (e).
Footnotes:
1 H.R. Rep. No. 95-294, at 203-04 (1977), reprinted in 1977 U.S.C.C.A.N 1077, 1282.
2 Areas designated as mandatory Class I Federal areas (or Class I for short) consist of national parks exceeding
6,000 acres, national wilderness areas and national memorial parks exceeding 5,000 acres, and all international parks
that were in existence on August 7, 1977. See 42 U.S.C. § 7472(a).
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
Comment: Luminant provided background information on establishing reasonable
progress goals. [Luminant (0061) p.7]
Luminant explained that under EPA's regulations, when a State has a Class I area within its
borders, that State must establish a reasonable progress goal ("RPG") for the Class I area. EPA's
regulations provide: "For each mandatory Class I Federal area located within the State, the State
must establish goals (expressed in deciviews) that provide for reasonable progress towards
14
-------
achieving natural visibility conditions."36 As directed by the statute, and reiterated in the
regulations, the State must evaluate four-factors to establish the reasonable progress goal for the
in-State Class I area.37 These factors are: (1) "costs of compliance;" (2) "the time necessary for
compliance"; (3) "the energy and non-air quality environmental impacts of compliance"; and (4)
"the remaining useful life of any potentially affected sources."38 EPA's guidance explains that
"[i]n this context we believe that the cost of compliance factor can be interpreted to encompass
the cost of compliance for individual sources or source categories... ,"39 Notably, while the
visibility improvement or benefit of any particular control measure is included as a factor in a
BART determination, it is not one of the statutory factors to be considered in establishing an
RPG. The BART requirements (as discussed below) are source-specific and intended to be
applied at individual sources, while the RPG addresses regional haze from a broad group of
sources analyzed using the four statutory factors.
Luminant Stated that the CAA does not mandate any specific rate of progress as "reasonable
progress" towards meeting the goal.40 Instead, in its regulations, EPA established an analytical
requirement for measuring reasonable progress towards attainment of the national goal by the
year 2064.41 Reasonable progress goals are not enforceable measures in a State's SIP,42 but
rather "interim goals that represent incremental visibility improvement over time."43 Ultimately,
States have discretion and considerable flexibility in setting their reasonable progress goals.44
Luminant Stated that EPA guidance explains the analytical process States should follow in
establishing their reasonable progress goals. States must first"[i Jdentify the key pollutants and
sources and/or source categories that are contributing to visibility impairment at each Class I
area," and then "identify the control measures and associated emission reductions that are
expected to result from compliance with existing rules and other available measures."45
Importantly, EPA explains that "[g]iven the significant emissions reductions that we anticipate to
result from BART, . . . and the implementation of other CAA programs, .... for many States
this will be an important step in determining [the reasonable progress goal], and it may be all
that is necessary to achieve reasonable progress in the first planning period for some States."46
Next, Luminant noted that States are to "[ajnalyze and determine the rate of progress needed to
attain natural visibility conditions by the year 2064[,]" which is known as the "uniform rate of
progress" or "URP,"47 The URP simply serves as an analytical benchmark that defines the rate of
progress that would achieve natural visibility by 2064. EPA's regulations provide that the URP
need not be achieved in practice and that a "reasonable" rate of progress may achieve natural
conditions at a slower rate, well past 2064. Under the regulations, "if the State establishes a
reasonable progress goal that provides for a slower rate of improvement in visibility than the
[uniform rate of progress]," EPA's regulations provide that the State "demonstrate, based on the
[four-factors], that the [uniform rate of progress] is not reasonable; and that the progress goal
adopted by the State is reasonable."48
Luminant Stated that the regulations further define how reasonable progress goals are measured.
Reasonable progress goals "must provide for an improvement in visibility for the most impaired
days over the period of the implementation plan and ensure no degradation in visibility for the
least impaired days over the same period."49 The term "most impaired days" is defined as "the
average visibility impairment (measured in deciviews) for the twenty percent of monitored days
15
-------
in a calendar year with the highest amount of visibility impairment," and the term "least impaired
days" is defined as "the average visibility impairment (measured in deciviews) for the twenty
percent of monitored days in a calendar year with the lowest amount of visibility impairment."50
Thus, the regulations require that the reasonable progress goal must provide improvement on the
20 percent worst days and no degradation on the 20 percent best days.
According to Luminant, States' "RPGs are interim goals that represent incremental visibility
improvement over time."51 Thus, unlike BART determinations for individual sources (discussed
below), which are one-time determinations, States submit their reasonable progress SIPs in
phases. The first submittal covers the "first planning period" from 2008 to 2018. States then
update their reasonable progress goals and long-term strategies in the form of SIP revisions at
ten-year intervals thereafter (thus, the first update is due on July 31, 2018).52 For each revision
for the applicable planning period, the State "must evaluate and reassess all of the elements
required [by § 51.308(d)], taking into account improvements in monitoring data collection and
analysis techniques, control technologies, and other relevant factors."53 A State's revision must
address the following: (1) "[cjurrent visibility conditions" and "actual progress made towards
natural conditions"; (2) the "effectiveness of the long-term strategy"; and (3) "[ajffirmation of, or
revision to, the reasonable progress goal," including an evaluation of the reasonableness of the
goal if a "slower rate of progress" had originally been adopted.54 The time period for the Texas
SIP submission under review here is the first planning period—2008 to 2018—although, as
explained below, EPA's proposed FIP erroneously and unlawfully seeks to reach outside the first
planning period to 2020.
In addition to 10-year SIP revisions, Luminant noted that each State must submit a progress
report to EPA at five-year intervals beginning Five-years after the submission of the initial
regional haze SIP, evaluating the State's progress towards meeting the reasonable progress goals
for each Class I area within the State and each affected out-of-State Class I area.55 Depending on
the conclusions in the progress report, the State may need to revise its SIP or take other action.56
Footnotes:
36 40 C.F.R. § 51.308(d)(1).
37 42 U.S.C. § 7491(g)(1); 40 C.F.R. § 51.308(d)(l)(i)(A).
38 40 C.F.R. § 51.308(d)(l)(i)(A); see also 42 U.S.C. § 7491(g)(1).
39 EPA Reasonable Progress Guidance at 5-1 (emphasis added); EPA, Additional Regional Haze Questions 9 (Sept.
27, 2006) ("Reasonable progress is not required to be demonstrated on a source-by-source basis.").
40 42 U.S.C. § 7491(b)(2).
41 See 40 C.F.R. § 51.308(d)(l)(ii).
42 Id. § 51.308(d)(l)(v).
43 EPA Reasonable Progress Guidance at 1-2.
44 The legislative history of the regional haze program also supports this flexible approach. For instance, in the
conference committee that reconciled the House and Senate versions of the 1977 amendments the term "reasonable
progress" was specifically changed from term "maximum feasible progress." See 1 Legislative History of the Clean
Air Act Amendments 1977 Pub. L. No. 95-
95 155 (1977) ("The term 'maximum feasible progress' is changed to read 'reasonable progress' whenever it
appears in the section.").
45 EPA Reasonable Progress Guidance at 2-3.
46 Id. at 4-1 (emphasis added).
47 40 C.F.R. § 51.308(d)(l)(i)(B).
48 Id. § 51.308(d)(l)(ii).
49 Id. § 51.308(d)(1) (emphasis added).
16
-------
50 Id. § 51.301.
51 EPA Reasonable Progress Guidance at 1-2 (emphasis added).
52 40 C.F.R. § 51.308(f).
53 Id.
54 Id. § 51.308(f)(l)-(3).
55 Id. § 51.308(g).
56 Id. § 51.308(h).
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis. Consideration of visibility, our basis for
promulgating a FIP that includes source specific controls under reasonable progress, and our
authority to require installation of controls beyond the first planning period are discussed
elsewhere in this document.
Comment: Earthjustice et al., provided background on BART. [Earthjustice (0067) p.7]
Earthjustice et al., explained that, as one element of making progress towards the natural
visibility goal, BART controls are required at fossil fuel-fired power plants and other major
stationary sources that "may reasonably be anticipated to cause or contribute to any impairment
of visibility in any mandatory Class I Federal area," and were in existence in 1977, but were not
in operation before 1962. 42 U.S.C. § 7491(b)(2)(A); 40 C.F.R. § 51.308(e). The term "major
stationary source" is defined to include any source that has the potential to emit 250 tons per year
or more of any pollutant, and falls within one of 26 categories of industrial sources defined by
the Act. 42 U.S.C. § 7491(g)(7). BART is defined as "an emission limitation based on the
degree of reduction achievable through the application of the best system of continuous emission
reduction for each pollutant which is emitted by an existing stationary facility." 40 C.F.R. §
51.301 (emphasis added).
When determining BART, the States and EPA must analyze "the best system of continuous
emission control technology available" by taking into consideration five factors: (1) the costs of
compliance, (2) the energy and non-air quality environmental impacts of compliance, (3) existing
pollution controls at the source, (4) the remaining useful life of the source, and (5) the degree of
visibility improvement from pollution controls. Id. § 51.308(e)(l)(ii)(A). BART is an essential
component of the regional haze program because Congress largely grandfathered these
antiquated sources into many of the Clean Air Act's requirements. See 70 Fed. Reg. 39,104,
39,111 (July 6, 2005). BART compels these older, disproportionately-polluting sources to install
up-to-date and cost-effective pollution controls.
BART is a mandatory measure that must be implemented to achieve reasonable progress toward
restoration of natural visibility conditions. The Clean Air Act expressly requires States to adopt
SIPs that contain "emission limits, schedules of compliance and other measures as may be
necessary to make reasonable progress toward meeting the national goal . . . including1' the
installation and operation of BART at eligible sources that emit any air pollutant which may
reasonably be anticipated to cause or contribute to visibility impairment in any Class I area. 42
U.S.C. § 7491(b)(2) (emphasis added). The only permissible exemption from BART is when
EPA, by rule promulgated with sufficient notice and opportunity for public comment, determines
that (1) the source does not "by itself or in combination with other sources" cause or contribute
17
-------
to significant visibility impairment at a Class I area; (2) if the power plant has a design capacity
of 750 megawatts or greater, the owner or operator must demonstrate to EPA that the plant is
located at such a distance from all Class I areas that it does not "by itself or in combination with
other sources" emit pollution that may reasonably be anticipated to cause or contribute to
significant visibility impairment at a Class I area; and (3) the affected Federal Land Managers
concur with the BART exemption. Id. § 7491(c).4
Footnote:
4See also 40 C.F.R. § 51.308(e) (requiring BART for each eligible source "that may reasonably be anticipated to
cause or contribute to any impairment of visibility in any mandatory Class I Federal area, unless the State
demonstrates that an emissions trading program or other alternative will achieve greater reasonable progress toward
natural visibility conditions.").
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
Comment: Luminant provided background information on the process for determining
BART. [Luminant (0061) p. 10]
According to Luminant, under the BART requirement, States must develop "emission limitations
representing BART" for certain major stationary sources of air emissions.63 A State's BART
determination "must be based on an analysis of the best system of continuous emission control
technology available and associated emission reductions achievable for each BART-eligible
source that is subject to BART within the State."64 The State evaluates five factors when making
a BART determination: (1) "the costs of compliance"; (2) "the energy and non-air quality
environmental impacts of compliance"; (3) "any existing pollution control technology in use at
the source"; (4) "the remaining useful life of the source"; and (5) "the degree of improvement in
visibility which may reasonably be anticipated to result from the use of such technology."65 This
fifth factor distinguishes the BART analysis from the reasonable progress analysis, which does
not contain visibility improvement as a factor.
Luminant explained that BART-eligible sources are those that "were 'in existence' on August 7,
1977 but were not 'in operation' before August 7, 1962" and "have potential emissions greater
than 250 tons per year."66 As required by the statute, EPA has developed BART guidelines to
assist States in making BART determinations.67 For fossil fuel units that have a "total generating
capacity greater than 750 megawatts," the State is required to follow EPA's BART guidelines.68
According to Luminant, in lieu of source-specific BART, EPA's regulations specifically
authorize participation by States in a regional trading program instead of requiring "sources
subject to BART to install, operate, and maintain BART."69 The regulations provide that the
emissions trading program "must achieve greater reasonable progress than would be achieved
through the installation and operation of BART."70 As discussed more fully below, EPA has
found that both the Clean Air Interstate Rule ("CAIR") and the replacement Cross State Air
Pollution Rule ("CSAPR"), which limit SO2 and NOx emissions from EGUs, are "better-than-
BART" and operate as BART alternatives for States subject to those programs, including Texas.
18
-------
Footnotes:
65 42 U.S.C. § 7491(g)(2); see 40 C.F.R. § 51.308(e)(l)(ii)(A).
66 40 C.F.R. pt. 51, app. Y.
67 See id.
68 Id. § 51.308(e)(l)(ii)(B).
69 Id. § 51.308(e)(2); see 70 Fed. Reg. 39,104, 39,138-43 (July 6, 2005); Util. Air Regulatory Grp. v. EPA, 471
F.3d 1333, 1339-41 (D.C. Cir. 2006) (upholding "EPA's substitution of CAIR for BART").
70 40 C.F.R. § 51.308(e)(2).
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis. For comments relating to visibility as a
consideration, we have addressed those specific comments elsewhere.
Comment: [Earthjustice (0067) p.l], EPA must finalize these disapprovals in order to fulfill its
statutory responsibility to review State implementation plan ("SIP") submittals and disapprove
proposed plans that do not comply with Clean Air Act requirements. Earthjustice et al., stated
that EPA must also finalize a Federal implementation plan ("FIP") to remedy the deficiencies
EPA is disapproving in the Texas and Oklahoma plans.
Response: We are finalizing the majority of our SIP disapproval and FIP in this action as
detailed in our final rule. As discussed in the final action and other parts of our responses to
comments, given the uncertainty arising from the remand of Texas' CSAPR budgets, we have
concluded that it would not be appropriate to finalize our proposed determination to rely on
CSAPR as an alternative to SO2 and NOx BART for EGUs in Texas at this time. Because of the
CASPR remand and resulting uncertainty regarding SO2 and NOx BART for EGUs, we have
also decided not to finalize our proposed approval of Texas' PM BART determination. Finally,
today's action does not finalize the portion of our proposed FIP addressing Texas' interstate
visibility transport obligations because that portion of the FIP would have partially relied on
CSAPR.
Comment: [Earthjustice (0067) p. 19] Earthjustice et al., stated that Texas' do-nothing approach
does not comply with applicable Clean Air Act requirements, and EPA has no choice but to
disapprove it. We explain in detail below why EPA must disapprove Texas' regional haze SIP
and Oklahoma's reasonable progress goals (which unreasonably failed to account for any
emission reductions from Texas sources that impact Class I Areas in Oklahoma) and why the
SO2 emissions limits EPA proposes are reasonable, as are the revised reasonable progress goals
for Texas and Oklahoma.
Response: While we generally agree with the necessity to disapprove the Texas and Oklahoma
SIPs, we take no position as to specific statements made in this comment. We appreciate
commenters support with regards to the FIP SO2 limits and revised reasonable progress goals.
Comment: EPA Must Disapprove Texas' Inadequate Regional Haze SIP. [Earthjustice
(0067) p. 19, 59] The Texas SIP failed to satisfy the haze requirements in the Clean Air Act on
multiple grounds, and EPA's proposed disapproval of the Texas SIP is therefore proper.
19
-------
Earthjustice et al., urged EPA to finalize a strong regional haze plan for Texas, which is vital to
restoring clean air for the people and iconic national parks and wilderness areas in Texas and the
region.
Response: We appreciate commenters' support of our finalization of the disapproval and the
FIP. We take no position on any other specific statements made in the comment.
Comment: [Earthjustice (0067) p. 1-2], EPA's proposed measures are actually conservative
in light of the fact that it could - and should - have disapproved additional sections of
Texas' SIP and required pollutant reductions from additional sources under the proposed
FIP
As discussed in the accompanying report by Victoria Stamper (0068), EPA's proposed controls
are even more cost-effective than EPA has calculated. Earthjustice et al., stated that the EPA's
proposed SO2 emission limits on 15 Texas sources is a critically important, albeit conservative
measure, and EPA should finalize the proposed reduction requirements and evaluate additional
measures to make greater gains in visibility at the effected in and out of State Class I areas. The
proposed FIP represents a critically important, minimum set of controls necessary for Texas to
make reasonable progress toward eliminating haze at its own and out-of-State parks and
wilderness areas. These controls would not only fulfill the statutory mandate to clean up our
national parks and wilderness areas [sic]. As discussed in the accompanying report by Dr.
George Thurston, the proposed controls would yield billions of dollars in public health benefits
by avoiding the premature deaths, respiratory conditions, and other health problems caused by
the emissions that also contribute to haze.
Response: We agree with commenters that we have the authority to finalize a FIP where our
review of a SIP merits a disapproval. We appreciate commenters support regarding our FIP. We
address the general issues concerning the costs of our controls and the ancillary health benefits of
our final action elsewhere.
Comment: EPA Should Clarify that Each Flaw in Texas' Haze SIP Provides an
Independent Basis for EPA's SIP Disapproval. [Earthjustice (0067) p.5]
Earthjustice et al., stated that the EPA properly proposes to disapprove Texas' regional haze SIP
because (1) Texas did not accurately calculate natural visibility conditions at Big Bend and
Guadalupe Mountains, (2) Texas' statewide reasonable progress analysis is not supported by the
record, is not well-reasoned, and does not comply with Clean Air Act requirements, (3) Texas'
reasonable progress goals for Big Bend and the Guadalupe Mountains violated the Clean Air Act
and the regional haze regulations, and (4) Texas' long-term strategy was based on a technically
inadequate consultation with Oklahoma and did not require the control measures needed for
reasonable progress at the Wichita Mountains. Each of these significant flaws in Texas' haze
SIP is fatal and each flaw provides an independent ground for EPA to disapprove the SIP. In the
final rule, EPA should clarify this fact and make clear that each flaw provides an independent
basis for EPA's SIP disapproval. For example, even if EPA had approved Texas' calculation of
natural visibility conditions, EPA would still be required to disapprove the SIP's inadequate
RPGs and long-term strategy.
20
-------
Response: We thank commenters for the support. We also agree with the comment that items
one through four listed above provide separate bases for disapproval. We have addressed those
comments elsewhere in this document.
2. State and Federal Roles in the Regional Haze Program
Comment: Commenters argue that EPA's proposal to disapprove Texas and Oklahoma regional
haze SIPs disregards the primary role of the States under the CAA, the Regional Haze Rule, and
relevant case law, therefore, EPA acted outside its authority. Some commenters argue that the
Clean Air Act is based on principles of cooperative Federalism that require EPA to defer to their
States in development of implementation plans, so long as necessary statutory requirements are
met. Commenters stated that EPA's proposal ignores such limits and would impose FIPs that
ignore the primary implementation role given to Texas and Oklahoma
Response: We do not agree. Our action does accord with the CAA. Congress crafted the CAA
to provide for States to take the lead in developing implementation plans, but also required EPA
to review SIPs for compliance with statutory and regulatory requirements. We recognize that
States have the primary responsibility of drafting an implementation plan to address the
requirements of the regional haze program. We also recognize that we have the responsibility of
ensuring that the State plans, including Regional Haze (RH) SIPs, conform to the CAA
requirements. We have determined that the Texas and Oklahoma SIPs do not meet certain
elements of the Federal regional haze requirements, and we discuss in our final actions those
portions of the SIP that we will disapprove.
Additionally, our review of SIPs is not limited to a ministerial review and approval of a State's
submittal. We disagree with the commenters arguments regarding cooperative Federalism.
Under this framework, the CAA directs us to act if a State fails to submit a SIP, submits an
incomplete SIP, or submits a SIP that does not meet the Federal requirements. Thus, the CAA
provides us with a critical oversight role in ensuring that SIPs meet the Act's requirements.
Comment: Commenters Stated that Texas' plan was complete by operation of law, met all
requirements, and that EPA had no authority to impose a FIP.
Response: We disagree. The commenters confuse the actions of merely submitting a SIP and
having it deemed complete, with the process of reviewing that SIP for compliance with the
applicable Federal requirements. We agree that the States are given flexibility in developing a
SIP, but in doing so, they are required to adopt SIPs that meet Federal requirements. EPA must
review a State's submittal and determine whether it meets Federal requirements. If it does not,
EPA has authority to impose a FIP to fill in the gaps. In this instance, we determined that
portions of the State's submittal did not meet many of our regional haze requirements. These
determinations are discussed elsewhere in this document and final action.
21
-------
Comment: Some commenters argued that the EPA's Regional Haze Rule (RHR) established the
remedy for a substantially inadequate plan as periodic updates, not a Federal plan.
Response: We do not agree with commenters' asserted position that the remedy for an
unapprovable Regional Haze SIP is periodic updates. The Regional Haze Rule's requirements
for comprehensive periodic revisions (see 40 CFR 51.308(f)) and periodic progress reports (see
40 CFR 51.308(g)) are very different from the authority to impose a FIP when there is a
determination that a SIP is not approvable. As we have stated elsewhere, we have the authority
and obligation to impose a FIP to fill in such gaps. The provisions of the Regional Haze Rule
regarding States' ongoing responsibility to periodically revise their RH SIPs do not override this
responsibility.
Comment: Earthjustice provided background on EPA's authority regarding regional haze.
[Earthjustice (0067) p. 10]
Earthjustice Stated that the Clean Air Act's regional haze program provides States with the
initial opportunity to develop regional haze SIPs that clean up the air in our nation's national
parks and wilderness areas in accordance with the Regional Haze Rule and EPA guidance. See
40 C.F.R. § 51.308. Where a State's regional haze plan fails to establish a haze program that
complies with the applicable legal requirements, the Clean Air Act's cooperative Federalism
provisions require EPA to exercise Federal oversight by disapproving the State plan and issuing
a Federal implementation plan ("FIP") in its place. 42 U.S.C. § 7410(c)(1); Oklahoma v. EPA,
723 F.3d 1201, 1207-10 (10th Cir. 2013).
According to Earthjustice, Congress gave EPA the final say on whether a State's regional haze
SIP complies with the Clean Air Act. 42 U.S.C. §§ 7410(c)(1), (k)(3), (1), 7491. As the courts
have recognized, EPA has broad oversight authority over the regional haze program and is
obligated to disapprove State haze SIPs that fail to comply with the Act. North Dakota v. EPA,
730 F.3d 750, 760-62 (8th Cir. 2013); Oklahoma v. EPA, 723 F.3d at 1207-10. Congress gave
EPA this broad oversight authority to prevent recalcitrant States from undermining the purposes
of the Clean Air Act through inadequate SIPs. See, e.g., Alaska Dep't ofEnvtl. Conservation v.
EPA, 298 F.3d 814, 819-20 (9th Cir. 2002) (Congress gave EPA oversight authority because of
"disappointing State response[s] to air pollution concerns" and its recognition that "States
experience^ internal industry 'pressure ... to relax their standards'"), aff'd, 540 U.S. 461
(2004).
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
Comment: EPA's proposal unlawfully ignores the primary role and flexibility that the
clean air act gives to States to determine reasonable progress. [Luminant (0061) p. 56]
Luminant Stated that the EPA's proposal to disapprove the Texas and Oklahoma regional haze
SIPs disregards the primary role of the States under the Clean Air Act generally and under the
22
-------
Regional Haze Rule in particular and is thus outside of EPA's authority. EPA "is a creature of
statute" and "has no power to act unless and until Congress confers power upon it."398 EPA's
proposal, and the underlying methodology and criteria that EPA employs, are not authorized by
either the statute or the regulations, and thus EPA's proposal is unlawful.
According to Luminant, the Clean Air Act gives the States the primary role and substantial
discretion in formulating plans for meeting the statutory goals and requirements, including in
particular in the regional haze program. The Clean Air Act "establishes a comprehensive
program for controlling and improving the nation's air quality through State and Federal
regulation."399 Congress chose a "cooperative Federalism" structure to implement the statute,
dividing authority between the Federal government and the States.400 Within this division, "air
pollution prevention ... is the primary responsibility of States and local governments."401
Luminant Stated, consistent with this structure, as to visibility protection, EPA's job is to
"promulgate regulations to assure . . . reasonable progress toward meeting the national goal" of
preventing future and remedying existing visibility impairment in Class I areas, but it is the SIP
that contains the "measures" "necessary to make reasonable progress."402 EPA's role in
reviewing SIP provisions developed by States to make reasonable progress is limited. As the
Fifth Circuit has explained: "The great flexibility accorded the States under the Clean Air Act is .
. . illustrated by the sharply contrasting, narrow role to be played by EPA."403 In that narrow role,
where a SIP meets the basic requirements of the statute, EPA must approve it.404 "With regard to
implementation, the Act confines the EPA to the ministerial function of reviewing SIPs for
consistency with the Act's requirements."405 Indeed, "the EPA has no authority to question the
wisdom of a State's choices of emission limitations if they are part of a SIP that otherwise
satisfies the standards set forth in 42 U.S.C. § 7410(a)(2)."406 The Clean Air Act "supplies the
goals and basic requirements of [SIPs], but the States have broad authority to determine the
methods and particular control strategies they will use to achieve the statutory requirements."407
EPA's regional haze regulations and guidance confirm the States' primary role in determining
"reasonable progress." EPA's implementing regulations "call[] for States to play the lead role in
designing and implementing regional haze programs... ,"408 And the regulations, EPA explained
at the time of their adoption, are "based on the principle that States should have considerable
flexibility in adopting visibility improvement goals and in choosing the associated emission
reduction strategies for making 'reasonable progress' toward the national visibility goal."409
Accordingly, Luminant noted "[t]he final regional haze rule . . . provide[s] the States
considerable discretion in establishing reasonable progress goals for improving visibility in the
Class I areas."410 "The State must address regional haze" by "establishing a reasonable progress
goal" for each Class I area in the State.411 In doing so, "the State must. . . [cjonsider" the four
statutory factors—"the costs of compliance, the time necessary for compliance the energy and
non-air quality environmental impacts of compliance, and the remaining useful life of any
potentially affected sources... ,"412 As EPA has recently explained, States have discretion and
considerable flexibility in setting their reasonable progress goals based on these factors:
The RHR does not mandate specific milestones or rates of progress, but instead
calls for States to establish goals that provide for "reasonable progress" toward
achieving natural (i.e., "background") visibility conditions . . . States have
23
-------
significant discretion in establishing RPGs, but are required to consider the [four]
factors established in section 169A of the CAA and in EPA's RHR at 40 CFR
51.308(d)(l)(i)(A). . . States have considerable flexibility in how they take these
factors into consideration... ,413
EPA, by contrast, is not given authority to consider these factors independently, but instead "[i]n
determining whether the State's goal for visibility improvement provides for reasonable progress
... the Administrator will evaluate the demonstrations developed by the State... ,"414 Indeed,
EPA itself has explained that, "[a]s long as this evaluation is done adequately and the States
provide a reasoned basis for their decisions, EPA will defer to the State" in its reasonable
progress determinations.415 As we discuss below, EPA has abandoned that approach—the
approach it has taken with respect to every other State's regional haze SIP—in its review of the
Texas SIP, without explanation or justification.
Luminant explained that EPA guidance has consistently confirmed the discretion and flexibility
given to States that is inherent in determining reasonable progress. EPA explained when issuing
its regional haze regulations: "The final rule provides States flexibility in determining the
amount of progress that is 'reasonable' in light of the statutory factors, and also provides
flexibility to determine the best mix of strategies to meet the reasonable progress goal they
select."416 And in its regional haze guidance to the States for developing their SIPs for the first
planning period, EPA again emphasized that the regulations "give[] States wide latitude to
determine additional control requirements" and, in applying the four statutory factors, States
"have flexibility in how to take into consideration these statutory factors and any other factors
that you have determined to be relevant."417 A State's discretion extends to all aspects of the
reasonable progress determination, including "whether given emission reduction measures are
reasonable" and whether "the uniform glide path is not reasonable based on the application of the
factors."418
Luminant asserted that the EPA's proposed disapproval thus flies in the face of the CAA, Texas'
primary role, and substantial flexibility States are given in determining reasonable progress, even
by EPA's own admission and guidelines. EPA does not dispute that Texas applied the four
statutory factors, and did so in a manner consistent with EPA's regulations and guidance. EPA
does not dispute that, following multi-year planning and consultation among CENRAP States,
Texas and the other States agreed on coordinated goals and strategies for improving visibility at
the Class I areas. But EPA's proposal credits none of that. Instead, EPA simply asserts that
Texas and Oklahoma should have conducted their analysis and State coordination differently.
The Clean Air Act, however, does not authorize EPA to second-guess the States' choices, so
long as the statutory standards are met, as they are here.419 The nature of the reasonable progress
test is to allow the regulator to exercise discretion.420 TCEQ here exercised such judgment and
discretion. The facts were presented and TCEQ reached its conclusions, which were made based
not just on the cost of compliance, but with consideration of the four-factors along with
additional information that was pertinent. EPA may not like the result, but that is irrelevant
when, as here, the State is given discretion by the Clean Air Act, the regional haze rule, and
EPA's reasonable progress guidance, and its exercise of that discretion was reasonable and
adequately explained. EPA may believe its goals and analysis are more reasonable than Texas's.
24
-------
But, even if that were true (and, as we show below, it is not), "the CAA requires only that a State
establish reasonable progress, not the most reasonable progress."421
Footnotes:
398 Elec. Power Supply Ass'n v. FERC, 753 F.3d 216, 220 (D.C. Cir. 2014).
399 BCCA Appeal Grp. v. EPA, 355 F.3d 817, 821-22 (5th Cir. 2003).
400 Michigan v. EPA, 268 F.3d 1075, 1083 (D.C. Cir. 2001); see also Fla. Power & Light Co. v. Costle, 650 F.2d
579, 581 (5th Cir. 1981) ("Congress chose a balanced scheme of State-Federal interaction to implement the goals of
the [Clean Air] Act.").
401 42 U.S.C. § 7401(a)(3). See also North Dakota, 730 F.3d at 760-6 ("[T]he CAA grants States the primary role
of determining the appropriate pollution controls within their borders .. . .").
402 42 U.S.C. § 7491(b).
403 Fla. Power & Light Co., 650 F.2d at 587.
404 See, e.g., 42 U.S.C. § 7410(k)(3) ("[T]he Administrator shall approve [a SIP or SIP revision] as a whole if it
meets all of the applicable requirements of this Act.").
405 Luminant Generation Co. LLC, 675 F.3d at 921 (emphasis added) (citing § 7410(k)(3)) ("[T]he [EPA]
Administrator shall approve [a SIP or SIP revision] as a whole if it meets all of the applicable requirements of this
Act." (emphasis added))); Fla. Power &
Light Co., 650 F.2d at 587 ("The great flexibility accorded the States under the Clean Air Act is . .. illustrated by
the sharply contrasting, narrow role to be played by EPA.").
406 CleanCOALition v. TXU Power, 536 F.3d 469, 472 n.3 (5th Cir. 2008).
407 BCCA Appeal Grp., 355 F.3d at 822.
408 Am. Corn Growers Ass'n., 291 F.3d at 2.
409 EPA, Response to Petitions for Reconsideration of Regional Haze Rule 11 (Jan. 10, 2001) ("Response to
Petitions").
410 Id. at 12 (Jan. 10,2001).
411 40 C.F.R. § 51.308(d)(l)(i).
412 Id. § 51.308(d)(l)(i)(A) (emphasis added). See also 64 Fed. Reg. at 35,731 ("Today's final rule requires the
States to determine the rate of progress for remedying existing impairment that is reasonable, taking into
consideration the statutory factors, and informed input from all stakeholders.").
413 77 Fed. Reg. at 30,251 (emphasis added).
414 40 C.F.R. § 51.308(d)(l)(iii) (emphasis added).
415 77 Fed. Reg. 40,150, 40,156 (July 6, 2012) (emphasis added).
416 64 Fed. Reg. at 35,736 (emphasis added).
417 EPA Reasonable Progress Guidance at 4-2, 5-1.
418 EPA, Additional Regional Haze Questions 10 (Sept. 27, 2006).
419 See Luminant Generation, 675 F.3d at 921; Cf. Alaska Dep't of Envtl. Conservation v. EPA, 540 U.S. 461, 491
(2004) ("Only when a State agency's BACT determination is 'not based on a reasoned analysis may EPA step in to
ensure that the statutory requirements are honored." (internal citations omitted)).
420 Reasonable Progress Guidance at 5-1.
421 North Dakota, 730 F.3d at 768.
Response: We do not agree that our partial disapproval of the Texas Regional Haze SIP is
contrary to the CAA, the Regional Haze Rule, or relevant case law. Congress crafted the CAA
to provide for States to take the lead in developing implementation plans, but balanced that
decision by requiring us to review the plans to determine whether a SIP meets the requirements
of the CAA. EPA's review of SIPs is not limited to a ministerial review and approval of a State's
decisions. EPA must review the State's SIP submittal and determine whether its meets the
requisite Federal requirements. If it does not, the CAA grants EPA authority to impose a Federal
plan to fill in such gaps.
Nothing in the CAA indicates that EPA's role is less important in the context of the regional haze
program than under other CAA programs. On the contrary, CAA Section 110(a)(2)(J) explicitly
25
-------
requires that SIPs "meet the applicable requirements" of Part C of Title I of the CAA including
the requirements for visibility protection set forth in sections 169A and 169B.6 Pursuant to Section
169A(b), EPA is required to promulgate visibility protection regulations that apply to "each
applicable implementation plan" (i.e., each SIP or FIP)7 for each State containing one or more
Class I areas and each State "emissions from which may reasonably be anticipated to cause or
contribute to any impairment of visibility in any [Class I area] " 8 The CAA specifies that these
regulations (including the Regional Haze Rule) must require each such SIP or FIP to "contain such
emission limits, schedules of compliance and other measures as may be necessary to make
reasonable progress toward meeting the national goal," as determined by the State (or by us in the
case of a FIP).9 Thus, we have an undiminished obligation to review State plan submissions for
compliance with Regional Haze Rule requirements. As with all other cases of our action on plan
submissions, we are bound to follow the statutory direction set forth in CAA Section 110(k).
Pursuant to CAA Section 110, States must submit SIPs to us for review and we must review SIPs
for compliance with the Act's requirements and may not approve any SIP revision that "would
interfere with any applicable requirement" of the Act.10 Furthermore, the Act mandates that we
promulgate a FIP when we find that a State has failed to submit a required SIP to the Agency,
failed to submit a complete SIP, or where we disapprove a SIP in whole or in part.11 Thus, the
CAA provides us with a critical oversight role in ensuring that SIPs meet the Act's requirements.
The cases cited by Luminant do not support an argument that our role as a reviewer is any less
critical in the regional haze context than it is in reviewing other SIP components. We agree that
the CAA places the requirements for developing Regional Haze plans and determining RPGs on
States. As discussed above, our role is to review the Regional Haze SIP submittal including the
RPG determinations and determine if the State met the applicable statutory and regulatory
requirements. While the court in American Corn Growers found that we had impermissibly
constrained State authority, it did so because it found that we forced States to require BART
controls without first assessing a source's particular contribution to visibility impairment. This is
not the case with our action. EPA must ensure that Regional Haze SIPs comply with the applicable
statutory and regulatory provisions. Oklahoma v. EPA, 723 F.3d 1201, 1208 (10th Cir. 2013).
Likewise, in the present action, we are disapproving portions of the Texas and Oklahoma RH SIPs
because those pieces did not meet the requirements of the Regional Haze Rule, and our detailed
responses to those portions of our disapproval are explained elsewhere.
6 CAA sections 110(a)(2)(J), 169A and 169B 42 U.S.C. 7410(a)(2)(J), 7491 and 7492.
7 Under the CAA, "applicable implementation plan" is defined as "the portion (or portions) of the implementation
plan, or most recent revision thereof, which has been approved under [CAA section 110], or promulgated under
[CAA section 110](c) * * * and which implements the relevant requirements of [the CAA]." CAA section 302(q),
42 U.S.C. 7602(q). In other words, an "applicable implementation plan" is an EPA-approved SIP or Tribal
Implementation Plan, or an EPA-promulgated FIP.
8 42 U.S.C. 7491(b)(2). In promulgating the Regional Haze Rule, EPA determined that "all States contain sources
whose emissions are reasonably anticipated to contribute to regional haze in a Class I area and, therefore, must
submit regional haze SIPs."64 FR 35720; see also 40 CFR 51.300(b)(3).
9 42 U.S.C. 7491(b)(2).
10 CAA section 110(a)(1), (k)(3) and (1), 42 U.S.C. 7410(a)(1), (k)(3) and (1).
11 See id. Section 7410(c)(1).
26
-------
We have the authority to issue a FIP either when we have made a finding that the State has failed
to timely submit a SIP or where we have found a SIP deficient. Here, we have authority on the
latter grounds, and we have approved those pieces of the SIP that meet Federal regional haze
requirements and adopted a FIP to fill the remaining gaps. Our action today is consistent with
the statute.
Several cases are cited by Luminant asserting that the State has the primary role in air pollution
prevention and that CAA confines us to the ministerial function of reviewing SIPs for consistency
with the Act's requirements, including Luminant Generation v. EPA, 675F.3d917 (5 th Cir. 2012),
Florida Power & Light Co. v. Costle, 650 F.2d 579 (5th Cir. 1981), and N.I), v. United States
EPA, 730 F.3d 750 (8th Cir. 2013).
Luminant Generation v. EPA involved our disapprovals of SIP revisions involving Texas' minor
new source review (NSR) program. As noted by the Luminant court, "because 'the Act includes
no specifics regarding the structure or functioning of minor NSR programs' and because the
implementing regulations are 'very general [,]... SIP-approved minor NSR programs can vary
quite widely from State to State.'" 12 By contrast, Regional Haze SIPs are subject to detailed
requirements set forth in CAA sections 169A and the Regional Haze Rule. In Luminant, the
Fifth Circuit found that we failed to tie our disapproval to any requirement of the CAA or our
implementing regulations. 13 In Florida Power & Light Co. v. Costle, the court held that we
must approve a SIP that "meets statutory criteria." In this case, our partial disapproval is based
on the SIP's failure to comply with portions of the Federal regional haze requirements as detailed
elsewhere in our responses to comments and in our final action.
In N.D. v. EPA, Luminant quoted the court as stating, "[T]he CAA grants States the primary role
of determining the appropriate pollution controls within their borders..." However, this quote is
not fully reproduced, leading to an inaccurate characterization of the court's opinion. The full
sentence of the court Stated, "Although the CAA grants States the primary role of determining
the appropriate pollution controls within their borders, EPA is left with more than the ministerial
task of routinely approving SIP submissions."(emphasis added) 14
The court in North Dakota also held that the CAA and the States operate under a framework of
"cooperative Federalism."15 Under this framework, the court stated that the CAA left the
individual States to make pollution restriction for particular emitters within that State. "But, if a
State fails to submit a SIP, submits an incomplete SIP, or submits a SIP that does not meet the
statutory requirements, EPA is obligated to implement its own FIP to correct the deficiency in
the SIP, unless the State can correct the deficiency itself and EPA can approve that correction
within two years. 42 U.S.C. 7410(c). This is commonly referred to as cooperative Federalism,
and both Section 169A and Section 110 operate under this framework." 16
12 Luminant Generation Co. LLC v. EPA, 675 F.3d 917, 921 (5th Cir. 2012) (citing 74 FR 51418, 51421 (Oct. 6,
2009).
13 Id. at 924, 929; 690 F.3d at 679, 682, 686.
14 730 F.3d at 760-61.
15 Id. at 57.
16 Id.
27
-------
Luminant also places emphasis on our prior Statements. These Statements are not as supportive
of Luminant's position as it suggests. For example, "some flexibility" does not suggest
unfettered flexibility; a report's suggestion that a cooperative approach would make sense does
not suggest that we will or must approve unilateral decision-making by a State no matter what.
We agree that the States are given flexibility in establishing reasonable progress goals, but they
are required by the CAA to consider certain factors. Whether one characterizes our role as
limited or not limited in reviewing RH SIPs, we must determine if the State's SIP meets the
applicable statutory and regulatory requirements. The States' analyses and determinations were
not approvable for reasons discussed elsewhere in this notice and the proposed rulemaking.
While States have the authority to exercise different choices in setting RPGs, such decisions
must be reasonable and consistent with statutory and regulatory requirements. Here, the States'
errors were such that we cannot conclude that each State's decision met this standard. Our
disapproval of portions of the Regional Haze SIP has an appropriate basis in our CAA authority.
With regard to the comment that our guidance only requires a four-factor analysis for potentially
affected sources, we note that our RPG Guidance States the following:
In determining reasonable progress, CAA § 169A(g)(l) requires States to take into
consideration a number of factors. However, you have flexibility in how to take
into consideration these statutory factors and any other factors that you have
determined to be relevant. For example, the factors could be used to select which
sources or activities should or should not be regulated, or they could be used to
determine the level or stringency of control, if any, for selected sources or activities,
or some combination of both. 17
Our guidance for setting RPGs also provides that:
The RHR gives States wide latitude to determine additional control requirements,
and there are many ways to approach identifying additional reasonable measures;
however, you must at a minimum, consider the four statutory factors. Based on the
contribution from certain source categories and the magnitude of their emissions
you may determine that little additional analysis is required to determine further
controls are not warranted for that category. 18
Although the State has flexibility in how to consider the four statutory factors, it must do so in a
reasonable manner. As we discuss in our proposal,19 and elsewhere in this response to comments
we do not believe that Texas complied with these requirements. Thus, we disagree with
Luminant's assertion that, "EPA does not dispute that Texas applied the four statutory factors,
and did so in a manner consistent with EPA's regulations and guidance." We also disagree with
Luminant that the Regional Haze Rule confers approval as long as the State addresses the
17 See EPA's Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program (June 1, 2007),
Section 5.0.
18 See EPA's Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program (June 1, 2007),
Section 4.2.
19 See discussion beginning on 79 FR 74838.
28
-------
requirements in some form or fashion. A State's reasonable progress demonstration must satisfy
the Act and our regulations, and Texas' demonstration failed to do so.
Luminant also stated that EPA's proposal did not credit the consultation and coordination among
the CENRAP States regarding visibility improvement strategies. We disagree with this
assertion. Please see our responses to commenters regarding consultation elsewhere in this
document.
Comment: [Associations (0059) p. 2-4, 21] The Associations Stated that the EPA, in this
proposal, seeks to dramatically increase its own authority over the regional haze program at the
expense of the States and Tribes to whom Congress gave a primary role in implementing the
regional haze program.
The Associations Stated that the EPA's proposal to disapprove of Texas' and Oklahoma's State
Implementation Plans ("SIPs") and impose Federal Implementation Plans ("FIPs") in their place
far exceeds EPA's legal authority under the Clean Air Act and would fundamentally transform
the structure of the regional haze program from a State-driven program based on cooperative
Federalism to a centralized, Federal program with little real input from States or regulated
entities. Nothing in the Clean Air Act or the administrative record supports EPA's determination
to supersede the States' judgment in mandating $2 billion in new emission controls that will have
no perceptible impact on visibility. If finalized, EPA's proposal could create dangerous
precedent that could be used by EPA in the future to disregard the decisions made by other States
under the regional haze program, impose requirements found nowhere in the Clean Air Act or
EPA's own regulations, and require States and industry to undertake significant and costly
regulatory burdens disproportionate to any visibility benefit.
The Associations noted, in recognition of diminished visibility at national parks and other scenic
areas, Congress enacted the Clean Air Act's regional haze provisions with a long-term goal of
returning these areas to a State of natural visibility. At the same time, however, Congress
realized such changes could not be fully realized immediately and adopted an approach by which
States would make incremental improvements over time. Texas and Oklahoma have invested
significant time and resources to understand the sources of regional haze related to their States,
the effect of existing Federal and State programs to reduce emissions from such sources, and to
cooperate with each other and other nearby States to improve visibility. These efforts have been
successful, and measured improvements in visibility conditions at the Big Bend, Guadalupe
Mountain, and Wichita Mountain Class I areas have exceeded the proposed reasonable progress
goals EPA would set for these three areas.
The Associations Stated, despite the fact that real-world, measured air quality demonstrates that
the States are on track to meet the visibility improvements contemplated by Congress and EPA,
EPA has unreasonably proposed to disapprove their SIPs and impose FIPs in their place. The
FIPs would impose emission control requirements on a handful of sources in Texas at significant
cost, based on counterfactual projections that regional haze will somehow get worse,
notwithstanding expected further emissions reductions from levels that have achieved the desired
target today. And even then, EPA projects that these costly emissions controls would achieve
29
-------
only de minimis visibility improvements in 2018 that would not be perceptible to the human eye
and, under EPA's own standards, would round to zero.
The Associations Stated that the EPA's proposal, which, in essence, second guesses the reasoned
decisions made by Oklahoma and Texas in their SIPs, is both unlawful and flatly inconsistent
with EPA's prior administration of the regional haze program where it has routinely approved
SIPs that were functionally equivalent to those of Texas and Oklahoma without subjecting them
to the same level of scrutiny. The Clean Air Act gives States primacy in implementing the
regional haze program and limits EPA's review of regional haze SIPs to an analysis of whether
or not the State has complied with statutory and regulatory requirements. Despite the fact that
Texas and Oklahoma followed all applicable regulatory requirements for developing regional
haze SIPs, EPA here proposes unlawfully to second-guess the States' decisions and to substitute
its own judgment for that of the States. EPA compounds this error by applying an unlawful
methodology that focuses on emission controls at individual sources rather than source
categories and that places undue reliance on visibility benefits to the detriment of the statutory
factors mandated by Congress.
Response: We take no position of the Associations' characterization of the intent of Congress.
The role of States in the regional haze context, disapproval of SIPs, and cooperative Federalism
has already been addressed above. We disagree with the Associations that our proposal was
"based on counterfactual projections that regional haze will somehow get worse, notwithstanding
expected further emissions reductions from levels that have achieved the desired target today."
Our modeling does not project that visibility will worsen. In addition, the intent of the Regional
Haze Rule is not limited to the maintenance of visibility. As we state in the Regional Haze
Rule,20 "The reasonable progress goals must provide for an improvement in visibility for the
most impaired days over the period of the implementation plan and ensure no degradation in
visibility for the least impaired days over the same period." We determined (among other things)
that Texas' reasonable progress and long-term strategy demonstration was flawed. Our FIP
corrects these and other flaws in Texas' demonstration and will result in additional visibility
improvement at the Class I areas in Texas and Oklahoma.
We address the issues of perceptibility, the significance of the anticipated visibility benefits, and
source versus source category impact analysis in our responses to comments that raise these
issues in more detail.
Comment: EPA's proposal unlawfully ignores State primacy in developing regional haze
plans. [Associations (0059) p. 7-9, 12-13]
The Associations Stated that the EPA's proposal to disapprove Texas and Oklahoma's regional
haze SIPs is unlawful because it disregards the primary role given to States in implementing the
regional haze program. The Clean Air Act is built on principles of cooperative Federalism that
require EPA to defer to States in developing implementation plans so long as necessary statutory
requirements are met. EPA's proposal ignores those limits and would impose FIPs that ignore
the primary implementation role given to Texas and Oklahoma. Not only is this approach
20 64 FR 35766.
30
-------
inconsistent with the Clean Air Act and EPA's past practice, it would give EPA unprecedented
power arbitrarily to substitute its own judgment for that of the States at virtually any stage of the
implementation process.
The Associations noted that, as courts have recognized, the Clean Air Act was intended by
Congress to be "a model of cooperative Federalism." Sierra Club v. Korleski, 681 F.3d 342, 343
(6th Cir. 2012); see also Michigan v. EPA, 268 F.3d 1075, 1083 (D.C. Cir. 2001); Florida
Power & Light Co. v. Costle, 650 F.2d 579, 581 (5th Cir. 1981). Under this structure, Congress
specifically found that "air pollution prevention ... is the primary responsibility of States and
local governments." 42 U.S.C. § 7401(a)(3). Relying on that finding, the Eighth Circuit
recently noted that "the [Clean Air Act] grants States the primary role in determining the
appropriate pollution controls within their borders." North Dakota, 730 F.3d at 760-61. Within
that context, "States have broad authority to determine the methods and particular control
strategies they will use to achieve the statutory requirements." BCCA Appeal Group v. EPA, 355
F.3d 817, 822 (5th Cir. 2003).
In contrast, after it has established broad emission standards, the Associations asserted that the
EPA's role is limited to ensuring that the States' implementation of those standards is consistent
with the Act. See Florida Power & Light, 650 F.2d at 587 ("The great flexibility accorded the
States under the Clean Air Act is ... illustrated by the sharply contrasting, narrow role to be
played by EPA."). In that narrow role, EPA is confined to "the ministerial function of reviewing
SIPs for consistency with the Act's requirements." Luminant Generation Company LLC v. U.S.
EPA, 675 F.3d 917, 921 (5th Cir. 2012). In other words, EPA cannot second-guess the States,
but must approve any SIP that complies with basic statutory requirements. See 42 U.S.C. §
7410(k)(3) ("The Administrator shall approve [a SIP or SIP revision] as a whole if it meets all of
the applicable requirements of this chapter.") (emphasis added). When, as here, an agency is
given a mandatory command (e.g., "shall") to base its decision on a limited set of factors
prescribed by statute, it cannot depart from Congress' direction by considering additional factors
not listed in the statute. See National Ass'n of Home Builders v. Defenders of Wildlife, 551 U.S.
644 (2007) (rejecting interpretation of Clean Water Act that would have allowed EPA to deny
transfer of permitting authority to State agencies based on additional factors when the statute
Stated that EPA "shall approve" transfer where "nine specified criteria are satisfied").
The Associations Stated that the EPA has previously recognized its limited role in implementing
the regional haze program and explained that "[t]he final regional haze rule ... provide[s] States
considerable discretion in establishing reasonable progress goals for improving visibility in Class
I areas." EPA, Response to Petition for Reconsideration of Regional Haze Rule 11 (Jan. 10,
2001). Thus, rather than establishing strict implementation requirements, EPA's rule "requires
States to determine the rate of progress for remedying existing impairment that is reasonable,
taking into consideration the statutory factors, and the informed input from all stakeholders." 64
Fed. Reg. at 35,731; see also 40 C.F.R. § 51.308(d)(l)(i)(A). In its guidance to States, EPA
further emphasized that the regional haze rule "gives States wide latitude to determine additional
control requirements" and, in applying the required statutory factors, the States "have flexibility
in how to take into consideration these statutory factors and any other factors that you have
determined to be relevant." EPA, Guidance for Setting Reasonable Progress Goals Under the
Regional Haze Program 4-2, 5-1 (June 1, 2007). Thus, EPA has explained that "[a]s long as this
31
-------
evaluation is done adequately and the States provide a reasoned basis for their decision, EPA will
defer to the State" with respect to reasonable progress determinations. 77 Fed. Reg. 40,150,
40,156 (July 6, 2012).
In this proposal, however, the Associations noted that the EPA ignores these well-established
principles of cooperative Federalism, disregards the reasoned judgment of Texas and Oklahoma,
and seeks to impose a radically different implementation plan based on its own independent
analysis. This is contrary to the Clean Air Act and unlawful. In the absence of any showing that
Texas and Oklahoma failed to comply with the statutory and regulatory requirements for
developing reasonable progress goals and long- term strategies, EPA has no choice under the
Clean Air Act but to approve these SIPs.
The Associations Stated that by going beyond its limited authority under the Clean Air Act
strictly to review States' compliance with applicable statutory and regulatory requirements in
establishing reasonable progress goals and long-term strategies, EPA would fundamentally
transform the regional haze program into a Federal program over which State decisions could be
vetoed for virtually any reason by EPA. Given the nature of regional haze issues, States spend
significant resources modeling and tracking emissions from a wide variety of sources and
consulting with neighboring States prior to developing SIPs to address regional haze. Those
efforts will be largely futile if EPA can later substitute its own judgment for that of a State's
decision by conducting what is in essence a de novo review of a State's reasonable progress
goals and long-term strategy. Such an approach deprives States of any certainty in implementing
the regional haze program and is in direct contradiction to the cooperative Federalism principles
upon which the regional haze program was based. If EPA proceeds to disapprove Texas and
Oklahoma's SIPs, it will set a dangerous precedent that will threaten the primacy of all States to
use their judgment to establish reasonable progress goals and long-term strategies that will
improve visibility over time while respecting other State interests that are recognized by the
Clean Air Act.
Response: The cases cited in this comment regarding cooperative Federalism, EPA's role in
reviewing SIP submittals and approving SIPs, and EPA's Regional Haze guidance have already
been addressed elsewhere.
Comment: [TCEQ/PUCT (0056) p. 1-3] The TCEQ opposed the proposed partial disapproval
of Texas' 2009 Regional Haze (RH) SIP or adoption of the proposed FIP. The EPA's proposed
partial SIP disapproval and FIP ignores the flexibility the Clean Air Act (CAA) provides to
States in crafting regional haze plans and thus is arbitrary, capricious, and an abuse of discretion
The EPA should withdraw this proposal and propose to approve the TCEQ's 2009 RH SIP as
meeting the statutory and regulatory requirements for regional haze.
The TCEQ submitted a RH SIP that meets all requirements of the CAA and the regional haze
rule (RHR). The 2009 RH SIP includes a detailed analysis of each requirement of a regional
haze plan, as identified in CAA section 169A(b)(2) including: a determination of which sources
are subject to Best Available Retrofit Technology (BART); reasonable progress goals for the
State's Class I areas, based on the four statutory factors; calculations of baseline and natural
32
-------
visibility conditions; consultations with States; and a long-term strategy and a monitoring
strategy.
The TCEQ argued that EPA bears the burden to show Texas' judgment was unreasonable or does
not meet the statutory requirements. As the U.S. Supreme Court opined in Alaska Dept. of
Environmental Conservation v. EPA (540 U.S. 461, 484-89 (ADEC)): in reviewing an EPA
disapproval of a State's exercise of discretion, courts must defer to State judgments, and the EPA
bears the burden of establishing that those judgments were unreasonable. States are due even
greater deference under CAA, section 169A (USC 7491) than under the standard articulated
under the Supreme Court's decision in ADEC1 The RHR and EPA guidance suggest that States
have a large degree of flexibility in crafting regional haze plans.
The TCEQ stated that the EPA's determination that the TCEQ did not meet all applicable
requirements of the CAA regarding regional haze is flawed. The State plan submitted in 2009
followed all the EPA rules and guidance and contains a thorough analysis and justification for its
conclusions for each statutorily required element. The EPA States that the TCEQ did not
'reasonably consider' the four statutory factors in developing the reasonable progress goals
(RPG) for its Class I areas, Big Bend and Guadalupe Mountains National Parks. The CAA
requires States to develop RPGs "tak[ing] into consideration" the factors listed in section
169A(g)(l). Texas' plan does this. The EPA's complaint is that it would have considered these
factors differently than Texas. This is not a valid basis for disapproval of the Texas plan. The
EPA proposes to find that it would have developed certain elements of the visibility plan
differently, thus holding Texas to a different standard of compliance than what is provided for in
statute and rule. This is the very nature of an arbitrary and capricious action. The EPA also
proposed that the Texas uniform rate of progress (URP) is faulty because it assumes the TCEQ's
natural visibility conditions estimate is incorrect.2 This is an estimate that was developed by the
TCEQ following the EPA's own guidance and rules that provide the States broad flexibility and
discretion in their calculation. Again, it appears the EPA prefers a different outcome than that of
the Texas plan. The EPA's proposed disapproval of the long-term strategy for Wichita
Mountains in Oklahoma is based on new and unfounded interpretations without basis in the CAA
or its rules. First, the EPA claims that the four statutory factors for RPGs apply to the long-term
strategy. This is not found in the statute and is not supported by the RHR. The EPA also
proposes disapproval of the long-term strategy and State consultations - in which both States
agreed with the reductions calculated for sources in Texas that impacted the Wichita Mountains-
because Oklahoma's 'progress goal' established for Wichita Mountains must be "approved or
approvable" in order for Texas to rely on it in its own plan.
It appears that the EPA has carried out the process of developing its proposed partial SIP
disapproval and proposed partial FIP in the following sequence: First, the EPA decided to find a
way to impose additional control requirements beyond those in Clean Air Interstate Rule (CAIR)
on multiple electric generating units (EGU) in Texas. The EPA then analyzed the Texas 2009
RH SIP using new approval criteria that were not in place in either the RHR or in the EPA's
guidance when it was submitted in 2009. Again, the EPA's proposed partial SIP disapproval and
FIP is an attempt to force its preferred outcome for specific sources in Texas. This is arbitrary
and capricious and does not comport with the CAA.
Footnotes:
33
-------
1 See American Corn Growers Assn. v. EPA, 291 F3d. 1 (2002).
2"... we propose to find the TCEQ has calculated this rate of progress on the basis of, and compared baseline
visibility conditions to, a flawed estimation of natural visibility conditions for the Big Bend and Guadalupe
Mountains, as we describe above. Therefore, we propose to disapprove the TCEQ's calculation of the URP needed
to attain natural visibility conditions by 2064." 79 FR 74818, 74833
Response: Section 169A and EPA guidance have already been addressed above. Commenter
cites to Alaska Dept. of Environmental Conservation v. EPA (540 U.S. 461 at 502 (ADEC)) to
support the contention that Congress structured the CAA to limit our authority and that EPA
disapprovals or partial disapprovals are arbitrary and capricious. At issue in the ADEC case was
the State's "new source" permitting SIP, not its Regional Haze SIP, and a special statutory
provision concerning oversight of State permitting decisions made pursuant to its SIP21. This
case is not relevant to our approval of a Regional Haze SIP. Here, our disapproval is based on
the State's failure to satisfy several regional haze requirements as detailed elsewhere in our
responses to comments. Our role in the SIP review process has already been addressed at length
above.
We disagree with the TCEQ's apparent suggestion that we disapproved parts of the Texas SIP
merely because we preferred something different. The bases for our disapproval were
determined by our substantive review of the SIP submittal against the requirements specified by
the CAA and our regulations, not from arbitrary preferences. The specific bases for disapproval
were identified in our proposal and have been further explained at length in responding to
comments and in our final action. To take one example that we elaborate on more fully
elsewhere, we disagree that Texas' natural conditions calculation followed our guidance. The
core reason we proposed to disapprove Texas' natural conditions calculation was because it
failed to adequately support the assumption that 100% of the fine soil and coarse mass that
contributed to visibility impairment during the baseline period at its Class I areas was entirely
natural.22 We disagree with the TCEQ that we analyzed the Texas SIP with new approval
criteria that was not in place at the time Texas submitted its SIP. We explained our rationale and
why we believe our proposal comports with the CAA and our Regional Haze Rule beginning in
Section IV of our proposal.23
EPA deference to the State in developing regional haze SIPs has already been addressed in this
section above. We disagree with TCEQ regarding our disapproval of Texas' reasonable progress
goals and its reasonable progress four-factor analysis for the reasons we have detailed in our
proposal, final and as further discussed in our responses to other more detailed comments
concerning these issues. We have explained in detail elsewhere how the four-factor analyses,
21 The Court held recognized that while States have wide discretion in formulating their plans; SIPs must include
certain measures Congress specified to assure that national ambient air quality standards are achieved. Id. at 470.
The Court also held that the EPA has supervisory authority over the reasonableness of State permitting authorities
and may issue stop construction orders if a BACT selection is not reasonable. Id. at 502. The Court also held that
EPA's stop construction orders were neither arbitrary nor capricious. Id.
22 "States are free to develop alternative approaches that will provide natural visibility conditions estimates
that are technically and scientifically supportable. Any refined approach should be based on accurate, complete, and
unbiased information and should be developed using a high degree of scientific rigor." Guidance for Estimating
Natural Visibility Conditions Under the Regional Haze Rule, U.S. EPA, September 2003, pg 1-11.
23 79 FR 74823.
34
-------
reasonable progress goals, and long-term strategies are inextricably linked. With regard to the
consultation, we also discuss that elsewhere in the responses to comments.
Comment: [TCEQ/PUCT (0056) p. 16] The TCEQ Stated that the EPA's proposed FIP is
contrary to authority provided in the CAA. The statute provides the EPA with authority to
address State plans that it believes are substantially inadequate to comply with the Act's
requirements. The EPA Regional Haze Rule identifies periodic reviews and plan updates as the
remedy for addressing RH SIPs that are inadequate.
The TCEQ noted that in order to promulgate a FIP, the CAA requires that the EPA disapprove a
State plan in whole or in part for not meeting the applicable requirements of §110(k). Texas'
plan was complete by operation of law and met all requirements. The EPA has no authority to
impose a FIP that merely replaces the EPA's judgment for Texas' but does not correct an error or
is not based on a failure of Texas' plan to meet the requirements of the Regional Haze Rule or
CAA.13
The TCEQ Stated that the EPA's Regional Haze Rule established the remedy for a substantially
inadequate plan as periodic updates, not a Federal plan.14 The nature of regional haze and the
statutory requirement for reasonable progress and long-term solutions to visibility impairment
require regular updates and reviews of State plans by the States themselves. Thus, the very
nature of regional haze planning recognizes that the solution to plans that don't make adequate
progress towards the natural visibility condition goal is an update of the plan, not a FIP.
Footnotes:
13 See Train, 421 U.S. 60, 79 "The CAA gives the [EPA] no authority to question the wisdom of a State's choice of
emission limitations if such choices are part of a plan which satisfies the standards of 110(a)(2)."
14 See 64 FR 3574S: "... section 110(a)(2)(F) of the CAA provides that SIPs are to require 'periodic reports on the
nature and amounts of emissions and emissions-related data' and 'correlation of such reports .... with any emission
limitations or standards establish pursuant to this chapter.' Moreover, section 110(a)(2)(H) requires SIPs to provide
for revision when found to be substantially inadequate to 'comply with any additional requirements established
under... [the CAA]."'
Response: In this seminal case, the Supreme Court recognized the basic principle that we must
review SIP revision submittals for compliance with the requirements of CAA Section
110(a)(2).24
Our action does not contradict the Supreme Court's decision in Train. States have significant
responsibilities in implementation of the CAA and meeting the requirements of the Regional
Haze Rule. We recognize that States have the primary responsibility of drafting an
implementation plan to address the requirements of the Federal regional haze program. We also
recognize that we have the responsibility of ensuring that the State plans, including RH SIPs,
24 See Train, 421 U. S. 60, 79 ("Under § 110(a)(2), the Agency is required to approve a State plan which provides for
the timely attainment and subsequent maintenance of ambient air standards, and which also satisfies that section's
other general requirements. The Act gives the Agency no authority to question the wisdom of a State's choices of
emission limitations if they are part of a plan which satisfies the standards of section 110(a)(2) . . ." (emphasis
added)).
35
-------
conform to the CAA requirements. We cannot approve a RH SIP that fails to address adequately
several elements of the Federal regional haze program.
Contrary to commenter's assertion, we have not destroyed the State's primacy. In fact, we have
approved those portions of the State's submittal that we believe meet our Federal regional haze
requirements. We are only disapproving the portions of the Texas' submittal that do not meet
the Act and EPA rules. As explained in a comment response above, we do not agree with
commenters' asserted position that the remedy for an unapprovable RH SIP is periodic updates.
The RHR's requirements for comprehensive periodic revisions (see 40 CFR 51.308(f)) and
periodic progress reports (see 40 CFR 51.308(g)) are very different from the authority to impose
a FIP when there is a determination that a SIP is not approvable. As we have stated elsewhere,
EPA has the authority and obligation to impose a FIP to fill in such gaps. The provisions of the
RHR do not override this responsibility.
As explained in our proposal, we identified several deficiencies in the Texas and Oklahoma SIP
submittals and proposed to disapprove those. As such, we have a FIP obligation to cure such
deficiencies with a Federal plan to fill in those gaps.
We agree with the TCEQ the plan was complete by operation of law. We disagree with the
TCEQ that just because the Texas' plan was complete by operation of law, this means we are no
longer allowed to review it to ensure it has met all requirements, and that we have no authority to
impose a FIP. The TCEQ confuses the action of merely submitting a SIP and having it deemed
complete, with the process of reviewing that SIP for compliance with the applicable Federal
requirements.
Texas submitted a Regional Haze SIP revision on March 31, 2009, which later became complete
by operation of law. However, this only means the Texas SIP was deemed "an official
submission for purposes of review." This determination has no bearing on whether the SIP met
the requirements of the Regional Haze Rule. As we explain in our proposal in detail, the Texas
SIP failed to do so in a number of areas.
Furthermore, while we agree that the procedural requirements for promulgation of a FIP under
110(c) are set forth in CAA Section 307(d), we do not agree that our action violates that
provision in any way. We have in fact met those requirements, as explained in our proposed
notice and comment rulemaking. Consistent with the requirements of that section, our proposal
included a summary of the factual data on which our proposed FIP was based, as well as the
methodology used in obtaining the data and in analyzing the data and the major legal
interpretations and policy considerations underlying the proposed action and FIP. 25 In addition,
we provided a detailed evaluation of the Texas and Oklahoma RH SIPs' analyses for the relevant
units, which formed the basis for our proposed action on those portions of the Texas and
Oklahoma Regional Haze SIPs. 26 This final rulemaking includes similar information with
respect to the SIP and the FIP, as well as "an explanation of the reasons for any major changes in
25 See CAA section 307(d)(3), 42 U.S.C. 7607(d)(3).
26 The SIP portion of our action is subject to the procedural requirements of section 553(b) of Administrative
Procedure Act (APA), 5 U.S.C. 553(b), rather than the requirements of CAA subsection 307(d), 42 U.S.C. 7607(d).
36
-------
the promulgated rule from the proposed rule" and "a response to each of the significant
comments, criticisms, and new data submitted in written or oral presentations during the
comment period."27 Therefore, our action complies with the applicable procedural requirements
of the CAA.
Comment: [TCEQ/PUCT (0056) p. 16-17] The TCEQ Stated that the CAA gives States
authority to develop regional haze plans that reflect State needs. The EPA should not get
deference for its own choices in its FIP over those of Texas.
The TCEQ Stated that the EPA's interpretation of its authority to review regional haze
submissions under CAA, §169A is flawed. While the EPA review and State revision of regional
haze SIPs is a component of §110, the CAA also provides an independent grant of authority to
States, and specific language identifying the EPA authority to establish goals and guidance for
regional haze. The use of the word "guideline" in the in §169A evidences a clear congressional
intent that States be granted wide latitude in decision-making here. CAA, §169A inherently
limits the EPA's SIP approval and review authority in §110.
The TCEQ Stated that the EPA's only complaint regarding the 2009 Texas SIP is that it would
have taken a different approach to meet the statutory and regulatory requirements. The EPA's
suggested reliance on the NCII default values in estimating natural visibility conditions at Big
Bend and Guadalupe Mountains rather than the FLM's 80% approach was not adequately
justified and therefore is unreasonable.
Response: The issue of States' latitude in decision making under CAA Section 169A has already
been addressed above.
We agree that the CAA places the requirements for developing Regional Haze plans on States.
As discussed above, EPA's role is then to review the Regional Haze SIP submittal for
compliance with the applicable statutory and regulatory requirements. While the court in
American Corn Growers found that EPA had impermissibly constrained State authority, it did so
because it found that EPA forced States to require BART controls without first assessing a
source's particular contribution to visibility impairment. This is not the case with our action. We
are not forcing Texas to adopt a particular measure or to weigh the statutory factors in a
particular way. Rather, we are disapproving portions of Texas' Regional Haze SIP because the
State's analysis and conclusions were flawed for those portions and failed to support their final
determinations. See Oklahoma v. EPA, 723 F.3d 1201, 1210 (10th Cir. 2013) (holding that the
CAA requires that the BART determination complies with the guidelines, and because the EPA
monitors SIPs for compliance with the statute, it must monitor BART determinations for
compliance with the guidelines). In this situation we are obligated to promulgate a FIP that
addresses the deficiencies in the SIP to fill those gaps.
We disagree with the TCEQ's assertion that our "reliance on the NCII default values in
estimating natural visibility conditions at Big Bend and the Guadalupe Mountains rather than the
FLM's 80% approach was not adequately justified and therefore is unreasonable." First, Texas
27 CAA section 307(d)(6)(A) & (B), 42 U.S.C. 7607(d)(6)(A) and (B).
37
-------
did not rely on an "80% approach" in calculating its natural conditions. Texas relied on an
approach in which 100% of fine soil and coarse mass was assumed to be natural. That
assumption was not supported or adequately documented. The FLMs suggested to Texas that a
value of 80%) of fine soil and coarse mass was natural may be more reasonable, but the SIP
submittal persisted with TCEQ's "own refined estimates" that contain the 100%> assumption. We
respond at length to comments on default values in responding to other comments on the
estimation of natural visibility conditions.
Comment: [CCP (0075) p. 1-2] CCP Stated that the Proposed Rule unreasonably concludes that
Coleto Creek Unit 1 must be retrofitted with a wet Flue Gas Desulfurization (WFGD) scrubber
to control emissions of SO2 to meet EPA goals for reasonable progress toward controlling
regional haze in three Class I areas in Texas and Oklahoma. The Proposed Rule is arbitrary and
unreasonable because it rejects a reasonable regional haze SIP developed by the Texas
Commission on Environmental Quality (TCEQ) and Oklahoma that is fully compliant with
Federal CAA requirements without giving the States the deference they are due in developing
regional haze programs. In proposing to disapprove aspects of the regional haze SIPs, EPA
arbitrarily disregards substantial progress made on actual visibility conditions in Class I areas;
wholly ignores significant contributions to haze from natural conditions and foreign sources; and
impermissibly substitutes its judgment for that of the States in establishing reasonable progress
goals (RPGS) inconsistent with its own rules, guidance, and prior SIP actions. While EPA
recognizes that the Uniform Rate of Progress (URP) glide path need not be met for this planning
period, EPA proposes to reject the States' proposed RPGs in favor of more aggressive EPA
calculated RPGs as justification for the controls it seeks. EPA's actions usurp State authority and
impose costly controls totaling nearly $2 billion despite the fact that the monitoring data now
available demonstrates that EPA's RPGs have already been attained.
Response: For responses that include the referenced State deference issue and substituting EPA
or State judgment for establishing RPGs, please see above responses. We address CCP's
comments on monitoring data and its assertion that Coleto Creek should not install a SO2
scrubber, in our responses to its more detailed comments on these issues elsewhere.
Comment: States have wide discretion to develop their own Regional Haze SIPs. [CCP
(0075) p. 2-3]
CCP Stated that, under the CAA, "air pollution prevention . . . and air pollution control ... is the
primary responsibility of States and local governments." 42 U.S.C. § 7401(a)(3); see also 42
U.S.C. § 7407(a) ("Each State shall have the primary responsibility for assuring air quality
within the entire geographic area comprising such State. . . ."). States have "wide discretion" in
formulating a SIP. Union Elec. Co. v. EPA, 427 U.S. 246, 250 (1976). Once a State submits a
SIP, EPA's role is limited to determining whether the plan satisfies the applicable statutory and
regulatory criteria. 42 U.S.C. § 7410(k)(3). If a SIP satisfies these requirements, the CAA
mandates EPA approval. Id. ("[T]he Administrator shall approve such submittal as a whole if it
meets all of the applicable requirements of this chapter."); see also Train v. Natural Res. Def.
Council, Inc., 421 U.S. 60, 79 (1975) (EPA has "no authority to question the wisdom of a State's
38
-------
choices of emission limitations if they are part of a plan which satisfies the [Act's] standards.");
Luminant Generation Co., LLC v. EPA, 675 F.3d 917, 926 (5th Cir. 2012) ("EPA may consider
only the requirements of the [CAA] when reviewing SIP submissions. . . . [T]he agency [has] no
discretion to do anything other than ensure that a State's submission meets the CAA's
requirements and, if it does, approve it before the passage of [EPA's] statutory deadline.").
CCP Stated that Congress emphasized the primary role of States in CAA regional haze
provisions. For example, Congress directed EPA to "provide guidelines to the States" so that
States, not EPA, could develop SIPs to implement the program. 42 U.S.C. § 7491(b)(1)
(emphasis added); see also American Corn Growers Ass'n v. EPA, 291 F.3d 1 (D.C. Cir. 2002)
(the regional haze rule "calls for States to play the lead role in designing and implementing
regional haze programs").
CCP stated that, under this framework, it falls primarily on the States to make the determinations
required under the program. For each Class I area within a State, these determinations include
the identification of baseline and natural visibility conditions, a calculation of a Uniform Rate of
Progress ("URP") required to achieve those conditions, and a calculation of Reasonable Progress
Goals ("RPGs") for meeting natural visibility conditions. 40 C.F.R. § 51.308(d)(1). If the
States conclude that it is not reasonable to attain the URP within the planning period, the State
must demonstrate why meeting the URP is "not reasonable" and establish an alternative RPG
that is "reasonable." Id. at § 51.308(d)(l)(ii).
CCP stated that the EPA has consistently recognized the States' wide discretion in implementing
the regional haze program and making the required or available determinations:
• "If the State determines that the amount of progress identified through the analysis is
reasonable ... the State should identify this amount of progress as its reasonable
progress goal for the first long-term strategy . . ."64 Fed. Reg. 35,732 (July 1,
1999) (EPA Regional Haze Rule ("RHR")) (emphasis added);
• "As noted in EPA's Reasonable Progress Guidance, the States have wide latitude to
determine appropriate additional control requirements for ensuring reasonable
progress, and there are many ways for a State to approach identification of additional
reasonable measures." 77 Fed. Reg. 11,468 (Feb. 27, 2012) ("Georgia SIP
Approval") (emphasis added);
• "States have significant discretion in establishing RPGs." 76 Fed. Reg. 78,197 (Dec.
16, 2011) ("Kentucky SIP Approval").
CCP stated that the EPA may not simply disapprove the Texas and Oklahoma SIPs because it
disagrees with the States. Rather, courts recognize that EPA has limited discretion when
rejecting a SIP. See Oklahoma v. EPA, 723 F.3d 1201, 1213 n.7 (10th Cir. 2013) ("[w]e
recognize that the EPA has less discretion when it takes actions to reject a SIP than it does when
it promulgates a [Federal Implementation Plan ("FIP")])."
39
-------
Response: Our SIP approval authority and Congressional intent have already been addressed
above.
Comment: EPA failed to afford the States appropriate discretion to develop their own
reasonable URPs and RPGs. [CCP (0075) p. 3]
CCP Stated, despite the substantial deference owed to the States in developing regional haze
SIPs, EPA repeatedly and unreasonably rejected the States' reasonable determinations in the
Proposed Rule simply because it disagreed with their approach. This was impermissible under
the CAA. In particular, while EPA agreed with the States that meeting the URPs for this
planning period was "not reasonable," it disagreed with the "reasonable" URPs and alternative
RPGs developed by the States in order to impose substantial and costly new controls.
CCP Stated that the EPA's rejection of the States' URPs and RPGs is unreasonable for three
fundamental reasons: (1) EPA provides no justification for rejecting TCEQ's refined
assumptions regarding natural conditions for setting its URPs and EPA's reversion back to the
unsupported use of "default" values; (2) the RPGs developed by the States were reasonable, as
confirmed by recent actual monitoring data supporting the attainment of the States' RPGs; and
(3) EPA arbitrarily rejects the States' four-factor analysis in determining RPGS in favor of its
own flawed approach.
Response: Our deference to the State in developing regional haze SIPs and progress
determinations has already been addressed above. We disagree with CCP regarding our
disapproval of Texas' natural conditions, Texas' reasonable progress goals, and its reasonable
progress four-factor analysis for the reasons we have detailed in our proposal and as further
discussed in our response to other more detailed comments concerning these issues.
Comment: [Xcel Energy (0064) p. 6] Xcel Energy stated that the CAA does not allow EPA to
simply substitute its judgment for Texas in establishing a regional haze program. Particularly
where, as here, EPA's Proposal fails to meet the statutory obligations, imposes dramatically
higher costs on Texas sources and electric consumers, and produces admittedly imperceptible,
meaningless visibility improvement in Texas' two Class I areas, EPA cannot justify rejecting
Texas' well-reasoned regional haze plan. EPA should rescind its rejection of the Texas SIP and
approve major portions of the SIP.
Response: The scope of our authority under the CAA has already been addressed above.
Further, we disagree with Xcel that our proposal failed to meet our statutory obligations.
Environmental regulatory measures, including measures to reduce air pollution, do not
necessarily increase the cost of electricity, but even assuming that they may in this case, such an
effect would provide no exemption from CAA requirements. We disagree that our proposal
resulted in meaningless visibility improvement. The controls in our final rule will result in
visibility improvement at the Wichita Mountains and other Class I areas and assist those Class I
areas in attaining the national goal Congress has established for a return to natural visibility
conditions.
40
-------
Comment: EPA is required to accord appropriate deference to Texas in reviewing its
Regional Haze SIP [Xcel Energy (0064) p. 7-9]
Xcel Energy Stated that in developing a "cooperative Federalism" framework under the CAA,
Congress purposely limited EPA's authority by creating a statute in which "air pollution
prevention ... and air pollution control ... is the primary responsibility of States and local
governments." 42 U.S.C. § 7401(a)(3); see also 42 U.S.C. § 7407(a) ("Each State shall have the
primary responsibility for assuring air quality within the entire geographic area comprising such
State.... ").
Xcel Energy stated that the primary avenue for States to implement their responsibility under the
CAA is the promulgation of a SIP. States have "wide discretion" in formulating a SIP. Union
Elec. Co. v. EPA, 427 U.S. 246,250 (1976). Once a State submits a SIP, EPA's role is limited
to determining whether the plan satisfies the applicable statutory and regulatory criteria. 42
U.S.C. § 7410(k)(3). If a SIP satisfies these requirements, the CAA mandates EPA approval.
!d. ("[T]he Administrator shall approve such submittal as a whole if it meets all of the applicable
requirements of this chapter."); see also Train v. Natural Res. Def Council, Inc., 421 U.S. 60, 79
(1975) (EPA has "no authority to question the wisdom of a State's choices of emission
limitations if they are part of a plan which satisfies the [Act's] standards."); Luminant Generation
Co., LLC v. EPA, 675 F.3d 917, 926 (5th Cir. 2012) ("EPA may consider only the requirements
of the [CAA] when reviewing SIP submissions ... [T]he agency [has] no discretion to do
anything other than ensure that a State's submission meets the CAA's requirements and, if it
does, approve it before the passage of [EPA's] statutory deadline.").
Xcel Energy Stated that the CAA's grant of authority to States under Section 169 A is even
broader than in other parts of the CAA, so EPA's deference to the States should be even greater
in the context of regional haze SIPs. Section 169 A of the CAA establishes as a national goal the
prevention of any future, and the remedying of any existing, impairment of visibility in
mandatory Class I Federal areas, where the impairment is the result of manmade air pollution.
42 U.S.C. § 7491(a)(1). But Congress placed extra emphasis on the primary role of States in
CAA regional haze provisions. Congress directed EPA to "provide guidelines to the States" so
that States, not EPA, could develop SIPs to implement the program. 42 U.S.C. § 7491(b)(1)
(emphasis added); see also American Corn Growers Ass'n v. EPA, 291 F.3d 1 (D.C. Cir. 2002)
(the regional haze rule "calls for States to play the lead role in designing and implementing
regional haze programs").
Xcel Energy stated that the legislative history of the CAA confirms Congress's intent to
emphasize State primacy in the regional haze context. Two of the primary sponsors of the
visibility provisions, Senator McClure (R-ID) and Senator Muskie (D-ME), addressed the issue
directly during the legislative debate over passage of the Clean Air Act Amendments of 1977:
Mr. McClure: Under the conference agreement, does the State retain sole authority for
identification of sources for the purpose of visibility issues under this section?
41
-------
Mr. Muskie: Yes; the State, not the Administrator, identifies a source that may impair
visibility....
A Legislative History of the Clean Air Act Amendments Of 1977, at 374-75 (1979) (emphasis
added); see also H.R. REP. No. 95-564, at 155 (1977).
Xcel Energy Stated that the EPA also has repeatedly reaffirmed that States have wide discretion
to make regional haze reasonable progress goal ("RPG") determinations and that States have the
primary role in identifying and addressing RPGs:
• "If the State determines that the amount of progress identified through the analysis is
reasonable ... the State should identify this amount of progress as its reasonable
progress goal for the first long-term strategy .... " EPA Regional Haze Rule, 64 Fed.
Reg. 35,714, 35,732 (July 1, 1999) ("RHR") (emphasis added);
• "As noted in EPA's Reasonable Progress Guidance, the States have wide latitude to
determine appropriate additional control requirements for ensuring reasonable
progress, and there are many ways for a State to approach identification of additional
reasonable measures." Proposed Georgia SIP Approval, 77 Fed. Reg. 11,452,
11,468 (Feb. 27, 2012) (emphasis added);
• EPA has recognized the problems of a rigid requirement to meet a long-term goal.
"EPA made clear in the RHR that the RPG is not a mandatory standard which must
be achieved by a particular date." Id. at 11,473; see RHR, 64 Fed. Reg. at 35,733
("[T]he [RPG] is a goal and not a mandatory standard which must be achieved by a
particular date.");
• "States have significant discretion in establishing RPGs." Kentucky SIP Approval,
76 Fed. Reg. 78,194, 78,197 (Dec. 16, 2011).1
To be consistent with the CAA and EPA's prior SIP determinations under the regional haze
program, Xcel Energy stated that the EPA must review Texas' SIP with appropriate deference
and not simply disapprove the SIP because EPA disagrees with Texas' assumptions,
methodologies, or long-term strategy. Indeed, courts have recognized the distinction between
EPA's limited authority to reject a SIP and its authority to promulgate a FIP. See Oklahoma v.
EPA, 723 F.3d 1201, 1213 n.7 (10th Cir. 2013) ("[w]e recognize that the EPA has less discretion
when it takes actions to reject a SIP than it does when it promulgates a [FIP]"). EPA should be
cautious when it unreasonably rejects a State's first, reasonable, detailed technical conclusion.
Id. at 1225 (J. Kelly, dissenting) (citing Lockheed Martin Corp. v. Admin. Review Bd., US.
Dep't ofLabor, 111 F.3d 1121, 1128-29 (10th Cir. 2013)). Such actions are ripe for judicial
review and challenge. EPA has repeatedly and unreasonably rejected Texas' well-reasoned
technical conclusions in the Proposal. Accordingly, EPA should withdraw its FIP and approve
Texas' SIP.
Footnotes:
42
-------
1 The emphasis on State primacy also is evident in the Best Available Retrofit Technology ("BART") context of
regional haze programs. See BART Rule, 70 Fed. Reg. 39,104, 39,137 (July 6, 2005) ("[T]he Act and legislative
history indicate that Congress evinced a special concern with insuring that States would be the decision makers").
Response: Cooperative Federalism and State discretion in the CAA's regional haze program
have already been addressed above.
Xcel expressed its view that its arguments were reinforced by legislative history of the 1977
CAA amendments. The commenter referred to Statements of Senator Edmund Muskie regarding
the conference agreement on the provisions for visibility protection in those amendments. This
issue was also argued in the Tenth Circuit Court of Appeals in Oklahoma v. EPA, 723 F.3d 1201,
and similarly, the full text of Senator Muskie's Statements was not reproduced. We reproduce it
here, along with Judge Briscoe's ruling:
The Senate discussion about the Conference Report also highlighted the role that
the guidelines play in BART determinations for large power plants:
[Senator] McClure. And while those existing sources are limited to the 28 major
sources contained in the Senate bill's definition of major emitting facilities,
exempting any such source which has the maximum potential to emit less than 250
tons per year, Federal guidelines apply only to fossil-fuel fired generating plants in
excess of 750 megawatts?
[Senator] Muskie. That is correct.
[Senator] McClure. Under the conference agreement, does the State retain sole
authority for identification of sources for the purpose of visibility issues under this
section?
[Senator] Muskie. Yes; the State, not the Administrator, identifies a source that
may impair visibility and thereby falls within the requirement of section 128.
[Senator] McClure. And does this also hold true for determination of "Best
Available Retrofit Technology"?
[Senator] Muskie. Yes; here again it is the State which determines what constitutes
"Best Available Retrofit Technology," as defined in section 128. The Federal
guidelines apply only to the large power plants we have described.
123 Cong. Rec. S26, 854 (daily ed. Aug. 4, 1977) (emphasis added). The last
sentence — omitted by petitioners in their brief — makes clear that the statute
requires that the BART determination here comply with the guidelines. See Pet.
Opening Br. at 15. And because the EPA monitors SIPs for compliance with the
statute, it must monitor BART determinations for compliance with the guidelines.
To be sure, the guidelines themselves might somehow conflict with the statute. But
the petitioners have not argued that any conflict exists here. We therefore hold that
43
-------
the EPA had the authority to review Oklahoma's BART determination with respect
to these two power plants.
As discussed above, the CAA requires the States to follow our BART Guidelines28 when
proposing BART determinations. When States fail to do so, we have authority to disapprove the
BART determinations and promulgate a FIP to cure the deficiency. Similarly, because we have
determined that the Texas and Oklahoma SIPs do not conform with Section 51.308(d) and are
not approvable, we are authorized and at this time required to promulgate a FIP.
Comment: EPA may not supplant Texas' SIP with what EPA believes is a more reasonable
FIP. [GCLC (0063) p. 2-3]
GCLC Stated that Texas' SIP submission meets all statutory and regulatory regional haze
requirements. EPA's analysis of Texas' submission is based on an inappropriate and unsupported
interpretation of the CAA and associated regulations that ignores State primacy and flexibility
provided to the States in the CAA and in the regional haze program in particular.
According to GCLC, Congress delegated to the States the primary responsibility of air pollution
control in the CAA.3 When a SIP meets the basic requirements of the CAA, EPA is required to
approve the SIP submission.4 Section 169A of the CAA specifically places the burden of
developing SIPs and leaves determining whether "reasonable progress" has been achieved to the
States. For example, regarding the four-factor analysis that is central to a State establishing
reasonable progress, EPA itself has recognized that "States have considerable flexibility in how
they take these factors into consideration, as noted in EPA's Guidance for Setting Reasonable
Progress Goals under the Regional Haze Program." 5 EPA has further noted in the recent
regional haze SIP rulemaking for Nebraska that regarding the visibility benefit, best available
retrofit technology ("BART"), and reasonable progress determinations of the State in its SIP,
"[a]s long as this evaluation is done adequately and the States provide a reasoned basis for their
decisions, EPA will defer to the State. "6 EPA regional haze regulations reinforce this deference
to State authority, including (as explained by EPA when issuing those regulations) that "[t]he
final [regional haze] rule provides States flexibility in determining the amount of progress that is
'reasonable' in light of the statutory factors, and also provides flexibility to determine the best
mix of strategies to meet the reasonable progress goal they select." 7
GCLC noted that it is the clear intent of the CAA to provide States flexibility, which has long
been recognized by EPA. This is apparently completely forgotten by EPA in this Proposed FIP.
Reviewing EPA's rule preamble and its Technical Support Documents ("TSD"), including its
Texas Regional Haze State Implementation Plan ("TX TSD"),8 Oklahoma and Texas Regional
Haze Federal Implementation Plans TSD ("FIP TSD"),9 and other supporting documents, it is
immediately apparent that EPA's decision to disapprove Texas' SIP is not because Texas' SIP has
not demonstrated reasonable progress, but rather, it is because EPA believes its FIP will achieve
more reasonable progress than Texas. 10
28 70 FR 39104. See BART Guidelines, Appendix Y to Part 51, beginning on page 39156; Oklahoma v. EPA, 723
F.3d 1201, 1210 (10th Cir. 2013).
44
-------
GCLC asserted that EPA's approach in this rulemaking, to supplant Texas' SIP with one that it
believes is more reasonable, is a direct affront to the cooperative Federalism central to the CAA,
undermines the purpose behind the lead role that States take in the actual implementation of
CAA requirements, and conflicts with Federal court precedent. As Stated by the 8th Circuit
Court of Appeals, "the CAA requires only that a State establish reasonable progress, not the most
reasonable progress." 11 EPA's critiques of the Texas reasonable progress analysis are limited to
questions of the reasoned judgment of the State, and EPA attempts to apply standards that are
simply not supported by the CAA or EPA's implementing regulations. Therefore, EPA should
recognize Texas' primacy, withdraw its FIP, and approve Texas' SIP submission.
Footnotes:
3 See 42 USC§ 7401(a)(3), which States that "Congress finds ... that air pollution prevention (that is, the reduction
or elimination, through any measures, of the amount of pollutants produced or created at the source) and air
pollution control at its source is the primary responsibility of States and local governments ... " (emphasis added).
4 42 USC § 7 41 0(k)(3), which States that "the Administrator shall approve [a SIP] submittal as a whole if it meets
all of the applicable requirements of the CAA.
5 See Approval and Promulgation of Implementation Plans; State of Idaho; Regional Haze State Implementation
Plan, Proposed Rule, 77 Fed. Reg. 30248, 30251 (May 22, 2012).
6 See Approval, Disapproval and Promulgation of Implementation Plans; State of Nebraska; Regional Haze State
Implementation Plan; Federal Implementation Plan for Best Available Retrofit Technology Determination, Final
Rule, 77 Fed. Reg. 40150,40156 (July 6, 2012) ("Nebraska SIP Final Rule").
7 Regional Haze Regulations, Final Rule, 64 Fed. Reg. 35714, 35736 (July I, 1999).
8 U.S. EPA, Technical Support Document for the Texas Regional Haze State Implementation Plan (Nov. 2014) ("TX
TSD").
9 U.S. EPA, Technical Support Document for the Oklahoma and Texas Regional Haze Federal Implementation
Plans (Nov. 2014) ("FIP TSD").
10 Once example of EPA's approach can be found in its TX TSD, in which EPA Stated: "We believe that in
performing its control analysis, the TCEQ should have given greater consideration to the flexibility in the CAIR
trading program and the resulting uncertainty in the projected emissions. In other words, the TCEQ could have
recognized that implementation of reasonable controls under the Regional Haze Rule would likely not be in addition
to anticipated reductions due to CAIR predicted by IPM, but would replace or complement any controls predicted by
IPM." TX TSD at 22. (emphasis added).
11 See North Dakota v. EPA, 730 F.3d 750, 768 (8th Cir. 2013).
Response: EPA guidance, reasonable progress, and State flexibility have already been addressed
above.
Comment: EPA may not issue this FIP prior to providing Texas the opportunity to submit
a SIP responsive to EPA's determination that Texas' 2009 SIP submission was inadequate.
[GCLC (0063) p. 19]
GCLC noted, in contravention of the language and intent of the CAA, EPA is attempting to
disapprove the SIP and immediately move to a FIP. The CAA provides opportunities to States to
correct deficiencies in SIPs, providing the Administrator up to two years to promulgate a FIP in
response to a finding that a SIP was inadequate. 75 This is particularly relevant in this FIP, as
EPA has taken numerous novel steps, including new and unprecedented interpretations of
existing regulations, in order to disapprove this SIP. While EPA's proposed disapproval of
Texas' SIP and its proposed FIP lack merit, Texas still must be given the opportunity to respond-
45
-------
and if appropriate- revise elements of its SIP prior to EPA issuing this FIP. 76 To do otherwise
violates the CAA and unfairly burdens Texas EGUs.
Further, GCLC noted that EPA in its recent rulemaking regarding Arkansas' regional haze SIP
submittal "elect[ed] to not promulgate a FIP" at the time it issued its partial SIP
approval/disapproval "in order to provide Arkansas time to correct [the] deficiencies" indicated
by EPA. This is another example of the unfair and unequal treatment of Texas.
Footnotes:
75 42 USC 7410(c).
76 EPA provided such opportunity to Arkansas regarding its regional haze SIP submittal, where EPA "elect[ed] to
not promulgate a FIP" at the time it issued its partial SIP approval/disapproval "in order to provide Arkansas time to
correct [the] deficiencies" indicated by EPA In fact, it took over three years for EPA to proceed with a FIP. See
Approval and Promulgation of Implementation Plans; Arkansas; Regional Haze State Implementation Plan;
Interstate Transport State Implementation Plan To Address Pollution Affecting Visibility and Regional Haze, Final
Rule 77 Fed. Reg. 14604, 14672 (Mar. 12, 2012); see also Promulgation of Air Quality Implementation Plans;
State of Arkansas; Regional Haze and Interstate Visibility Transport Federal Implementation Plan, 80 Fed. Reg.
18944 (Apr. 8,2015).
Response: We disagree with this comment. As explained in our TSD and elsewhere in this
document and final action, we cannot approve the portions of a State's Regional Haze submittal
that do not meet the CAA and EPA RHR. Therefore, we are obligated to promulgate a FIP to
address these requirements, and we are doing so in today's action.
The requirement for FIP promulgation was triggered because of our 2005 finding that Texas did
not make a timely SIP submission, and the expiration of the 24-month "FIP clock" under CAA
Section 110(c). We may disapprove the SIP and promulgate a FIP in the same action. In
Oklahoma v. EPA, we finalized a rule that partially approved and partially disapproved
Oklahoma's SIP, and in the same action, we promulgated a FIP.29 The Court held that "[o]nce
the EPA issued the finding that Oklahoma failed to submit the required SIP under the Regional
Haze Rule, EPA had an obligation to promulgate a FIP."30 Additionally, the court agreed with
us that a rule requiring us to delay its promulgation of a FIP until it rules on a proposed SIP
"would essentially nullify any time limits the EPA placed on States. States could forestall the
promulgation of a FIP by submitting one inadequate SIP after another."31
This action is distinguishable from the Arkansas regional haze SIP in that the "FIP clock" has
already expired for the Texas Regional Haze plan. We are required to promulgate a FIP for any
disapproved portion of the SIP. Our action fulfills this duty.
Comment: The proposal does not appropriately recognize State primacy under the CAA.
[EEI (0076) p. 7-8]
EEI stated that EPA's proposal raises serious State primacy concerns with its proposed partial
disapproval of Texas' and Oklahoma's SIPs and the imposition of its own FIPs. As a general
29 Oklahoma v. EPA, 723 F.3d 1201.
30 Id. at 62.
31 Id.
46
-------
matter, EPA must respect the role of the States in the regional haze process as authorized by the
CAA, and properly submitted State SIPs have primacy over EPA's FIPs. In this instance, EPA's
proposal does not reflect the flexibility and discretion that States are granted as part of the
regional haze SIP process; instead, the proposed FIP is inflexible by comparison, with
unreasonably tight timelines and strict emissions limitations. As EPA recognizes in its regional
haze guidance to the States, the regulations "give[] States wide latitude to determine additional
control requirements" and, in applying the four statutory factors, States "have flexibility in how
to take into consideration these statutory factors and any other factors that you have determined
to be relevant."4 Indeed, EPA itself has explained that, "[a]s long as this evaluation is done
adequately and the States provide a reasoned basis for their decisions, EPA will defer to the
State" in its reasonable progress determinations. See 77 Fed. Reg. at 40,150, 40,156. EPA's
proposed FIPs depart from the Agency's still-effective guidance.
EEI noted that Texas' and Oklahoma's SIPs were well considered and EPA should approve such
State-led regulatory processes. In support of its proposed FIP, EPA's main rationale is that Texas
and Oklahoma did not properly confer regarding RPGs. This forms the basis for the Agency's
disapproval of Texas' and Oklahoma's SIPs and the imposition of its own FIP. Id. at 74,823.
However, Texas and Oklahoma's conferral regarding the RPGs was fully consistent with the
CAA and the regional haze regulations, and, during that process, the States agreed on which
regulatory programs were needed for reasonable progress. 5 EPA must respect this decision and
recognize that the Agency's imposition of a FIP is unreasonable and unauthorized under the
CAA given these State actions.
Footnotes:
4 EPA, Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program, June 1, 2007 at 4-2, 5-1.
5 See March 25, 2008 letter from Susana M. Hildebrand, Air Quality Division, Texas Commission on
Environmental Quality to Eddie Terrill, Air Quality Division, Oklahoma Department of Environmental Quality; and
May 12, 2008 letter from Eddie Terrill, Air Quality Division, Oklahoma Department of Environmental Quality to
Susana M. Hildebrand, Air Quality Division, Texas Commission on Environmental Quality.
Response: State latitude to determine control requirements, State flexibility regarding
consideration statutory factors, and our remedies available to address an insufficient SIP have
already been addressed above.
Comment: [Nucor Steel (0058) p. 2] Nucor Steel Stated that EPA's interpretation is contrary to
the Clean Air Act's clear direction that each State is to determine its own emission limits,
schedules of compliance and other measures for sources in that State for purposes of visibility
protection under §169A. EPA's interpretation would effectively give one State the power to
control another State's regional haze SIP decisions, including its BART determinations.
Moreover, in this case, EPA improperly overrides the SIP decisions by both States involved,
Texas and Oklahoma.
Response: We explain in our responses elsewhere the roles and obligations of upwind and
downwind States in conducting and establishing their four-factor analyses, reasonable progress
47
-------
goals, and long-term strategies, as well as consultation. We also explain this interrelationship in
detail in our proposal and today's final action.32
Comment: EPA's Proposal disregards the primacy and considerable flexibility and
discretion that Texas is supposed to have under the CAA, and the EPA Regional Haze rules
and guidance, in determining what constitutes reasonable progress [AECT (0074) p. 2]
AECT Stated that EPA's Regional Haze rules and guidance clearly provide that each State has
the primary role in determining what constitutes reasonable progress and that EPA is supposed to
provide - each State considerable flexibility and discretion in making that determination. EPA's
Regional Haze rules "call for States to play the lead role in designing and implementing regional
haze" SIPs.33 Further, EPA previously stated that under its Regional Haze rules, each State has
considerable flexibility and discretion in determining what constitutes reasonable progress.34
Moreover, EPA previously stated that it will defer to a State's determination as to what
constitutes reasonable progress if such determination has a reasoned basis.35
AECT noted, notwithstanding the foregoing, in developing its Proposal, EPA usurped Texas'
primary role in determining what constitutes reasonable progress, did not allow Texas any
flexibility or discretion in making that determination, and provided Texas with no deference
regarding that determination. EPA does not dispute that in determining what constitutes
reasonable progress, Texas applied the four reasonable progress factors specified in the CAA and
EPA's Regional Haze rules or that Texas had consultations with Oklahoma in accordance with
EPA's Regional Haze rules and guidance. EPA merely determined that it would have conducted
the reasonable progress analysis differently and made a different reasonable progress
determination than Texas made, and based on those determinations, EPA issued its Proposal.
Doing so is contrary to EPA's Regional Haze rules and guidance since they clearly provide that
EPA is supposed to provide Texas with considerable flexibility, discretion, and deference in
determining what constitutes reasonable progress, and EPA is not supposed to second guess
Texas' reasonable progress determination and replace Texas' determination with its own
reasonable progress determination.
In light of the foregoing, AECT requested that EPA defer to Texas' reasonable progress
determination that is specified in its Regional Haze SIP.
Response: State flexibility, discretion, and deference in determining what constitutes reasonable
progress has already been addressed above and is discussed in detail elsewhere in the responses
to comments on those issues.
32 See our discussion regarding this, in Section IV, beginning on 79 FR 74823.
33 American Corn Growers Ass'n. v. EPA 291F.3d 1, 2 (D.C. Cir. 2002).
34 EPA's Response to Petition for Reconsideration of Regional Haze Rule, at 11-12 (Jan. 10, 2001) (The Regional
Haze rules are "based on the principle that States should have considerable flexibility in adopting visibility
improvement goals and in choosing the associated emission reduction strategies for making 'reasonable progress'
toward the national visibility goal", and those rules "provide each State with considerable discretion in establishing
reasonable progress goals for improving visibility in the Class I areas.").
35 77 Fed. Reg. 40150, 40156 (July 6, 2012) ("... as long as a State's reasonable progress determination "is done
adequately and the State provides a reasoned basis for [it], EPA will defer to the State.")
48
-------
3. Our Clarified Interpretation of the Reasonable Progress and Long-Term Strategy
Requirements
Comment: EPA's Interpretation is the Best Reading of 40 C.F.R. Section 51.308(d)(3).
[Earthjustice (0067) p.32]
Earthjustice et al., stated that if the "progress goal" in Section 51.308(d)(3)(ii) is not an
"approved or approvable" one, it is, almost by definition, a progress goal that does not comply
with the requirements of the Clean Air Act. In essence, the agency's interpretation does nothing
more than restate what the rule already requires—namely, that the respective states' reasonable
progress goals comply with all applicable requirements of the Act and its implementing
regulations.39
The contrary interpretation of the regulations40 would be that a state may lawfully develop its
long-term strategy to meet a reasonable progress goal proposed by another state even if the
progress goal does not comply with the Clean Air Act and cannot be approved by EPA. This
interpretation of the existing regulations cannot be reconciled with the overall regulatory and
statutory scheme in which SIPs must be reviewed and approved by EPA before becoming valid
as federal law, and EPA can approve a SIP only if it complies with all Clean Air Act
requirements. See 42 U.S.C. § 7410(c), (k), (1). If EPA must disapprove a progress goal that
does not meet the requirements of the Act, then EPA must also disapprove a long-term strategy
that is designed to meet an unapprovable progress goal. Where, as here, one state contributes
significantly to visibility impairment in another state's Class I area, and both states fail to
properly evaluate and include in their respective SIPs "all measures necessary" to achieve a
reasonable (i.e., approvable and lawful) progress goal at a downwind state Class I area, neither
SIP can be approved.41
According to Earthjustice et al., that EPA applied the requirements of the Regional Haze Rule
and the Clean Air Act in the context of a SIP disapproval does not undermine EPA's
interpretation, or require the agency to give the states an opportunity to revise their own
implementation plans before disapproving them and issuing federal plans to replace them. See
EPA v. EMEHomer City Generation, L.P., 134 S. Ct. 1584, 1601 (2014) ("A SIP's failure to
satisfy" the Act's requirements "without more, triggers EPA's obligation to issue a federal plan
within two years," and "EPA is not obliged to wait two years or postpone its action even a single
day" before issuing a federal plan). Indeed, Congress gave EPA, not the states, authority to
determine what the Clean Air Act requires, and whether a SIP fully complies with the Act and its
implementing regulations. 42 U.S.C. § 7410(c), (1); Oklahoma, 723 F.3d at 1204, 1207-10.
That is exactly what EPA proposes to do.
Footnotes:
39 Cf. 42 U.S.C. § 7410(1) (EPA may not approve any plan that "would interfere with any applicable requirement"
of the Act).
40 Even if EPA's interpretation of its existing regulations could be characterized as issuance of a new rule, which it is
not, EPA's action would be proper. EPA may break "new ground" in notice-and-comment rulemaking, so long as
the agency "display awareness" and provides notice of the new position. See FCC v. Fox Television Stations, Inc.,
49
-------
556 U.S. 502, 515-517 (2009). EPA has provided notice and an opportunity to comment on its interpretation of the
existing regulations. Furthermore, EPA's proposal thoughtfully and carefully explains how the Regional Haze
Rule's consultation and transport provisions fit together to achieve the core purpose of the rule and the Clean Air
Act's visibility provisions.
41 40 C.F.R. § 51.308(d)(3)(ii).
Response: We generally agree with this comment, but take no position as to specific statements.
Comment: EPA's interpretation of 40 C.F.R. § 51.308(d)(3)(ii) is plainly erroneous and
cannot support EPA's proposal [Luminant (0061) p. 87]
Luminant stated that EPA's regional haze regulations provide that a state's "long-term strategy
must include enforceable emissions limitations, compliance schedules, and other measures as
necessary to achieve the reasonable progress goals established by States having mandatory Class
I Federal areas."588 EPA does not dispute that Texas' long-term strategy includes enforceable
emissions limitations, compliance schedules, and other measures as necessary to achieve the
reasonable progress goal established by Oklahoma for Wichita Mountains through the CENRAP
process. Nor does EPA assert that Oklahoma requested, and Texas refused, to include any such
measures.589 Instead, EPA concludes, as the basis for its proposed disapproval and FIP, that
Texas' long-term strategy was flawed and violates the regulations because it was not "sufficient
to obtain [Texas'] share of reductions needed to meet an approved, or approvable, progress goal"
for Wichita Mountains—i.e., the one that EPA is proposing in its FIP for Oklahoma.590 This is
not a requirement of the statute or the regulations and, thus, cannot support EPA's disapproval.
Luminant noted, contrary to EPA's assertion, the phrase "progress goal" in 40 C.F.R. §
51.308(d)(3)(ii) plainly refers to "the reasonable progress goals established by States having
mandatory Class I Federal areas," not to goals that have been approved by EPA or may be
approved in the future (and certainly not to a reasonable progress goal proposed by EPA
itself).591 Thus, in developing its long-term strategy "for each mandatory Class I Federal area
located outside the State which may be affected by emissions from the State," as Texas has done
here for Wichita Mountains, a state is only required to include "enforceable emissions
imitations, compliance schedules, and other measures as necessary to achieve the reasonable
progress goals established by States having mandatory Class I Federal areas."592 EPA's
interpretation of the phrase "progress goal" in a subsequent sentence in § 51.308(d)(3)(ii) ignores
the plain language of the preceding provisions that define the "progress goal" that a state must
consider in developing its long-term strategy—i.e., the one "established by" Oklahoma. Thus,
EPA's proposal to disapprove Texas' long-term strategy because it relies on the reasonable
progress goal established by Oklahoma for the Wichita Mountains593—and not some other
progress goal that EPA now claims is reasonable or that Texas developed on its own—is
unlawful and contrary to the plain meaning of the regulations. There is no dispute that Texas
consulted with Oklahoma, participated with Oklahoma in an approved RPO, and developed a
long-term strategy to meet the reasonable progress goal for Wichita Mountains that was
established by Oklahoma. That is all the regulations require for Texas' submission, and EPA
must approve it.
50
-------
Nor does EPA's novel interpretation of § 51.308(d)(3)(ii) and its application of that
interpretation here make logical sense. EPA does not, in this proposal, claim that Oklahoma
substantively erred in setting the RPG for Wichita Mountains. Instead, EPA is simply claiming
to "reset Oklahoma's RPGs based on our [Texas small group] analysis."594 In other words, EPA
is proposing to disapprove Oklahoma's RPG based on the long-term strategy that EPA claims
Texas should have adopted. EPA's logic is thus circular and assumes what it seeks to prove:
EPA claims Oklahoma's RPG for Wichita Mountains is not approvable because it doesn't
consider EPA's proposed FIP for Texas' long-term strategy, but EPA justifies that FIP for
Texas' long-term strategy on its conclusion that the Oklahoma RPG is not approvable. EPA
meets itself coming and going. The truth is that EPA would interpret the regulations to give it
plenary authority over any aspect of a state's regional haze SIP at any time it suits EPA. But that
is not how the regulations are written, nor is it how they have been implemented up until today.
Footnotes:
588 40 C.F.R. § 51.308(d)(3) (emphasis added).
589 79 Fed. Reg. at 74,856.
590 TX SIP TSD at 49 (emphasis added).
591 40 C.F.R. § 51.308(d)(3).
592 Id. (emphasis added).
593 79 Fed. Reg. at 74,857.
594 Id. at 74,889.
Response: We disagree with this comment. Under the commenter's interpretation of 40 CFR
51.308(d)(3)(ii), the obligation of an upwind State is merely to include in its long-term strategy
the emission limitations necessary to achieve the reasonable progress goal set by a downwind
State, irrespective of whether we have approved or could approve that goal. If we disapproved
the downwind State's reasonable progress goal because it did not provide for reasonable
progress, but were powerless to disapprove the upwind State's four-factor analysis and long-term
strategy on which the downwind State unreasonably relied, then the downwind State would be
left without recourse. The commenter seeks support for its interpretation by pointing to language
that says reasonable progress goals are "established by States," not by the EPA. This language,
like similar language throughout the CAA and our implementing regulations, merely reflects that
States are tasked with developing SIPs in the first instance. It is black-letter law that where a
State fails to meet applicable CAA requirements, we are required to step into the shoes of the
State by promulgating a FIP.
The commenter also alleges that we have used circular logic, but this is not the case. The fact is
that four-factor analyses, reasonable progress goals, and long-term strategies are inextricably
linked. Congress created this paradigm when it required us to promulgate regulations that
required both upwind and downwind States to include emission limitations in their SIPs to
achieve reasonable progress, defined reasonable progress as a four-factor analysis, and required
US to develop criteria to measure reasonable progress. See CAA Sections 169A(b)(2) and
169B(e). The scheme that we developed in the Regional Haze Rule (RHR) operates in three
steps: (1) upwind and downwind States conduct four-factor analyses to determine what control
measures are reasonable; (2) States with Class I areas calculate reasonable progress goals (i.e.,
criteria for measuring reasonable progress), measured in deciviews, that reflect the level of
visibility improvement that will result once the controls measures are implemented; and (3)
upwind and downwind States include enforceable emission limitations in their long-term
51
-------
strategies to ensure that the reasonable progress goals are achieved. In this instance, Texas failed
to conduct a reasonable four-factor analysis at Step 1. As a result, Oklahoma set a reasonable
progress goal that did not reflect a reasoned consideration of available controls in Texas at Step
2, and Texas developed a long-term strategy that did not include sufficient emission limitations
to achieve reasonable progress at Wichita Mountains at Step 3. We acknowledge that the States
were developing their SIPs simultaneously, which complicated their ability to proceed in an
orderly step-wised fashion, but this does not excuse the States from having to satisfy the
applicable requirements of the CAA and the Regional Haze Rule.
Comment: EPA's interpretation of 40 C.F.R. § 51.308(d)(3)(iii) is plainly erroneous and
cannot support EPA's proposal [Luminant (0061) p. 88]
Luminant stated that EPA further proposes "to find that the technical basis on which Texas relied
to determine its apportionment of emission reduction obligations necessary for achieving
reasonable progress in the Wichita Mountains was inadequate" and thus Texas' long-term
strategy does not meet the requirements of 40 C.F.R. § 51.308(d)(3)(iii).595 That provision
provides that "the State must document the technical basis, including modeling, monitoring and
emissions information, on which the State is relying to determine its apportionment of emission
reduction obligations necessary for achieving reasonable progress in each mandatory Class I
Federal area it affects."596 It further provides: "The State may meet this requirement by relying
on technical analyses developed by the regional planning organization and approved by all State
participants."597
Luminant noted, EPA concedes that Texas' long-term strategy "rel[ied] on technical analyses
developed by CENRAP and approved by all state participants" and further that Texas
"performed an additional analysis building upon the work of the regional planning organization
in order to evaluate additional controls."598 Nevertheless, EPA contends that Texas should have
done more and independently "considered] the four-factors used in determining reasonable
progress [in § 51.308(d)(l)(i)(A)] in developing the technical basis for . . . downwind [i.e., out-
of-state] Class I areas," including Wichita Mountains.599 In other words, EPA concludes that
Texas' long-term strategy, in addressing Wichita Mountains, should not have considered the
reasonable progress goal for the area as established by Oklahoma using the four statutory factors,
but instead Texas should have itself conducted a four-factor analysis for Wichita Mountains to
determine what was reasonable progress for the area.
Luminant asserted that this interpretation is wrong and backwards. EPA's claim that a state must
conduct an analysis of the four statutory factors for "both their own Class I areas and downwind
Class I areas" is plainly erroneous and contrary to the statute and regulations.600 The statute and
the regulations plainly provide that states must only analyze the four-factors for their own Class I
areas and only in setting the reasonable progress goal for the area. The statute provides that "in
determining reasonable progress there shall be taken into consideration the costs of compliance,
the time necessary for compliance, and the energy and nonair quality environmental impacts of
compliance, and the remaining useful like of any existing source subject to such
requirements."601 And EPA's implementing regulations only reference the four-factors in §
51.308(d)(1), which provides that a state must consider these factors "[i]n establishing a
52
-------
reasonable progress goal for any mandatory Class I Federal area within the State . . . ,"602 In stark
contrast, § 51.308(d)(3), which contains the requirements for a state's long-term strategy that
addresses Class I areas "outside the State," contains seven (not four) entirely different factors
that "[t]he State must consider" "in developing its long-term strategy."603 Indeed, the long-term
strategy regulations do not speak in terms of establishing reasonable progress goals (as EPA's
interpretation would have it), but instead of "apportionment of emission reduction obligations"
needed to meet the goals already established.604
Luminant stated that EPA's interpretation is nonsensical. Texas, in developing its long-term
strategy, is not required to determine what is reasonable progress for Class I areas outside of
Texas but only to ensure that its plan includes the measures necessary to achieve its share of
emission reductions needed to meet the reasonable progress goals established by the states in
which such areas are located.605 Thus, Texas is not required to consider whether additional cost-
effective controls on its sources could achieve more progress than is deemed reasonable at areas
outside of the state, as EPA's proposal would require.606 The regulations—in defining the
"technical basis" for the state's demonstration of "apportionment"—do not include cost
information as among the documentation that the state must provide and thus clearly do not
contemplate a review of the four reasonable progress factors. Indeed, in issuing its final regional
haze regulations, EPA specifically rejected the idea of applying the four statutory factors
(including cost-effectiveness) to the long-term strategy requirements.607 EPA cannot disapprove
Texas' SIP on the basis of an "interpretation" that is plainly erroneous and inconsistent with the
agency's binding regulations.608
Further, Luminant noted that EPA's regulations plainly provide that a state may meet the
requirement in § 51.308(d)(3)(iii) to "document the technical basis" of its long-term strategy by
"relying on technical analyses developed by the regional planning organization and approved by
all State participants."609 This is exactly what Texas did here. Texas participated with Oklahoma
and other states in CENRAP regional haze planning over a multi-year period. CENRAP
developed air quality modeling, including point source apportionment modeling (by ENVIRON),
and data on the availability and Cost-effective ness of additional SO2 controls on Texas sources
(by Alpine Geophysics). EPA finds no error in CENRAP's data development610 and, indeed,
relies on CENRAP's technical analysis in its own proposal. EPA completely ignores this
provision of its regulation in its proposal here. And there is no qualifier in § 51.308(d)(3)(iii)
that would limit a state's ability to rely on an RPO's technical analyses, other than that it be
approved by all State participants,"611 as was the case here. Indeed, such incorporation of and
reliance on the technical work of CENRAP is expressly encouraged and permitted in the
Regional Haze Rule. As the Tenth Circuit has recently held, under EPA's regulations, a state
may "base their determination of reasonable progress on the [regional planning organization's]
assessments," as Texas did here.612 Thus, EPA's attempt to limit Texas' ability to rely on
CENRAP's air quality modeling and Cost-effective ness data is contrary to the plain language of
the regulations and cannot support EPA's proposed disapproval.613
Footnotes:
595 Id. at 74,822.
596 40 C.F.R. § 51.308(d)(3)(iii) (emphasis added).
597 Id.
598 79 Fed. Reg. at 74,857.
53
-------
599 Id. at 74,861; TX SIP TSD at 65.
600 79 Fed. Reg. at 74,829.
601 42 U.S.C. § 7491(g)(1) (emphasis added).
602 40 C.F.R. § 51.308(d)(l)(i) (emphasis added). EPA has explained that § 51.308(d)(1) contains the requirements
"for States to establish 'reasonable progress goals' for each Class I area within the State." 64 Fed. Reg. at 35,730
(emphasis added).
603 40 C.F.R. § 51.308(d)(3)(v)(A)-(G); see also 64 Fed. Reg. at 35,737 ("In their regional haze SIP submissions,
States must describe how each of these seven factors is taken into account in developing long-term strategies.").
604 40 C.F.R. § 51.308(d)(3)(iii) (emphasis added).
605 Id. § 51.308(d)(3).
606 Fortius same reason, EPA's interpretation of 40 C.F.R. § 51.308(d)(3)(v)(C) is plainly erroneous as well. EPA
contends, TX SIP TSD at 71, that Texas' long-term strategy did not meet the requirements of 40 C.F.R. §
51.308(d)(3)(v)(C), which provides: "The State must consider, at a minimum, the following factors in developing its
long-term strategy: . . . (C) Emission limitations and schedules for compliance to achieve the reasonable progress
goal. . . ." EPA's basis for this contention is that "Texas did not adequately consider the emissions limitations and
schedules for compliance needed to achieve reasonable progress in Big Bend, Guadalupe Mountains, or Wichita
Mountains." 79 Fed. Reg. at 74,822; TX SIP TSD at 71. However, the plain language of § 51.308(d)(3)(v)(C) only
requires the state to consider "the reasonable progress goal" already established by the state, not to independently
determine what is reasonable progress as part of its long-term strategy. EPA's disapproval of Texas' long-term
strategy is unlawful for this reason as well.
607 64 Fed. Reg. at 35,736-37 ("We have decided not to include the five proposed items that are derived from
section 51.306(g) [RAVI] [as factors to consider for long-term strategies], because four of these items are included
on the list of 'reasonable progress' factors in section 51.308(d)(l)(i)(A) of the final rule . .. .") (emphasis added).
608 For these same reasons, EPA's proposed FIP is unlawful and contrary to the statute and regulations. EPA's FIP
"simultaneously conduces] RP and LTS analyses using the 'four-factor analysis' outlined in 40 C.F.R §
51.308(d)(1)(A)..." FIP TSD at 12. But, as explained above, the reasonable progress "four-factors" are not
appropriately used in developing a long-term strategy. EPA's FIP thus relies on factors that Congress did not intend
the agency to consider and is unlawful. Luminant Generation, 675 F.3d at 925.
609 40 C.F.R. § 51.308(d)(3)(iii).
610 Indeed, EPA concedes that "[t]he CENRAP states' modeling .. . was developed consistent with our guidance."
TX SIP TSD at 55.
611 40 C.F.R. § 51.308(d)(3)(iii).
612 WildEarth Guardians, 770 F.3d at 944.
613 79 Fed. Reg. at 74,857.
Response: We disagree with this comment. The commenter's interpretation of 40 CFR
51.308(d)(3)(iii) is not only in conflict with that provision's plain language, it also contradicts
the statute and the process that Texas itself followed when developing its SIP. 40 CFR
51.308(d)(3)(iii) states that "the State must document the technical basis, including modeling,
monitoring and emissions information, on which the State is relying to determine its
apportionment of emission reduction obligations necessary for achieving reasonable progress in
each mandatory Class I Federal area it affects." (emphasis added). CAA Section 169A(g)(l)
defines "reasonable progress" to be a consideration of the four-factors. Thus, the plain language
of 40 CFR 51.308(d)(3) requires States to conduct a four-factor analysis and document the
technical basis for that analysis for both upwind and downwind Class I areas. Even if the
regulation were unclear, the statute is not. CAA Section 169A(b)(2) plainly requires both
upwind and downwind States to include emission limits in their SIPs that are necessary to
achieve reasonable progress. Under the commenter's interpretation, a State without a Class I
area would have no obligation to conduct a four-factor analysis for its sources at all. This is not
what Congress intended. The commenter attempts to bolster its position by citing to the
preamble of the Regional Haze Rule, where we explained that we chose not to include the four-
factors in 40 CFR 51.308(d)(3)(v). However, we explained that our reason for doing this was to
54
-------
avoid redundancy, not to absolve upwind States of an obligation to consider the four-factors. In
any event, commentary on 40 CFR 51.308(d)(3)(v) says nothing about the plain language of 40
CFR 51.308(d)(3)(iii).
While the commenter correctly points out that a State can rely on technical analyses performed
by RPOs to satisfy the requirement in 40 CFR 51.308(d)(3)(iii), the commenter fails to account
for situations where these analyses are limited in scope because the participating states wished to
perform for themselves certain of the required analyses, particularly the LTS analysis for
whether there additional reasonable controls were available. The Tenth Circuit decision cited by
the commenter merely stands for the proposition that States can pool their resources with
neighbor States to conduct a multi-state analysis in lieu of conducting independent analyses, not
that every multi-state analysis will be per se sufficient for every State. Tellingly, the
commenter's cramped theory does not even square with reality. Texas itself recognized the need
for additional information and supplemented the CENRAP analysis by conducting a four-factor
analysis that included both upwind and downwind Class I areas.36 Texas stated several times in
its response to comment document for its SIP that this four-factor analysis was required by the
Regional Haze Rule.37
At bottom, the commenter's views are premised on a fundamental misunderstanding of the
regional haze planning process. The commenter seems to suggest that downwind States first set
reasonable progress goals and that upwind States then develop a long-term strategy to ensure that
they achieve their apportionment of emission reductions. On the contrary, the first step in the
process is the four-factor analysis, the results of which determine the suite of reasonable control
measures each State will adopt. Reasonable progress goals are criteria for measuring the amount
of visibility improvement that is projected to result from the installation of these controls, not a
random number that States generate in a knowledge vacuum that then becomes the lodestar for
later control determinations.
Comment: EPA's new interpretations, taken together, create a distortion of the regional
haze program [Luminant (0061) p. 91]
Luminant stated, taken together, EPA's new interpretations of these provisions [40 CFR
51.308(d)(3)(ii) and (iii)] reflect a fundamental overreach of authority by EPA and, if finalized,
would alter the entire nature and intent of the reasonable progress and long-term strategy
requirements. EPA's regulations and EPA guidance encourage states to develop their reasonable
progress goals and long-term strategies in close coordination with each other and through
participation in RPOs, as Texas and Oklahoma did.614 Yet, through its new and unfounded
regulatory interpretations, EPA would discourage coordination, encourage conflict among the
36 See, e.g., the Texas Regional Haze SIP at 10-5 ("The TCEQ used the CENRAP modeling to estimate the impact
that the control strategy would have on the Class I areas impacted by Texas' emissions."). Following this statement,
Texas presents Table 10-6, which summarizes its calculations of the improvements from its control suite at its Class
I areas, plus Breton Isle in Louisiana, Caney Creek in Arkansas, and Carlsbad Caverns in New Mexico.
37 See, e.g., Appendix 2-2 to the Texas Regional Haze SIP at 24 ("Further, a four factor analysis is necessary for the
set of sources in the respective areas of influence that impact each of the Class I areas that Texas' emissions
impact.") (emphases added) ("The TCEQ has used the four factor analysis, as required, for the set of Texas sources
impacting Class I areas, to determine whether all reasonable reductions have been required.") (emphasis added).
55
-------
states, and disregard years of regional planning and consultation—all performed at EPA's
urging.
Luminant asserted that EPA's interpretation of these provisions—which would require Texas to
divine EPA's views on Oklahoma's reasonable progress goals before those goals are even
submitted to EPA—is also inconsistent with the structure and intent of EPA's regional haze
regulations and makes no sense as a practical matter. The regulations establish a coordinated and
step-wise process in which states set their reasonable progress goals based on modeling
"performed early in the [regional planning] process," and then other contributing states develop
their long-term strategies to meet those goals.615 And, even where it is later determined that the
contributing state's long-term strategy is not, in fact, adequate to meet the established goal, that
is "not grounds for disapproving either [states'] SIP" and issuing a FIP, as EPA itself found in
approving Nebraska's SIP.616 Rather, the proper course is for the states "to consider whether
other reasonable control measures are appropriate to ensure reasonable progress during
subsequent periodic progress reports and regional haze SIP revisions. "617 Thus, EPA's
attempt here to "simultaneously conduct[] reasonable progress and long-term strategy analyses"
is fundamentally at odds with the regulations and EPA's prior application of those regulations.618
EPA's conflating of these two distinct analytical steps is contrary to the plain language of the
current regulations and results in an analysis and outcome that are nowhere contemplated by the
regulations and exceed EPA's authority.
According to Luminant, EPA has all but conceded that its regional haze regulations would
require formal amendment in order to accommodate its new approach. EPA submits these
changes here as "interpretations" of its existing regulations, but in fact EPA is simultaneously
working to formally amend its regulations to authorize EPA's new approach.619 Unless and until
EPA formally amends its current regulations—the plain text of which do not permit EPA's
interpretations—EPA is without authority to enforce its unprecedented interpretations against
Texas and affected Texas sources here for the first time.
Footnotes:
614 64 Fed. Reg. at 35,735 ("The EPA expects that much of the consultation, apportionment demonstrations, and
technical documentation will be facilitated and developed by regional planning organizations. We expect, and
encourage, these efforts to develop a common technical basis and apportionment for long-term strategies that could
be approved by individual State participants, and translated into regional haze SIPs for submission to EPA.").
615 77 Fed. Reg. at 40,155.
616 Id.
617 Id. (emphasis added).
ei8 pip TSD at 5.
619 See EPA, Pre-Meeting Materials for the EPA-FLM-RPO-States-Tribes Meeting on the Future of the Regional
Haze Program (Feb. 3,2015).
Response: We disagree with this comment. Contrary to the commenter's assertion, we are not
providing a "new" interpretation of our regulations. As we discuss in our proposal, we are
merely clarifying how several complex and interconnected regulatory provisions operate to carry
out the Congressional directives in CAA Sections 110(a)(2)(D)(i)(II) and 169A(b)(2). We are
not suggesting that States should not coordinate their regional haze SIPs or not participate in
RPOs. We are reiterating that States have an obligation to include reasonable controls measures
56
-------
in their SIPs so that sufficient progress toward the national goal is achieved at every Class I area
and that upwind States cannot shirk this responsibility for downwind Class I areas.
As we explained in our prior response, the commenter has mischaracterized the regional haze
planning process. States do not start by conducting regional modeling and setting their
reasonable progress goals. To conduct modeling, States first must identify what controls
measures their sources will install. To identify these control measures, the CAA and our
implementing regulations require States to conduct four-factor analyses. We understand that
many States made rudimentary judgments about what control measures they would require their
sources to install before conducting regional modeling, but this reality does not excuse those
States from their statutory obligation to include in their SIPs "such emission limits, schedules of
compliance and other measures as may be necessary to make reasonable progress toward
meeting the national goal." CAA Section 169A(b)(2). If we determine that a State's four-factor
analysis failed to properly take into consideration one of the statutory factors or was somehow
unreasonable, then the CAA requires that we disapprove the SIP. As we explain in response to a
later comment in this section of the document, we have taken a voluntary remand of our action
on the Nebraska long-term strategy, so the quote by the commenter is irrelevant.
Finally, the commenter is incorrect that we are amending our regulations "to authorize this
approach." We are currently soliciting stakeholder feedback on a number of issues related to the
second planning period for regional haze. These issues include whether we should revise
portions of the Regional Haze Rule related to SIP submission deadlines, the form of five-year
progress reports, the RAVI regulations at 40 CFR 51.300-307, and the calculation of the uniform
rate of progress. We have also solicited feedback on new guidance to aid States in conducting
the modeling, control assessments, and other technical work that will be needed to make
reasonable progress going in future planning periods.
Comment: EPA's new interpretations of regulations found in 40 CFR § 51.308(d)(3)
conflict with the plain language of the regulations and are no basis for disapproving Texas'
SIP. [GCLC (0063) p. 13-14]
GCLC stated that EPA has proposed new interpretations of the LTS-related regulations, all of
which EPA relies on to disapprove certain elements of Texas' SIP. EPA's interpretations, as
discussed below, conflict with the plain language of the LTS regulations. EPA is also effectively
pulling a bait-and-switch on Texas, redefining regulations that have consistently been applied, in
a way that arbitrary and capriciously prejudices Texas. These new and unfounded interpretations
are an unlawful basis for EPA's proposal and do not warrant any deference, as claimed by EPA.
Reinterpretation of "Progress Goal." According to GCLC, EPA seeks to reinterpret the
requirements found at 40 CFR § 51.3 08(d)(3)(ii) by requiring that Texas' demonstration must be
based on progress goals that are "approved or approvable" by EPA. 59 This conflicts with the
plain language of the regulations that require the upwind state (Texas) to base its long-term
strategy on the "progress goal" "established by" the downwind state (Oklahoma); at no time
during the states' coordination process is EPA approval of the progress goal necessary or even
possible. Through Texas' coordination with Oklahoma and its agreed upon progress goals,
57
-------
Texas' LTS met the reasonable progress goal established by Oklahoma for the Wichita
Mountains. Regardless of the actions that EPA is attempting to take regarding the Oklahoma
SIP, Texas' submission fully complies with the requirements of the CAA and associated
regulations and must be approved.
Interpretations requiring four-factor analysis for intra and interstate Class I Areas.
According to GCLC, EPA's proposal to require Texas to conduct the four-factor analysis for the
Wichita Mountains, a Class I area outside of the state, completely contradicts the language and
intent of the CAA and implementing regulations. Oklahoma has already conducted that analysis,
and EPA is in effect asking for Texas to develop its own reasonable progress goal for the
Wichita Mountains. This is a duplicative analysis that is not required by law. There is nothing
in 40 CFR § 51.308(d)(3)(ii) that requires a four-factor analysis. The LTS regulations state that:
"[t]he State must document the technical basis, including modeling, monitoring and emissions
information, on which the State is relying to determine its apportionment of emission reduction
obligations necessary for achieving reasonable progress in each mandatory Class I Federal area it
affects. The State may meet this requirement by relying on technical analyses developed by the
regional planning organization and approved by all State participants."60 This is exactly what
Texas did through its participation with the states in its regional planning organization ("RPO"),
and to which EPA has no apparent objection.
GCLC asserted that the regional haze regulations only require a State to conduct a four-factor
analysis for Class I areas within the state.61 The LTS regulations discussing out-of-state impacts
to Class I areas, as discussed above, only require the apportionment of emissions reductions
obligations to meet other states' RPGs that have already been established.62 Further, the LTS
includes a list of factors for consideration in the LTS for Class I Areas outside the state, but this
is a list of seven factors that are completely different, in both language and intent, than the
reasonable progress four-factors. 63
GCLC noted, ultimately, EPA "believe[s] the record supports a finding that [Texas'] analysis is
inadequate as it does not provide the information necessary to determine the reasonableness of
controls at those sources in Texas that significantly impact visibility at the Wichita Mountains."64
But as discussed above, Oklahoma has all information necessary to determine the reasonableness
of controls and EPA's attempt to reinterpret the rules does not impose any additional burdens.
Texas has met its LTS obligations.
Footnotes:
59 79 Fed Reg. at 74,829.
60 40 CFR § 51.308(d)(3)(iii). (emphasis added).
61 See 40 CFR § 51.308(d)(l)(i), which only lists the four-factors in relation to "establishing a reasonable progress
goal for any mandatory Class I Federal area within the State." (emphasis added).
62 See 40 CFR § 51.308(d)(3)(iii).
63 40 CFR 51.308(d)(3)(V)(A-G).
64 Id.
Response: We disagree with this comment. See our responses to the comments from Luminant
above.
58
-------
Comment: [Associations (0059) p. 11-12] The Associations stated that, after proposing to
disapprove Oklahoma's reasonable progress goals, EPA goes on to propose disapproval of
Texas' long-term strategy on the basis that it is not consistent with the modified reasonable
progress goals that EPA would impose through a FIP. This is both inconsistent with the Clean
Air Act and EPA's implementing regulations and patently unreasonable. Texas' long-term
strategy fully complies with the Clean Air Act and EPA's regulations because it meets Texas'
obligations with respect to Oklahoma and all other States whose visibility may be impacted by
emissions from Texas sources. Under EPA regulations, to satisfy the long-term strategy
requirement, a State that "has participated in a regional planning process ... must ensure that it
has included all measures needed to achieve its apportionment of emissions reduction obligations
agreed upon through that process." 40 C.F.R. § 51.308(d)(3)(ii) (emphasis added). Texas did so
here. EPA concedes that "Oklahoma did not specifically request any additional reductions from
Texas sources," 79 Fed. Reg. at 74,856, meaning that the "agreement" between Texas and
Oklahoma did not require any new emissions controls to be added to Texas' long-term strategy.
Instead, EPA ignores this agreement between the States and proposes to disapprove Texas' long-
term strategy because it is not consistent with the additional emission reductions proposed by
EPA in its Oklahoma FIP. This conclusion is flatly inconsistent with EPA regulations, which
require consistency with reasonable progress goals "established by states" during the SIP
process, not with alternative reasonable progress goals developed at a later date by EPA. 40
C.F.R. § 51.308(d)(3)(ii).
The Associations contended that the EPA simply ignores the relevant provision of Section
51.308(d)(3)(ii) and instead bases its proposed disapproval on a supposed "interpretation" of a
different portion of that rule, which provides that a State's SIP must "include enforceable
emissions limitations, compliance schedules, and other measures as necessary to achieve the
reasonable progress goals established by States having mandatory Class I Federal areas." 40
C.F.R. § 51.308(d)(3). Engrafting a requirement nowhere found in the text of the regulation,
EPA now proposes to "interpret" the phrase "progress goal" to instead mean a reasonable
progress goal that is "approved or approvable" by EPA. 79 Fed. Reg. at 74,829. But no amount
of supposed "Auer deference" permits EPA to "interpret" a regulation to impose a requirement
that is contrary to the regulation itself. EPA simply ignores the fact that the "progress goals"
phrase it cites refers to the sentence that precedes it: "the reasonable progress goals established
by States having mandatory Class I Federal areas." 40 C.F.R. § 51.308(d)(3).
The Associations argued that the EPA's proposed disapproval is unreasonable. Texas'
obligation to develop a long-term strategy must be based on the agreements reached among the
States at the time their SIPs are submitted. Texas—and other States—cannot be expected to
divine whether EPA will disagree with another State's reasonable progress goals years in the
future and then anticipate and preemptively incorporate into its long-term strategy the revised
reasonable progress goals EPA may decide to include in a subsequent FIP. Thus, even if EPA
were to disapprove a State's reasonable progress goals, it is not reasonable to demand that
neighboring States adjust their long-term strategies until the next review period.3
Footnotes:
3 EPA has previously agreed with this position, explaining that when a State's final action with respect to reasonable
progress goals "deviate[s] from what was included in the [regional] modeling," the remedy is for affected States to
59
-------
"consider asking [the contributing state] for additional emission reductions" "during subsequent period progress
reports and regional haze SIP revisions." 77 Fed. Reg. at 41,155-56.
Response: We disagree with this comment. See our responses to the comments from Luminant
above.
Comment: [TCEQ/PUCT (0056) p. 5-7] The TCEQ stated that the EPA's interpretation of the
RHR is unprecedented, incorrect, and unreasonable. The TCEQ argued that the EPA exceeded
its authority in disapproving Texas' long-term strategy.
The TCEQ asserted that the EPA has misinterpreted the requirements in §§51.308(d)(1) and
(d)(3) and improperly gives meaning to a phrase in order to fill a perceived gap in their own
regulations. The RHR requires upwind states to consult with downwind states and develop
coordinated strategies to address the upwind state's share of impairment in the downwind state's
Class I areas that are impacted. Texas met these long-term strategy requirements. As the EPA
admits on 79 FR 74856, in its evaluation of the consultation with Oklahoma, both states agreed
with the 2009 Texas plan. Therefore Texas met its obligation under the RHR for the long-term
strategy assessment for Class I areas outside the state, specifically Wichita Mountains. The EPA
may be correct that its own rules do not address situations where a downwind state's RPG for an
area is not properly set, but that does not give the EPA the authority to arbitrarily revise its rules
ad hoc, without the proper notice and comment procedures; nor does the flaw in the EPA's rules
mean that the Texas plan addressing the long-term strategy is deficient.
The TCEQ stated that the EPA exceeded its authority in disapproving Texas' long-term strategy.
First, the EPA bases its proposed disapproval of the RPG and long-term strategy on a new
interpretation of §51.308(d)(3)(ii) that the 'progress goal' established by a downwind state, i.e.
Oklahoma, must be "approved or approvable." This new definition in 2014 of the term progress
goal in order to justify the proposed disapproval of the 2009 RH SIP is arbitrary and capricious.
The EPA is proposing to disapprove Texas' portion of the RPG calculation for Wichita
Mountains, not because of a flaw in Texas' analysis, but because the EPA does not agree with
Oklahoma's RPG. The EPA maintains that in this case, it must disapprove both Texas and
Oklahoma's plans regarding Wichita Mountains. This interpretation is not found in the rule or
statute and is not legally valid for reviewing Texas' long-term strategy or RPG. In fact, the,
§51.308(d)(1) standard for determining the acceptability of the RPG is "it must provide for an
improvement in visibility for the most impaired days over the period of the implementation plan
and ensure no degradation in visibility for the least impaired days over the same period." The
EPA agrees that both Texas' RPGs for Big Bend and Guadalupe Mountains and Oklahoma's
RPG for Wichita Mountains meet this requirement (79 FR 74834).5
The TCEQ explained that, in developing its long-term strategy for impacts to Wichita
Mountains, Texas relied on an agreed upon approach to emission reductions. Oklahoma and
Texas both agreed to the Texas SIP long-term strategy during consultation. Texas' long-term
strategy was based partly upon meeting the RPG for Wichita Mountains established by
Oklahoma. That plan and those consultations are what the EPA must review for compliance
with the CAA. The EPA also relies on an incorrect interpretation of the long-term strategy
requirements in (d)(3). Texas is not required to consider the four statutory factors for Class I
60
-------
areas outside the state. These factors are considered in the determination of 'reasonable progress'
in CAA §169A(g)(l) for Class I areas located in the state. For Class I areas located outside the
state, Texas is required to consult with those 'downwind' states in developing coordinated
emissions management strategies as may be necessary to achieve the RPGs established by the
host state6. In establishing its long-term strategy, the TCEQ properly relied on its consultation
and concurrence with Oklahoma at the time the Texas 2009 RH SIP was developed. That
consultation resulted in concurrence that controls - additional to those already required under
existing regulations - were not reasonable for Texas sources. The EPA is changing the rules
after the fact to give a never before used meaning to 'progress goal' that those goals for
Oklahoma must be approved or approvable in order to approve Texas' long-term strategy. The
EPA cannot rely on the deference from the courts as this interpretation is inconsistent with the
regulation and clearly not found in the RHR.
Footnotes:
5 Once again, the EPA engages in creative interpretation of its rules that is not based iu the CAA. The EPA
maintains that "ODEQ's RPGs for the Wichita Mountains are consistent with minimum requirements of
§51.308(d)(1) " (emphasis added) This section of the rule makes no mention of a minimum level of progress
and in fact provides all of the requirements for what the RPG must provide.
6 For Wichita Mountains, the host state is Oklahoma. See 40 CFR §51.308(d)(3).
Response: We disagree with this comment. See our responses to the comments from Luminant
above. We also note that the commenter's statement that "Texas is not required to consider the
four-factors for Class I areas outside the state," is directly contradicted by the analysis the
commenter actually performed and by statements the commenter made in its own response to
comment document.38 Finally, the commenter mischaracterizes the consultation discussions with
Oklahoma. On multiple occasions, Oklahoma indicated that Texas' sources had outsized
impacts on visibility at Wichita Mountains, that Oklahoma could not meet the glidepath without
emission reductions from Texas, and that Oklahoma did not believe it had the authority to
require those reductions, but instead had to rely on Texas or the EPA. In this final rule, we have
clarified that States should not hesitate to ask their neighbors for additional emission reductions
if the evidence suggests that cost-effective controls are available, and, at a minimum, should
document their disagreements regarding the proper apportionment of emission reductions with
all available evidence so that the we can properly evaluate each State's SIP.
Comment: [TCEQ/PUCT (0056) p. 14-15] The TCEQ disagreed with the EPA's position that
Texas did not adequately address the documentation requirements in 40 CFR 51.308(d)(3)(iii)
regarding the technical basis for Texas' long-term strategy.
The TCEQ noted that the proposal quotes the RHR:
The State must document the technical basis, including modeling, monitoring and
emissions information, on which the State is relying to determine its apportionment of
38 See, e.g., Appendix 2-2 to the Texas Regional Haze SIP at 24 ("Further, a four factor analysis is necessary for the
set of sources in the respective areas of influence that impact each of the Class I areas that Texas' emissions
impact.") (emphasis added) ("The TCEQ has used the four factor analysis, as required, for the set of Texas sources
impacting Class I areas, to determine whether all reasonable reductions have been required.") (emphasis added).
61
-------
emission reduction obligations necessary for achieving reasonable progress in each
mandatory Class I Federal area it affects. The State may meet this requirement by relying
on technical analyses developed by the regional planning organization and approved by
all State participants (79 FR 74861).
Texas documented the modeling, the monitoring, and emissions information data used for the
2009 RH SIP. The modeling was completed by CENRAP and available for all states. The
monitoring data were available from the IMPROVE monitors and the emissions data had been
previously approved by the EPA. The preamble contains a lengthy discussion - over eight
Federal Register pages, plus the Technical Support Document - of Texas' consultation with
Oklahoma, Colorado, Arkansas, and New Mexico, the CENRAP process and modeling and the
TCEQ's supplemental analysis of CENRAP's technical analysis. This discussion belies the
EPA's claim that the TCEQ did not adequately meet the requirements in 40 CFR
51.308(d)(3)(iii) to document the technical basis for the TCEQ's apportionment determination.
The EPA and Oklahoma cannot fairly argue that not all relevant data was available to inform
them of Texas source's visibility impact on neighboring Class I areas and the reasoned analysis
that additional controls would not be necessary to reduce visibility impairment outside Texas.
Response: As we have discussed in our responses to other comments, the mere fact that Texas
addressed a requirement in the Regional Haze Rule does not mean that it did so in a reasonable
fashion and therefore satisfied that requirement. Similarly, Texas cannot rely on CENRAP's
technical evaluation if that evaluation is limited in scope. Also, the TCEQ should not mistake
the length of our documentation of how Texas addressed a particular portion of our regulations
as being indicative of our assessment of the quality of those consultations. Furthermore, as we
also discuss at length in our proposal and in the response to comments herein, we disagree with
the TCEQ's conclusion that Texas provided Oklahoma with enough information with which to
make an informed calculation of its reasonable progress goal for Oklahoma. As we state in our
proposal:39
Because it only estimated the visibility benefit of all the controls together, the
TCEQ was not able to assess the potential benefit of controlling individual sources
with significant, and potentially cost-effective, visibility benefits. Also, we believe
that individual benefits were masked by the inclusion of those controls with little
visibility benefit that only served to increase the total cost figures. Thus, Oklahoma
was not armed with adequate information with which to make an informed decision
concerning the benefits and costs of controlling sources in Texas.
Comment: [Associations (0059) p. 2-4, 21] The Associations stated the EPA's proposal stands
in stark contrast to EPA's prior interpretation of the Clean Air Act and its own regional haze
regulations, both in guidance and in its review of prior SIP submissions from other States. If
finalized, EPA's new interpretation would dramatically expand EPA's authority while unfairly
minimizing the role of the States in determining how to best balance competing interests while
improving visibility at national parks. As the Supreme Court recently explained, an agency must
"provide more substantial justification when its new policy rests upon factual findings that
39 79 FR 74838.
62
-------
contradict those which underlay its prior policy; or when its prior policy has engendered serious
reliance interests that must be taken into account." Perez v. Mortgage Bankers Ass'//, Case No.
13-1041 (S. Ct. Mar. 9, 2015), Slip op. at 13 (internal citation omitted).
The Associations noted that, here, EPA has failed entirely to address its change in interpretation,
let alone provide "substantial justification" for it. Not only would EPA's approach needlessly
impose nearly $2 billion in unnecessary costs on Texas utilities despite Texas' reliance on EPA's
prior policy when preparing its SIP, it would create harmful precedent that could be used by EPA
in the future to ignore States' reasoned judgments and impose significant and excessive costs on
the Associations' members. Therefore, we urge EPA to withdraw the proposal and to recognize
the reasoned judgment of Texas and Oklahoma by fully approving their regional haze SIPs.
[Associations (0059) p. 21] The Associations concluded that the EPA's proposal to disapprove
Texas and Oklahoma's SIPs and impose FIPs to establish reasonable progress goals and long-
term strategies is unlawful, arbitrary, and capricious. The Associations urged EPA to approve
Texas and Oklahoma's SIPs as consistent with the Clean Air Act and EPA's Regional Haze
Rule.
Response: The commenter is incorrect that we are creating a new interpretation or new policy.
As we explained in the proposal, after evaluating the Texas and Oklahoma regional haze SIPs,
we found it necessary to clarify our interpretation of certain statutory and regulatory provisions,
many of which are highly complex and interconnected, in order to provide States with clarity as
to their respective roles and obligations when addressing visibility transport.
Comment: EPA cannot disapprove the plans on the basis of a new interpretation that is
contrary to the plain language of EPA regulations on state consultations. [NRG (0078) p.
7]
NRG stated that the EPA has proposed to interpret its rules at 40 C.F.R. §§ 51.308(d)(3) to
effectively require states such as Texas to guess correctly as to whether EPA will in the future
disagree with another state's long-term strategy. 24 This proposed interpretation is also part of
the basis for EPA's proposed disapproval of Texas' plan.
NRG disagreed with this interpretation, as it is contrary to the plain language of the regulations,
as described in the following table:
Rule
Plain Meaning EPA Proposed Interpretation
63
-------
40C.F.R. 51.308(d)(l)(iv): "(1)
Reasonable progress goals. For
each mandatory Class I Federal
area located within the State, the
State must establish goals
(expressed as deciviews) that
provide for reasonable progress
towards achieving natural
visibility conditions....
(iv) In developing each
reasonable progress goal, the
State must consult with those
States which may reasonably be
anticipated to cause or
contribute to visibility
impairment in the mandatory
Class I Federal area."
The state containing the Class I
area, which is responsible for
setting the reasonable progress
goal, must consult with upwind
states that might affect the Class
I area.
The regulation that calls for
upwind states to initiate
consultation in furtherance of
"coordinated emission
management strategies" (40
C.F.R. § 51.308(d)(3)(i)) only
applies once the reasonable
progress goal has been set.
"[A]s a corollary to Section
51.308(d)(l)(iv), upwind states
must consult with [downwind]25
States(s) in order to develop
coordinated management
strategies."
40C.F.R. § 51.308(d)(3)(ii)
(emphasis added): "Where other
States cause or contribute to
impairment in a mandatory
Class I Federal area, the State
must demonstrate that it has
included in its implementation
plan all measures necessary to
obtain its share of the emission
reductions needed to meet the
progress goal for the area."
The term "progress goal" refers
to the reasonable progress goal
established by the state
containing the Class I area, per
subsection 51.308(d)(1) which
provides that "For each
mandatory Class I Federal area
located within the State, the
State must establish goals
(expressed In deciviews) that
provide for reasonable progress
towards achieving "natural
visibility conditions."
The statute and regulations do
not provide that a state must
guess what an "approved or
approvable" reasonable progress
goal would be beyond the plain
meaning of EPA's rules.
"[W]e interpret the term
'progress goal' in Section
51.308(d)(3)(ii) as an approved
or approvable progress goal."26
NRG stated that the proposed interpretations offered by EPA appear calculated to provide a basis
for EPA to disregard the Texas-Oklahoma consultations that underlie this action, including the
Wichita Mountains reasonable progress goals. However, they are contrary to the plain language
of the rules themselves. Notably, EPA can point to no flaw in the consultations between Texas
and Oklahoma other than EPA's subjective determination that the consultations did not result in
the same suite of emission controls that EPA now plans to impose. The distinguishing feature of
such an analysis is the proposed insertion of extra-statutory and extra-regulatory requirements,
developed only for the Texas action, such as "coordinated management strategies" and
"approved or approvable" goals. In fact, the required consultations occurred, as described in
EPA's Federal Register notice. 79 Fed. Reg. at 74,854.Thus, inadequacy of consultation does
not provide a basis for EPA's action.
64
-------
Footnotes:
25The "downwind" brackets appear in EPA's preamble.
2679 Fed. Reg. at 74,829.
Response: We disagree with this comment. We are not requiring States to guess whether we
will approve a State's reasonable progress goal in the future. The Regional Haze Rule requires
States to conduct adequate four-factor analyses as the first step in the process of establishing
reasonable progress goals and long-term strategies. For the reasons explained in the preamble to
this final rule, Texas failed to conduct a proper four-factor analysis for either its own Class I
areas or the Wichita Mountains. In addition, we disagree that 40 CFR 51.308(d)(3)(i) requires
States to consult concerning coordinated emission management strategies only after their
reasonable progress goals have been established. The plain language of that provision contains
no such requirement. Rather, the commenter appears to have divined this sequence of events
from the mere fact that 40 CFR 51.308(d)(3) follows 40 CFR 51.308(d)(1) in the Code of
Federal Regulations. As we have explained elsewhere, four-factor analyses, reasonable progress
goals, and long-term strategies are inextricably linked. The consultation requirements in 40 CFR
51.308(d)(l)(iv) and 40 CFR 51.308(d)(3)(i) are two sides of the same coin. In regards to other
aspects of the commenter's preferred interpretation of the provisions at 40 CFR 51.308(d), see
our responses to the comments from Luminant and others above.
Comment: [TCEQ/PUCT (0056) p. 5] The TCEQ stated that the EPA's action is based not on
current law or guidance but rather the agency's preference of what the law and guidance should
be. This is apparent from recent meetings the EPA has conducted with regional planning
organizations (RPOs), federal land managers (FLMs), and states on possible changes to the RHR
and guidance - changes that in many ways would codify the approach that the EPA has taken in
proposing disapproval of the Texas and Oklahoma SIPs.
The TCEQ noted that the EPA has indicated intentions to revise the RHR and guidance and is in
the process of holding meetings with relevant stakeholders such as states, FLMs, and RPOs to
receive feedback and input on what these revisions should entail. This is the correct approach
for an agency considering making changes to properly promulgated rules. Several stakeholders
have already expressed to the EPA that the agency needs to more clearly articulate expectations
in the rule or guidance for how to consider the four statutory factors used in setting RPGs. The
EPA has posed a series of questions to stakeholders on how to revise the RHR and guidance,
including how states should address each RPG factor. For example, the EPA asks if the RPG
analysis should include a presumption that certain controls are needed for reasonable progress.
This is precisely what the EPA has done in reviewing the Texas 2009 RH SIP and developing the
proposed FIP, an action that is without a basis in the current regulations. If the EPA finds that in
its review of state RH plans there are flaws in its own rules, the appropriate mechanism for
correcting those flaws is not disapproving those plans; it is through prospective, CAA-compliant
rulemaking. The EPA must base its review of the Texas 2009 RH SIP on what the rule and
guidance required at the time Texas submitted the plan in 2009. Changes to the law must be
properly made through notice and comment rulemaking and not imposed prematurely and
without notice to states after plans are submitted. It is arbitrary and capricious, as well as
contrary to current case law, to require a state to guess what the EPA may choose to require from
65
-------
a state for an approvable plan. The EPA had appropriate rules and guidance, these were
correctly and appropriately followed by the TCEQ in developing the 2009 RH SIP, and the EPA
is obligated to follow its own rules and guidance that were in place when the plan was developed
as it evaluates the merits of the submission.
Response: We disagree with this comment. We are currently soliciting stakeholder feedback on
a number of issues related to the second planning period for regional haze. These issues include
whether we should revise portions of the Regional Haze Rule related to SIP submission
deadlines, the form of five-year progress reports, the RAVI regulations at 40 CFR 51.300-307,
and the calculation of the uniform rate of progress. We have also solicited feedback on new
guidance to aid States in conducting the modeling, control assessments, and other technical work
that will be needed to make reasonable progress going forward. We have not solicited feedback
on revising the regulatory provisions related to four-factor analyses, interstate consultation, or
long-term strategies, as the commenter suggests, because we do not believe these provisions
require revision. Instead, we have clarified our existing interpretation of these provisions in this
rulemaking.
Comment: EPA's "guidance" on interstate consultation is procedurally flawed and
inconsistent with the regional haze rule. [UARG (0065) p. 5-9]
UARG stated that the EPA acknowledges that Oklahoma and Texas engaged in the interstate
consultation process required by EPA's regional haze rule and that Oklahoma did not request any
additional emission reductions from Texas. 79 Fed. Reg. at 74,822-23. EPA suggests, however,
that Oklahoma did not understand its right under the regional haze program to seek more
emission reductions from Texas or that it failed to properly exercise that right. Id. at 74,872.
EPA proposes to conclude that it must step in and force the Texas emission reductions that it
believes Oklahoma should have requested. Id. For the reasons described below, EPA has no
basis for doing so. EPA further concludes that because, in its view, Oklahoma did not
understand the regional haze program's consultation process, new nationally applicable guidance
on interstate consultation, and on how those requirements apply in the "visibility transport"
context, is needed. Id. at 74,823. EPA's proposed "guidance" on this issue is irretrievably
flawed and should be withdrawn.
As an initial matter, UARG stated that the EPA purports to provide nationally applicable
guidance with respect to visibility transport and interstate consultation but fails to provide legally
adequate notice that this rulemaking addresses issues relevant to any state other than Texas and
Oklahoma. See id. At 74,828-30 (EPA's interpretation of 40 C.F.R. § 51.308(d)(1) and (d)(3));
id. at 74,888 (proposed determination of nationwide scope and effect). The title of the proposed
rule does not refer to nationally applicable guidance, and there is no reason to believe that the
general public, including other states, have been apprised that this proposed rule includes
significant new interpretations of EPA rules that are intended to govern the future
implementation of the regional haze program outside of Texas and Oklahoma. For that reason,
EPA's proposed rule is procedurally flawed and the proposed nationally applicable visibility
transport guidance should be withdrawn.
66
-------
UARG stated that the guidance EPA proposes is, moreover, inconsistent with EPA's regional
haze rule and the Agency's existing guidance on interstate consultation. EPA begins its
discussion of its new "interpretation" of 40 C.F.R. § 51.308(d)(1) and (d)(3) by stating that its
regulations:
do not explicitly address situations where the control measures in an upwind state's long-
term strategy are sufficient to obtain its share of reductions needed to meet a RPG
included in a downwind state's SIP, but the goal itself is flawed precisely because the
upwind state never proposed sufficient control measures to ensure reasonable progress in
the first place. Id. at 74,829.
UARG noted that, in such a circumstance, EPA proposes to conclude that it must "disapprove
both the downwind state's goal and the upwind state's long-term strategy." Id. But EPA's
proposed action proceeds from a false premise. The regional haze rule does not specifically
address this situation because that rule is designed in such a way that these circumstances will
not arise. Under the rule, states are empowered to establish RPGs and are generally required to
do so based on an analysis of the four reasonable progress factors. So long as states undertake
that analysis, EPA cannot engage in second-guessing and substitute its own RPGs for those
adopted by a state. EPA's novel proposed guidance would eviscerate the state discretion
provided under the regional haze rule to establish RPGs and would allow EPA to substitute its
own RPG preferences, based on nothing more than an assertion that a downwind state would
have set a different RPG if it had been (from EPA's perspective) more assertive in its dealings
with an upwind state.
UARG stated that EPA's proposed guidance is erroneous is confirmed by the interstate
consultation provisions EPA purports to interpret, which provide no authority for EPA to second-
guess state determinations made through the interstate consultation process. The regional haze
rule expressly provides that if a downwind state believes an upwind state should do more to
reduce its emissions and the upwind state is uncooperative, the downwind state is to document
the consultation and inform EPA of the situation. 40 C.F.R. § 51.308(d)(l)(iv). In the absence
of a state conclusion that an impasse in interstate consultations has been reached, EPA has no
authority to revisit the interstate determinations that have been made. EPA's proposed new
interpretation is therefore inconsistent with the regional haze rule.
UARG stated that the EPA's proposed new interpretation is also inconsistent with EPA's
existing guidance on these matters. The proposed rule cites and describes a 2006 document,
"Additional Regional Haze Questions," in which EPA "encouraged the early identification of
any potential disputes to allow all parties ample opportunity to address and document any
disagreements" and in which EPA explained that when states disagree over measures necessary
to contribute adequately to reasonable progress, the disagreement should be brought to EPA's
attention as early as possible. 79 Fed. Reg. at 74,827. This guidance thus makes clear that
determinations as to the adequacy of apportionment of emission reduction obligations among
upwind and downwind states are for states to make through the interstate consultation process
except where states conclude that agreement cannot be reached. Where a state does not invoke
the provision of the regional haze rule that authorizes it to inform EPA of a dispute that that state
has with other states and where that state uses its discretion under the rules and the CAA to
67
-------
establish an RPG, EPA has no right to assume or conclude that the state meant to or should have
invoked that provision but improperly failed to do so for some unarticulated reason.2 Because
EPA's proposed guidance takes the opposite position, it is inconsistent with the CAA and the
existing regional haze rule and must be withdrawn.
In addition, UARG noted that the EPA's proposed guidance includes a statement that contradicts
the regional haze rule's provisions concerning regional planning organizations ("RPOs"). EPA
acknowledges that its rules allow states to satisfy the RPG analysis technical documentation
requirement by relying on RPO technical analyses approved by all state participants. Id. at
74,829.3 EPA goes on to state, however, that "[i]n situations where a regional planning
organization's analyses are limited, incomplete or do not adequately assess the four-factors ...
states must fill in any remaining gaps to meet this requirement." Id. EPA does not explain what
it means by this statement, but the vague assertion that reliance on RPO technical analyses might
not be appropriate contradicts the plain language of the regional haze rule (in particular, 40
C.F.R. § 51.308(d)(3)(iii)). Because the proposed guidance is inconsistent with the regional
haze rule, it should be withdrawn.
UARG asserted that the EPA in any event cannot apply its novel interpretation - which, in effect,
constitutes new SIP requirements - retroactively. See Bow en v. Georgetown Univ. Hosp., 488
U.S. 204, 208-09 (1988). If EPA wants to change the rules to which states must adhere in
developing SIPs and discharging their implementation responsibilities under the regional haze
program, EPA must conduct national rulemaking - clearly designated as such - and may only
make any changes effective on a prospective basis. EPA thus cannot use its proposed new
guidance as a basis for disapproving any part of the Texas SIP or the Oklahoma SIP.
Footnotes:
2 EPA cites the floor statements of three senators to support its argument that, in enacting the regional haze program,
"Congress was motivated in part by the dilemma of Vermont and other downwind states." 79 Fed. Reg. at 74,830.
By citing Vermont, EPA appears to be referring to its discussion of Vermont v. Thomas, 850 F.2d 99 (2d Cir.
1988), and the dispute between Vermont (the downwind state) and upwind states that, in Vermont's view, were not
adequately controlling emissions from their sources. As an initial matter, the floor statements of individual
members of Congress do not constitute the views of Congress. See, e.g., Kenna v. U.S. Dist. Court for C.D. Cal.,
435 F.3d 1011, 1015 (9th Cir. 2006). Moreover, in Vermont, the downwind state unequivocally stated its
opposition to the emission control decisions and policies of upwind states and in fact brought litigation to seek
resolution of its dispute. That set of facts bears no resemblance to the facts here.
3 EPA adds that "regional haze is a regional problem that requires regional solutions." 79 Fed. Reg. at 74,830. This
statement would seem to acknowledge the important and legitimate role played by the RPOs and that RPO-
developed regional solutions are more appropriate than FIP requirements imposed by EPA.
Response: We disagree with this comment. As an initial matter, we reiterate that we are not
providing "new" interpretations or "new" guidance. We are merely clarifying our existing
interpretation of various statutory and regulatory requirements because the record in this action
revealed the difficulties States can encounter when dealing with visibility transport. We believe
that the clarifications provided in the proposed rule and in this final rule may prove useful to
other States as they prepare for the second planning period. To the extent that we were required
to provide notice of our clarification of existing interpretations, we did so. We published the
proposed rule in the Federal Register. The commenter suggests that notice was inadequate due
to the title of the proposed rule, but points to no authority for this proposition.
68
-------
The commenter is also incorrect that we re "second-guessing" States' reasonable progress goals.
We evaluated Texas' four-factor analysis and found it to be flawed in several respects. As a
result, we proposed to disapprove Texas' four-factor analysis and determination that no new
controls were necessary to achieve reasonable progress at any Class I area. At the same time, we
proposed our own four-factor analysis and proposed to find that there were cost-effective SO2
controls available for several large Texas EGUs that have significant visibility impacts, including
at the Wichita Mountains. Once we made these findings, we necessarily had to propose to
disapprove Oklahoma's reasonable progress goals for the Wichita Mountains because they did
not account for the SO2 reductions that can reasonably be achieved by Texas' EGUs. We have
largely confirmed these determinations in the final rule.40 Therefore, it was our disapproval of
Texas' four-factor analysis and control determinations that served as the predicate for our
disapproval of Oklahoma's RPGs, not the reverse. We acknowledge, however, that the two are
closely intertwined, which is the reason why we provided clarification of these statutory and
regulatory requirements in the proposed rule.
The commenter also mischaracterizes our 2006 Q & A document. There, we encouraged States
to document their disagreements so that we have a sufficient record to determine the adequacy of
each State's SIP and resolve disputes between States. This document does not suggest, however,
that so long as States agree on the apportionment of emission reductions, we are powerless to
review that apportionment or the underlying analyses. If this were the case, States would be free
to collude to do nothing to address visibility impairment in their Class I areas. Similarly, an
upwind State could pressure a downwind State to acquiesce to the upwind State's proposal to do
nothing to address visibility impairment even though the downwind State had concerns. Neither
the CAA nor our implementing regulations contemplate such an unfair system. Instead, States
are charged to conduct four-factor analyses to provide a sound factual predicate to their control
determinations. If a State fails to conduct a four-factor analysis or fails to do so reasonably, then
we are required to disapprove that analysis and step into the shoes of the State. Therefore, while
we agree with the commenter that States should work together through the consultation process
to determine their apportionment of emission reductions, and we prefer to defer to States'
decisions wherever possible, we disagree that we must approve any apportionment of reductions
the States agree upon, regardless of whether it is reasonable or not.
Furthermore, the commenter is incorrect that 40 CFR 51.308(d)(3)(iii) provides States with
unbridled discretion to rely on an RPO's technical analyses in place of their own. The first
sentence of 40 CFR 51.308(d)(3)(iii) makes clear that each State "must" justify its apportionment
of emission reductions with a technically supported four-factor analysis. Our regulations allow
States to work together in RPOs, which provide for enhanced communication and pooled
resources. As a result, the second sentence of 40 CFR 51.308(d)(3)(iii) provides that States
40 As discussed elsewhere, the D.C. Circuit recently remanded without vacating a number of CSAPR's state
emissions budgets. EME Homer City Generation v. EPA, 795 F.3d 118 (D.C. Cir2015). We are in the process of
acting on the Court's remand. As a result, at this time we cannot ensure that CSAPR will continue to be an
appropriate alternative to BART for Texas EGUs. Given the uncertainty arising from the remand of some of the
state CSAPR budgets, we have decided not to finalize that portion of our FIP relying on CSAPR as an alternative to
S02 and NOx BART for EGUs in Texas. As the question of how best to address the BART requirements for these
significant sources of emissions of visibility impairing pollutants remains undecided, we have also concluded
that our proposed action to establish new RPGs for Oklahoma should also be addressed in a future rulemaking.
69
-------
"may" satisfy the first sentence's requirement by conducting the requisite analysis in
coordination with their neighbor States in the RPO process. In other words, the second sentence
of 40 CFR 51.308(d)(3)(iii) merely allows States to rely on coordinated analyses instead of a
State's own analysis. It does not diminish the requirement that the analyses must be complete
and well-reasoned. To the extent that RPO analyses do not address all the RH requirements for a
particular State, then the State is required to make up the difference. Finally, the EPA is not
applying a novel interpretation or creating new SIP requirements. We are applying existing
requirements to a complex factual scenario. Therefore, the commenter's citation to Bow en is
inapposite.
Comment: EPA must approve Texas' long-term strategy for the same reasons EPA
approved Nebraska's. [Luminant (0061) p. 83]
Luminant stated that EPA's actions with respect to Nebraska and South Dakota unravel EPA's
new theory of the need and justification for "simultaneous" action on Texas' and Oklahoma's
regional haze plans, and its interpretation of 40 C.F.R. § 51.308(d)(3)(ii) that underlies the
theory (discussed below). Commenters on EPA's action for Nebraska asserted that, where one
state's RPG is determined to be "not sufficient" (i.e., not "approvable"), "each state participating
in the regional planning process for the applicable Class I area [must] be required to re-evaluate
their LTS and make appropriate revisions to ensure they met their apportionment of emission
reduction obligations necessary for achieving reasonable progress."565 EPA disagreed with this
comment, and explained that the process works in a fundamentally different way. EPA
explained that states look to "air quality modeling performed by the RPOs" "[t]o set RPGs."566
EPA further explained that there is "an inherent amount of uncertainty in the assumed emissions
from all sources" and that when a state's final action "deviate[s] from what was included in the
modeling," the remedy is for affected states to "consider asking [the contributing state] for
additional emission reductions" "during subsequent periodic progress reports and regional haze
SIP revisions."567
Luminant noted that EPA's proposed action here cannot be squared with its actions in connection
with the Nebraska and South Dakota SIPs. EPA claims here that "[t]o properly assess whether
Oklahoma had satisfied the reasonable progress requirements," it "had to review and evaluate
Texas' regional haze SIP before proposing action on Oklahoma's RPGs."568 But this is
fundamentally at odds with EPA's explanation in its Nebraska/South Dakota action that states
fulfill their statutory obligations by consulting and making assumptions together to develop their
regional haze SIPs, and then make adjustments in future planning periods as necessary. As
discussed above, EPA's prior explanation of the correct process is consistent with its regional
haze regulations, and its current and novel interpretation is not. Further, while EPA now claims
that "[i]n order to address these intricately intertwined issues between Oklahoma and Texas, it is
appropriate to review them simultaneously,"569 Oklahoma and Texas are no more "intertwined"
than any other two states that must consult over out-of-state impacts. Indeed, the long distances
between Texas sources and Oklahoma's Class I area—and the negligible visibility impacts
involved—make Texas and Oklahoma less "intertwined" than most other states. There is no
justification for EPA's approach here. And if EPA wants to require states to follow its new
approach to regional haze planning, it must amend its regulations to establish a new consultation
70
-------
process, but it cannot impose one by fiat and use it to retroactively judge the Texas and
Oklahoma submissions.
Footnotes:
565 77 Fed. Reg. at 40,155.
566 Id.
567 Id. at 40,155-56 (emphasis added).
568 79 Fed. Reg. at 74,821.
569 Id. at 74,822.
Response: We disagree with this comment. While the commenter is correct that our statements
in the Nebraska haze action appear to be inconsistent with our explanation of the regional haze
requirements in this rulemaking, the Eighth Circuit granted our request for a voluntary remand of
our action on the Nebraska long-term strategy on March 19, 2015. In our motion for a voluntary
remand, we explained that "EPA is concerned that its present explanation could potentially be
construed in a manner that is inconsistent with EPA's interpretation of the relevant statutory and
regulatory requirements. Remand is therefore appropriate so that EPA has the opportunity to
amend or further explain its rationale for declining to require additional controls as part of the
FIP's long-term strategy, to more fully respond to comments submitted by the public, and to take
further action if necessary." Therefore, the statements relied upon by the commenter are no
longer relevant. We are currently reconsidering the Nebraska long-term strategy as it relates to
the South Dakota reasonable progress goals and will take appropriate action in the future.
While the commenter is correct that we acted on most regional haze SIPs independently, in some
instances, we acted on multiple regional haze SIPs simultaneously. See, e.g., 78 FR 59,825
(Sept. 30, 2013) (disapproving Michigan and Minnesota SIPs with respect to taconite facilities);
78 FR 8705 (Feb. 6, 2013) (issuing a FIP for Michigan and Minnesota taconite facilities); 77 FR
33,642 (June 7, 2012) (issuing limited disapprovals for multiple States' SIPs due to reliance on
CAIR and promulgating FIPs that relied on CSAPR instead). More importantly, however, even
when we take individual action on a single State's SIP, we consider the record developed during
the RPO process and other State's SIP submissions in order to fully evaluate the adequacy of the
SIP. In this instance, we chose to act on the Oklahoma RPGs and the Texas SIP simultaneously
for good reasons. Texas' sources impact the visibility at the Wichita Mountains more than
Oklahoma's own sources do, and the consultation record revealed misunderstandings over the
relevant roles of upwind and downwind States in addressing visibility transport issues. As a
result, we determined that this action provided an appropriate vehicle to clarify to the States our
interpretation of various statutory and regulatory requirements.
Comment: [Luminant (0061) p. 147] Luminant concluded that the EPA's proposed disapproval
of Texas' and Oklahoma's regional haze SIPs and EPA's proposed FIPs are contrary to the
Clean Air Act and EPA's regulations and cannot be finalized. EPA should take final action to
approve Texas' and Oklahoma's regional haze SIPs in full. To the extent EPA would have
Texas, Oklahoma, and other states address regional haze issues in the new and different manner
EPA is now proposing, EPA must first amend its regulations consistent with the statute prior to
the second planning period so that all states are subject to the same standards and consistent
treatment.
71
-------
Response: We disagree with this comment for the reasons provided in responses to earlier
comments from Luminant above.
Comment: EPA's Interpretation of the Clean Air Act is Reasonable, and Indeed is the
Best Reading of the Statute. [Earthjustice (0067) p.33]
Earthjustice et al., stated that, under the Clean Air Act, EPA must ultimately decide, among other
things, whether the states' respective SIPs ensure "reasonable progress" towards the national
visibility goal,42 and whether those SIPs "contain adequate provisions prohibiting . . . any source
or other type of emissions activity within the State from emitting any air pollutant in amounts
which will interfere with measures required to be included in the applicable implementation plan
for any other State." 42 U.S.C. § 7410(a)(2)(D)(i)(II) (emphasis added). In making that
determination, EPA evaluates whether any interstate "consultation" regarding the transport of
visibility pollution meets the requirements of the rule, and includes "all measures necessary" to
achieve each states' respective reasonable progress goals. EPA is not required to rubberstamp a
cursory and unreasonable visibility-transport "consultation" that undermines the core purposes
and requirements of the haze program. In such circumstances, EPA has the authority—indeed,
an obligation—to disapprove an inadequate SIP, and explain what is required to comply with the
transport and consultation provisions of the Clean Air Act and its implementing regulations.
EPA's interpretation of the Clean Air Act's good neighbor and visibility provisions, as well as
the Regional Haze Rule, fits comfortably within that well-established scope of authority.43
Earthjustice et al., stated that, even if its implementing regulations did not exist, EPA's proposed
clarification of its existing regulations would be a valid interpretation of - indeed, would be
compelled by - the text of the Clean Air Act. EPA is not powerless to let states exploit any
silences or ambiguities in EPA's own regulations in ways that undermine the statutory text and
its purposes. For example, it would be inconsistent with the haze provisions to interpret the
statute such that EPA had to approve a consultation simply because Texas proposed the control
measures that Oklahoma used as part of the basis for its proposed reasonable progress goal.
Such an interpretation would flip the statute on its head and eviscerate the oversight role
Congress gave to EPA by allowing states to have the final say on what constitutes reasonable
progress. Such an interpretation cannot be reconciled with the statutory scheme in which states
submit their haze plans to EPA, and EPA determines whether the plans comply with the Clean
Air Act.
Earthjustice et al., stated that if the agency could not take the action it proposes here, states
would be free to game the Clean Air Act and issue SIPs that defeat the purposes of the Clean Air
Act whenever EPA regulations do not explicitly address every conceivable issue. That is clearly
not what Congress intended when it gave EPA the ultimate oversight over state implementation
plans. Additionally, such an interpretation would create an incentive—or an obligation—for
EPA to issue exceedingly detailed regulations, which would ultimately limit state discretion
contrary to the CAA's cooperative federalism structure.44
Footnotes:
42 40 CFR 51.308(d)(3)(ii)
72
-------
43 Cf. EPA v. EMEHomer City Generation, L.P., 134 S. Ct. 1584, 1600-01 (2014) (observing that "once EPA has
found a SIP inadequate, the Agency has a statutory duty to issue a FIP 'at any time' within two years," and
upholding EPA's Transport Rule promulgated in the context of, and subsequent to, disapproving the good neighbor
SIP provisions of 23 states, in part, because "nothing in the statute places EPA under an obligation to provide
specific metrics to States before they undertake to fulfill their good neighbor obligations.").
44 We agree with EPA that any petitions for review of the rule must be filed in the D.C. Circuit Court of Appeals.
Even if EPA were not clarifying its interpretation of the existing haze rule in this action, EPA's rule takes action on
two SIPs, the review of which would normally occur in two separate circuits. For this reason alone, review of the
rule must occur in the D.C. Circuit.
Response: We generally agree with this comment, but take no position as to specific statements.
Comment: [TCEQ/PUCT (0056) p. 13] The TCEQ stated that the RHR does not require that a
downwind state's RPG must be "approved or approvable" in order to determine if the upwind
state's long-term strategy meets the statute or the rule. This is a new and illegal change to the
RHR without going through notice and comment rulemaking as required by the Administrative
Procedures Act and is thus an arbitrary and capricious determination by the EPA.
The TCEQ stated that the EPA's proposed disapproval of the state consultation requirements is
based upon Oklahoma's determination, subsequent to submittal of the Texas 2009 RH SIP, that it
required further reductions from Texas. The EPA has not justified its determination that Texas
failed to meet the requirements of §51,308(d)(3)(i) and in fact the record shows that the process
as laid out in the SIP and as required by the rule was followed by Texas. The EPA's
determination is based on a new definition of progress goal in subsection (d)(3)(ii) and a
misstatement of the actual rule itself in subparagraph (i).
The TCEQ stated that Texas met the consultation requirements in §51.308(d)(3)(i). Texas
determined where emissions were reasonably anticipated to contribute to visibility impairment in
Oklahoma. Texas consulted with Oklahoma. The EPA asserts that the TCEQ should have
provided information necessary to identify reasonable reductions, which is not required by the
RHR. Oklahoma requested information on controls identified by CENRAP. Oklahoma had
information on control upgrades contained in the proposed Texas 2009 RH SIP. Yet, it did not
request additional controls on Texas sources or disagree with Texas' determination that
additional controls were not warranted during the first planning period. It was only after
consultation with Texas that Oklahoma argued that it needed controls that they did not have
authority to require from Texas sources. Oklahoma's after-the-fact change in position and the
EPA's subsequent proposed disapproval of their RPGs for Wichita Mountains does not provide
the legal basis for proposed disapproval of Texas' long-term strategy consultations. The RHR
does not require that a downwind state's RPG must be "approved or approvable" in order to
determine if the upwind state's long-term strategy meets the statute or the rule. This is a new and
illegal change to the RHR and is thus an arbitrary and capricious determination by the EPA.
Response: We disagree with this comment. As explained in response to prior comments in this
section, we are not making a change to the Regional Haze Rule, nor are we making a "new
definition of progress goal." We merely clarified that the progress goal referred to in 40 CFR
51.308(d)(3)(ii) must be approved or, at a minimum, approvable for a State to satisfy that
provision. This is sound logic. For example, imagine a provision that required State A to
73
-------
provide half of the emission reductions necessary to achieve natural visibility conditions in a
Class I area in State B. Together, State A and State B determined that each State would need to
provide 1,000 tons of emissions reductions to achieve this goal. As a result, State A and State B
each developed a SIP that required 1,000 tons of emission reductions. When reviewing the SIPs,
however, we determined that the States had miscalculated the amount of emission reductions
necessary to achieve natural visibility conditions at State B's Class I area. In reality, each State
should have provided 2,000 tons of emission reductions. Even though State A thought its SIP
was doing enough to achieve natural conditions at State B's Class I area, that conclusion was
based on faulty analysis. As a result, State A has not satisfied the provision because State A's
SIP does not provide for half of the emission reductions necessary to achieve natural conditions
in State B's Class I area. Common sense dictates that we must disapprove State A's SIP. Under
the commenter's theory, however, we should approve State A's SIP anyway, even though the
SIP was based on faulty analysis, does not satisfy the regulatory provision in question, and will
not permit State B's Class I area to achieve natural conditions.
In addition to noting this faulty analysis, we take issue with a statement made by the TCEQ. The
TCEQ holds up its statement that it, "determined where emissions were reasonably anticipated to
contribute to visibility impairment in Oklahoma," as satisfying section 308(d)(3)(i). First, we
note that because Oklahoma has only one Class I area—the Wichita Mountains, there is no
question as to the location of the impact of Texas' emissions. Second, merely identifying the
location of the impacts, and to the extent the TCEQ meant—the magnitude of these emissions,
does not satisfy this regulation. As section 308(d)(3)(i) states, the purpose of the consultations is
to " develop coordinated emission management strategies." Consequently, the TCEQ's
statement that it need not have, "provided information necessary to identify reasonable
reductions" in order to satisfy this regulation, is incorrect. At the heart of any coordinated effort
is the exchange of information. In this instance, this exchange of information necessarily
required the locations, magnitude, and costs of individual sources controls. As we noted in our
proposal, Texas' abbreviated source analysis that it references here, did not provide Oklahoma
with the information necessary in order to make an informed decision. Texas' analysis did not
include individual source impacts and was missing many source control costs. Our analysis
filled that knowledge gap.
Lastly, the TCEQ states that Oklahoma made a post-consultation request for controls, and took
the position that it did not have the legal authority to request those controls from Texas. The
TCEQ states such a request does not mean that Texas was obligated to supply those controls
because the request occurred after the consultation process. The TCEQ does not provide any
citation for its analysis. We can only assume it is referring to a number of statements in the
Oklahoma SIP acknowledging the magnitude of Texas' emissions on the visibility at the Wichita
Mountain and/or the following statements made at the Oklahoma public hearing:
Western Farmers' Electric Cooperative - In a letter received by DEQ on December
16, 2009, signed by Gerald Butcher.
25. COMMENT: The DEQ has determined the impact of out-of-state emissions
(primarily from the State of Texas) on visibility in the WIMO are significant.
Conversely, Texas recently submitted its Regional Haze SIP Revision to EPA and
74
-------
therein indicated emissions originating from Texas do not impact visibility in the
WIMO. Therefore, there appears to be a significant disagreement between the
findings from each State. How does the DEQ propose to resolve this issue?
RESPONSE: DEQ stands by its assessment that Texas emissions significantly
impair visibility at the Wichita Mountains. EPA can evaluate both SIPs and will
be ultimately responsible for determining which findings are supported by the
technical demonstrations included in each SIP.
26. COMMENT: Did the DEQ advise Texas that additional emission reductions
from Texas sources would not be needed to help Oklahoma meet the WIMO
reasonable progress goals, and if so, on what basis was such determination made?
RESPONSE: DEQ advised Texas of its finding during the consultation process that
Oklahoma would be unable to meet the uniform rate of progress without additional
reductions, including those from Texas sources. However, DEQ does not have the
regulatory authority require emissions reductions in other states. Only Texas and
EPA can require those reductions.
27. COMMENT: ... Based on the above and the fact that"... even the elimination
of all anthropogenic sources within Oklahoma is not sufficient to comply with
uniform rate of progress", the DEQ concluded "any effective strategy for managing
visibility impairment at the Wichita Mountains must address outside sources
including regional and international transport." However, the Revision is silent as
to how such outside sources will be addressed.
RESPONSE: See response to previous comment.
We disagree with Oklahoma that it did not have authority to request additional controls from
Texas. It is required under Section 2.1(c) to secure all the authority it needs to adopt and
implement its regional haze SIP, and to provide that authority with its SIP submission, which it
did. Also, as we discuss in our proposal, we do not hold Oklahoma blameless in its failed
consultations with Texas. On the contrary, we proposed to find:
We do not agree, however, with the ODEQ's approach to consultation to address
impacts from emissions from Texas. At the time that Oklahoma was developing
its SIP, it had (1) abundant information showing the impact of Texas sources on
visibility at the Wichita Mountains, particularly from EGU sources in northeast
Texas, and (2) evidence that cost-effective controls on these sources were likely
available. Despite this information, the ODEQ requested neither that the TCEQ
further investigate controls at these sources nor did it request additional reductions
from Texas sources to address the impacts of emissions from these sources at the
Wichita Mountains. The Regional Haze Rule requires states to use the consultation
process under Sections 51.308(d)(l)(iv) in the development of RPGs to ensure that
all states, including downwind states, take a hard look at what measures are
necessary for ensuring reasonable progress towards improving and maintaining
75
-------
visibility at Class I areas. Lacking development of critical information during its
consultations with Texas, we believe that Oklahoma did not have adequate
information to reasonably establish its RPG for the Wichita Mountains, and, as
explained below, should have requested that the TCEQ further investigate these
sources or requested additional reductions from Texas sources to ensure that all
reasonable measures to improve visibility were included in Texas' LTS and
incorporated into Oklahoma's RPG for the Wichita Mountains.
4. Consideration of Visibility in the Reasonable Progress Analysis
Comment: [Luminant (0061) p. 1] and [Luminant (0061) p. 62]
Luminant stated that the EPA's proposal is further contrary to law because it attempts to impose
a non-statutory factor on Texas. EPA's proposal is not based on an analysis of the four statutory
factors for "reasonable progress," but instead hinges on a non-statutory factor—"visibility
benefit"—to determine whether additional emission reductions should be required and which
sources must incur the costs. Federal Land Managers have recently warned EPA against using
"visibility benefit" as "a fifth factor" since EPA has no "statutory mandate to do so." 5
Nevertheless, EPA treats this non-statutory factor as the primary consideration in its analysis
here. Thus, "EPA [has] overstepped the bounds of its narrow statutory role in the SIP approval
process" and acted "ultra vires" by relying on a "factor [ ] which Congress has not intended [the
EPA] to consider." 6
Luminant Stated that the EPA's only so-called fault with Texas' analysis is "how it analyzed and
weighed the four reasonable progress factors,"439 and specifically the manner in which Texas
considered the potential visibility benefits from the control strategies that it examined.440 EPA
contends that Texas should have "separately evaluate[d] the visibility benefit from the
implementation of [individual] controls]." 441 EPA's statutory role does not extend to dictating
"how" a State considers the four actors. As EPA itself has explained, "States have considerable
flexibility in how they take these factors into consideration ..." 442 There is no requirement in
the statute, regulations, or guidance that Texas considers the visibility benefit from the
implementation of individual controls in the manner EPA would—or even to consider visibility
at all in its four-factor analysis. Indeed, EPA has approved other States' four-factor analyses,
noting specifically that they did not perform this type of visibility analysis.443
Moreover, Luminant asserted that visibility benefit is not even one of the statutory factors
required to be considered for reasonable progress, in stark contrast to the statutory factors for
source-specific BART, which include visibility benefit as a fifth factor. Texas reviewed the
visibility benefits of its proposed control scenario purely as a discretionary matter, and EPA may
not disapprove Texas' SIP because Texas did not conduct some different visibility analysis that
is not required by either the statute or the regulations. Indeed, even in the BART context where
visibility benefit is a statutory factor, EPA's previous attempt to elevate visibility benefits above
the other statutory factors was squarely rejected by the D.C. Circuit as contrary to the statute.444
EPA's attempt here to require a "separate" visibility analysis and use that as an additional factor
"in such a dramatically different fashion" than the statutory factors Texas considered is therefore
76
-------
doubly unlawful.445 In the reasonable progress context, visibility improvement is not one of the
statutory factors for reasonable progress, and it certainly is not the determining factor as EPA
treats it here. EPA's proposal here is thus contrary to the text and structure of the statute because
it "isolates [the visibility] benefit calculation and constrains authority Congress conferred on the
States." 446
Luminant stated that EPA's approach to visibility benefit, which is the only basis for its proposed
disapproval and FIP—is thus inconsistent with the statute and regulations. EPA's approach
elevates visibility benefit to the determinative factor—above the statutory factors. Indeed, EPA
uses its visibility analysis to identify the sources that are then reviewed for costs. This is
backwards and plainly not authorized by or consistent with the statute or regulations and is not a
proper application of the four statutory factors.
Footnotes:
5USDA Forest Serv., Recommendations for Improved Implementation of the Regional Haze Program 5 (May 2014),
available at http://tinyurl.com/FederalLandrec.
6 Luminant Generation Co. LLC v. EPA, 675 F.3d 917, 925, 926 (5th Cir. 2012) (internal citations omitted)
(internal quotations omitted).
439 TX SIP TSD at 18 (emphasis added).
440 79 Fed. Reg. at 74,838-39.
441 Id. at 74,839.
442 77 Fed. Reg. at 30,251 (emphasis added).
443 See, e.g., 78 Fed. Reg. 10,546, 10,553 (Feb. 14, 2013) (approving Alaska's reasonable progress goals and
recognizing in response to comments that "the SIP submission does not specifically identify the contribution of coal-
combustion sources to visibility impairment inDenali National Park . .. ."); 77 Fed. Reg. 70,693, 70,702 (Nov. 27,
2012) (approving New Mexico's reasonable progress analysis that did not evaluate the contribution from individual
EGUs). See also WildEarth Guardians v. EPA, 770 F.3d at 944 (affirming EPA's approval of New Mexico's
reasonable progress analysis and holding: "Neither the Clean Air Act nor the Regional Haze Rule requires source-
specific analysis in the determination of reasonable progress.").
444 Am. Corn Growers Ass'n, 291 F.3d at 6 ("The Haze Rule's splitting of the statutory factors is consistent with
neither the text nor the structure of the statute.").
445 Id. at 6.
46 Id. at 9.
Response: We disagree with this comment. The commenter appears to be stating that States (or
EPA when promulgating a FIP) cannot consider visibility in any way in determining reasonable
progress and that we must approve a State's reasonable progress goals and long-term strategy as
long as all four mandatory reasonable progress factors are analyzed to some degree. This
cramped view is at odds with the overarching purpose of the CAA's visibility provisions.
Congress declared in CAA Section 169A(a)(l) a national goal of the "prevention of any future,
and the remedying of any existing, impairment of visibility in mandatory Class I Federal areas
which impairment results from manmade air pollution." CAA Section 169A(b)(2) required the
Administrator to promulgate regulations to assure "reasonable progress toward meeting the
national goal." Thus, the entire purpose of the reasonable progress mandate is to achieve the
national goal of natural visibility conditions. A reasonable progress analysis that does not take
visibility into account in some fashion would be directly at odds with the analysis's core purpose.
CAA Section 169A(g)(l) goes on to state that, in determining "reasonable progress," States must
consider four-factor: "the costs of compliance, the time necessary for compliance, and the energy
and nonair quality environmental impacts of compliance, and the remaining useful life of any
77
-------
existing source subject to such requirements." This consideration is commonly referred to as the
"four-factor analysis."41 The crux of the commenter's argument seems to be that, because this list
of factors does not include visibility, States can ignore visibility altogether or, if they choose,
consider it in any fashion they want.
While we agree that visibility is not one of the four mandatory factors explicitly listed for
consideration in Section 169A(g)(l) or 40 CFR 51.308(d)(l)(i)(A), the term "reasonable
progress" itself means reasonable progress towards the national goal of natural visibility
conditions. The Supreme Court has stated that, "[i]n determining whether Congress has
specifically addressed the question at issue, a reviewing court should not confine itself to
examining a particular statutory provision in isolation. The meaning—or ambiguity—of certain
words or phrases may only become evident when placed in context. It is a 'fundamental canon
of statutory construction that the words of a statute must be read in their context and with a view
to their place in the overall statutory scheme.' A court must therefore interpret the statute 'as a
symmetrical and coherent regulatory scheme' and 'fit, if possible, all parts into an harmonious
whole.'"42
To ensure a coherent regulatory scheme, we believe that states (or EPA when promulgating a
FIP) can consider visibility when determining reasonable progress in at least two ways. First,
states can consider the visibility impacts of sources when determining what sources to analyze
under the four-factor framework. CAA Section 169A(b)(2) does not provide any direction
regarding which sources or source categories a state should analyze when determining
reasonable progress. Similarly, CAA Section 169A(g)(l) refers to "any existing source subject
to such requirements," but unlike the BART provisions, does not identify which existing sources
or source categories should be subject to reasonable progress requirements. Given this statutory
ambiguity, we believe that allowing states to consider visibility impacts when determining the
scope of the reasonable progress analysis is a reasonable interpretation of the statute "as a
harmonious whole." As such, states can develop screening metrics that target those sources with
the greatest visibility impacts for further analysis. Our 2007 guidance advocated this approach,
and nearly all states, including Texas, used metrics like Q/d to consider the potential visibility
impacts of their sources and screen out those sources with low visibility impacts.43 We followed
41 Correspondingly, under Section 51.308(d)(1) of the Regional Haze Rule, promulgated in response to this mandate,
States must "establish goals (expressed in deciviews) that provide for reasonable progress towards achieving natural
visibility conditions" for each Class I area within a State. RPGs are interim goals that represent measurable,
incremental visibility improvement over time toward the goal of natural visibility conditions. Section
51.308(d)(l)(i)(A) requires States to consider the four statutory factors when establishing their RPGs.
42 FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 132-33 (2000) (quoting Davis v. Michigan Dept. of
Treasury, 489 U.S. 803, 809 (1989), Gustafsonv. Alloyd Co., 513 U.S. 561, 569 (1995), and FTC v. Mandel
Brothers, Inc., 359 U.S. 385, 389 (1959)). The Court delineated this canon of statutory construction within "step
one" of its Chevron analysis, recounting that under Chevron, a reviewing court must first ask "whether Congress has
directly spoken to the precise question at issue," and that if Congress has done so, the inquiry is at an end and the
court "must give effect to the unambiguously expressed intent of Congress." Chevron U.S.A. Inc. v. Natural
Resources Defense Council, Inc., 467 U.S. 837.
43 For example, in VISTAS states, to select the specific point sources that would be considered for each Class I
area, VISTAS first identified the geographic area that was most likely to influence visibility in each Class I area and
then identified the major SO2 point sources in that geographic area. The distance-weighted point source SO2
emissions (Q/d) were combined with the gridded extinction-weighted back-trajectory residence times. The distance-
78
-------
this same approach in our FIP by using both Q/d and a second metric based on a source's
modeled percent contribution to total visibility impairment at impacted Class I areas. If states or
we could not consider visibility impacts as a way of identifying which sources should be
considered for additional controls, then states would have no rational way to differentiate
between hundreds of sources that vary in distance from Class I areas, emit different visibility
impairing pollutants in varying amounts, and are subject to diverse meteorological conditions
that affect the transport of visibility-impairing pollutants. The result would be a cumbersome
analysis encompassing hundreds of sources (or in the case of Texas, well over a thousand), many
of which may have little if any impact on visibility in Class I areas. Congress could not have
intended such an incongruous result.
Allowing consideration of visibility improvement is appropriate for several reasons. Most
importantly, it aligns with Congress' national goal, which is to remedy existing impairment of
visibility in Class I areas. While Section 169A(g)(l) of the CAA contains a list of factors states
must consider when determining reasonable progress, we do not believe that list is exclusive. As
the Eighth Circuit Court acknowledged in North Dakota v. EPA, states can take visibility
improvement into account when evaluating reasonable progress controls so long as they do so in
a reasonable way.44 We have iterated this position in previous regional haze actions. For
example, in our final rule on the Montana regional haze SIP, we stated, "We agree that visibility
improvement is not one of the four factors required by CAA Section 169A(g)(l) and 40 CFR
51.308(d)(l)(i)(A), however, it (along with other relevant factors) can be considered when
determining controls that should be required for reasonable progress."45 Similarly, in our final
rule on the Arizona regional haze SIP, we concluded that, "while visibility is not an explicitly
listed factor to consider when determining whether additional controls are reasonable, the
purpose of the four-factor analysis is to determine what degree of progress toward natural
visibility conditions is reasonable. Therefore it is appropriate to consider the projected visibility
benefit of the controls when determining if the controls are needed to make reasonable progress."
Second, once a universe of sources has been identified for analysis, we believe that States have
the option of considering the visibility improvement that will result from potential control
options when weighing the four statutory factors. Allowing consideration of visibility
improvement is appropriate for several reasons. First, it aligns with Congress' national goal,
which is to remedy existing impairment of visibility in Class I areas. Second, while Section
169A(g)(l) of the CAA contains a list of factors States must consider when determining
reasonable progress, we do not believe that list is exclusive. As the Eighth Circuit acknowledged
in North Dakota v. EPA, States can take visibility improvement into account when evaluating
reasonable progress controls so long as they do so in a reasonable way.46 We have iterated this
position in previous regional haze actions. For example, in our final rule on the Montana
regional haze SIP, we stated, "We agree that visibility improvement is not one of the four-factor
weighted (Q/d) gridded point source SO2 emissions were then multiplied by the total extinction-weighted back-
trajectory residence times on a cell-by-cell basis and then normalized. VISTAS Area of Influence Analyses, 2007,
is available in the docket for this action.
44 North Dakota v. EPA, 730 F.3d 750, 766 (8th Cir. 2013).
45 77 FR 57864, 57899, 57901; see also Montana Proposed Rule, 77 FR 23988, 24062.
46 North Dakota v. EPA, 730 F.3d 750, 766 (8th Cir. 2013).
79
-------
required by CAA Section 169A(g)(l) and 40 CFR 51.308(d)(l)(i)(A), however, it (along with
other relevant factors) can be considered when determining controls that should be required for
reasonable progress."47 Similarly, in our final rule on the Arizona regional haze SIP, we
concluded that, "while visibility is not an explicitly listed factor to consider when determining
whether additional controls are reasonable, the purpose of the four-factor analysis is to determine
what degree of progress toward natural visibility conditions is reasonable. Therefore it is
appropriate to consider the projected visibility benefit of the controls when determining if the
controls are needed to make reasonable progress."48 Third, allowing States to consider visibility
improvement in addition to the four statutory factors ensures that only those cost-effective
controls that will achieve sufficient benefits are required. If States were not permitted to
consider visibility improvement when conducting their control determinations, then States would
have to require all cost-effective controls (assuming no limiting energy or non-air quality
environmental impacts) regardless of whether some of those controls would be more beneficial
than others. Oddly, the commenter appears to be suggesting that if we had not considered
visibility benefits in our analysis, we would not have controlled certain sources. On the contrary,
we decided not to require certain cost-effective controls because they would not achieve as much
benefit as other controls. If the commenter is correct and the consideration of visibility benefits
is impermissible in a four-factor analysis, then we would have required all cost-effective
controls, including those at the Parrish and Welsh facilities. Fourth, we note that Congress did
not provide any guidance as to how States should consider "the costs of compliance." One
possible way a State could "consider" costs is to compare them to prospective benefits. In other
words, we believe the first statutory factor is capacious enough to allow for a comparison of
cost-effectiveness to visibility improvement. Finally, we note that our 2007 guidance explicitly
permits States to consider other relevant factors when conducting a four-factor analysis,49 and
many States, including Texas, did so. In conclusion, we believe that States are permitted, but not
required, to consider visibility improvement in addition to the four statutory factors when making
their reasonable progress determinations.
The commenter alludes that visibility improvement is irrelevant to a four-factor analysis because
Congress did not include it as one of the four-factor but did include it as a factor to be considered
in determining BART. We do not find this reasoning to be persuasive. The sources that
Congress subjected to the BART requirement (i.e., sources grandfathered from the PSD
requirement) were not necessarily sources that would have an impact on visibility impairment.
As such, Congress included specific language in CAA Sections 169A(b)(2)(A) and 169A(g)(2)
to ensure that only those grandfathered sources that cause or contribute to visibility impairment
and that would result in visibility improvement if controlled would be required to install BART.
On the other hand, the national goal of achieving natural visibility conditions is central to the
notion of reasonable progress, so Congress had no need to include language regarding visibility
improvement in CAA section 169A(g)(l).
47 77 FR 57864, 57899, 57901; see also Montana Proposed Rule, 77 FR 23988, 24062.
48 79 FR 9318, footnote 137 (finalized based on this same reasoning at 79 FR 52420); TX TSD, at 7, footnote 6; FIP
TSD, at 12; 79 FR 74874.
49 "In determining reasonable progress, CAA §169A(g)(l) requires States to take into consideration a number of
factors. However, you have flexibility in how to take into consideration these statutory factors and any other factors
that you have determined to be relevant." 2007 Guidance at 5-1.
80
-------
We also disagree with the commenter that we cannot disapprove a State's SIP where the State
has considered visibility improvement in an unreasonable fashion. As the Eighth Circuit
explained in North Dakota, "[although the State was free to employ its own visibility model and
to consider visibility improvement in its reasonable progress determinations, it was not free to do
so in a manner that was inconsistent with the CAA."50 Like the State of North Dakota, Texas
chose to evaluate visibility improvement alongside the four statutory reasonable progress factors,
but did so by using a model that employed degraded background conditions. As a result, Texas'
conclusion that the costs of additional controls were not worth the benefits was unreasonable,
and we appropriately disapproved this portion of Texas' SIP. The fact that Texas' decision to
evaluate visibility improvement was "discretionary" does not mean that Texas was free to
exercise that discretion in an unreasonable manner. We discuss several ways that Texas'
consideration of visibility improvement in its reasonable progress determinations was
unreasonable elsewhere in our proposal, and in our Texas TSD.51 One point worth mentioning
here, however, is that Texas estimated the visibility improvement of potential controls by making
comparisons to degraded background conditions instead of to natural background conditions,
which is precisely the same mistake that North Dakota made.52
We note that the Tenth Circuit's decision in WildEarth Guardians v. EPA does not address the
issues present in this case.53 There, the Tenth Circuit merely held that the CAA does not require a
State to conduct a source-specific reasonable progress analysis. The Court did not hold that a
State is free to conduct any type of analysis irrespective of whether or not the analysis is
reasonable. Nor did the Court hold that the CAA prevents States or EPA from conducting a
source-specific analysis if that approach is determined to be appropriate.
Finally, we disagree with the commenter that we elevated visibility improvement to a place of
primary importance, either in disapproving Texas' SIP or in promulgating our FIP. The flaws
with Texas' visibility modeling were only one aspect of our disapproval. Moreover, we stated
on multiple occasions in our proposal that we considered all four statutory factors in our analysis.
Our analysis does not give greater weight to one factor over another; rather, we considered all
four factors fully, revealing that the cost factor, which included visibility improvement
consideration, was the most determinative in our decisions and that costs and visibility
improvement were the two most important factors in our decisions. The commenter's citation to
American Corn Growers is also inapposite. There, the D.C. Circuit Court faulted how EPA
assessed the statutory fifth factor of visibility improvement in a BART determination (not a
reasonable progress determination) by using a regional, multi-source, group approach to
assessing the visibility improvement factor, while assessing the other four statutory BART
50 North Dakota, 730 F.3d at 766.
51 See Section B.2 of the Texas TSD and Section V.C.3 of our proposal (79 FR 74818).
52 In contrast, Texas conducted a proper visibility analysis using natural background conditions elsewhere in its SIP
when the state assessed the visibility impacts of its BART sources. See Texas Regional Haze SIP, Appendix 9-5 at
2-11 ("The source's HI [haze index] is compared to natural conditions to assess the significance of the source's
visibility impact. EPA guidance lists natural conditions (bnatural) by Class I area in terms of Mm-1 (EPA, 2003b)
and assumes clean conditions with no anthropogenic or weather interference. The visibility significance metric for
evaluating BART sources is the change in deciview (del-dv) from the source's and natural conditions haze
indices.").
53 WildEarth Guardians v. EPA, 770 F.3d 991, 944 (10th Cir. 2014).
81
-------
factors on a source-specific basis. Here, not only is the analysis at issue not being performed
under BART, but we did not give greater weight to our consideration of visibility improvement
within the cost factor, or consider the cost factor in a different fashion from the other three
reasonable progress factors.
Comment: [Luminant (0061) p. 68] Luminant Stated, more fundamentally, States are not
required to consider visibility benefit as a fifth factor at all, much less with regard to individual
controls at individual sources, as EPA's proposal asserts. EPA cannot disapprove a SIP revision
for "failing" to consider a factor that is not required by the Clean Air Act.487 Visibility benefit
from individual controls is not one of the four statutory factors that States must consider when
evaluating controls for reasonable progress.488 As EPA has explained, "[t]he final regional haze
rule clearly provides the States with the flexibility to establish a reasonable progress goal based
on its analysis of the statutory factors." 489 Thus, Texas was not required to "separately evaluate
the visibility benefit from the implementation of [] controls]" either at individual sources or for
source categories, as EPA claims. That Texas chose to review the costs of controls in relation to
projected benefits was purely discretionary on Texas' part and not a required element of the
analysis upon which EPA may base its disapproval.
Footnotes:
487 See supra note 405.
488 See 42 U.S.C. § 7491(g)(1).
489 Response to Petitions at 13.
Response: As we further discussed above, while we agree that the statute and regulations do not
include visibility benefit as a mandatory factor for the reasonable progress analysis, or that the
statue mandates any inclusion of visibility benefit be on an "individual control at individual
source" basis, we note that we do have authority to disapprove a SIP revision for failing to meet
the requirements of the CAA, which is the basis of our proposed disapproval and fully explained
there. As we have explained above, while the RHR does provide flexibility to the States, and to
the EPA, in the "four-factor analysis" to determine reasonable progress, when a State considers
visibility benefit in reasonable progress determinations, EPA's review of the State's
determinations must ensure the analysis is reasonable within the purpose and explicit national
goal established by Congress in CAA Section 169A. We did not compel Texas to consider
visibility benefit in its reasonable progress/RPG four-factor analysis; Texas made that
determination on its own.54 We reviewed Texas' reasonable progress four-factor analysis to
determine whether the State's determination provides for reasonable progress towards natural
visibility conditions, and proposed to find that its determination does not.55 As we note above,
regardless of the approach taken, the State must engage in some rational method for making this
assessment that complies with the requirements in the regional haze rule. As we further discuss
elsewhere, Texas' approach was highly flawed. Texas' approach effectively had the effect of
obscuring the cost-effective and available controls on those sources with the largest visibility
impacts, which other approaches would not have done, such as the individual source analysis we
chose to perform. This flaw, considered with the other flaws we identified in Texas' four-factor
54 79 FR 74837.
55 79 FR 74838, 74841, 74843, 74872.
82
-------
analysis, caused us to conclude that Texas' reasonable progress demonstration under Section
51.308(d)(l)(i)(A) was not approvable.
Comment: [Luminant (006) p. 106] Luminant Stated that neither the visibility impact of a
source nor the benefit from an individual emission control at any is one of the statutory factors
that States must consider in determining reasonable progress.1 Unlike in the BART context,
visibility improvement is not one of the four statutory factors that States must consider when
evaluating controls for reasonable progress and thus is not a lawful basis upon which EPA may
disapprove a State's plan.2 While EPA previously has acknowledged that States may consider
visibility improvement (or other factors) as an additional factor when making reasonable
progress determinations,3 such consideration is purely discretionary with the State. Certainly,
then, Texas was not required to conduct such modeling of individual controls to determine which
are "reasonable" or "feasible," nor was the absence of such modeling valid grounds for EPA to
disapprove. EPA's assessment of Texas' RPGs and Texas' decision that additional controls are
not reasonable during this planning period must be based on the statutory factors alone, as EPA
has previously recognized,4 and EPA may not disapprove Texas' submission on the basis that
Texas did not consider a non-statutory factor in the manner EPA would like.
Further, Luminant noted that EPA's methodology unlawfully elevates visibility benefits (a non-
statutory factor) above the statutory factors and uses it as a threshold factor to determine which
sources to evaluate under the four statutory factors. At most, visibility benefits are only modeled
or estimated by a State after it has determined what control measures or set of controls measures
are reasonable by weighing the four statutory factors. As EPA's guidance clearly explains,
States should estimate "the improvement in visibility that would result from implementations of
the control measures you have found to be reasonable and compare this to the uniform rate of
progress." 5 EPA's approach here for Texas is thus backwards and contrary to the statute, and
EPA admits there is no "prior precedent" for it.6 Here, EPA used visibility modeling to identify
and select the individual source controls for which it would analyze the costs and other statutory
factors.7 That is not a proper application of the four-factor. Thus, not only did EPA consider a
non-statutory factor in its analysis (and fault Texas for not doing so in the same manner as EPA
did), it elevated that non-statutory factor to the primary consideration in its analysis and used it
as the litmus test for which sources would be further regulated and which would not. EPA's
proposal here is contrary to the text and structure of the statute because it "isolates [the visibility]
benefit calculation and constrains authority Congress conferred on the States."8 Further, "EPA
[has] overstepped the bounds of its narrow statutory role in the SIP approval process" and acted
"ultra vires" by relying on a "factor[ ] which Congress has not intended [the EPA] to consider."
9
Footnotes:
1 79 Fed. Reg. at 74,874 (stating that EPA is "weighing the cost of compliance against the projected visibility
benefit," even though visibility benefit is not one of the statutory factors).
2 Compare 42 U.S.C. § 7491(g)(1) (four reasonable progress factors) with id. § 7491(g)(2) (BART factors
including "the degree in improvement in visibility which may reasonably be anticipated to result from the use of
such technology"); see also 40 C.F.R. § 51.308(d)(l)(i)(A) (four reasonable progress factors).
3 See North Dakota, 730 F.3d at 765.
4 77 Fed. Reg. 20,894, 20,934 (Apr. 6, 2012) (Under our regulations, we determine whether a State's rejection of
reasonable progress controls is reasonable based on the reasonable progress factors.").
83
-------
5 EPA Reasonable Progress Guidance at 203 (emphasis added).
6 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'n v. McCarthy, No. 11-01548, at 5 (D.D.C. 2014).
7 See 79 Fed. Reg. at 74877—78.
8 Am. Corn Growers Ass'n, 291 F.3d at 9. Indeed, Federal Land Managers have warned EPA against using
visibility benefit as a "fifth" factor in reasonable progress determination since EPA has no "statutory mandate to do
so." USDA Forest Serv., Recommendations for Improved Implementation of the Regional Haze Program 6 (May
2014).
9 Luminant Generation, 675 F.3d at 926 (alteration in original) (internal citations omitted) (internal quotations
omitted).
Response: We disagree with this comment, as further discussed above, as we believe our
proposal is in conformance with law and Congressional intent, is based on the "four-factor
analysis," and does not treat visibility benefit as a factor in the reasonable progress analysis, but
as a consideration within the cost factor,. As we further discussed above, we agree that visibility
impact of a source or the visibility benefit from an individual source is not a mandatory factor
States must consider in determining reasonable progress, or that any inclusion of such must
always be on an individual source basis. However, visibility is relevant to the purpose of the
statutory subsections and corresponding regulations at issue—assuring reasonable progress
towards the national goal of natural visibility conditions.
As we have discussed more fully above, while visibility or visibility benefit is not an explicitly
listed factor, we believe consideration of visibility within reasonable progress is a reasonable
interpretation of the statute in at least two ways, including when determining scope of the
analysis and what cost-effective controls should be required for reasonable progress. Also, as we
have discussed above, while States have discretion in evaluating reasonable progress, when a
State considers visibility improvement in evaluating control options, as Texas did here, that
consideration must still be reasonable in light of the explicit goals established by Congress in
CAA Section 169A. The State's discretionary inclusion of other appropriate considerations into
the reasonable progress four-factor analysis can only be approvable if the resulting analysis does
not subvert the overarching goal and intent of the statutory requirement—otherwise, undertaking
analysis of the mandatory factors could be rendered mere lip service to the statutory
requirements. As we discussed above, we have considered the Federal Land Managers
recommendation regarding visibility benefit, and still find that while the statue does not mandate
consideration of visibility benefit as one of the four factors listed in Section 169(g), we, and the
State, can consider visibility benefit if we, or the State, finds it relevant to achieving the statutory
goal of reasonable progress towards natural visibility conditions.
We agree that Texas was not required by the four-factor analysis to consider visibility or conduct
visibility modeling of individual controls to determine which are "reasonable" or "feasible" for
reasonable progress. We did not compel Texas to consider visibility impact and visibility benefit
in its reasonable progress and RPG four-factor analysis; Texas made that determination on its
own, both screening out sources prior to the four-factor analysis based in part on a Q/d analysis,
and comparing costs of controls to visibility benefit within the four-factor analysis.56 We
reviewed Texas' reasonable progress and RPG four-factor analysis to determine whether the
State's determination provides for reasonable progress towards natural visibility conditions, and
56 79 FR 74835, 74837.
84
-------
proposed to find that its determination does not.57 As we further discussed above, Texas'
approach was highly flawed. Texas' approach effectively had the effect of obscuring the cost-
effective and available controls on those sources with the largest visibility impacts, which other
approaches would not have done, such as the individual source analysis we chose to perform.
This flaw, considered with the other flaws we identified in Texas' four-factor analysis, caused us
to conclude that Texas' reasonable progress demonstration under Section 51.308(d)(l)(i)(A) was
not approvable. In sum, we did not disapprove Texas' analysis because it did not undertake the
same analysis we later determined was appropriate in our own FIP proposal; we proposed
disapproval because Texas' analysis, given all flexibility we recognize as appropriate within the
four-factor analysis, did not meet the requirements and intent of the statute.
Luminant's assertion that reasonable progress "must be based on the statutory factors alone, as
EPA has previously recognized," with citation to our North Dakota RH action is also
incompletely reproduced and taken out of context. The full quote is actually:58
As we have noted, our regulations require consideration of four factors in
reasonable progress determinations; visibility improvement is not one of the
specified factors. As we have indicated, when a state considers visibility
improvement as an additional factor in evaluating single-source control options,
that consideration must be reasonable in light of the explicit goals established by
Congress in CAA section 169A.
Thus, our statement in our North Dakota action actually supports the consideration of visibility.
We disagree with Luminant's interpretation of Sam Coleman's declaration. While we did say that
there was no prior precedent, this was in regards to the particular type of modeling undertaken;
Luminant takes Coleman's Statement out of context. Coleman's Statement is related to the
additional modeling we determined was appropriate due to the large distances involved and the
large number of sources being analyzed, which was a unique set of facts not encountered by us in
the Regional Haze context before. Luminant conflates this context and over broadens the scope
of Coleman's Statement beyond the modeling to our reasonable progress analysis overall, and
visibility in particular. Coleman's Statement was not stating there was no prior precedent for our
four-factor analysis, nor was it stating there was no prior precedent for consideration of visibility
or visibility benefit.
Our analysis neither treats the four factors differently from each other, nor elevates visibility
above the four statutory factors. We do not believe the case cited by Luminant, American Corn
Grower, is applicable to our proposal for the reasons discussed elsewhere, as the analysis at issue
is reasonable progress, not BART, and our analysis neither treats the statutory factors differently
from each other, nor elevates visibility above the four factors. Our proposal analyzed all four
factors in evaluating reasonable progress/RPGs in accordance with the statute. Our proposal did
not give greater weight to one factor over another; rather our proposal considered all factors
fully, revealing that the cost factor, including the consideration of visibility benefit, was the most
57 79 FR 74838, 74841, 74843, 74872.
58 77 FR 20934.
85
-------
determinative factor, i.e. had the most limiting effect, on the final determination.59 Thorough
analysis revealed that some factors had little to no limiting effect on what controls were available
for reasonable progress, while others did have greater effect, and therefore were more
determinative. This certainly does not mean consideration of visibility improvement was
weighted, or given main or primary status. As further discussed above, our proposal included a
reasonable interpretation of the statute, which included visibility in our analysis in two ways: 1)
when evaluating to what degree sources/source categories contributed to visibility impacts in
Class I areas; and 2) within our analysis of the four statutory factors, specifically comparing the
costs of compliance against the projected visibility benefit in Class I areas. As we have
discussed elsewhere, the consideration of visibility impact and visibility benefit in the two ways
our proposal included them is reasonable given the facts at hand and consistent with previous
statements in other State regional haze actions, further underscoring that we disagree that this
proposal was backwards from previous statements, guidance, or law. In fact, Texas also
screened out sources through a Q/d like analysis, and considered visibility benefit within their
four-factor analysis, thereby directly contradicting Luminant's assertion that visibility benefit
can and has only been estimated after determination of the four-factor analysis is complete.
Luminant's citation to us allegedly directly contradicting our guidance to support Luminant's
assertion that our approach is backwards and contradictory ignores the following paragraph. The
cited approach that ends in modeling visibility improvement of controls found reasonable, to
compare to the uniform rate of progress, is specifically stated in the guidance as one potential
approach. The next potential approach outlined in the guidance begins with dispersion modeling
to estimate visibility impacts prior to the four-factor analysis, described in the document as a
"back out" approach. The full context of the guidance document supports our approach as
consistent and reasonable.
As discussed further elsewhere, Luminant's comparison of reasonable progress and BART
regarding visibility in each respective analysis ignores the context of the statute at issue, which
disregards a key canon of statutory interpretation. As further elaborated on above, Sections
169(A) and 169(B) require reasonable progress towards the national goal of visibility in Class I
Federal areas. Visibility is inherently included overarchingly within the statutory text and intent,
and the statute's mandate to establish regulatory criteria for measuring reasonable progress. Our
interpretation avoids potentially absurd, in the case of consideration of visibility within scoping
of the analysis, or overly burdensome, in the case of consideration of visibility benefits within
the cost factor, results, as discussed further above, and is reasonable within the statue's and our
regulations' purpose and overarching scheme. Also discussed further above, while differences
between reasonable progress and BART are not irrelevant, we have consistently noted that there
is substantial overlap in the statutory and regulatory requirements applicable to BART and
reasonable progress—the ultimate purpose of requiring controls for both types of sources is to
make progress toward the national goal of eliminating man-made visibility impairment.
Therefore, we believe it is appropriate for analyses of potential controls for reasonable progress
sources to resemble BART analyses in many respects.
As discussed further elsewhere, we disagree with Luminant's application of Luminant
Generation Co. LLC v. EPA to our proposal. The court in that case found that we had stepped
59 79 FR 74874; FIP TSD, at 8, 12-15
86
-------
outside of our role in the SIP review process because the court found that we had based our
disapproval on whether the regulation at issue was in conformance with State law, rather than if
it was in conformance with the CAA. As we have explained above, we believe our interpretation
of reasonable progress towards the national goal of visibility improvement, including
consideration of visibility within the reasonable progress analysis, is in conformance with the
statutory language and Congressional intent of the CAA.
Comment: [Associations (0059) p. 16-17]
The Associations stated that the EPA's proposal is also unlawful because it would require States
to include visibility benefits as a mandatory, if not preeminent, factor in setting reasonable
progress goals. Under the Clean Air Act, States are required to consider the four statutory
factors in setting reasonable progress goals: "the costs of compliance, the time necessary for
compliance, the energy and nonair quality environmental impacts of compliance, and the
remaining useful life of any existing source subject to such requirements." 42 U.S.C. §
7491(g)(1). Visibility benefits are not included among the statutory factors. As EPA has
previously explained, "[t]he final regional haze rule clearly provides States with the flexibility to
establish a reasonable progress goal based on its analysis of the statutory factors." EPA,
Response to Petition for Reconsideration of Regional Haze Rule 13 (Jan. 10, 2001). Federal
land managers have concurred in this conclusion and have urged EPA not to include visibility
impacts as a mandatory fifth factor in state reasonable progress goals because EPA has no "clear
statutory mandate to do so." See U.S. Forest Service, Recommendations for Improved
Implementation of the Regional Haze Program 5 (May 2014). In this respect, the State's
obligations to establish reasonable progress goals are substantially different from those for
BART, where visibility benefits play an important role. See 42 U.S.C. § 7491(g)(2) (including
"the degree of improvement in visibility which may reasonably be anticipated to result from the
use of such technology" as a mandatory factor in BART determinations). This difference
reflects a clear Congressional intent that States cannot be compelled to include visibility benefits
in determining reasonable progress goals and certainly cannot be compelled to consider them in
the manner that EPA would require in this proposal.
The Associations noted that, nevertheless, EPA proposes to disapprove Texas' reasonable
progress goals based on a failure to consider visibility benefits alongside the required statutory
factors. See, e.g., 79 Fed. Reg. at 74,839 (asserting that Texas should have "separately
evaluate[d] the visibility benefit from implementation of [emission] controls]"); id. at 74,838
(asserting that "individual benefits were masked by the inclusion of those controls with little
visibility benefit that only served to increase the total cost figure"). Simply put, visibility
benefits are not mandatory statutory factors that States are compelled to consider when
establishing reasonable progress goals. As a result, it is unlawful for EPA to disapprove a State's
reasonable progress goals based on a failure to evaluate visibility benefits.
The Associations stated that the EPA's FIP proposal exacerbates the unlawful nature of EPA's
action by elevating visibility benefits above the four statutory factors. In applying its individual
source-based approach, EPA relies on visibility benefits as a threshold test to determine which
individual sources it will review for costs—a statutorily required factor. An approach that
87
-------
focuses first on visibility benefits will unnecessarily constrain States' ability to use their
discretion to apply the four statutory factors and establish reasonable progress goals as intended
by Congress. In fact, even in the context of BART determinations, where visibility benefits must
be considered, courts have rejected EPA's attempts to elevate visibility above the other statutory
factors. See American Corn Growers Ass'n v. EPA, 291 F.3d 1, 6-7 (D.C. Cir. 2002) ("The
Haze Rule's splitting of the statutory factors is consistent with neither the text nor the structure
of the statute."). There the court found that EPA's regulation was unlawful because it "isolate[d]
[the visibility] benefit calculation and constrained] authority Congress conferred on the States."
Id. at 8-9.
The Associations further stated that the EPA's undue reliance on visibility benefits in its
proposed disapproval of Texas' reasonable progress goals and proposed FIP is unlawful and
raises serious concerns regarding the implementation of reasonable progress goals. By elevating
visibility benefits to a primary, threshold role in establishing reasonable progress goals, EPA
would distort the statutory analysis envisioned by Congress and, contrary to cooperative
Federalism principles, would unnecessarily constrain States' ability to use their discretion to
consider the four reasonable progress factors that are mandated by Congress. Moreover, if EPA
is permitted to disapprove of State reasonable progress goals on the basis of this non-statutory
factor, States and regulated entities would face the risk of becoming subject to significant—and
potentially disproportionate—emission control costs if EPA perceives that such emission
controls would confer some miniscule visibility benefit.
Response: We disagree with this comment, as further discussed above, as we believe our
proposal is in conformance with law and Congressional intent, is based on the "four-factor
analysis," and does not treat visibility benefit as a "preeminent" or primary factor in our analysis
but rather as a consideration within the cost factor. As we further discussed above, we agree that
the statute and regulations do not include visibility benefit as a mandatory factor for reasonable
progress, or that any inclusion of such be on an individual source basis. We note that we do have
authority to disapprove a SIP revision for failing to meet the requirements of the CAA, as
discussed further above, which is the basis of our disapproval. Also, as we have discussed
above, while States have flexibility in evaluating reasonable progress, when a State considers
visibility improvement in evaluating control options, as Texas did here, that consideration must
still be reasonable in light of the explicit goals established by Congress in CAA Section 169A.
As discussed further above, while visibility is not an explicitly listed factor to consider when
determining whether additional controls are reasonable, visibility is relevant to the purpose of the
statutory subsections and corresponding regulations at issue—assuring reasonable progress
towards the national goal of natural visibility conditions. As we have discussed more fully
above, we believe consideration of visibility benefit within reasonable progress is a reasonable
interpretation of the statute, including when determining what controls should be required for
reasonable progress. As we discussed above, we have considered the Federal Land Managers60
60 The Commenters cite to one FLM document. The National Park Services comments do support the use of Q/d
and state: "We agree with EPA that' ... based on their visibility impacts, a smaller subset of the facilities
that we have initially analyzed should be further evaluated to determine ... if cost-effective
controls are available ...
88
-------
recommendation regarding visibility benefit, and still find that while the statute does not mandate
consideration of visibility benefit as one of the four factors listed in Section 169(g), we, and the
State, can consider visibility benefit if we, or the State, finds it relevant to achieving the statutory
goal of reasonable progress. We also note that the Forest Service commented in regards to our
proposal that, despite the concern referenced by Associations, it finds "the methodology and
metrics that EPA used are the most comprehensive seen to date for any SIP/FIP in the country
that we have reviewed, and should serve as a model for future efforts to consider the contribution
and/or potential benefits of individual sources to visibility."
We did not compel Texas to consider visibility benefit in its RPG four-factor analysis; Texas
made that determination on its own, both considering visibility in its Q/d analysis and in its
comparison of costs of controls to visibility benefit.61 We agree with Texas that, while visibility
is not an explicitly listed factor to consider when determining whether additional controls are
reasonable, the purpose of the four-factor analysis is to determine what degree of progress
toward natural visibility conditions is reasonable.62 We reviewed Texas' reasonable
progress/RPG four-factor analysis to determine whether the State's reasonable progress/RPG
determination provides for reasonable progress towards natural visibility conditions, and
proposed to find that its determination does not.63 As we further discuss above, Texas' approach
was highly flawed. Texas' approach effectively had the effect of obscuring the cost-effective
and available controls on those sources with the largest visibility impacts, which other
approaches would not have done, such as the individual source analysis EPA chose to perform.
This flaw, considered with the other flaws we identified in Texas' four-factor analysis, caused us
to conclude that Texas' reasonable progress demonstration under Section 51.308(d)(l)(i)(A) was
not approvable. In sum, we did not disapprove Texas' analysis because it did not undertake the
same analysis we later determined was appropriate in our own FIP proposal; we proposed
disapproval because Texas' analysis, given all flexibility we recognize as appropriate within the
four-factor analysis, did not meet the requirements and intent of the statute. The basis for our
disapproval and authority to do so is fully explained in our proposal and supporting documents,
final action, and elsewhere in this response to comments document.
As we discussed further above, our proposal analyzed all four statutory factors in evaluating
reasonable progress/RPGs in accordance with the statute, regulations, and our guidance.64 As
further discussed above, our proposal included a reasonable interpretation of the statute, which
included visibility in our analysis in two ways: 1) when evaluating to what degree sources/source
categories contributed to visibility impacts in Class I areas; and 2) within our analysis of the four
statutory factors, specifically comparing the costs of compliance against the projected visibility
benefit in Class I areas. Our proposal did not give greater weight to one factor over another;
USDA Forest Service states: "In summary, while the USDA Forest Service has expressed concern to EPA that the
use of visibility as a factor to be considered within the reasonable progress context may be outside the statutory
framework established for RP (see Clean Air Act, Section 169A (g)(1)), the methodology and metrics that EPA used
are the most comprehensive seen to date for any SIP/FIP in the country that we have reviewed, and should serve as a
model for future efforts to consider the contribution and/or potential benefits of individual sources to visibility."
61 79 FR 74837.
62 79 FR 74838, 74840; TX TSD, at 18.
63 79 FR 74838, 74841, 74843, 74872.
64 79 FR 74872-77, 74883, 74886; FIP TSD.
89
-------
rather our proposal considered all four factors fully, revealing that the cost factor, including
consideration of visibility benefit, was the most determinative factor, i.e. had the most limiting
effect, on the final determination.65 Thorough analysis revealed that some factors had little to no
limiting effect on what controls were available for reasonable progress, while others did have
greater effect, and therefore were more determinative. This certainly does not mean visibility
improvement was weighted, or given main or primary status.
The Associations argue that Congress did not intend reasonable progress to include visibility as a
factor for consideration, pointing out a difference between those factors and the inclusion of
visibility in the listed factors for BART. As discussed further above, we did not include
visibility as a factor. Furthermore, we believe the Associations Statement ignores the context of
the statute that these terms are used in, which disregards a key canon of statutory interpretation.
We do not believe the case cited by the Associations, American Corn Grower, is applicable to
our proposal for the reasons discussed in greater detail above, as the analysis at issue is
reasonable progress, not BART, and our analysis neither treats factors differently from each
other, nor elevates visibility above the four statutory factors.
Comment: [CCP (0075) p. 10] CCP Stated that visibility is not a specific statutory factor to
consider prior to the establishment of RPGs under CAA Section 169A. See 77 Fed. Reg.
20,894, 20,934 (Apr. 6, 2012) ("Nevada SIP Approval") ("As we have noted, our regulations
require consideration of four-factor in reasonable progress determinations; visibility
improvement is not one of the specified factors."). Nonetheless, EPA recommends looking at
visibility in order to determine whether RPGs "are reasonable." EPA's RPG guidance notes that
States need only look at "available measures for the sources and source categories that contribute
significantly to visibility impairment." See Guidance for Setting Reasonable Progress Goals
Under the Regional Haze Program, at 2-3 (June 1, 2007) ("RPG Guidance"). As discussed
above, EPA is statutorily constrained in how much weight it may give visibility and has accorded
it far too much weight in its proposed disapproval and FIP. However, even if EPA had properly
considered visibility as one component in the statutory cost analysis, as TCEQ did, it failed to
show that there will be appreciable visibility benefits. Because no benefit will be derived from a
mandate to install controls that will not improve visibility conditions, the proper focus of the
States and EPA is on the identification of sources that significantly impair visibility and of
controls that may significantly improve visibility conditions.
CCP Stated that visibility may be appropriately considered on a cumulative basis, as TCEQ did
for all sources that are candidates for control. Using a cumulative approach, Texas appropriately
concluded there were insignificant cumulative visibility benefits, measured in deciviews, from
requiring additional controls. See 76 Fed. Reg. 74,387 Table 10 (identifying estimated
deciview improvements ranging from 0.16 dv in Big Bend to 0.36 dv in Wichita Mountains).
EPA previously rejected similar cumulative visibility improvements of 0.254 dv and 0.273 dv in
the New York SIP as too "small" to justify controls. 77 Fed. Reg. 24,818.
Response: We address CCP's allegation of inconsistency with Nevada (actually North Dakota)
in a separate response. As we discussed further above, our proposal analyzed all four statutory
65 79 FR 74874; FIP TSD, at 8, 12-15
90
-------
factors in evaluating reasonable progress/RPGs in accordance with the statute, regulations, and
our guidance.66 The portion of our guidance CCP quotes is a portion of step 2 of the suggested
approach for State's setting RPGs. In full, step 2 states: "Identify the control measures and
associated emissions reductions that are expected to result from compliance with existing
rules and other available measures for the sources and source categories that contribute
significantly to visibility impairment"67 (emphasis added). Step 3 then states "[djetermine what
additional control measures would be reasonable based on the statutory factors and other
reasonable factors for the sources and/or source categories you have identified."68 Step 1,
preceding these steps, States, "Identify the key pollutants and sources and/or source categories
that are contributing to visibility impairment." As further discussed above, our proposal
included a reasonable interpretation of the statute, and followed this interpretation, by including
visibility in our analysis in two ways: 1) when evaluating to what degree sources/source
categories contributed to visibility impacts in Class I areas; and 2) within our analysis of the four
statutory factors, specifically comparing the costs of compliance against the projected visibility
benefit in Class I areas. As discussed elsewhere, our proposal did not give greater weight to one
factor over another; rather our proposal considered all four factors fully, revealing that the cost
factor, including consideration of visibility benefit, was the most determinative factor, i.e. had
the most limiting effect, on the final determination, but that other factors had impact as well.69
We believe our proposal properly focuses on a reasonable interpretation of what controls (and
corresponding significant visibility improvement) are required for reasonable progress towards
the national goal of natural visibility conditions in Class I areas. We agree with the commenter
that the proper focus during the first implementation period in a State as geographically large and
source-numerous as Texas should be on the identification of sources or groups of sources that
significantly impair visibility, and controls on those sources that result in the largest visibility
improvement. As we discuss in a separate response to comment, we disagree with the
commenter and have demonstrated that the required controls are cost-effective and result in
significant visibility benefits towards the goal of reaching natural visibility conditions.
Comment: Texas properly conducted the four-factor analysis required by the CAA; there
is no statutory requirement to consider a "fifth" visibility factor. [GCLC (0063) p. 5-7]
GCLC Stated, as an initial matter, EPA does not have the statutory right to dictate "how" a State
analyzes the four-factor and, in this instance, has far overstepped its bounds in its questioning of
Texas' four-factor analysis. EPA itself has acknowledged, "[sjtates have considerable flexibility
in how they take these factors into consideration."21 There is no required emissions or visibility
target, but rather, the requirement that States analyze the four-factors; the statute therefore
requires that the EPA approve a State's reasonable progress goals so long as the required analysis
was performed. Therefore, under even a strict statutory standard, Texas has fully met this
statutory burden, performed the required analysis, 22 and considering the flexibility that the CAA
provides to the States (as recognized by EPA), EPA simply has no basis to deny its submission.
66 79 FR 74872-77, 74883, 74886; FIP TSD.
67 RGP Guidance, at 2-3.
68 RGP Guidance, at 2-3.
69 79 FR 74874; FIP TSD, at 8, 12-15.
91
-------
GCLC asserted that EPA has no real substantive complaints regarding Texas' analysis of the
four-statutory factors. Rather, EPA's denial was based on Texas' alleged failure to analyze a
"fifth" factor in its analysis- visibility- in a manner that EPA prefers. As Stated by EPA:
"While visibility is not an explicitly listed factor to consider when determining
whether additional controls are reasonable, the purpose of the four-factor analysis
is to determine what degree of progress toward natural visibility conditions is
reasonable. Therefore, we consider it appropriate to consider the projected
visibility benefit of the controls when determining if the controls are needed to
make reasonable progress."23
GCLC Stated that EPA's determination that visibility is an "appropriate" consideration is
completely without statutory basis. In fact, it contradicts the statutory language itself. The CAA
defines the four-factor to be reviewed regarding reasonable progress, which does not include
visibility. This is not an oversight or accidental, but rather, an intentional act of omission by the
Congress. Immediately following the criteria for determining "reasonable progress" are the
criteria for determining 'BART," which includes a fifth visibility factor. 24 If Congress intended
the States to consider visibility on par with the other factors, or to provide EPA the authority to
impose this consideration on the States as EPA attempts to do here, this fifth factor would have
been included in the reasonable progress criteria. Ultimately, "EPA overstepped the bounds of
its narrow statutory role in the SIP approval process" and acted "ultra vires" by relying on a
"factor[ ] which Congress has not intended [EPA] to consider. "25
According to GCLC, while EPA cannot require Texas to consider this fifth factor, Texas, in its
discretion, did consider visibility, and thus EPA's proposal is further in error. Regarding the
visibility analysis that Texas did conduct as part of its SIP, it is important to note that this was an
exercise of a purely discretionary matter. A right afforded to Texas under the flexibility of the
CAA, it does not provide EPA a basis to disapprove of a discretionary analysis that is required
by neither statute nor regulations.
GCLC noted, even if EPA did have the ability to impose a fifth "visibility factor," Texas' choice
of a 0.5 deciview ("dv") threshold as a benchmark for total visibility improvement was entirely
reasonable. For example, in recently reviewing and approving Idaho's reasonable progress goals,
EPA "independently evaluated whether there are reasonable control measures available for
sources located within Idaho's regulatory jurisdiction" and concluded that facilities with visibility
impacts of 0.5 dv or less at the nearest Class I area were "relatively small."26 Therefore, EPA
ultimately concluded in Idaho that additional controls for "reasonable progress purposes [were]
not reasonable at [that] time, because even though there [were] Cost-effective controls identified,
visibility improvement [was] anticipated to be relatively small."27
Footnotes:
21 Idaho SIP Approval Proposal, 77 Fed. Reg. at 30251.
22 See TX TSD at 55, stating that "[t]he CENRAP States' modeling, described in Section 8 of the Texas Regional
Haze SIP, was developed consistent with our guidance."
23 TX TSD at 22.
24 Compare language at 42 USC§ 7491(g):
92
-------
(1) in determining reasonable progress there shall be taken into consideration the costs of compliance, the time
necessary for compliance, and the energy and non-air quality environmental impacts of compliance, and the
remaining useful life of any existing source subject to such requirements;
(2) in determining best available retrofit technology the State (or the Administrator in determining emission
limitations which reflect such technology) shall take into consideration the costs of compliance, the energy and
nonair quality environmental impacts of compliance, any existing pollution control technology in use at the source,
the remaining useful life of the source, and the degree of improvement in visibility which may reasonably be
anticipated to result from the use of such technology;
25 See Luminant Generation Co. LLC v. EPA, 675 F.3d 917, 925, 926 (5th Cir. 2012) (internal quotations and
citations omitted).
26 Idaho SIP Approval Proposal, 77 Fed. Reg. at 30256.
27 Id.
Response: We note that we do have authority to disapprove a SIP revision for failing to meet
the requirements of the CAA, as discussed further above, which is what our proposed approval
did here. As discussed further above, while visibility is not an explicitly listed factor to consider
when determining whether additional controls are reasonable, visibility is relevant to the purpose
of the statutory subsections and corresponding regulations at issue—assuring reasonable progress
towards the national goal of natural visibility conditions. As we have discussed more fully
above, we believe consideration of visibility within reasonable progress is a reasonable
interpretation of the statute in at least two ways, including when determining what controls
should be required for reasonable progress. Also, as we have discussed above, while States have
flexibility in evaluating reasonable progress, when a State considers visibility improvement in
evaluating control options, as Texas determined to do here, that consideration must still be
reasonable in light of the explicit goals established by Congress in CAA Section 169A. We did
not require Texas to consider visibility; Texas made that determination on its own. Flexibility
does not give States the ability to prepare an analysis in such a way that it obscures reasonable
controls through misapplication of reasonable progress factors and considerations, rendering the
analysis merely a paperwork exercise. As we noted above, Texas' four-factor analysis was
flawed in multiple ways and not consistent with the purpose of the statute, and consequently not
reasonable in light of the goals of the regional haze program, thus subject to disapproval by us
for not meeting the requirements of the CAA.
GCLC argues that Congress did not intend reasonable progress to include visibility as a factor
for consideration, pointing out a difference between those factors and the inclusion of visibility
in the listed factors for BART. As discussed further above, we believe GCLC's comment
ignores the context of the statute that these terms are used in, which disregards a key canon of
statutory interpretation. We do not believe the case cited by the GCLC, Luminant, is applicable
to our proposal for the reasons discussed above. We address GCLC's allegation of inconsistency
with our Idaho action and the choice of a dv threshold in separate responses.
Comment: Selection of sources for reasonable progress analysis [NPS (0077) p. 2-3, 4-5]
The NPS agreed with EPA that"... based on their visibility impacts, a smaller subset of the
facilities that we have initially analyzed should be further evaluated to determine ... if cost-
effective controls are available ..."
93
-------
Response: We thank you for your comment and support for this portion of our proposal.
Comment: [NPS (0077) p. 3-4] The NPS agreed that"... the cost of compliance is the
dominant factor ..." in a reasonable progress four-factor analysis. However, we disagree with
EPA's decision to, "... consider visibility benefits in weighing the factors and to assist in its
consideration of the cost of compliance" and consider "their projected visibility benefits [in
determining] which, if any controls should be proposed." As EPA notes, "visibility is not an
explicitly listed factor to consider when determining whether additional controls are reasonable
..." It is clear from the statute that Congress did not intend that visibility be added as a pseudo-
fifth-factor in the reasonable progress analysis. The problem of regional haze is the contribution
of numerous emission sources, and any given one may be "insignificant," while their aggregate
impact is significant. By using the Q/d screening metric, EPA has already taken potential
visibility impacts (and benefits of control) into account. While we agree that it is appropriate to
evaluate the overall benefits of the control strategies selected by the four-factor analyses (as EPA
did), it is not proper to introduce visibility as a fifth-factor in the decision-making process once a
source has been selected for analysis. It is also not proper to create a fifth reasonable progress
factor to act solely as an "off-ramp" for sources that would otherwise be controlled.
[NPS (0077) p. 5] The NPS suggested that EPA should conduct four-factor reasonable progress
analyses for the 38 facilities identified by EPA for further evaluation in its Q/D analysis. While
we agree that it is appropriate to evaluate the overall benefits of the control strategies selected by
the four-factor analyses (as EPA did), it is clear from the statute that Congress did not intend that
visibility be added as a pseudo-fifth-factor in the reasonable progress analysis once a source has
been selected for analysis. If the application of the four statutory factors results in a conclusion
that controls are reasonable, those controls should be required.
Response: We disagree with the Park Service's criticism of our use of visibility in proposing
which sources to control as reasonable progress. As we discussed above, we believe our
reasonable progress analysis is in conformance with the statute, is consistent with Congressional
intent, and is a reasonable interpretation of the statute and regulations. As further discussed
above, our proposal included a reasonable interpretation of the statute, which is to include
visibility in our analysis in two ways: 1) when evaluating to what degree sources/source
categories contributed to visibility impacts in Class I areas; and 2) within our analysis of the four
statutory factors, specifically comparing the costs of compliance against the projected visibility
benefit in Class I areas.
As discussed further above, while visibility benefit is not an explicitly listed factor to consider
when determining whether additional controls are reasonable, visibility is relevant to the purpose
of the statutory subsections and corresponding regulations at issue—assuring reasonable progress
towards the national goal of natural visibility conditions. As further explained elsewhere, once a
universe of sources has been identified for analysis, we believe that States, and/or EPA, have the
option of considering the visibility improvement that will result from potential control options
when weighing the four statutory factors. Allowing consideration of visibility improvement is
appropriate for several reasons, including that 1) it aligns with Congress' national goal, which is
94
-------
to remedy existing impairment of visibility in Class I areas; 2) consideration of only the four-
factor in a vacuum could also lead to the unwieldy and overly burdensome determination that
most or all further controls (beyond those required by non-Regional Haze regulations and
BART/CSAPR) must be considered reasonable, or none at all, even within the first planning
period;70 3) our 2007 guidance permits States to consider visibility improvement when
conducting a four-factor analysis, and several States, including Texas,71 did so. Therefore, we
believe consideration of "visibility benefit," potential visibility improvement towards the
national goal, in the cost factor of the four-factor analysis for controls is also a reasonable
interpretation of the statute as a harmonious whole, resulting in a reasonable progress
determination that reasonably splits up controls over the multiple implementation phases into
manageable amounts, and focuses on those controls at sources that have the largest visibility
benefits. In fact, the Court of Appeals for the Eighth Circuit agrees that visibility improvement
can be a consideration in a State's reasonable progress determination 72
We agree with the Park Service that regional haze is caused by the contribution of numerous
emission sources. We also agree with the Park Service that some sources may have very small
visibility impacts, but the aggregate of all these small sources may have a significant visibility
impact. However, we disagree with the Park Service's subsequent use of that statement. While
there are undoubtedly thousands of sources within Texas that are insignificant or very small
contributors to regional haze, there are many sources that have relatively large visibility impacts.
In fact, the overall strategy we employed in our reasonable progress and long-term strategy
analysis was to identify the most significant sources that impact visibility, determine if cost-
effective controls were available, and balance the cost of those controls against their visibility
benefits. Adopting the approach implicit in the Park Service's comments would have led to a
"control everything" strategy during the first implementation phase (e.g., [consideration of
visibility improvement is] "an "off-ramp" for sources that would otherwise be controlled"; or a
"divide and exempt approach"). The logistical problems aside, obviously controlling any source
in Texas that could be retrofitted or upgraded with controls that we have found to be cost-
effective in other actions, regardless of their respective visibility benefits, would lead to an
unwieldy and overly burdensome result. As we explained above, our consideration of visibility
benefit allows the analysis to remain within the intent and purpose of the CAA, by focusing on
those controls that are a reasonable first step towards the national goal, within the first phase. As
these sources are controlled, other sources or source categories will be identified as the most
significant sources impacting visibility in future planning periods.
The Q/d analysis only considers emissions and distance and does not consider how
meteorology, chemistry or stack parameters influence the potential to impact visibility. We used
it as a way to initially identify the group of sources that could potentially impact visibility based
70 We do not consider such a strategy "reasonable in light of the explicit goals established by Congress in CAA
section 169A," as this interpretation could completely overly burden the first phase of the multi-phased approach
contemplated by the statute and corresponding regulations. We believe that such rigid application was clearly not
intended by Congress, in contrast to our interpretation that meets the plain meaning and statutory context of
reasonable progress. We therefore consider visibility benefit as a reasonable marker available within the four-factor
analysis when determining what was reasonable to control at this time and what was not, to fulfill the statutorily
required first step towards the ultimate goal of natural visibility conditions.
71 79 FR 74838, 74840; TX TSD, at 18.
72 North Dakota v. United States EPA, 730 F.3d 750, 766 (8th Cir. 2013).
95
-------
on their proximity to the Class I area and their emission level. As discussed in a separate
response to comment within the modeling section and in the FIP TSD, we determined it was
appropriate to use photochemical modeling to assess the visibility impact from those sources
identified by our Q/d analysis. After the initial Q/d analysis, we tagged emissions from 38
facilities in order to evaluate the potential of emissions from a facility to impact visibility. This
is a more refined approach than the initial Q/d analysis performed by both Texas and EPA
because unlike a Q/d analysis that only considers emissions and distance, this accounts for
emissions, location, stack parameters, meteorological conditions, and models both chemistry and
transport to the Class I areas. The results of this modeling indicated that a subset of the 38
facilities were the primary contributors to visibility impairment at each Class I area. Therefore,
we determined it was reasonable to eliminate some of the small impacting sources from a full
four-factor analysis for this planning period based on facility-level visibility impacts and
consideration of estimated unit level impacts, as described in detail in the FIP TSD and a
separate response to comment. As we discuss above, we also considered visibility benefit as a
reasonable marker within the four-factor analysis modeling cost-effective controls to determine
their visibility impacts.
Comment: [USDA Forest Service (0083) p. 2] While the USDA Forest Service has expressed
concern to EPA that the use of visibility as a factor to be considered within the reasonable
progress context may be outside the statutory framework established for reasonable progress (see
Clean Air Act, Section 169A (g)(1)), the methodology and metrics that EPA used are the most
comprehensive seen to date for any SIP/FIP in the country that we have reviewed, and should
serve as a model for future efforts to consider the contribution and/or potential benefits of
individual sources to visibility.
Response: We disagree with the Forest Service that our consideration of visibility is outside the
statutory framework of the CAA for reasonable progress. As we discuss previously, while
visibility benefit is not an explicitly listed factor when determining whether additional controls
are reasonable, consideration of visibility benefit within the cost factor properly focuses on a
reasonable interpretation of what controls (and corresponding significant visibility improvement)
are required for reasonable progress towards the national goal of natural visibility conditions in
Class I areas. As we have discussed more fully above, we believe consideration of visibility
within reasonable progress is a reasonable interpretation of the statute in at least two ways,
including when determining what controls should be required for reasonable progress. We thank
the Forest Service for its characterization of the thoroughness of the methodology and metrics we
used in our analysis.
Comment: [Nucor Steel (0058) p. 3] Nucor Steel Stated that EPA proposed rule and FIP also is
improperly based on a "visibility benefit" factor in determining reasonable progress, a factor that
is not one of the four-factor authorized under the Clean Air Act. See, 42 U.S.C. 7491(g)(1); 40
CFR §51.308(d)(l)(i)(A).
Response: As discussed further above, we note that, while visibility benefit is not an explicitly
listed factor to consider when determining whether additional controls are reasonable, visibility
96
-------
is relevant to the purpose of the statutory subsections and corresponding regulations at issue—
assuring reasonable progress towards the national goal of natural visibility conditions. As we
have discussed more fully above, we believe consideration of visibility benefit within reasonable
progress is a reasonable interpretation of the statute.
Comment: [AECT (0074) p. 4-5] According to AECT, EPA's proposed requirement that a small
number of Texas EGUs use additional and costly SO2 emissions controls for Texas to meet the
reasonable progress requirement is based on a factor— visibility improvement— that is not
required by the CAA or EPA's Regional Haze rules. In developing that proposed requirement,
to narrow down the number of emissions sources for which EPA might require additional SO2
emissions controls, EPA considered the visibility improvement that would be predicted to occur
if additional SO2 emissions controls were required for those sources, and it identified the sources,
all of which are EGUs, that it would further evaluate to determine if it would require additional
SO2 emissions control. EPA then determined the likely visibility improvements that might occur
if different types of additional SO2 emissions controls were used for each of those EGUs. For
each EGU for which EPA determined that a type of additional SO2 emissions control would
provide for a "reasonable" or "significant" visibility improvement,8 EPA is proposing to require
that the EGU use that type of additional SO2 emissions control.
AECT commented that visibility improvement is not just used as a factor on which EPA based
its proposed requirement that a small number of Texas EGUs have to use additional and costly
SO2 emissions controls for Texas to meet its reasonable progress requirement, visibility
improvement was the main factor that EPA used as its basis for that proposed requirement.
EPA's use of visibility improvement as a factor, much less as the main factor, in developing that
proposed requirement is not allowed under the CAA or EPA's Regional Haze rules. Both the
CAA and EPA's Regional Haze rules specify the factors that States must consider in determining
reasonable progress requirements, and visibility improvement is not one of those factors.
Congress' inclusion of visibility improvement as one of the factors to be used in determining
what constitutes BART in contrast to its non-inclusion of visibility improvement as one of the
factors to be used in determining reasonable progress requirements, clearly demonstrates that
Congress meant for visibility improvement to not be one of the factors that States must consider
in making reasonable progress determinations, and certainly not the main factor. Federal case
law supports that conclusion. Federal courts, including the U.S. Supreme Court, have held that it
should be assumed that Congress acted intentionally when it included particular language in one
section of a statute but omitted it in another section of the same statute.9 The similar non-
inclusion in EPA's Regional Haze rules of visibility improvement as a factor that States must
consider in determining reasonable progress requirements further demonstrates that EPA should
not have used visibility improvement as factor, much less the main factor, in evaluating Texas'
reasonable progress determination and in developing its proposed requirement that a small
number of Texas EGUs use additional and costly SO2 emissions controls for Texas to meet the
reasonable progress requirement.
AECT noted, moreover, even if it was appropriate for EPA to have used visibility improvement
as a factor in developing that proposed requirement, it would not be appropriate for EPA to base
that proposed requirement on whether the visibility improvement that would result from the
97
-------
additional SO2 emissions controls would be "reasonable" or "significant". That is because
neither of those words is used, much less defined, in the CAA Regional Haze provisions or
EPA's Regional Haze rules.
Therefore, AECT requested that EPA re-analyze Texas' reasonable progress determination by
considering Texas' evaluation of the four-factor specified in the CAA and EPA's Regional Haze
rules, and by considering Texas' evaluation of visibility improvement as part of its consideration
of the costs of compliance factor, rather than as a separate factor.
Footnotes:
8 79 Fed. Reg. 74884 (Dec. 16,2014)
9 White Stallion Energy Ctr., LLC v. EPA, 748 F.3d 1222, 1237 (D.C. Cir. 2014) and Barnhart v. Sigmon Coal
Co., 534 U.S. 438, 452, 151 L. Ed. 2d 908, 122 S. Ct. 941(2002), both citing Russello v. United States, 464 U.S.
16, 23, 104 S. Ct. 296, 78 L. Ed. 2d 17 (1983)
Response: Our methods and analysis are discussed fully in our proposal. As we discussed
further above, our proposal analyzed all four statutory factors in evaluating reasonable
progress/RPGs in accordance with the statute, regulations, and our guidance.73 As discussed
further above, while visibility benefit is not an explicitly listed factor to consider when
determining whether additional controls are reasonable, visibility is relevant to the purpose of the
statutory subsections and corresponding regulations at issue—assuring reasonable progress
towards the national goal of natural visibility conditions. As we have discussed more fully
above, we believe consideration of visibility within reasonable progress is a reasonable
interpretation of the statute and regulations in at least two ways.
As further discussed above, our proposed FIP follows our guidance and includes visibility in our
analysis in two ways: 1) when evaluating to what degree sources/source categories contributed to
visibility impacts in Class I areas; and 2) within our analysis of the four statutory factors,
specifically comparing the costs of compliance against the projected visibility benefit in Class I
areas. Our proposal did not give greater weight to one factor over another; rather our proposal
considered all four factors fully, revealing that the cost factor, including the consideration of
visibility benefit, was the most determinative factor, i.e. had the most limiting effect, on the final
determination.74 Thorough analysis revealed that some factors had little to no limiting effect on
what controls were available for reasonable progress, while other did have greater effect, and
therefore were more determinative. This certainly does not mean visibility improvement was
weighted, or given main or primary status. We believe our proposal properly focuses on a
reasonable interpretation of what controls are required for reasonable progress towards the
national goal of natural visibility conditions in Class I areas.
AECT argues that Congress did not intend reasonable progress to include visibility as a factor for
consideration, pointing out a difference between those factors and the inclusion of visibility in
the listed factors for BART. AECT also generally references certain case law, which states
"[where] Congress includes particular language in one section of a statute but omits it in another
section of the same Act, it is generally presumed that Congress acts intentionally and purposely
73 79 FR 74872-77, 74883, 74886; FIP TSD.
74 79 FR 74874; FIP TSD, at 8, 12-15
98
-------
in the disparate inclusion or exclusion."75 While we agree that the Court's Statement here is an
important component of statutory interpretation, the Court specifically says, "it is generally
presumed." As discussed further above, we believe AECT's comment ignores the context of the
statute that these terms are used in, which disregards a key canon of statutory interpretation. 76
As further elaborated on above, Sections 169(A) and 169(B) require reasonable progress towards
the national goal of visibility in Class I Federal areas. Visibility is inherently included
overarchingly within the statutory text and intent, and the statute's mandate to establish
regulatory criteria for measuring reasonable progress. Our interpretation avoids potentially
absurd results from eliminating the purpose of the statute entirely from the scoping of the
required analysis, as discussed further above, and is thus reasonable within the statue's and our
regulations' purpose and overarching scheme. Also discussed further above, while differences
between reasonable progress and BART are not irrelevant, we have consistently noted that there
is substantial overlap in the statutory and regulatory requirements applicable to BART and non-
BART sources—the ultimate purpose of requiring controls for both types of sources is to achieve
reasonable progress toward the national goal of eliminating man-made visibility impairment.
Therefore, we interpret the listed factors for each within the context of the statute as a whole.
AECT asserted that we should not have "based"77 requirement of additional SO2 emissions
controls for reasonable progress on whether they corresponded to "reasonable" or "significant"
visibility improvement. We note that use of the term "reasonable" occurs in our proposal either
in the context of agency statutory interpretation or as shorthand in determining what controls are
required for reasonable progress through the four-factor analysis, i.e. reasonable controls or
reasonable progress controls. We do not consider it appropriate to base our proposal or final
action on "unreasonable" visibility benefits, as we have an obligation for administrative
decisions to be reasonable, and thus we disagree with any suggestion that we are prohibited from
using the word reasonable according to its ordinary meaning in the context of administrative
decision-making. We also disagree that the word reasonable is extra-statutory, because it is part
of the statutory term "reasonable progress," see CAA Section 169A(g). To the extent we used
75 Russello v. United States, 464 U.S. 16, 23 (1983).
76 Furthermore, EPA believes that Congress explicitly listed visibility improvements as a required BART factor
because CAA 169A(b)(2)(A) otherwise could easily be interpreted to preclude consideration of visibility
improvements, given that it brings into the BART process any source (of a specified type and age) that emits any air
pollutant anticipated to cause or contribute to any visibility impairment in any Class I area. Moreover, the term
"best available" in "best available retrofit technology" does not necessarily indicate consideration of potential
visibility improvements, as it could refer to best control efficiency only. Congress therefore made clear its intention
by listing visibility improvements as a required factor for BART determinations, so that States would consider the
balance between visibility and the other four-factor listed for BART. However, the CAA provision requiring state
plans to provide for reasonable progress does not contain such absolute statements, and so Congress did not need to
counteract any such absolute Statements by explicitly listing visibility improvements as a factor that must be taken
into consideration. When compared to the RP analysis, no specific sources are delineated by the statute for
evaluation, meaning we, or the State during the multi-phased RP process need to determine which sources/source
categories undergo a more detailed analysis for reasonable controls for each implementation period. We conclude
that this nuance between the two programs reveals several reasons why the statute refers to visibility as the
overarching goal within the RP analysis statutory subsection while specifically lists visibility under the BART
subsection.
77 We note again that we disagree with this characterization and that our proposal, after undergoing the four-factor
analysis, determined that certain cost-effective controls that corresponded with significant visibility benefit were
reasonable progress towards the national goal of natural visibility conditions.
99
-------
the term "significant," we have done so according to its ordinary meaning and consistent with
our guidance and previous actions.78 We consider our use of it as reasonable when determining
the scope of sources or source categories to analyze in accordance with the statute, and when
determining which controls are necessary for reasonable progress within the first implementation
period in a State as geographically large and source numerous as Texas. Moreover, we consider
it appropriate to State the FIP will yield significant improvements in visibility, just as it is
appropriate to say the improvements will be meaningful, sizeable, not insignificant, or any other
fitting synonym. Finally, many Federal register actions, including actions upheld on judicial
review, reinforce our expectation that we have used the words that have concerned the
commenter in appropriate ways.
Comment: [Stamper (0068) p. 4] Although benefits to visibility are not a specific criteria of the
40 CFR 51.308(d)(1) "four-factor" analysis, the purpose of these measures is the timely
attainment of natural background visibility measures in the Class I areas affected by the air
emissions from sources within a State.
Response: We thank you for your comment, and agree regarding the purpose of the statutory
requirements.
Comment: [Earthjustice (0067) p.5] Congress required States and EPA to consider four-factor
in determining the pollution controls and other measures that define reasonable progress. 42
U.S.C. § 7491(g)(1). The statute does not list visibility improvement as a fifth factor in the
reasonable progress analysis. Id
[Earthjustice (0067) p.21]
30 As discussed previously, visibility is not a fifth factor in reasonable progress analyses.
Accordingly, visibility benefits may not be used to screen out reasonable progress controls.
Response: As discussed further above, we note that, while visibility benefit is not an explicitly
listed factor to consider when determining whether additional controls are reasonable, visibility
is relevant to the purpose of the statutory subsections and corresponding regulations at issue—
assuring reasonable progress towards the national goal of natural visibility conditions. As we
have discussed more fully above, we believe consideration of visibility benefit within reasonable
progress is a reasonable interpretation of the statute. As further explained elsewhere, once a
universe of sources has been identified for analysis, we believe that States, and/or EPA, have the
option of considering the visibility improvement that will result from potential control options
when weighing the four statutory factors. Allowing consideration of visibility improvement is
appropriate for several reasons, including that 1) it aligns with Congress' national goal, which is
to remedy existing impairment of visibility in Class I areas; 2) consideration of only the four
statutory factors in a vacuum could also lead to the unwieldy and overly burdensome
determination that most or all further controls (beyond those required by non-Regional Haze
regulations and BART/CSAPR) must be considered reasonable, or none at all, even within the
first planning period; 3) our 2007 guidance permits States to consider visibility improvement
78 See, e.g., RPG Guidance at 3-2.
100
-------
when conducting a four-factor analysis, and several States, including Texas,79 did so. Therefore,
we believe consideration of "visibility benefit," potential visibility improvement towards the
national goal, in the cost factor of the four-factor analysis for controls is also a reasonable
interpretation of the statute as a harmonious whole, resulting in a reasonable progress
determination that reasonably splits up controls over the multiple implementation phases into
manageable amounts, and focuses on those controls at sources that have the biggest visibility
benefits. In fact, the Court of Appeals for the Eighth Circuit agrees that visibility improvement
can be a consideration in a State's reasonable progress determination.80
5. Consultation between Oklahoma and Texas
Comment: Earthjustice et al., stated that the EPA properly proposes to disapprove Texas'
long-term strategy, in part because Texas failed to consult meaningfully with Oklahoma.
[Earthjustice (0067) p.27]
Earthjustice et al., stated that the EPA properly proposes to disapprove Texas' long-term
strategy, in part because Texas failed to consult meaningfully with Oklahoma. 79 Fed. Reg. at
74.854-57. The purpose of the regional haze program is to restore natural visibility at all Class I
areas in the nation. 42 U.S.C. § 7491(a)(1). As Congress recognized, visibility-impairing air
pollution often crosses States lines and out-of-State sources often cause visibility impairment at
Class I areas. See, e.g., id. § 7492(c). Accordingly, the Regional Haze rule requires States to
consult with each other and to implement a long-term strategy containing the emissions
limitations and control measures necessary to reduce visibility impairment at both in-State and
out-of-State Class I areas. 40 C.F.R. § 51.308(d)(l)(iv), (d)(3); see also 64 Fed. Reg. at 35,732
("Because haze is a regional problem, States are encouraged to work together to develop
acceptable approaches for addressing visibility problems to which they jointly contribute.").
Earthjustice et al., stated that Texas sources are the primary cause of visibility impairment at
Wichita Mountains Wilderness Area in Oklahoma. 79 Fed. Reg. at 74,856. In fact, Texas
sources cause more visibility impairment at Oklahoma's only Class I area than Oklahoma's own
sources. Id. When Texas developed its regional haze SIP, it "consulted" with Oklahoma and
other States with Class I areas impacted by Texas sources. But Texas did so in a cursory manner
and without providing Oklahoma and the other States the technical data necessary to identify the
worst sources of visibility impairment in Texas and potential pollution controls for these sources.
Id. Ultimately, Oklahoma informed Texas that Oklahoma's reasonable progress goals for
Wichita Mountains would assume no additional pollution reductions from Texas sources, even
though Wichita Mountains was far off the 2064 glide path and Texas sources were the primary
reason why Wichita Mountains was not on track to restore natural visibility by 2064. Id. at
74.855-56.
Earthjustice et al., stated that, as EPA correctly recognizes, Texas' "consultation" with
Oklahoma violated both the Regional Haze Rule's consultation provision and the Rule's
requirement that States adequately document the technical basis for the emission reductions
79 79 FR 74838, 74840; TX TSD, at 18.
80 North Dakota v. United States EPA, 730 F.3d 750, 766 (8th Cir. 2013).
101
-------
necessary to achieve reasonable progress in downwind States' Class I areas. Id. at 74, 829,
74,856, 74,861. A cursory consultation with another State based on inadequate data is not the
kind of consultation required by the Clean Air Act. EPA's conclusion that the regional haze
regulations require a meaningful consultation based on sufficient technical analyses is reasonable
and consistent with the visibility program's purposes. See id. at 74,828-30, 74,856, 74,861.
Earthjustice et al., stated that the EPA also properly disapproved Texas' long-term strategy
because Texas did not include the necessary control measures to obtain its share of the pollution
reductions needed for Wichita Mountains. 79 Fed. Reg. at 74, 829, 74,856-57, 74,861. As
discussed above, Texas sources are the largest contributors to visibility impairment at Wichita
Mountains, and "the impact from sources in Texas is several times greater than the impact from
Oklahoma's own sources." Id. at 74,823. Yet Texas did not require a single source to install any
pollution controls. EPA correctly explained that CENRAP's and Texas' technical analyses "did
not provide the information needed to evaluate the reasonableness of controls on those sources
with the largest potential to impact visibility at the Wichita Mountains." Id. at 74,857. For
example, although both Oklahoma and Texas knew generally that certain Texas coal plants have
large visibility impacts at Wichita Mountains, Texas' technical analysis did not provide
sufficient details on the visibility impacts of individual sources or cost-effective pollution
controls on these individual sources. Id. at 74,861.
Response: Please see our response to comments elsewhere regarding long-term strategy
consultation for our general agreement with commenter that the CAA requires consultation
between the two States and an exchange of sufficient technical analysis in order to ensure that
reasonable progress is achieved at Wichita Mountains.
Comment: EPA Arbitrarily Disapproves the Consultation between Oklahoma and Texas.
[Luminant (0061), p. 2, iii]
Luminant Stated that the regulations require that Texas' long-term strategy reflect the emission
reductions requested and agreed to by the other Central Regional Air Planning Association
("CENRAP") States. Texas fully met this obligation with respect to Oklahoma through a
cooperative and mutually agreeable process. EPA's unlawful disapproval of that consultation
would be the first time in history it has disapproved a State regional haze consultation.
[Luminant (0061), p. iii] Luminant Stated that the Clean Air Act's regional haze program
requires States to work cooperatively to develop State plans that achieve reasonable progress
toward the goal of improved visibility in national parks and other Federally protected areas
(called "Class I areas"). To comply, Texas worked with neighboring States over a multi-year
period to model and project haze impacts, review State emissions, and develop coordinated plans
to achieve reasonable progress.
According to Luminant, Texas and its neighbors, including Oklahoma, consulted on the emission
reductions that each would include in its plan to improve visibility in the Federal areas in each
State. These plans are working. As confirmed by recent monitoring data, Texas and its
neighbors have already achieved substantial progress in improving visibility, and, in fact,
visibility improvements have surpassed even the most aggressive projections and goals.
102
-------
Luminant asserted that yet EPA now brushes aside this cooperation among States and disregards
the substantial improvements that have been achieved. Instead of using the same objective
criteria and standards applied to every other State's regional haze plan, EPA inexplicably created
standards out of whole cloth to review the Texas and Oklahoma plans.
Response: Section 51.308(d)(3)(i) of the Regional Haze Rule requires that Texas must consult
with Oklahoma because it has emissions that are reasonably anticipated to cause visibility
impairment at Oklahoma's Wichita Mountains. Next, Texas must demonstrate that it has
included in its RH SIP submittal all measures necessary to obtain its share of the emission
reductions needed to meet the progress goal for Wichita Mountains. In addition, Texas must
document the technical basis upon which it is relying to determine its apportionment of
emissions reductions obligations necessary for achieving reasonable progress in Wichita
Mountains. While we expect that much of the consultation, apportionment demonstrations, and
technical documentation will be facilitated and developed by regional planning organizations
(RPO), we disagree with Commenters that participation alone in an RPO process (here
CENRAP) will always be enough to meet the requirements for consultation under the RHR.
The rule does not negate the requirement that a State have a complete and technically adequate
analysis so that what results from consultations is well informed.
We believe that the consultation process should start with an exchange of all appropriate
technical information so that States can "develop coordinated emissions strategies," and proceed
with the required consultation process on an informed basis. Properly informed downwind
States then can assess whether any additional upwind emissions reductions are necessary to
achieve reasonable progress at their Class I area. The RHR provides that States may meet this
requirement by relying on the technical analyses developed by the RPO and approved by all
State participants. Thus, States have the option of meeting this requirement by relying on the
four-factor analyses and associated technical documentation prepared by a regional planning
organization on behalf of its member States to the extent that such analyses and documentation
were conducted. On the other hand, CENRAP was not required, nor did it provide, state-specific
analyses and information on the cost-effectiveness and visibility benefits of potential control
strategies under consideration by each state to address the specific sources or groups of sources
within that state that have the largest visibility impacts. Rather, CENRAP provided more general
information on overall projected visibility conditions, potential controls and associated costs for
some sources and the potential benefit of regional emission reductions to inform the
development of potential control strategies that may require additional analysis.81 For example,
while the CENRAP analysis identified that impacts from EGUs in Texas were significant, it did
not provide a refined analysis to fully assess the cost-effectiveness and visibility benefits of
controlling those sources, including not providing information on the cost-effectiveness of
scrubber upgrades for those sources with existing, underperforming scrubbers. As Texas states
in its regional haze SIP, "While Texas participates in CENRAP and benefits from the technical
81 CENRAP conducted a control sensitivity analysis to evaluate the impact of point source emission reductions
across all CENRAP states given a maximum dollar per control level of $5,000/ton; however, the results "were
intended to be a starting point for control discussions that would require much greater refinement." Technical
Support Document for CENRAP Emissions and Air Quality Modeling to Support Regional Haze State
Implementation Plans, September 12, 2007 at 2-37).
103
-------
work coordinated by the RPO, Texas has sole responsibility and authority for the development
and content of its Regional Haze SIP."82 Therefore, participation in a RPO does not
automatically satisfy a State's obligation to "[demonstrate that it has included all measures
necessary to obtain its share of the emissions reductions needed to meet the progress goal" for a
Class I area. [74856],
Texas although participating in CENRAP retained the duty to do whatever additional analysis
was necessary to address fully the requirements of the RH rule for RPG and LTS. While the
LTS requirements say, you may rely on the RPO technical analysis that is true only to the extent
that it provides the necessary information. Any gaps in that analysis must be addressed by the
State. For a State that has little impact on Class I areas outside of the State, the gap is to evaluate
your sources impacts on your own Class I areas (RP analysis). For Texas, the gap existed not
only for the RP analysis for Texas Class I areas, but also for the LTS development for addressing
large impacts at the Wichita Mountains.
Recognizing that the information made available by CENRAP indicated the significant impact of
Texas emissions and potential for cost-effective controls, Texas used the CENRAP analysis as a
starting point, and performed supplemental analysis for both its reasonable progress and long-
term strategy demonstrations. In short, the control analysis performed by CENRAP was a
starting point for identifying reasonable controls and developing the LTS.
Texas therefore using the CENRAP analysis as a starting point, attempted to supplement that
analysis, for both its reasonable progress and long-term strategy demonstrations ("[t]he TCEQ
used the control strategy analysis as the starting point for the analysis of additional controls." Pg.
10-4 TCEQ SIP). [74857], However, the additional technical analysis performed by TCEQ was
flawed and therefore did not provide the type of information necessary to fully evaluate the
reasonableness of controls at Texas sources with the largest potential to impact visibility at the
Wichita Mountains.[ 74861], Allowing this lack of information to continue was a critical
misstep for ODEQ in setting its RPG and a critical misstep for TCEQ when determining its fair
share of emissions reductions under the LTS requirement. [74857],
As stated in our proposal, given the plain language of the CAA, this requires States to consider
the four-factor used in determining reasonable progress in developing the technical basis for both
their own Class I areas and downwind Class I areas. Such documentation is necessary so that
interstate consultations can proceed on an informed basis, and so that downwind States can
properly assess whether any additional upwind emissions reductions are necessary to achieve
reasonable progress at their Class I areas. Therefore, Texas had an obligation to provide
appropriate information to Oklahoma so it could establish a proper progress goal for the Wichita
Mountains. Further, Texas had an obligation to conduct an appropriate technical analysis, and
demonstrate through that analysis (required under (d)(3)(ii)), that it provided its fair share of
emissions reductions to Oklahoma. In summary, Texas was required through the RPG and LTS
consultations' processes to provide Oklahoma the information it needed to establish the RPG for
the Wichita Mountains, and it failed to do so. We address Oklahoma in the next response.
82 2009 Texas Regional Haze SIP at 3-1.
104
-------
As for comments on consideration of recent monitoring data, we address those comments
elsewhere. However, we emphasize here that the glidepath is not a safe-harbor. Favorable
monitored conditions will not necessarily correspond with permanent reductions or conditions
that can extrapolate forward to all future years; thus, the RHR requires examination of those
permanent and enforceable emissions limitations that are reasonable and cost-effective for the
first planning period. The feasibility and reasonableness of those controls should not be
obscured by a State's analytical approach. Reductions should instead be obtained at the earliest
opportunity with an eye toward new, additional candidates in future plan submissions and
revisions. The glidepath is not a yardstick to measure progress in any definitive sense, but rather
a source of guidance on the suite of controls that may invite closer examination. The outcome of
the Texas SIP was to not require new controls on larger sources that were not already and
otherwise required by CAA measures. Reasonable controls were not fairly identified even as the
technical record for the FIP firmly establishes those controls can be found and that they can and
should be required to achieve reasonable progress.
Comment: Texas' long-term strategy meets all statutory and regulatory requirements, and
EPA must approve it. Texas and Oklahoma fully met the consultation requirement, and
EPA has no authority to second-guess their regional agreement. [Luminant (0061) p. 77]
According to Luminant, Regional consultation among the States is central to the long-term
planning process, as provided in EPA's regulations. EPA's regulations specifically provide that,
to meet the long-term strategy requirement, "[i]f the State has participated in a regional planning
process, the State must ensure it has included all measures needed to achieve its apportionment
of emission reduction obligations agreed upon through that process,"526 Texas fully met this
obligation with respect to Oklahoma and all other CENRAP States. As EPA concedes,
"Oklahoma did not specifically request any additional reductions from Texas sources."527 Thus,
the "agreement" between Oklahoma and Texas is that no further reductions, beyond those
required by programs already in place as projected by the CENRAP modeling, would be
apportioned to Texas. EPA does not dispute this. Instead, it would look behind the agreement
reached by the two States. But there is no requirement or justification that the States agree to
any particular amount of reduction, beyond what they both consider to be reasonable.
Luminant Stated that nor is EPA's underlying and unfounded premise—that Oklahoma was
somehow deprived of data about Texas sources, their impacts, and the costs of controls528—even
remotely correct. In fact, even EPA asserts that during the Texas / Oklahoma consultations,
Oklahoma "had (1) abundant information showing the impact of Texas sources on visibility at
the Wichita Mountains . . . and (2) evidence that cost-effective controls on these sources were
likely available."529 And EPA further concedes that "the analyses developed by CENRAP [and
used by Texas and Oklahoma in their consultations] provide[d] a great deal of information on
contributions to visibility impairment and a set of potential add-on controls and cost associated
with those controls . . . "53° The truth is that Oklahoma possessed more than adequate
information about impacts and potential controls but correctly decided it was not reasonable to
request any further reductions from Texas sources during the first planning period. EPA may
disagree with that choice in hindsight and may wish Oklahoma's and Texas' agreement were
different, but that is an unlawful basis for disapproving Texas' long-term strategy.531
105
-------
Luminant asserted that EPA's analogy to a situation where two States disagree about how to
apportion impacts and reductions also does not support EPA's proposal.532 The situation here is
not one where EPA must step in to mediate a dispute between two States or where one State is
refusing to make reductions requested by another. There was complete agreement between
Oklahoma and Texas, and Texas' long-term strategy correctly reflects "all measures needed to
achieve its apportionment of emission reduction obligations agreed upon through [the regional
planning] process."533 And, as discussed elsewhere in these comments, those agreed-to
reductions have resulted in more progress than even EPA believes is reasonable by 2018.
Footnotes:
526 40 C.F.R. § 51.308(d)(3)(ii) (emphasis added).
527 79 Fed. Reg. at 74,856.
528 Id.
529 Id. at 74,867.
530 Id. at 74,872; see also id. at 74,867 (discussing "[c]ontrol cost data developed by Alpine Geophysics, and shared
by Oklahoma during consultations").
531 40 C.F.R. § 51.308(d)(3)(ii) (requiring only that a State's long-term strategy include "all measures needed to
achieve its apportionment of emission reduction obligations agreed upon through that [regional planning] process").
532 79 Fed. Reg. at 74,827, 74,872.
533 40 C.F.R. § 51.308(d)(3)(ii) (emphasis added).
Response: We describe above our basis for disapproving the Texas long-term strategy
consultation for Wichita Mountains. Here, we discuss our basis for disapproving the Oklahoma
RPG consultation with Texas for Wichita Mountains. As previously discussed in our clarified
interpretation of the RHR category, the four-factor analyses, reasonable progress goals, and long-
term strategies are inextricably linked. The consultation requirements in 40 CFR
51.308(d)(l)(iv) and 40 CFR 51.308(d)(3)(i) are two sides of the same coin. The consultation
record for Oklahoma revealed misunderstandings over its role in addressing visibility transport
issues.
While we agree with the commenter that Oklahoma possessed more than adequate information
from the CENRAP analyses about impacts from and potential controls for Texas sources, we do
not agree that it was reasonable for Oklahoma to stop at this point. Oklahoma had, based upon
the CENRAP analyses, (1) abundant information showing the impact of Texas sources on
visibility at the Wichita Mountains, particularly from EGU sources in northeast Texas, and (2)
evidence that cost-effective controls on these sources were likely available. Despite this
information, however, the ODEQ requested neither that the TCEQ further investigate controls at
these sources nor did it explicitly request Texas to obtain additional reductions from Texas
sources to address the impacts of emissions from these sources at the Wichita Mountains. The
Regional Haze Rule requires Oklahoma to use the consultation process under Section
51,308(d)(l)(iv) in the development of RPGs in tandem with Texas. Oklahoma failed to request
specifically that the TCEQ further investigate these sources. It also did not explicitly request
from Texas that Texas require additional reductions from Texas sources to ensure that all
reasonable measures to improve visibility were included in Texas' LTS and incorporated into
Oklahoma's RPG for the Wichita Mountains. Failing to do this resulted in the development of
an improper RPG for Wichita Mountains because it did not include the consideration of the
Texas impacts.
106
-------
We address comments concerning recent monitoring data in detail elsewhere.
Comment: [Luminant (0061) p. 79] Luminant noted that EPA's new approach of second-
guessing regional agreements—years after they are reached and implemented—would
undermine and chill the regional planning process, and discourage States from participating. In
issuing its regional haze regulations, EPA actively encouraged States "to work together" in
regional planning organizations "to develop a common technical basis and apportionment for
long-term strategies that could be approved by individual State participants . . . ,"539 Indeed,
Congress provided funding for the development of the RPOs, including CENRAP, and EPA was
"actively involved" in the establishment of the RPOs and "participate[d] early and actively in
regional planning efforts"540 Yet, despite its active involvement over the multi-year CENRAP
process, in which Texas' and Oklahoma's agreement was evident and well known, EPA never
raised any of the concerns it asserts today. And EPA never second-guessed the process or the
data that the States were developing—as it does now, years after that process has been completed
and on the eve of the next planning period. In truth, Texas and Oklahoma did exactly what EPA
encouraged them to do, and EPA has no legitimate basis to reject the States' resulting agreement.
Certainly, EPA's regulations do not permit it. And EPA cites no examples (and we have found
none) where EPA has similarly disapproved a regional agreement. There is no basis for EPA to
do so here, nor should EPA take such an unprecedented approach if it intends for States to
continue to work together in subsequent planning periods.
Footnotes:
539 64 Fed. Reg. at 35,732, 35,735. See also Response to Petitions at 7 ("The EPA further notes that in addition to
the general flexibility of the regional haze rule, EPA has consistently encouraged States and tribes to continue to
work together to better understand the regional haze problems in their respective regions and to develop effective
emission reduction strategies to address haze.").
540 Response to Petitions at 7 (internal quotations omitted).
Response: We disagree that this is a new approach on the consultation requirements nor that our
position undermines or chills the regional planning process. We agree that Texas and Oklahoma
worked together in the CENRAP RPO process. Nevertheless, throughout the consultations,
Oklahoma never explicitly asked Texas for reductions even though there was clear evidence
from the CENRAP analyses that Texas sources were impacting the Wichita Mountains and cost-
effective controls were likely available on some of these sources. Armed with this evidence,
Texas performed additional RPG/LTS technical analysis for the two Texas areas and Wichita
Mountains but it was flawed. While our regulations allow States to work together in RPOs, like
CENRAP, this is not a stopping point for States to fall back on as a rationale not to meet the Act
and RH rules. In our Clarified Interpretation of the RHR category, we discuss our interpretation
of the Act and the legally required role of these two States during their RPG and LTS
consultation with each other. We have not disapproved other States' RPG/LTS consultation
processes because the particular facts of Texas and Oklahoma did not arise. These new facts
required us to provide our clarification of the Act and the RHR in our proposal. We
acknowledge that the States were developing their SIPs simultaneously, which complicated their
ability to proceed in an orderly step-wised fashion, but this does not excuse the States from
having to satisfy the applicable requirements of the CAA and the Regional Haze Rule.
107
-------
With regard to the comment that we are only now raising concerns, the commenter is incorrect.
In our comment letter to Texas during its public comment period, we said that Texas should
update its Oklahoma consultation information. Specifically, we asked that Texas demonstrate it
has included in its implementation plan all measures necessary to obtain its share of the emission
reductions needed to meet the progress goals for the Class I areas it affects. We also said that
Texas should document its technical basis, including modeling, monitoring and emissions
information, on which it relies to determine its apportionment of emission reduction obligations
necessary for achieving reasonable progress in each area it affects. Furthermore, we said that the
Texas RPG/LTS proposed technical analysis raised concerns about whether it appropriately
evaluated whether there were additional reasonable controls available to help reduce its impact
on the Wichita Mountains in order to achieve progress at the Wichita Mountains.
Finally, in our 2008 comment letter to Texas, we said:
EPA has submitted these comments on the Texas draft Regional Haze State
Implementation Plan (RH SIP) with the intention of addressing the more
significant issues that could be identified considering the review time available.
Due to time, resource constraints, and the fact that that the TCEQ has elected not
to submit a paper copy of the SIP (which consists of approx. 50 separate
electronic files), it has not been possible to conduct a completely thorough review,
particularly with regard to modeling. It is possible that additional concerns, not
discovered during the review of this draft, will surface during the review of the
final version of this SIP.
We also sent a 2009 comment letter to Oklahoma during its public comment period stating it
does not appear that ODEQ actually requested reductions from Texas and we urged Oklahoma to
ensure Texas is aware of its sources' impact and encourage reductions as necessary. Again, as in
our comment letter to Texas, we said the same to Oklahoma that additional concerns will surface
during the review of the final SIP submittal.
Our job under the Act is to review a SIP submittal and determine if it meets the Act and rules,
regardless of whether we commented or not on a State's proposed SIP during its State
rulemaking process. There is no requirement in the Act that EPA must review, evaluate, and
comment on a State's proposed SIP revision.
Comment: Oklahoma's reasonable progress goal for Wichita Mountains meets all statutory
and regulatory requirements, and EPA must approve it. [Luminant (0061) p. 80]
Luminant Stated that EPA also has no legitimate basis for disapproving the reasonable progress
goal set by Oklahoma for the Wichita Mountains. As did Texas, Oklahoma relied on modeling
and cost data developed by CENRAP, consulted with neighboring States, and analyzed the four
statutory factors.541 EPA does not dispute Oklahoma's modeling or its analysis of the four
statutory factors. Instead, EPA cites as its basis for disapproving Oklahoma's RPG "an
incomplete consultation with Texas . . . ,"542 But, as discussed above, neither the statute nor
EPA's regulations establish any criteria for consultation that Oklahoma and Texas failed to meet,
108
-------
nor do they authorize EPA to judge what is and is not a "complete" consultation. The opposite is
true—the regulations only authorize EPA to consider a "disagreement" between States "in
determining whether the State's goal for visibility improvement provides for reasonable progress
towards natural visibility conditions."543 Here, Oklahoma and Texas were in agreement on the
goal and measures for the Wichita Mountains, and EPA thus has no authority to disapprove
Oklahoma's RPG.
Further, Luminant noted that even if EPA's disapproval of Oklahoma's RPG were authorized
and supported, that disapproval does not allow EPA to disapprove Texas' long-term strategy. As
discussed in Sections III. A and IV, Texas' obligation under the regulations is to include in its SIP
those measures necessary "to achieve its apportionment of emission reduction obligations agreed
upon through [the regional planning] process ,"544 And the benchmark for such apportionment
is "the reasonable progress goals established by States having mandatory Class I Federal
areas."545 Here, regardless of EPA's view of Oklahoma's RPG for Wichita Mountains, it is
undisputed that Texas' SIP includes the measures necessary to secure Texas' agreed-to
apportionment of emission reductions to meet the RPG for Wichita Mountains established by
Oklahoma, and thus Texas' SIP must be approved. Moreover, to the extent that Oklahoma's
RPG is adjusted by EPA's FIP or in some other manner, the matter is properly addressed in the
second planning period, not by disapproving Texas' long-term strategy. That is the same
approach EPA has taken as to other States, and there is no basis for treating Texas and Oklahoma
any differently.546
Footnotes:
541 Oklahoma TSD at 9.
542 Id. at 11.
543 40 C.F.R. § 51.308(d)(l)(iv).
544 Id. § 51.308(d)(3)(ii) (emphasis added).
545 Id. § 51.308(d)(3) (emphasis added).
546 See, e.g., 77 Fed. Reg. at 40,155-56 (explaining, in review of Nebraska long-term strategy, that the issue of
"whether other reasonable control measures are appropriate to ensure reasonable progress" would be considered
"during subsequent periodic progress reports and regional haze SIP revisions," and the downwind South Dakota
"may at that time consider asking Nebraska for additional emission reductions").
Response: See our previous responses on why we believe the CAA and RHR provide us with
the legal authority to disapprove Oklahoma and Texas for failure to meet the RPG/LTS
consultation requirements. As discussed in depth elsewhere, we have determined that Texas'
analysis is inadequate because it does not provide the information necessary to determine the
reasonableness of controls at those sources in Texas that significantly impact visibility at the
Wichita Mountains in Oklahoma, or the Texas Class I areas. Oklahoma and Texas discussed the
significant contribution of sources in Texas to visibility impairment at the Wichita Mountains
during the interstate consultation process required by the Regional Haze Rule. The results of the
CENRAP analysis demonstrated that Texas point sources, and in particular EGUs in northeast
Texas, have significant visibility impacts on the Wichita Mountains and that cost-effective
controls were potentially available for some of these sources. However, Oklahoma did not
pursue the point in its consultations with Texas under Section 51.308(d)(l)(iv). Oklahoma did
not have adequate information to establish its reasonable progress goal for the Wichita
Mountains, and should have requested that the TCEQ further investigate these sources or
requested additional reductions from Texas sources to ensure that all reasonable measures to
109
-------
improve visibility were included in Texas' long-term strategy and incorporated into Oklahoma's
reasonable progress goals for the Wichita Mountains. Furthermore, because of the flawed
consultations with Texas, Oklahoma did not consider the emission reduction measures necessary
to achieve the uniform rate of progress for the Wichita Mountains and did not adequately
demonstrate that the reasonable progress goals it established were reasonable based on the four
statutory factors under 51.308(d)(l)(ii). 83
We disagree that disapproval of Oklahoma's RPG for the Wichita Mountains does not allow
EPA to disapprove Texas' long-term strategy. Our disapproval of Oklahoma's reasonable
progress goal for Wichita Mountains has nothing to do with our disapproval of Texas' LTS. We
are disapproving the Texas LTS because the analysis underlying it is technically flawed.
Because of these flaws, Texas' SIP submittal does not include all the measures necessary to
secure its apportionment of the emission reductions needed to meet the progress goal that should
account for all reasonable control measures for the Wichita Mountains, or its own Class I areas.
We are disapproving the Oklahoma RPG for the Wichita Mountains not because of the
technically flawed Texas LTS but because Oklahoma's consultations with Texas were flawed
which prevented it from adequately developing its reasonable progress goals for the Wichita
Mountains. Because Oklahoma's consultations with Texas were flawed, Oklahoma did not
adequately consider the emission reduction measures necessary to achieve the uniform rate of
progress for the Wichita Mountains and did not adequately demonstrate that the reasonable
progress goals it established were reasonable based on the four statutory factors.
See our responses on our evaluation of Texas long-term strategy elsewhere. We address
Luminant's allegation that we are treating Texas differently in relation to our Nebraska and
South Dakota actions in the consistency section of this document.
Comment: The Proposed Rule arbitrarily disapproves of Texas and Oklahoma's
consultation efforts. [CCP (0075) p. 14-15]
CCP stated that, under the regional haze program, a State "must consult with those States which
may reasonably be anticipated to cause or contribute to visibility impairment in the mandatory
Class I Federal area." 40 C.F.R. § 51.308(d)(l)(iv); see also 40 C.F.R. §§ 51.308(d)(3)(i)-(iii)
(outlining consultation requirements for the long-term strategy). Interstate consultation requires
that States work together and "take a hard look at what measures are necessary" to meet RPGs.
79 Fed. Reg. 74,818.
CCP stated that the following factors support Texas and Oklahoma's consultation process:
• Both States were active participants in the Central Regional Air Planning Association
("CENRAP") regional planning process, and separately coordinated with each other
based on CENRAP modeling of visibility impacts between the States.
• Texas and Oklahoma met numerous times, held various phone calls, and exchanged
correspondence regularly on the impacts of each State's sources on Class I areas. See
79 Fed. Reg. 74,865-67 (reviewing consultation efforts); Texas SIP App. 4-2.
83 79 FR 74871, 74872.
110
-------
• EPA recognized that Texas went above and beyond the regional planning process for
interstate consultation purposes. Id. at 74,857 ("In fact, the TCEQ went beyond the
CENRAP analysis by contemplating additional controls, applying a lower Cost-
effective ness threshold and estimating the visibility benefit from the identified
control set.").
• As part of its consultation discussions, Texas will allow Oklahoma to comment on
Texas' evaluation of Best Available Control Technology for new and modified
sources. See 79 Fed. Reg. 74,866.
Nonetheless, CCP stated that the EPA arbitrarily disapproved of the consultation process
between Oklahoma and Texas without any reference to its rules, guidance and prior SIP
approvals. The Proposed Rule never details what information Oklahoma lacked in establishing
its RPGs, and EPA must provide a more adequate explanation of how additional information
would have changed Oklahoma's ultimate determination that additional controls on Texas
sources would not move the Wichita Mountains perceptibly closer to its regional haze goals.
Response: See our previous responses. Moreover, we disagree that we did not provide what
information Oklahoma lacked in establishing its RPG for Wichita Mountains. Texas in its
flawed analysis tried to provide the lacking information. Our analysis provides the lacking
information. We also disagree that Oklahoma ultimately determined no additional controls on
Texas sources were necessary.
Finally, the commenter mischaracterizes the consultation discussions with Oklahoma. On
multiple occasions, Oklahoma indicated that Texas' sources had outsized impacts on visibility at
Wichita Mountains, and that Oklahoma could not meet the glidepath without emission reductions
from Texas. Furthermore, Oklahoma indicated in their response to comments that it did not
believe it had the authority to require those reductions, but instead had to rely on Texas or the
EPA. Oklahoma should not have hesitated to ask Texas outright for additional emission
reductions because the evidence available during the consultations clearly suggested that cost-
effective controls were available and impacted the Wichita Mountains.
Comment: [Associations (0059) p. 10-11] The Associations Stated that Oklahoma and
Texas fully complied with EPA's consultation requirements for cross-State emissions
through participation in a regional planning process. Based on encouragement from EPA,
Texas and Oklahoma worked with nearby States to establish CENRAP to assist the States with
technical issues associated with their regional haze SIPs. In particular, CENRAP assisted the
States in developing emissions inventories and modeling, including models for a 2002 base case
for visibility in Class I areas, projections for 2018 emissions and visibility, estimates of natural
conditions, and cost/benefit analyses for emission controls. Through the CENRAP process and
subsequent consultation meetings with Oklahoma, Texas shared significant amounts of
information with Oklahoma regarding SO2 and NOx emissions from Texas that could affect
Class I areas in Oklahoma. Texas also responded to several requests from Oklahoma and agreed
111
-------
to make changes to its New Source Review program to ensure that potential visibility impacts in
Oklahoma were accounted for. At the conclusion of the consultation period, Texas requested
"Oklahoma's concurrence on this assessment and verification that [Oklahoma] is not depending
on any additional reductions from Texas sources in order to meet [Oklahoma's] reasonable
progress goals." Letter from Susanna M. Hildebrand, Air Quality Director, TCEQ, to Eddie
Terrill, Air Quality Division Director, ODEQ 2 (Mar. 25, 2008). Oklahoma agreed with Texas'
assessment and did not request further reductions from Texas beyond those expected from
existing programs agreed to or implemented by Texas. See Letter from Eddie Terrill to Susanna
Hildebrand, supra. As a result, Oklahoma established reasonable progress goals that did not
require additional emissions controls from Texas facilities, and Texas developed a long-term
strategy that did not incorporate additional emissions controls to improve visibility in Oklahoma.
Despite the depth of coordination and consultation between Oklahoma and Texas, the
Associations noted that the EPA's proposal ignores the reasoned conclusions that these States
reached and rejects Oklahoma's reasonable progress goals and Texas' long-term strategy
because, in EPA's separate and distinct judgment, Oklahoma should have sought additional
information about potential emissions controls from certain sources in Texas. By second-
guessing these States, EPA's proposal is unlawful, arbitrary and capricious. EPA does not
dispute the quality of Oklahoma's modeling or its analysis of the four statutory factors required
by 40 U.S.C. §7419(g)(l). Instead, it relies on what it considers to be "an incomplete
consultation with Texas." EPA, Oklahoma Technical Support Document at 11. This conclusion
is not supported by the record. EPA points to no flaws in the CENRAP regional planning
process in which Texas and Oklahoma participated together. Nor does it point to any specific
flaws in the subsequent consultation process between the States. In fact, EPA concedes that, as a
result of consultations between the States, Oklahoma "had (1) abundant information showing the
impact of Texas sources on visibility at the Wichita Mountains [and] (2) evidence that cost-
effective controls on these sources were likely available." 79 Fed. Reg. 74,867. EPA goes on to
acknowledge that "the analysis developed by CENRAP [and used by Texas and Oklahoma in
their consultations] provide[d] a great deal of information on contributions to visibility
impairment and a set of potential add-on controls and costs associated with those controls." Id. at
74,872.
Although the proposal acknowledges the wealth of information shared between the two States,
the Associations noted that the EPA nonetheless asserts that Texas somehow deprived Oklahoma
of relevant information about Texas sources and emission reduction options. In doing so, EPA
fails to identify any specific information that Texas failed to provide. Instead, EPA simply
reinterprets the data and modeling available to Oklahoma while it developed its SIP and reaches
a different conclusion— that Oklahoma's reasonable progress goals should require additional
emission controls from a handful of Texas sources in order to provide small, imperceptible
improvements to visibility in Class I areas in Oklahoma. In doing so, EPA clearly would exceed
its statutory authority by putting aside a review of Oklahoma's compliance with statutory and
regulatory requirements and instead substituting its own judgment for that of the State. Not only
is this an unlawful usurpation of Oklahoma's discretion under the regional haze program to
consider the four statutory factors and establish reasonable progress goals, it is also arbitrary,
capricious, and unsupported by the record in light of the expansive consultation and information
sharing process that took place between Texas and Oklahoma.
112
-------
Response: In reviewing Oklahoma's SIP, we evaluated whether Oklahoma met the statutory and
regulatory requirements of the Regional Haze Rule. Adoption of reasonable progress goals
includes evaluation of the four statutory factors, consultation with other States that may
reasonably be anticipated to cause or contribute to visibility impairment in the mandatory Class I
area, and analyze the rate of progress needed to attain natural visibility conditions. While we
agree that Oklahoma possessed more than adequate information from the CENRAP analyses
about impacts from Texas sources at a certain level of aggregation, and some knowledge
concerning potential controls for some of these sources, we do not agree that it was reasonable
for Oklahoma to stop at this point. Despite the information it did have, Oklahoma never
explicitly asked Texas for reductions even though there was clear evidence from the CENRAP
analyses that Texas sources, particularly EGUs in northeast Texas, were significantly impacting
the Wichita Mountains and that cost-effective controls were likely available on some of these
sources.
The Regional Haze Rule required that Oklahoma use the consultation process under 40 CFR
51.308(d)(l)(iv) in the development of reasonable progress goals in tandem with Texas.
Nevertheless, throughout the consultations, Oklahoma failed to explicitly request that Texas
further investigate whether reasonable controls were available or that Texas reduce emissions
from these significantly impacting sources to ensure that all reasonable measures to improve
visibility were included in Texas' long-term strategy and incorporated into Oklahoma's
reasonable progress goals for the Wichita Mountains. This failure resulted in the development of
improper reasonable progress goals for the Wichita Mountains
We are disapproving Oklahoma's adoption of its reasonable progress goals for Wichita
Mountains because it failed to meet the consultation requirements. Oklahoma did not consult
fully with Texas regarding Texas' sources impacts on the Wichita Mountains. As noted in our
proposal, at the time that Oklahoma was developing its SIP, it had (1) abundant information
showing the impact of Texas sources on visibility at the Wichita Mountains, particularly from
EGU sources in northeast Texas, and (2) evidence that cost-effective controls on these sources
were likely available. Despite this information, the ODEQ requested neither that the TCEQ
further investigate controls at these sources nor did it explicitly request Texas to obtain
additional reductions from Texas sources to address the impacts of emissions from these sources
at the Wichita Mountains. The Regional Haze Rule requires States to use the consultation
process under Sections 51.308(d)(l)(iv) in the development of RPGs to ensure that all States,
including downwind States, take a hard look at what measures are necessary for ensuring
reasonable progress towards improving and maintaining visibility at Class I areas.
Based upon the above knowledge and without further analysis from Texas, Oklahoma did not
have adequate information to establish its RPG for the Wichita Mountains. It therefore should
have requested clearly that Texas further investigate these sources or explicitly requested Texas
to obtain additional reductions from Texas sources to ensure that all reasonable measures to
improve visibility were included in Texas' LTS and incorporated into Oklahoma's RPG for the
Wichita Mountains.
113
-------
Please see the responses regarding state and federal roles in the Regional Haze program for the
comment on EPA substituting its judgment for Oklahoma.
Comment: [TCEQ/PUCT (0056) p. 13] The TCEQ Stated that the EPA's proposed
disapproval of the State consultation requirements is based upon Oklahoma's
determination, subsequent to submittal of the Texas 2009 RH SIP, that it required further
reductions from Texas.
The EPA has not justified its determination that Texas failed to meet the requirements of
§51.308(d)(3)(i) and in fact the record shows that the process as laid out in the SIP and as
required by the rule was followed by Texas. The EPA's determination is based on a new
definition of progress goal in subsection (d)(3)(ii) and a misstatement of the actual rule itself in
subparagraph (i).
The TCEQ stated that Texas met the consultation requirements in §51.308(d)(3)(i). Texas
determined where emissions were reasonably anticipated to contribute to visibility impairment in
Oklahoma. Texas consulted with Oklahoma. The EPA asserts that the TCEQ should have
provided information necessary to identify reasonable reductions, which is not required by the
RHR. Oklahoma requested information on controls identified by CENRAP. Oklahoma had
information on control upgrades contained in the proposed Texas 2009 RH SIP. Yet, it did not
request additional controls on Texas sources or disagree with Texas' determination that
additional controls were not warranted during the first planning period. It was only after
consultation with Texas that Oklahoma argued that it needed controls that they did not have
authority to require from Texas sources. Oklahoma's after-the-fact change in position and the
EPA's subsequent proposed disapproval of their RPGs for Wichita Mountains does not provide
the legal basis for proposed disapproval of Texas' long-term strategy consultations.
Response: Long-Term Strategy consultation under 51.308(d)(3)(i) provides that where the State
has emissions that are reasonably anticipated to contribute to visibility impairment in any
mandatory Class I area in another State, the State must consult with the other State in order to
develop coordinated emission management strategies. Reasonable Progress Goal consultation
under 51.308(d)(l)(iv) requires that a State must consult with those States which may reasonably
be anticipated to cause or contribute to visibility impairment in the mandatory Class I areas in
establishing its reasonable progress goal. Texas had sufficient evidence that its sources were
impacting visibility at the Wichita Mountains area and that cost-effective controls on some of
these sources were likely available. Because of this evidence, it performed additional analysis,
which was technically flawed.
Although Texas participated in CENRAP, it retained the duty to do whatever additional analysis
was necessary to address fully the requirements of the Regional Haze Rule for addressing its
long-term strategy and setting its reasonable progress goals. While the long-term strategy
requirements allow a state to rely on the RPO technical analysis that is true only to the extent it
provides the necessary information. A state must address any gaps in that analysis. For Texas,
inadequate information existed not only for the reasonable progress analysis for its own Class I
areas, but also for the long-term strategy development for addressing significant impacts at the
114
-------
Wichita Mountains. CENRAP was not required, nor did it provide state-specific analyses and
information on the cost-effectiveness and visibility benefits of potential control strategies under
consideration by each state to address the specific sources or groups of sources within that state
that have the largest visibility impacts. Rather, CENRAP provided more general information on
overall projected visibility conditions, potential controls and associated costs for some sources
and the potential benefit of regional emission reductions to inform the development of potential
control strategies that may require additional analysis.84 For example, while the CENRAP
analysis identified that impacts from EGUs in Texas were significant, it did not provide a refined
analysis to assess fully the cost-effectiveness and visibility benefits of controlling those sources,
including not providing information on the cost-effectiveness of scrubber upgrades for those
sources with existing, underperforming scrubbers. As Texas states in its regional haze SIP,
"While Texas participates in CENRAP and benefits from the technical work coordinated by the
RPO, Texas has sole responsibility and authority for the development and content of its Regional
Haze SIP."85
Recognizing that the information made available by CENRAP indicated the significant impact of
Texas emissions and potential for cost-effective controls, Texas used the CENRAP analysis as a
starting point, and performed supplemental analysis for both its reasonable progress and long-
term strategy demonstrations. However, that additional technical analysis performed by Texas
was flawed and therefore did not provide the type of information necessary to fully evaluate the
reasonableness of controls at Texas sources with the largest potential to impact visibility at its
own Class I areas and the Wichita Mountains. Allowing this lack of adequate information to
continue was a critical misstep for ODEQ in setting its reasonable progress goals, and a critical
misstep for Texas when determining its fair share of emissions reductions under the long-term
strategy requirement. The plain language of the CAA requires that states consider the four
factors used in determining reasonable progress in developing the technical basis for the
reasonable progress goals both in their own Class I areas and downwind Class I areas. Such
documentation is necessary so that interstate consultations can proceed on an informed basis, and
so that downwind states can properly assess whether any additional upwind emissions reductions
are necessary to achieve reasonable progress at their Class I areas. Therefore, Texas had an
obligation to provide appropriate information to Oklahoma so it could establish a proper progress
goal for the Wichita Mountains. Further, Texas had an obligation to conduct an appropriate
technical analysis, and demonstrate through that analysis (required under (d)(3)(ii)), that it
provided its fair share of emissions reductions to Oklahoma. In summary, Texas was required
through the consultation process to provide Oklahoma the information it needed to establish its
reasonable progress goals for the Wichita Mountains, and it failed to do so.
Comment: [TCEQ/PUCT (0056) p. 14] The TCEQ Stated that the EPA's finding that the
TCEQ did not meet the long-term strategy consultation requirements of 40 CFR 51.308(d)(3)(i)
84 CENRAP conducted a control sensitivity analysis to evaluate the impact of point source emission reductions
across all CENRAP states given a maximum dollar per control level of $5,000/ton; however, the results "were
intended to be a starting point for control discussions that would require much greater refinement." Technical
Support Document for CENRAP Emissions and Air Quality Modeling to Support Regional Haze State
Implementation Plans, September 12, 2007 at 2-37).
85 2009 Texas Regional Haze SIP at 3-1.
115
-------
and (ii) ignores the voluminous and detailed consultation record contained in the Texas 2009 RH
SIP. The EPA holds Texas to a different standard of review than it has with other similar
regional haze SIPs.
The TCEQ noted that §51.308(d)(3) requires, (i) that Texas consult with other States if its
emissions are reasonably anticipated to contribute to visibility impairment at that State's Class I
areas(s), and (ii) if so, it must demonstrate that it has included in its SIP all measures necessary
to obtain its share of emission reductions needed to meet the RPG for that Class I area.
The TCEQ noted, as the EPA acknowledges, that the TCEQ relied on CENRAP source
apportionment modeling and its own supplemental analysis, available to all affected States,
FLMs, and tribes, to evaluate and identify reasonable controls. The TCEQ did include additional
controls or measures in its SIP, beyond those required to meet other programs, and every State in
the consultation, including Oklahoma, concurred. For Wichita Mountains, additional controls
were not deemed reasonable given that the CENRAP modeling - agreed to by all the States -
showed that the visibility impairment contributions from Texas go down during the planning
period (2002- 2018). The EPA's preamble, and Table 26 acknowledge this.12 Most importantly,
Oklahoma did not request additional controls from Texas during consultation. The EPA ignores
the record and proposes to hold the Texas plan to a standard that is not found in the RHR. The
EPA merely disagrees with the TCEQ's conclusions and attempts to apply a 'reasonableness"
standard to §51,308(d)(3)(ii) where none exists. That section only requires that the TCEQ
demonstrate that all controls necessary to meet the progress goal, for Wichita Mountains, are
included. Oklahoma agreed that no additional controls were needed at the time, and the evidence
that the contribution to visibility improvement from emission reductions at Texas sources during
the planning period is a sufficient basis for these conclusions.
Footnote:
12 "The contributions from Texas sources on total visibility impairment decreases from 2002 to 2018 at all impacted
Class I areas shown in the tables below." 79 FR page 74860.
Response: All of these issues are covered elsewhere in our responses to other comments. In
particular, see the consistency portions of this document for our responses to allegations that we
have been inconsistent in our treatment of Texas in comparison to our other actions.
Comment: Texas and Oklahoma satisfied the Interstate Consultation Requirements, and
EPA therefore has no authority to disapprove the SIPs based on EPA's unfounded
assertion that the States failed to meet those requirements. [UARG (0065) p. 9]
UARG noted that the EPA's proposed rule describes the interstate consultation undertaken by
Texas and Oklahoma. 79 Fed. Reg. at 74,843-44. EPA's proposed determination that Texas
and Oklahoma failed to satisfy their interstate consultation obligations with respect to one
another is without foundation and cannot serve as a basis for disapproving either State's RPGs or
any other part of their regional haze SIPs.
Response: See our previous responses on consultation.
116
-------
Comment: Texas Satisfied Its Interstate Consultation Obligations. [UARG (0065) p. 9-12]
UARG stated that the EPA correctly proposes to find that Texas fulfilled its consultation
obligations under the regional haze rule with respect to States whose emissions affect visibility in
Texas Class I areas. Id. at 74,844. EPA proposes, however, to disapprove Texas' regional haze
SIP, in part, because EPA does not believe that Texas fulfilled its interstate consultation
obligations with respect to Oklahoma. The proposed rule describes in considerable detail the
consultation between Texas and Oklahoma, id. at 74,854-57, and demonstrates that Texas
engaged in a comprehensive consultation process that satisfied all of the regional haze rule's
applicable requirements, see id. at 74,855.
UARG Stated that process concluded with both States agreeing that no emission reductions from
Texas in addition to those modeled by the Central Regional Air Planning Association
("CENRAP") would be needed to ensure reasonable progress for Oklahoma. Id. Under the
regional haze rule, that should be the end of the matter. EPA nevertheless proposes to
disapprove Texas' consultation because EPA "believe[s] that the technical analysis developed by
Texas did not provide the information necessary to identify reasonable reductions from its
sources, and inform consultations in order to develop coordinated management strategies with
Oklahoma." Id. at 74,856. EPA, however, provides no evidence for any such conclusion and
does not and cannot support its proposed determination that Texas failed to satisfy its obligation
to consult with Oklahoma and to provide any information needed for consultation. To the
contrary, the record demonstrates that both Texas and Oklahoma participated in CENRAP and
that both Texas and Oklahoma included in their SIP submissions documentation and emission
reduction measures necessary to effectuate the RPO's recommendations. The proposed rule
simply asserts that participation in an RPO "does not automatically satisfy a State's obligation to
'demonstrate that it has included in its implementation plan all measures necessary to obtain its
share of the emission reductions needed to meet the progress goal' for a Class I area." Id. No
such Statement appears in the regional haze rule. In fact, the rule refutes EPA's proposed
determination here by providing that:
[w]here other States cause or contribute to impairment in a mandatory Class I Federal
area, the State must demonstrate that it has included in its implementation plan all
measures necessary to obtain its share of the emission reductions needed to meet the
progress goal for the area. If the State has participated in a regional planning process,
the State must ensure it has included all measures needed to achieve its apportionment of
emission reduction obligations agreed upon through that process. 40 C.F.R. §
51.308(d)(3)(ii) (emphasis added).
UARG stated that the same conclusion is supported by the regional haze rule's technical-
documentation provision:
The State must document the technical basis, including modeling, monitoring and
emissions information, on which the State is relying to determine its apportionment of
emission reduction obligations necessary for achieving reasonable progress in each
mandatory Class I Federal area it affects. The State may meet this requirement by relying
117
-------
on technical analyses developed by the regional planning organization and approved by
all State participants. Id. § 51.308(d)(3)(iii) (emphasis added).
UARG asserted that far from providing or suggesting that emission control measures agreed on
through the RPO process are merely a starting point for analysis, the regional haze rule equates
adoption of all measures agreed on through the RPO process to the demonstration that a SIP
contains all necessary measures. EPA's Statements to the contrary in this proposed rule are
inconsistent with its own regulations and cannot support disapproval of the interstate
consultation component of the Texas SIP.
UARG Stated, further, as EPA notes, "[i]n fact, the TCEQ [Texas Commission on
Environmental Quality] went beyond the CENRAP analysis by contemplating additional
controls, applying a lower cost-effectiveness threshold and estimating the visibility benefit from
the identified control set." 79 Fed. Reg. at 74,857. Moreover, Texas relied on that supplemental
analysis - in addition to CENRAP's analyses and conclusions - "to inform its decision not to
control any additional sources, including those that impact the visibility at the Wichita
Mountains and other Class I areas in other States." Id.
UARG Stated that, ironically, despite the fact that Texas "went beyond" the requirements of the
regional haze rule, id., EPA proposes to conclude that Texas did not prepare enough analyses to
evaluate completely its reasonable progress obligations or to permit Oklahoma to establish in a
fully informed way the RPGs for Wichita Mountains. Id. at 74,861, 74,862. EPA does not
establish a legal basis for disapproving Texas' SIP on these grounds. Indeed, EPA does not
explain how the CENRAP modeling was inadequate or how it differed from the modeling
conducted by the other RPOs and does not explain how or why Texas' additional modeling was
also inadequate. EPA only asserts vaguely that additional "refine[ment] from a high level State"
was necessary and that "it [is] necessary to undertake a cost/control and visibility analysis which
is presented in our FIP TSD [Technical Support Document]" to support adequate reasonable
progress determinations. Id. at 74,861. No support exists in the regional haze rule for any such
requirements, and EPA never provided Texas with guidance suggesting that refined modeling
and additional analyses were necessary components of or prerequisites to an approvable regional
haze SIP. A State (or EPA, when it is authorized to promulgate a FIP) may choose to undertake
such additional analyses, but they are not legally required of a State. Thus, the absence of such
analyses can provide no basis for disapproving Texas' regional haze SIP.
Response: See our previous responses.
Comment: Oklahoma Satisfied Its Interstate Consultation Obligations.
[UARG (0065) p. 12-14]
UARG stated that the EPA's proposed disapproval of Oklahoma's RPGs for Wichita Mountains
is unsupported and contrary to the CAA and EPA's regional haze rule. As it does with respect to
Texas, the proposed rule makes clear that Oklahoma undertook every action required by the
regional haze rule, including reliance, as contemplated by the regional haze rule, on CENRAP's
modeling. Id. at 74,864-65. Oklahoma's consultation, as described in the proposed rule, was
118
-------
extensive. Id. at 74,865-67. As a result of the consultation, Oklahoma requested three specific
actions from Texas: (1) that Texas require new and modified sources subject to EPA's
prevention of significant deterioration ("PSD") program to conduct analyses of their impacts on
visibility at Wichita Mountains; (2) that Texas give Oklahoma an opportunity to review and
comment on PSD determinations regarding "best available control technology" for proposed
projects likely to affect visibility at Wichita Mountains; and (3) that Texas extend evaluations of
visibility impacts from within 100 kilometers of Wichita Mountains to within 300 kilometers of
Wichita Mountains. Id. at 74,866. As EPA's proposed rule explains, Texas agreed to
Oklahoma's first two requests and committed to working with the Federal Land Managers
("FLMs") and with Oklahoma to develop a protocol to determine when a proposed PSD source
should conduct a Class I area review. Id.
UARG noted that the EPA goes on to describe Oklahoma's consideration of the reasonable
progress factors and its assessment of the reasonableness of the URP during the first planning
period. Id. at 74,868- 69. EPA states:
After considering the URP, the results of the CENRAP modeling and the four reasonable
progress factors ... [Oklahoma] determined that meeting the URP goal for 2018 was not
reasonable. It then adopted the 2018 projected visibility conditions from the CENRAP
photochemical modeling as the RPGs for the 20% best days and 20% worst days for the
Wichita Mountains. Id. at 74,869.
UARG Stated that as a result of the regional consultations and Oklahoma's consideration of the
URP and the reasonable progress factors, Oklahoma developed RPGs for Wichita Mountains that
comply with the regional haze rule requirements that RPGs "provide for an improvement in
visibility for the most impaired days over the period of the SIP and ensure no degradation in
visibility for the least impaired days over the same period." Id. at 74,865.
UARG stated that, nevertheless, EPA proposes to disapprove the Oklahoma RPGs due to a
purported "incomplete consultation" with Texas "that resulted in inadequate reasonable progress
towards the national visibility goal." Id. The source, nature, meaning, and parameters of EPA's
putative "completeness" criterion are utterly unclear and undefined. Ultimately, however, the
supposed shortcoming in Oklahoma's consultation efforts, according to EPA, is that "the
technical analysis developed by Texas did not provide the information necessary to identify
reasonable reductions from its sources, and inform consultations in order to develop coordinated
management strategies with Oklahoma." Id. at 74,871. Apparently, EPA's theory is that if
Texas had provided more analysis, or if Oklahoma had demanded it, Oklahoma would have
realized that Texas was not offering emission reductions consistent with reasonable progress
requirements and that Oklahoma would have asked EPA to press Texas for additional emission
reductions.
According to UARG, this EPA rationale for proposing to disapprove the Oklahoma consultation
component of its regional haze SIP and, as a result, its RPGs is fatally flawed for at least two
reasons. First, there was no shortcoming in the technical analyses Texas provided to Oklahoma.
As noted above, the information Texas provided not only satisfied the regional haze rule's
requirements, it exceeded them. Indeed, elsewhere in the proposed rule EPA States that, in its
119
-------
view, information in the record "showed that cost-effective controls on Texas sources were likely
available" and that information in the record documented the impact of Texas sources on Wichita
Mountains. Id. Given those acknowledgements by EPA, EPA cannot logically maintain that
Oklahoma lacked the information from Texas that was necessary to apprise Oklahoma that Texas
might be able to contribute additional emission reductions. Second, EPA is not empowered to
intervene in the interstate consultation component of the regional planning process that the CAA
directs. States and RPOs conduct the technical analyses needed to inform decision-making, and
then States must decide for themselves whether they believe the consultation process has been
successful or whether, instead, to seek EPA resolution of interstate disagreement.
UARG Stated that Oklahoma had all of the information it required, consistent with the regional
haze rule, to make appropriate RPG determinations through the interstate consultation process.
EPA, accordingly, has no basis for finding that Oklahoma's interstate consultation with Texas
was incomplete or that its RPGs for Wichita Mountains fail to represent reasonable progress.
Response: See our previous responses to Luminant and others in this section. We have
addressed the comments on the Texas technical analysis elsewhere.
Comment: EPA arbitrarily imposed new, more onerous State consultation requirements in
its Proposal than it has applied to other SIP reviews [Xcel Energy (0064) p. 22-24]
Xcel Energy noted, as part of a State's development of RPGs, a State "must consult with those
States which may reasonably be anticipated to cause or contribute to visibility impairment in the
mandatory Class I Federal area." 40 C.F.R. § 51.308(d)(l)(iv); see also 40 C.F.R. §§
51,308(d)(3)(i)-(iii) (outlining consultation requirements for the long-term strategy). As EPA
notes, all that is required for interstate consultation is that States work together and "take a hard
look at what measures are necessary" to meet RPGs. 79 Fed. Reg. at 74,867.
Xcel Energy stated that it is undeniable that Texas and Oklahoma took the requisite "hard look"
required as part of interstate consultation on regional haze. Most importantly, both States were
active participants in the CENRAP regional planning process. Because CENRAP modeling
showed Texas and Oklahoma sources affecting each State's Class I areas, Texas and Oklahoma
commenced a separate interstate consultation process. Indeed, Texas and Oklahoma provided
substantial documentation of the numerous meetings, phone calls, and correspondence
exchanged on the impacts of each State's sources on Class I areas. See 79 Fed. Reg. at 74,865-
67 (reviewing substantial consultation efforts); Texas SIP, at App. 4- 2. Based on a request
from Oklahoma as part of this consultation process, Texas even specifically agreed to allow
Oklahoma an opportunity to comment on Texas' evaluation of Best Available Control
Technology for new and modified sources. See 79 Fed. Reg. at 74,866.
Nonetheless, Xcel Energy noted that the Proposal disapproves of Oklahoma's and Texas'
interstate consultation, and concludes that "Oklahoma did not have adequate information to
reasonably establish its RPG for the Wichita Mountains, and, as explained below, should have
requested that the TCEQ further investigate these sources." Id. at 74,867; see also id. at 74,857
(disapproving of Texas' consultation under long-term strategy requirements). The Proposal
120
-------
never details what information Oklahoma lacked in establishing its RPGs, nor did EPA provide
an adequate explanation of how additional information would have changed Oklahoma's ultimate
determination that additional controls on Texas sources would not move the Wichita Mountains
perceptibly closer to its regional haze goals.
Xcel Energy Stated that the EPA is requiring significantly more from Oklahoma and Texas with
respect to the consultation process than it has previously required via rule, guidance and other
SIP approvals. 8 In adopting its regional haze regulatory guidelines, EPA noted that, if States
determine that no further controls are needed in a particular planning period, States must merely
document "any consultations with other States in support of their conclusions." RHR, 64 Fed.
Reg. at 35,721-22. Texas and Oklahoma did this. See Texas SIP, at App. 4-2.
Xcel Energy stated that the EPA suggests that Texas and Oklahoma are simply relying on their
participation in CENRAP to meet their consultation requirements. See 79 Fed. Reg. at 74,856
(noting that "[participation in a regional planning organization does not automatically satisfy a
State's obligation to" consult). This is inconsistent with the substantial record of consultation
between the States. Id at 74,865-67. Even EPA acknowledges that Texas went above and
beyond the regional planning process that EPA deems sufficient for interstate consultation in the
long-term strategy context. Id at 74,857 ("In fact, the TCEQ went beyond the CENRAP analysis
by contemplating additional controls, applying a lower Cost-effective ness threshold and
estimating the visibility benefit from the identified control set."). This was more than what is
required in EPA regulations on consultation. 40 C.F.R. § 51.308(d)(iii) (noting that "The State
may meet this requirement by relying on technical analyses developed by the regional planning
organization and approved by all State participants.").
Xcel Energy noted that the EPA also appears to be arbitrarily taking a harder line in reviewing
Texas' and Oklahoma's consultation efforts than it has taken with other States. In other recent
regional haze SIP actions, EPA concluded that interstate consultation requirements were met,
even though there was substantially less consultation than the discussions between Oklahoma
and Texas:
• In Michigan, EPA found adequate consultation even when Michigan did not offer
additional controls for a Class I area not meeting its glide path until 2209. See
Proposed Michigan SIP Approval, 77 Fed. Reg. at 46,917 ("By coordinating with the
MRPO and other RPOs, Michigan has worked to ensure that it achieves its fair share
of overall emission reductions").
• In Arkansas, EPA concluded that Arkansas met consultation requirements based on
three calls with States and concurrence in the conclusion that controls in other States
are not necessary. Proposed Arkansas SIP Approval/Disapproval, 76 Fed. Reg.
64,186, 64,196 (Oct. 17,2011).
• In Kentucky, EPA found that Kentucky adequately addressed the consultation
requirements by determining that sources were meeting more stringent requirements
than regional MANE-VU recommendations. Kentucky SIP Approval, 76 Fed. Reg.
at 78,213.
121
-------
Xcel Energy argued that the EPA cannot make inconsistent conclusions on the adequacy of the
consultation process between Oklahoma and Texas as compared to other States without any
reference to its rules, guidance and prior SIP approvals.
Response: Please see our responses to the consistency sections of this document for our
responses to allegations that we have been inconsistent in our treatment of Texas in comparison
to our other actions.
Comment: Texas and Oklahoma consulted as required on the reasonable progress goals for
the Wichita Mountains. [NRG (0078) p. 9]
NRG Stated that Texas and Oklahoma met the consultation requirements of EPA's rules. EPA's
preamble documents the communications between the States' agencies, concluding with a
succinct description of facts indicating that the States agreed on the key point of the reasonable
progress goal for the Wichita Mountains:
The TCEQ concluded by requesting ODEQ's concurrence on that assessment, and, "that
your State is not depending on any additional reductions from Texas sources in order to
meet your reasonable progress goal(s)."
On May 12, 2008, the ODEQ responded to that letter and concurred with the
"information in that letter." (79 Fed. Reg. at 74,855)
NRG stated that this record makes it clear that Texas and Oklahoma consulted. EPA's regional
haze regulations require States to consult with each other, as they did, but do not require the
States to consult with EPA:
In developing each reasonable progress goal, the State must consult with those States
which may reasonably be anticipated to cause or contribute to visibility impairment in the
mandatory Class I Federal area. In any situation in which the State cannot agree with
another such State or group of States that a goal provides for reasonable progress, the
State must describe in its submittal the actions taken to resolve the disagreement. In
reviewing the State's implementation plan submittal, the Administrator will take this
information into account in determining whether the State's goal for visibility
improvement provides for reasonable progress towards natural visibility conditions. (40
C.F.R. § 51.308(d)(l)(iv))
By dismissing the States' consultation, NRG stated that the EPA's proposed disapproval of the
Wichita Mountains reasonable progress goal is inconsistent with the process described by EPA's
own regulations. EPA must accept the reasonable progress goals identified by Texas and
Oklahoma for the Wichita Mountains.
Response: Please see our previous responses to Luminant and others in this consultation
section.
122
-------
Comment: Texas has met all consultation requirements found in the regional haze long-
term strategy regulations. [GCLC (0063) p. 9]
GCLC noted that the EPA believes that Texas "did not adequately address the requirement in
Section 51.308(d)(3)(i) to 'consult with the other State(s) in order to develop coordinated
emission management strategies.'"38 This is largely based on EPA's belief that "the technical
analysis developed by Texas to evaluate controls for Texas sources did not provide the
information necessary to identify reasonable reductions from its sources, and inform
consultations in order to develop coordinated management strategies with Oklahoma." 39 This is
simply not true and directly conflicts with the numerous meetings, interactions, and affirmative
approvals, between Texas and Oklahoma administrative agencies.
GCLC Stated that Texas has fully complied with the consultation requirements outlined in the
LTS regulations40 with regard to the Central Regional Air Planning Association ("CENRAP")
States, including Oklahoma. Texas and Oklahoma engaged in lengthy and detailed consultation
in the development of their regional haze SIPs. GCLC listed examples of consultation from
2007 to 2008 (see comment 0063 for details).
[GCLC (0063) p. 11] GCLC asserted that EPA has already recognized the extensive
information that was in Oklahoma's possession as it was developing its SIP. In an attempt to
disagree with the reasonable decisions made by Oklahoma in developing its SIP, EPA Stated
"[a]t the time that Oklahoma was developing its SIP, it had ... abundant information showing the
impact of Texas sources on visibility at the Wichita Mountains." 52 While this impact did not
warrant additional controls, as reasonably determined by both Texas and Oklahoma
cooperatively, EPA's Statement clearly illustrates that Oklahoma was in possession of "abundant
information." Further, EPA has admitted that "Oklahoma did not specifically request any
additional reductions from Texas sources." 53
GCLC noted that Oklahoma had all necessary information to develop an appropriate regional
haze SIP, and in its reasonable discretion, chose not to request additional controls in Texas
beyond those required by existing programs. While EPA may disagree with that choice - though
GCLC believes that Oklahoma's choice was valid - it is not a valid legal basis to disapprove
Texas' long-term strategy, because the evidence is clear that there was extensive consultation
between Oklahoma and Texas. Texas met and exceeded all consultation requirements and all
regulatory requirements, and Texas' SIP includes all measures "to achieve its apportionment of
emission reduction obligations agreed upon through [the regional planning] process." 54
Footnotes:
38 TX TSD at 49.
39 Id.
40 40 CFR § 51.308(d)(93)(ii).
52 Proposed FIP, 79 Fed. Reg. 74867.
53 Id. at 74856
54 40 CFR § 51.308(d)(3)(ii). (emphasis added).
Response: Please see our previous responses to Luminant and others in this section.
123
-------
Comment: Legal and Factual Background [UARG (0065) p. 2-5] Finally, States whose
emissions may cause visibility impairment in another State's Class I area and States with Class I
areas that may experience visibility impairment caused by emissions from other States may be
subject to an interstate-consultation requirement. Id. § 51.308(d)(l)(iv). The purpose of that
requirement is to provide a forum for States to decide collaboratively on reasonable emission
reductions and appropriate apportionment of responsibility for reducing emissions during each
planning period of the regional haze program.
Response: Please see our previous responses to Luminant and others in this section.
Comment: Earthjustice et al., provided background on developing a long-term strategy.
[Earthjustice (0067) p.7]
Earthjustice et al., stated that a regional haze implementation plan must, among other things,
include emission limits, schedules of compliance, and "all measures necessary" to make
reasonable progress towards achieving natural visibility conditions. 40 C.F.R. §
51.308(d)(3)(ii). In developing a long-term strategy, a State must look beyond major stationary
sources to area, mobile, and minor sources, id. § 51.308(d)(3)(iv), as well as a number of other
sources of impairment such as construction, agricultural, and forestry practices. Id. §
51.308(d)(3)(v). The long-term strategy must be sufficient to achieve reasonable progress for
both the Class I areas within a State's borders as well as the out-of-State areas affected by the
State's emissions. Id. § 51.308(d)(3). To ensure that each State does its part to address regional
haze, a State that contributes to impairment at another State's Class I area must consult with the
State home to the Class I area. Id. § 51.308(d)(3)(i).
According to Earthjustice et al., Consultation proceeds based in part on analyses of how much
impairment at a given Class I area is due to emissions from each State. Each State then must
document the technical basis by which it determines its share of the emissions reductions
necessary to make reasonable progress at a Class I area. Id. § 51.308(d)(3)(iii). After fulfilling
the procedural requirements for consultation, a State must ensure that its haze plan satisfies the
primary substantive requirement for a long-term strategy: to include the enforceable measures
necessary to meet the reasonable progress goals for each Class I area affected by the State's
emissions. Id. § 51.308(d)(3).
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
Comment: Luminant provided background information on establishing a long-term
strategy. [Luminant (0061) p. 9]
Luminant explained that in addition to the RPGs, a State must also develop a long-term strategy
that addresses visibility impairment for both in-State Class I areas and out-of-State Class I areas
"which may be affected by emissions from the State."57 The long-term strategy must "include
124
-------
enforceable emissions limitations, compliance schedules, and other measures as necessary to
achieve the reasonable progress goals established by States having mandatory Class I Federal
areas."58
Luminant noted that in order to develop its long-term strategy, the State must "consult with the
other State(s) in order to develop coordinated emission management strategies."59 And "the State
must demonstrate that it has included in its implementation plan all measures necessary to obtain
its share of the emission reductions needed to meet the progress goal for the area."60 In
developing the long-term strategy, the State must evaluate seven specific factors that are distinct
from the four statutory factors for reasonable progress assessments. Among these factors are
"[ejmissions limitations and schedules for compliance to achieve the reasonable progress goal,"
mitigation of construction activities, source retirement and replacement, smoke management
techniques, and net changes in visibility projected from changes in point, area, and mobile
sources.61 Thus, the long-term strategy broadly encompasses multiple source types and actions
by the State and is not intended to be a source-specific analysis.
According to Luminant, of particular relevance to EPA's proposal here, EPA's regulations
specifically provide that, to meet the long-term strategy requirement, "[i]f the State has
participated in a regional planning process, the State must ensure it has included all measures
needed to achieve its apportionment of emission reduction obligations agreed upon through that
process."62 As discussed elsewhere, Texas met this requirement by participating in a regional
planning organization and including in its long-term strategy all emission reductions agreed upon
through that process.
Footnotes:
57 Id. § 51.308(d)(3)
58 Id. (emphasis added).
59 Id. § 51.308(d)(3)(i).
60 Id. § 51.308(d)(3)(ii).
61 Id. § 51.308(d)(3)(v).
62 Id. § 51.308(d)(3)(ii) (emphasis added).
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
Comment: Luminant provided background information on Texas' approach to
establishing the State's long-term strategy. [Luminant (0061) p. 23]
Luminant noted that Texas developed its long-term strategy to address regional haze visibility
impairment at each Class I area within the State and at each Class I area outside the State that
may be affected by emissions from Texas sources. Consistent with EPA's regulations, Texas
used CENRAP's PSAT modeling to determine the "apportioned impact of different areas and
pollutants to visibility impairment" for in-State Class I areas (Big Bend and Guadalupe
Mountains) and out-of-State Class I areas impacted by emissions from Texas.177
Luminant Stated that for the in-State areas (Guadalupe Mountains and Big Bend), the PSAT
modeling showed that Kansas, Louisiana, New Mexico, and Oklahoma "contribute] to visibility
125
-------
impairment" at these Class I areas.178 Texas' long-term strategy relies on reductions that these
States projected from their sources.179 Texas concluded that "[bjased on their plans and
commitments elicited through the consultation process, the commission has determined that the
emissions reductions these States are projecting are reasonable for contributing to progress in
reducing their contributions to visibility impairment at the two Class I areas in Texas."180
In regard to Class I areas outside Texas affected by emissions from Texas sources, Luminant
noted that Texas and the surrounding States engaged in the required consultation through
CENRAP. Texas evaluated its emissions based on the CENRAP PSAT modeling and
specifically consulted with Arkansas, Missouri, Oklahoma, New Mexico, Louisiana, and
Colorado to determine "whether emission reductions projected in Texas by 2018 are sufficient to
meet Texas' apportionment of the impact reduction needed to meet the reasonable progress goal
for each Class I area in each State."181 Based on the consultation process "none of these States ...
asked Texas for further emission reductions to help the State meet its reasonable progress goals
for its Class I area(s)."182
As to the Wichita Mountains, Luminant noted that the consultation process between Oklahoma
and Texas specifically acknowledged certain impacts on the Wichita Mountains. Relying on
CENRAP modeling, Oklahoma's reasonable progress goal for the Wichita Mountains "reflects
visibility improvement resulting from emissions reduction programs associated with the Federal
CAA and Oklahoma CAA, including long-term strategies of Oklahoma, Texas, and other States
and presumptive emissions reductions from the Oklahoma BART rule."183
To set its RPG, Luminant noted that Oklahoma relied on the Alpine modeling that was
commissioned by CENRAP. This modeling demonstrated that "[e]ven if all CENRAP member
States compelled sources to install and use controls as effectively as this scenario envisions, then
the Wichita Mountains still would fall significantly short of meeting the uniform rate of progress
glide path for the worst quintile days in 2018."184 Oklahoma thus conducted the statutory four-
factor analysis to determine the reasonable progress goal for Wichita Mountains.185 Oklahoma
established the reasonable progress goal for the Wichita Mountains based on projected visibility
of 21.47 deciviews in 2018 for the worst 20 percent days.186 Using this goal, Oklahoma
estimated that "natural conditions will be met circa 2102."187
Footnotes:
177 Id. at 11-1.
178 Id. at 11-7.
179-182
183 ODEQ, Regional Haze Implementation Plan Revision 104 (Feb. 2, 2010) ("2010 Oklahoma SIP Narrative").
184 Id. at 109.
185 Id. at 111-12, 114.
186 Id.
187 Id. at 104.
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
126
-------
Comment: Luminant provided a summary of EPA's proposed disapproval of Texas' long-
term strategy. [Luminant (0061) p. 47]
Luminant noted that the EPA proposes to disapprove Texas' long-term strategy in three
respects.335 First, EPA proposes "to find that Texas' long-term strategy does not include all
measures necessary to obtain the State's share of emission reductions needed to make reasonable
progress in the Wichita Mountains Class I area in Oklahoma," despite the fact that Texas
included all emission reductions that were agreed to in the CENRAP process and Texas /
Oklahoma consultations.336 EPA bases this conclusion on its finding that, although Texas and
Oklahoma engaged in extensive consultations through CENRAP, Oklahoma "did not specifically
request any additional reductions from Texas sources," and Texas "did not adequately" consult
with Oklahoma.337 EPA further "do[es] not agree . . . with the ODEQ's approach to consultation
to address impacts from emissions from Texas."338 Specifically, even though Oklahoma had
"abundant information" about impacts from Texas sources at Wichita Mountains and "evidence
[on] Cost-effective controls on these sources," EPA believes that Oklahoma should have
requested that Texas "further investigate controls at these sources."339 For these reasons, EPA
concluded that Texas has failed to meet the consultation requirement in 40 C.F.R. §
51.308(d)(3)(i) and (ii).340
Second, EPA proposes "to find that the technical basis on which Texas relied to determine its
apportionment of emission reduction obligations necessary for achieving reasonable progress in
Wichita Mountains was inadequate" and thus Texas' long-term strategy does not meet the
requirements of 40 C.F.R. § 51.308(d)(3)(iii).341 EPA bases this conclusion on its finding that,
although Texas' long-term strategy "rel[ied] on technical analyses developed by CENRAP and
approved by all State participants" and further "performed an additional analysis building upon
the work of the regional planning organization in order to evaluate additional controls," Texas
should have independently "considered] the four-factor used in determining reasonable progress
[in 51.308(d)(l)(i)(A)] in the developing the technical basis for . . . downwind [i.e., out-of-State]
Class I areas," including Wichita Mountains.342 In other words, EPA contends that Texas should
have developed its own reasonable progress goal for Wichita Mountains located in Oklahoma.
Third, EPA proposes "to find that Texas did not adequately consider the emissions limitations
and schedules for compliance needed to achieve reasonable progress in Big Bend, Guadalupe
Mountains, or Wichita Mountains,"343 and thus Texas' submission does not meet 40 C.F.R. §
51,308(d)(3)(v)(C) in EPA's view. As to Big Bend and Guadalupe, EPA's finding of
inadequacy relies on EPA's finding that Texas 'RPGs are inadequate.344 As to Wichita
Mountains, EPA "believe[s] the record supports a finding that [Texas's] analysis is inadequate as
it does not provide the information necessary to determine the reasonableness of controls at those
sources in Texas that significantly impact visibility at the Wichita Mountains." 345
Footnotes:
335 Id. at 74,822.
336 Id.
337 Id. at 74,856.
338 Id. at 74,867.
339 Id.
340 Id. at 74,856; see TX SIP TSD at 48-49.
341 79 Fed. Reg. at 74,822.
127
-------
342 Id. at 74,857, 74,861; see TX SIP TSD at 65.
343 79 Fed. Reg. at 74,822.
344 TX SIP TSD at 71.
345 Id.
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
6. Source Category and Individual Source Modeling
Comment: EPA is applying an unlawful standard to Texas. [Luminant (0061) p. 1]
Luminant Stated that the EPA proposes to disapprove Texas' regional haze SIP and impose a FIP
for the sole reason that, in EPA's view, Texas was required to conduct a source-specific analysis
of certain facilities to meet the reasonable progress requirement.2 But, as the Tenth Circuit has
recently held: "Neither the Clean Air Act nor the Regional Haze Rule requires source-specific
analysis in the determination of reasonable progress."3 And EPA's own regional haze guidance
provides that "[reasonable progress is not required to be demonstrated on a source-by-source
basis."4 EPA's proposal thus relies on the wrong legal standard and is contrary to law.
Footnotes:
279 Fed. Reg. 74,818, 74,839 (Dec. 16, 2014) (explaining that EPA is proposing to disapprove Texas' SIP
"[b]ecause individual sources were not considered by the TCEQ").
3 See WildEarth Guardians v. EPA, 770 F.3d 919, 944 (10th Cir. 2014).
4 EPA, Additional Regional Haze Questions 9 (Sept. 27, 2006), available at http://tinyurl.com/EPARHquestions.
Response: This comment does not accurately describe the proposed basis for disapproval of
Texas' reasonable progress analysis and mischaracterizes what Texas had in fact submitted for
EPA review. We reviewed that analysis under the requirements of 40 CFR 51,308(d)(l)(i)(A)
mindful also that conducting a proper analysis is critical to meeting requirements in Section
51.308(d)(3). Our disapproval hinges on deficiencies in the provided analysis, but these
deficiencies did not specifically or necessarily come about from the SIP not providing source-
specific analyses or not demonstrating reasonable progress on a source-by-source basis. For
purposes of Texas meeting the requirements, the rule sets forth that reasonable progress goals
must be set for each mandatory Class I Federal area within the State of Texas. Notably, the
technical record for this FIP does not purport to conduct a source-by-source analysis for all
sources in Texas; instead, it identified a group or subset of impactful sources that justified further
analysis. In inviting comments on and determining whether controls are appropriate for
individual sources, the study of the appropriateness of those controls naturally becomes more
"source-specific" in its emphasis. However, even with an appropriate analysis of a group of
sources or a study of sector-wide emissions, this is necessarily how the analysis would proceed
when controls appear to be warranted on some or all sources. Accordingly, while our FIP does
consider and ultimately apply controls to individual sources to assure reasonable progress, this is
128
-------
consistent with our regulations, our case law, and past EPA guidance (including those cited by
the commenter).86
Our proposal had observed "that individual sources were not considered by the TCEQ,"
however, we note this statement concerned the depth, granularity and coherency of the State's
analysis. It would be incorrect to assert that the submitted analysis had no individual source
analysis. In fact, our proposal's discussion section bore the title, "The TCEQ's Weighing of the
Four-factors for Individual Sources." 79 FR at 74838. Plainly, Texas submitted an analysis that
employed a large potential control set consisting of a mix of large and small sources. (See
Section 10.4.1 of the SIP submittal). It is apparent from the Texas SIP itself (see, for example,
Appendix 10-1: Analysis of Control Strategies and Determination of Reasonable Progress Goals,
and accompanying tables) that Texas identified individual sources as possible candidates for
controls under reasonable progress. However, Texas used an inappropriate and flawed
demonstration to reach its determination that no reasonable controls were required on those
sources, and this determination conflicts with the technical record before the State and as further
developed for this FIP. We proposed to determine the analysis was deficient and not approvable
because the large control set was not appropriately refined, targeted or focused on those sources
having significant and potentially cost-effective visibility benefits. Consistent with our proposal,
we conclude that control set was over-inclusive. It included controls on sources that would
increase total cost figures with little visibility benefit. As was noted in our proposal, the SIP
adopted this approach despite evidence in the record of identified cost-effective controls that
would result in large emission reductions on certain EGUs and despite source apportionment
modeling identifying large impacts from EGU sources in northeast Texas. As was stated in the
proposal, this approach had the potential to "mask" benefits that might be obtained. Moreover,
as was also noted in our proposal, the submitted analysis failed to study or consider scrubber
upgrade candidates, and this proved to be an inappropriate and unreasonable omission that
overlooked an important aspect of the regional haze problem and further obscured an accurate
picture of the potential for highly cost-effective emission reductions that provide progress toward
natural visibility conditions. We find it necessary to disapprove the submitted analysis because
these deficiencies materially affected the outcome of the State's SIP process.
Comment: EPA's "individual control" and "individual source" requirement for reasonable
progress is an unlawful basis for disapproval of Texas' goals. [Luminant (0061) p. 64]
Luminant Stated that EPA's proposal is based on an unlawful standard for applying the four
statutory factors for "reasonable progress." EPA's proposal—including EPA's proposed
disapproval of Texas' reasonable progress goals and long-term strategy and its proposed Federal
Implementation Plan ("FIP")—is premised on the unfounded assertion that Texas was required
to review and analyze the four statutory factors in 42 U.S.C. § 7491(g)(1) on an individual
control basis to a small group of facilities and that Texas is prohibited from applying the factors
on a source-category basis.453 EPA would disapprove Texas' RPGs because, EPA finds,
86 40 CFR 51.308(d)(l)(i)(A) expressly sets forth that the analysis will deal with "compliance" time and costs and
repercussions for "potentially affected sources." To the extent this or any other comment would suggest the analysis
can or should fully depart from all consideration of source impacts on Class I Federal areas, the analysis—whatever
its organizational structure or emphasis—would have no useful content.
129
-------
"TCEQ's analysis is insufficient to determine the visibility benefit of controlling the source or
subset of sources with the most effective controls for improving visibility conditions . . . ,"454
EPA's finding is an unlawful basis for its proposal.
Footnotes:
453 79 Fed. Reg. at 74,838 ("[BJecause the TCEQ did not evaluate controls on a source-by-source basis, source-
specific factors related to the evaluation of the reasonable progress four-factor analysis could not be considered.");
id. at 74,839 ("Because individual sources were not considered by TCEQ, we found it necessary to conduct an
additional analysis . . . .").
454 Id. at 74,841.
Response: We do not agree with how this comment characterizes the basis of our proposed
disapproval. Although we find fault with the State's analysis, which did not precisely engage the
four statutory factors on "an individual control basis to a small group of facilities," we do not
consider this the core of Texas' flawed approach. Nor are we disapproving the submitted
analysis for applying the factors on a source-category basis. Rather, the submitted analysis is
faulty for not having or not appropriately considering critical information. The State's analysis
had the effect of obscuring achievable visibility benefits. This is no specific consequence of the
labeling we would give to the structure of the State's analysis. The submitted analysis did not
appropriately define a subset of sources to produce an effective source-category or grouped
analysis. The source-category analysis also did not have the proper information to discern
obtainable visibility benefits or the most effective controls. These deficiencies do not implicate
that we demand uniform application of factors on an individual control basis. However, in the
case of this SIP submittal, we are left to conclude that these deficiencies are significant to an
extent that they would materially affect the outcome of the State's SIP process, and that this
component of the SIP thus warrants disapproval.
Comment: EPA's disapproval is contrary to the clean air act and EPA's implementing
regulations. [Luminant (0061) p. 64]
According to Luminant, EPA's basis for disapproval is contrary to the statute, regulations, and
guidance. Texas is not required to "determine the visibility benefit of controlling the source or
subset of sources with the most effective controls for improving visibility conditions," as EPA
contends.455 The statute, regulations, and EPA guidance all support and permit TCEQ's source-
category analysis and its review of costs on a source category basis as a reasonable basis for
developing its RPGs. EPA's individual source and individual control analysis of a small group
of sources is, by contrast, not required, nor is it the sensible way to evaluate the factors.456 As the
Tenth Circuit has specifically held, "[njeither the Clean Air Act nor the Regional Haze Rule
requires source-specific analysis in the determination of reasonable progress."457 Indeed, until
this proposed action, EPA has consistently taken the position that the reasonable progress factors
should be applied to "source categories" and that "[reasonable progress is not required to be
demonstrated on a source-by-source basis."458
Luminant Stated that EPA's entirely new and unprecedented interpretation and application of the
four reasonable progress factors in its review of Texas' SIP submission is inconsistent with the
structure of the Clean Air Act. The statute itself speaks of "classes or categories of sources" that
130
-------
impact visibility and directs EPA to issue its regulations taking that approach into account.459
Structurally, the statute regulates individual sources under the best available retrofit technology
("BART") provisions and the reasonably attributable visibility impairment ("RAVI") provisions,
but does not direct regulation of individual sources or the evaluation of individual controls under
the reasonable progress provisions.460 EPA's regulations, likewise, focus the reasonable progress
requirements and the four-factor on numerous sources, not individual sources or a small number
of sources. For example, the four reasonable progress factors are considered by a State in
"addressing] regional haze,"461 which is defined as "visibility impairment that is caused by the
emission of air pollutants from numerous sources located over a wide geographic area."462 In
contrast, the RAVI provisions of the regulations,463 which are separate and distinct from the
reasonable progress provisions, address "visibility impairment that is caused by the emission of
air pollutants from one[] or a small number of sources." 464 EPA's regulations incorporate this
important distinction throughout.465 Thus, EPA's "additional analysis" of "a small number of
sources"466 is an unlawful standard by which to disapprove Texas' reasonable progress analysis
and contrary to the statute and regulations. Texas' comprehensive analysis, by contrast, fully
met the reasonable progress requirements as contained in the statute and regulations.467
Luminant Stated that EPA has repeatedly confirmed that the statute and its implementing
regulations allow the four-factor analysis be done on a source category basis. For example, in
"guidance to the State air pollution control agencies and the general public on meeting the
regional haze SIP requirements," EPA explained in no uncertain terms:
Unlike the technical demonstration for CAIR or BART, the reasonable progress
demonstration involves a test of a strategy. The strategy includes a suite of
controls that has been identified through the identification of pollutants and
source categories of pollutants for visibility impairment - the possible controls for
these pollutants (and their precursors) and source categories - the application of
four statutory factors and how much progress is made with a potential strategy
with respect to the glide path. Modeling occurs with a strategy and is not a
source-specific demonstration like the BART assessment....
Reasonable progress is not required to be demonstrated on a source-by-source
basis. It is demonstrated based on a control strategy developed from a suite of
controls that has been assessed with the four statutory factors and the uniform rate
of progress.468
According to Luminant, in other guidance focused on the reasonable progress requirements for
SIPs, EPA again explained that States "have flexibility in how to take into consideration these
statutory factors and any other factors that you have determined to be relevant."469 As to the first
factor—"costs of compliance"—EPA specifically explained that "we believe that the cost of
compliance factor can be interpreted to encompass the cost of compliance for individual sources
or source categories.. ."470 Indeed, EPA repeatedly makes clear in its guidance that States may
apply the factors to "sources and/or source categories,"471 And EPA has explained that when
"EPA steps into the State's shoes" in issuing a FIP, it has "flexibility to make technical
judgments within the bounds of the [regional haze] rule, and . . is not statutorily obligated to
131
-------
impose source-specific controls,"472 EPA cannot require of Texas what EPA concedes is not
required of itself.
Luminant asserted that this EPA-endorsed approach is exactly what Texas followed in the
CENRAP process and in its SIP submission,473 and it must be approved. Yet, EPA judges
Texas' submission by the very standard that it has said does not apply.474 EPA's new
interpretation is not entitled to judicial deference here because, among other things, it squarely
conflicts with EPA's prior interpretation and without any explanation 475 And EPA's new
approach is arbitrary and capricious because its prior guidance endorsing a source category
approach "has engendered serious reliance interests that must be taken into account."476 If EPA
seeks to change its long-established interpretation of its regulations, which Texas and other
States have relied on, EPA may only do so prospectively to future regional haze submissions and
not retroactively to Texas'2009 submission.477
Footnotes:
455 Id.
456 Although EPA contends that it conducted an individual source-by-source analysis, Id. at 74,839, EPA in fact did
no such thing. Instead, EPA targeted 12 Texas facilities out of more than 1,600 for its "individual source" analysis
of costs and potential controls. See Cost TSD at 1. Thus, EPA's four-factor analysis, not TCEQ's, is incomplete
and does not adequately consider the four-factor for all Texas sources or even all Texas source categories.
457 WildEarth Guardians, 770 F.3d at 944.
458 EPA, Additional Regional Haze Questions 9 (Sept. 27, 2006).
459 42 U.S.C. § 7491(a)(3), (b)(1).
460 Id. §7491(b)(2)(A), (c).
461 40 C.F.R. § 51.308(d).
462 Id. § 51.301. See also 77 Fed. Reg. 30,248, 30,248 (May 22, 2012) ("[T]he Act and EPA's rules .. . require
States to prevent any future and remedy any existing anthropogenic impairment of visibility in mandatory Class I
areas caused by emissions of air pollutants from numerous sources located over a wide geographic area (also
referred to as the 'regional haze program')."); id. At 30,249 ("Regional haze is impairment of visual range or
colorization caused by emission of air pollution produced by numerous sources and activities, located across a broad
regional area.").
463 40 C.F.R. §§ 51.302-.306.
464 Id. § 51.301.
465 Id. § 51.300(a) (explaining that visibility impairment has "two principal forms"—"impairment attributable to a
single source/small group of sources" and "regional haze" "from a multitude of sources").
466 79 Fed. Reg. at 74,839.
467 EPA concedes as much by recognizing that "TCEQ constructed a large potential control set consisting of a mix
of large and small sources, located at various distances from Class I areas, with a large geographical distribution."
Id. at 74,838. And EPA further concedes that Texas and Oklahoma considered the impacts from individual "sources
of particular interest to Wichita Mountains." Id. at 74,855.
468 EPA, Additional Regional Haze Questions 9 (Sept. 27, 2006) (emphasis added).
469 Reasonable Progress Guidance at 5-1.
470 Id. (emphasis added).
471 Id. at 2-3 (emphasis added).
472 77 Fed. Reg. at 40,164 (emphasis added).
473 Texas RH SIP at 10-6 to 10-7.
474 Compare EPA, Additional Regional Haze Questions 9 (Sept. 27, 2006) ("Reasonable progress is not required to
be demonstrated on a source-by-source basis.") with 79 Fed. Reg. at 74,839 ("TCEQ did not separately evaluate the
visibility benefit from the implementation of this control [at Big Brown], or appropriately weigh the four reasonable
progress factors in determining the reasonableness of this individual control." (emphasis added)).
475 See Christopher v. SmithKline Beecham Corp., 132 S. Ct. 2156, 2166 (2012) ("Deference is undoubtedly
inappropriate, for example, when the agency's interpretation is plainly erroneous or inconsistent with the regulation.
And deference is likewise unwarranted when there is reason to suspect that the agency's interpretation does not
132
-------
reflect the agency's fair and considered judgment on the matter in question. This might occur when the agency's
interpretation conflicts with a prior interpretation . .. (internal citations omitted) (internal quotations omitted)).
476 Perez v. Mortgage Bankers Ass'n, 135 S. Ct. 1199, 1209 (2015) (internal quotations omitted).
477 Indeed, separately, EPA is has "planned" "revisions" to its regional haze regulations that would require
consideration of individual sources in the reasonable progress four-factor analysis. See EPA, Pre-Meeting Materials
for the EPA-FLM-RPO-States-Tribes Meeting on the Future of the Regional Haze Program, Topic IV: Reasonable
Progress and Long Term Strategy (Four-factor) 6 (Feb. 3, 2015) (explaining that "[t]he [current] RHR does not
explicitly require a State to consider visibility impacts of individual sources when selecting additional controls for
inclusion in its long-term strategy" and soliciting input on regulatory changes that would). But a prospective
rulemaking cannot justify EPA's approach to reviewing Texas'2009 regional haze SIP.
Response: Even as it is permissible to conduct a reasonable progress goal analysis according to
source categories and groups of sources, mere reliance on this organizational structure for the
analysis does not assure it will be approvable. It is our statutory task to review the submitted
analysis and the entire submitted SIP to ensure it conforms with the requirements of the Clean
Air Act, including—in this case—40 CFR 51.308(d)(l)(I)(A). We proposed to disapprove the
submitted analysis for, inter alia, being over-inclusive of sources with limited impacts at the
pertinent Federal Class I area and over-inclusive of sources that were relatively poor candidates
for making progress according to the statutory factors. We do not contend the submitted analysis
had to at all conform to the structure or depth of analysis requisite under BART or RAVI rules;
different rules address those sources or source requirements. But our acknowledgement of these
contrasting requirements does not mean the submitted analysis is shielded from substantive
scrutiny or that individual sources cannot permissibly have compliance requirements as an
outcome of the analysis to establish RPGs. Whether or not one cites analytical components as
being site-specific or generalizable across a source category (and whether or not a factor is
applied collectively to a group of sources or applied to an individual source), the submitted
analysis missed an important aspect of the problem of regional haze for the Federal Class I areas
at issue, specifically Guadalupe Mountains and Big Bend.
We are within the framework of past guidance where we stated that that the factors apply
(individually or together) to sources or source categories. In this case, the submitted analysis
was overgeneralized and not discerning of the information most important to source category and
individual source impacts at the relevant Federal Class I areas. This may have been
accomplished in the context of a source category analysis that was open to and discriminating of
the most critical and relevant information. However, regardless of how the analysis is classified,
the demonstration fails for not giving appropriate focus to the sources or a subset of sources that
were most critical to the matter of visibility protection at the Federal Class I areas addressed by
our rule.
We acknowledge many of the citations provided by the comment—although several are presented
outside their proper context87—but the comment begins with the faulty premise that the
disapproval is based on the SIP's use of source-category analysis. This is not correct. The
87 For example, in citing the language of 77 Fed. Reg. at 40,164, the comment erroneously places importance on a
stray EPA Statement, made under a FIP obligation in an unrelated rulemaking, that EPA was "not statutorily
obligated to impose source-specific controls." However, this Statement was made to emphasize statutory flexibility,
not statutory limitations in imposing controls. The cited notice thereafter explained EPA's flexibility to instead
require an alternative program under 40 CFR 51.308(e)(3). If anything, the text there undermines the commenter's
position.
133
-------
disapproval instead rests on the SIP's flawed analysis, the State's reliance on which has
frustrated the requirement that the State implementation plan "contain emission limits, schedules
of compliance, and other measures as may be necessary to make reasonable progress." CAA
Section 169A(b)(2). To the extent the comment advocates that emission controls cannot or can
never be required to make reasonable progress, the comment is plainly mistaken. By the statute
and by regulation, such controls, including "source-specific controls," may be required. Notably,
the Texas SIP submission does not argue otherwise; doing so (i.e., agreeing with the commenter
that only BART and RAVI sources are subject to controls in the regional haze program) would
quite likely have constituted an additional basis for disapproval. Contrary to the positions of
many commenters, it is apparent that the Texas, by its submission, presupposes that individual
source controls (beyond BART and RAVI controls) can be required to make reasonable progress.
Even as we find it necessary to disapprove part of the Texas SIP, we note this area of agreement
on an underpinning of the regional haze program.
We believe that in some instances, it is possible for a State to approach reasonable progress on a
source category basis. For instance, a State may be able to demonstrate that due to the
magnitude of their emissions and distance from any Class I areas, EGUs or some other source
category are not significant contributors to regional haze, and thus should not be controlled
during this planning period. However, although that was possible and appropriate for other
sources categories in Texas, it was not possible for Texas EGUs. The technical record before
Texas demonstrated that its EGUs are a very significant contributor to regional haze at a number
of Class I areas, and our additions to the technical record reinforce this fact. In fact, both the
Texas and Oklahoma regional haze SIPs are suffused with data points and information on these
significant contributions. Still, that fact in and of itself did not derail Texas' strategy. Instead,
the decision by Texas to effectively ignore (by obscuring or otherwise) this information is what
caused its plan to not be approvable.
Once we determined that Texas' submission was not approvable and we were required to
propose a Federal plan, it is possible that our plan could have emphasized source categories in
lieu of specific sources. For instance, having identified EGUs as the primary contributing source
category, and also having identified SO2 as the primary contributing pollutant, we could have
presented a plan that focused on further controlling SO2 across the EGU source category.
However, we believe that approach would have been less satisfactory. In this instance, unless
evaluated on a source-by-source basis, we would have had no basis for determining which of
these EGUs would be most beneficial to control. We would in fact be in the same place we note
in our responses to objections to our consideration of visibility—an "all or nothing" proposition,
arguably over-controlling sources or arguably formulating an arbitrary proposal. Our Final Rule
is consistent with the Congressional goal set in the Clean Air Act, and ensures that affected
sources had the benefit of a thorough study of the appropriateness of controls before those
controls were imposed.
Comment: EPA's disapproval ignores that Texas did consider individual sources and
found that Individual controls were not reasonable. [Luminant (0061) p. 69]
134
-------
Luminant Stated that EPA's contention that Texas ignored individual sources and their impacts
contradicts the record before EPA. Texas and Oklahoma, as part of the CENRAP process,
considered "sources of particular interest to Wichita Mountains," including "their emissions and
location with the [areas of impact]" modeled by CENRAP.495 Contrary to EPA's apparent belief,
Texas not only assessed controls for the appropriate source categories consistent with EPA
guidance, it also considered and reviewed the costs of compliance for individual units, such as
Luminant's Big Brown units, to determine that no additional controls were reasonable.496 EPA is
thus wrong when it asserts that "individual sources were not considered by the TCEQ."497
Indeed, not only did TCEQ include in its analysis the cost of controls at individual units
developed by CENRAP, TCEQ included in its analysis "additional individual sources" that were
"not part of the CENRAP AirControlNET dataset,"498 thus going above and beyond its
obligation under the regulations 499
Luminant asserted that EPA's conclusions about Texas' process and the information it developed
are simply not correct and cannot support EPA's proposal. An agency decision fails to pass legal
muster if the administrative record reveals that the agency "offered an explanation for its
decision that runs counter to the evidence . . . ." 500 And an agency must demonstrate that it has
"examine[d] the relevant data and articulate[d] a satisfactory explanation for its action including
a 'rational connection between the facts found and the choice made.'" 501 Here, EPA's proposal
fails these standards. Contrary to EPA's apparent belief, the record shows that Texas did in fact
identify the sources most likely to contribute to visibility impairment at the Class I areas of
interest and reviewed costs of controls for those sources. EPA's contention that it needed to step
in and provide its own analysis, to fill in perceived gaps left by Texas, is thus contrary to the
record and arbitrary and capricious.
Footnotes:
495 79 Fed. Reg. at 74,855.
496 See, e.g., Texas RH SIP at 10-7; Alpine Geophysics, SO2 Costs Per Ton, Dock. ID No. EPA-R06-OAR-2014-
0754-013-13 (May 9, 2006).
497 79 Fed. Reg. at 74,839.
498 2009 Texas SIP Narrative at 10-7.
499 See 40 C.F.R. § 51,308(d)(3)(iii) ("The State may meet this requirement by relying on technical analyses
developed by the regional planning organization. . . .").
500 Motor Vehicle Mfrs. Ass'n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983).
501 Id. (quoting Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168 (1962)).
Response: To the extent the comment observes that aspects of an individual source analysis
were submitted—and in some ways can even said to be characteristic of the submitted four-
factor analysis—we agree. This and many other public comments appear to be fixed on the
proposal's Statement that "individual sources were not considered by the TCEQ." This
Statement in our proposal follows a paragraph in which we point out that the TCEQ declined to
control Big Brown, despite its own finding that Cost-effective controls were available, that these
controls could be expected to result in approximately 40,000 tons per year of SO2 reductions, and
that EGUs in east Texas caused large visibility impacts. Consequently, we could have instead
referred to the TCEQ's efforts as "individual sources were not effectively considered by the
TCEQ." Nevertheless, our statement remains accurate in the important sense that the analysis
was not conducted with appropriate emphasis on the most important data and the most important
sources or subset of sources. To reiterate the example offered in our proposal: TCEQ had record
135
-------
evidence that Big Brown units would be especially promising candidates for controls that would
potentially yield visibility improvement benefits through SO2 reductions and could even do so
quite cost-effectively, yet the "TCEQ did not separately evaluate the visibility benefit from the
implementation of this control." This omission and related omissions critically undermined the
overall analysis and demonstration.
As much as we appreciate Luminant's assertion that the submitted analysis arguably does
consider individual sources, this viewpoint only reinforces our role in reviewing the adequacy of
the State's consideration or lack of consideration of the factors for establishing reasonable
progress goals, an enterprise that inherently requires the assessment and consideration of
potential controls for individual sources. (See, e.g., Section 10.4.1 of the SIP submittal:
"... pollutants of primary concern were determined to be SO2 and NOx from point sources...).
Moreover, the comment validates the authority of States to consider and ultimately impose
controls where justified on individual sources in establishing reasonable progress goals. We
agree with this point and note that it also validates the authority of EPA to impose such controls,
when appropriate, when EPA is subject to its obligation to issue a FIP, as it is here.
Comment: EPA's unprecedented approach is bad policy. [Luminant (0061) p. 70]
Luminant stated that EPA's proposed unprecedented approach is not sensible. By relying on a
source category analysis—and not just targeting a small number of sources as EPA does—
Texas' regional haze and reasonable progress analysis is fully consistent with the statutory and
regulatory requirements and provides a comprehensive look at numerous sources.502 That is the
essence and goal of a regional haze/reasonable progress analysis, which should take into account
"numerous sources and activities, located across a broad regional area." 503 EPA's focus on only
a "small number" of sources, while ignoring others, does not provide for a comprehensive look
at all potential contributors and opportunities for improvement.504 Individual sources, moreover,
are expressly considered under other aspects of the regional haze program—namely the RAVI
analysis and BART—not the reasonable progress analysis. Texas' reasonable progress analysis,
not EPA's, meets the substance, intent, and plain language of the regulations.
Footnotes:
502 See, e.g., 77 Fed. Reg. at 30,251 ("States should consider all types of anthropogenic sources of visibility
impairment in developing their LTS, including stationary, minor, mobile, and area sources.").
503 Id. at 30,249.
504 Not only is EPA conducting the wrong analysis for the reasonable progress factors, EPA's conclusions contradict
its prior finding "there is no existing visibility impairment [at Big Bend or Guadalupe] that is reasonably attributable
to specific sources" in Texas. 54 Fed. Reg. 7,767, 7,769 (Feb. 23,1989).
Response: We disagree with the comment's suggestion that individual sources are not subject to
control as part of the measures necessary to make reasonable progress. The statute and
regulation make clear that reasonable progress goals are established and selected in consideration
of—to borrow the comment's extra-statutory phrasing—all potential contributors and
opportunities for improvement. See CAA Section 169A(b)(2); 40 CFR 51.308(d)(l)(i)(A). It is
not happenstance that the final rule focuses on a small number of sources, but is instead and
rather the product of the thoughtful application of the factors used to determine reasonable
progress. See CAA Section 169A(g). Under a refined and, in fact, comprehensive analysis of
136
-------
numerous sources, these sources are determined to be the key contributors and the key
opportunities for improvement that the Texas analysis unreasonably overlooked. Neither Texas,
nor we, take the position that these sources are beyond regulation under the regional haze
program under the argument that these sources are not presently subject to an applicable RAVI
or BART analysis. Texas identified sources through Q/d and cost information, but the state's
analysis mistakenly overlooked scrubber upgrades and obscured a realistic picture of the
contributors and opportunities for improvement. We also note that even those sources subject to
BART or RAVI controls can be subject to new and later assessment (and more effective or
stringent future controls) as may be necessary to make reasonable progress. It is not evident that
the comment embraces this understanding, even though it is entirely natural to the aims of the
regional haze goals set by Congress.
Comment: EPA's proposed individual source-based approach for reasonable progress
goals is unlawful. [Associations (0059) p. 13-14]
The Associations Stated that the EPA's proposal to disapprove Texas' reasonable progress goals
and long-term strategy and replace them with a FIP is also unlawful because EPA adopts an
individual source-based approach to setting reasonable progress goals inconsistent with the
Clean Air Act. Unlike other aspects of the regional haze program, reasonable progress goal
provisions are intended to address contributions from a wide range of sources that can be best
addressed on a source-category basis. In this respect, they are fundamentally different from
other provisions such as those for BART and reasonably attributable visibility impairment
("RAVI"), which are specifically designed to address individual sources. Moreover, EPA's
adoption of an individual source-based approach is inconsistent with EPA's past practice and
with EPA's strict uniformity rule for regional offices.
The Associations Stated that the EPA rejects Texas' source category-based approach for
establishing reasonable progress goals in favor of an approach that focuses on potential
emissions controls for individual sources. In particular, EPA asserts that "TCEQ's analysis is
insufficient to determine the visibility benefit of controlling the source or subset of sources with
the most effective controls for improving visibility conditions." 79 Fed. Reg. at 74,841; see also
id. at 74,838 ("[BJecause TCEQ did not evaluate controls on a source-by-source basis, source-
specific factors related to the evaluation of the reasonable progress four-factor analysis could not
be considered."). EPA then determined it was "necessary to conduct an additional analysis"
because "individual sources were not considered by the TCEQ." Id. at 74,839. After conducting
an individual source and individual emission control analysis of a small subset of sources within
Texas, EPA concluded that several sources should be required to install additional control
technologies. Not only does this approach ignore State primacy in establishing reasonable
progress goals, it unlawfully shifts the focus of the reasonable progress goals from source
categories to individual sources.
The Associations Stated that the Clean Air Act and EPA's implementing regulations draw a clear
line between source category-based reasonable progress goals and other source-specific regional
haze provisions. As the Tenth Circuit recently explained, "[njeither the Clean Air Act nor the
Regional Haze Rule requires source-specific analysis in the determination of reasonable
137
-------
progress." WildEarth Guardians v. EPA, 770 F.3d 919, 944 (10th Cir. 2014). Instead, the
Clean Air Act discusses "classes or categories of sources" that "may reasonably be anticipated to
cause or contribute significantly to impairment of visibility" and directs EPA to promulgate rules
to address them. 42 U.S.C. § 7491(a)(3), (b)(1). In contrast, under the Clean Air Act, BART
and RAVI provisions are applied individually to "each major stationary source" meeting certain
threshold criteria. Id. § 7491(b)(2)(A), (c).
According to the Associations, this structural distinction is maintained in EPA's regulations. For
example, under the Regional Haze Rule, SIPs that incorporate reasonable progress goals are
intended to address "regional haze." 40 C.F.R. § 51.308(d). Regional haze is defined by EPA as
"visibility impairment that is caused by the emission of air pollutants from numerous sources
over a wide geographic area." Id. § 51.301 (emphasis added). In contrast, EPA defines BART
as an "emission limitation [that] must be applied on a case-by-case basis ..." Id. Likewise,
EPA explains that its RAVI provisions are designed to address "visibility impairment that is
caused by the emission of air pollutants from one or a small number of sources." Id. Thus,
EPA's regulations draw a stark distinction between reasonable progress goals on the one hand,
which address emissions from a broad range of sources, and BART and RAVI provisions, which
address individual sources.
The Associations stated that the EPA's own guidance also confirms this interpretation of both the
Clean Air Act and EPA's implementing regulations. In defining the reasonableness of controls,
EPA expressly contrasts reasonable progress goals with BART:
Unlike the technical demonstration for CAIR or BART, the reasonable progress
demonstration involves a test of a strategy. The strategy includes a suite of controls that
has been identified through the identification of pollutants and source categories of
pollutants for visibility impairment—the possible controls for these pollutants (and their
precursors) and source categories—the application of four statutory factors and how
much progress is made with a potential strategy with respect to the glide path. Modeling
occurs with a strategy and is not a source-specific demonstration like the BART
assessment. EPA, Additional Regional Haze Questions 9 (Sept. 27, 2006).
The Associations stated that the EPA goes on to explain more explicitly that"Reasonable
progress is not required to be demonstrated on a source-by-source basis. It is demonstrated
based on a control strategy developed from a suite of controls that has been assessed with the
four statutory factors and the uniform rate of progress." Id. (emphasis added). In describing its
own obligations with respect to establishing reasonable progress goals in FIPs, EPA explained
that it has "flexibility to make technical judgments within the bounds of the [regional haze] rule
and ... is not statutorily obligated to impose source-specific controls77 Fed. Reg. at 40,164
(emphasis added). When establishing a FIP, "EPA steps into the State's shoes," id., and EPA
cannot disapprove a SIP—as it proposes to do here—for failing to conduct an analysis that EPA
concedes is not statutorily required. See 42 U.S.C. § 7410(k)(3) ("The Administrator shall
approve [a SIP or SIP revision] as a whole if it meets all of the applicable requirements of this
chapter."); National Ass 'n of Home Builders, 551 U.S. at 664 (finding similar language forbids
EPA from denying approval on the basis of non-statutory factors).
138
-------
Second, EPA fails to provide a reasoned explanation for disapproving Texas' reasonable
progress goals based on Texas' failure to conduct a source-by-source analysis of emission
controls when EPA has never required an individual source-based approach in the past. In its
prior reviews of State reasonable progress goals EPA has uniformly approved States' reliance on
source category-based analyses, even in the face of public comments urging a source-based
approach. For example, EPA approved Alaska's regional haze SIP in which the State asserted
that "it is reasonable to conduct the four-factor analysis on the general source categories rather
than on individual sources." Alaska, SIP Narrative 9-9 (2011); 78 Fed. Reg. 10,546 (Feb. 14,
2013) (approving Alaska SIP). Likewise, EPA approved Oregon's reasonable progress goals
after the State explained that it "looked at key pollutants and certain source categories and the
magnitude of their emission in applying the four-factor." Oregon, SIP Narrative 163 (2011); 77
Fed. Reg. 50,611 (Aug. 22, 2012) (approving Oregon SIP). EPA also approved Washington's
reasonable progress goals after the "State decided to focus its four-factor analyses on ... 10
specific industries and emission source categories." Washington, SIP Narrative at 9-5 to 9-7
(2010); 79 Fed. Reg. 33.439 (June 11, 2014) (approving Washington SIP).
It is a well-established tenet of administrative law that "[rjeasoned decision making ...
necessarily requires the agency to acknowledge and provide an adequate explanation for its
departure from established precedent." Dillmon v. NTSB, 588 F.3d 1085, 1089-90 (D.C. Cir.
2009) (citing FCC v Fox Television Stations, Inc., 129 S. Ct. 1800, 1811 (2009)). Indeed, given
that the submitting States relied upon EPA's established guidance and precedents in crafting
their SIPs, EPA is required to provide a "more substantial justification" for its disapprovals.
Perez v. Mortgage Bankers Ass'//, supra, slip op. at 13. In its proposal to disapprove Texas'
reasonable progress goals, EPA fails to even acknowledge, let alone provide a reasoned
explanation for, its departure from past precedent where it has approved multiple SIPs based on
analyses that were, in all relevant respects, identical to that conducted by Texas.
Response: We disagree and observe a significant misunderstanding in the comment. While we
(and at least one court) have explained that reasonable progress is not required to be
demonstrated on a source-by-source basis (or via "source-specific analysis"), this emphatically
does not mean that individual sources are free from any possibility of required new or stricter
controls as part of establishing reasonable progress goals or what is necessary as part of a long-
term strategy. In other words, an appropriately focused and detailed analysis using a so-called
category based approach may (and in some cases manifestly should) result in a determination
that source-specific controls are needed to make reasonable progress. To conclude otherwise
vitiates the statutory purpose and asks for an entirely ineffectual analysis which could not result
in possible emission limitations as expressly envisioned in the statute. See CAA Section
169A(b)(2); 40 CFR 51.308(d)(1) and (3).
In this case, we have determined the Texas analysis submitted to meet the requirements of 40
CFR 51.308(d)(1) was critically flawed, because the demonstration that was made to select the
reasonable progress goals was not appropriately focused and had obscured the potential for cost-
effective improvements in visibility. It is not inappropriate to apply the factors given by CAA
Section 169A(g)(l) to a group of sources on a collective basis, provided that the analysis and the
focus of the analysis is technically sound. In this case, on substantive review, it is evident that
the analysis is flawed and not consistent with what is required by the CAA. The SIP submission
139
-------
therefore does not, in fact, meet "all of the applicable requirements of the [Clean Air Act],"
which puts us in the position of "approving] the plan revision in part and disapproving] the plan
revision in part." CAA Section 110(k)(3). Nothing in the disapproval rests on non-statutory
factors (or the demands of regulations required by the CAA).
See the consistency section of this document for our responses to allegations that we have been
inconsistent in our treatment of Texas in comparison to our other actions.
Comment: [TCEQ/PUCT (0056) p. 10-11] The TCEQ argued that the EPA has no basis to
disapprove the State's RPGs because the TCEQ did not examine the four statutory factors on a
unit-by-unit basis. The TCEQ's analysis of the statutory factors using a source category
approach was consistent with the statute, the RHR, and the existing EPA guidance.
According to the TCEQ, neither CAA §169A, the RHR, nor the guidance available in 2009
required a unit-by-unit four-factor analysis even where the State's RPGs would improve visibility
less than the URP. The statute simply provides that in determining reasonable progress, the four
statutory factors shall be taken into consideration (§7491(g)(1)). The statute does not direct how.
The RHR provides the same in 40 CFR 51,308(d)(l)(i)(A). In addition, the EPA's RPG
guidance does not refer to a unit-by-unit four-factor analysis but instead says that States have
"flexibility" in how to consider the factors. The EPA has failed to establish that Texas' RPGs do
not meet the RHR for improvement in visibility for the most impaired days and no degradation
for least impaired days. The EPA also fails to establish that Texas' determination that additional
controls are unnecessary and that they would not provide a discernable visibility improvement
for the added cost is unreasonable based on the text of the CAA and the EPA regulations.10 The
EPA itself supported the non-source specific four-factor analysis approach in reviewing New
Mexico's regional haze plan. In a challenge to New Mexico's plan, the EPA "points out that
[§51.308(d)(l)(i)(A)] does not require a source-specific analysis." 11 The 10th circuit agreed that
"[N]either the Clean Air Act or the Regional Haze Rule requires source-specific analysis in
determination of reasonable progress." (id) The EPA has also ignored its own words from the
RHR preamble: " .... EPA is not specifying in this final rule what specific control measures a
State must implement in its initial SIP for regional haze. That determination can only be made
by a State once it has conducted the necessary technical analyses of emission, air quality, and the
other factors that go into determining reasonable progress" (64 FR 35721).
Footnotes:
10 Dissent in Oklahoma et al v. EPA (challenges to the EPA's SIP disapproval and FIP of Oklahoma's RH BART
determinations.) 10th circuit July 2013, pages 4-5: "Finally, it is worth noting that the EPA's regional haze program
is distinct in the amount of power given to the States There are a number of reasons for this approach, not the
least of which is that its goals and standards are purely aesthetic rather than directly related to health and safety.
The EPA's rule here requires OG&E to make a $1.2 billion investment over the next Five-years that will, even
under EPA's estimate, result in no appreciable change invisibility .... Although the EPA has at least some authority
to review BART determinations within a State's SIP, it has no authority to condition approval of a SIP based simply
on a preference for a particular control measure. Texas v. EPA 690 F3d 670,684 (5th Cir. 2012)
see EME Homer City Generation L.P. v. EPA 696 F3d 7, 29 (D.C. Cir. 2012) (reviewing a different rule and
concluding that the CAA 'prohibits EPA from using the SIP process to force States to adopt specific control
measures'). Oklahoma considered the cost and resulting benefit of such a large investment in scrubbers, and its
conclusion was not unreasonable."
11 See Wildearth Guardians v. EPA, 770 F3d 919, 944
140
-------
Response: We agree with the language cited by the comment from the RHR preamble. 40 CFR
51.308 plainly does not dictate specific control measures that a State must implement. A State
must first conduct the necessary technical analyses of the factors that go into determining
reasonable progress. We have found adequate the analyses of other States that used a so-called
source category approach in their SIP submissions, but this does not eliminate our mandated role
in reviewing the soundness of the Texas submission. We acknowledge the proclaimed outcome
of Texas' analysis, viz. the asserted determination that additional controls are unnecessary and
the asserted determination that requiring controls could not provide visibility benefits warranted
by the added costs, but this conclusion lacks an appropriate technical foundation and scope. As
noted in our proposal, we found that Texas' conclusion for no additional controls was
inappropriate because its analysis obscured the benefits that could be realized under a more
focused analysis of grouped or individual sources. We acknowledge we stated in our proposal
that "individual sources were not considered by the TCEQ."88 However, this Statement in our
proposal follows a paragraph in which we point out that the TCEQ declined to control Big
Brown, despite its own finding that cost-effective controls were available, that these controls
could be expected to result in approximately 40,000 tons per year of SO2 reductions, and that
EGUs in east Texas caused large visibility impacts. Consequently, we could have instead
referred to the TCEQ's efforts as "individual sources were not effectively considered by the
TCEQ." The core of our objection to Texas' approach to its reasonable progress demonstration
lies not in whether it selected an approach based on source category, individual, or some hybrid
of the two, but rather in how it judged the information it had, and that armed with that
information its decision to not pursue it any further. We have found the submitted analysis to not
be tailored to the task of determining reasonable progress; it avoids focus on the most suitable
candidates for new measures to make reasonable progress by grouping them with less suitable or
entirely not suitable candidates. These circumstances have not been in issue with EPA's
approval of RPGs or 40 CFR 51.308(d)(l)(i)(A) analyses submitted by other States, and we
necessarily acknowledge that Texas, by its State-specific profile of emissions adding to the
problem of regional haze, has an important challenge and responsibility in meeting this
requirement. We are required to review the determinations within the SIP and determine that
they are supported by the requisite analyses. Upon finding the SIP deficient, we are obliged to
determine the requirements of the FIP; our disapproval is not based on or conditioned on a
preference for particular control measures.
Comment: [TCEQ/PUCT (0056) p. 11] The TCEQ disagreed with the EPA's assertion that an
analysis of controls for a group of sources should not have been performed because this grouped
analysis hid potential improvements of smaller-costing controls from individual equipment.
The TCEQ stated that site-specific analyses were not considered necessary because visibility
improvements from a group were not perceptible. Thus, a subset of the sources could not result
in a better-controlled approach or improvement in the visibility predicted by the larger group.
The TCEQ performed a grouped source analysis because it was allowed under the EPA's rule
and the guidance available at the time the analysis was performed.
88 79 FR 74839.
141
-------
Response: As we noted in our responses to other comments, Texas did in fact perform some
analysis on individual sources, albeit in a flawed manner. As was outlined in our proposal and
TSD, the methodology employed in the SIP assumed all emission reductions within a large
geographic region would have the same effectiveness in reducing visibility impairment. This
contrasts with what was confirmed by our own source apportionment modeling, which found that
individual sources had significantly different impacts. As was additionally noted, and expanded
upon in our responses to other comments herein, even visibility improvements that are not
perceptible may still be determined to be significant and beneficial for purposes of meeting
regional haze program requirements, and the threshold Texas applied to its estimated visibility
benefits was not appropriate Furthermore, Texas dismisses the visibility benefit in comparison
to the overall cost, but as explained elsewhere, the analysis was over-inclusive and included
controls that served to increase the total cost but provide little to no visibility benefit.89 A more
focused analysis, identifying a smaller group of sources, would have resulted in an overall more
cost-effective control set resulting in significant visibility benefits. Thus, consistent with our
proposal, it remains the case that reliance on a flawed analysis materially affected the State's
conclusion.
Comment: [TCEQ/PUCT (0056) p. 12] The TCEQ disagreed with the EPA's position that it
was unreasonable for Texas not to ask for site-specific data to perform a site-specific analysis
because the TCEQ does not have the legal authority to require companies to submit the
information necessary to properly evaluate flue gas desulfurization (FGD) scrubber upgrades. It
is unreasonable for the EPA to expect the TCEQ to perform an analysis of scrubber upgrades on
the specific EGUs when only the EPA has the legal authority to obtain the necessary information
to conduct such an analysis.
The TCEQ noted that the EPA Stated in its Cost Technical Support Document and in the Federal
Register notice that the nature of acceptable scrubber upgrades is site-specific and the data were
not publicly available. Under CAA §114(a), the EPA required companies to submit detailed
information regarding the facilities' current scrubber systems and any improvements that have
been made since initial installation. The EPA indicated the information was necessary in order
to properly evaluate the potential for upgrades to the FGD scrubbers (79 FR 74876).
The TCEQ agreed that such extensive knowledge of the existing scrubber systems is necessary to
properly evaluate the viability of upgrading an FGD scrubber. However, the TCEQ does not
have any authority equivalent to the EPA's authority under CAA §114(a) to require submission
of cost data or design requirements for a suite of potential scrubber upgrades at individual sites.
The TCEQ cannot require the companies to provide the information that the EPA admits is
necessary to evaluate FGD scrubber upgrades. There are many possible control strategies the
TCEQ could have considered, but it can only evaluate controls for which it has credible and
defensible information to support. Additionally, the TCEQ is not aware if this information was
even available at the companies in 2008 when this portion of the SIP was developed.
89 The analysis was also "under-inclusive," because the potential for scrubber upgrades was not examined, although
they prove cost-effective.
142
-------
According to the TCEQ, it is unreasonable for the EPA to disapprove a SIP submittal on the
basis of the State failing to perform an analysis when only the EPA has the legal authority to
require submission of the necessary information for such an analysis. The EPA should not hold
the States to a standard for SIP approvability that only the EPA is capable of meeting.
Response: As was noted in the TSD, some information that our proposal had used to evaluate
controls, including FGD scrubber upgrades, was requested by us with reference to our CAA
Section 114 authority. As a side-note, Texas has the ability to request delegation of this
authority. See CAA Section 114(b). TCEQ, as the State's air quality planning agency, has other
authorities and avenues to request information from sources, develop information of its own,
and, if such information is in our possession, from us. TCEQ can and could adequately study
FGD scrubber upgrades as a potential control option with other information, e.g., using available
records and technical resources. As the Congressional vision of the Clean Air Act operates, the
State is charged with primary responsibility for "air pollution control at its source," and by virtue
of submitting SIPs, the State affirms its position that it has sufficient legal and technical capacity
to meet CAA requirements and address the issue of air pollution, including haze-causing
pollution. We disagree with this comment's suggestion that TCEQ could omit or overlook the
evaluation of controls based on a proclaimed lack of credible and defensible information. That
information can and should be developed under the State's regulatory authority and the SIP
process, including by requesting information from sources, developing the information using its
own technical resources, and soliciting information in public comments. We are not
disapproving the SIP for any standard that the State cannot meet. Even as the State may rightly
assert it does not have authority "equivalent" to CAA Section 114(a) authority, the absence of
that authority or of precise information gained under that authority is not related to our basis for
disapproval.
Comment: [TCEQ/PUCT (0056) p. 12-13] The TCEQ Stated that the EPA's finding that the
TCEQ should have considered scrubber upgrades in the 2009 RH SIP is arbitrary and capricious.
While the EPA did comment on the TCEQ's proposed 2009 RH SIP, the EPA did not suggest in
any way in those comments that the TCEQ should consider scrubber upgrades in the control
strategy analysis for reasonable progress goals. The EPA is attempting to hold Texas to a
standard created five-years after the TCEQ submitted the 2009 RH SIP.
The TCEQ noted that the EPA states in the proposed FIP that it was "unreasonable" for Texas to
not perform an analysis of potential scrubber upgrades on coal-fired units in Texas that were
already equipped with FGD scrubbers (79 FR 74841). However, in the comments (dated
February 15, 2008) that the EPA submitted on the proposed 2009 RH SIP, the EPA did not
suggest the TCEQ consider scrubber upgrades as a possible control strategy or indicate in any
manner that not considering this potential measure would be grounds for the EPA proposed
disapproval of the SIP. Furthermore, in the agency's comments (dated September 30, 2013) on
the proposed 2014 Five-Year Texas RH SIP Revision, the EPA again did not mention the subject
of FGD scrubber upgrades. The EPA had multiple opportunities to inform the TCEQ that
considering FGD scrubber upgrades was as critical as the EPA now claims it to be; however, the
143
-------
EPA did not even mention the subject of scrubber upgrades in any of the formal comments it
submitted to the TCEQ during the comment period for the 2009 RH SIP.
The TCEQ Stated that the EPA attempts to back-fill its lack of any notice to Texas regarding the
consideration of FGD scrubber upgrades by citing Statements made by the EPA in the 2005 final
BART rulemaking recommending that States consider scrubber upgrades for BART analysis
purposes (Technical Support Document for the Cost of Controls Calculations for the Texas
Regional Haze Federal Implementation Plan, page 26). However, the EPA's Statements in the
final BART rulemaking were made solely in the context of BART analysis (70 FR 39171). As
Texas was included in the CAIR in 2008 and the EPA determined that CAIR was better than
BART, the EPA's comments regarding scrubber upgrades and BART were not relevant to Texas.
Furthermore, the EPA did not mention in the 2005 BART rulemaking that States should also
consider scrubber upgrades for reasonable progress purposes even if the State's BART -eligible
EGUs were subject to CAIR.
The TCEQ contended that the EPA is attempting to hold Texas to a standard of SIP approvability
arbitrarily created by the EPA Five-years after the TCEQ submitted the SIP revision. The EPA
is creating impossible standards for SIP approvability by expecting States' SIP revisions to meet
requirements created by the EPA after the States are required to submit the SIP revision.
Response: Comment letters provided by EPA staff regarding State agency rules under
development are intended to aid rule development; they do not constitute any kind of agency
action or purport to review the in-progress SIP provisions under the authority and requirements
of CAA Section 110(k). Furthermore, the Act does not require us to provide comments on a
State's proposal during its public comment period. Therefore, we disagree with any suggestion
that TCEQ can be shielded from a disapproval action for failing to meet an applicable
requirement of the Clean Air Act because a point regarding that CAA requirement was not made
or emphasized in earlier comment letters to the State. The comment appears to acknowledge that
FGD scrubber upgrades are a known way of reducing emissions. This is not a hidden or obscure
measure for reducing emissions. Wherever a written reference to it may be located does not
change the fact that FGD scrubber upgrades are a control measure known before and after 2005;
"notice" or the asserted "creation" of a standard are not in issue. If we in reviewing the SIP had
unreasonably overlooked the matter of FGD scrubber upgrades just as TCEQ had overlooked
them in its control strategy analysis for reasonable progress, it would not change the fact that the
analysis had deficiently overlooked a critical option for controlling emissions—only that we had
overlooked another basis for disapproval (at least until we reviewed comments that would have
alerted us to the deficiency and the obstacle to approval).
Comment: [UARG (0065) p. 18] As part of their argument that the EPA does not provide any
lawful basis for disapproving the RPGs for Big Bend and Guadalupe or the Texas LTS, UARG
asserted that States are not required to conduct reasonable progress assessments on a source-by-
source basis. This is confirmed by EPA itself in the Reasonable Progress Guidance, which
States that the cost factor can be evaluated "for individual sources or source categories "
Reasonable Progress Guidance at 5-1 (emphasis added). State discretion to conduct a reasonable
progress assessment without analyzing controls on a source-by-source basis has been confirmed
144
-------
by the U.S. Court of Appeals for the Tenth Circuit. WildEarth Guardians v. EPA, 770 F.3d 919,
944 (10th Cir. 2014) ("Neither the Clean Air Act nor the Regional Haze Rule requires source-
specific analysis in the determination of reasonable progress."). EPA fails to acknowledge this
established law and cites no support for its conclusion.
Response: Our disapproval is not based on the claim that Texas had to conduct its reasonable
progress assessments on a source-by-source basis, but rather on the flaws in the analysis, as
discussed in our proposal. Nothing in the sources cited by the comment eliminates our
obligation to review the submitted assessment. We are required to review whether the
conclusions are technically supported (or supportable). Moreover, nothing in these sources
relieves us of our duty to disapprove a part of SIP revision that does not meet the Act and EPA
regulations. In this case, the State conducted a flawed evaluation of the factors to establish
reasonable progress, which under CAA Section 110(k) has the consequence of the partial
disapproval of the SIP revision.
Comment: [Nucor Steel (0058) p. 3] Nucor Steel Stated that EPA's interpretation also is
contrary to the CAA because it improperly focuses on case by case, source specific analyses of
individual sources located in Texas to achieve Oklahoma's reasonable progress goals which
would require more stringent analysis and controls than authorized under the Clean Air Act. See,
WildEarth Guardians v. EPA, 770 F.3d 919. (10th Cir. 2014).
Response: Individual sources may be subject to controls and source specific analysis when
establishing reasonable progress goals. The comment misinterprets the cited 10th Circuit case.
Nothing in that decision states that individual source controls cannot be required when
establishing reasonable progress goals. When a State conducts the requisite analysis—whether it
begins it by analyzing sources individually or by analyzing source categories—it should
encompass the possibility of applying individual source controls, when they are warranted and
reasonable. If, for example, a source category analysis were supplemented with information
showing that one particular point source was a leading obstacle to reasonable progress for a
particular Class I areas, then it would be proper for the State to focus its analysis on that source,
including with a source-specific assessment. Whatever flexibility is accorded in how the
analysis is shaped, it may not be unreasonable. It is not reasonable or a technically adequate
analysis if the State ignores or obscures the most promising candidate(s) for establishing
reasonable progress, and as our proposal indicated and we now conclude, the Texas analysis was
critically flawed in this respect. Incidentally, the cited 10th Circuit case had considered the claim
that the State of New Mexico failed to conduct source-specific analyses for its non-BART
sources with specific attention to the Escalante plant. However, there was no record evidence
that this plant, remote from any relevant Class I areas, had notable impacts for regional haze
purposes for this initial planning period. In future regional haze planning periods, depending on
technical evidence, that plant (and others in New Mexico) may very well be subject to controls
as may be required to meet reasonable progress.
Comment: EPA's use of source-specific analysis treats Texas sources differently from other
States. [NRG (0078) p. 4]
145
-------
NRG stated that the proposal evaluates the reasonable progress goals for the relevant areas on the
basis of source-by-source contributions to visibility impairment.
However, NRG contended that this approach exceeds what EPA can require of a State under the
statute and regulations. Regional haze regulations establish a process and criteria to be
considered in setting reasonable progress goals, but provide no mention of a source-specific
analysis. See generally 40 C.F.R. § 51.308. By contrast, the regional haze program only calls
for source-specific evaluations under the "reasonably attributable visibility impairment" program
(for which EPA has proposed to approve Texas' plan) and the Best Available Retrofit
Technology ("BART") program.
NRG Stated that source-specific analysis also has not been applied by EPA in practice in the
reasonable progress context:
EPA recently defended a State's reasonable progress goals that did not rely on a
source-specific analysis, and prevailed before the court: "[W]e reject the
environmental groups' argument that the EPA had to engage in a source-specific
analysis for a reasonable-progress determination. Nothing in the Regional Haze Rule
or the Clean Air Act required New Mexico to conduct a four-factor analysis of the
Escalante plant." WildEarth Guardians v. EPA, 770 F.3d 919, 944.
EPA has repeatedly used interstate emission trading rules such as the Clean Air
Interstate Rule ("CAIR") and the Cross-State Air Pollution Rule ("CSAPR") to
satisfy reasonable progress goals.11 As EPA is proposing to use CSAPR to satisfy
Texas' BART obligations,12 it would only be logical to also use CSAPR to satisfy
Texas' reasonable progress obligations as well.
Footnotes:
11 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77 Fed. Reg.
19,098 (March 30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,185 (June
27, 2012) (North Carolina); 77 Fed. Reg. 38,509 (June 28, 2012) (South Carolina); 77 Fed. Reg. 24,392 (April 24,
2012) (Tennessee); 77 Fed. Reg. 16,937 (March 23, 2012) West Virginia).
12 79 Fed. Reg. at 74,844.
Response: We disagree that source-specific controls to achieve reasonable progress are not
contemplated in 40 CFR 51.308. The statutory factors in CAA Section 169A(g) are directed to
sources, including individual stationary sources. Those factors can be addressed, on a collective
basis, for sources that are subject to a logical grouping. For example, a source category such as
sources involved in cement production may have unique considerations as compared to other
types of source. Or perhaps a source category or source is not demonstrated to be impactful of
relevant Class I areas as was the case for the Escalante Plant. However, even under a grouped or
so-called source category approach to analyzing the requisite factors, a State cannot obscure or
ignore potential candidates for closer study or potential control options. We have the
responsibility to disapprove an unreasonable and flawed analysis, as is the case here. Even
though we are required to prepare our own analysis to do a FIP, this does not mean our analysis
and conclusions are precisely what Texas was required to follow or itself conclude to meet the
146
-------
CAA requirements. We address the specific allegations of inconsistency in the consistency
section of this document.
Comment: There is no statutory basis to disapprove of Texas' SIP based on "additional
analysis" on a "small group" of individual sources. [GCLC (0063) p. 7-9]
GCLC Stated that EPA does not have the authority to disapprove Texas' SIP submission based
on an unfounded assertion that Texas was required to conduct a four-factor analysis on an
individual-source basis. EPA States in the Proposed FIP that:
"Because individual sources were not considered by the [Texas Commission on
Environmental Quality ("TCEQ")], we found it is necessary to conduct an
additional analysis to determine whether this approach materially affected the
outcome of the TCEQ analysis. As we demonstrate in detail in our FIP TSD, by
analyzing sources individually, we believe we have identified a small number of
sources that are responsible for much of Texas' collective visibility impact on the
Texas' Class I areas, which if controlled, would provide for visibility benefit at
Texas' Class I areas."28
EPA goes on to add in its TX TSD that:
"Because it only estimated the visibility benefit of all the controls together, the
TCEQ was not able to assess the potential benefit of controlling individual
sources with significant, and potentially cost-effective, visibility benefits."29
According to GCLC, EPA contends that Texas should have "separately evaluate[d] the visibility
benefit from the implementation of [individual] controls]."30 Besides the fact that visibility is
not one of the factors that Texas is required to analyze, as discussed above, there is simply no
basis for this type of review in statutory law, regulatory provisions, guidance, EPA's recent
regional haze SIP submittal decisions, and recent case law.
GCLC asserted, reviewing the statutory structure of the CAA, it is clear that there is no
requirement to evaluate and impose individual controls as part of a reasonable progress analysis.
This is echoed in EPA's regulations, which focus on reasonable progress requirements viewed as
whole from numerous source categories, not individual sources or even a small number of
sources. For example, the regulations require States to consider the four-factor to "address
regional haze,"31 which is defined as "visibility impairment that is caused by the emission of air
pollutants from numerous sources located over a wide geographic area. Such sources include,
but are not limited to, major and minor stationary sources, mobile sources, and area sources."32
GCLC noted that EPA's guidance has supported this conclusion that the reasonable progress
requirements do not require a source-specific analysis, stating:
Unlike the technical demonstration for CAIR or BART, the reasonable progress
demonstration involves a test of a strategy. The strategy includes a suite of
147
-------
controls that has been identified through the identification of pollutants and
source categories of pollutants for visibility impairment - the possible controls for
these pollutants (and their precursors) and source categories - the application of
four statutory factors and how much progress is made with a potential strategy
with respect to the glide path. Modeling occurs with a strategy and is not a
source-specific demonstration like the BART assessment.
Reasonable progress is not required to be demonstrated on a source-bv-source
basis. It is demonstrated based on a control strategy developed from a suite of
controls that has been assessed with the four statutory factors and the uniform rate
of progress. 33
GCLC Stated that EPA has also, on numerous recent occasions, approved of SIPs that did not
require source-specific requirements, but rather, looked more broadly at source categories. For
example, EPA recently approved the reasonable progress goals submitted by Idaho based on the
State's "general level of review for the major source categories."34 Finally, and likely most
importantly, EPA's decision to require source-specific analysis and limitations conflicts with the
decision of the U.S. Court of Appeals for the Tenth Circuit, which recently held "[njeither the
Clean Air Act nor the Regional Haze Rule requires source-specific analysis in the determination
of reasonable progress."35 Ultimately, EPA has no authority to impose source-specific emissions
limits for SO2, or any other pollutant, on any source in Texas. EPA must approve Texas' SIP.
Footnotes:
28 Proposed FIP, 79 Fed. Reg. at 74839.
29 TX TSD at 19.
30 Proposed FIP, 79 Fed. Reg. at 74839.
31 40 CFR§ 53.308(d).
32 40 CFR§ 51.301.
33 See EPA, Additional Regional Haze Questions, 9 (Sept. 27, 2006 Revision). Available at: https:/ /www.tceg.
texas.gov/assets/public/implementation/air/sip/bart/EPAQA-Haze.pdf
34 See Idaho SIP Approval Proposal, 77 Fed. Reg. at 30256; see also Approval and Promulgation of Implementation
Plans; State of Idaho; Regional Haze State Implementation Plan, 77 Fed. Reg. at 66929, 66929 (Nov. 8, 2012).
35 WildEarth Guardians v. EPA, 770 F.3d 919,944 (10th Cir. 2014).
Response: We disagree with GCLC's assertion that we cannot use visibility in our assessment
and address this issue in our responses to other comments. This comment, as with many others,
confuses observations offered in the proposal with the proposed basis for disapproval. Our TSD
revealed that Texas' approach was not appropriately refined and left unidentified the few, key
sources that were responsible for much of the collective visibility impact at Texas' Class I areas.
We proposed that this analysis was technically deficient in several respects such that it materially
affected the outcome and the State's conclusions. We did not propose to disapprove it because
of an alleged failure to be sufficiently source-specific. As focused on the sources impacting the
Class I areas, the TSD also supports our conclusion that controls are warranted to meet
reasonable progress. The comment is not correct that such individual controls cannot be required
by the Clean Air Act. This argument is contradicted by the terms of the statute and the logic of
the Congressional goal for natural visibility, CAA Section 169A(b)(2), and it is notably
inconsistent with the underlying assumptions of the SIP itself. To be clear: the State's analysis
overtly examines the question of "additional controls" and proposed controls for individual
148
-------
sources. The Tenth Circuit decision cited by the comment speaks to what the Clean Air Act does
or does not require for the shape of the analysis, but it does not purport to make unreasoned or
technically deficient analyses permissible. Moreover, it manifestly does not state that individual
controls cannot be required to be meet reasonable progress. The control strategy needed to
demonstrate reasonable progress would necessarily implicate new or additional controls for
individual stationary sources. If the CAA's requirements were such that only sources subject to
controls for BART or RAVI could be looked to for emission reductions to promote reasonable
progress, then States, or EPA acting as necessary in the place of a State, would have little to no
room for additional progress and even less need for sequential planning periods to build on past
progress. Source-specific emission limitations, as promulgated in our FIP, are manifestly
envisioned by the statute. We address comments alleging that we have been inconsistent in in
the consistency section of this document.
7. Constitutional Law
Comment: [Texas Governor (0066) p. 3-4] The Texas Governor stated that the EPA's FIP
violates the Commerce Clause, U.S. Const, art. I, section 8, cl. 3. The Commerce Clause gives
Congress power "[t]o regulate Commerce with foreign Nations, and among the several States,
and with the Indian Tribes." Id. According to the commenter, while modern court decisions have
expanded that text far beyond its plain or original meanings, the Clause still imposes meaningful
limits on what Congress and administrative agencies can do. As the Fifth Circuit has held:
Neither the plain language of the Commerce Clause, nor judicial decisions
construing it, suggest that. . . Congress may regulate activity (here, Cave Species
takes) solely because non-regulated conduct (here, commercial development) by
the actor engaged in the regulated activity will have some connection to interstate
commerce. ... To accept [such an] analysis would allow application of otherwise
unconstitutional statutes to commercial actors, but not to non-commercial actors.
There would be no limit to Congress' authority to regulate intrastate activities, so
long as those subjected to the regulation were entities which had an otherwise
substantial connection to interstate commerce. GDF Realty Inv., Ltd. v. Norton,
326 F.3d 622,634 (5th Cir. 2003).
The commenter asserts that this is exactly what EPA has interpreted the Clean Air Act to allow.
According to the commenter, EPA concedes that the majority of "regional haze" in Big Bend and
the Guadalupe Mountains comes from non-regulated conduct—namely, emissions from Mexico
and from natural sources (such as dust storms and fires). See 79 Fed. Reg. al 74,844
("Approximately half of the 2002 visibility impairment at Big Bend is due to Mexico and other
international sources."); Id. at 74,885 ("We agree that dust storms and other blown dust from
deserts are a significant contributor to visibility impairment at the Texas Class I areas that may
not be captured accurately by our default method."). The commenter contends that, having made
such concessions, EPA cannot then regulate "regional haze" on the theory that regulated
conduct—like carbon emissions from coal-fired power plants—will have some effect on
interstate commerce.
149
-------
Response: We disagree with this comment. Coal-fired power plants are owned by companies
engaged in interstate commerce and that release emissions substantially affecting interstate
commerce. Furthermore, modeling confirms that the haze-forming emissions from these coal-
fired plants travel to other states; they do not merely remain in Texas.
The U.S. Supreme Court has made clear that, "[w]here economic activity substantially affects
interstate commerce, legislation regulating that activity will be sustained." United States v.
Lopez, 514 U.S. 549 (1995). The Court has further held that, "the power conferred by the
Commerce Clause [is] broad enough to permit congressional regulation of activities causing air .
. . pollution . . . that may have effects in more than one State." Hodel v. Va. Surface Mining &
Reclamation Ass'//, 452 U.S. 264, 282 (1981). The owners and operators of the Texas sources
subject to this regional haze FIP are engaged in economic activities (the operation of coal-fired
power plants) that cause haze-forming air pollution to travels into other states and substantially
affect interstate commerce. Accordingly, our regulation of these sources pursuant to the Clean
Air Act is squarely within the federal government's Commerce Clause authority.
The Governor's comment seems to suggest that the regulation of emissions from coal-fired
power plants is unconstitutional because non-commercial activity {i.e., dust storms) and
international commercial activity {i.e., emissions from sources in Mexico) cause and contribute
to regional haze. Further, the Governor seems to suggest that, because of the aforementioned
contributing factors, it is an abuse of the Commerce Clause to suggest that emissions from coal-
fired power plants have an effect on interstate commerce, and goes so far as to suggest that
emissions from coal-fired power plants are purely intrastate and outside the stream of interstate
commerce.
The U.S. Supreme Court has recognized that the power of Congress over interstate commerce is
not confined to the regulation of commerce among the states, but extends to activities that have a
substantial effect on interstate commerce. See NFIB, 132 S. Ct. 2566, at 2585-56. Congress's
power is not limited to regulation of an activity that by itself substantially affects interstate
commerce, but also extends to activities that do so only when aggregated with similar activities.
Id. at 2586 (describing the "expansive scope" of Commerce Clause authority). Accordingly,
perhaps unsurprisingly, even the principal case cited by this commenter held that the Commerce
Clause extends to putatively "intrastate" activities that have an aggregated effect on interstate
commerce. See GDF Realty Inv., Ltd v. Norton, 326 F.3d 622, 651 (5th Cir. 2013).
Our regulation of emissions from coal-fired power plants, which cause and contribute to regional
haze in multiple states, seeks to fulfill the regional haze provisions of the Clean Air Act, which
in turn are constitutional exercises of Congress's power under the Commerce Clause of the U.S.
Constitution.
Comment: [Texas Governor (0066) p. 4] The Texas Governor contended that the EPA's
"regional haze" rule suffers from a non-delegation problem. The Constitution vests "[a]ll
legislative Powers herein granted ... in a Congress of the United States." U.S. Const, art. I, § 1
The commenter stated that if Congress wants to delegate its power to an administrative agency,
then Congress must "lay down by legislative act an intelligible principle to which the person or
150
-------
body authorized to [act] is directed to conform." J. W. Hampton, Jr., & Co. v. United States, 276
U.S. 394, 409 (1928). The commenter quoted a law review article in support of the assertion that
Congress cannot enact "a statute creating the Goodness and Niceness Commission and giv[e] it
power 'to promulgate rules for the promotion of goodness and niceness in all areas within the
power of Congress under the Constitution.'" The commenter cited Whitman v. Am. Trucking
Assns. 531 U.S. 457, 472 (2001), for the proposition that, where Congress violates the non-
delegation doctrine, the agency cannot "cure [the] unlawful delegation of legislative power by
adopting in its discretion a limiting construction of the statute."
The Texas Governor stated that the EPA has crowned itself the proverbial Goodness and
Niceness Commission. In the Clean Air Act, Congress "declare[d] as a national goal the
prevention of any future, and the remedying of any existing, impairment of visibility in" places
like Big Bend and the Guadalupe Mountains. 42 U.S.C. section 7491(a)(1). That is a vacuous
delegation, and EPA has exacerbated it by exercising standardless discretion to approve some
SIPs and disapprove others based on illegitimate criteria, inaccurate science, and faulty methods.
Response: We disagree with this comment. In applying the intelligible-principle test to
congressional delegations, the courts have been "driven by a practical understanding that in our
increasingly complex society, replete with ever-changing and increasingly technical problems,
Congress cannot simply do its job absent an ability to delegate power under broad general
directives." Mistretta v. United States, 488 U.S. 361, 372 (1989). Thus, for example, the Court
has explained that it is "constitutionally sufficient if Congress clearly delineates [1] the general
policy, [2] the public agency which is to apply it, and [3] the boundaries of this delegated
authority." Id. at 372-73. In other words, Congress must provide an "intelligible principle" to
guide the Executive Branch. See Whitman, 531 U.S. at 472 (citing J. W. Hampton, Jr. & Co. v.
United States, 276 U.S. 394, 409 (1928)).
The U.S. Supreme Court has only twice struck down a federal law for violating the non-
delegation doctrine—both times in 1935. Mistretta, 488 U.S. at 373-74 (recounting that
history). Of the two statutes that have failed this test, one statute "provided literally no guidance
for the exercise of discretion, and the other of which conferred authority to regulate the entire
economy on the basis of no more precise a standard than stimulating the economy by assuring
'fair competition.'"Id. at 474.
By contrast, the Court has rejected all other non-delegation-based challenges, including to
provisions of the Clean Air Act. See, e.g., Whitman v. Am. Trucking Assns., 52 U.S. 457 (2001)
(unanimous). In Whitman, the Supreme Court addressed whether our interpretation of CAA
Section 109(b)(1) violated the non-delegation doctrine. That CAA provision requires us to set
the primary NAAQS at a level "the attainment and maintenance of which in the judgment of the
Administrator, based on such criteria and allowing an adequate margin of safety, are requisite to
protect the public health." The Court agreed that section 109(b)(1) at least requires that, "for a
discrete set of pollutants and based on published air quality criteria that reflect the latest
scientific knowledge," we must set the NAAQS "at a level that is requisite [i.e., sufficient to, but
no more than necessary] to protect public health from the adverse effects of that pollutant in the
ambient air." Id. at 473. The Court held that the scope of discretion provided to us by CAA
151
-------
section 109(b)(1) was "well within the outer limits of [the Court's] non-delegation precedents."
Id. at 474.
Like the provision at issue in Whitman, the CAA's visibility provisions provide extensive
intelligible principles that guide our exercise of discretion. In addition to the statement of policy
quoted by the Texas Governor, CAA Section 169A required us to promulgate regulations
directing the States to revise their SIPs to include emission limits and other measures as
necessary to make "reasonable progress." 42 U.S.C. 7491(b)(2). Congress defined reasonable
progress to be the consideration of four statutory factors, including cost and energy impacts. Id.
at 7491(g)(1). Congress also directed EPA's regulations to require best available retrofit
technology, or BART, for a specific universe of older sources, and again provided a set of
statutory factors States must consider when determining what control technology constitutes
BART. Id. at 7491(b)(2)(A) & (g)(2). These two sets of statutory factors, among several other
provisions and definitions in CAA Section 169A that provide specific instructions to EPA and
States, clearly constitute intelligible principles under the framework set forth in Mistretta and
Whitman. The Regional Haze Rule, which we promulgated pursuant to the statutory mandate in
Section 169A, reflects these same intelligible principles and has been upheld by the D.C.
Circuit.
Comment: [Texas Governor (0066) p. 4-5] The Texas Governor explained that the EPA has
commandeered the States in violation of the Fifth Amendment. "[T]he question whether the
Constitution should permit Congress to employ state governments as regulatory agencies was a
topic of lively debate among the Framers" - and the Framers emphatically rejected the idea. New
York v. United States, 505 U.S. 144, 163 (1992). Thus, in New York, the Court invalidated a
statute that purported to give the States "latitude ... to implement Congress' plan" for disposing
of nuclear waste. Id. at 176. In particular, the statute at issue gave the States a "choice" to either
take title to the waste or to enact a series of state regulations. But the Court held that was no
"choice" at all because "[n]o matter which path the State chooses, it must follow the direction of
Congress." Id. at 177; see also, e.g., NFIB v. Sebelius, 132 S. Ct.2566, 2601-05 (2012); Printz v.
United States, 521 U.S. 898,926(1997).
The Texas Governor explained that, similarly, the EPA has purported to offer the State a
"choice" between two unpalatable and federally controlled outcomes. First, the State can submit
a SIP that EPA will scrutinize like a teacher grading a pupil's exam answers, approving some and
disapproving others. By turning the SIP-FIP process into a paper-grading exercise, EPA has
effectively turned the States into subordinate administrative agencies-in direct contravention of
the Framers' constitutional design. See New York, 505 U.S. at 163. Second, the State can forgo
a SIP and face draconian penalties-including the loss of highway funds, loss of support for air
pollution planning and control programs, and so-called "offset penalties." See 42 U.S.C. section
7509. Moreover, if the State chooses to forgo the SIP process, the statute (and EPA's
implementation of it) blurs the accountability for clean-air regulations by making it appear that
the State is somehow responsible for not staving off EPA's draconian response. See Printz, 521
U.S. at 929-30 (Tenth Amendment forbids statutory schemes that shift costs and perceived
responsibilities to the States). That is precisely the sort of coercion that the Tenth Amendment's
anti-commandeering principle forbids. See NFIB, 132 S. Ct. at 2601-05.
152
-------
Response: We disagree with this comment. The U.S. Supreme Court has held that, "the Federal
Government may not compel the States to implement federal regulatory programs." Printz v.
United States, 521 U.S. 898, 925 (1997). The CAA in no way compels a state to implement
federal regulatory programs. The CAA, instead, authorizes us to promulgate and administer a
FIP if a state fails to submit an adequate SIP. 42 U.S.C. 7410(c)(1)(A). The adequacy of Texas'
regional haze SIP is discussed elsewhere in this response to comments.
The Supreme Court has "repeatedly affirm[ed] the constitutionality of federal statutes that allow
States to administer federal programs but provide for direct federal administration if a State
chooses not to administer it." See Texas v. EPA, 726 F.3d 180, 196-7 (D.C. Cir. 2013) (citing
New Yorkv United States, 505 U.S. 144, 167-8, 173-4 (1992); Hodel v. Va. Surface Mining &
Reclamation Ass % Inc., 452 U.S. 264m 288 (1981)). If a state chooses not to submit a SIP, the
"full regulatory burden will be borne by the Federal Government." Va. Surface Mining
&Reclamation Ass'//, 452 U.S. at 288.
In National Federation of Independent Business v. Sebelius (NFIB), the Court struck down a
provision of the Affordable Care Act that expanded the scope of the Medicaid program and
increased the number of individuals that the states had to cover. 132 S.Ct. 2566, 2582 (2012).
In NFIB, states refusing to expand Medicaid risked losing all existing federal Medicaid funds—
over 10-percent of some states' budgets. Id. at 2604-05. By contrast, in South Dakota v. Dole,
the Court upheld a federal law that would withhold 5-percent of one state's highway funding
unless the state raised its drinking age to 21. See generally Dole, 483 U.S. 203 (1987). Unlike
the figurative "gun to the head" that was the threat of losing all Medicaid funding in NFIB, the
Court in Dole found the potential loss of federal funds to be merely be "mild encouragement"
that the state functionally had a choice to reject. See NFIB, 132 S.Ct. at 2604.
Here, we note that we have never imposed sanctions in the regional haze program. Furthermore,
the mandatory sanctions set forth in part D of the CAA apply only where a state fails to make a
required SIP submittal, we disapprove a SIP submittal applicable to a nonattainment area, or we
make a finding of failure to implement any approved part of a SIP. This is not the situation here.
Texas made the required SIP submittal; this required RH SIP submittal is not a requirement for
nonattainment areas; and we are not finding the Texas RH SIP is not being implemented. See
42 U.S.C. 7509(b).
8. Stay of Effective Date, Consolidated Appropriations Act, and Executive Orders
Comment: [GCLC (0063) p. 1-2, 20]
GCLC contests the legality of EPA's proposed action, but should EPA finalize it, it requests that
EPA establish a sufficiently delayed effective date for the final rule and its compliance timelines
to allow judicial vetting of EPA's novel legal theory that it can compel such extraordinary actions
from a class of Texas operators who are BART-compliant, BART-exempt, or both. The
irreparable harm that will befall operators and the State of Texas as result of EPA's proposal
demands such forbearance.
153
-------
Response: We understand this comment to request that any final rule imposing controls on
Texas sources also set effectiveness and compliance dates far enough into the future to allow for
judicial review of any petitions challenging those controls before implementation. We are
obliged to give effect to the requirements of the Clean Air Act, and those requirements often—as
here—implement controls to carry out the objectives of the Act. Because we have demonstrated
all the controls to be cost-effective in this case, and because the comment offers no information
that viability of continued operations is in issue for any specific operators, we cannot accede to
what the comment asserts regarding "irreparable harm." Ordinarily, having to comply with a
government regulation, while it can entail costs, does not equate with irreparable harm.
The comment identifies no specific authority for delaying the effectiveness of our final action
outside the customary period. Customarily, EPA rules under the Clean Air Act take effect within
30 to 60 days of publication in the Federal Register. This convention accords with practices as
established by the Administrative Procedure Act (5 U.S.C. § 553(d)). Moreover, as provided by
Clean Air Act section 307(d)(7)(B), we note that even the Administrator's reconsideration of a
rule (according to appropriate procedural formalities) would only allow a stay of effectiveness
"not to exceed three months"; this counsels against any exceptional measure to delay
effectiveness of the final rule as requested by the comment.
Our timing in finalizing the rule and taking action satisfies a judicially supervised consent decree
deadline. Nothing in that consent decree dictates the substance of our action, but we are mindful
of our obligation to ensure that the applicable requirements of the CAA are carried into effect.
Although the CAA expressly provides an opportunity for judicial review of this particular final
rule and action (see CAA section 307(b)), we find no basis of authority for granting the unusual
request made in this comment. Given that the filing of a petition for reconsideration of a pending
rule does not "affect the finality" or "postpone the effectiveness" of a rule pending judicial
review, see CAA section 307(b)(1), it would also appear to be inappropriate to institute a delay at
the outset for an identified purpose of "judicial vetting," which we note the comment does not
define. Accordingly, we disagree with the comment. As noted in the language of the Federal
Register, our action and final rule is effective 30 days from publication. As to the requested
extension of the compliance dates for controls, the compliance dates are feasible and reasonable,
and we conclude that adjusting those dates to allow for judicial vetting would be irregular and
not appropriate.
Comment: [Luminant (0061) p. 147] Luminant contests the legality of EPA's proposal. Should
EPA proceed to finalize its proposal, Luminant requests that EPA stay the effective date of its
final rule and the compliance dates for Texas sources pending judicial review, given the
immense expense and negligible benefit associated with EPA's proposal. Since the Class I areas
have already met EPA's visibility goals for 2018 and because new emission limits for these very
units are now in place by virtue of EPA's CSAPR and MATS, the public interest weighs heavily
in favor of staying EPA's rule pending judicial review of EPA's admittedly new regulatory
interpretations.
Response: We understand this comment to request that we delay the effective date of the final
rule and the compliance dates for Texas sources until judicial review of any petitions challenging
154
-------
the final rule are complete. We are not aware of any precedent for such an accommodation, and
we believe the comment is deficient for not providing or citing any basis of authority and
justification for instituting a delay "pending judicial review." We note that even when we grant
a petition for administrative reconsideration of a finalized rule, that process does not "postpone
the effectiveness of the rule." CAA Section 307(d)(7)(B). Moreover, granting an administrative
petition for reconsideration only allows possibility for a stay of the effectiveness of the rule "for
a period not to exceed three months." This statutory limitation on staying the effectiveness of
the rule applies to both the Administrator and a court, see CAA Section 307(d)(7)(B), and
counsels against our ability to find specific authority in the CAA to delay or stay effectiveness of
the requirements of the CAA that would be implemented by our final rule and action. We also
note that even the Congressional Review Act (5 U.S.C. 801-808), which empowers Congress to
overrule certain rules, envisions that those rules would take effect after 60 days absent
Congressional action. Accordingly, what the comment requests departs significantly from
established and customary practices in carrying out the rulemaking objectives set by Congress.
The request, if granted, would also put into question the finality of our action or, at least, make
the applicable regional haze requirements indefinite in a way that is contrary to established air
quality planning processes. Of related note, we have previously disapproved part of a SIP
provision that purported to take effect upon EPA's disapproval of other SIP-specified controls
and after the exhaustion of all administrative and judicial appeals following that disapproval.
With that disapproval (incidentally, a disapproval to part of the Oklahoma RH SIP), we
explained that the "contingent SIP" was "predicated on speculative actions and outcomes of
review by EPA and courts, and [did] not comport with established SIP planning and approval
processes under the CAA." 76 FR 81737. Similarly, here, where we consider our final action to
carry out certain due, if not overdue, requirements of the CAA, it is not appropriate that we delay
the effectiveness of those requirements by a mechanism that incentivizes litigation and
perpetuates the specific environmental harms that are directed to be remedied under the CAA.
For this reason, even as we note and are aware that 5 U.S.C. 705, may provide a basis for
exercising stay authority in certain situations, we would not invoke it here, because 1) there is no
pending litigation, and 2) justice does not, in fact, require that we delay the implementation of
the applicable requirements of the CAA at issue in our rulemaking. Moreover, to the extent the
comment invokes or alludes to the "public interest" or other factors that are conventionally
considered by Courts on a judicial application for a stay of an agency action, we disagree that
those factors are demonstrated90 or that it is appropriate that we engage them here and now.
Among other things, since we consider our final action to be reasoned and defensible upon our
thorough consideration of public comments, we have no known basis to think a legal challenge
to our rule would be likely to succeed. We thus believe it is unwarranted and contradictory to
make the effectiveness and compliance dates on the rule, at its very outset, subject to
postponement and non-effectiveness pending Court review. Accordingly, we disagree with the
comment and cannot grant the request sought by the comment.
90 For example, putting aside any inaccuracy in the commenter's contention about controls that are "in place" for
MATS or CSAPR, any such controls would be required by other provisions of the CAA. It is in the public interest
that all CAA requirements are carried into effect.
155
-------
Comment: [Page 4 of Edison Electric Comments] Comment on the Consolidated Appropriations
Act of 2014. See H.R. 3547, P.L. 113-76 [comment from EEI]
EEI noted that as EPA and the states begin to implement the next rounds of the regional haze
program to continue reasonable progress ... it will be necessary to use the most up-to-date and
accurate implementation tools available as EPA and the states move forward with the regional
haze reasonable progress determinations. To that end, EPA should update both its atmospheric
modeling platforms as part of the upcoming Appendix W rewrite and the cost manual in order to
support reasonable future assessments of visibility impacts and appropriate control strategies.
The Agency should do this consistent with the Consolidated Appropriations Act of 2014. See
H.R. 3547, P.L. 113-76. EPA also should consider the latest available visibility monitoring data
for Class I areas when assessing potential controls needed to meet RPGs.
Response: While not specific to this final action, but rather for future planning and
implementation of reasonable progress determinations, we agree that we should use the most up-
to-date and accurate implementation tools available for future reasonable progress
determinations. And, while not specifically referenced in the comment, we assume updates to
the atmospheric modeling platforms and cost control manual relate to the committee report that
accompanies the Appropriations Act of 2014, as copied below:
Regional Haze. — The process for reviewing State implementation plans is well-
served when EPA, States, and industry work collaboratively to ensure that
dispersion models are continually improved and updated to ensure the most
accurate predictions of visibility impacts, as well as a uniform set of cost
estimates. To that end, EPA shall begin development of a seventh edition of the
document entitled "EPA Air Pollution Control Cost Manual." The Administrator
shall consult, and seek comment from State, local, and tribal departments of
environmental quality during development of such seventh edition, and provide
opportunity for public comment. In addition, EPA shall publish in the Federal
Register a notice to solicit comment on revising the Agency's "Guideline on Air
Quality Models" under appendix W to part 51 of title 40, Code of Federal
Regulations, to allow flexible modeling approaches and to adopt updates to the
CALPUFF modeling system (or portions thereof) or other modeling tools as may
be appropriate under such Guideline. Within six months of enactment of this Act,
if EPA finds the requirements above cannot be accomplished without causing
delay in the approval of State implementation plans, the Agency shall certify such
to the Committees. The certification from EPA shall include documentation on
how the directives would cause delay in a particular State and also an estimate of
when the directives can be carried out without causing delays in the program. 160
Cong. Rec. H475, H979 (Jan. 15,2014). (Full committee report for 2014:
https://www.congress.gov/crec/2014/01/15/CREC-2014-01-15-pt2-PgH475-
2.pdf).
As a general matter, wherever possible, EPA intends to follow such committee report
instructions even where not specifically incorporated by reference into the Act itself. We are
currently working to update the Agency's "Guideline on Air Quality Models" under appendix W
156
-------
to part 51 of title 40, Code of Federal Regulations and as of the date of responding to this
comment, we have also proposed updates to chapters with the Control Cost Manual. As to
updates to Appendix W, we proposed updates on July 29, 201591
As to the comment regarding consideration of the latest available visibility monitoring data for
Class I areas when assessing potential controls needed to meet reasonable progress goals, we
address that elsewhere.
Comment: Executive Orders 12898,12866, and 13045 [Earthjustice (0067) p. 17] and
Commenter 0053-53
Two commenters focused on the importance of EPA including children, minority, underserved,
vulnerable, and low-income populations in those receiving air health benefits. Commenter
Earthjustice et al., agrees that the FIP, as proposed, is not subject to either Executive Order
12866 or Executive Order 13045 because it will result in environmental and public health
benefits to affected populations. Earthjustice noted that that Executive Order 12898 establishes
federal executive policy identify and address, as appropriate, disproportionately high and adverse
human health or environmental effects of their programs, policies, and activities on minority
populations and low-income populations in the United States. Similarly, Executive Order 13045
establishes a policy of evaluating and addressing any rule that: (1) Is determined to be
economically significant as defined under Executive Order 12866; and (2) concerns an
environmental health or safety risk that we have reason to believe may have a disproportionate
effect on children. Earthjustice stated that if the proposal was changed in the final rule to
conclude that the proposed SO2 reductions at one or more of the affected facilities are not
necessary to assure reasonable progress, EPA must conduct an analysis of the impacts of such a
reversal on environmental justice populations and children. Citing Dr. Thurston's analysis as
demonstrating that emissions from these 14 sources disproportionately impact metropolitan areas
throughout the central United States, including Houston, Dallas, Tulsa, Oklahoma City, and even
Chicago and St. Louis,92 and emphasizing that those urban areas are likely to have higher
populations of children, minority, and low-income populations, Earthjustice stated that EPA
would be required to re-evaluate the impacts to those populations from a less protective rule.
Another Commenter (0053-56) suggested that polluters need to reconsider a business model that
straps low income communities and, more specifically, low income communities of color with
the [air pollution] burdens required for them to profit, and that EPA is accountable to the many
communities in Texas that are constantly being ignored, which are usually low income,
underserved, and vulnerable communities. Commenter 0053-56 stated that air is a basic human
right that does not know class, color, or success.
Response: Thank you for your comments. Since we are finalizing emission limits as proposed
at affected facilities to assure reasonable progress, our reasoning remains that Executive Order
13045 does not apply. To the extent this proposed rule will limit emissions of SO2, the final rule,
like the proposed rule, will have a beneficial effect on children's health by reducing air pollution.
Executive Order 13045: Protection of Children From Environmental Health Risks and Safety
91 80 FR 45340
92 Thurston Report, Table 4.
157
-------
Risks applies to any rule that: (1) Is determined to be economically significant as defined under
Executive Order 12866; and (2) concerns an environmental health or safety risk that we have
reason to believe may have a disproportionate effect on children. This action is not subject to
Executive Order 13045 because we do not believe the environmental health or safety risks
addressed by this action present a disproportionate risk to children. Moreover, "regulation" or
"rule," is defined in Executive Order 12866 as "an agency statement of general applicability and
future effect." E.O. 12866 does not define "statement of general applicability," but this term
commonly refers to statements that apply to groups or classes, as opposed to statements which
apply only to named entities. The proposed FIP therefore is not a rule of general applicability
because its requirements apply and are tailored to only eight individually identified facilities.
Thus, it is not a "rule" or "regulation" within the meaning of E.O. 12866.
The proposed rule, to the extent it will limit emissions of SO2, will also increase the level of
environmental protection and beneficial effect on human health for all affected populations
without having any disproportionately high and adverse human health or environmental effects
on any population, including any minority or low-income population.
Comment: [TCEQ/PUCT (0056) p. 20] Commenter (TCEQ) contends that EPA has improperly
avoided analyzing and evaluating potential energy impacts of the proposed rule on reliability and
prices of electricity in Texas and the ERCOT region, despite Executive Order 13211 requiring
such evaluation. The TCEQ contended that the EPA is using a loophole in Executive Order
12866 to avoid evaluating the potential energy impacts of the proposed action as required by
Executive Order 13211. TCEQ stated that the proposed FIP affects a significant portion of
Texas' base load power generation fleet and the EPA should evaluate and consider the impacts of
the proposed FIP on the reliability and price of electricity in Texas.
TCEQ states that the proposed rule is inconsistent in claiming the rule is both of national scope
and effect and not of general applicability. The TCEQ takes issue with EPA's determination that
the rule is not of general applicability to such that Executive Orders 12866 and 13211 do not
apply, while at the same time determining that the rule is of nationwide scope and effect in an
effort to have any petitions for review be filed in the United States Court of Appeals for the
District of Columbia (79 FR 74888). Specifically, TCEQ contends that EPA determined that the
rule is of national scope for purposes of legal challenges, but then also determined that the rule is
of limited scope for the purposes of avoiding Executive Orders 12866 and 13211 without any
explanation of how this action can have two contradictory scopes. The TCEQ asserts that the
scope of the regulatory action proposed by the EPA must be either nationwide or limited to
Texas; it cannot be both.
Specifically, the TCEQ disagreed with the EPA's position that the proposed action is of
nationwide scope (79 FR 74888). However, the TCEQ also disagreed with the EPA position that
the potential impact to the supply, distribution, and use of energy does not need to be considered
in this proposed action. According to TCEQ, the annualized cost for the scrubber retrofits
portion of the proposal is estimated to be approximately $238 million per year, greatly exceeding
the $100 million per year threshold established under Executive Order 12866. Furthermore, the
TCEQ asserts that EPA's proposed FIP would meet Executive Order 13211 criteria for being
158
-------
"likely to have a significant adverse effect on the supply, distribution, or use of energy" based on
the guidance provided in Office of Management and Budget (OMB) Memoranda 01-27, July 13,
2001 Guidance for Implementing Executive Order 13211. Section 4 of the OMB Memoranda
01-27 provides a number of examples of adverse effects for the purpose of Executive Order
13211. One of the listed examples is a reduction in electricity production in excess of 1 billion
kilowatt-hours or in excess of 500 megawatts (MW) of installed capacity. According to a recent
ERCOT report included in Appendix 1 to the TCEQ's comments, ERCOT's modeling indicates
that approximately 1,800 MW of capacity from the affected coal-fired EGUs are expected to
retire due to the EPA's proposed Regional Haze FIP requirements, exceeding the threshold in the
OMB guidance for an adverse effect.18 Also, with the exception of the San Miguel facility, each
of the units subject to the EPA's proposed FIP is greater than 500 MW. If just one of these units
is no longer economically viable as a result of the EPA's FIP, it would result in the reduction of
more than 500 MW of installed capacity.
The TCEQ stated, according to OMB Memoranda 01-27, the basic purpose of Executive Order
13211 is to ensure that agencies "appropriately weigh and consider the effects of the Federal
Government's regulations on the supply, distribution, and use of energy." The EPA's
interpretation of Executive Orders 12866 and 13211 would mean that a national rule applying to
all coal-fired EGUs in the country with an annualized cost of $100 million per year that might
result in the loss of only 500 MW of a capacity would require an energy impact analysis because
it may have a significant adverse effect on the supply, distribution, or use of energy. However,
according to the EPA's interpretation, a rule costing more than twice that cost threshold and
potentially resulting in the loss of more the three times the capacity but focused within a discrete
electric reliability region in a single state that has limited connections to the rest of the United
States' grid does not require any analysis or consideration of the possible adverse impacts on
energy. In other words, the EPA's position is that the Federal Government does not need to
concern itself with a potentially severe impact of this proposed rule on the supply, distribution,
or use of energy within ERCOT because the impact is limited to a single state. Such an
interpretation and outcome is illogical and clearly contrary to the stated intent of Executive Order
13211. The potential for adverse effects from the EPA's proposed rule is actually increased, not
lessened, because the costs and impacts of the rule are focused within a smaller region.
Additionally, the TCEQ noted that CAA §169A(g) requires that the State and the Administrator
consider the energy and non-air quality environmental impacts of compliance when determining
the best available retrofit technology. While the EPA's guidance on evaluating energy impacts
for BART analyses does not specifically address considering electrical grid reliability and
electricity prices, the guidance does make allowance for considering indirect energy impacts as
well as potential impacts such as locally scarce fuels and significant economic disruption or
unemployment (70 FR 39169). Furthermore, the EPA recommends that states consider the
BART guidelines when evaluating the energy and non-air environmental impacts for reasonable
progress goal purposes. 19
The TCEQ stated that the proposed action affects almost 10,000 MW of generation capacity in
Texas and almost 8,8oo MW of that capacity is within the ERCOT region. The affected units in
ERCOT represent approximately 11% of region's 2015 total capacity based on ERCOT's Report
on Capacity, Demand, and Reserves for the ERCOT Region, 2015 - 2024.20 Based on the
159
-------
significant portion of the Texas electrical grid affected by the EPA proposal and the projected
retirements estimated by ERCOT to result from this action, the EPA should analyze and consider
the possible impacts of the proposed rule on the reliability and prices of electricity in Texas,
regardless of the applicability of Executive Orders 12866 and 13211.
Footnote:
18 See ERCOT Report Impacts of Environmental Regulations in the ERCOT Region, December 16, 2014, page 27
(Attached to comment 0056).
19 See Guidance for Setting Reasonable Progress Goals under the Regional Haze Program, June 1, 2007, page 5-3;
79 FR 74874.
20 See http://www ercot.com/gridiufo/resource; December 1, 2014.
Response: As explained in our final action and below, Executive Order 13211 does not apply as
this action is not a rule of general applicability under Executive Order 12866. This is not
inconsistent with our determination that the rule is of national scope and effect, as these are
different determinations that we fully evaluated under their respective standards, and are not
directly comparable. Consequently, we did not "utilize a loophole" in the applicability
provisions of Executive Order 12866 to avoid consideration of the concerns raised in this
comment.
Specifically, we do not agree that our proposed FIP is avoiding the requirements of any
applicable Executive Orders (E.O.s) or statutes, or that the proposed FIP is inconsistent or
contradictory in determination of scope. This proposed action is not subject to Executive Order
13211 because it is not a "significant regulatory action" under Executive Order 12866 (58 FR
51735, October 4, 1993). Under E.O. 12866, "Regulatory Action" is defined as "any
substantive action by an agency . . . that promulgates or is expected to lead to the promulgation
of a final rule or regulation." "Regulation" or "rule," in turn, is defined as "an agency statement
of general applicability and future effect." E.O. 12866 does not define "statement of general
applicability," but this term commonly refers to statements that apply to groups or classes, as
opposed to statements which apply only to named entities. The proposed FIP therefore is not a
rule of general applicability because its requirements apply and are tailored to only eight
individually identified facilities. Thus, it is not a "rule" or "regulation" within the meaning of
E.O. 12866 and this action is not a "regulatory action" subject to 12866. Since E.O. 13211 (79
FR 74890) applies only to "significant regulatory actions" under E.O. 12866, this action is not
subject to review under E.O. 13211. Evaluation of the proposal under E.O. 13211 's criteria is
therefore not required.
A determination of whether the proposal is of "nationwide scope or effect" requires a specific,
separate statutory analysis than a determination of whether the proposal is a "rule" or
"regulation" of "general applicability" under E.O. 12866. As discussed in detail elsewhere,
Section 307(b)(1) of the CAA provides, in part, that petitions for review must be filed in the U.S.
Court of Appeals for the District of Columbia Circuit when such action is locally or regionally
applicable if "such action is based on a determination of nationwide scope or effect and if in
taking such action the Administrator finds and publishes that such action is based on such a
determination."93 Section 307(b)(1) reflects Congress's intent that some local and regional
actions should be reviewed in the D.C. Circuit only. Additionally, in the 1977 CAA
93 42 U.S.C. Section 7607(b)(1).
160
-------
Amendments that revised CAA Section 307(b)(1), Congress noted that the Administrator's
determination that an action is of "nationwide scope or effect" would be appropriate for any
action that has "scope or effect beyond a single judicial circuit."94 First, this final rule involves
two States in two separate judicial circuits and, as the EPA has explained in detail above and
elsewhere, the final rule includes statutory and regulatory interpretations that are nationally
significant. Second, for judicial review to lie in the D.C. Circuit, the EPA must also make an
affirmative determination and publish that finding, which we did here. Consequently, EPA's
determination of nationwide scope and effect is "consistent with the legislative history of the
CAA, which evinces a clear congressional intent to 'centralize review of 'national' SIP issues in
the D.C. Circuit.'"95 Contrastingly, as discussed above, the proposed FIP is not a rule of general
applicability because its requirements apply and are tailored to only eight individually identified
facilities, and thus is not a "rule" or "regulation" within the meaning of E.O. 12866.
Accordingly, this proposal is a rulemaking of nationwide scope or effect, while not being a
"rule" or "regulation" of "general applicability."
Furthermore, as explained in our final action, while the E.O. 13211 does not apply here, as this
action is not a rule of general applicability under Executive Order 12866, EPA did consider
TCEQ's concerns regarding grid reliability and price of electricity, as discussed more fully
elsewhere and in our final action, so EPA did not utilize a loophole in Executive Order
applicability to avoid consideration of the concerns in this comment.
The TCEQ is correct that the BART Guidelines does not require that grid reliability or energy
prices be considered, but that energy impacts should be considered. Consideration of "the
energy and non-air quality environmental impacts of compliance," also falls under Section
51.308(d)(l)(i), which is within the reasonable progress portion of the Regional Haze Rule that
governs the subject cost analysis. Our Reasonable Progress Guidance considers this factor from
the standpoint of whether it would cause the reconsideration of the control in question due to
energy or non-air quality penalties. In other words, the energy and non-air quality environmental
impacts of compliance serve as a potential modifier to the particular control being considered.
This modifier does not consider grid reliability or electricity pricing, but rather the potential for
additional costs due to energy and non-air quality considerations. The "energy" portion of this
factor, refers to potential costs due to the additional consumption of energy or switching to
another energy source,96 not grid reliability or electricity pricing. The only facility in which
energy and non-air quality consideration was relevant was Tolk, due to our consideration of
potential water limitations (addressed in our response to other comments).
94H.R. Rep. No. 95-294 at 323-24, reprinted in 1977 U.S.C.C.A.N. 1402-03.
95 Texas v. EPA, 2011 U.S. App. LEXIS 5654 at *15 (5th Cir. Feb. 14, 2011) (citing Admin. Conference of the
U.S., Recommendations on Judicial Review Under the Clean Air Act, 41 FR 56767, 56769 (Dec. 30, 1976)
(Comments of G. William Frick)).
96 See our Reasonable Progress Guidance, page 5-2: "The third factor is "energy and non-air environmental
impacts." In assessing energy impacts, you may want to consider whether the energy requirements associated with a
control technology result in energy penalties. For example, controls on diesel engines may decrease the engine's
fuel efficiency, leading to an increase in diesel fuel consumption. Or, a particular control may require a fuel
unavailable in the area. To the extent that these considerations are quantifiable they should be included in the
engineering analyses supporting compliance cost estimates.
161
-------
Notwithstanding our Executive Order comments above, we nevertheless fully considered the
commenter's concerns regarding grid reliability. As discussed more fully elsewhere, we note
that controls achieving the level of control that we are requiring are highly cost-effective, are in
wide use in the industry, and thus should not require a source to shut down to comply. However,
we contracted with Synapse Energy Economics, Inc., a nationally recognized firm with particular
expertise in the subject area to review ERCOT's report. We reviewed and accept our
contractor's finding and adopt its conclusion that ERCOT's report contained significant flaws
and does not support a determination that there is likely to be any significant, adverse effect on
the supply, distribution, or use of energy. During our comment period, we received no, non-
speculative information to validate claims that sources would retire rather than install
demonstrably cost-effective controls. Commenters who have alleged grid reliability concerns in
response to our proposed controls have not provided adequate documentation for their assertions.
9. Controls in Addition to CAIR/CSAPR, and CSAPR Better than BART
Comment: EPA is unlawfully attempting to double-burden sources already complying with
BART requirements and attempts to apply beyond-BART requirements to sources that are
explicitly exempted from source-specific BART requirements. [GCLC (0063) p. 3-5]
GCLC noted, in the context of BART, there are two types of EGUs in Texas: 1) those EGUs that
currently comply with BART requirements and are fully controlled under the BART provisions
through compliance with the Cross-State Air Pollution Rule ("CSAPR"), and previously the
Clean Air Interstate Rule ("CAIR"); and 2) those EGUs that are excluded from BART criteria
due to the specific age exclusions (or other exclusions) found in the CAA.
GCLC stated that EPA has attempted a regulatory end-around in this FIP proposal by attempting
to impose overly burdensome beyond-BART requirements via the reasonable progress goal
("RPG") and long-term strategy ("LTS") provisions of the CAA to units that have either already
achieved BART compliance through compliance with CAIR and then CSAPR and/or are
statutorily excluded from BART requirements.
GCLC noted that Texas' EGUs are currently subject to CSAPR, which EPA has already found to
be "better-than-BART."12 Since these units comply with, and actually exceed, BART limitations,
the CAA does not contemplate or require that states impose additional controls on these units as
they have already exceeded relevant statutory requirements. 13 This is not just the position of
GCLC, but is EPA's own stated position. As stated in the Reasonable Progress Guidance, "it is
reasonable to conclude that any control requirements imposed in the BART determination also
satisfy the RPG-related requirements for source review in the first RPG planning period. Hence,
[a State] may conclude that no additional emissions controls are necessary for these sources in
the first planning period." 14 Indeed, we have found no instance where EPA has imposed a FIP
for the first planning period requiring additional reasonable progress controls on EGUs that
relied on implementation of CAIR or CSAPR to satisfy BART. 15 Therefore, by EPA's own
admission and prior regional haze SIP approvals, EPA has already pre-emptively found that
Texas' SIP submission regarding BART-subject EGUs is sufficient for the first planning period.
162
-------
Therefore, EPA is without basis to require additional controls on those units already fully
meeting and exceeding BART requirements via CSAPR compliance.
GCLC stated, regarding non-BART units, Congress, acknowledging the specified and
burdensome nature of BART requirements, provided EPA the authority to impose those source-
specific limitations in only limited instances. Specifically, BART could only be applied to those
units that were "in existence on August 7, 1977, but which ha[d] not been in operation for more
than fifteen years as of such date." 16 By providing this limitation, Congress was ensuring that
unit-specific BART limitations could only be applied to those selected units. Despite this
limitation, EPA is proposing a rule that imposes beyond-BART limitations, through source-
specific SO2 emissions limits, on specifically exempted EGUs in direct contravention of the
CAA.
Furthermore, GCLC stated that the CAA includes separate definitions for determining what is
"reasonable progress" and for determining what is "best available retrofit technology." 17 Though
similar, by drafting two different definitions, the intent of the Congress is clear that these terms
are to be applied separately and distinctly. If Congress intended the regional haze provisions to
be applied in a way to require BART limitations on non-BART units, let alone limitations that
are even more strenuous than BART, as EPA's proposal here would do, Congress would have
said so. This fact, and the limitations that Congress put on EPA's actions, have been ignored by
EPA in this Proposed FIP. EPA must withdraw its proposed emission limits on both the BART
and non-BART EGUs.
Footnotes:
12 Regional Haze: Revisions to Provisions Governing Alternatives to Source-Specific Best A vail able Retrofit
Technology (BART) Determinations, Limited SIP Disapprovals, and Federal Implementation Plans, Final Rule, 77
Fed. Reg. 33,642, 33,648 (June 7, 2012). GCLC supports EPA's finding that compliance with CSAPR satisfies the
BART requirements but also strongly believes that complying with CAIR adequately satisfied BART, as well.
Given still some lingering uncertainties regarding CSAPR, due to legal challenges, if in a future legal action or by
EPA's own volition, CS APR's application is delayed, remanded, or vacated nationally or within Texas, EPA must
recognize that compliance with CAIR- or whichever avenue the Court chooses to impose following that decision -
automatically be recognized as compliance with BART.
13 See 42 USC§ 7491(b)(2)(A).
14 Memorandum from William L. Wehrum, Acting Assistant Administrator to Regional Administrators, EPA
Regions 1-10, Guidance for Setting Reasonable Progress Goals under the Regional Haze Program, 4-2-4-3 (June 1,
2007); see also id. at 5-1, stating that "the significant emissions reductions that we anticipate to result from BART,
the CAIR, and the implementation of other CAA programs ... may be all that is necessary to achieve reasonable
progress in the first planning period for some States."
15 See 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77 Fed.
Reg. 34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098 (March
30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26, 2012)
(Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012) (Ohio); 79
Fed. Reg. 24,340 (April 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South Carolina); 77 Fed.
Reg. 24,392 (April 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77 Fed. Reg. 16,937
(March 23, 2012) (West Virginia).
16 42 USC§ 7491(b)(2)(A).
17 Id. at§ 7491(g)(l-2).
Response: We acknowledge the background information provided by the GCLC, but we take no
position with respect to it. We note that BART is a part of reasonable progress but a BART
163
-------
determination is not a shield from additional review under the reasonable progress and long-term
strategy provisions of the Regional Haze Rule, specifically CAIR/CSAPR better-than-BART
credit:97
Our determination that CAIR makes greater reasonable progress than BART for
EGUs is not a determination that CAIR satisfies all reasonable progress
requirements in CAIR affected States. Each State, whether in the CAIR region or
not, is required to set reasonable progress goals for each Class I area within the
State as required in regional haze rule section 308(d)(1), and to develop long term
strategies, considering all anthropogenic sources of visibility impairing pollutants,
as required by section 308(d)(3).
GCLC references our Reasonable Progress Guidance. We disagree that this guidance exempts
BART sources from being analyzed under reasonable progress. The reasonable progress
provisions must still be followed. If a state desires to remove a BART source from consideration
under reasonable progress, it must do so according to our regulations. Our Regional Haze Rule
itself makes that clear:98 "After a State has met the requirements for BART or implemented
emissions trading program or other alternative measure that achieves more reasonable progress
than the installation and operation of BART, BART-eligible sources will be subject to the
requirements of paragraph (d) [reasonable progress] this section in the same manner as other
sources."
We disagree with GCLC that we are either imposing BART analysis on these sources in question
or that we are otherwise attempting to circumvent BART with reasonable progress.
Concerning the legal uncertainties of CSAPR, although we proposed to rely on CSAPR to
address the BART requirements for EGUs in Texas, it is not finalizing that proposed action. On
July 28, 2015, the D.C. Circuit Court's issued its decision inEMEHomer City Generation v.
EPA, 795 F.3d 118 (D.C. Cir 2015), upholding CSAPR but remanding without vacating a
number of the Rule's state emissions budgets. Specifically, the court invalidated a number of the
Phase 2 ozone season NOx budgets and found that the SO2 budgets as to four states resulted in
overcontrol for purposes of Section 110(a)(2)(D). Texas' ozone season NOx budget and its SO2
budget are both implicated in this remand. EPA is in the process of acting on the Court's
remand. As a result, at this time we cannot ensure that CSAPR will continue to be an
appropriate alternative to BART for Texas EGUs. However, in the absence of CSAPR,
compliance with CAIR is not compliance with BART. EPA has already issued a limited
disapproval of the Texas regional haze SIP in 2012 because of Texas' reliance on CAIR to meet
certain requirements of the regional haze program. 77 FR 33642. Given the uncertainty arising
from the remand of Texas' CSAPR budgets, we have concluded that it would not be appropriate
to finalize our proposed determination to rely on CSAPR as an alternative to SO2 and NOx
BART for EGUs in Texas at this time. Should we determine in the future that it is necessary to
perform source-specific BART determinations for these sources instead of relying on CSAPR,
we anticipate that the SO2 controls we are finalizing today, which are currently the most stringent
available, will also be sufficient to satisfy the SO2 BART requirement.
97 70 FR 39143.
98 Section 51.308(e)(5).
164
-------
Concerning the non-BART units, we disagree that the source-specific SO2 emissions limits we
are requiring on non-BART EGUs is in direct contravention of the CAA. The CAA allows for
source-specific controls as appropriate to meet the reasonable progress requirements on any
source, regardless of its BART eligibility. CAA Section 169A(b)(2). Further, EPA has imposed
similar controls in other states as necessary to ensure the reasonable progress requirements are
met. See e.g. 79 FR 52420 (Arizona); 79 FR 5032 (Wyoming); and 77 FR 20894 (North
Dakota).
We address GCLC's allegations of inconsistencies in the consistency section of this document.
Comment: Background on alternatives to BART. [Earthjustice (0067) p.8]
Earthjustice et al., stated that, under the statute and EPA's implementing regulations, the default
approach to meeting the BART requirements is for a state to consider the five statutory factors
on a case-by-case basis. See 42 U.S.C. § 7491(b)(2)(A) (requiring BART "for each major
stationary source"); 40 C.F.R. § 51.308(e)(l)(ii)(A) ("[t]he determination of BART must be
based on an analysis of the best system of continuous emission control technology available and
associated emission reductions achievable for each BART-eligible source"). However, EPA's
regulations purport to allow states to waive the BART requirements through "an emissions
trading program or other alternative measure" rather than by setting BART limits on a case-by-
case basis if the applicable standards for using an alternative are met. 40 C.F.R. § 51.308(e)(2).
According to Earthjustice et al., an alternative to BART "must achieve greater reasonable
progress than would be achieved through the installation and operation of BART." 40 C.F.R. §
51.308(e)(2). A state may demonstrate that an alternative program makes greater reasonable
progress than BART by proving that under the alternative program (1) visibility does not decline
in any Class I area and (2) there is an overall improvement in visibility compared to BART at all
affected Class I areas. Id. § 51.308(e)(3)(i)-(ii).
Trading programs give sources emission allocations, and then allow sources to trade the
allowances. One such trading program is the Cross State Air Pollution Rule ("CSAPR"). EPA
issued CSAPR to promote downwind attainment of national health standards for ozone and
particulate matter by reducing upwind emissions of NOx and SO2. EPA v. EME Homer City
Generation L.P., 134 S. Ct. 1584,1596 (2014). CSAPR sets state budgets for emissions of NOx
and SO2, then authorizes sources to make intrastate and interstate emission trades to meet
compliance obligations. 76 Fed. Reg. 48,208 (Aug. 8, 2011); see also EPA, Cross-State Air
Pollution Rule (CSAPR)—Basic Information, available at
http://www.epa.gov/airtransport/CSAPR/basic.html. CSAPR, which was to be fully
implemented by 2014, replaced CAIR, which had been remanded following previous litigation,
although the D.C. Circuit left the rule in place pending further action by EPA. EPA v. EME
Homer City Generation, L.P., 134 S. Ct. at 1595-96 (citing North Carolina v. EPA, 550 F.3d
1176, 1178 (D.C. Cir. 2008) (per curiam)).
165
-------
Earthjustice et al., stated that the EPA later issued a rule finding that CSAPR was "better-than-
BART," meaning that EPA concluded it achieved "greater reasonable progress towards the
national goal of achieving natural visibility conditions in Class I areas than source-specific
[BART] in those states covered by the Transport Rule." 77 Fed. Reg. 33,642, 33,643 (June 7,
2012). To support its conclusion that CSAPR makes greater reasonable progress than BART,
EPA compared CSAPR as it existed in 2011 to presumptive BART. However, EPA
subsequently made significant changes to both the emission limits and the compliance deadlines
for CSAPR. To date, EPA has not updated its "better than BART" analysis to examine whether
the revised version of CSAPR makes greater reasonable progress than BART.
Earthjustice et al., stated that BART is a necessary component of establishing progress, but it is
just one component and cannot substitute for the other elements of a regional haze plan. In
particular, it is not a replacement for the analysis of "any potentially affected sources" required
by 40 C.F.R. § 51,308(d)(l)(i)(A). The Regional Haze Rule requires states to look beyond
BART for additional emissions reductions that provide for "reasonable progress" toward 2064
natural visibility goal for Class I national parks and wilderness areas. 42 U.S.C. §
7491(b)(2)(B); 40 C.F.R. § 51.308(d).5
Even where, as here, EPA has determined that a nationally applicable "emissions trading
program or other alternative will achieve greater reasonable progress toward natural visibility
conditions" than BART, that finding does not exempt BART sources from emission control
requirements to advance reasonable progress if they continue to cause or contribute to visibility
impairment. Id. § 51.308(e). In (hiI. Air Regulatory Group v. EPA, 471 F.3d 1333 (D.C. Cir.
2006), the D.C. Circuit upheld EPA's "better than BART" determination for the Cross State Air
Pollution Rule's predecessor, but the court in no way reduced the states' or EPA's authority and
obligation to require updated pollution controls to ensure reasonable progress at each Class I
area. The court stated:
[UJnless there is some reasonable excuse, [a regional haze plan's reasonable]
progress must be sufficient to attain natural visibility conditions at every single
Class I area by 2064. Indeed, EPA emphasized in its briefs that because "the
regulatory scheme as a whole (and all the regulations promulgated pursuant to it)
must be designed to achieve the goal [of reasonable progress] at every Class I
area," states must, if CAIR is substituted for BART and is not likely to achieve
that goal, take "other measures as necessary to achieve reasonable progress goals
including at each Class I area."
Id. at 1340 (internal citations omitted); see also 70 Fed. Reg. at 39,138 n.73 ("The reasonable
progress test in the Regional Haze Rule remains as a separate test from [CAIR's] better than
BART" determination.). The court recognized that BART and "better than BART" alternatives
are merely one mechanism for achieving the overarching reasonable progress requirements of the
Clean Air Act.
Earthjustice et al., stated that, under a trading program such as CSAPR, some sources may
reduce emissions significantly, while others may not reduce emissions at all or may even
increase emissions. When EPA originally authorized alternatives to BART, EPA recognized that
166
-------
trading programs can create "hot spots:" areas where emissions fail to decrease, or even increase.
71 Fed. Reg. 60,612, 60,627 (Oct. 13, 2006). To address this problem, EPA inserted a
provision whereby states that opt to use an alternative program and waive the BART
requirements can supplement the alternative program to address any pollution not adequately
controlled by the alternative program. EPA calls this provision a "geographic enhancement." 40
C.F.R. § 51.308(e)(2)(v).
Due to the flexible, market driven nature of a trading program, Earthjustice et al., stated that
there is no assurance that benefits in reduced pollution will be realized at the very places
intended for restoration under the Clean Air Act's visibility protection mandate. This can lead to
visibility hot spots when a source near a Class I area purchases emission allowances from a
distant source rather than reduces emissions, or when a group of sources in the same region
purchase emission allowances rather than reducing emissions. Reasonable progress controls are
particularly important under emission trading programs because those trading programs do not
require installation of pollution controls.
Footnotes:
5 Under Section 51.308(e)(5) of the Regional Haze Rule, "[a]fter a State has met the requirements for BART or
implemented emissions trading program or other alternative measure that achieves more reasonable progress than
the installation and operation of BART, BART-eligible sources will be subject to the requirements of paragraph (d)
of this section in the same manner as other sources."
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis. Comments concerning our finding that
CSAPR is "better-than-BART" (77 FR 33642) are outside the scope of today's rulemaking.
Comment: EPA Should Reconsider Its Approval of Texas' Exclusion of All BART-Eligible
Sources and EPA's Proposed Reliance on CSAPR as an Alternative to BART. [Earthjustice
(0067) p.52]
Earthjustice et al., stated that in crafting a program to rid our national parks and wilderness areas
of visible air pollution, Congress specified that some of the largest sources of pollution must
install and operate the best available pollution controls, or BART. See 42 U.S.C. §
7491(b)(2)(A). To ensure that states require the best pollution controls on these sources,
Congress prescribed the factors that must be considered when determining BART. Id. §
7491(g)(2). Moreover, Congress instructed states to follow EPA Guidelines for determining
BART for large power plants. Id. § 7491(b)(2). Given that BART sources are responsible for a
significant amount of the haze pollution at Class I areas, BART controls are a cornerstone of
each haze plan. Texas submitted a haze plan that proposes no controls for any BART-eligible
source in the state. Texas reached this result by first taking certain power plants' NOx and SO2
emissions off the table, because Texas relied on CAIR in lieu of source-specific BART for power
plants subject to BART. SIP at 9-1. Texas then used a variety of methods to conclude that no
other source was subject to BART. Id. at 9-10. As a result, Texas did not conduct a five-factor
BART analysis for a single one of its 126 BART-eligible sources. EPA proposes to approve
Texas' determination that a five-factor BART analysis does not need to be completed for any
BART eligible source. 79 Fed. Reg. at 74,854. EPA also proposes a FIP that relies on CSAPR
167
-------
as an alternative to BART for Texas power plants. 79 Fed. Reg. at 74,853.65 EPA should revisit
both of these proposed actions.
Earthjustice et al., stated that the EPA should disapprove Texas' determination to exclude all
BART-eligible sources from being subject to BART. Texas' screening of BART-eligible
sources is based in part on the unsupported assumption that 0.5 deciviews is the threshold for
contribution to visibility impairment. Moreover, Texas erred in its analysis excluding certain
sources from a five factor BART analysis. Second, for power plants, if EPA does not finalize the
SO2 controls it proposed for reasonable progress in the FIP, then EPA should assess and require
them as BART controls. EPA should also conduct source-specific analyses as to the NOx
controls that represent BART for these power plants.
Earthjustice et al., stated that the two Class I areas in Texas, Big Bend and Guadalupe
Mountains, are not on track to attain natural visibility in 2064—even if EPA finalizes all of the
controls it is proposing. The controls EPA is proposing are critical to making reasonable
progress. However, to achieve needed progress beyond that proposed in its FIP, EPA should
reverse course, taking a fresh look at BART-eligible source previously exempt from control.
Footnotes:
65 After the D.C. Circuit remanded CAIR to EPA, the agency disapproved the portion of Texas'
SIP that relied on CAIR as a BART alternative. 77 Fed. Reg. at 33,654. However, EPA has not yet finalized a FIP
to fill the gap left by disapproving Texas' reliance on CAIR.
Response: We acknowledge the background information provided by Earthjustice. We take no
position with respect to the commenter's synopsis. We proposed to approve Texas' screening of
BART sources from full BART analysis. We are finalizing our approval of Texas' BART
determinations for non-EGUs under Section 51.308(e). The comment does not include any
information to support disapproval of this portion of Texas' BART determinations. We find that
Texas' assumption that 0.5 deciviews is the threshold for contribution to visibility impairment is
in accordance with the BART guidelines. 99 In setting a threshold, states should consider the
number of BART-eligible sources within the state and the magnitude of each source's impacts.
TCEQ provided an adequate rationale for choosing 0.5 dv as the threshold for determining
BART eligibility. We are approving Texas' determination of which non-EGU sources in the
state are BART-eligible and the determination that none of the state's BART-eligible non-EGU
sources are subject to BART because they are not reasonably anticipated to cause or contribute
to visibility impairment at any Class I areas. We reviewed the various modeling techniques
utilized by the TCEQ in evaluating and screening out the BART-eligible non-EGU sources and
we concur with the results of its analysis.100
We also proposed a FIP that relies on CSAPR as an alternative to BART for Texas power plants.
We took partial action on Texas' regional haze SIP in 2012 when we partially disapproved Texas'
reliance on Clean Air Interstate Rule (CAIR) to meet the SO2 and NOx BART requirements
99 The BART Guidelines state that "the appropriate threshold for determining whether a source contributes to
visibility impairment' may reasonably differ across States," but, "[a]s a general matter, any threshold that you use for
determining whether a source contributes' to visibility impairment should not be higher than 0.5 dv." BART
Guidelines, 40 CFR part 51, appendix Y, section III.A.l.
i°° 79 pR 74844.
168
-------
for its EGUs. 77 FR 33642 (June 7, 2012). In our proposal, we stated that we intended to
address this deficiency in Texas' plan by relying on CAIR's replacement, CSAPR, which we
previously determined would provide for greater reasonable progress than BART. 79 FR at
74821 (Dec. 16,2014). On July 28, 2015, the D.C. Circuit Court's issued its decision in EME
Homer City Generation v. EPA, 795 F.3d 118 (D.C. Cir 2015), upholding CSAPR but
remanding without vacating a number of the Rule's state emissions budgets. Specifically, the
court invalidated a number of the Phase 2101 ozone season NOx budgets and found that the SO2
budgets as to four states resulted in overcontrol for purposes of section 110(a)(2)(D)(i) of the
CAA. Texas' ozone season NOx budget and its SO2 budget are both implicated in this remand.
We are in the process of acting on the Court's remand. As a result, at this time we cannot ensure
that CSAPR will continue to be an appropriate alternative to BART for Texas EGUs. Given the
uncertainty arising from the remand of some of the state CSAPR budgets, we have decided not to
finalize that portion of our FIP relying on CSAPR as an alternative to SO2 and NOx BART for
EGUs in Texas. We will address the question of appropriate SO2 and NOx BART limits for
EGUS in Texas in a future rulemaking once EPA has determined how best to respond to the
remand of some of the CSAPR state budgets. This analysis may include a review of which
EGUs are subject to BART and require a five-factor analysis. We note that a few of the sources
for which we are finalizing SO2 controls as part of the Texas long-term strategy are also BART-
eligible. Should we determine in the future that it is necessary to perform source-specific BART
determinations for these sources instead of relying on CSAPR, we anticipate that the SO2
controls we are finalizing today, which are currently the most stringent available, will also be
sufficient to satisfy the BART requirements.
Comment: If EPA Does Not Finalize Reasonable Progress Controls for BART Sources,
Then EPA Should Impose Source-Specific BART Controls on Those Sources.
[Earthjustice (0067) p.53]
Earthjustice et al., stated that the EPA should finalize the SO2 controls it has proposed for 15
units as reasonable progress controls. See 79 Fed. Reg. at 74,891. However, if EPA does not
require those controls as reasonable progress measures in the final rule, then EPA should require
the proposed controls as BART.
Earthjustice et al., stated that the five factors that must be considered when establishing BART
are the same factors that EPA has used in this rulemaking to evaluate reasonable progress
controls. Three of the five BART factors are included in the reasonable progress factors, which
EPA used here. Compare 42 U.S.C. 7491(b)(2)(A) (requiring BART to be based on cost,
energy and nonair quality environmental impacts and remaining useful life) with 79 Fed. Reg. at
74,873-84 (considering cost, energy and nonair quality environmental impacts, and remaining
useful life). EPA considered the remaining two BART factors—existing pollution controls and
the visibility improvement from controls—in its reasonable progress analyses. 79 Fed. Reg. at
74,874 ("we are including in our evaluation a consideration of any control technology that may
already be installed at the facility"); id. ("we are also considering the projected visibility benefit
101 CSAPR's effective date was stayed by the D.C. Circuit during a portion of the litigation over the rule. Thus, the
Rule's Phase 1 effective date of January 1, 2012 was delayed until January 1, 2015, and the Rule's Phase 2 effective
date of January 1, 2014 was delayed until January 1, 2017.
169
-------
in our analysis"). Thus, in developing the proposed reasonable progress controls, EPA
considered all of the factors that are required to be considered when determining BART. Given
that EPA evaluated reasonable progress controls using the factors that it would use to set BART
limits, EPA's analysis of the proposed SO2 controls would support finding that these limits
represent BART. In other words, the required five-factor BART analysis would include the
information EPA developed in support of the proposed reasonable progress controls for Texas
sources.
Based on EPA's analysis for the proposed rule, Earthjustice et al., stated that the EPA should
find that the limits listed below are BART if the agency does not require the units to meet these
limits to satisfy the reasonable progress requirements:
Unit
SOj Limit
pCMtoY reHiiig
§1 #1 ip,
ttts/MMBfu)
Big Brown 1
' 1
Big Brow 2
11 r >4
M.vtm Lake 1
H 12
Mwm Lake 2
Martin Like 1
Oil
MoRtiedlo "
11 14
Moiitiedlo 2
s 104
San Mimiel
0 V
For the EGUs listed below, Earthjustice et al., noted that the EPA is proposing reasonable
progress controls, but it is unclear whether the units are subject to BART for SO2 and NOx
emissions. Texas did not make publicly available the results of its BART eligibility survey, and
Texas did not determine whether EGUs subject to CAIR are subject to BART for SO2 and NOx
emissions. Thus, for the units listed below, EPA should determine whether the units are subject
to BART. If the units are subject to BART, and if EPA does not finalize the proposed reasonable
progress controls for these units, then EPA should find that the SO2 limits EPA is proposing
under its reasonable progress authority are BART for all of the following units that EPA finds
are subject to BART.
Unit
50i Limit I
(3Mj.y rolling
" «igf,
fe/MMBtii)
rrt-K 1
11U4
Lur.e-tone 1
0 08
LuneMMie 2
0 08 1
Momcelio 3
0.0-5 1
Sandow 4
0.20 1
Toft 171B
006
1 o.lV. J t
0.06
Earthjustice et al., argued that EPA did not conduct a detailed, four-factor analysis for which
NOx controls should be required to achieve reasonable progress. Thus, EPA has no preexisting
170
-------
analysis to use for setting NOxBART limits. Instead, EPA should conduct source-specific
analyses of the NOx limits that represent BART for the BART units listed above.
Response: We acknowledge the background information provided by the Earthjustice but take
no position on it. We acknowledge the similarity of some aspects of our analysis with the type
of analysis that is required under our BART Guidelines. However, we disagree that we have the
ability to change our proposal to make the SO2 controls we proposed under the reasonable
progress and long-term strategy provisions of the Regional Haze Rule apply to or substitute for
BART, without first re-proposing them as such and taking comment. Additionally, we are
finalizing controls for reasonable progress as proposed.
Comment: EPA's Proposed Reliance on CSAPR as a BART Alternative is Unlawful.
[Earthjustice (0067) p. 55]
Earthjustice et al., stated that the EPA's proposal to rely on CSAPR as an alternative to BART is
unlawful for three reasons. First, EPA's proposal exempts sources from BART requirements
without complying with the statutory prerequisites for such an exemption. Second, even if EPA
could relieve the sources of the obligation to install BART controls, the "Better than BART" rule
upon which EPA relies is flawed. Third, the "Better than BART rule" is no longer valid given
the substantial changes in CSAPR allocations and compliance deadlines. We incorporate and
resubmit their 2012 comments, which remain relevant to EPA's proposal to rely on CSAPR as a
BART alternative. EPA Docket ID EPA-HQ-OAR-2011-0729, Letter from McCrystie Adams &
Michael Hiatt, Earthjustice to EPA (Feb. 28, 2012). [This document is attached to comment
0067 -Item 5],
First, Earthjustice et al., stated that EPA's proposal is unlawful because it exempts sources from
installing BART controls without going through the exemption process Congress prescribed.
The visibility protection provisions of the Clean Air Act include a "requirement" that certain
sources "install, and operate" BART controls. 42 U.S.C. § 7491(b)(2)(A). Congress specified
the standard by which sources could be exempted from the BART requirements, which is that the
source is not "reasonably [] anticipated to cause or contribute to a significant impairment of
visibility" in any Class I area. Id. § 7491(c)(1). Appropriate federal land managers must concur
with any proposed exemption. Id. § 7491(c)(3). EPA has not demonstrated that any of the Texas
EGUs subject to BART meet the standards for an exemption, nor has EPA obtained the
concurrence of federal land managers. Therefore, EPA must require source-specific BART for
each power plant subject to BART.
Second, Earthjustice et al., stated that even if EPA could use a BART alternative without going
through the statutory exemption process, EPA could not rely on CSAPR because of flaws in the
rule that purports to show that CSAPR makes more reasonable progress than BART (the "Better
than BART" rule). EPA's regulations purport to allow the use of an alternative program in lieu
of source-specific BART only if the alternative makes "greater reasonable progress" than would
BART. 40 C.F.R. § 51.308(e)(2). To demonstrate greater reasonable progress, a state or EPA
must show that the alternative program does not cause visibility to decline in any Class I area
and results in an overall improvement in visibility relative to BART at all affected Class I areas.
171
-------
Id. § 51.308(e)(3)(i)-(ii). Here, EPA claims that its 2012 "Better than BART" rule demonstrated
that CSAPR achieves greater reasonable progress than BART. &e77Fed. Reg. 33,642.
As explained in detail in their 2012 comments, Earthjustice et al., stated that EPA compared
CSAPR to BART in the Better than BART rule by using CSAPR allocations that are more
stringent than now required as well as by using presumptive BART limits that are less stringent
than are actually required under the statute. See Letter from McCrystie Adams, Earthjustice to
EPA at 13-16 (Feb. 28, 2012). In short, EPA used assumptions that made CSAPR appear more
effective than it is, and made BART appear to be less effective than it is, which ultimately tilted
the scales in favor of CSAPR. It would be arbitrary and capricious for EPA to rely on such an
inaccurate, faulty comparison to conclude that CSAPR will achieve greater reasonable progress
than will BART. Even under EPA's skewed comparison, CSAPR achieves barely more
visibility improvement than BART at Big Bend and Guadalupe Mountains.
Earthjustice et al., stated that if EPA modeled accurate BART limits and up-to-date CSAPR
allocations, then EPA would likely find that CSAPR would lead to less visibility improvement
than BART. Indeed, SO2 emissions allowed under CSAPR for Texas EGUs are 91% higher than
would be allowed under BART. Id. at 17. Similarly, the NOx emissions allowed under CSAPR
from Texas EGUs are higher than would be allowed under BART. Id. at 17-18. This was true
even before EPA revised CSAPR to increase the emissions allocations for all Texas EGUs.
Third, Earthjustice et al., stated that EPA cannot lawfully rely on the "Better than BART" rule
because the rule is based on a version of CSAPR that no longer exists. Accordingly, any
conclusion that EPA made in the 2012 Better than BART rule regarding whether CSAPR
achieves greater reasonable progress than BART is no longer valid. Since 2012, EPA has
significantly changed the allocations and the compliance deadlines for CSAPR. Of particular
relevance here, after 2012, EPA dramatically increased the CSAPR allocations for every covered
EGU in Texas. 77 Fed. Reg. 10,324 (Feb. 21, 2012), Final Revisions Rule State Budgets and
New Units Set-Asides TSD at 3. EPA later withdrew the February 21, 2012 rule revision, 77
Fed. Reg. 28,785 (May 16, 2012), but issued a new rule that included both the changes in the
February 21, 2012 rule as well as additional changes to state budgets. 77 Fed. Reg. 34,830
(June 12, 2012). By the time that EPA finalized the Better than BART rule in June 2012, EPA
had changed the state emissions budgets by tens of thousands of tons, yet EPA proceeded to
finalize the Better than BART rule based solely on the emissions budgets in the original, 2011
CSAPR rule.
Earthjustice et al., stated that EPA also extended the compliance deadlines by three years, such
that the phase 1 emissions budgets take effect in 2015-2016 and the phase 2 emissions budgets
take effect in 2017 and beyond. 79 Fed. Reg. 71,663 (Dec. 3, 2014); see also 79 Fed. Reg. at
74,853. And yet more changes may occur as a result of the pending challenges to CSAPR in the
D.C. Circuit Court of Appeals. EME Homer City Generation, LP v. EPA, No. 11-1302 (D.C.
Cir. filed 2011).
In short, Earthjustice et al., concluded that EPA purports to satisfy the regulatory requirements
for a BART alternative by relying on its 2012 finding that CSAPR makes more reasonable
progress than BART. But EPA's 2012 finding was based on a prior version of CSAPR, not the
172
-------
version that is in effect in 2015 at the time of EPA's proposal and will be in effect at the time of
the final Texas rule and in the future. To rely on CSAPR as an alternative to BART, EPA must
demonstrate that the version of CSAPR that is now in effect, and will be in effect at the time of
the final rule, makes greater reasonable progress than BART. Having failed to make that
demonstration, EPA has not met its burden to show that CSAPR will achieve greater reasonable
progress than source-specific BART. See 40 C.F.R. § 51.308(e)(2), (3).
Response: We took partial action on Texas' regional haze SIP in 2012 when we partially
disapproved Texas' reliance on the Clean Air Interstate Rule to meet the SO2 and NOx BART
requirements for its EGUs. 77 FR 33642 (June 7, 2012). In our proposed FIP, we stated that we
intended to address this deficiency in Texas' plan by relying on CAIR's replacement, CSAPR,
which we previously determined would provide for greater reasonable progress than BART. 79
FR at 74821 (Dec. 16,2014). On July 28, 2015, the D.C. Circuit Court's issued its decision in
EMEHomer City Generation v. EPA, 795 F.3d 118 (D.C. Cir 2015), upholding CSAPR but
remanding without vacating a number of the Rule's state emissions budgets. Specifically, the
court invalidated a number of the Phase 2102 ozone season NOx budgets and found that the SO2
budgets as to four states resulted in overcontrol. Texas' ozone season NOx budget and its SO2
budget are both implicated in this remand. We are in the process of acting on the Court's
remand. As a result, at this time we cannot ensure that CSAPR will continue to be an
appropriate alternative to BART for Texas EGUs. Given the uncertainty arising from the
remand of some of the state CSAPR budgets, we have decided not to finalize that portion of our
FIP relying on CSAPR as an alternative to SO2 and NOx BART for EGUs in Texas. We will
address the question of appropriate SO2 and NOx BART limits for EGUS in Texas and the
remaining issues in a future rulemaking once EPA has determined how best to respond to the
remand of some of the CSAPR state budgets. Therefore, these comments are beyond the scope
of today's final action. Comments concerning our finding that CSAPR is "better-than-BART"
(77 FR 33642) are outside the scope of today's rulemaking. Additionally, comments concerning
our previous finding that CSAPR is "better-than-BART" are outside the scope of today's action.
Comment: Luminant provided background information on Texas' BART determination.
[Luminant (0061) p. 24]
Consistent with EPA's prior determination, Luminant stated that Texas concluded that
"participation in CAIR is equivalent to BART" for EGUs.188 Accordingly, Texas did not conduct
a source-specific BART analysis for SO2 and NOx for EGUs.189
Luminant noted that for the remaining non-EGU sources and for EGUs with respect to
particulate matter ("PM"), Texas conducted a screening analysis to determine which sources
were "BART-eligible."190 Using a screening analysis, Texas concluded that 254 sites were
"potentially BART-eligible based on distance and actual emissions."191 These 254 sources were
surveyed to determine "any equipment built or reconstructed during the applicable time period or
if the PTE [potential to emit] of their site were less than 250 tpy."192 If BART eligible equipment
102 CSAPR's effective date was stayed by the D.C. Circuit during a portion of the litigation over the rule. Thus, the
Rule's Phase 1 effective date of January 1, 2012 was delayed until January 1, 2015, and the Rule's Phase 2 effective
date of January 1, 2014 was delayed until January 1, 2017.
173
-------
did not exist at the site, the sources "were not asked to supply any further information and were
considered not BART-eligible."193 If the source potentially had BART-eligible equipment, a
"detailed survey asked whether each piece of equipment at the site was built or reconstructed
between the applicable dates" and if the "PTE of their BART-eligible equipment exceeded the
250 tpy threshold."194 The results of Texas' survey revealed that "over 100 sources were
identified as BART-eligible."195
Luminant stated that Texas conducted an analysis of which of those BART eligible sources were
"subject to BART."196 Texas used three methods, endorsed by EPA's BART guidelines, for
determining whether the screened sources were "subject to BART": (1) "the use of model plants
to exempt sources with common characteristics"; (2) "a cumulative modeling analysis to show
that groups of sources are not subject to BART"; and (3) "an individual source attribution
approach."197 The cumulative modeling analysis utilized CAMx PSAT and resulted in 72 sources
being screened out from further review.198 For the remaining sources, the individual source
modeling first relied on California Puff Model ("CALPUFF") and single-source CAMx.199 The
CALPUFF modeling resulted in 29 additional sources being screened out, and the single-source
CAMx modeling resulted in six sources being screened out.200 Texas further removed 22 sources
from BART based on exemption requests relying on distance to Class I areas and levels of haze-
producing emissions.201 Based on the screening analysis and the exemption requests that were
granted, Texas ultimately concluded that "no Texas sources remained subject to BART."202
Footnotes:
188 2009 Texas SIP Narrative at 9-1.
189-190
191 Id. at 9-3. The following sources that EPA proposes to regulate in this current proposal are BART-eligible and
subject to BART controls: Big Brown, Sandow Unit 4, Martin Lake, and Monticello. The following sources are not:
Limestone, San Miguel, Coleto Creek, and Tolk.
192-195
196 Id. at 9-6.
197 Id.
198 Id. at 9-6 to 9-8.
199 Id. at 9-6, 9-13 to 9-15.
200 Id. at 9-6, 9-14 to 9-15.
201 Id. at 9-15 to 9-16.
202 Id. at 9-17.
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis.
Comment: EPA's treatment of BART-compliant sources in Texas is unprecedented and
unlawful [Luminant (0061) p. 121]
Luminant stated that EPA's proposed additional emission limitations for BART-compliant
sources in Texas under the guise of reasonable progress is unlawful and inconsistent with EPA's
own proposed FIP. Many of the sources that EPA includes in its "small group" analysis
(including all of the Luminant units) are BART-eligible and fully controlled under the BART
provisions755—by inclusion in CAIR and now CSAPR—and thus are inappropriately targeted for
additional controls in the first planning period. EPA's consistent and established practice is that
174
-------
"it is reasonable to conclude that any control requirements imposed in the BART determination
also satisfy the RPG-related requirements for source review in the first RPG planning period."756
Under EPA's own finding and precedent, Texas' decision not to further regulate these BART-
complaint sources is reasonable and must be approved.
Luminant noted that EPA's proposal to target these BART-compliant sources in Texas is all the
more arbitrary given EPA's proposed FIP that finds CSAPR's emission limits on these sources
are better-than-BART757 By definition, CSAPR "achieve[s] greater reasonable progress than
would be achieved through the installation and operation of BART."758 Thus, Texas' compliance
with CSAPR amply demonstrates reasonable progress for these sources in the first planning
period and fully justifies Texas' decision not to impose further controls on these units at this
time. Indeed, until now, EPA has agreed with this standard. In reviewing other states'
reasonable progress goals for the first planning period, EPA has repeatedly and systematically
excluded sources that otherwise comply with the BART requirements, including complying
through a regional trading program that is better-than-BART, from further emission controls in
the first planning period without even conducting an "additional analysis" like it performs here
for certain Texas sources.759 In every instance Luminant located, EPA concluded that the
CSAPR-is-better-than-BART program is all that is needed for these sources to make reasonable
progress in the first planning period.761 EPA gives no explanation of its failure to apply the same
standard here to Texas' BART-compliant sources.
Luminant asserted that EPA's error is compounded by its failure to even include in its modeling
and reasonable progress analysis the very emission reductions mandated by CSAPR that make
these sources BART-compliant and that achieve reasonable progress. EPA has arbitrarily
refused to include in its analysis the emission reductions and limitations that are imposed on
these Texas sources by CSAPR, which became effective on January 1, 2015.762 Thus, not only
has EPA arbitrarily refused to exclude these BART-compliant sources from further analysis (as it
has done for BART sources in other states), it has further arbitrarily refused to even consider the
emission limitations that these sources must comply with under current regulatory programs.
Luminant noted EPA's own prior modeling of CSAPR clearly demonstrates that reasonable
progress will be achieved at the three Class I areas of interest here, without any further controls
beyond CSAPR's emission limits and other existing limitations. EPA's own prior modeling
shows that the Wichita Mountains, Big Bend, and Guadalupe Mountains all meet the RPG for
2018 that EPA is proposing based on the implementation of CSAPR—even without the
additional controls in this proposal. For its CSAPR-better-than-BART finding, EPA modeled the
visibility in all Class I areas to determine the visibility (in deciviews) at each area in 2014 with
the implementation of CSAPR and the application of BART in states not subject to CSAPR
(including Oklahoma) 763 This included Big Bend, Guadalupe Mountains, and Wichita
Mountains. EPA used the same model—CAMx—that it now uses to judge the Texas SIP. Table
9 shows the visibility at the three Class I areas of interest here that EPA modeled for 2014,
assuming CSAPR emission budgets in Texas and other CSAPR states and BART controls in
non-CSAPR states.
175
-------
TABLE 9: EPA MODELED1 VISIBILITY AT THREE CLASS I AREAS WITH CSAPR IN PLACE
Class 1 Area
EPA Modeled Visibility
Conditions with CSAPR in
Place (20 H) (20% Worst
Days):
EPA Proposed RPG
(2018) (20% worst
days)'
Big Bend
1 5 _ »i.
Guadalupe Mountains
14 i ui
Wichita Mountains
?! ! ? d.
Thus, Luminant stated that based on EPA's own modeling (which was validated)766 CSAPR will
achieve greater progress than even EPA believes is reasonable for these three areas by 2018.
Indeed, EPA's own prior modeling shows visibility at Big Bend and Guadalupe Mountains better
than their respective URP, well before 2018. It was arbitrary for EPA to ignore this prior
validated modeling, arbitrary for EPA to exclude CSAPR from its new modeling, and unlawful
and unreasonable for EPA to conclude that Texas' compliance with CSAPR's better-then-BART
emission limits does not satisfy the reasonable progress requirement for the first planning period.
Footnotes:
755 According to EPA, the following sources that EPA proposes to regulate under reasonable progress are BART-
eligible and subject to BART controls: Big Brown, Sandow, Martin Lake, Monticello. The following sources are
not: Limestone, San Miguel, Coleto Creek, and Tolk.
756 Reasonable Progress Guidance at 4-2 to 4-3; see also id. at 5-1 ("[T]he significant emissions reductions that we
anticipate to result from BART, the CAIR, and the implementation of other CAA programs . .. may be all that is
necessary to achieve reasonable progress in the first planning period for some States."). See also 77 Fed. Reg.
30,454, 30,460 (May 23, 2012) (explaining that a state's long-term strategy for the first planning period should
include "BART and any additional controls for non-BART sources" (emphasis added)).
757 79 Fed. Reg. at 74,853; see also 77 Fed. Reg. 33,642, 33,648 (June 7, 2012).
758 40 C.F.R. § 51.308(e)(2) (emphasis added).
759 Never, in its review of regional haze SIPs for the first planning period, has EPA imposed a FIP requiring
additional "reasonable progress" controls on EGUs that relied on the implementation of CAIR/CSAPR to satisfy
BART. See 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77
Fed. Reg. 34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098
(Mar. 30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26,
2012) (Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012)
(Ohio); 79 Fed. Reg. 24,340 (Apr. 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South
Carolina); 77 Fed. Reg. 24,392 (Apr. 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77
Fed. Reg. 16,937 (Mar. 23, 2012) (West Virginia).
760 Id.
761 Id.
762 FIP TSD at A-45.
763 BART Alternative TSD at 25-30, tbl.3-3.
764 Id. at tbl. 3-3.
765 79 Fed. Reg. at 74,887, tbl 43.
766 See BART Alternative TSD at 9.
Response: Although EPA proposed to rely on CSAPR to address the BART requirements for
EGUs in Texas, it is not finalizing that proposed action. On July 28, 2015, the D.C. Circuit
176
-------
Court's issued its decision inEMEHomer City Generation v. EPA, 795 F.3d 118 (D.C. Cir
2015), upholding CSAPR but remanding without vacating a number of the Rule's state emissions
budgets. Specifically, the court invalidated a number of the Phase 2103 ozone season NOx
budgets and found that the SO2 budgets as to four states resulted in overcontrol for purposes of
section 110(a)(2)(D). Texas' ozone season NOx budget and its SO2 budget are both implicated
in this remand. EPA is in the process of acting on the Court's remand. As a result, at this time
we cannot ensure that CSAPR will continue to be an appropriate alternative to BART for Texas
EGUs. As a result, the comments that the emission limitations under reasonable progress should
not be required for EGUs that are already compliant with BART is no longer applicable.
Even assuming, however, that EME Homer City had not invalidated the CSAPR NOx and SO2
budgets for Texas and that EPA were taking final action to address the BART requirements
through reliance on CSAPR, we do not agree that EPA has been inconsistent in its treatment of
Texas as compared to other states. As explained in our proposed rulemaking, allowing Texas to
rely on CSAPR to meet its reasonable progress obligations is not appropriate, and the fact that
other states did not require additional reasonable progress controls beyond CAIR (or CSAPR)
does not automatically mean all states should not require any additional controls. Such a
simplistic comparison ignores the meaningful differences between Texas and the states cited.
These include the significant impacts that sources in Texas have on the visibility at the Wichita
Mountains in Oklahoma, the quality of the Texas technical evaluation, and the quality of the
consultations between Texas and Oklahoma. We address comments concerning other states'
reasonable progress goals for the first planning period more fully elsewhere in this document.
We note that the modeling that was performed for the CSAPR Better-than BART demonstration
was developed for the final CSARPR notice and did not include the supplemental adjusted
budgets for Texas. Consistent with the requirements for determining whether a BART
alternative provides for better reasonable progress than BART, the CSAPR Better-than-BART
demonstration was focused on assessing the improvements in visibility on average across all
Class I areas compared to BART level controls. We note that this modeling projection was
based on different baseline conditions and emissions than the CENRAP modeling. The
sensitivity analysis104 performed that supported the CSAPR Better-than-BART rule
demonstrated that the CSAPR budgets including the supplemental budget adjustment in Texas
still were Better-than-BART but visibility at Class I areas impacted the most by Texas EGU
emissions are not expected to achieve the same amount of visibility benefit as modeling based on
the CSAPR budgets prior to the adjustments. The sensitivity analysis relied on a conservative
assumption that visibility at the Class I areas nearest Texas saw no visibility benefits due to
CSAPR reductions. Further, as discussed above, these budgets have been invalidated because
they "required Texas to reduce emissions by more than the amount necessary to achieve
attainment in every downwind State to which it is linked."105
103 CSAPR's effective date was stayed by the D.C. Circuit during a portion of the litigation over the rule. Thus, the
Rule's Phase 1 effective date of January 1, 2012 was delayed until January 1, 2015, and the Rule's Phase 2 effective
date of January 1, 2014 was delayed until January 1, 2017.
104 See Appendix C of the Technical Support Document for Demonstration of the Transport Rule as a BART
Alternative. Available at http://www.regulations.gov/#!documentDetail;D=EPA-HQ-OAR-2011-0729-0014
105 EME Homer City, 795 F.3d at 129.
177
-------
Comment: [Associations (0059) p. 16] The Associations stated that the EPA's individual
source-based approach unlawfully creates inconsistent treatment of sources by subjecting them
to different standards under BART and reasonable progress SIPs. Several of the sources that
EPA evaluates and targets for additional emission controls are BART-eligible sources. Because
Texas is currently subject to CSAPR, BART-eligible sources can be controlled through Texas'
implementation of that rule because EPA has concluded that CSAPR's emission limits on those
sources are more stringent than BART. 77 Fed. Reg. 33,642, 33,648 (June 7, 2012). Further,
by definition, CSAPR "will achieve greater reasonable progress than would have resulted from
the installation and operation of BART." 40 C.F.R. § 51.3108(e)(2). As a result, Texas'
compliance with CSAPR should be sufficient to meet reasonable progress goals with respect to
BART-eligible sources. EPA's proposal to require separate and additional controls for BART-
eligible sources on a source-by-source basis is fundamentally incompatible with EPA's BART
and CSAPR regulations and further underscores the fact that reasonable progress goals should
not be developed in an individual source-specific manner.
Response: See our responses to other comments concerning our proposed control of sources that
are also BART eligible.
Comment: Although there should be no need to promulgate a CSAPR=BART FIP -
Because EPA had no sound basis for disapproving Texas' CAIR=BART SIP in 2012 -
Reliance on CSAPR=BART in Texas is appropriate at this time given that CSAPR is now
in effect. [UARG (0065) p. 30-32]
UARG noted that EPA in a separate rulemaking promulgated limited disapprovals of the
regional haze SIPs of 14 states, including that of Texas, on the grounds that those SIPs -
consistent with EPA's rules (specifically, 40 C.F.R. § 51.308(e)(4) as in effect at that time and
as affirmed by the U.S. Court of Appeals for the District of Columbia Circuit in Utility Air
Regulatory Group v. EPA, 471 F.3d 1333 (D.C. Cir. 2006)) - relied on those states'
participation in CAIR to satisfy BART requirements with respect to EGUs' SO2 and NOx
emissions. &e79Fed. Reg. at 74,821. After Texas and other states submitted their
"CAIR=BART" SIPs, EPA promulgated CSAPR to replace CAIR. 76 Fed. Reg. 48,208 (Aug.
8, 2011). EPA reasoned that because CAIR would no longer be in effect upon CSAPR's
implementation, it was necessary to disapprove SIPs that had lawfully relied on CAIR and to
promulgate FIPs replacing reliance on CAIR with reliance on CSAPR for states that were subject
to both CAIR and CSAPR. See 77 Fed. Reg. 33,642, 33,643 (June 7, 2012). In the same June
7, 2012 rule in which EPA promulgated the limited SIP disapprovals, EPA also promulgated so-
called "CSAPR=BART" FIPs for several states, but not for Texas. Id. at 33,643, 33,654. In the
present rulemaking, EPA proposes a CSAPR=BART FIP for Texas to replace Texas' reliance on
CAIR with reliance on CSAPR. 79 Fed. Reg. at 74,823, 74,844, 74,853-54, 74,888.
For the reasons described in UARG's comments on the proposed version of EPA's 2012 rule
promulgating the limited disapprovals of the regional haze SIPs of Texas and other states,7
UARG asserted that the EPA had no authority and no sound reason to promulgate limited
disapprovals of SIPs that relied on CAIR to satisfy BART. At the same time, UARG recognizes
178
-------
that the June 2012 rule in which EPA promulgated its limited SIP disapproval actions is not
directly at issue in the present rulemaking and is the subject of pending petitions for judicial
review in the D.C. Circuit. UARG also recognizes that because CSAPR is currently in effect,
reliance on participation in CSAPR to satisfy BART requirements for EGUs' SO2 and NOx
emissions is appropriate pursuant to 40 C.F.R. § 51.308(e)(4) as revised.8
Footnotes:
7 Comments of the Utility Air Regulatory Group on EPA's Proposed Rule: Regional Haze: Revisions to Provisions
Governing Alternatives to Source-Specific Best Available Retrofit Technology (BART) Determinations, Limited
SIP Disapprovals, and Federal Implementation Plans (Feb. 28, 2012), Doc. ID No. EPA-HQ-OAR-2011-0729-
0298 ("UARG 2012 Comments").
8 In addition, in light of EPA's determination that "the overall EGU emission reductions from CSAPR are larger
than the EGU emission reductions that would have been achieved by CAIR," EPA was correct in its conclusion that
it should not, and could not, "disapprove the reasonable progress goals in any of the regional haze SIPs for their
reliance on CAIR, including those for Texas." 79 Fed. Reg. at 74,853.
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis. To the extent that the commenter
comments on our limited disapproval of SIPs that relied on CAIR, that comment is outside the
scope of today's rulemaking. We agree with the commenter that we previously concluded not to
disapprove the reasonable progress goals in any of the regional haze SIPs for their reliance on
CAIR, including those for Texas.79 FR at 74851. As discussed in an earlier response in this
section, we are not finalizing our proposed action on EGU BART for Texas at this time. We will
address the question of appropriate SO2 and NOx BART limits for EGUS in Texas in a future
rulemaking once EPA has determined how best to respond to the remand of some of the CSAPR
state budgets. We note that a few of the sources for which we are finalizing SO2 controls as part
of the Texas long-term strategy are also BART-eligible. Should we determine in the future that
it is necessary to perform source-specific BART determinations for these sources instead of
relying on CSAPR, we anticipate that the SO2 controls we are finalizing today, which are
currently the most stringent available, will also be sufficient to satisfy the BART requirement.
Comment: [Xcel Energy (0064) p. 10] First, EPA claims that the CAIR program could not be
used to demonstrate the RPGs because CAIR would not require emissions reductions to be
achieved within Texas. But Texas is not relying on CAIR to meet reasonable progress goals or
relying on either CAIR, or its successor rule, the Cross-State Air Pollution Rule ("CSAPR"), in
lieu of meeting the RPGs. Rather, Texas simply used the indicative cost level of $2,700 from
CAIR as a basis for deciding not to include some sources under its Regional Haze SIP. Using
$2,700 as a basis for deciding not to require controls for some sources in regional haze
requirements is not the same as allowing sources to avoid regional haze requirements by
importing allowances. Further, EPA's concern about Texas importing unlimited allowances so
as to avoid reducing emissions is not valid here because CSAPR, which replaced CAIR, limits
the use of out-of-state allowances that can be used for compliance, and EPA has expressly
allowed other states to rely on CSAPR to meet RPGs. See Proposed Michigan SIP Approval, 77
Fed. Reg. 46,912, 46,919 (Aug. 6, 2012) (the regional planning organization's "analysis shows
emission reductions equivalent to the scale of CAIR are needed to meet reasonable progress
goals. . . . EPA believes that with CSAPR providing the reductions that Michigan expects to
obtain from CAIR, Michigan's long-term strategy can in fact be expected to achieve the state-
179
-------
adopted reasonable progress goals that Michigan established."). The fact that it is impossible to
meet the RPGs in Texas does not make it less legitimate to rely on the cost thresholds used in
CAIR for identifying reasonable controls.
Xcel Energy stated that despite EPA's action in Michigan, in this Proposal, EPA arbitrarily
argues that CAIR could be used "in lieu of BART" but not used in the RPG context. While
BART and RPGs are distinctive components of a regional haze strategy, EPA provides no
reasoned basis for allowing consideration of CAIR in the BART context and rejecting it in the
RPG context. 3 That EPA would allow CAIR or CSAPR to substitute for BART, which is a
unit-specific standard with unit-specific performance criteria, but not for demonstrating
reasonable progress, which is a state-wide, multi-source program aimed at reducing the
pollutants of concern for regional haze, is illogical, as well as arbitrary and capricious. In fact,
EPA has done the exact opposite in other RPG determinations and re-affirmed States' reliance on
BART-equivalent analyses. For example, as stated in its proposed approval of the Georgia SIP,
"EPA believes it is reasonable to conclude that any control requirements imposed in the BART
determination also satisfy the RPG related requirements for source review in the first
implementation period." See Proposed Georgia SIP Approval, 77 Fed. Reg. 11,452, 11,469
(Feb. 27, 2012); see also Final Georgia SIP Approval, 77 Fed. Reg. 38,501 (Jun. 28,2012). In
North Dakota, EPA specifically rejected modeling for RPGs that was not conducted in the same
way as BART on the basis that the "ultimate goal is the same." Proposed North Dakota SIP
Approval/Disapproval, 76 Fed. Reg. at 58,629 n. 85; see also Final North Dakota SIP
Approval/Disapproval, 77 Fed. Reg. at 20,906-07.
Commenter's References:
3 It does not make sense to disregard CAIR or CSAPR in the context of RPGs, because all electric generating units
in the state are subject to their emission limitations, while only some electric generating units are subject to BART.
Response: Xcel raises concerns that we have allowed other states to rely on CSAPR to meet
RPGs, citing to EPA's action on Michigan. As discussed elsewhere, Michigan had less than a
5% impact on visibility in downwind states. With respect to its own Class I areas, Michigan
assessed the contribution of three Midwestern states together and concluded that the 47%
reduction in SO2 emissions was sufficient for reasonable progress. A number of non-EGUs also
have significant impacts on visibility in Michigan's Class I areas and BART determinations (by
the State and by EPA) require additional reductions from these sources. More generally, with
respect to Xcel's comment that EPA provides no reasoned basis for rejecting consideration of
CAIR in the reasonable progress context, we disagree that this is what we proposed to do in
Texas. In our proposal we took into account the impact of CSPAR/CAIR on emissions in Texas,
but determined that participation in this program would not ensure reasonable progress at the
Wichita Mountains, Big Bend, and the Guadalupe Mountains. It is important to remember that
the test for determining whether a BART alternative provides for greater reasonable progress
than BART is based on improvements in visibility on average across all Class I areas. We
noted in 2005 that the determination that CAIR provided for greater reasonable progress than
BART did not answer the question of whether more than CAIR would be required in a regional
haze SIP:
180
-------
Our determination that CAIR makes greater reasonable progress than BART for EGUs is not a
determination that CAIR satisfies all reasonable progress requirements in CAIR affected States.
Each State, whether in the CAIR region or not, is required to set reasonable progress goals for
each Class I area within the State as required in regional haze rule section 308(d)(1), and to
develop long term strategies, considering all anthropogenic sources of visibility impairing
pollutants, as required by section 308(d)(3). In setting the reasonable progress goals, the State is
to consider the amount of visibility improvement needed to achieve a uniform rate of progress
towards natural background conditions in the year 2064. (This uniform rate of progress is
sometimes referred to as the default glide-path). The State is also to consider the statutory
reasonable progress factors contained in CAA section 169A(g)(l).85 In doing so, we anticipate
that States will take into account the degree to which CAIR emissions reductions are projected to
bring visibility conditions at its Class I areas in line with the default glide path. In some States,
the improvements expected from CAIR, combined with the application of the reasonable
progress factors to other source sectors, may result in a determination that few additional
emissions reductions are reasonable for the first long-term strategy period. Nonetheless, each
State is required to set its reasonable progress goals as provided by the regional haze rule and
cannot assume that CAIR will satisfy all of its visibility-related obligations. 70 FR 39104, 39143
(July 6, 2005). This is consistent with our statement regarding the Georgia SIP cited by the
commenter, where we noted that any control requirements imposed as BART would also satisfy
reasonable progress. EPA's determination that CSPAR would provide for greater reasonable
progress than BART did not result in the imposition of any control requirements imposed as
BART.
Our response concerning Texas' use of the cost level of $2,700 from CAIR as a basis for
deciding not to include some sources under its Regional Haze SIP is discussed more fully in
consistency section of this document.
10. Installation of Controls Beyond the First Planning Period
Comment: [NERA (0061) p. 6]
In the report prepared for Luminant, NERA also considered the Cost-effective ness of the
proposed scrubber retrofits. Those retrofits cannot be considered for reasonable progress
because, as EPA admits, they cannot be installed by the end of the first planning period (2018).12
However, even if these controls could be installed by 2018, they would only degrade the
economic Cost-effective ness of EPA's proposed FIP.
Footnotes:
12 79 Fed. Reg. at 74,874.
Response:
The FIP imposes cost effective controls as an overall matter. We do not agree with the implied,
artificial differentiation suggested by the comment, because "even if these controls could be
installed by 2018," they would in fact still be deemed cost effective. The FIP accommodates a
reasonable construction timeframe by allowing that emission limitations would be met after
2018, and they in fact are considered for reasonable progress as part of the long-term strategy,
181
-------
which differs from whether they warrant inclusion in the numerical reasonable progress goals.
Thus, the comment is not accurate in stating what "EPA admits" and the point about
"degrad[ing]" cost-effectiveness has no identifiable bearing on the content of the final rule.
Comment: EPA's proposal exceeds its regulatory authority. [Luminant (0061) p. 4]
Luminant argued that the EPA has no legal authority to require the installation of scrubbers at
Luminant's units in 2020 as part of its proposed FIP. Per EPA's regulations, the Texas SIP
submission at issue—which was submitted by the Texas Commission on Environmental Quality
("TCEQ") to EPA in 2009—covers only the first regional haze planning period (2008-2018) and
addresses emission reductions needed to meet the 2018 interim goal. EPA concedes that the
scrubbers it is proposing would take at least five years to construct and thus cannot be
operational by 2018.15 EPA's proposal thus clearly runs afoul of the statutory factors for
reasonable progress—which require consideration of "the time necessary for compliance" 16—
and exceeds EPA's FIP authority, which is only to fill the gaps in a SIP, not to add regulation
outside the required scope of the SIP submission.17 Whether these units must install controls to
meet subsequent goals for later years is a matter fundamentally to be decided by Texas in the
second planning period, not by EPA in the current period.
Footnotes:
15 EPA, Technical Support Document for the Oklahoma and Texas Regional Haze Federal Implementation Plan (FIP
TSD) 7 (Nov. 2014) ("FIP TSD").
16 42 U.S.C. § 7491(g)(1)
17 Id. § 7602(y).
Response:
We disagree with the comment. The disapproval of the SIP submission authorizes and mandates
that we take action to promulgate a FIP. See CAA Section 110(c). When we act under FIP
authority we act in the place of the State, including by imposition of enforceable emission
limitations, to ensure that CAA requirements are met and to ensure the inadequacy of the SIP is
addressed. See CAA Section 302(y). As a practical matter, the enforceable emission limitations
or control measures required by a FIP may not be implemented and put into effect at the earliest
time directed by Clean Air Act. The timely installation of controls is ideal and would prove
workable had such controls been required under an approvable SIP, but this fact does not excuse
overdue controls from being installed and operated. Even as we note there is no prohibition
against early compliance for sources that would meet the FIP requirements, we also anticipate
the construction timeframes that are reasonably necessary for new controls. Thus, our FIP does
properly take account of the "time necessary for compliance." The comment suggests that
consideration of this statutory factor should exempt controls or that it necessitates special
coordination with TCEQ's next obligation to submit a comprehensive periodic SIP revision for
regional haze; however, we have determined that our specified FIP controls are required by the
CAA for the current regional haze implementation period. These are due, if not overdue,
emission limitations that should be met as soon as reasonable construction timeframes would
allow. With the promulgation of enforceable emission limitations in the FIP, the TCEQ will
have no difficulty in crediting these FIP controlled sources for purposes of studying what
additional controls can and should be required in the next planning period. In other words, the
182
-------
FIP is no obstacle to Texas planning and requirements that will be appropriate for the second
regional haze planning period.
Comment: EPA has no authority to require emission controls in 2020 for the interim 2018
reasonable progress goal [Luminant (0061) p. 144]
Luminant stated that EPA steps well outside its authority in attempting to require the installation
of new scrubbers at seven Texas EGUs in 2020. The time period at issue in Texas' submittal,
and thus the scope of EPA's review and authority, is limited to the first regional haze planning
period—2008 to 20 1 8.879 As EPA has explained: "The RHR [regional haze rule] requires control
strategies to cover an initial implementation period extending to the year 2018, with a
comprehensive reassessment and revision of those strategies, as appropriate, every 10 years
thereafter."880
Thus, Luminant stated that the only relevant question in this submittal is whether Texas' SIP will
achieve interim goals for reasonable progress in 2018—what happens in 2020 is irrelevant for
present purposes. Indeed, the "time necessary for compliance" is one of the factors to be
considered in establishing reasonable progress.881 And EPA concedes, as it must, that "typical
SO2 scrubber installations can take up to five years to plan, construct and bring to operational
readiness."882 And EPA further concedes that it "cannot assume that the SO2 controls we are
proposing will be installed and operational within this planning period, which ends in 2018."883
Luminant noted, accordingly, EPA correctly does not take into account any emission reductions
from the scrubber installations it is proposing to require in setting the reasonable progress goals
for its proposed FIP.884 But EPA errs and steps outside of its authority when it nevertheless
asserts the authority to impose the reductions in 2020. EPA's regional haze guidance explains
that time constraints which "preclude the installation of controls ... by 2018" "should be
considered in setting the RPG and in establishing the SIP requirements to meet the RPG,"885
And EPA has further explained that "[i]n setting the RPGs, states must also consider the rate of
progress needed to reach natural visibility conditions by 2064 (referred to hereafter as the
'Uniform Rate of Progress (URP)') and the emission reduction measures needed to achieve
that rate ofprogress over the 10-year period of the SIP,"886 This limitation extends to the
contents of a long-term strategy for regional haze as well: "The LTS [long-term strategy] is the
compilation of all control measures a state will use during the implementation period of the
specific SIP submittal to meet applicable RPGs."887 In no prior regional haze action that we have
found has EPA ever finalized a FIP for the first regional haze planning period that required
reasonable progress controls after 20 1 8.888 It may not lawfully do so here.
Luminant stated that EPA "is a creature of statute" and thus "has no constitutional or common
law existence or authority, but only those authorities conferred upon it by Congress."889 If EPA
lacks authority under the CAA to promulgate a rule, "its action is plainly contrary to law and
cannot stand."890 Here, EPA's proposal to require scrubbers in 2020 contravenes the statute in at
least two respects. First, it disregards the "time necessary for compliance" statutory factor in 42
U.S.C. § 7491(g)(1). EPA concedes that these controls cannot be installed and operational
within this SIP period, and thus this factor dictates that consideration of the controls be deferred
183
-------
until the next planning period. That EPA's delay in reviewing Texas' SIP revision is the reason
for this fact does not make EPA's proposal lawful. EPA had Texas' full and complete
submission in early 2009 and thus could have completed its review in a timely manner. Second,
EPA's proposal asserts FIP authority beyond the scope of the required SIP submission. EPA's
authority to issue a FIP under 42 U.S.C. § 7410(c) is limited to filling the gap left in the SIP by
virtue of its disapproval.891 Indeed, the Clean Air Act delineates the scope of a FIP as "a plan (or
portion thereof) promulgated by the Administrator to fill all or a portion of a gap or otherwise
correct all or a portion of an inadequacy in a State implementation plan . . . ,"892 There is no gap
in the Texas regional haze SIP that requires emission controls in 2020, nor do such controls
address any "inadequacy" in Texas' submission for the first planning period. The required scope
of the state's submission extended only to 2018, and achieving reasonable progress in 2018, and
thus EPA's FIP authority extends no further.893 The need for any emission controls in 2020, if
any, is properly considered only in the second planning period, for which SIPs are not due until
July 31,2018.
Footnotes:
879 Id. at 74,818.
88° qq pe(j RCg at 30.252 (EPA proposed approval of Idaho reasonable progress goals and long-term strategy) 77
Fed. Reg. 30,454, 30,458 (May 23, 2012) (EPA proposed approval of Oregon reasonable progress goals and long-
term strategy).
881 42 U.S.C. § 7491(g)(1); 40 C.F.R. § 51.308(d)(l)(i)(A).
882 FIP TSD at 7.
883 Id.
884 79 Fed. Reg. at 74,886.
885 Reasonable Progress Guidance at 5-2.
886 76 Fed. Reg. 16,168, 16,173 (Mar. 22, 2011) (proposed approval of Oklahoma URP) (emphasis added).
887 77 Fed. Reg. at 30,251 (emphasis added).
888 See 77 Fed. Reg. 20,894, 20,944 (Apr. 6, 2012) (compliance deadline of July 31, 2018 for Antelope Valley
StationUnits 1 and 2); 77 Fed. Reg. 57,864, 57,916 (Sept. 18, 2012) (compliance deadline of July 31, 2018 forthe
Blaine County Compressor Station); 79 Fed. Reg. 52,420, 52426 (Sept. 3, 2014) (compliance deadline of
December 31, 2018 for Phoenix Cement Clarkdale Plant Kiln 4 and CalPortland Cement Rillito Plant Kiln 4).
889 Michigan v. EPA, 268 F.3d 1075, 1081 (D.C. Cir. 2001).
890 Id.
891 See Arizona v. EPA, 151 F.3d 1205, 1212 (9th Cir. 1998) (A FIP is "specifically meant to fill in the gaps where
a State has failed to submit an SIP or where the State's SIP does not satisfy minimum criteria under the CAA.").
892 42 U.S.C. § 7602(y).
893 See 64 Fed. Reg. at 35,734 ("[T]he final rule requires control strategies to cover an initial implementation period
extending to the year 2018, with a reassessment and revision of those strategies, as appropriate, every 10 years.").
Response: We understand the comment to argue that our authority to issue a FIP under 42
U.S.C. 7410(c) is limited to "filling the gap left in the SIP," and, by proposing requirements
outside of the scope of Texas' SIP submission, we are exceeding that authority. 42 U.S.C. §
7410(c) mandates that we promulgate a FIP within two years after the Agency disapproves a SIP
in whole or in part. The "gap" language to which Commenter refers is located at 42 U.S.C. §
7602(y), which defines FIP as "a plan (or portion thereof) promulgated by the Administrator to
fill all or a portion of a gap or otherwise correct all or a portion of an inadequacy in a [SIP.]" We
discern the comment to be claiming one of two things: (1) Our FIP authority is limited by the
scope of the SIP submission, or, (2) since the required RPGs are limited to the first planning
period, our FIP authority is likewise limited to the first planning period. We disagree with both
possible claims. The Clean Air Act clearly establishes our authority to promulgate a FIP that
184
-------
addresses the requirements of the Regional Haze program where a State's SIP submission fails to
meet the program requirements, including circumstances where the Long Term Strategy
necessitates controls that cannot be installed during the planning period.
We disagree with any assertion that our FIP authority is limited by the scope of the SIP
submission we could not approve. "Gap" or "inadequacy" refers not to a "gap" in the plan as
submitted but to a "gap" in the plan that fails to address the Regional Haze program requirements
contained in the statute and regulations. Where a SIP submission fails to address a major
component of the program (such as the establishment of RPGs), we are required to disapprove
the SIP to promulgate a FIP that does address that necessary component. A disapproved SIP's
exclusion of sources from regulation or air pollution controls does not limit or exclude those
sources from controls under a FIP. Any argument to the contrary misapprehends the structure of
the CAA and the purposes and historic uses of FIP authority.
If Commenter's "gap" argument means to assert that the authority to impose FIP controls is
limited, in a temporal sense, by the requirement to set RPGs for the planning period, we also
disagree with the comment. It contravenes both the plain language of the statute and regulations
as well as the common understanding of the Regional Haze program. The Regional Haze
program includes three general components: "measures necessary to make progress toward
meeting the national goal," a long-term strategy, and BART.106 In issuing regulations related to
the first component, we require a SIP to include "reasonable progress goals."107 In our guidance
on setting RPGs, we provided a definition of RPGs, explaining that RPGs are "interim goals that
represent incremental visibility improvement over time toward the goal of natural background
conditions [... ,]"108 The first planning period, ending in 2018, includes RPGs specific to that
planning period; however, as discussed below, there is no such limitation to the first planning
period for provisions in the long-term strategy and BART requirements. As two of the three
major components of the Regional Haze program may extend beyond the first planning period, it
follows that EPA has FIP authority fill in "gaps" or "inadequacies" related to those components
irrespective of whether they are required during the first planning period or later.
We also disagree with comment's assertion that our proposed FIP disregards the "time necessary
for compliance" factor contained in 42 U.S.C. 7491(g)(1). The requirements for a long-term
strategy are not limited to the first planning period.
Congress declared that the "national goal" of the Regional Haze program is "the prevention of
any future, and the remedying of any existing, impairment of visibility in mandatory Class I
Federal areas which impairment results from man-made air pollution."109 To meet this goal,
Congress ordered us to issue regulations requiring States to implement SIPs "to contain such
emission, limits, schedules of compliance and other measures as may be necessary to make
reasonable progress toward meeting the national goal[,]" and the SIPs must include, among other
106 79 Fed. Reg. 74818,74824.
107 40 C.F.R. 51.308(d)(1).
108 OAQPS, EPA, Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program (June 1,
2007), 1-2.
109 42 U.S.C. § 7491(a)(1).
185
-------
requirements, "a long-term (ten to fifteen years) strategy for making reasonable progress toward
meeting the national goal[.]"110 Congress stated that "in determining reasonable progress there
shall be taken into account [...] the time necessary for compliance[.]"m Thus, the Clean Air Act
does not restrain EPA from requiring sources to install controls outside of the initial planning
period in the long-term strategy, but rather, it requires us to consider the time necessary for
compliance when determining what constitutes "reasonable progress" toward the national
goal.112
We issued regulations including, in addition to BART requirements, four "core requirements"
including reasonable progress goals, calculations of baseline and natural visibility conditions, a
long-term strategy and a monitoring strategy and other implementation plan requirements.113 As
part of these SIP requirements, a State must "consider [...] the time necessary for compliance"
when establishing RPGs.114 The long-term strategy component "must include enforceable
emission limitations, compliance schedules, and other measures necessary to achieve the
[RPGs]" however, this "must include" language does not limit the long-term strategy to only
those measures to achieve RPGs.115 Rather, it mandates inclusion of certain requirements to meet
the RPGs without limiting planning authority to include other requirements to meet the national
goal. Thus, there is no explicit statutory or regulatory requirement to limit control options to the
first planning period, and further, EPA's long-standing interpretation of the Clean Air Act and its
regulations explicitly provide for the type of post-planning period controls to which Commenter
objects.
In issuing the Regional Haze rule and consistent with 42 U.S.C. 7491(b)(2)(B), we interpreted
"long-term strategy" as the "control measures that are needed to ensure reasonable progress,
together with a demonstration that those measures will provide for reasonable progress during
the 10 to 15 year period."116 Again, the reference to "reasonable progress," as opposed to RPG,
refers to the national goal, and as noted above, the regulation provides for flexibility to achieve
the national goal. In issuing its guidance on RPGs, the very situation to which Commenter
objects is laid clear:
The second factor is the "time necessary for compliance." It may be appropriate
for you to use this factor to adjust the RPG to reflect the degree of improvement
in visibility achievable within the period of the first SIP if the time needfor full
implementation of a control measure (or measures) will extend beyond 2018.
For example, if you anticipate that constraints on the availability of construction
labor will preclude the installation of controls at all sources of a particular
category by 2018, the visibility improvement anticipated from installation of
controls at the percentage of sources that could be controlled within the strategy
110 Id. at 7491(b)(2).
111 Id. at 7491(g)(1).
112 While the CAA does provide the statutory factors, including "time necessary for compliance," we note that it
does not delineate 10 year planning periods. These are specified by the RHR.
113 40 C.F.R. § 51.308(d).
114 Id. at 51.308(d)(1).
115 Id. at 51.308(d)(3).
116 64 Fed. Reg. 35714,35734.
186
-------
period should be considered in setting the RPG and in establishing the SIP
requirements to meet the RPG.117 (emphasis added)
Accordingly, we are required only to "consider [...] time necessary for compliance" when
establishing RPGs setting requirements to meet the RPGs, and the proposed FIP accords with our
guidance.118 By noting that scrubber retrofits may take up to five years to install, yet also noting
that scrubber upgrades may be installed within three years, we proposed RPGs that only account
for the visibility benefits associated with scrubber upgrades.119 In doing so, we considered the
time necessary for compliance in establishing RPGs, yet exercised our authority to propose a
long-term plan including emission limits that likely require controls that may not be operational
during the planning period.
We also note that, due to sequencing of the SIP process, including submission, review, revisions,
re-review, and FIP preparation, public notice and promulgation, the CAA anticipates a
substantial period of time between SIP submission and the promulgation of a FIP, as may
ultimately be essential for carrying out CAA requirements. Those sources that should be subject
to new controls under the CAA may take certain advantages from delays in having those control
requirements imposed, but they do not have the benefit of an exclusion from CAA requirements.
Instead, any emission limitations that prove to be required by the CAA for the first planning
period need to be achieved at their soonest opportunity, not delayed, deferred or avoided for later
planning periods at those times when other sources should be the focus of potential additional
controls. Accordingly, we do not think the comment takes proper account of the time
engagement required to promulgate a FIP within a planning period (if a FIP proves needed) or
the significance of the CAA's contemplated ten- to fifteen-year long-term strategy, which could
not be included in a FIP absent some degree of flexibility to extend FIP requirements beyond the
date set for the end of the first planning period.120
Finally, we note that we have proposed controls outside of the planning period in other FIPs,
including our proposed FIP for Arkansas and the FIP for Wyoming. Additionally, the Oklahoma
FIP was stayed during litigation, resulting in a revised compliance date beyond 2018.
Comment: EPA lacks authority to include emission controls that cannot be implemented
during the regional haze planning period. [Associations (0059) p. 19-20]
117 OAQPS, EPA, Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program (June 1,
2007), 5-2.
118 40 C.F.R. § 51.308(d)(1)(A).
119 79 Fed. Reg. 74818,74874.
120 Additionally, the State's responsibility for its SIP, or EPA's authority when needing to promulgate a FIP, is not
shown to be discontinuous or intermittent because of the requirement to conduct comprehensive periodic plan
revisions. Instead, there is ongoing authority to ensure reasonable progress from sources within the State. As
demonstrated by the progress report requirements of the RHR, the regional haze program expressly contemplates
that a State may determine the need for "further revision of the existing implementation plan" because the existing
plan "is or may be inadequate"; in such situations, the rule sets out, deficiencies are to be addressed promptly
("within one year"), not deferred to the next comprehensive periodic plan revision, as the comment thinks to be
necessary. See 40 CFR 51.308(h).
187
-------
The Associations argued that the proposed FIP exceeds EPA's legal authority under the regional
haze program because it would require individual sources to install new emission control devices
in 2020, after the 10-year regional haze planning period has concluded. Under EPA's regional
haze rule, States must prepare SIPs that adopt control strategies over an initial implementation
period from 2008 to 2018 and must then conduct "a comprehensive reassessment and revision of
those strategies, as appropriate, every 10 years thereafter." 77 Fed. Reg. at 30,252. EPA
recognizes the limited scope of the SIP in the preamble, noting that the proposal "addresses
regional haze for the first planning period from 2008 through 2018." 79 Fed. Reg. at 74,818.
According to the Associations, focusing exclusively on emissions controls that can be
implemented during the interim 2008 to 2018 planning period is consistent with both the
reasonable progress goals and long-term strategy components of the States' regional haze plans.
For example, EPA guidance directs States to focus on emissions controls at sources that "could
be controlled within the strategy period" when "setting the RPG and ... establishing the SIP
requirements to meet the RPG." EPA, Guidance for Setting Reasonable Progress Goals Under
the Regional Haze Program 5-2 (June 1, 2007). In evaluating SIPs submitted by other States,
EPA has further explained that "[i]n setting the RPGs, states must also consider the rate of
progress needed to reach natural visibility conditions by 2064 ... and the emissions reduction
measures needed to achieve that rate of progress over the 10-year period of the SIP." 76 Fed.
Reg. 16,168, 16173 (Mar. 22, 2011) (proposed approval of Oklahoma Uniform Rate of
Progress) (emphasis added). Likewise, EPA has previously explained that "[t]he [long-term
strategy] is the compilation of all control measures a state will use during the implementation
period of the specific SIP submittal to meet applicable RPGs." 77 Fed. Reg. at 30,251 (emphasis
added).
Despite recognizing these limits on the scope of SIPs under the interim planning period, the
Associations noted that the EPA proposes a FIP that would require emission controls that cannot
be implemented until at least 2020. See FIP TSD at 7 (acknowledging that "typical SO2
scrubbers take up to five years to plan, construct, and bring to operational readiness"). In fact,
EPA acknowledges that it "cannot assume that the SO2 controls we are proposing will be
installed and operational within this planning period, which ends in 2018." Id. In this respect
EPA's proposal is inconsistent with past regional haze actions, where EPA has consistently
limited the scope of FIPs to control measures that can be implemented during the interim
planning period. See, e.g., 77 Fed. Reg. 20,894, 20,944 (Apr. 6, 2012) (applying a July 31,
2018, compliance deadline in North Dakota FIP); 77 Fed. Reg. 57,864, 57,916 (Sept. 18,2012)
(applying a July 31, 2018, compliance deadline in Montana FIP); 79 Fed. Reg. 52,420, 52,426
(Sept. 3, 2014) (applying a December 31, 2018, compliance deadline in Arizona FIP).
The Associations stated that the EPA's proposal to require installation of emission controls in
2020 exceeds EPA's authority under the Clean Air Act and is, therefore, unlawful for at least two
independent reasons. First, EPA's proposed FIP ignores the statutory mandate to consider the
"time necessary for compliance" as a factor in determining reasonable progress. 42 U.S.C. §
7491(g)(1). Because the time necessary for compliance will extend into the next planning
period, beginning in 2018, EPA is obligated by statute to defer consideration of such emission
controls until the next planning period. Further, by proposing emission controls that cannot be
implemented until 2020, EPA's proposed FIP would unlawfully extend beyond the scope of
188
-------
Texas' required SIP submission. Under the Clean Air Act, the scope of EPA's FIP authority is
limited to preparing "a plan (or portion thereof)" that "fill[s] all or a portion of a gap or otherwise
corrects] all or a portion of an inadequacy in a State implementation plan." 42 U.S.C. §
7602(y). Yet, Texas was under no obligation under the Clean Air Act or Regional Haze Rule to
develop a SIP that extended beyond 2018. Because the scope of Texas' SIP obligation was
limited to achieving reasonable progress during the interim 2008 to 2018 planning period, EPA's
FIP authority is necessarily subject to the same limits. Thus, determination of whether additional
emission controls are needed after 2018 must, by statute, be deferred until the next planning
period.
Response: Many of the concerns expressed in this comment are addressed by the response we
provided, above, to a previous comment in this section. Commenter expresses concerns
addressed above in this section.
We acknowledge the claim that the FIP for Texas is inconsistent with previous FIPs because
other FIPs are confined to measures that can be implemented within the first planning period.
We disagree with this claim, because we have proposed or promulgated FIPs with control
measures that will be implemented beyond the first planning period. Examples are referenced
above. Our FIP imposes emissions limitations that we conclude to be necessary under the CAA
for the first planning period. Ideally, these controls would be installed and the emission
limitations achieved, so the visibility improvements can be realized and built on in a subsequent
comprehensive periodic SIP revision (see 40 C.F.R. 51.308(f)). We cannot agree that any
delays in promulgating the FIP would create exclusions or deferments from meeting the emission
limitation requirements of the CAA. Instead, it is our duty to ensure that sources subject to
emission limitations for the first planning period meet those limitations in as timely a way as
circumstances allow. As discussed above, the planning period deadline is relevant to RPGs but
not to the long-term strategy, and our acknowledgement of reasonably necessary construction
timeframes is fully in keeping with the consideration of the time necessary for compliance.
We further address the Associations' comments alleging inconsistencies with our previous
actions in a separate response.
Comment: [UARG (0065) p.22-24] As part of their argument that EPA's proposed FIPs for
Texas and Oklahoma are unlawful, UARG noted that the EPA's FIP would impose requirements
that cannot be achieved during the first planning period of the regional haze program. This is a
problem of EPA's own making. It is, moreover, inconsistent with EPA's regulations and
requires that the FIP be withdrawn. According to EPA, the Agency received the Texas regional
haze SIP on March 31, 2009. Pursuant to CAA § 110(k)(l)(B), that plan was deemed complete
by operation of law on September 30, 2009. Pursuant to CAA § 110(k)(2), therefore, EPA was
obligated to take final action on the Texas SIP by September 30, 2010. EPA did not publish
even its proposed action on the Texas SIP until December 16, 2014, and EPA is not scheduled to
take final action on that SIP until September 4, 2015, nearly five years after the statutory
deadline. After years of delay, EPA now proposes to disapprove (in part) Texas' regional haze
SIP for the first planning period of the regional haze program, which runs from 2008 through
2018, and to impose requirements that, by EPA's admission, cannot be implemented until 2020,
189
-------
well after the end of the program's first planning period and, thus, well after the date by which
the RPG is to be achieved. EPA has no authority to take this action.
UARG noted, consistent with EPA's regulations, the Texas SIP submission covers only those
emission reductions needed in the first planning period. The CAA and EPA's regional haze rule
state that, in assessing measures needed to achieve reasonable progress, states (and EPA) must
take into account "the time necessary for compliance" with those measures. CAA § 169A(g)(l);
40 C.F.R. § 51.308(d)(l)(i)(A). One critical consideration is whether measures intended to
ensure reasonable progress during the first planning period can in fact be implemented during
that planning period. Indeed, EPA's Reasonable Progress Guidance expressly states that "[i]t
may be appropriate" for states, in assessing the time-necessary-for-compliance factor, "to use
this factor to adjust the RPG to reflect the degree of improvement in visibility achievable within
the period of the first SIP if the time needed for full implementation of a control measure (or
measures) will extend beyond 2018." Reasonable Progress Guidance at 5-2. Accordingly,
reasonable progress requirements for emission reductions that could not or would not be required
to be achieved within the first planning period should be adjusted, consistent with the RPGs that
are intended to reflect those emission reductions, to mandate only those reductions that in fact
can be achieved within the first planning period. As EPA explained in the Reasonable Progress
Guidance, where time constraints "preclude the installation of controls ... by 2018, the visibility
improvement anticipated from installation of controls at the ... sources that could be controlled
within the strategy period [z.e., by 2018] should be considered in setting the RPG and in
establishing the SIP requirements to meet the RPG." Id. at 5-2 (second emphasis added).
UARG stated that the EPA candidly acknowledges that its FIP would impose requirements that
extend beyond the first planning period. 79 Fed. Reg. at 74,874 ("we cannot assume that the
SO2 controls we are proposing will be installed and operational within this planning period,
which ends in 2018"). Instead of adjusting its proposed reasonable progress requirements,
however, EPA proposes to maintain its post-2018 emission reduction requirements but to
exclude the effects of the post- 2018 emission reductions from the calculation of the RPGs it
proposes:
We note that we do not anticipate implementation of the identified scrubber retrofits by
the end of 2018. Therefore, we are only adjusting the RPGs established by the states to
reflect the additional anticipated visibility benefit from the scrubber upgrades over the
2018 projected visibility conditions. Id. at 74,866.
UARG stated that although EPA's proposed RPGs attempt to maintain the illusion that EPA is
limiting its rulemaking action to the first planning period, the proposed FIP itself plainly imposes
requirements applicable in the second planning period. The proposed FIP thus would violate the
CAA's SIP and FIP provisions, which limit EPA's promulgation of FIPs to filling any "gap" left
by a SIP. CAA § 302(y). Texas' SIP was properly limited to emission reduction requirements
that could be implemented during the first planning period; Texas' obligation to submit a SIP to
address reasonable progress in the second planning period lies in the future, and thus there is at
this time no gap that a FIP could lawfully fill with respect to the second planning period. Only if
Texas, in the future, defaults on its obligation to submit an approvable SIP addressing reasonable
progress for the second planning period could EPA even conceivably have authority to take
190
-------
action with respect to any regional haze requirements that cannot be, and are not required to be,
implemented by 2018.
Response: This comment is highly similar to earlier comments to which we already provide
responses, above. Despite any delays in finalizing our action on the Texas SIP or in
promulgating the FIP, our duty to act on the SIP and our FIP authority to impose the controls
required by the CAA persists and is not forfeited. After our deadline for action on the SIP, we
became subject to a judicially supervised deadline that acknowledges our continuing obligation
to take action. Our authority to impose emission limitations needed for the first planning period
or for a long-term strategy also continues after rulemaking or litigation delays would project that
achievement of those limitations would fall after 2018. To assert otherwise is fundamentally at
odds with the objectives of the CAA, and the long-term goal of the regional haze program. Air
quality control authority—whether exercised under a SIP or a FIP—is not intermittent such that
sources avoid or defer the control requirements that are due under the CAA because of an
unapprovable SIP or a delayed FIP promulgation. Even as the RHR establishes a program of
"comprehensive periodic" SIP revisions for regional haze, it too does not have limited windows
for the imposition of controls. On the contrary, it expressly envisions that deficiencies may—
and even must—be addressed even in the middle of a planning period. See 40 CFR
51.308(g)(4).
Comment: EPA unreasonably required controls at Coleto Creek Unit 1 despite
acknowledging that visibility benefit will not be obtained in the applicable planning 2008-
2018 period. [CCP (0075) p. 9]
CCP stated that the "time of compliance" is an express factor required for development of RPGs
under CAA Section 169A(g). Texas' SIP covers a planning period of 2008-2018. EPA admits
that its requirement to install WFGD scrubbers at Coleto Creek Unit 1 will have no impact on
Texas' plan to meet 2018 goals. EPA has no authority to require installation of WFGD
scrubbers in TCEQ's current plan when there will be no benefit until after the applicable
planning period. Indeed, even EPA "requests] that Oklahoma and Texas consider the additional
visibility improvements anticipated from any proposed FIP controls implemented after 2018 with
the submission of their next regional haze SIPs due July 13, 2018." Technical Support Document
at 7. Thus, to the extent a scrubber retrofit on Coleto Creek Unit 1 is considered at all, it should
only be evaluated as part of the next planning period.
Response: We disagree that Coleto Creek Unit 1 should only be evaluated in the next planning
period. Under the technical record outlined by our proposal, controls are warranted and should
be imposed for the first planning period. Our reference to Oklahoma and Texas taking account
of FIP controls is a simple matter of coordination and taking account of emission limitations and
visibility improvements that are in place or soon to be achieved. Taking account of the
imposition of FIP controls, Oklahoma and Texas will be able to focus on other candidates for
new controls as it establishes requirements for the next planning period and continues the work
toward the national goal. The fact that the benefits of any controls may not be realized until after
2018 does not disprove our determination that such controls were needed for the first planning
191
-------
period or deprive our authority to impose the emission limitations so they are timely achieved
under the circumstances.
Comment: Time Necessary for Compliance [GCLC (0063) p. 15]
According to GCLC, EPA has proposed that seven Texas coal-fired EGUs will be required to
install scrubbers in 2020. This is beyond 2018, the year when the first regional haze period ends.
This violates the "time necessary to comply" factor. Since these controls cannot be installed by
the end of this first planning period, consideration of these types of controls must be deferred
until the conclusion of the next planning period. While EPA may want to now act quickly, it is
not the fault of Texas or its operators that a plan submitted in 2009 is only now being reviewed
and EPA is only now proposing its FIP. EPA's delay does not excuse, or provide reason, to go
beyond the limits of the phased implementation schedule. Furthermore, since a scrubber cannot
be installed by 2018, that means there is no requirement to install such scrubber on the targeted
Texas units. The question of whether these units must install controls, therefore, can only be
addressed by Texas in the second planning period, not by EPA in this current period. EPA has
acted too soon and without legal authority.
Response: We disagree with the comment for the reasons provided in responses to other
comments above in this section. We do not agree with the claim that "since a scrubber cannot be
installed by 2018, that means there is no requirement to install such scrubber on the targeted
Texas units." The prospect of delayed implementation of CAA requirements provides no
exclusion from CAA requirements. The FIP establishes enforceable emission limits within the
boundaries of our statutory and regulatory authority, including emission limits that may not be
achieved until 2020 to allow for additional time required to install scrubbers.
Comment: The proposed timing in the FIP is unjustified [EEI (0076) p. 4-5]
EEI stated that the Agency's proposed FIP raises serious concerns in its attempt to require
reductions to meet the 2018 RPGs for three federally-protected Class I visibility areas in Texas
(Big Bend National Park and Guadalupe Mountains National Park) and Oklahoma (Wichita
Mountain Wilderness Area). Id.
EEI asserted that the Agency is attempting to require controls well beyond the Texas SIP that it
knows will not be in place or effective within the first planning period, i.e., by 2018. EPA's
proposal thus exceeds the scope of the state's submission and the Agency's authority to issue a
FIP. EPA's proposal is at variance with the "time necessary for compliance" statutory factor in
42 U.S.C. § 7491(g)(1) for determining reasonable progress. Moreover, it also is unreasonable
to require the proposed emissions controls on a tight regulatory timeline.
Given that the regional haze program's target extends to 2064, EEI noted that EPA's insistence
on requiring extensive control equipment installations that will not be technically in place or
effective to meet a 2018 milestone is unjustified at this time. Instead, these controls should only
be considered by the states during the next planning period that runs between 2019 and 2028.
This also is a more reasonable approach that would allow for the consideration of updated data,
192
-------
use contemporaneous monitoring and meteorological conditions, and would avoid overreliance
on data and modeling that are more than 10 years out of date. Further, such an approach would
allow the coordination of these important investment and regulatory decisions with the
implementation of other pending regulations. Finally, this approach would give states and
regulated entities the opportunity to conduct integrated compliance planning in ways that are
consistent with provision of reliable and affordable electric power.
[EEI (0076) p. 11] EPA's insistence in the proposed FIP on requiring extensive control
equipment that will not technically be in place or effective to meet a 2018 milestone is
problematic. Instead, the proper course is for the states to consider these and other control
options as part of the suite of possible actions that can be taken during the next planning period
that runs between 2019 and 2028.
Response: We disagree with the comment for reasons explained in response to other comments.
In addition, we note our disagreement with the comment for suggesting that a RH SIP that
warrants disapproval is neither suited nor required to be addressed by a FIP. On the contrary, the
FIP is meant to address the deficiencies identified in the SIP for the current period. The fact that
the RH SIP for the current planning period is not approvable is not a basis for the deferring and
exempting requirements and measures to make reasonable progress during that planning period.
Instead, the FIP will impose the enforceable emission limitations, and having addressed these
sources, we have full confidence that Texas will be better able to focus on other candidates for
making reasonable progress in its next comprehensive SIP revision. Overall, we disagree that
our FIP presents a tight regulatory timeline. It is an appropriate timeline for highly cost effective
control measures needed for reasonable progress, and it is notable that the comment has not
suggested or acceded to any concrete alternate timeframe for our consideration. The comment's
assertions regarding electric reliability are addressed in responses to specific comments on that
topic.
Comment: [Nucor Steel (0058) p. 3] Nucor Steel stated that EPA's interpretation and rule is
contrary to the CAA because it seeks to establish reductions that it deems necessary to meet
reasonable progress goals beyond the 2018 planning period addressed by the states of Texas and
Oklahoma in their SIPs.
Response: For the reasons discussed in our responses to other similar comments, we disagree
with this comment. The FIP establishes reductions to meet reasonable progress goals as well as
the long-term strategy.
Comment: [OG&E (0057) p. 1-2, 4] OG&E noted, as evidenced by the ODEQ SIP and the
Proposed Rule, Oklahoma and EPA do not believe further emission reductions are needed from
Oklahoma sources at this time and may not be needed in the near future. Oklahoma is currently
making significant emissions reductions from its coal-fired fleet and visibility monitoring in the
Wichita Mountains Class I area is improving even before a large slate of emission reduction
programs by the power sector has been fully implemented over the next roughly 10 years.
OG&E believes the final rule should affirm it is not necessary to impose additional emission
control requirements at this time from utilities in Oklahoma.
193
-------
OG&E supported the EPA Proposal that no additional emission controls for Oklahoma sources
are currently needed2 and that the cumulative actions being taken by Oklahoma and Texas as
well as other upwind states to comply with this and other EPA rules including the Cross State
Air Pollution Rule ("CSAPR") be accounted for in the next iteration of the evaluation of the
Regional Haze program in Oklahoma.
OG&E stated that utilities in Oklahoma are well underway toward complying with the Regional
Haze State Implementation Plan (SIP) and FIP for Oklahoma which, in addition to actions being
taken by Oklahoma sources to comply with the Mercury and Air Toxics Standards ("MATS")
rule will achieve additional reductions in relevant emissions and impacts over 75%3 of the coal-
fired generation in the state through the installation of controls, conversion to gas, or retirements.
Furthermore, visibility in the Wichita Mountains National Wildlife Refuge ("Wichita
Mountains") is improving4 as visibility monitors in the Wichita Mountains are already close to
meeting the Uniform Rate of Progress goal ("URP") and achieving 2018 Reasonable Progress
Goals ("RPG") developed by Oklahoma in the 2011 SIP5. These actions will continue to reduce
emissions dramatically in the current Regional Haze planning period ending in 2018 and into the
next 10-year planning period ending in 2028.
OG&E asserted that promulgating a requirement for further emission controls related to visibility
in the Wichita Mountains appears unnecessary at this time. Not only are sources in Oklahoma
reducing emissions as a result of a number of federal requirements12, but sources in contributing
states will be as well. It would be prudent for EPA to allow states to assess the efficacy of these
and any additional reductions in a future planning period.
Footnotes:
12 Such requirements include Regional Haze/BART, MATS, CSAPR, ozone NAAQS, and SO2 NAAQS, among
others.
Footnotes:
2 79 Fed. Reg. 74818 (December 16, 2014) "Approval and Promulgation of Implementation Plans; Texas and
Oklahoma; Regional Haze State Implementation Plans; Interstate Transport State Implementation Plan to Address
Pollution Affecting Visibility and Regional Haze; Federal Implementation Plan for Regional Haze and Interstate
Transport of Pollution Affecting Visibility"
3 EPA Air Markets Program 2014 available at http://ampd.epa.gov/ampd
4 IMPROVE Summary Data website at:
http://vista.cira.colostate.edu/DataWarehouse/IMPROVE/Data/SummaryData/RHR_2013/SIA_group_means_7_14.
CSV
5 76 Fed. Reg. 16176 (March 22, 2011)
Response: We agree that further emission reductions are not needed from Oklahoma sources
during this planning period, and we appreciate OG&E's supportive comments for this aspect of
our final rule.
194
-------
Comment: EPA's proposed FIP requirement for some Texas EGUs to install new scrubbers
is inappropriate because those scrubbers could not be installed until after the end of the
2008-2018 period covered by the FIP [AECT (0074) p. 9]
AECT stated that EPA's Proposal would require some Texas EGUs to install new scrubbers, but
EPA has concluded that those scrubbers could not be installed until sometime after 2018.30 In
light of that, the requirement to add those new scrubbers cannot reasonably be part of EPA's
proposed FIP for Texas since the scopes of the Texas Regional Haze SIP and the proposed EPA
FIP are limited to the 2008-2018 planning period. In light of that, and of the "time necessary for
compliance" factor, it is inappropriate for EPA to require that any Texas EGU install a new
scrubber as part of the proposed FIP.
Footnotes:
30 79 Fed. Reg. at 74874 ("typical 502 scrubber installations can take up to five years to plan, construct and bring to
operational readiness", which will be much later than 2018)
Response: We disagree with the comment for the reasons discussed in our responses to other
comments, above in this section.
Comment: The Timeline for Compliance Should Be Shorter Than EPA Proposed.
[Earthjustice (0067) p. 39; Stamper (0068) p. 5]
Stamper noted that the first factor in establishing controls measures and reasonable progress
goals for a Class I area is the time necessary for compliance. EPA has indicated that, due to
delays in processing the Texas and Oklahoma regional haze SIPs, EPA "cannot assume that the
SO2 controls [EPA is] proposing will be installed and operational within this planning period,
which ends in 2018." FIP TSD at 7. Thus, EPA's proposed reasonable progress goals for
Oklahoma and Texas only accounted for the proposed scrubber upgrades to be completed by
2018. Id. EPA assumes that scrubber retrofits could not be accomplished by the end of 2018, or
within the next approximately 3.75 years. However, EPA provided no support for this
assumption.
Earthjustice et al., and Stamper stated that EPA should require implementation of all of its
proposed reasonable progress requirements by the end of 2018, and EPA should revise the RPGs
of each affected Class I area to account for these control measures being operational by the end
of the first planning period in 2018. Earthjustice et al., stated that the EPA assumes that scrubber
retrofits could not be accomplished by the end of 2018, or within the next approximately 3.75
years, but provided no support for this assumption in the proposed rule. Earthjustice et al., and
Stamper noted that the EPA's statements in the proposal conflict with its statements during the
adoption of the Mercury and Air Toxics Standards ("MATS") that EGUs could install required
controls, including scrubbers, within 3 years. Specifically, EPA stated in 2011 that "[ujnits that
choose to install dry or wet scrubbing technology should be able to do so within the compliance
schedule required by the [Clean Air Act] as this technology can be installed within the 3-year
window." 76 Fed. Reg. 24976, 25054 (May 3, 2011).
195
-------
In support of this claim, Stamper noted that the EPA references a letter to Senator Carper dated
November 3, 2010, in which David Foerter, executive director of the Institute of Clean Air
Companies (ICAC), stated that wet scrubbers could be installed in 36 months, dry scrubbing
technology could be installed in 24 months, and dry sorbent injection could be installed in 12
months. Id., fn 172. ICAC's claims were based on 7 years of pollution control installation at
coal-fired EGUs under the Clean Air Interstate Rule (CAIR) and under the NOx SIP Call. The
ICAC letter states that, between 2008 to 2010, flue gas desulfurization (FGD) controls were
installed at numerous EGUs with combined capacity of 60 gigawatts (GW) while, concurrently,
selective catalytic reduction was installed at roughly 20 GW of EGUs.4 During that timeframe of
significant pollution control installation, there were no labor shortages.5
Stamper noted that in analyses conducted for the MATS rule, EPA similarly found that there
were significant FGD and SCR retrofits in recent years, and stated:
These data depict a recent ramp-up in the [Air Pollution Control] deployment
capabilities of all industry participants, including plant owners, the [Air Pollution
Control] supply chain, and state and local permitting agencies. These expanded
[Air Pollution Control] capabilities are still active today....6
Stamper noted that the compliance deadline for the MATS rule is April 2015, with the possibility
of a one-year extension until April 2016. Because the first haze planning period ends in 2018,
well after April 2016, there should not be any labor shortages for installing SO2 controls to meet
EPA's proposed reasonable progress requirements at the Texas EGUs.
Earthjustice et al., and Stamper noted that the EPA's statements in the MATS rulemaking are
proving to be true, as many scrubbers are being installed to meet MATS within three years to no
longer than four years, as shown in Table 2 of the attached Stamper Report (0068)).
Examples of FGD System Installation Timeframes
(provided by Stamper as Table 2 to 0068)
State
Facility
Unit
Time to Install FGD
MI
Dan E Karn
1 and 2
Contract for design and supply for dry scrubbers was
issued in August 2011.7 According to CAMD, dry lime
scrubber began operation at Unit 1 on June 6, 2014. The
scrubber on Unit 2 will be operational by April 2015.8
TN
Gallatin
1,2, 3,4
FGD design for all four units began in September 2011.
The FGD at Unit 4 was expected to be in operation by
April 2015, Unit 3 by June 2015, Unit 1 by November
2015, and Unit 2 by January 2016.9
PA
Homer City
1 and 2
Construction of FGDs began in 2012 and final tie-in to be
completed by end of third quarter of 2015.10
MI
JH Campbell
2,3
Engineering for the Unit 2 FGD began in late 2012 and
the FGD is expected to be installed and operational by
early 2016.11
KS
La Cygne
1,2
Contract for design and supply of wet FGD systems issued
in December 2011.12 Installation of wet FGD systems to
be completed by June 1, 2015.13
196
-------
IN
Michigan City
12
Planning for the dry FGDs began in 2011 with final
operation scheduled for 1st quarter 2016 for Unit 12.14
IN
RM Schahfer
14, 15
Co-located with the Michigan City Plant, FGD systems
were installed and became operational at Unit 14 on
November 1, 2013 and at Unit 15 on October 26, 2014
according to CAMD.15
PA
Homer City
1 and 2
Construction of FGDs began in 2012 and final tie-in to be
completed by end of third quarter of 2015.10
MI
IH Campbell
2,3
Engineering for the Unit 2 FGD began in late 2012 and
the FGD is expected to be installed and operational by
early 2016.11
KS
La Cygne
1,2
Contract for design and supply of wet FGD systems issued
in December 2011.12 Installation of wet FGD systems to
be completed by lune 1, 2015.13
IN
Michigan City
12
Planning for the dry FGDs began in 2011 with final
operation scheduled for 1st quarter 2016 for Unit 12.14
Stamper noted that for those plants that will have multiple units at which scrubbers are to be
installed under EPA's proposed rule, those plants will benefit from coordinated and shared
engineering design and FGD fabrication, economies of scale benefits with FGD suppliers, and
more consistent staffing levels with on-site contractors for concurrent FGD installations.16 This
is demonstrated in the table above. For example, the Gallatin plant will have four FGDs installed
within approximately four years from design to operation.
Earthjustice et al., noted that at a public hearing on the proposed rule, Texas argued that EPA
cannot impose new SO2 emission limits that go into effect in 2020, given the end of the interim
planning period in 2018. This argument fails for several reasons.
First, Earthjustice et al., stated that Texas' argument proceeds from a faulty premise. Contrary to
Texas' argument, it is feasible for all of the proposed FIP controls—both the scrubber
installations and the scrubber upgrades—to be completed by the end of 2018. EPA should revise
the FIP to require installation and operation of all proposed controls by the end of the first
planning period.
Second, Earthjustice et al., stated that Texas' argument mischaracterizes the law. The regional
haze regulations do not require controls to be installed by the end of the planning period except
in the case of BART alternatives, 40 C.F.R. § 51.308(e)(2)(iii), and the 5 year requirement for
BART installation. 42 U.S.C. § 7491(g)(4). The statute notes that long-term strategies can be
implemented over a 10- 15 year time frame. Id. § 7491(b)(2)(B).
Third, Earthjustice et al., stated that it would unreasonable for Texas sources to avoid installing
controls because of the time it took to disapprove Texas' plan and develop a federal plan. Texas
could have avoided the need for such a delay if its 2009 plan submittal had complied with the
Clean Air Act. Texas sources would then have had 9 years (from 2009 through 2018) to install
controls before the end of the first planning period—surely, a feasible length of time. Texas
should not benefit from delays that result from Texas' failure to meet its legal obligations.
Footnotes:
197
-------
4 See November 3, 2010 letter from David C. Foerter, ICAC to Senator Carper, at 4 (Ex. 1).
5 Id.
6 See U.S. EPA, An Assessment of the Feasibility of Retrofits for the Toxics Rule, March 9, 2011, at 5. (Ex. 2).
7 See August 3, 2011 "B&W gets contract for dry scrubber project at Karn coal plant." (Ex. 4A).
x See December 17, 2014 Extension Request for Consumers Energy Company's D.E. Karn Plant (SRN B2840)
Units 1 & 2 for Compliance with the Mercury and Air Toxics Standard (40 CFR 63 Subpart UUUUU) and the
Michigan Mercury Rule (R336.2501) at 2 (Ex. 4B).
9 See July 9, 2014 TVA - Gallatin Fossil Plant (GAF) - Request for Compliance Extension - Mercury and Air
Toxics (MATS), Enclosure at page 4 (Ex. 5).
10 See November 5, 2013 Request for One-Year Extension of the Compliance Deadline for the Mercury and Air
Toxics Standards and of the Expiration Date of the Plan Approval for the Installation of Flue Gas Desulfurization
Units at 1-2 (Ex. 6).
11 See October 4, 2012 Construction Extension for Consumers Energy Company's JH Campbell Facility Pursuant to
the Mercury and Air Toxics Standard (40 CFR 63 Subpart UUUUU, also known as MATS) as well as the Michigan
Mercury Rule (R336.2501, etseq), Exhibit B, Figures B-lc and B-ld (Ex. 7).
12 See "Hitachi Power Systems America Awarded Contract to Supply Pollution Controls Equipment for KCP&L."
(Ex. 8A).
13 See June 22, 2012 Request for Extension of the Mercury and Air Toxics Standards (MATS) Compliance Deadline
KCP&L La Cygne, Source
ID No. 1070005, at 1 (Ex. 8B).
14 See January 30, 2013 NIPSCO - Michigan City and R.M. Schahfer Generation Stations Request for Extension of
Time to Comply with the Utility MATS NESHAP at 1. (Ex. 11).
15 See EPA's Clean Air Markets Database for RM Schahfer. See also January 30, 2013 NIPSCO - Michigan City
and R.M. Schahfer Generation Stations Request for Extension of Time to Comply with the Utility MATS NESHAP
at 1.
16 See, e.g., testimony of Mr. Chad Teply, PacifiCorp, before the Wyoming Public Service Commission at 8-9,
which outlines of the benefits of installation of multiple SCRs at the Jim Bridger Power Plant. (Ex. 3).
Response: We agree with Commenter's assertion that Regional Haze regulations do not require
controls to be installed by the end of the planning period. Additionally, we agree that, if Texas'
SIP had complied with the requirements of the rule such that we could have approved it in full as
submitted, Texas sources would have ample time to install controls, including the scrubber
retrofits, within the first planning period.
We note that Commenter asserts that the scrubber retrofits can be installed within three years'
time. We agree that in some cases scrubber retrofits can and have been installed in less than
three years. However, we do not have the level of detailed information necessary to make that
evaluation for each facility. Thus, we proposed an installation timeframe that has been used in
BART successfully in the past, which although conservative, ensures that any facility has the
necessary time in which to install its controls.121
11. Cost
Comment: EPA's proposed determination that Texas analyzed the "costs of compliance"
factor incorrectly is not supported by the CAA, EPA's Regional Haze rules, or EPA's
reasonable progress guidance [AECT (0074) p. 2-4]
121 This timeframe may also allow for additional coordination in terms of maintenance, installations, and shutdowns
across multiple units and facilities. This timeframe may also better account for lead-in time to go out for bids and
secure funding and other actions that do not necessarily fall between contract award and completion, a timeframe
that was the seeming focus of many of the commenter's examples.
198
-------
AECT stated that EPA's proposed determination that Texas analyzed the "costs of compliance"
factor incorrectly is not supported by the CAA, EPA's Regional Haze rules, or EPA's guidance.
Texas' analysis of the "costs of compliance" factor, however, is supported by the CAA, EPA's
Regional Haze rules, and EPA's guidance.
According to AECT, EPA proposes to determine that Texas should have evaluated the costs of
compliance factor individually for each of a select subset of Texas electric generating units
("EGUs") that were identified in EPA's screening process, rather than evaluating that factor
collectively for all of those EGUs. That proposed determination is contrary to the CAA and
EPA's own Regional Haze rules, as the Tenth Circuit Court of Appeals recently held.4
Further, that proposed determination is contrary to EPA's own guidance; in fact, EPA's
guidance clearly supports Texas' evaluation of the costs of compliance factor collectively for
all of the identified EGUs. EPA states on page 5-1 of its guidance document entitled
"Guidance for Setting Reasonable Progress Goals under the Regional Haze Program" (June
1, 2007) ("Reasonable Progress Guidance") that each state has the discretion to interpret
the costs of compliance factor "to encompass the cost of compliance for ... source categories
...". Further, EPA states on that same page that in applying the costs of compliance factor,
states may use EPA's BART guidelines,5 which provide that "states have flexibility in
how they calculate costs" of compliance, and may choose to apply the costs of compliance
factor collectively for all of the sources in a source category.6 Moreover, EPA states in other
Regional Haze guidance that "reasonable progress is not required to be demonstrated on a
source-by-source basis".7 In light of the foregoing, AECT requests that EPA defer to Texas'
decision to evaluate the costs of compliance factor collectively for all of the identified EGUs.
In addition, AECT noted that Texas' analysis of the costs of compliance factor, which fully
considered the costs of additional sulfur dioxide ("SO2") and nitrogen oxides ("NOx") emissions
controls at the identified EGUs relative to the predicted visibility improvement due to such
additional controls, is consistent with the CAA, EPA's Regional Haze rules, and EPA's
guidance. Indeed, Texas' costs of compliance factor analysis is more supportable under the
CAA, EPA's Regional Haze rules, and EPA's guidance than is EPA's costs of compliance factor
analysis, which considered the cost of the additional SO2 emissions control per ton of SO2
emissions it would reduce (in $/ton SO2 emissions reduced) for only a handful of sources, which
EPA refers to as the "cost effectiveness" of each such control. Texas' costs of compliance factor
analysis is consistent with, and is more supportable under, the CAA, EPA's Regional Haze rules,
and EPA's guidance because "compliance" for purposes of the reasonable progress requirements
in the CAA and EPA's Regional Haze rules should be viewed in relation to the visibility
improvement that would be predicted to occur due to the additional SO2 and/or NOx emissions
controls, rather than on the reduction in tons of SO2 and/or NOx emissions that would occur due
to those additional emissions controls. On Page 5-2 of its Reasonable Progress Guidance, EPA
states that because different pollutants impair visibility differently, in evaluating the costs of
compliance factor, analyzing the costs of possible additional emissions controls relative to the
visibility improvement that would be predicted to occur due to those controls may be more
meaningful and appropriate than evaluating the cost effectiveness of such emissions controls in
terms of their costs per ton of emissions they would reduce. Based on the foregoing, AECT
requests that EPA defer to Texas' analysis of the costs of compliance factor, which involved
199
-------
consideration of the costs of additional SO2 and NOx emissions controls at the identified EGUs
relative to the predicted visibility improvement due to such additional controls.
Footnotes:
4 WildEarth Guardians v. EPA, 770 F.3d 919, 944 (10th Cir. 2014) ("Neither the Clean Air Act nor the Regional
Haze Rule requires source-specific analysis in the determination of reasonable progress.")
5 70 FR 39104 (July 6, 2005)
6 Id., at 39127
7 EPA's "Additional Regional Haze Questions" {September 27, 2006 Revision), available at
http://tinvurl.com/EPARHquestions
Response: We agree that neither the CAA nor the Regional Haze Rule requires a source-by-
sources analysis, and we did not propose to disapprove Texas' reasonable progress and long-term
strategy analyses on that basis. See our responses to other more detailed comments on this issue.
AECT states that Texas' cost of compliance factor analysis is more supportable under the CAA,
the Regional Haze Rule, and our regional haze guidance than is our cost of compliance factor,
which AECT understands to be our cost effectiveness determinant of $/ton. AECT provides no
documentation for this assertion except for stating that "compliance" for purposes of the
reasonable progress requirements in the CAA and our Regional Haze rule should be viewed in
relation to the visibility improvement that would be predicted to occur due to the additional SO2
and/or NOx emissions controls, rather than on the reduction in tons of SO2 and/or NOx
emissions that would occur due to those additional emissions controls. AECT points to our
Reasonable Progress Guidance as support for its position. First, we disagree with AECT that our
proposed control set was based solely on a $/ton analysis. As we discuss in our Cost and FIP
TSDs, we balanced our use of $/ton against the modeled visibility benefit.122 In fact, the sources
for which we conducted a cost analysis were selected because they were in fact the ones that had
the most visibility impacts at the Wichita Mountains, Big Bend, and the Guadalupe Mountains
Class I areas. Second, we believe AECT's reference to our Reasonable Progress Guidance,
which is reproduced below, is misplaced:
In considering the cost of compliance factor, you should keep in mind that
different pollutants differently impact visibility impairment. For example, on a
ton basis, sulfur dioxide related particles have a greater impact on visibility
impairment than crustal material. Therefore, in assessing additional emissions
reduction strategies for source categories or individual, large scale sources, simple
cost effectiveness estimates based on a dollar-per-ton calculation may not be as
meaningful as a dollar-per-deciview calculation, especially if the strategies reduce
different groups of pollutants.
Because all of our analysis was done on the basis of SO2, this reference to the visibility impact of
multiple pollutants is moot. AECT seems to imply that Texas' approach to balancing cost and
visibility was superior to our own. However, as we discuss in more detail in other responses, we
disapproved Texas's approach in part because it masked the potential benefits of controlling
individual sources.123 Thus, we disagree with AECT that Texas' analysis of the costs of
122 See discussion in our Cost TSD beginning on page 5. See discussion in our FIP TSD, beginning on page 27.
123 79 FR 74838.
200
-------
compliance approach is more supportable under the CAA, the Regional Haze Rule, and our
guidance that our own approach.
Comment: Time needed for scrubber upgrades.
NRG [NRG (0078) p. 13] attached a confidential report by Sargent & Lundy that indicates that
significant modifications would be required to the existing Limestone plant scrubbers to achieve
substantially lower SO2 emission rates. Such modifications would likely include new dampers,
modifications to existing duct work and spray towers, replacement of induced-draft fans, and
reheat system upgrades or new chimneys. Therefore, to accommodate the detailed engineering
analysis, procurement of long lead-time items such as fans, demolition or renovation of existing
equipment followed by installation and commissioning of the new systems, additional time
beyond EPA's proposed 3 year implementation may be required. If EPA requires upgraded SO2
controls at Limestone to address regional haze, NRG requests that the compliance date for be set
42 months following the effective date of the rule such that the outages necessary to complete the
modifications can be scheduled following the 2018 summer peak period.
Commenter 0054-39 (Sierra Club) suggested that the rule should be implemented sooner than the
3 to 5 years that EPA is proposing. The commenter noted that these plants have already had 30
to 40 years of getting off of the hook.
Commenter 0054-54 stated that a compliance time of 3 years for scrubber upgrades is too long
and recommended 2 years to increase efficiency to 99 percent with a limit of 0.06. The
commenter suggested compliance in 3 years (instead of 5 years) for retrofitted scrubbers that are
99 percent efficient.
Response: We respond to comment concerning our proposed emission limits in other comments.
As we discuss in our proposal, we believe that three years is appropriate for these units, as we
based our cost analysis on upgrading the existing wet FGD scrubbers of these units, which we
believe to be less complex and time consuming that the construction of a new scrubber. We
solicited comments on alternative timeframes: from two years up to five years from the effective
date of our final rule. In so doing, we must judge the comments we receive based on the level of
documentation that accompanies them.
We have reviewed the report referenced in the comment. Since the report was claimed as
Confidential Business Information (CBI) under 40 C.F.R.§ 2.203(b), we have responded to the
comments contained within it in a separate document that is not a part of our posted docket but
will be available for review by NRG. As we have noted elsewhere in our response to comments,
we are requiring that NRG upgrade the scrubber systems at the Limestone facility. That
referenced S&L report does not contain any information concerning how long NGR projects it
will take to upgrade the Limestone scrubber systems. Except for this comment, we have not seen
any information from NRG that discusses the need for additional time to complete the scrubber
upgrades. We expect that our final rule will be effective in January, 2016, resulting in a required
operational date for the scrubber upgrades (or an alternative method for achieving the SO2 limits)
201
-------
of approximately January, 2019.124 NRG states that it wishes to complete the required upgrades
following the summer 2018 peak period, but does not provide any reason why the upgrades
could not be completed during one of its other scheduled outings, such as that following the
summer 2017 peak period.
In addition, two commenters objected to the 3 years we proposed for the installation of scrubber
upgrades. These commenters were neither specific as to the facility, nor did they provide any
support for their allegations that the scrubber upgrades could be completed in a shorter period of
time.
As a consequence, because (1) we have received no documentation from the two commenters to
support their allegations that the scrubbers upgrades could be completed in a shorter timeframe,
and (2) we have received no documentation from NRG that the additional time is necessary, or a
reason why NRG could not complete the required upgrades during an earlier outage, we disagree
with these commenters.
Comment: Implementation status for the regional haze rule in Oklahoma
[OG&E (0057) p. 2-3]
OG&E noted, in 2011, EPA acted to partially approve and partially disapprove the Regional
Haze SIP developed by Oklahoma and promulgated a FIP for the portions disapproved6. In that
action, EPA stated that the NOx controls adopted by the state meet the Clean Air Act Best
Available Retrofit Technology ("CAA BART") requirements; the SO2 BART controls proposed
in the FIP, in addition to the state adopted NOx controls, would lead to significant improvement
in visibility and meet the CAA BART requirements; additional NOx controls would not be cost
effective; and additional pollutant controls are not needed to meet the CAA BART
requirements7. OG&E and other source operators in the state have begun to implement the
compliance plans.
According to OG&E, by this summer, low NOx burner technology will have been installed on
the four affected coal-fired units in the OG&E fleet, reducing NOx emissions by over 50% from
the coal units. Low NOx burners will be installed by the April 2017 compliance deadline on the
remaining three SIP-affected units (gas-fueled boiler units). By 2019, OG&E will install dry flue
gas desulfurization ("DFGD") systems on two of the four affected coal units, reducing permitted
SO2 emissions by approximately 95%. OG&E will also convert two coal units to natural gas,
virtually eliminating SO2 emissions from these units after 2018. Emissions reductions of NOx
and SO2 in addition to that from the OG&E system in Oklahoma will occur due to the announced
retirement of two Public Service of Oklahoma ("PSO") coal-fueled units, one in 2016 and one in
2026 as well as from the retirement of one of the Grand River Dam Authority ("GRDA")'s units
at its Choteau facility.
Footnotes:
6 76 Fed. Reg. 81728 (December 28, 2011)
7 Id. at 81754
124 See our final action for more details and specificity.
202
-------
Response: We thank OG&E for its work in reducing regional haze.
Comment: [San Miguel (0060) p. 1-2] San Miguel stated that their principal business is the
production of electric energy in South Central Texas. San Miguel operates only one power
generation facility, which includes one (1) lignite-fired power plant and one (1) lignite mine in
South Central Texas. This lignite-fired power plant has a net capacity of 391 Megawatts and is a
base load unit. The generating unit fires only lignite provided by the lignite mine. This one
lignite-fired unit comprises 100% of SMEC's generating capacity and average yearly output is
2.9 million megawatt hours. San Miguel is a small business as defined by the FERC. Being a
not-for-profit cooperative, San Miguel will be forced to pass along, to its consumer-owners, all
costs of meeting any new requirements that may result from the implementation of the Proposed
FIP.
San Miguel stated that they have a significant interest in the outcome of this rulemaking. San
Miguel's lignite fired electric generating facility is a major source of electrical generation to our
member cooperatives, under long term wholesale power contracts for 100% of the generation of
the San Miguel Generating Station.
Response: We applaud San Miguel for upgrading its scrubber system, and demonstrating that
even considering the extremely high sulfur content of the coal it burns (way in excess of any
other unit in our proposal) that its scrubber is capable of essentially the same efficiency as we
assumed for the other units in our proposal. As we note in our proposal, we believe that based
on the scrubber upgrades it has recently performed and its demonstrated ability to maintain an
emission rate below the value we proposed that it can consistently achieve this emission level we
proposed without further upgrades. We address San Miguel's specific comments on its emission
limit in another comment.
Comment: The Proposed FIP Will Result in Additional Economic Benefits.
[Earthjustice (0067) p. 17]
Earthjustice et al stated that requiring antiquated power plants and other sources to invest in
modern pollution controls is a job-creating mechanism in itself, as each installation creates short-
term construction jobs, as well as permanent operations and management positions.18 As EPA
has explained, installing BART "will require well-paid, skilled labor which can potentially be
drawn from the local area and support local growth." 77 Fed. Reg. 57,864, 57,909 (Sept. 18,
2012) (final Montana regional haze FIP).
Footnote:
18 Ceres, New Jobs - Cleaner Air: Employment Effects Under Planned Changes to the EPA's Air Pollution Rules 1-
3 (2011), available at http://www.ceres.org/resources/reports/new-jobscleaner- air.
Response: We take no position on whether the facilities included in our FIP are antiquated. We
do agree that the installation of controls required by our final decision will require the same
kinds of well-paid, skilled labor which can potentially be drawn from the local area and support
203
-------
local growth, as would the installation of BART controls.
Comment: [Earthjustice (0067) p. 17] Earthjustice et al., stated that the regional haze program
protects national parks and wilderness areas, which are of great natural and cultural value, in
addition to serving as engines for sustainable local growth. A National Park Service study found
that national park visitors contribute approximately $30 billion to local economies and support
300,000 jobs, that every dollar invested in park operations generates about $10 in local
communities, and that every two Park Service jobs yield one job outside the parks.19 Nearly 300
million people visit national parks every year, and communities near national parks enjoy
greater-than-average economic growth due to the economic benefits of park visitors and related
businesses.20 Indeed, national parks attract businesses and individuals to the local area; the
resulting economic growth in areas near national parks is 1 percent per year greater than
statewide rates over the past three decades.21
Earthjustice et al., stated that Texas's two national parks are important components of west
Texas's economy. In 2014, over 314,000 people visited Big Bend and over 166,000 people
visited Guadalupe Mountains.22 Tourism at Big Bend in 2010 supported 372 jobs and resulted in
over $16.6 million in visitor spending.23 Tourism at Guadalupe Mountains that same year
supported 258 jobs and resulted in over $13.3 million in visitor spending.24 Studies show that
national park visitors highly value clean air and prioritize the enjoyment of beautiful scenery
when visiting national parks.25 Moreover, national park visitors readily perceive haze, enjoy their
visit less when haze is bad, and are willing to cut short visits to national parks based on their
perception of air quality.26 A decrease in visits means less time and money spent in Texas's
national parks and surrounding communities.
Footnotes:
19 Daniel J. Stynes, Mich. State Univ., Economic Benefits to Local Communities from National Park Visitation and
Payroll, 2010, at page v (2011), available at http://nature.nps.gov/
socialscience/docs/NPSSystemEstimates2010.pdf; see also NPS, National Park System - Summary: 1990 to 2008
(in 2008, National Park Service units received over 274 million visits, accounting for over $2.5 billion in
expenditures and revenue), available at http://www.census.gov/compendia/statab/2010/tables/10sl215.pdf.
20 See Jared Hardner & Bruce McKenney, Hardner & Gullison, The U.S. National Park System,
An Economic Asset at Risk 5 (2006).
21 Id.
22 Nat'l Park Serv., Annual Recreation Visitation Summary report for 2014, available at
https://irma.nps.gov/Stats/SSRSReports/National%20Reports/Annual%20 Visitation%20Summary%20Report%20(l
979%20-%20Last%20Calendar%20Year).
23 Headwater Economics, Nat'l Park Serv. Units: Economic Impacts of Visitation &
Expenditures, available at http://headwaterseconomics.org/apps-public/nps/impacts/.
24 Id.
25 Abt Assocs. Inc., Out of Sight: The Science and Economics of Visibility Impairment, at ES-7
(2000), available at http://www.abtassociates.com/reports/ES-clear.pdf.
26 Id.
Response: We acknowledge that today's action may have positive economic impacts, as described
by Commenter, although we are not charged to specifically take these benefits into account when
making our determination. Congress recognized the need for improvement in visibility at Class I
areas and set forth requirements in the Clean Air Act that directed the states and EPA to secure
204
-------
that improvement. We agree that our Class I areas and other national parks and wilderness areas
are a source of jobs and contribute to the economies of their respective states.
Comment: [Earthjustice (0067) p. 18] Earthjustice et al., stated that the regional haze program
also provides important environmental benefits. In addition to impairing visibility, NOx, SO2,
and PM pollution harm plants and animals, soil health, and entire ecosystems. NOx and SO2 are
the primary causes of acid rain, which acidifies lakes and streams and can damage certain types
of trees and soils. Acid rain also accelerates the decay of building materials and paints, including
irreplaceable buildings and statues that are part of our nation's cultural heritage.27 In addition,
nitrogen deposition—caused by wet and dry deposition of nitrates derived from NOx
emissions—causes well-known adverse impacts on ecological systems. At times, nitrogen
deposition exceeds "critical loads" beyond the tolerance of various ecosystems.28 NOx is also a
precursor to ozone, and ground-level ozone impacts plants and ecosystems by interfering with
plants' ability to produce and store food, and increasing their susceptibility to disease and
insects.29
Footnotes:
27 EPA, Effects of Acid Rain, available at http://www.epa.gov/acidrain/effects/index.html (last visited Sept. 30,
2013).
2SSee, e.g., William D. Bowman el al., Nitrogen Critical Loads for Alpine Vegetation and Soils in Rocky Mountain
Xational Park. 103 Journal of Envtl. Mgmt. 165-71 (2012); NPS, Nitrogen
Deposition: Issues and Effects in Rocky Mountain National Park (2005), available at
http://www.nps.gov/romo/parkmgmt/upload/romo_n_fact_final.pdf; see also National Park
Service, NPS Critical Loads and Deposition, available at
http://www.nature.nps.gov/air/studies/criticalloads/ecoregions/images/nps_cl_ecoregion_tdeplO
12.pdf.
29 EPA, Ground-level Ozone - Ecosystem Effects, http://www.epa.gov/groundlevelozone/ecosystem.html.
Response: We agree that the regional haze program provides environmental benefits that go
beyond regional haze.
Comment: [San Miguel (0060) p. 2, 5-6] San Miguel encouraged the EPA to withdraw the
Proposed FIP and allow Texas to continue with its implementation of the Texas SIP. If EPA
continues with the Proposed FIP, San Miguel requested the EPA remove San Miguel from the
list of EGUs with source-specific SO2 emissions limit. Further, if EPA maintains San Miguel's
inclusion on that list of EGUs, San Miguel requested that EPA increase the San Miguel emission
limit and increase average period to an annual average to insure the limit and average period
appropriately reflect base load operations and periods when the sulfur in the fuel increases to
elevated levels, as was the case between 2009 and 2011.
[San Miguel (0060) p. 3] San Miguel also stated that Table 1 shows how sulfur and heating
value vary over the past 14 years and how these two values determine the inlet quantity of SO2
into the wet flue gas scrubber system. Table 1 is referenced as Attachment A to comment 0060
which was separately hand-delivered to EPA's Region 6 office as it contains Confidential
Business Information.
[San Miguel (0060) p. 3-4] San Miguel expressed appreciation that the EPA has seen that San
205
-------
Miguel has been proactive in improving its SO2 scrubber system to operate at the highest level.
San Miguel also appreciates EPA's understanding that the lignite burned from the San Miguel
Mine has a naturally occurring and varying sulfur content. EPA has further recognized that San
Miguel is using the best technology available to remove sulfur at the highest level expected.
Thus the EPA has not proposed any further control for San Miguel.
According to San Miguel, the EPA has specifically solicited comments on the proposed SO2
emission limit for San Miguel and the potential need for a higher limit to provide sufficient
operational headroom. In response, San Miguel believes that a higher limit and an annual
emission limit is necessary to provide operational headroom to demonstrate compliance.
San Miguel stated that they are a mine mouth plant. Its only source of fuel is the lignite that is
delivered to the plant directly from the San Miguel Lignite Mine. This fuel varies significantly
in its heating value and sulfur content depending on the lignite seam and mining geographic
location. The lignite delivered is a blend of three seams of lignite from two geographical
separated areas in the mine. Each area and each seam has its own unique qualities - sulfur and
heating value. The mine attempts to provide a fuel that provides the specified heating value for
the lignite designed steam generator. The mining areas have moved approximately 14 miles
since San Miguel started commercial operation in 1982. The sulfur in the lignite has varied in
the mine as mining operations have moved. Table 1 [separately delivered to EPA as CBI
material] shows how sulfur and heating value vary over the past 14 years and how these two
values determine the inlet quantity of SO2 into the wet flue gas scrubber system.
San Miguel noted that the equivalent SO2 is the inlet SO2 loading to the scrubber system. This
value is directly related to the amount of SO2 that can be removed by the scrubber system.
According to San Miguel, Table 2 shows the relationship between actual annual emission rate
and the calculated emission limit based on a 94% removal rate of the average fuel.
Modifications to the scrubber have increased efficiency from 90% in the early 2000's to the
current maximized efficiency of 94%.
In the discussion of long-term strategy for San Miguel the EPA stated: "We believe that based on
the scrubber upgrades it has recently performed and its demonstrated ability to maintain an
emission rate below this value on a monthly basis from December 2013 to June 2014 that it can
consistently achieve this emission level." While it is true that San Miguel did achieve the
emission level discussed in the Proposed FIP, this is not typical of San Miguel's operations and
was due to outstanding events that contributed to achieving that emissions rate. During that time
period: 1) San Miguel was only operating for 3,332 hours out of the possible 5,064 hours; 2) San
Miguel, a base load unit, operated only one out of the seven months as a base load unit during
that time period. These changes in operations were due to economic and other external factors
and are at variance with long-time historic operations so they should be considered an exception,
not the norm of our operations for purposes of assessing long-term compliance feasibility. For
example, during all other months of 2014 San Miguel was operating in a load following mode,
cycling between full load and minimum load.
San Miguel stated that when they return back to operating as a base load unit, the emission limit
206
-------
will be unachievable, especially during the peak summer months. A higher limit in combination
with a limit based on an annual average, rather than the proposed 30 day rolling emission
average, will help to address these concerns.
San Miguel asserted that another reason the limit should be increased is due to variations in the
sulfur content and heat rate of the lignite. As demonstrated in Table 2, when the sulfur content
of the fuel was at its peak from 2009 through 2011, and the scrubber operating at 94% efficiency,
the limit could not be met. This indicates a higher limit would be necessary to demonstrate
compliance.
Table 2. Actual vs. Calculated Emission Rate:
flitjl# 1
v
Actual Months
Intel $02
C11 leu luff? cf
Efftlssioft
Actual Emission
Average Emissions
Year
lii:$/ff»m®tu
Rate Ife/mmBtu
Rate ibs/mmitu
Al". ,« •» 1 v'tnmitt.
2(100
*
7.386
¦§,443
0.68
1 ;
2001
8.2,17
0.494
0.746
11
•1AA**
Ilmi
8.721
0.523
0.731
11
2003
7.483
0,449
0.531
2004
7.623
0.457
0.486
2
2005
7.SS1
0.453
0.733
2006
9.040
0.542
0.657
2007
9.408
0.564
mm
3 1
| 2008
9.983
0.599
0.581
7
| 2009
10,152
0.609
0.628
7
I 2010
10,615
0.638
0.62
8
1 2011
In n
0.625
0.607
6
| 2012
9.616
0.577
0.621
i
| 201$
9.138
0.548
0.581
4
1 2014
8.669
0.520
0.482
Comment: Averaging period [San Miguel (0060) p. 4-5]
San Miguel stated that 2014 was an aberration and not in line with San Miguel's historic
operations. EPA should, therefore, raise the emissions limit to reflect the actual operation of the
San Miguel facility, while also increasing the averaging period to one year, rather than the
currently proposed 30-day rolling average. We understand that EPA has voiced a preference for
shorter averaging periods due to seasonal variations of emissions and visibility impacts.
However, since San Miguel is a base load unit and intends to operate year-round, these concerns
are minimized. That being said, if EPA does intend to proceed with a 30-day rolling SO2
emissions limit, San Miguel's limit should be raised to an even higher level than discussed above.
207
-------
This would account for fuel variability leading to higher than normal emissions over a period of
time greater than 30 day.
Response: For the reasons discussed in our proposal and in our response to comments, we
believe that we must finalize our proposed FIP controls for San Miguel. In our Cost TSD, we
noted that prior to the start of its scrubber upgrade program, the San Miguel FGD system required a
DBA level of approximately 1,400 ppm to achieve 94.75% SO2 removal while firing lignite with an
average sulfur content of 9.6 lb/MMBtu. San Miguel provided a 2013 report that detailed the
scrubber upgrades that it had performed. In that report, its contractor, URS outlined two options
for improving the scrubber efficiency, involving improvements to its scrubber system. Option 1
was designed to achieve the same 94.75% SO2 removal with a higher 10.5 lb/MMBtu coal with a
lower DBA concentration of about 725 ppm. Option 2 was designed to achieve the same 94.75%
SO2 removal with 10.5 lb/MMBtu lignite with an even lower DBA concentration of about 125 ppm.
San Miguel elected to initially install the Option 1 scrubber upgrade in all four of its absorber
modules during its spring 2010 outage, and the Option 2 scrubber upgrade during its spring outage in
2012. San Miguel subsequently further upgraded the scrubber system by making improvements to
the tank agitators in 2011 - 2012, and again replacing its trays with an improved design during the
spring 2014 outage. URS stated that it appears the FGD system is currently operating as intended
both from a chemical and physical design standpoint and that the FGD system was achieving
approximately 94% SO2 removal efficiency at absorber DBA concentrations of about 400 ppm.
San Miguel requests that should we finalize our FIP, we increase its emission limit and increase its
averaging period to an annual average to insure the limit and average period appropriately reflect
base load operations.
We proposed that San Miguel's SO2 emissions not exceed 0.60 lbs/MMBtu based on a 30 day boiler
operating day average. In our proposed Cost TSD, we presented San Miguel's historical sulfur
content and Btu coal values reported by it. Below we extend that information to 2014,125 and we also
calculate the resulting uncontrolled SO2 emissions, and a 94% control (what we assumed in our
proposal) of the SO2 emissions:
Year
% Sulfur
Btu/lb
Lbs
S02/MMBtu
94%
control
2009
2.68
5,280
10.15
0.609
2010
2.82
5,303
10.64
0.638
2011
2.75
5,280
10.42
0.625
2012
2.49
5,179
9.62
0.577
2013
2.38
5,209
9.14
0.548
2014
2.25
5,210
8.64
0.518
We calculated the uncontrolled SO2 emissions as follows, using year 2014 as an example:
Uncontrolled SO2 Emissions = (2.25 % sulfur X lb/5,210 Btu) X (1,000,000 Btu/MMBtu) X (2
molecules S02/molecule S) X (1.0/100%) = 8.64 lbs/MMBtu
125125 \ye used the data reported by San Miguel to the Energy Information Administration in Form EIA-923,
available here: http://www.eia.gov/electricity/data/eia923/
208
-------
As can be seen from the above table, 94% control of the coal that San Miguel has burned since 2012
would have resulted in an annual average SO2 emission limit below our proposed FIP limit of 0.60
lbs/MMBtu. In addition, we note that the coal sulfur content of the coal burned by San Miguel has
trended steadily downward since 2010, when San Miguel's average annual uncontrolled SO2
emissions was 10.64 lbs/MMBtu. San Miguel states with reference to its Table 2 that 2014 was an
aberration in its operations, but we note that it had only finished upgrading its scrubber system after
the 2014 spring outage, so we view the information prior to this period as not fully reflecting its
scrubber capabilities.
We also examined the monthly SO2 emission data for San Miguel. We update that information
below:
Year
Month
Operating
Time
(hours)
SO2
(tons)
SO2
Emission
Rate
(lbs/MMBtu)
2013
1
744
886.184
0.544
2013
2
671.5
759.849
0.569
2013
3
49.75
74.948
0.710
2013
4
668.5
753.295
0.587
2013
5
344.75
383.038
0.594
2013
6
720
917.028
0.637
2013
7
731
922.075
0.581
2013
8
649
795.605
0.626
2013
9
379
439.158
0.530
2013
10
728.75
967.086
0.552
2013
11
720
1104.451
0.624
2013
12
744
982.405
0.545
2014
1
536
510.134
0.425
2014
2
511
554.017
0.519
2014
3
1
0.047
0.079
2014
4
127
126.051
0.572
2014
5
743.5
925.666
0.580
2014
6
720
811.227
0.521
2014
7
743.25
870.983
0.576
2014
8
744
920.341
0.592
2014
9
672.5
419.734
0.442
2014
10
611.25
483.517
0.494
2014
11
630.25
545.097
0.480
2014
12
698.5
697.135
0.482
2015
1
709.5
407.832
0.330
2015
2
672
617.49
0.486
2015
3
168.75
272.642
0.712
2015
4
366.25
192.262
0.279
209
-------
2015
5
492.25
445.863
0.419
2015
6
702
493.456
0.358
We have typically conditioned our FIPs based on 30 day averaging periods and continue to believe
that a 30 day averaging period is appropriate. Our goal in proposing emission limits for pollution
control devices in regional haze federal plans is that the emission limits be attainable within the
specified averaging periods, considering start-up or shutdown issues. For instance, in our Texas FIP
we proposed that the emission limits employ a 30 Boiler Operating Day (BOD) average, as specified
by the regional haze rule for BART determinations, since our control analysis was similar to what we
would have conducted for BART. A 30 BOD average is inherently easier for a typical EGU to meet,
because it does not over-weigh emission spikes caused by start-up or shutdown events, as a straight
30 calendar day average has the potential to do. Below are the results of a 30 day BOD average for
San Miguel:
Sari Miguel S02 Emissions
We constructed this graph126 by downloading San Miguel's SO2 emission data from our website,127
filtering out any days when the boiler was not operating (BOD SO2), and calculating a running 30
day BOD average (Running 30 BOD SO2 Avg.). We also downloaded San Miguel's monthly coal
sulfur data from EIA128 and calculated San Miguel's average monthly uncontrolled SO2 emission
limit (Avg. Monthly Uncontrolled SO2). The above graph shows that following San Miguel's spring
2014 outage, when it further upgraded its scrubber system, its running 30 BOD SO2 average has been
trending downward, while its coal sulfur content has been slightly trending upward. Focusing on the
128 See the file, "San Miguel Daily Einissions.xlsx" in our final docket.
ir http://ainpd.epa.gov/ampd/
128 http://www.eia.gov/electricity/data/eia923/
210
-------
peaks in San Miguel's running 30 BOD SO2 average that occurred on 12/18/14 and 3/6/15, we note
that San Miguel's running 30 BOD SO2 average was 0.559 lbs/MMBtu and 0.541 lbs/MMBtu,
respectively. During these months, San Miguel's average calculated uncontrolled SO2 was 8.87
lbs/MMBtu and 9.33 lbs/MMBtu, respectively. This results in a control level of approximately 94%.
We reaffirm our proposed conclusion that based on the coal that San Miguel has historically burned
over the last several years, and its demonstrated ability to remove 94% of the sulfur from that coal,
that it should be able to comfortably meet our proposed emission limit of 0.60 lbs/MMBtu based on a
30 BOD average.
San Miguel states that its ability to achieve our proposed FIP emission rate was due to outstanding
events that are not typical of its operations. It states that it was only operating for 3,332 hours out of
the possible 5,064 hours, and that it operated only one out of the seven months as a base load unit
during that time period. San Miguel does not fully explain how it believes these circumstances
should impact its ability to achieve our proposed SO2 emission limit. We assume it has to do with
more frequent startup and shut down emission fluctuations. However, our experience has been that
when an EGU operates as a base load unit, its emissions tend to be the most stable and its emissions
are the lowest due to minimal startup and shut down emission fluctuations. Our above analysis is
largely independent of those issues, because as we explain above, the BOD averaging method only
considers those days when San Miguel's boiler was operating. Thus, emission fluctuations due to
startup and shut down events are minimized. Also, to the minor extent those fluctuations do impact
emissions, we believe it would impair — not aid San Miguel in achieving our proposed SO2
emission limit. As a consequence, we do not believe that San Miguel has demonstrated that
intermittent or load following operation impairs its ability to achieve our proposed SO2 emission
limit.
San Miguel also requests that should we finalize our FIP, we increase its SO2 emission limit in case
it burns coal with higher sulfur levels, similar to that it burned between 2009 and 2011, when its
calculated annual average uncontrolled SO2 emissions reached 10.64 lbs/MMBtu. First, we
realize that San Miguel is a mine mouth facility and does not have rail access to outside sources
of coal. However, we are unaware of another facility in the United States that burns a coal with
a higher sulfur content than San Miguel, which causes San Miguel's visibility impact to remain
significant even when controlled to 94%. We believe this places a special burden on us to ensure
that we appropriately analyze San Miguel under the reasonable progress and long-term strategy
provisions of our FIP.
Second, we note that applying a 94% removal efficiency to a coal with an uncontrolled emission
rate of 10.64 lbs/MMBtu results in a controlled emission rate of 0.59 lbs/MMBtu. Although this
is just under our proposed FIP emission limit of 0.60 lbs/MMBtu, the uncontrolled emission rate
calculation of 10.64 lbs/MMBtu was based on an annual average sulfur content and does not
allow for normal monthly variations in sulfur content. However, we note that one of San
Miguel's stated purposes for upgrading its scrubber system was to reduce its DBA usage and
thus realize a savings in annual operating costs. As we note above, prior to its scrubber upgrade,
San Miguel previously used a DBA level of approximately 1,400 and that URS reports that
following the spring 2014 upgrade, San Miguel's scrubber system was operating as expected and
was achieving approximately 94% SO2 removal efficiency using DBA concentrations of about
400 ppm. Thus, we believe that some spare capacity exists in San Miguel's scrubber system.
211
-------
However, to ensure that San Miguel can meet our proposed FIP emission limit, even if it must
again burn its historically higher sulfur coal, we will offer San Miguel the following option, if it
cannot achieve our proposed emission limit of 0.60 lbs/MMBtu based on a 30 day BOD average:
• Install a CEMS at the inlet of the scrubber system. The 30 BOD SO2 average from the
existing outlet CEMS must read at or below 6.0% (94% control) of a 30 BOD SO2
average from the inlet CEMS. San Miguel must inform us in writing of its decision to
select this option for compliance by no later than their compliance date.
This option is specified in more detail in our final rule,
ll.a. Use of Confidential Business Information
Comment: [Luminant (0061)] The following attachments to the April 2015 Sargent & Lundy
report included in the Luminant comments were submitted separately as CBI:
Attachment 1 - Approach to Estimating O&M Costs
Attachment 2 - Scrubber Upgrade Capital Cost Adjustment and Cost Effectiveness
Response: We have responded to Luminant's CBI comments in a separate attachment that is not
a part of our posted docket, but is available for viewing by Luminant.
Comment: [NRG (0078) p. 12] NRG provided a confidential report by Sargent & Lundy titled,
S&L Comments to EPA Assessment of Limestone FGD Capability (Apr. 15, 2015).
Response: We have responded to NRG's CBI comments in a separate document that is not a part
of our posted docket, but is available for viewing by NRG.
Comment: [TCEQ/PUCT (0056) p. 17] The TCEQ stated that the EPA's cost analysis for the
proposed FIP is not adequate, in particular regarding the FGD scrubber upgrades. The EPA
cannot use the claim of confidential business information to circumvent its obligation to provide
the public with adequate information regarding the economic analysis of its regulatory actions or
to defend its decision to disapprove the Texas 2009 RH SIP.
As part of their comments on the proposed FIP requirements, the TCEQ noted that the EPA cites
the companies' claims of confidential business information to defend its complete lack of any
cost information regarding upgrades to scrubbers and merely claims that all the scrubber
upgrades were less than $600 per ton (79 FR 74877). Confidential business information is not a
justification for failing to provide proper cost impact information of a proposed rule. The EPA
could have provided example cost information for each type of scrubber upgrade considered
without disclosing any specific information claimed confidential by the companies. The EPA
has not even provided a total cost for all the scrubber upgrades. Additionally, while the proposal
preamble and Technical Support Document for the Cost of Controls Calculations for the Texas
Regional Haze Federal Implementation Plan include detailed information on the costs of the
scrubber retrofits, the EPA also did not provide a total cost estimate of the seven EGUs that EPA
212
-------
has proposed standards that would require installation of new FGD scrubbers. The only total
cost estimate provided by the EPA for the proposed FIP is the approximate $2 billion provided
by EPA staff in informal discussions with the TCEQ.
The TCEQ noted that the EPA claims the TCEQ should have considered scrubber upgrades as a
cost-effective control measure in the Texas 2009 RH SIP revision. Yet, even with the proposed
FIP, the EPA has not provided the TCEQ or the public with any information to evaluate the cost-
effectiveness of scrubber upgrades. Neither the TCEQ nor the public is required to accept the
EPA's unsubstantiated claim that the cost-effectiveness of the scrubber upgrades is less than
$600 per ton. The EPA is using the cost -effectiveness of scrubber upgrades as a basis for
disapproving the Texas 2009 RH SIP and must provide adequate information for evaluating the
basis of the EPA's decision. The EPA should provide cost information for all scrubber upgrade
methodologies considered by the agency.
Response: We disagree with the TCEQ that we should have provided more information
concerning the cost of the scrubber upgrades we analyzed. Our scrubber upgrade cost
information was based on information supplied by the affected facilities in response to requests
for information under section 114(a) of the CAA. The affected companies claimed that
information as Confidential Business Information (CBI) under 40 C.F.R.§ 2.203(b) by the
authority of the CAA as amended (42 U. S. C. 7401, 7411, 7412, 7414, 7416, 7601) therefore
we are generally prohibited from making it available for public review. Any excerpts or
summaries must also be treated as being subject to CBI claims and therefore we are prohibited
from it disclosure or release to another party. Accordingly, although this information is being
used to support our decision making, because all of the affected companies claimed that
information as CBI, it cannot be included in our docket for public review and can only be
disclosed by us to the extent permitted by Section 114(c) of CAA and our regulations governing
treatment of CBI as set out at 40 CFR Part 2, Subpart B.
We also disagree with the following statement made by the TCEQ:
The EPA has not even provided a total cost estimate of the seven EGUs that EPA
has proposed standards that would require installation of new FGD scrubbers.
The only total cost estimate provided by the EPA for the proposed FIP is the
approximate $2 billion provided by EPA staff in informal discussions with the
TCEQ.
Our Cost TSD provides much more than just cost summaries for all of our proposed
scrubber retrofits (what the TCEQ refers to as "installation of new FGD scrubbers"). We
have provided detailed information concerning every aspect of the cost analyses of our
proposed scrubber retrofits, including our choice of cost models, all the inputs, results
summaries, detailed discussion of the results, and the accompanying cost spreadsheets.
All of this information and more is present in the docket for this action.
Comment: EPA's RPG analysis impermissibly fails to consider site-specific factors
[CCP (0075) p. 8]
213
-------
Despite an express requirement to consider "costs of compliance, the time necessary for
compliance, the energy and non-air quality environmental impacts of compliance, and the
remaining useful life of any potentially affected sources," EPA failed to consider site-specific
factors applicable to the Coleto Creek Unit 1 and other sources in adopting its one-size-fits all
wet FGD ("WFGD") scrubber control strategy.1 There is no excuse for such oversight because
EPA specifically solicited information from CCP via a CAA Section 114 information request
seeking specific information related to installation of additional controls at Coleto Creek Unit 1
and arbitrarily decided to ignore it. For example, Coleto Creek produced confidential business
information to EPA indicating that capital cost for a WFGD scrubber would greatly exceed
EPA's estimated capital cost for CCP of $262,435,000 as stated in the Cost TSD. Consistent
with EPA's statement at page 12 of its TSD, CCP is limited in what it may publicly state in these
comments without compromising its claim of Confidential Business Information under 40 C.F.R.
Part 2, Subpart B. Therefore, CCP refers EPA to the Phase I Screening Study labeled
COL_CONFIDENTIAL_007212 for information indicating that the capital costs of a WFGD
scrubber is much higher than EPA estimates.2
Footnotes:
1 EPA also fails to consider more general factors applicable to use of WFGD scrubbers. For example, EPA fails to
consider the impact and costs associated with managing and disposing of scrubber and ash residue in light of the
proposed Coal Combustion Residue rule. EPA is required by law to account for such "non-air environmental
impacts." 40 C.F.R. § 50.308(d)(1).
2 EPA does appropriately determine that any implementation of controls stemming from RPGs should allow for an
implementation period of five years for Coleto Creek Unit 1.
Response: We have reviewed the document that CCP has referenced in the above comment.
The only wet FGD cost information we could find in that document was contained in a summary
table that produced a single number for the total capital cost of a wet FGD. This number was not
accompanied by a line-item break down of the costs or any backup whatsoever. Thus, we do not
know what equipment or assumptions went into the generation of that number, or if the costs
contained items that are disallowed by our Control Cost Manual, such as AFUDC or owner's
costs. As a consequence, that cost assertion is not reliable for a regulatory cost determination.
Comment: [Stamper (0068) p. 39] The fact that all 15 of the Texas EGUs with SO2 controls
proposed by EPA have provided detailed cost estimates, emission reductions, and scrubber
upgrade measures in publicly available documents does call into question whether it is justifiable
for the owners/operators of the Texas EGUs evaluated by EPA for scrubber upgrades to withhold
such information from the public as CBI.
Response: We disagree with the comment that we should have provided more information
concerning the cost of the scrubber upgrades we analyzed. Our scrubber upgrade cost
information was based on information supplied by the affected facilities in response to requests
for information under section 114(a) of the CAA. The affected companies claimed that
information as Confidential Business Information (CBI) under 40 C.F.R.§ 2.203(b) by the
authority of the CAA as amended (42 U. S. C. 7401, 7411, 7412, 7414, 7416, 7601) therefore
we are generally prohibited from making it available for public review. Any excerpts or
summaries must also be treated as being subject to CBI claims and therefore we are prohibited
from it disclosure or release to another party. Accordingly, although this information is being
214
-------
used to support our decision making, because all of the affected companies claimed that
information as CBI, it cannot be included in our docket for public review and can only be
disclosed by us to the extent permitted by Section 114(c) of CAA and our regulations governing
treatment of CBI as set out at 40 CFR Part 2, Subpart B.
11.b. General cost comments
Comment: [Luminant (0061) p. 5] Luminant operates nine of the fifteen units in Texas that
EPA proposes to regulate in this rulemaking, and Luminant would bear over half of the more
than $2 billion in costs that EPA seeks to impose—for no perceptible improvement in visibility,
according to EPA's projections.
Response: We believe, for reasons we have outlined in our proposal and elsewhere in our
response to comments, that the controls we proposed under our FIP will result in significant
improvements in visibility at a number of Class I areas. We do not agree that the consideration
of visibility improvement must directly reflect human perception. As we explain in our FIP
TSD, we have linked many aspects of our reasonable progress and long-term strategy analyses to
our BART Guidelines.129 The CAA and the RHR require, as part of each BART analysis,
consideration of "the degree of improvement in visibility which may reasonably be anticipated to
result from the use of such technology." The regulations do not require that the improvement
anticipated to result from a particular technology at a particular source be perceptible by a single
human being in order to be relevant as part of a BART determination. As we explained in the
preamble to the BART Guidelines:
Even though the visibility improvement from an individual source may not be
perceptible, it should still be considered in setting BART because the contribution
to haze may be significant relative to other source contributions in the Class I
area. Thus, we disagree that the degree of improvement should be contingent
upon perceptibility.
Therefore, in our visibility improvement analysis, we have not considered perceptibility as a
threshold criterion for considering improvements in visibility. Rather, we have considered
visibility improvement in a holistic manner, taking into account all reasonably anticipated
improvements in visibility expected to result at all Class I areas, and the fact that in the
aggregate, improvements from controls on multiple sources will contribute to visibility progress.
Visibility impacts below the thresholds of perceptibility cannot be ignored because regional haze
is produced by a multitude of sources and activities which are located across a broad geographic
area.
129 To assist in interpreting these reasonable progress factors, we will rely on our reasonable progress Guidance. 3
Our Reasonable Progress Guidance notes the similarity between some of the reasonable progress factors and the
BART factors contained in Section 51.308(e)(l)((ii)(A), and suggests that the BART Guidelines be consulted
regarding cost, energy and non-air quality environmental impacts, and remaining useful life. We are therefore
relying on our BART Guidelines for assistance in interpreting those reasonable progress factors, as applicable. FIP
TSD page 6.
215
-------
Comment: [NERA (0061) p. 1] In a report prepared for Luminant, NERA provided an analysis
of certain economic aspects of EPA's proposal. The report demonstrates that the additional
controls not required by Texas's SIP, but proposed by EPA in its FIP, make no sense from the
standpoint of rational decision making, and are significantly less cost effective than controls that
EPA has declined to require in other regional haze actions that EPA has finalized for this same
planning period. This set of technical comments complements and supplements the many other
legal and other analytical issues identified in Generation Company LLC's (Luminant's)
comments on the proposed FIP and is submitted in support of Luminant's comments.
According to NERA, in support of its proposal, EPA has developed an arbitrary rationale to
support which emissions sources should be required to take action under the proposed FIP and
which other emissions sources are not required to take action. There are numerous
inconsistencies and flaws in the logic applied in EPA's screening and quantitative analyses.
Ultimately, however, rational and sound decision making must demonstrate that the resulting set
of actions produce reasonable visibility benefits in return for the costs that they will impose on
society. That is the essence of a cost-benefit analysis. EPA has not demonstrated that here.
[NERA (0061) p. 3] NERA stated that this report and its statements are based on EPA's own
cost estimates for the upgrades and retrofits in the proposed FIP. The fact that we use EPA's
cost estimates to demonstrate the lack of cost effectiveness of those proposed controls does not
imply that we are endorsing them. Luminant contends in other parts of its comments that these
estimates are not correct and that EPA understates the costs of its proposal. To the extent that
actual costs are higher than EPA has estimated, our conclusion that the proposed FIP is
exceptionally cost ineffective would be further reinforced.
Response: We disagree that our rationale is arbitrary and that there are numerous inconsistencies
and flaws in logic in our analysis. Our action is based on an appropriate and thorough study of
the cost considerations deemed necessary by the Regional Haze Rule. We respond to these
generalized comments in other our responses to other comments in which these issues are more
specifically discussed.
Comment: [TCEQ/PUCT (0056) p. 15-16] The TCEQ stated that that its analysis of potential
additional controls is adequate and approvable. It stated that the EPA's proposed finding that a
specific type of unit-by-unit cost and effectiveness analysis was necessary to have an approvable
long-term strategy and an approvable consultation with Oklahoma contradicts the EPA's own
June 1, 2007 Guidance for Setting Reasonable Progress Goals Under the Regional Haze
Program. The EPA's methodology of evaluating possible additional controls on existing EGUs
is not required by the RHR or by the guidance in place at the time Texas prepared its 2009 RH
SIP.
The TCEQ noted that the EPA's own guidance, Chapter 4: Identify Control Measures for
Contributing Source Categories for the First Planning Period, page 4-2, states:
The Regional Haze Rule gives States wide latitude to determine additional control
216
-------
requirements, and there are many ways to approach identifying additional control
measures; however, you must at a minimum, consider the four statutory factors.
The TCEQ stated that it prepared its analysis of the cost and effectiveness of additional controls
by selecting sources and controls that met a $2,700 per ton threshold. This threshold amount was
used in CAIR, as well as used by the EPA in preparing its BART rules and guidance.
The TCEQ noted that the control package Texas considered included SO2 controls at 24 facilities
from 15 sites. The NOx controls included 24 facilities at 15 sites. The calculated haze index
improvements at affected Class I areas from the additional controls ranged from a low of 0.04
deciview at Wheeler Peak in New Mexico to 0.36 deciview at Wichita Mountains in Oklahoma.
The estimated annualized cost for the controls necessary to achieve these calculated benefits was
$324 million. Texas determined that this cost is unreasonable for a visibility improvement that is
below the threshold of perception and below the 0.5 deciview criteria the EPA used for
"contribute to."
The TCEQ stated that on page 4-2, the guidance refers to the EPA's AirControlNET database as
a source of $324 million a year. In its analysis, Texas relied on the cost and effectiveness
information supplied by AirControlNET regarding control techniques for specific source
categories. In preparing the 2009 RH SIP, Texas did use appropriate areas of influence; it did
consider controls from the EPA's AirControlNET database; and it did consider the four statutory
factors in considering whether additional controls were reasonable to implement.
The TCEQ stated that the EPA's preference for a different analysis procedure that reaches a
similar conclusion about cost and effectiveness is not a justifiable basis for the EPA to
disapprove Texas' process in developing its 2009 RH SIP submittal nor is it a justifiable basis for
the EPA to disapprove the Texas-Oklahoma consultation about Texas' impact on visibility
impairment at Wichita Mountains.
Response: We disagree with the TCEQ that its analysis of potential controls was adequate for
the reasons discussed in our proposal and TSDs. We also disagree, as discussed in depth
elsewhere, with the TCEQ that we required a specific type of analysis or that we disapproved the
SIP based on the supposed use of a source category-based analysis by Texas. We chose our
analysis strategy to remedy specific flaws in the TCEQ's analysis. As we state in our
proposal:130
The TCEQ constructed a large potential control set consisting of a mix of large
and small sources, located at various distances from Class I areas, with a large
geographical distribution. Because of the variation in size, type, and location of
these sources, the potential to impact visibility and potential benefit from controls
at a given Class I area can vary greatly between the identified sources. This
potential control set identified by the TCEQ included controls on some sources
that would likely result in significant visibility benefits, but also included controls
on many sources with much less anticipated visibility benefits. Because it only
estimated the visibility benefit of all the controls together, the TCEQ was not able
130 79 FR 74838.
217
-------
to assess the potential benefit of controlling individual sources with significant,
and potentially cost-effective, visibility benefits. Also, we believe that individual
benefits were masked by the inclusion of those controls with little visibility
benefit that only served to increase the total cost figures. For example, the TCEQ
identified SO2 controls at Big Brown to be approximately $l,500/ton,
significantly less than its $2,700/ton threshold. These controls were estimated to
achieve greater than 40,000 tpy SO2 emission reductions. Despite this evidence in
the record of an identified cost-effective control that results in large emission
reductions, and source apportionment modeling identifying large impacts from
EGU sources in northeast Texas, the TCEQ did not separately evaluate the
visibility benefit from the implementation of this control, or appropriately weigh
the four reasonable progress factors in determining the reasonableness of this
individual control.
Because the TCEQ's analysis masked the potential visibility benefits from controlling specific
sources or groups of sources, our analysis was designed to investigate whether individual source-
specific controls were justified.
The fact that the TCEQ used the cost effectiveness tool we mentioned in our Reasonable
Progress Guidance, AirControlNET, does not mean that its resulting analysis is shielded from
our review. As the Regional Haze Rule states:131
In determining whether the State's goal for visibility improvement provides for
reasonable progress towards natural visibility conditions, the Administrator will
evaluate the demonstrations developed by the State pursuant to paragraphs
(d)(l)(i) and (d)(l)(ii) of this section.
Thus, regardless of the specific tools a state uses in its SIP, we must analyze its demonstration.
We agree with the TCEQ that our Reasonable Progress Guidance, and the Regional Haze Rule
itself, provides states with wide latitude in complying with many aspects of the Regional Haze
Rule. However, that latitude concerns how states comply with certain aspects of the Regional
Haze Rule, not whether states must comply with the Regional Haze Rule. As we discuss in our
proposal, we determined that cost effective controls were available that would provide
significant visibility benefits to the Wichita Mountains Class I areas and other Class I areas that
Texas should have required. Instead, despite the fact that sources within Texas impact the
visibility at the Wichita Mountains in Oklahoma more than do sources in Oklahoma, Texas
required no additional controls other than CAIR/CSAPR and those already on its books. We
conclude that the flaws in Texas's analyses cannot be approved as latitude and/or flexibility in
how a proper analysis might be conducted under the Regional Haze Rule.
We address comments concerning no perceptible visibility benefits and Texas' consideration of
visibility benefits along with the use of a 0.5 dv threshold in the section of this document
concerning cost versus visibility benefits. We address comments on Texas' selection and
application of the $2,700/ton threshold in the consistency section of this document.
131 40 CFR Section 51.308(d)(l)(iii).
218
-------
Comment: Scrubber upgrades on NRG's plants would be unreasonable
[NRG (0078) p. 12]
NRG stated that the EPA has proposed to require scrubber upgrades at two of NRG's assets,
Units 1 and 2 at the Limestone power plant. 79 Fed. Reg. at 74,884. As indicated throughout
these comments, NRG does not agree that additional SO2 controls on these units are warranted to
address regional haze.
NRG stated that the EPA's proposed SO2 emission rate for Limestone would be inappropriate,
because it would not meaningfully improve visibility and relies on an erroneous cost-
effectiveness analysis. NRG requests that EPA withdraw its proposed SO2 emission rate for the
Limestone plant.
In addition, NRG supported the exclusion of the W. A. Parish plant from EPA's proposed SO2
emission limits for similar reasons. NRG disagrees that emission reductions from the Parish
plant would be cost-effective from a visibility improvement perspective.
Response: We disagree with NRG that scrubber upgrades at the Limestone facility are not
warranted. We address NRG's general comments concerning our cost analyses and its emission
limit in our responses to its specific cost comments. Many of these comments were submitted by
NRG as Confidential Business Information. We have responded to NRG's CBI comments in a
separate document that is not a part of our posted docket, but is available for viewing by NRG.
Comment: Texas's Aggregate Analysis Masked Cost-Effective Controls.
[Earthjustice (0067) p.24]
Earthjustice et al., stated that the EPA also properly rejected Texas' reasonable progress control
analysis because Texas compared the costs of controls against benefits in such a way that led the
State to reject controls as too costly even though the costs met the State's own cost threshold.
EPA rightly concluded that Texas's analysis is unreasonable because it led Texas to overlook
reasonable, cost-effective controls.
Earthjustice et al., stated that Texas decided that reasonable progress controls costing $2,700 per
ton or less were cost-effective; it screened controls over that amount out of its analysis. Texas
then analyzed in the aggregate reasonable progress controls costing $2,700 or less by comparing
the total cost of controls at all reasonable progress sources to the total visibility benefits. In
essence, Texas applied two cost-effectiveness tests - one based on costs per ton to screen out
specific controls, and the second based on a cumulative dollars per deciview ($/dv) argument to
eliminate all controls evaluated.
With regard to the second cost-effectiveness test, Earthjustice et al., stated that Texas's approach
guaranteed that the state could overlook controls that would have been reasonable if the State
had conducted a more fine-grained analysis. EPA properly concluded that Texas's analysis was
unreasonable because the methodology masked the availability of more effective, lower cost
219
-------
controls for specific sources by lumping them together with less effective, higher cost controls
Response: We agree that Texas' long-term strategy and reasonable progress demonstration was
flawed.
Comment: EPA's Conservative Assumptions Underestimate the Cost-Effectiveness of the
Proposed Controls. [Earthjustice (0067) p.36, Stamper (0068) p.4]
Earthjustice et al., stated that the attached Stamper Report (0068) confirms EPA's conclusion
that the controls proposed in the FIP are cost-effective. Ms. Stamper46 conducted her own
analyses of the proposed controls and concludes that the proposed controls are even more cost-
effective than EPA considered, and most of the proposed controls even meet the very low
$2,700/ton cost-effectiveness threshold that Texas adopted.47 The Stamper Report confirms
EPA's conclusion that the controls proposed in the FIP are cost-effective.
Stamper stated that the EPA has proposed to find that SO2 controls on 15 EGUs are justified to
achieve reasonable progress towards the national goal of eliminating anthropogenic visibility
impairment in Class I areas. The Stamper report (0068) evaluates EPA's four-factor analyses of
its proposed reasonable progress controls, and provides additional support for EPA's finding that
these controls are cost effective. Further, this report shows that SO2 control measures on 5 other
EGUs evaluated by EPA are cost effective and should be part of EPA's FIP, and that controls or
emission limitations on additional units would be meaningful and should have been evaluated.
Footnotes:
46 Victoria R. Stamper is an independent air quality consultant and engineer with extensive experience spanning
government and the private sector. Ms. Stamper's experience includes ten years working in EPA's Region VIII
NSR Program and significant work on regional haze and Class I air quality matters, including work on permit and
plan review and analysis. Ms. Stamper's Curriculum Vitae is included as Attachment A to her expert report.
47 To be clear, Texas's cost-effectiveness threshold is not adequately supported in the record and is unreasonable.
We reference Texas's threshold solely to point out that Texas's methodology caused the State to overlook controls
that meet Texas's own cost-effectiveness threshold.
Response: In performing our cost analyses, we attempted to choose conservative input values
where possible. We agree that our cost analyses are conservative and cost effective. We
disagree that we should have proposed controls for the 5 other EGUs mentioned in the Stamper
report. We address Earthjustice, et al.'s general comments concerning control measures on
additional EGUs in our responses to specific comments on each of these EGUs.
Comment: Cost Analyses Using More Reasonable Assumptions [Earthjustice (0067) p. 38;
Stamper p. 37,43]
Earthjustice et al., noted that Ms. Stamper revised EPA's analyses to be based on a 5-year
annual average emissions baseline, 5-year annual average SChrate in lb/MMBtu, and 5-year
average gross heat rate and MW-hrs generated, based on actual operating data from 2009 to
2013. She also presents a 4th control option for the EGUs: a NID™ circulating dry scrubber,
based on the SDA IPM cost module but assuming SO2 control efficiency of 98% or an SO2 limit
220
-------
of 0.04 lb/MMBtu, whichever is more stringent.
As shown in the FIP TSD, EPA's calculated costs of scrubber retrofits range from $l,255/ton to
$3,500/ton of SO2 removed. Earthjustice et al., stated that EPA's costs for SO2 controls represent
a worst case cost estimate, since EPA's costs are based on highest uncontrolled sulfur content
and maximum MW-hours generated per year. Ms. Stamper's revised cost analyses for scrubber
retrofits, which follow a methodology consistent with prior EPA analyses, are all lower than
EPA's cost estimates stated in the FIP TSD, and range from $l,103/ton to $3,459/ton. All of
these costs are reasonable. Other similar sources have had to bear similar costs for pollution
control to address regional haze as described in the Stamper Report.
Stamper stated that EPA's proposed scrubber upgrades are cost effective. Stamper noted that the
EPA provided extensive information on the various types of scrubber upgrades in its Cost TSD,
and provided justification that such upgrades were very cost effective. Cost TSD at 31-52.
However, EPA did not provide the specifics details of the scrubber upgrades at each EGU
because the information collected by EPA on each units' specific scrubber and scrubber
upgrades was withheld as Confidential Business Information (CBI). Cost TSD at 55. EPA also
did not provide any of the cost information for the scrubber upgrades, due to the information
being withheld as CBI. Id. EPA did state that each of the EGUs had engineering firms evaluate
SO2 scrubber upgrade options and costs which EPA was able to review and which, in some
cases, were very detailed. Id. at 53. Thus, clearly the scrubber upgrades and costs were based
on site-specific analyses and data, which should mean that the scrubber upgrade details are
feasible and the costs are realistic.
Because EPA did not provide costs for any of the scrubber upgrades, Stamper only estimated the
annualized costs of these upgrades. EPA stated that the cost effectiveness for all of the scrubber
upgrades at the EGUs listed in Table 19 of the Cost TSD were less than $600/ton of SO2
removed. Cost TSD at 55. From that $600 per ton value and the SO2 emission reductions due to
the scrubber upgrades listed in Table 19 of the Cost TSD, one can estimate a range of annualized
costs of these upgrades of $1,050,000 per year up to $12,473,400 per year. It must be noted that
this range could be very inaccurate, since we do not know with any certainty the cost per ton of
SO2 removed for scrubber upgrades at each unit. This just provides a low end and a high end
estimate of the annualized costs for these controls.
Stamper gathered other available data on SO2 scrubber upgrades required to meet BART to
determine if these costs seemed reasonable. Stamper attachment 48 is a compilation of that data,
and the information is summarized in the table below.
221
-------
T»W» 11. SwMwuy of Scrubber Upgrades and Cost Effectiveness Reqaired to Meet BART
ana utner Ktgieiiii nait rrovmoiis.
Plant
Unit
Description of Scrubber
Upgrades
Jt nmiiltTA/)
C os is in litlS
Cost
Effecrh hi*--
Jtm Bridger
1
Existing Wet Scrubbers:
Eliminated bypass, relocated
opacity Diomtof „ silicic lino*
and drains for wet stack
operation, aided new ED
used a refined soda ash
reagent m place of existing
SOdhmt
$2,423,840
$603
Jim Bridge
$2,423,840
$603
Tim Bridger
Jim Bridget
3
4
$2,413,322
$1,175,146
$602
$49?
Coal Creek
I
Existing wet scrubbers:
Eliminate bypass and modify
stack, new iimst eliminator,
liquid distribution ring, mi
fan, PLUS lignite coal drying
$11,973,S31
$57?
Coal Creek
2
$11,973,831
$577
_____
4
Replacing existing wet
scrubber tower thai only
treated 30% of flue gas, with a
new wet scrubber that treats
100% of floe gas
$9,149,759
----
1125*11
1
Existing lime spray dryer:
AM spare atomize aid
increase in reagent.
$124,894
$2,047
Hayden
2
§124,8S4
$3,202
O olstnp
1
Existing wet venturi
scrubbers: Additional
scrubber vessel and add line
injection.
$3,960,664
$883
Colstrip
-j.
$3,960,Si4
$959
m SmSOI Scratober Upgrade Costs Data (Ex. 4% foci
of EttvinxBOBtbtl Quality BART AfficitioiiAitlva'- ix JimBrukfi Pchk Plant fx 4v. D«cea$e! l'l 2007
Coal Owk Station Units I ad 2 Beet Amiable Retrofit Technology AaatvsU (Ex 50); haam 20011.411
Analysts f®f Cbolk Umt 4 (Ex 51); ColoradoDepart** of ?ubl;c jr.d Eir v«amxt I
Retrofit Technology (BART) Aasims of Control Opticas for Public Smim C: ^ EPA's
Cclstnp ttet 2 <0: £mr ,;ol- rad t >:n Summary (Ex SS).
Earthjustice et al., stated that the costs of scrubber upgrades proposed by EPA in the Texas
regional haze FIP are quite reasonable to justify these controls to meet regional haze
requirements, whether based on EPA's conservative cost estimates or the revised cost estimates
provided by Ms. Stamper. Based on the above information (provided as Table 11 of Stamper
(0068)), Earthjustice et al., and Stamper concluded that the cost effectiveness of scrubber
upgrades (including one complete replacement of a wet scrubber) that have been required to
meet BART or other regional haze requirements has ranged from $497 to $3,202 per ton of SO2
removed. A comparison of EPA's cost numbers for the 9 EGUs in Texas to the cost numbers
provided above from the Stamper Report demonstrates that the total annual costs for scrubber
upgrades assumed by EPA are well within the range that has been provided by other facilities.
Indeed, the cost effectiveness of the scrubber upgrades at the Texas EGUs of less than $600/ton
222
-------
of SO2 removed are quite reasonable, given the range of cost effectiveness that has been
considered reasonable to meet BART and reasonable progress requirements. Thus, Earthjustice
et al., and Stamper concluded that the EPA is justified in finding that the cost effectiveness for
the scrubber upgrades at Sandow Unit 4, Martin Lake Units 1-3, Monticello Unit 3, W.A. Parish
8, and Limestone Units 1 and 2 are reasonable, as other EGUs have had to bear similar costs to
meet regional haze requirements as well as requirements under other Clean Air Act programs.
Stamper and Earthjustice et al., stated that all of the SO2 scrubber control retrofits and upgrades
evaluated by EPA have been adequately justified as cost effective methods to reduce SO2
emissions from Texas sources, which EPA has identified as the primary contributor to regional
haze in the Class I areas impacted by Texas sources. FIP TSD at 3. The installation of a NID™
circulating dry scrubber can achieve the same levels of SO2 reduction as a wet scrubber, but with
much lower water use and lower costs. Thus, for those units for which water use is a concern,
such as the Tolk units, a NID system may be the best choice for SO2 control, and it is also a more
cost effective choice for SO2 control. Stamper stated that all of these measures are necessary to
provide reasonable progress towards the national visibility goal of attaining natural background
visibility conditions at the Wichita Mountains, Big Bend, Guadalupe Mountains, Caney Creek,
Salt Creek, and the other Class I areas with visibility impairment causes by Texas sources. Most
of the Class I areas impacted by Texas sources are not on track to attain natural background
visibility conditions by 2064. However, with EPA's proposed FIP along with the SO2 scrubber
retrofits at the W.A. Parish and Welsh units, air quality in these Class I areas will be improved
more quickly.
Stamper stated that not only are all of the scrubber retrofits and upgrades evaluated by EPA cost
effective and justified as reasonable progress controls, but there are additional measures that
EPA should consider to improve the rate of progress towards achieving natural conditions at all
of the Class I area that are impacted by air emissions sources in Texas. This is discussed in the
next section of this report.
Response: We confirm that we intended to construct conservative cost analyses. We agree that
our Cost TSD provided extensive information concerning various types of scrubber upgrades
typically performed and that our resulting scrubber upgrade costs are very cost effective. We
also agree that our scrubber upgrade cost analyses were based on site specific information that
was claimed as CBI by the respective companies that supplied it under our CAA section 114
requests, which prevented us from presenting our scrubber upgrade cost analyses. We believe
that our scrubber upgrades costs (all of which are wet FGD) are in line with the information Ms.
Stamper provides concerning other similar wet FGD scrubber upgrades that have been found by
us to be cost effective.
We also agree that the NID technology the comment cites is capable of a similar level of SO2
control. We take up the issue of whether we should have consider NID technology in another
comment. We address comments on additional measures in separate responses to comments.
Comment: EPA's Proposed Scrubber Upgrade Reasonable Progress Requirements.
[Stamper (0068) p. 33]
223
-------
Stamper stated that the EPA proposed to require upgrades to existing scrubbers at Sandow Unit
4, Martin Lake Units 1, 2, and 3, Monticello Unit 3, W.A. Parish WAP8, Limestone Units 1 and
2, and San Miguel. 79 Fed. Reg. 74884 (December 16, 2014). All of the existing scrubbers at
these units are wet scrubbers. Cost TSD at 25. With respect to the San Miguel facility, its
existing wet scrubber was recently upgraded, and EPA's proposed FIP simply imposes an SO2
emission limit reflective of those scrubber upgrades. 79 Fed. Reg. 74822-3, 74877 (December
16, 2014); FIP Cost TSD at 56-61.
Stamper stated that the EPA provided a comprehensive list of the types of upgrades to wet
scrubbers that can be done to improve SO2 removal efficiency and lower SO2 emissions. Those
types of upgrades include:
• Elimination of scrubber bypass
• Upgrades to the scrubber components (installation of liquid distribution rings, perforated
trays, redesign of spray header or nozzle configuration)
• Use of organic acid additives
• Improve/upgrade scrubber auxiliary system equipment.
Cost TSD at 26, referencing EPA's BART Guidelines.
Stamper agreed that the scrubber modifications/upgrades that EPA has described in its Cost TSD
have been commonly implemented, generally achieve very high SO2 removal efficiencies, and
are cost effective.
Stamper noted that URS estimated the range of costs for three categories of scrubber upgrades -
minor, moderate, and major - to reduce hazardous air pollutant (HAP) emissions for the EPA's
Mercury and Air Toxics Standards (MATS), since SO2 scrubbers also control certain HAPs.
Their categorization is as follows:134
Minor Upgrades: These consist of moderate changes to some of the internals of a
scrubber module, reusing most of existing process and structural components. The
primary focus is enhancement of gas/liquid contacting within the absorber. Examples of
applicable modifications to the configuration or type of reagent spray headers used (to
improve overall reagent coverage and overlap within the absorber). Typical minor
upgrades have been used to boost FGD performance to 92 - 97% SO2 removal.
Moderate Upgrades: These upgrades typically consist of major overhauls to the internals
of a scrubber module to replace poor performing or failing components. In many cases,
the existing internal process and structural components are removed; the absorber shell
and large equipment components are maintained. Modifications include installation of
new and improved reagent spray header arrays and absorber trays.
Major Upgrades: For some old FGD units, systems operating with lower-efficiency
technologies and units that have been out of commission for extended periods of time,
major upgrades are required to obtain desired levels of performance. Such upgrades
224
-------
typically involve the same activities described for moderate upgrades, but can also
include additional replacement of some large equipment along with modifications or
additions associated with balance-of-plant equipment. The latter can include additions or
modifications to reagent preparation systems (e.g., ball mills), byproduct dewatering
systems, and process slurry recirculation systems. In some cases, enhancements are
made to existing mist eliminators to reduce carry-over of particulate material from the
scrubber. Conversions from one FGD technology to another may include demolition of
old or unneeded process components and/or addition of new ones.
Stamper noted that URS concluded the capital cost ranges for each of these categories of
upgrades are as follows:
• Minor Upgrades: $5 - $10/kW
• Moderate Upgrades: $15 - $25/kW
• Major Upgrades: $50 - $100/kW
Stamper stated that these capital costs for scrubber upgrades are lower than the capital costs of a
new wet FGD or dry FGD system.
Footnotes:
134 URS, Assessment of Technology Options Available to Achieve Reductions of Hazardous Air Pollutants, 4/5/11,
pp. A-5 to A-7 (Ex. 43); Babcock Power Environmental, Wet Flue Gas Desulfurization Scrubber Upgrades, 2009,
(Ex. 44), available atwww.babcockpower.com/pdf/WetFlueGasDesulfurizationUpgradesSS.pdf.
Response: We agree with the commenter that scrubber retrofits are common and generally very
cost effective.
ll.c. Individual Cost Criticisms (AFUDC, lifetime, escalation, etc.)
Comment: EPA's cost control analysis is flawed and arbitrary and capricious
[Luminant (0061) p. 130]
Luminant stated that EPA's so-called cost-effectiveness analysis is flawed in several respects.
Luminant believes our proposal relies on significantly overstated cost-effectiveness values to
justify both the scrubber upgrades and the retrofits it is proposing. Luminant believes our
analysis omits substantial costs associated with these upgrades and retrofits, overstates the SO2
reductions that are achievable (further skewing its cost-effective values), and is internally
inconsistent in the methodology that is applied. Moreover, states Luminant, despite claiming
that visibility is a key factor in its analysis, we fail to consider costs with respect to the visibility
benefits that EPA contends will be achieved. EPA's analysis relies entirely on a cost-per-ton
metric but ignores the more meaningful cost-per-deciview metric. An analysis of the cost-per-
deciview of EPA's proposal shows that the costs EPA would impose here are well out of
proportion to the visibility benefits and well in excess of costs that EPA has concluded elsewhere
are not reasonable to impose for regional haze purposes.
Response: The items that Luminant summarizes in this comment, and the information provided
225
-------
in its attached Sargent and Lundy (S&L) report, regarding our scrubber upgrade costs are
detailed in a separate comment package submitted under the Confidential Business Information
(CBI) provisions of 40 C.F.R.§ 2.203(b). Within those CBI comments, S&L also provides its
own cost analyses for upgrading Luminant's scrubbers. We are prohibited from responding to
many of these summarized scrubber upgrade comments here with any specificity, because doing
so would involve citing and discussing in detail items that Luminant has claimed as CBI.
Accordingly, many of our responses to the scrubber upgrade comments are contained within a
separate document that is not a part of our posted docket, but will be available for review by
Luminant. The responses to comments that do not contain CBI information are contained within
this document.
With regard to our scrubber upgrade cost analysis, we generally disagree with Luminant that our
analysis was flawed. We used Luminant's own information, backed by independent contractors
hired by it, supplied by Luminant in response to our CAA section 114 requests for information.
This included cost estimates from well-known and respected contracting firms with a history of
many scrubber upgrades. In any event, criticisms regarding our use of this information are moot,
because S&L has provided its own cost analysis (under the CBI protections), which it offers as a
replacement to our own cost analyses. We have reviewed the scrubber upgrade cost analyses
performed by S&L and adopted its methodology. However, we noted many errors and
undocumented cost figures in its analyses. We corrected these errors and rejected some of the
undocumented assertions and/or costs in S&L's cost analyses. Nevertheless, in order to produce
a conservative scrubber upgrade cost analysis and set many of the issues that Luminant raises
aside, we incorporated many of Luminant's cost items. The resulting costs for Luminant's
scrubber upgrades increased slightly, resulting in a range of $368/ton to 910/ton for all of the
scrubber upgrades, well within a range that we believe is cost effective, given the significant
visibility benefits that will result from the installation of those controls.
Luminant did not submit any documentation that caused us to conclude that we overstated the
SO2 reductions that are achievable in either our scrubber upgrade or scrubber retrofit analyses.
Because Luminant did not submit its comments concerning our proposed scrubber retrofit costs
as CBI, we address them within this document as they arise.
Regarding Luminant's assertions about the cost per deciview metric, we disagree that it is a more
meaningful metric. As we note in our Oklahoma FIP:132
[T]he BART Guidelines require that cost effectiveness be calculated in terms of
annualized dollars per ton of pollutant removed, or $/ton.133 OG&E provided a
$/deciview analysis for its units and comparable BART determination performed
by us. In our analysis for our BART FIP for OG&E and AEP/PSO, we did not
evaluate $/deciview. We explain that the BART Guidelines list the $/deciview
metric as an optional cost effectiveness measure that can be employed along with
132 Response to Technical Comments for Sections E. through H. of the Federal Register Notice for the Oklahoma
Regional Haze and Visibility Transport Federal Implementation Plan, Docket No. EPA-R06-OAR-2010-0190,
12/13/2011, pdf 116.
133 70 FR 39167.
226
-------
the required $/ton metric for use in a BART evaluation. The metric can be useful
in comparing control strategies or as additional information in the BART
determination process; however, due to the complexity of the technical issues
surrounding regional haze, we have never recommended the use of this metric as
a cutpoint in making BART determinations. We note that to use the $/deciview
metric as the main determining factor would most likely require the development
of thresholds of acceptable costs per deciview of improvement for BART
determinations for both single and multiple Class I analyses. We have not
developed such thresholds for use in BART determination made by us. As
OG&E acknowledges, EPA did not use this metric as part of its proposed BART
determinations for either the Four Corners Power Plant FIP in AZ, or the San Juan
Generating Station FIP in NM. Generally speaking, while the metric can be
useful if thoughtfully applied, we view the use of the $/deciview metric as
suggesting a level of precision in the calculation of visibility impacts that is not
justified in many cases. While we did not use a $/deciview metric, we did,
however, consider the visibility benefits and costs of control together, as noted
above by weighing the costs in light of the predicted visibility improvement.
Our decision was reviewed and upheld in Oklahoma v. EPA, 723 F.3d 1201 by the Tenth Circuit
which ruled:
Oklahoma first suggests EPA should not have rejected the visibility analysis it
conducted in the SIP, which used the dollar-per-deciview method. This argument
is misguided. The EPA rejected the SIP because of the flawed cost estimates.
When promulgating its own implementation plan, it did not need to use the same
metric as Oklahoma. The guidelines merely permit the BART-determining
authority to use dollar per deciview as an optional method of evaluating cost
effectiveness. See 40 C.F.R. pt. 51 app. Y(IV)(E)(1).134
And in the final rule, the EPA explained why it did not use the dollar-per-
deciview metric used by Oklahoma. "Generally speaking, while the metric can be
useful if thoughtfully applied, we view the use of the $/deciview metric as
suggesting a level of precision in the calculation of visibility impacts that is not
justified in many cases." 76 Fed.Reg. at 81,747. The EPA has never mandated the
use of this metric, and has not developed "thresholds of acceptable costs per
deciview improvement." Id. While the federal land managers have developed
thresholds, these thresholds were apparently developed without input from the
EPA and without notice-and-comment review. EPA Br. at 54 n. 13. In light of
this, we do not find it arbitrary or capricious that the EPA chose not to use the
dollar-per-deciview metric in evaluating BART options in creating the FIP. We
134 We note, however, that in both its final rule and in its brief the EPA asserts that the guidelines require the use of
the dollar-per-ton metric in evaluating cost effectiveness. The guidelines themselves are a bit unclear. In the section
on cost effectiveness, the guidelines mention only the dollar-per-ton metric. 40 C.F.R. pt. 51 app. Y(IV)(D)(4)(c).
However, the guidelines later state that in evaluating alternatives, "we recommend you develop a chart (or charts)
displaying for each of the alternatives" that includes, among other factors, the cost of compliance defined as
"compliance — total annualized costs ($), cost effectiveness ($/ton), and incremental cost effectiveness ($/ton),
and/or any other cost-effectiveness measures (such as $/deciview)." Id. app. Y(IV)(E)(1) (emphasis added).
227
-------
therefore also conclude that any argument by the petitioners that the dollar-per-
deciview measurement proves the scrubbers are not cost effective lacks merit.
See Pet. Reply Br. at 16.
We see no reason, despite that fact that the facilities we evaluated in our proposed TX/OK FIPs
were done under the reasonable progress and long-term strategy sections of the Regional Haze
Rule, to deviate from our view of the dollar per deciview metric here. We also note that the use
of the dollar per deciview metric is further complicated in the present case due to our use of
CAMx modeling. As we discuss in our proposal and elsewhere in our response to comments,
there is no way to directly compare the CAMx modeling we used in our proposed TX/OK FIPs
with previous CALPUFF modeling results.135 Consequently, even if we were to use the dollar
per deciview metric in our TX/OK FIPs, we would be unable to effectively compare the results
against other modeling and cost analyses, the vast majority of which employed CALPUFF.
Comment: EPA's analysis overstates the cost effectiveness of the controls it proposes
[Luminant (0061) p. 131]
Luminant stated that EPA contends that the costs of the controls it would require are reasonable
and cost-effective when viewed under a cost-per-ton metric. For the proposed scrubber upgrades
EPA concluded that "in all cases, the cost effectiveness was less than $600/ton" of SO2
removed.798 For the scrubber retrofits at Luminant's units, EPA found that the cost effectiveness
of wet scrubbers at Big Brown Units land 2 would be $l,255/ton and $l,257/ton of SO2
removed with a total capital cost of $515,173,000, and the cost effectiveness of wet scrubbers at
Monticello Units land 2 would be $l,937/ton and $2,170/ton of SO2 removed with a total capital
cost of $504,981,000.™
Luminant asked Sargent & Lundy (S&L) to perform an independent evaluation of EPA's
determination of the cost effectiveness of the scrubber upgrades and scrubber retrofits. To
conduct this evaluation, S&L reviewed EPA's Upgrade Analysis Document, the Scrubber
Upgrades spreadsheet, and the confidential business information provided by us to EPA that
included historic scrubber upgrade studies performed by S&L.800 Based on this review, S&L
concluded that EPA substantially overstates the cost-effectiveness of the scrubber upgrades and
retrofits that are proposed (that is, EPA's estimated $/ton of SO2 removed are too low).801
Luminant goes on to summarize the points made by Sargent and Lundy in its report.
Luminant states that the limits EPA is proposing in its FIP are more stringent than even Best
Available Retrofit Technology ("BART") limits for existing EGUs that EPA has recently
approved.826 There is no basis, and EPA cites none, for imposing more stringent limitations than
the "best available retrofit technology" can achieve in order to achieve the lesser "reasonable
progress" standard.
Thus, Luminant stated that even EPA's simple (and misguided) cost-effectiveness analysis
suffers from multiple flaws and is incomplete. Indeed, EPA concedes that its calculations "may
contain some error," but it arbitrarily proceeds with its proposal nonetheless—claiming that it
considers the scrubber upgrades to be reasonable at any cost.827 This is the epitome of arbitrary
135 See our FIP TSD, beginning on page A-35.
228
-------
and capricious rulemaking, and it disregards one of the statutory reasonable progress factors. As
S&L's analysis shows, EPA's analysis contains significant errors—more error than can be
corrected by just minor adjustments—and thus the analysis cannot support EPA's proposal.
These errors pervade not only EPA's cost-effective analysis (one of the statutory factors EPA
claims to evaluate in reviewing Texas's SIP) but also EPA's choice of emission limits in the
proposed FIP. Thus, neither EPA's proposed disapproval nor its proposed FIP should be
finalized.
Footnotes:
798 Cost TSD at 55.
799 Id. at 24.
800 Sargent & Lundy LLC, Review of EPA's Cost Analysis for Proposed Action on Texas Regional Haze State
Implementation Plan and Proposed Federal Implementation Plan, Report No. SL012741, at 3 (April 20, 2015). The
results of Sargent & Lundy's analysis are summarized in these comments, and the full report is attached and
incorporated by reference into these comments. Portions of S&L's report with detailed cost information are
submitted to EPA as confidential business information in accordance with and under the protections of 40 C.F.R.
Part 2.
801 Id. atES-3.
802 Id.
803 Id. at 14.
804 Id.
805 Id. at 8.
806 Id. at 5-6.
807 Id. at 7.
808 Id. at 7-8.
809 Id. at 9.
810 Id.
811 Id. at 10.
812 Id.
813 Id.
814 Id. at 11.
815 Id.
816 Id. at 11-12.
817 Id. at 13.
818 Id.
819 Id. at 15-21.
820 Id. at 15-17.
821 Id. at 17-18.
822 Id. at 18-21.
823 Id. at 22.
824 Id. at 21.
825 Id. at 22.
826 See, e.g., Alaska: 0.30 lb/MMBtu for Healy Unit #1 (78 Fed. Reg. 10,546, 10,549 (Feb. 14, 2013)); Arizona:
0.23 lb/MMBtu for the Sundt Generating Station (79 Fed. Reg. 9318, 9325 (Feb. 18, 2014)), 0.15 lb/MMBtu for
Cholla and Apache, and 0.08 lb/MMBtu for Coronado (77 Fed. Reg. 72,512, 72,515 (Dec. 5, 2012)); Colorado:
0.11 lb/MMBtu for Tri-State Craig, 0.12 lb/MMBtu for Comanche Station, 0.13 lb/MMBtu for Hayden, and 0.13
lb/MMBtu and 0.26 lb/MMBtu for Martin Drake (77 Fed. Reg. 18,052, 18,073 (March 26, 2012)); Kansas: 0.10
lb/MMBtu for La Cygne, and 0.15 lb/MMBtu for Westar Jeffrey (75 Fed. Reg. 80,754, 80,758 (Dec. 27, 2011));
Montana: 0.08 lb/MMbtu for Colstrip Units 1 and 2, and 0.57 for JE Corette (77 Fed. Reg. 57,864, 57,915 (Sept.
18, 2012)); North Dakota: 0.15 lb/MMBtu for Leland Olds Station, Milton R. Young Station, and Coal Creek
Station, 0.24 lb/MMBtu and 0.16 lb/MMBtu for Stanton Station (76 Fed. Reg. 58,570, 58,595 (Sept. 21,2011));
Nevada: 0.15 lb/MMBtu for Reid Gardner (77 Fed. Reg. 17,334, 17,338 (March 26, 2012)); New York: 0.09
lb/MMBtu for Danskammer Generating Station (77 Fed. Reg. 51,915, 51,928 (Aug. 28, 2012)); Oregon: 0.40
lb/MMBtu for PGE Boardman (76 Fed. Reg. 38997, 39,002 (July 2, 2011)); South Dakota: 0.09 lb/MMBtu for Big
229
-------
Stone (77 Fed. Reg. 24,845, 24,848 (April 26, 2012)).
827 79 Fed. Reg. at 74,885.
[Sargent & Lundy (0061) p. 3] S&L reviewed the EPA's Upgrade Analysis Document3,
Scrubber Upgrades spreadsheet4, and documents that were provided by Luminant to EPA as
confidential business information (CBI). Reports prepared by S&L and submitted by Luminant
to EPA as CBI are listed below.
• TXU - Monticello FGD Flow Modeling Study, FGD Upgrade and ID Fan Addition,
Analysis of Proposed Retrofit, S&L Report No. SL008398, December 2, 2004
(LUMINANT_REGHAZ_3-000000334-000000571.pdf)
• TXU Monticello Steam Electric Station Unit 3 100% Scrubbing Study (FGD) Conceptual
Cost Estimate (LUMINANT_BB-000292725.pdf)
• Monticello Unit 3 & Sandow Unit 4 100% Flue Gas Scrubbing Study, S&L Report No.
SL008590, January 20, 2006 (LUMINANT_MO-000035024-000035057.pdf)
• Maximum Scrubbing Project, TXU Electric, Martin Lake Station Units 1-3, S&L Report
No. SL008303, May 26, 2006.
According to S&L, the goal of this review was to analyze EPA's use and interpretation of the
technical information and costs included in these reports, and to identify whether any
conclusions reached by EPA were inconsistent with S&L's experience with scrubber upgrade
projects, in general, and as they apply to Luminant's units in particular.
S&L stated that EPA relied upon costs developed in conceptual studies for which S&L
conducted limited engineering and which necessarily included design assumptions that either no
longer apply or were not confirmed by EPA in its analysis. The cost estimates included in both
the Monticello Unit 3 & Sandow Unit 4 100% Flue Gas Scrubbing Study and the Martin Lake
Station Units 1-3 Maximum Scrubbing Project are consistent with Class 4 estimates as defined
by the Association for the Advancement of Cost Engineering. Because the quantity of
engineering completed to support Class 4 estimates is small, these high-level costs can be
underestimated by as much as 50%.5 In fact, the reports specifically indicate that further analysis
is required to define the costs of scrubber upgrades at each unit.
To improve estimate accuracy of study-phase estimates, S&L frequently solicits budgetary
quotations from vendors to support cost development for specific equipment items, but very few
budgetary quotations were received from vendors with respect to the scrubber upgrade cost
estimates for Luminant's units. The S&L report for the Monticello 3 and Sandow 4 scrubber
upgrade cost estimates does not indicate that any cost items were based upon receiving vendor
quotations.6 The S&L report for Martin Lake Units 1 through 3 indicates that budgetary
quotations were received for CEMS equipment and ID fan costs only.7
S&L stated that another means of improving study-phase estimate accuracy, particularly for
upgrade projects that rely upon the use of existing equipment, is to conduct condition
assessments, or inspections, of existing equipment to assess other upgrades that may be required.
Performing condition assessments was not part of S&L's scope of work with Luminant's
predecessor and information from such inspections is not included in S&L's cost estimates.
230
-------
According to S&L, because the studies were conducted in the 2004 to 2006 timeframe, many of
the assumptions included in the reports are no longer valid. As is discussed in more detail in
Section 2.3, the S&L reports were based on specific design conditions while firing specific
design fuels. In the context of estimating scrubber upgrade performance, typically higher sulfur
design fuels are chosen to ensure outlet emission rates are achievable across a range of fuels
expected to be fired. However, for balance of plant impacts such as pressure drop and the
associated fan modifications and auxiliary power consumption, fuels having lower heating values
have a greater impact because, at the same boiler heat input, higher volumes of flue gas are
generated. The S&L reports do not consider lower heating value fuels that are frequently fired at
the Luminant units; therefore, conclusions regarding fan modifications, auxiliary power
consumption, and electrical system upgrades are likely no longer valid. In addition, as is
discussed later in the report, the absorber trays at Monticello 3 and Sandow 4 were removed to
accommodate larger volumes of flue gas while minimizing the impacts to the ID fans. The study
estimates did not include costs to reinstall the trays or analyze the impacts of the existing ID fans
with new trays installed.
S&L's reports clearly indicate that further definition is required to define the cost of the scrubber
upgrades and state the assumptions upon which the studies and cost estimates are based. For
Monticello 3 and Sandow 4, the report states that the retrofit of scrubber upgrades at either unit
"... will require a more in depth evaluation of the work proposed."8 Additionally, some systems,
such as the makeup water system were not evaluated, "[d]ue to the time constraints and broad
scope of this study."9 Finally, S&L notes that several conclusions are based upon informal
discussions as opposed to detailed engineering investigations. For the Stack option selected by
EPA in its analysis, S&L's Martin Lake report cautions about the ".. .risk associated with it
because it is a concept that has been demonstrated on smaller stacks," but not to the scale
required at Martin Lake.10 In addition, the S&L study included information from other projects
because some original design information was not available for the study.11 Finally, S&L
included a specific "Qualifications" section highlighting that while S&L's analysis was based
upon the information it had received, further work needed to be done to properly define the
scope.12
Additionally, S&L's report includes a construction schedule for the stack option EPA selected
for its analysis. This option uses a temporary stack and ductwork to divert flue gas from the
existing chimney in order to allow the existing chimney to be partially demolished and a new wet
chimney erected in its place. The reason this option was considered to be advantageous at the
time of the analysis was because it took advantage of the existing foundation and some of the
existing chimney structure. Since the temporary stack and ductwork are transferred to each unit
sequentially, the construction schedule in the study shows that a duration of 43 months is
required to complete this approach for all three units.13 Therefore, the scrubber upgrades
proposed by EPA at Martin Lake cannot be implemented before the first Regional Haze Planning
Period ends in 2018.
For all of the above reasons, it is clear that EPA's analysis ignores many of the important
qualifying factors S&L included in its reports. While S&L believes that our conceptual cost
estimates were appropriate for providing indicative costs of scrubber upgrades to Luminant's
predecessor for its intended use at the time, we do not believe they are appropriate for EPA to
231
-------
use in assessing actual scrubber upgrade project costs in 2015.
Footnotes:
3 TX Scrubber Upgrade Analyses-CBI-Luminant(FROMEPA).docx
4 CBI Scrubber Upgrades-Luminant(FROMEPA).xlsx
5 Cost Estimate Classification System - As Applied in Engineering, Procurement, and Construction for the Process
Industries, AACE International Recommended Practice No. 18R-97, CMstensen and Dysert, 2005, page 2.
6 Monticello Unit 3 & Sandow Unit 4 100% Flue Gas Scrubbing Study, S&L Report No. SL008590, Exhibit A.
7 Martin Lake Station Units 1-3 Maximum Scrubbing Project, S&L Report No. SL008303, page 6-5 and Appendix
B.
8 Monticello Unit 3 & Sandow Unit 4 100% Flue Gas Scrubbing Study, S&L Report No. SL008590, pg.l.
9 Id., atpg. 6.
10 Martin Lake Station Units 1-3 Maximum Scrubbing Project, S&L Report No. SL008303, page 1-1.
11 Id., atpg.4-1.
12Id., atpg. 6-5.
13Id., atpg. 9-5
Response:
The items that Luminant summarizes in this comment, and the information provided in its
attached Sargent and Lundy (S&L) report, regarding our scrubber upgrade costs are detailed in a
separate comment package submitted under the Confidential Business Information (CBI)
provisions of 40 C.F.R.§ 2.203(b). Within those CBI comments, S&L also provides its own cost
analyses for upgrading Luminant's scrubbers. We are prohibited from responding to many of
these summarized scrubber upgrade comments here with any specificity, because doing so would
involve citing and discussing in detail items that Luminant has claimed as CBI. Accordingly,
many of our responses to the scrubber upgrade comments are contained within a separate
document that is not a part of our posted docket, but will be available for review by Luminant.
The responses to comments that do not contain CBI information are contained within this
document.
With regard to our scrubber upgrade cost analysis, we generally disagree with Luminant that our
analysis was flawed. We used Luminant's own information, backed by independent contractors
hired by it, supplied by Luminant in response to our CAA section 114 requests for information.
This included cost estimates from well-known and respected contracting firms with a history of
many scrubber upgrades. In any event, criticisms regarding our use of this information are moot,
because S&L has provided its own cost analysis (under the CBI protections), which it offers as a
replacement to our own cost analyses. We have reviewed the scrubber upgrade cost analyses
performed by S&L and adopted its methodology. However, we noted many errors and
undocumented cost figures in its analyses. We corrected these errors and rejected some of the
undocumented assertions and/or costs in S&L's cost analyses. Nevertheless, in order to produce
a conservative scrubber upgrade cost analysis and set many of the issues that Luminant raises
aside, we incorporated many of Luminant's cost items. The resulting costs for Luminant's
scrubber upgrades increased slightly, resulting in a range of $368/ton to 910/ton for all of the
scrubber upgrades, well within a range that we believe is cost effective, given the significant
visibility benefits that will result from the installation of those controls.
We take up the issue of whether requiring controls that will not be operational until after 2018 is
a flaw in our proposed FIP in responding to another comment.
232
-------
Luminant and others state that we concede that our calculations "may contain some error," but
that we arbitrarily proceed with our proposal nonetheless—claiming that it considers the
scrubber upgrades to be reasonable at any cost. Luminant claims that this indicates we are being
arbitrary and capricious, and that we are disregarding one of the statutory reasonable progress
factors. Luminant then makes a general statement that our cost analysis is flawed it prevents us
from finalizing our disapproval of Texas' SIP and our FIP. The full quote from our proposal
regarding our acknowledgement that our control cost analysis may contain some error is as
follows:136
In our Cost TSD, we discuss how we calculated the SO2 removal efficiency of the
units we analyzed for scrubber upgrades. We note that due to a number of factors
we could not accurately quantify, our calculations of scrubber efficiency may
contain some error. Based on the results of our scrubber upgrade cost analysis,
we do not believe that any reasonable error in calculating the true tons of SO2
removed affects our proposed decision to require emission reductions, as all of the
scrubber upgrades we analyzed are cost-effective (low $/ton). In other words,
were we to make reasonable adjustments in the tons removed to account for any
potential error in our scrubber efficiency calculation, we would still propose to
upgrade these SO2 scrubbers. We believe we have demonstrated that upgrading
an underperforming SO2 scrubber is one of the most cost-effective pollution
control upgrades a coal fired power plant can implement to improve the visibility
at Class I areas.
However, our proposed FIP does specify a SO2 emission limit that is based on
95% removal in all cases. This is below the upper end of what an upgraded wet
SO2 scrubber can achieve, which is 98-99%, as we have noted in our Cost TSD.
We believe that a 95% control assumption provides an adequate margin of error
for any of the units for which we have proposed scrubber upgrades, such that they
should be able to comfortably attain the emission limits we have proposed.
However, for the operator of any unit that disagrees with us on this point, we
propose the following:
(1) The affected unit should comment why it believes it cannot attain the SO2
emission limit we have proposed, based on a scrubber upgrade that includes the
kinds of improvements (e.g., elimination of bypass, wet stack conversion,
installation of trays or rings, upgraded spray headers, upgraded ID fans, using all
recycle pumps, etc.) typically included in a scrubber upgrade.
(2) After considering those comments, and responding to all relevant comments in
a final rulemaking action, should we still require a scrubber upgrade in our final
decision making action we will provide the company the following option to seek
a revised emission limit after taking the following steps:
(a) Install a CEMS at the inlet to the scrubber.
136 79 FR 74885.
233
-------
(b) Pre-approval of a scrubber upgrade plan conducted by a third party
engineering firm that considers the kinds of improvements (e.g.,
elimination of bypass, wet stack conversion, installation of trays or rings,
upgraded spray headers, upgraded ID fans, using all recycle pumps, etc.)
typically performed during a scrubber upgrade. The goal of this plan will
be to maximize the unit's overall SO2 removal efficiency.
(c) Installation of the scrubber upgrades.
(d) Pre-approval of a performance testing plan, followed by the
performance testing itself.
(e) A pre-approved schedule for 2.a through 2.d.
(f) Should we determine that a revision of the SO2 emission limit is
appropriate, we will have to propose a modification to our decision
making to do so. It should be noted that any proposal to modify the SO2
emission limit will be based largely on the performance testing and may
result in a proposed increase or decrease of that value.
First, we are unaware of any cost estimate we have seen, including those submitted by
Luminant's contractor, that does not contain some acknowledgment of margin for error, hence
the use of the term, "estimate." Cost estimates all acknowledge this and account for it through
the use of contingencies and other factors. The mere fact that we have acknowledged that and
specifically provided the affected facilities the ability to benefit from a refined approach, when
warranted, does not constitute arbitrary and capricious behavior—in fact just the opposite. In the
above quote, we are acknowledging that our calculations of the efficiencies of those units that
have an underperforming scrubber system are estimations that may contain some error. We
explain that we believe it is relatively minor and inconsequential to our decision making because
even if we adjusted the control efficiency and cost-effectiveness, the costs would still be well
within a range that we find to be very cost-effective. Nevertheless, we offer the affected
facilities an opportunity to address any errors. Our approach is not arbitrary or capricious,
because we are proceeding rationally with available information and are expressly open to
making appropriate refinements in finalizing emission limits for these sources.
We address Luminant's comment alleging our FIP emission limits are more stringent than BART
limits for EGUs that we have recently approved in our consistency section. We address
Luminant's comments regarding our scrubber retrofit costs and those portions of Luminant's
scrubber upgrade cost that we can respond to outside of the CBI claims in separate responses to
those comments in this document.
Comment: Omission of equipment capital costs [Sargent & Lundy (0061) p. 5]
S&L stated that, to estimate the total capital cost of the proposed scrubber upgrades, EPA relied
234
-------
upon cost estimates included in study reports prepared by S&L in 2006. For Monticello Unit 3
and Sandow Unit 4, those reports contained costs to reconfigure the spray headers as well as
replacing the mist eliminators. However, EPA removed these costs from its Monticello Unit 3
and Sandow Unit 4 scrubber upgrade cost estimates. It is possible that EPA excluded those costs
based on EPA's misreading of Luminant's 2013 Use Determination Application to the Texas
Commission on Environmental Quality (TCEQ) which identified that new replacement tower
spray nozzles and mist eliminators had been installed. By removing these line items, EPA
apparently assumed that the new spray nozzles listed in Luminant's Use Determination
Application were capable of achieving 95% SO2 removal while processing 100% of the flue gas
flow through the absorbers, yet none of the documentation or information provided confirms that
assumption.
According to S&L, in point of fact, EPA's assumption was incorrect. Because spray nozzles
inside a wet FGD absorber are exposed to a slurry of liquid and ground solid material, they erode
over time and must be replaced periodically. The spray nozzles listed in the Use Determination
Application were replaced in-kind due to normal maintenance practices to maintain the current
level of performance, but were not designed to achieve 95% removal while treating 100% of the
flue gas, as EPA assumes. The cost item in S&L's report represented the cost to change the
configuration of the nozzles (including additional nozzles arranged in a new pattern), such that
the sprayed liquid slurry would contact more of the gas. This cost item also included the cost to
replace the spray headers, which are piping sections that feed the nozzles and which must be
replaced to accommodate more spray nozzles. These changes would be required to achieve
higher SO2 removal efficiencies while treating 100% of the flue gas flow through the absorbers.
Because the spray headers were not replaced and because the nozzle configurations were not
changed, EPA was in error when it removed this line item from its capital cost estimate.
Response: As we note above, the items that Luminant and S&L summarize in this comment,
and the information provided in its attached Sargent and Lundy (S&L) report that are outlined in
successive comments below regarding our scrubber upgrade costs, are detailed in a separate
comment package submitted under the Confidential Business Information (CBI) provisions of 40
C.F.R.§ 2.203(b). We are unable to respond to many of these summarized scrubber upgrade
comments here with any specificity, because doing so would involve citing and discussing in
detail items that Luminant has claimed as CBI. Accordingly, most of our responses to the
scrubber upgrade comments are contained within a separate document that is not a part of our
posted docket, but will be available for review by Luminant.
With respect to S&L's reference to our removal of the costs for upgraded spray headers for
Monticello 3 and Sandow 4, we did in fact remove these costs for the reasons S&L suggests. We
assumed wrongly that after having identified that its scrubber system could be upgraded cost
effectively, and having performed some of those modifications, Luminant would have simply
installed the new upgraded spray headers and nozzles rather than replace its worn out spray
header and nozzles with the less efficient design. However, we added these cost back into our
updated scrubber upgrade cost analyses and the result was a very minor increase in cost
effectiveness (higher $/ton). This did not affect our conclusion that upgrading the scrubbers for
these units is very cost effective.
235
-------
Comment: EPA's methodology to escalate costs and its escalation time frame
[Sargent & Lundy (0061) p. 6]
Costs included in the S&L reports that EPA used as the basis for its estimates are not in today's
dollars; therefore, S&L agrees with EPA's decision to include escalation in its capital cost
estimates. EPA applied the composite Chemical Engineering Plant Cost Index (CEPCI) to
escalate costs in S&L's reports "from year 2006 to year 2013 dollars."14 This approach does not
accurately account for control technology cost increases for at least two reasons: (1) the CEPCI
index should not be used to escalate costs beyond 5-years; and (2) the escalation period (2006 to
2013) is incorrect and should be corrected to 2005 to 2015.
S&L stated that the CEPCI composite index is updated monthly in Chemical Engineering
Magazine. The composite index is built from four sub-indexes (Equipment, Construction Labor,
Buildings, and Engineering & Supervision), and seven equipment component sub-indexes (e.g.,
process machinery, pipe, valves & fittings, electrical equipment, etc.). Although the CEPCI
indexes are commonly used in industry to escalate project costs on a year-to-year basis, there are
limitations to its use. One of the most important limitations is the general industry standard that
the composite CEPCI index should not be used to escalate costs beyond 5-years.15 Beyond the 5-
year window, cost indexes may not accurately reflect changes in relative weighing-factor of the
underlying costs, such as fabricated equipment prices, labor costs and productivity, and
commodity pricing. Studies have found that beyond a period of 5-years, the differences between
actual prices of equipment and labor and those predicted by a cost index become too great to
provide meaningful budgetary cost estimates.16 This limitation applies to escalations of capital
as well as operating costs.
S&L stated that EPA acknowledged this limitation in its Scrubber Upgrade Analysis, noting that
"[bjecause the 2006 Sargent & Lundy cost estimate lies beyond the five year customary rule of
thumb for escalating costs, we add an additional 10% to this cost...."17 EPA provides no
explanation or basis for this adjustment. Although proper application of the CEPCI index can be
used to escalate costs for periods of up to 5-years, the approach should not be used to escalate
costs beyond the 5-year timeframe. A more accurate approach would be to develop project costs
based on more recent projects, up-to-date commodity pricing, and more recent labor costs.
According to S&L, even assuming the CEPCI composite index can be used to escalate
equipment costs beyond the 5-year limitation, EPA escalated costs "from year 2006 to year 2013
dollars" which does not represent the appropriate amount of escalation. EPA incorrectly
assumed the capital cost estimates included in S&L's 2006 studies represented costs in 2006
dollars. However, the reports were first issued in 2005, and they represent costs in 2005 dollars.
Final versions of the reports were updated in 2006, including updates to the cost estimates;
however, no additional escalation was included in the revised estimates. Therefore, the cost-
dollar basis remained 2005 and was not converted to a 2006 dollar basis. The S&L reports
clearly state that "the estimates are based upon 2005 dollars."18
In addition, S&L noted that EPA escalated capital costs to year 2013 even though its FIP is
proposed to be final in late 2015. Accordingly, the total capital cost developed by EPA should
236
-------
have included 10 years of escalation (2005 to 2015) instead of 7 years of escalation (2006 to
2013). Therefore, even assuming the CEPCI composite index can be used to escalate costs over
a 10-year period, EPA underestimated escalation by using the shorter time period, making the
scrubber upgrade projects appear to be more cost-effective. In order to adjust EPA's estimate to
include sufficient escalation for this analysis, S&L determined the average rate of escalation
from EPA's evaluation, which was estimated to be approximately 1.6% per year and applied that
rate over 10 years.
Footnotes:
14 TX Scrubber Upgrade Analyses-CBI-Luminant, Section 1.8, page 14.
15 See, Vatavuk, W., "Updating the CE Plant Cost Index," Chemical Engineering, January 2002, pg. 62, at pg. 66.
16 Vatavuk, at pg. 66.
17 TX Scrubber Upgrade Analyses-CBI-Luminant, Section 1.8, pg. 14.
18 Monticello Unit 3 & Sandow Unit 4 100% Flue Gas Scrubbing Study, S&L Report No. SL008590, Section 4, pg
15.
Response: As we note above, the items that Luminant and S&L summarize in this comment, and
the information provided in its attached Sargent and Lundy (S&L) report that are outlined in
successive comments below regarding our scrubber upgrade costs, are detailed in a separate
comment package submitted under the Confidential Business Information (CBI) provisions of 40
C.F.R.§ 2.203(b). We are unable to respond to many of these summarized scrubber upgrade
comments here with any specificity, because doing so would involve citing and discussing in
detail items that Luminant has claimed as CBI. Accordingly, most of our responses to the
scrubber upgrade comments are contained within a separate document that is not a part of our
posted docket, but will be available for review by Luminant.
We accept S&L's statement that the costs developed in the 2006 reports it cites were developed
in 2005 and that an additional year of escalation is appropriate. However, we cannot agree with
S&L that we should have carried our escalation forward to 2015. The Control Cost Manual uses
the overnight cost methodology, in which costs are calculated based on current dollars. As we
have noted in our proposal, one of our basic tasks in conducting visibility modeling and control
cost analyses was to conduct a cost versus benefit analysis. This requires that both the visibility
modeling and cost analyses be synchronized. Our visibility and cost analyses were conducted in
2014. Both of these tasks required the use of emission data. We used the most recent emission
data that was available, for both the cost analyses and modeling, which was 2013 data. Were we
to escalate the costs to 2015, it would introduce a mismatch between the costs and the emission
data used to calculate the cost effectiveness ($/ton).
S&L objects to our addition of 10% to our cost escalation to account for the fact that the 2006
S&L cost estimate lies beyond the five year customary rule of thumb for escalating costs. S&L
states we provided no explanation or basis for this adjustment. It states that a more accurate
approach would be to develop project costs based on more recent projects, up-to-date commodity
pricing, and more recent labor costs. As with any cost estimator, we must work with the
information we have available to us, considering its quality, and when necessary employing
conservative assumptions to ensure the result fits the required accuracy. As we have discussed in
our response to other comments, our cost analyses employed many conservative assumptions and
this is no exception. Below is the CEPCI Composite Indices from 2005 to 2013:
237
-------
Percent
Change
over
Composite
Previous
Year
CE Index
Year
2004
444.2
2005
468.2
5.40%
2006
499.6
6.71%
2007
525.4
5.16%
2008
575.4
9.52%
2009
521.9
-9.30%
2010
550.8
5.54%
2011
585.7
6.34%
2012
584.6
-0.19%
2013
567.3
-2.96%
2014
576.1
1.55%
Average % Increase
2.78%
Total Increase 2005 - 2013
23.05%
The CEPCI index is used to escalate the cost of an applicable item by simply multiplying that
cost by the ratio of the CEPCI index of the year in which the cost was calculated by the CEPCI
index of the year in which the escalated is desired. For instance, if an item cost $100.0 in 2005,
it would be expected to cost $123.0 in 2014 ($100 X 576.1/468.2), or an increase of 23%.
However, as we indicated in our scrubber upgrade cost analyses (and as S&L notes here), the
commonly accepted rule of thumb for cost escalation is that costs not be escalated beyond five
years. As can be seen from the above table, the CECPI indices have ranged from an increase of
9.52% to a decrease of 9.30%, with an average annual increase of 2.78%. In fact, the CEPCI
index has actually decreased three times during this period. In light of this, we conclude that our
approach of adding an additional 10% to our escalated cost is reasonable and likely conservative.
Comment: Omitted project costs [Sargent & Lundy (0061) p. 7]
S&L stated that the EPA also omitted from its capital cost estimate certain indirect capital costs
that would be incurred by Luminant as part of the scrubber upgrade projects, including Owner's
Costs and project financing costs (or AFUDC). The term "total capital investment" is defined in
the Control Cost Manual (which EPA claims to follow in its proposal), includes all costs required
to purchase the equipment needed for the control system (purchased equipment costs), the costs
of labor and materials for installing that equipment (direct installation costs), costs for site
preparation and building, working capital, and off-site facilities, as well as indirect installation
costs "such as engineering costs; construction and field expenses (i.e., costs for construction
supervisory personnel, office personnel, rental of temporary offices, etc.); contractor fees (for
construction and engineering firms involved in the project); start-up and performance test costs
(to get the control system running and to verify that it meets performance guarantees); and
contingencies.19
238
-------
S&L stated that owner's Costs and AFUDC are both indirect capital costs that should be
included in a capital cost estimate prepared in accordance with the methodology described in the
Control Cost Manual. Owner's Costs include a variety of non-financial costs incurred by the
owner to support the air pollution control project. Owner's Costs are project-specific, but
generally include costs incurred by the owner to manage the project, hire and retain staff to
support the project, and costs associated with third party assistance associated with project
development and financing. Owner's Costs include, but may not necessarily be limited to:
• site investigations (geotechnical, hydrology, etc.) for project design;
• environmental permitting/approvals;
• insurance during construction;
• site security during construction;
• transmission interconnection (if applicable);
• fuel interconnection (if applicable);
• owner's mobilization costs;
• owner's project management and support staff;
• insurance advisor;
• labor relations consultant;
• tax consultant;
• financial advisor;
• legal advisor; and
• community relations/community outreach program.
S&L stated that owner's Costs are real costs that the owner will incur during the project and are
typically included in cost estimates prepared for large air pollution control retrofit projects. In
fact, EPA's Coal Quality Environmental Cost (CUECost) model includes Owner's Costs (or
"Home Office" costs) in its air pollution control system cost estimating workbook and
interrelated set of spreadsheets. CUECost uses a factor of 10% of the total installed cost to
estimate Engineering and Home Office Costs for limestone forced oxidation and lime spray
dryer control systems.20
Based on S&L's experience on large air pollution control system projects and given the nature of
the upgrades proposed by EPA, it is reasonable to estimate Owner's Costs using a factor of 5%
of the total direct costs, and it was incorrect for EPA to exclude Owner's Costs from the cost
estimate.
S&L stated that AFUDC accounts for the time value of money associated with the distribution of
construction cash flows over the construction period.21 AFUDC can represent a significant cost
on large construction projects with long project durations and can be calculated based on a
typical construction project cash flow and real rate of interest. EPA's CUECost model includes
AFUDC in its calculation of air pollution control technology capital costs using an AFUDC
factor and the total plant cost.22 For FGD upgrades, the project could be spread over a
construction period of 30 months or more, particularly given the time necessary for the
construction of a new wet chimney.
239
-------
S&L stated that AFUDC is a real cost that Luminant would incur to implement the proposed
scrubber system upgrades, and is a cost that is allowed by the constant dollar approach described
in the Control Cost Manual. Excluding AFUDC from the cost estimate underestimates the total
capital investment for capital intensive projects with extended project durations. By excluding
AFUDC from its calculation of total capital costs, EPA failed to follow the methodology
described in the Control Cost Manual and failed to include an important line-item in the capital
cost calculation. AFUDC should be included in the total capital investment for the FGD
upgrades. With a typical construction project cash flow for a new chimney, assuming a real
interest rate of 7%, S&L estimated AFUDC to be 10% of the total project costs, which is
consistent with AFUDC factors used in the CUECost and IPM cost algorithms (which EPA also
claims to be following here).
Footnotes:
19 Control Cost Manual, Section 1, Section 2, page 2-5.
20 Coal Utility Environmental Cost (CUECost) Workbook Development Documentation, Version 5.0, prepared by
William H. Yelverton, U.S. EPA Office of Research and Development, EPA/600/R-09-131, September 2009,
Appendix B, pages 38 and 41.
21 Id., atpg. 33.
22 Id., atpg. 17.
Response: As we have indicated in our Cost TSD, we did not blindly adopt the IPM cost
algorithms but made necessary changes to comply with our Control Cost Manual and other
aspects of a typical regional haze cost analysis.137 Also, as we have noted in a number of our
FIPs, AFUDC and Owner's Costs are not valid costs under our Control Cost Manual
methodology. We invite S&L to examine to examine our response to similar comments we
received in response to those actions.138
Comment: EPA underestimated the annualized capital cost of the proposed scrubber
upgrades - incorrect assumptions for estimating the operating life of the scrubber upgrades
and existing equipment [Sargent & Lundy (0061) p. 9]
In the report prepared for Luminant, S&L noted that, to calculate the cost-effectiveness of the
scrubber upgrades, EPA annualized the capital cost of the upgrades using a 30-year equipment
life23 This approach is inconsistent with the methodology described in the Control Cost Manual,
and does not take into consideration the actual scope of the proposed upgrades. As described in
S&L's 2006 reports, scrubber upgrades generally include modifications or upgrades to the spray
headers and mist eliminators, modifications to the existing duct work, and installation of turning
vanes, expansion joints, and dampers to provide better reactant-flue gas contact. Modifications
do not include rebuilding or replacing the entire reactor shell. Equipment included as part of the
scrubber upgrades (excluding the new stacks) would have a useful life of significantly less than
30-years.
137 Cost TSD, pages 11, 21, and 22.
138 See for instance our "Response to Technical Comments for Sections E. through H. of the Federal Register
Notice for the Oklahoma Regional Haze and Visibility Transport Federal Implementation Plan," Docket No. EPA-
R06-O AR-2010-0190, 12/13/2011.
240
-------
S&L stated that The Control Cost Manual describes a methodology for annualizing the capital
cost of the control system, as well as annualizing the cost of replacement materials.
Replacement materials include components such as nozzles, which last for more than a year but
are consumed by the system.24 The Manual notes that "[t]he useful life of replacement material is
generally less than the useful life of the rest of the control system - typically two to five years.
Consequently, the annualization of the system's replacement materials must be done separately
from the annualization of the control system itself."25 By applying a 30-year equipment life to
the scrubber upgrades, EPA failed to take into consideration the actual scope of the proposed
upgrades and failed to follow the methodology described in the Control Cost Manual.
Furthermore, S&L noted that the Control Cost Manual assumes an equipment life of less than
30-years for other air pollution control equipment and does not specify otherwise for scrubber
upgrades.26 EPA has used a 20-year scrubber useful life in several Regional Haze BART
determinations.27 It is also important to note that the majority of these evaluations were
completed for retrofit FGD systems (i.e., new systems) and not upgrades to existing scrubbers
and related equipment that has already been operating for more than 30 years. Thus, a 20-year
useful life is more than reasonable, and in line with EPA's Control Cost Manual, to use for the
upgrades to existing scrubbers that EPA is proposing.
According to S&L, by utilizing a 30-year useful life to annualize the capital cost of the scrubber
upgrades, EPA is significantly overestimating the capabilities and scope of the proposed
upgrades. These upgrades are intended to improve performance, and the scope of the upgrades
does not include the high cost of repair and replacement work which would be required to
maintain the existing scrubbers to continue operating for a period equivalent to an entirely new
FGD system. EPA's use of a 30-year evaluation period has significantly overestimated the cost-
effectiveness of the FGD upgrades, as the annualized costs do not include the additional
maintenance and capital expenditures that would be required for a 30-year evaluation period.
Footnotes:
23 See, TX Scrubber Upgrade Analyses-CBI-Luminant, Section 1.8, pg. 17.
24 OAQPS Control Cost Manual, Section 1, Chapter 2, pg. 2-33.
25 Id.
26 See, e.g., Control Cost Manual Section 4.2, Chapter 2, pg. 2-50 (20-year equipment life for selective catalytic
reduction control systems); Section 6, Chapter 1, pg., 1-48 (typical 20-year life for a fabric filter baghouse); and
Section 5, Chapter 1, pg. 1-28 (15-year life for wet acid gas scrubber).
27 See, e.g., Best Available Retrofit Technology (BART) Determination, American Electric Power, Northeastern
Power Plant, May 30, 2008. BART Five Factor Analysis, Kansas City Power & Light, La Cygne Generation
Station, August 2007.
Response: We disagree that our use of a 30 year life is inconsistent with our Control Cost
Manual. S&L is incorrect that our Control Cost Manual assumes a 20-year useful life for other
pollution control equipment retrofitted on coal-fired boilers and does not specify otherwise for
scrubber upgrades. We suspect S&L is referring to the Control Cost Manual's treatment of SCR,
in which it does not recommend a lifetime for an SCR, but rather sets out a calculation example
that uses a lifetime of 20 years. We agree that the life of the control has a significant impact on
241
-------
the cost effectiveness. However, as we state in our Oklahoma FIP,139 we noted that scrubber
vendors indicate that the lifetime of a scrubber is equal to the lifetime of the boiler, which might
easily be well over 60 years. We identified specific scrubbers installed between 1975 and 1985
that were still in operation.
S&L states that our Control Cost Manual differentiates between the cost annualization of capital
and replacement materials. We agree, and reproduce the full quote to which S&L refers:
The useful life of replacement materials is generally less than the useful life of the
rest of the control system - typically two to five years. Consequently, the
annualization of the system's replacement materials must be done separately from
the annualization of the control system itself. Furthermore, the annualized cost of
the pollution control system should be performed net of the cost of the
replacement materials needed at the beginning of operations to prevent double
counting. Replacement materials labor will vary, depending upon the amount of
the material, its workability, accessibility of the control device, and other factors.
The cost of replacement materials labor should be included in the cost of the
materials before annualization
Here, our Control Cost Manual refers to typical items that are included in Operations and
Maintenance (O&M) cost items, the costs of which are calculated annually. We agree that
scrubber systems include many O&M costs and we have separately calculated them just as
described above. However, the fact that some parts must be replaced or repaired at normal
service intervals does not mean that the pollution control system as a whole has that equipment
life. Facilities regularly replace and/or repair many of the parts of a scrubber system (e.g.,
nozzles, headers, mist eliminators, pumps, pipes, etc.) at regular intervals, often multiple times
over the life of the scrubber system, including those parts that S&L mentions. The life of the
scrubber system ends when the system can no longer be maintained. As we discuss above,
scrubber vendors indicate that the lifetime of a scrubber is equal to the lifetime of the boiler,
which might easily be well over 60 years.
Additionally, we are unaware of any additional capital and/or O&M costs that would have to be
considered in a 30 year versus 20 year operational life. S&L has recalculated the capital and
O&M costs of several of the scrubber upgrades (which we largely adopt). Thus, it has had ample
opportunity to identify these additional cost items it claims are necessary into the record, but has
failed to do so.
We address S&L's allegation that we have approved of 20 year equipment lives in other actions
in the consistency section of this document.
Because none of the facilities involved have entered into or offered to enter into enforceable
commitments to shut down the applicable units earlier, we will continue to use a 30 year
equipment life for scrubber upgrades, as we believe that is proper.
139 Response to Technical Comments for Sections E. through H. of the Federal Register Notice for the Oklahoma
Regional Haze and Visibility Transport Federal Implementation Plan, Docket No. EPA-R06-OAR-2010-0190,
12/13/2011. See discussion beginning on page 36.
242
-------
Comment: Arbitrary approach to estimating operating and maintenance costs
[Sargent & Lundy (0061) p. 10]
According to S&L, there are several errors in EPA's approach to estimating the operating and
maintenance costs related to the proposed upgraded scrubbers. These errors include:
• Incorrect reference to information from the URS Report upon which EPA relied to
develop annual O&M costs
• Arbitrary escalation of operating costs using indices
• Inaccurate application of URS operating costs
• Use of a report from URS containing insufficient detail
• Incorrect assumptions for developing Martin Lake O&M costs
These errors are discussed in more detail in Attachment 1 (submitted as confidential business
information). For the above reasons, S&L does not believe that the operating costs developed by
EPA in its Upgrade Analysis Document represent the real operating costs that Luminant would
incur to meet the requirements of the proposed FIP. As shown in Attachment 1, EPA
significantly underestimated the real operating costs. In order to provide a more accurate
estimate of the expected operating costs associated with the future operating scenario, S&L
developed utility consumption rates and operating costs using information provided by Luminant
regarding the current operation of the scrubbers and recent commodity pricing. Attachment 1
details how the increases in commodities were developed to estimate O&M costs consistent with
industry practice and contains the estimated increase in O&M costs associated with the FGD
upgrades and treating 100% of the flue gas.
Response: The items that Luminant summarizes in this comment, and the information provided
in its attached Sargent and Lundy (S&L) report, regarding our scrubber upgrade costs are
detailed in a separate comment package submitted under the Confidential Business Information
(CBI) provisions of 40 C.F.R.§ 2.203(b). Within those CBI comments, S&L also provides its
own cost analyses for upgrading Luminant's scrubbers. We are prohibited from responding to
many of these summarized scrubber upgrade comments here with any specificity, because doing
so would involve citing and discussing in detail items that Luminant has claimed as CBI.
Accordingly, many of our responses to the scrubber upgrade comments are contained within a
separate document that is not a part of our posted docket, but will be available for review by
Luminant. The responses to comments that do not contain CBI information are contained within
this document.
With regard to our scrubber upgrade cost analysis, we generally disagree with Luminant that our
analysis was flawed. We used Luminant's own information, backed by independent contractors
hired by it, supplied by Luminant in response to our CAA section 114 requests for information.
This included cost estimates from well-known and respected contracting firms with a history of
many scrubber upgrades. In any event, criticisms regarding our use of this information are moot,
because S&L has provided its own cost analysis (under the CBI protections), which it offers as a
replacement to our own cost analyses. We have reviewed the scrubber upgrade cost analyses
243
-------
performed by S&L and adopted its methodology. However, we noted many errors and
undocumented cost figures in its analyses. We corrected these errors and rejected some of the
undocumented assertions and/or costs in S&L's cost analyses. Nevertheless, in order to produce
a conservative scrubber upgrade cost analysis and set many of the issues that Luminant raises
aside, we incorporated many of Luminant's cost items. The resulting costs for Luminant's
scrubber upgrades increased slightly, resulting in a range of $368/ton to 910/ton for all of the
scrubber upgrades, well within a range that we believe is cost effective, given the significant
visibility benefits that will result from the installation of those controls.
Comment: [CCP (0075) p. 8-9] CCP noted that an example of EPA's failure to consider site-
specific factors is the total disregard of the costs and limited availability of water near Goliad,
Texas. Texas is suffering from one of the worst droughts on record and water conservation is
being encouraged. Yet EPA assumes that sufficient water can be obtained for a WFGD scrubber
and makes no attempt to evaluate specific costs or availability for Coleto Creek Unit 1. This
factor is especially important given that WFGD scrubbers consume tremendous amounts of
water, more than any other control option that CCP evaluated in its confidential Phase I
Screening Study labeled COL_CONFIDENTIAL_007212 (at page 007235).
Response: Please refer to our response to a more detailed comment from Xcel on this issue.
Comment: [CCP (0075) p. 9] CCP noted that the EPA also inconsistently assumes a 30-year
scrubber life for the WFGD scrubber retrofit on Coleto Creek Unit 1 when estimating the cost-
effectiveness of a scrubber, despite the fact that EPA guidance uses a 20-year lifetime in similar
assessments. See Cost Manual, Section 4.2 at 2- 50. This will over-estimate cost-effectiveness.
In fact, CCP estimated a useful life of less than 20 years for WFGD scrubbers, as reflected in
confidential business information provided to EPA pursuant to its Section 114 information
request in a Regulatory Changes Impacts document labeled COL CONFIDENTIAL 001767 (at
page 001778).
Response: We disagree that our use of a 30 year life is inconsistent with our Control Cost
Manual. CCP is referring to the Control Cost Manual's treatment of SCR, in which it does not
recommend a lifetime for an SCR, but rather sets out a calculation example that uses a lifetime of
20 years. We agree that the life of the control has a significant impact on the cost effectiveness.
See our response to a similar comment above in which we reference our Oklahoma FIP, where
we noted that scrubber vendors indicate that the lifetime of a scrubber is equal to the lifetime of
the boiler, which might easily be well over 60 years. We identified specific scrubbers installed
between 1975 and 1985 that are still in operation today. Because CCP has not entered an
enforceable commitment to guarantee a 20 year (or shorter) operational life or provided any basis
to question a longer equipment lifetime, it is rational to assume 30 years in our cost analyses.
Comment: EPA's Proposal Significantly Underestimates the Cost of Installation of
Scrubbers at Tolk. [Xcel Energy (0064) p. 5-6, 29; 0053-24 and 0054-4]
244
-------
[Xcel Energy (0064) p. 6] Xcel Energy stated that the EPA proposes to require Tolk to install
SO2 scrubbers at a cost of approximately $400 million in capital costs alone. These costs,
however, ignore the other important factors required for installing and operating scrubbers. First,
EPA ignored the extremely limited water supplies in the region and the costs that would be
required for Xcel Energy to acquire the necessary water rights for the addition of scrubbers.
EPA also ignored the costs resulting from making its fly ash unmarketable and the resulting cost
of landfilling its ash. EPA also failed to consider the environmental impacts of both further
accelerating the already-rapidly depleting aquifer and the requirement to create new landfills.
Xcel Energy stated that, in evaluating control costs for Tolk, EPA failed to consider site-specific
factors applicable to Tolk, despite an express requirement to do so under the RPG rules. See 40
C.F.R. § 51.308(d)(l)(i)(A) (requiring consideration of "costs of compliance, the time necessary
for compliance, the energy and nonair quality environmental impacts of compliance, and the
remaining useful life of any potentially affected sources"); see also 79 Fed. Reg. at 74,874 n.
292 (noting that Tolk was the "one exception" "merit[ing] special consideration of the energy
and non-air quality environmental impacts of compliance"). This is particularly true regarding
water consumption at Tolk. EPA wrongly assumes that Tolk currently has available
groundwater that would allow it to operate scrubbers. EPA also fails to consider the impact and
costs associated with managing and disposing of scrubber and ash residue in light of Tolk's
current 100% beneficial reuse of ash and the requirements that apply to a new landfill under the
final Coal Combustion Residue ("CCR") rule, 80 Fed. Reg. 21,302 (Apr. 17, 2015).11 Finally,
EPA made an unreasonable assumption regarding the likely and appropriate amortization period
for the scrubbers. All of these omissions result in a significant underestimation of costs.
Commenter's References:
11 At least in part due to these "special" considerations at Tolk, EPA did appropriately determine that any installation
of controls at Tolk stemming from RPGs should allow for an implementation period of five years.
Response: Each of these arguments are detailed in other comments below and we respond to
them there.
Comment: Water availability and cost were not appropriately considered [at Tolk].
[Xcel Energy (0064) p. 29-30]
Xcel Energy noted, in the FIP TSD, EPA recognizes that water scarcity in the region around the
Tolk facility is a serious concern for the implementation of a wet scrubber. FIP TSD, at 8, 30.
However, EPA fails to recognize that the issue is just as problematic for a dry scrubber. The
qualifier "dry" in dry scrubber is the difference in how the control device utilizes the sorbent
material for removal. It does not mean that the control device operates without water. A dry
scrubber still uses water to hydrate the removal media (generally lime) to operate properly. A
wet scrubber uses water sprays and a lime sludge recycle to effectuate removal. Dry scrubbers
may require approximately 50% percent less water to operate than wet scrubbers, but they still
use a significant amount of water.
Xcel Energy noted that the EPA stated in a meeting with Xcel Energy on February 4, 2015, in
the EPA Region 6 office in Dallas, Texas that use of a dry scrubber requires 3% to 5% more
245
-------
water than the same plant without a dry scrubber. Xcel Energy could not find any support for
this statement from EPA but, even if true, it is not true in the case of Tolk. Due to water scarcity
in the region, Xcel Energy has undertaken aggressive measures to minimize and reuse the limited
water that is available from the aquifer. All of Tolk's water comes from the underground
Ogallala aquifer, the same aquifer used by the farmers and communities in the panhandle of
Texas. To conserve available water, Tolk installed additional clarifying capacity to enhance the
current treatment facilities at Tolk, which allows the cooling tower blowdown from its
neighboring Xcel Energy facility to be sent to Tolk, recycled and reused. This saves billions of
gallons of fresh water by maximizing the use of the existing water.
Xcel Energy stated that, in addition, the plant monitors water treatment continuously to
maximize the water quality on the supply side to the cooling cycle so that maximum cycles of
concentration can be utilized to limit the amount of blowdown water on the discharge side of the
cooling cycle. Basically, Tolk "reuses" the water 20 to 25 times before a solubility limit is
reached. To maintain the proper chemistry, some of the cooling water is blowdown (average of
300-350 GPM) to evaporation ponds. This maximum use of existing water means the total plant
water "makeup" (water required for all of plant operations including cooling tower evaporation,
boiler water, auxiliary cooling, washdowns, etc.) is much less than other plants that do not
maximize water usage. Even though a dry scrubber sorbent preparation system could reuse part
of the blowdown as makeup water, the plant would still require additional fresh makeup water
for the dry scrubber process. As a result, installing dry scrubbers at Tolk would increase the
makeup water intake requirements for the complex by approximately 9-12%.
Xcel Energy stated that these increased water requirements are highly significant in light of the
critical lack of water in the region. This additional amount of water is simply not available at the
Tolk site. Xcel Energy estimates adding dry scrubbers would require approximately 1,165
acre/feet per year of water availability. Over 30 years, as amortized by EPA for this Proposal,
additional water requirements would be approximately 36,000 acre-feet for operation of the
scrubbers. To obtain the additional amount of water necessary to support the operation of dry
scrubbers, SPS would attempt to purchase significant water rights from existing farmers along
with a gathering system or look at other costly alternatives. Based on the historical cost of water
rights in the area, this is an additional cost of approximately $40 million that was not included in
EPA's cost estimates. This is also assuming that these water rights are available. The acquisition
of these water rights may require the purchase of neighboring agricultural businesses, which
could further increase acquisition costs. These costs for additional water rights, infrastructure or
alternatives are not included in EPA's overall cost estimate for dry scrubber installation. If these
costs were added, they would greatly increase the estimated cost of dry scrubbers at Tolk and the
associated cost per ton of emissions reduction.
Response: In this comment, Xcel states that adding dry scrubbers would require 1,165 ac-ft/yr
(722 gpm)140 of water availability. In subsequent correspondence with EPA, Xcel clarified that
"[t]his estimate is the incremental additional water that would be required to operate dry
scrubbers. This estimate already assumes (i.e. was reduced by) our expectation that we could
140 (1,165 ac-ft/yr)(0.61955 gpm/ac-ft/yr) = 722 gpm.
246
-------
reuse approximately 70% of the current blowdown water in the dry scrubbers."141 The Xcel
comment states that the average blowdown to evaporation ponds is 300 to 350 gpm. Thus, Xcel
estimates that it has 210 to 245 gpm available. In subsequent correspondence, Xcel estimates the
total water requirement for operating dry scrubbers at Tolk is 1,046 gpm, and the additional
water need as being 816 gpm.142
We compared Xcel's water estimate to estimates for comparable dry scrubbers made by others,
expressed on a gallon per pound of SO2 removed basis as the amount of water depends directly
on the amount of SO2 that must be removed. The results of this analysis are summarized and
discussed below.
Unit Dry Scrubber Water Use
PRB Coal
Source
gal/lb SO2
Tolk Units 1 & 2
14.3143
Flint Creek Unit 1
3.9
B&W
3.9-4.6
CBI Unit 1
3.4
CBI Unit 2
3.9
This table shows that Xcel's estimate of Tolk scrubber water demand is very high when
compared to dry scrubber water demand reported elsewhere.
First, Babcock & Wilcox (B&W), a major supplier of SO2 control systems, presented typical
examples of dry scrubber design in its handbook, Steam. Its Generation and Use. This handbook
includes a chapter on scrubber design that includes mass balances for the two general types of
dry scrubbers, recycle and single pass, for a typical PRB coal fired in a 500 MW boiler. These
mass balances indicate that to remove 4,118 lb/hr of SO2, the recycle system would require 317
gpm and the single pass system would require 266 gpm.144 Thus, the unit water use would range
from 3.9 to 4.7 gal/lb SO2 removed.145
141 Email from Jeff West, Xcel, to Joe Kordzi, EPA, September 1, 2015.
142 Email from Jeff West to Guy Donaldson, EPA, October 19, 2015. Xcel assumes the average of 210 and 245
gpm as being available. Thus 1,046 gpm - 230 gpm =816 gpm. We note that the report referenced in this email,
"Phase IAQCS Engineering Services Study - Harrington Generating Station Unit 1, 2, & 3 and Tolk Generating
Station Units 1 & 2, November 2011," assumes all of the estimated 300 gpm blowdown is available for makeup to
dry scrubbers, concluding that an additional 746 gpm is needed.
143 The unit water use for the Tolk units, per pound of S02 removed, is calculated as follows: The SDA Cost IPM
spreadsheets (SDA Cost IPM 5-13 Sources ver 2.xlsx) indicate that the proposed SO2 limit of 0.06 lb/MMBtu is
based on removing 9,195 ton/yr of SO2 at Tolk Unit 171B and 10,015 ton/yr of SO2 at Tolk Unit 172B, for a total
SO2 reduction of 19,210 ton/yr. Thus, the unit water use per pound of SO2 removed: 1,046 gpm)(60 min/hr)(24
hr/day)(365 day/yr)/(19,210 ton/yr)(2000 lb/ton) = 14.3gal/lb.
144 Babcock & Wilcox, Steam Its Generation and Use. Chapter 35: Sulfur Dioxide Control, Figure 13 and Table 8.
145 For the single pass system: (83+183 gpm)(60 min/hr)/(4381-263 lb SO2 removed/hr) = 3.88 gal/lb SO2 removed.
For the recycle system: (48+269 gpm)(60 min/yr)/(4381-263 lb SO2 removed/hr) = 4.62 gal/lb SO2 removed.
247
-------
Second, a study-level analysis, accurate to +/-20%, for Flint Creek Unit 1 (558 MW boiler
burning PRB coal), similar to the Tolk units (533 MW & 543 MW boilers burning PRB coal),
reported water use for four types of scrubbers designed to reduce an inlet SO2 of 0.8 lb/MMBtu
to 0.06 lb/MMBtu, removing 15,155 ton/yr SO2 as follows:146
• SDA:223 gpm
• CDS:228 gpm
• NID: 229 gpm
• WFGD: 250 gpm
The spray dryer absorber (SDA) is the type of dry scrubber that we evaluated for the Tolk units.
It has the lowest water demand of all reported scrubbers. The unit water demand for the SDA,
based on this Flint Creek analysis, is 3.9 gal/lb SO2removed.147
Third, we acquired data from two SDA designs which have been claimed as Confidential
Business Information (CBI). We have included a redacted version of a spreadsheet that shows
our calculations for the CBI data, but does not identify the facility or the source of the data.
These SDA designs are for units of similar size to Tolk, but burn coals containing higher sulfur
contents. These SDA designs have a much lower water demand than Xcel has claimed.
Assuming parity among the most significant impactors to SDA systems, we would expect that
the higher sulfur coals would require more lime and thus more water.
Any of the unit water demand estimates are three times lower than those we estimated above
using Xcel's reported scrubber water demand. These estimates by others indicate Xcel's
estimate is very high for an SDA, suggesting the scrubber is not designed to minimize water use
by, for example, selecting the most efficient spray nozzles. If the lower unit water demand
estimates reported elsewhere are used to estimate Tolk's scrubber water demand, the existing
Tolk water supply is almost adequate to support a dry scrubber, as summarized below:
Revised Tolk Water Demand
Based on Gal/lb SO2 Removed
Additional Water Beyond Boiler Slowdown
Source
gpm
Xcel Estimate
816
Revised, Flint Creek Unit 1
53
Revised, B&W
54-108
CBI Unit 1
19
CBI Unit 2
55
146 American Electric Power, Flint Creek Unit 1, Flue Gas Desulfurization (FGD) Project, Technology Evaluation
and Recommendation, October 25, 2011, pdf 12.
147 The unit water use for the Flint Creek SDA, per pound of SO2 removed, is calculated as follows: The Flint
Creek SDA is designed to remove 15,155 ton/yr SO2. AEP 2011, pdf 22. Thus, the unit water use per pound of SO2
removed: (223 gpm)(60 min/hr)(24 hr/day)(365 day/yr)/[(15,155 ton/yr)(2000 lb/ton)] = 3.87 gal/lb SO2 removed.
248
-------
Revised Water Demand Based on Babcock & Wilcox Case Studies
Assuming the new scrubbers are based on the recycle design, the Tolk scrubbers would require
338 gpm of water148 compared to Xcel's estimate of 816 gpm. As noted above, 230 gpm would
be available as boiler blowdown, requiring only an additional 108 gpm (338 -230 = 260).
Assuming the new scrubbers are based on the single pass design, the Tolk scrubbers would
require 284 gpm of water149 compared to Xcel's estimate of 816 gpm. As noted above, 230 gpm
would be available as boiler blowdown, requiring only an additional 54 gpm (284 - 230 = 54).
Revised Water Demand Based on Flint Creek and CBI Units Estimates
Assuming the Flint Creek water demand of 3.9 gal/lb, the new Tolk scrubbers would require
only 283 gpm.150 As noted above, 230 gpm would be available as boiler blowdown, requiring
only an additional 53 gpm. Using the CBI Unit 1 and 2 data, would require only 249 to 285
gpm.151 As noted above, 230 gpm would be available as boiler blowdown, requiring only an
additional 19 to 55 gpm.
These various estimates indicate that from 19 gpm to 108 gpm of additional water would be
required to operate dry scrubbers on the Tolk units, much less than Xcel's estimate of 816 gpm.
This water will be available when the Plant X units shutdown, as we discuss below.
Plant X Water Supply
In a rate making case in testimony before the Public Utility Commission of Texas, Alan
Davidson, SPS's Director of Regional Capital Projects in the Engineering and Construction
Department of Energy Supply, disclosed that Plant X Units 1 and 2 will be retired in 2019 and
2020, respectively, or before the new scrubbers would be operating. In addition, Plant X Unit 3
would be retired in 2024.152 Thus, water that is currently being used for these units would be
available for the new scrubbers.
148 Revised dry scrubber water demand, based on B&W recycle design: (4.62 gal/lb SO2 removed) (9,195+10,015
ton SO2 removed/yr)(2000 lb/ton)/(365 day/yr)(24 hr/day)(60 min/hr) = 338 gpm.
149 Revised dry scrubber water demand, based on B&W single pass design: (3.88 gal/lb SO2 removed)
(9,195+10,015 ton SO2 removed/yr)(2000 lb/ton)/(365 day/yr)(24 hr/day)(60 min/hr) = 284 gpm.
150 Tolk scrubber water demand, based on Flint Creek unit demand per pound of S02 removed: (3.87 gal/lb S02
removed)(9,195+10,015 ton SO2 removed/yr)(2000 lb/ton)/(365 day/yr)(24 hr/day)(60 min/hr) = 283 gpm.
151 Tolk scrubber water demand, based on CBI Unit 1 demand per pound of SO2 removed: (3.4 gal/lb SO2
removed)(9,195+10,015 ton SO2 removed/yr)(2000 lb/ton)/(365 day/yr)(24 hr/day)(60 min/hr) = 249 gpm. Using
the CBI Unit 2 demand per pound of SO2 removed: (3.9 gal/lb SO2 removed)(9,195+10,015 ton SO2
removed/yr)(2000 lb/ton)/(365 day/yr)(24 hr/day)(60 min/hr) = 285 gpm
152 Direct Testimony of Alan J. Davidson on Behalf of Southwestern Public Service Company, Docket No. 43695,
Application of Southwestern Public Service Company for Authority to Change Rates, Public Utility Commission of
Texas, December 1, 2014 (Davidson Testimony), p. 40.
249
-------
The record does not disclose the current amount of water used by Plant X, but it can be estimated
from the total water demand for both plants combined, minus the water demand for Tolk, which
are separately reported.
Correspondence with Xcel indicates that the combined annual water demand of Plant X and Tolk
ranges between 3.5 billion gallons per year (BG/yr) (6,700 gpm) and 5.2 BG/yr (9,900 gpm).153
The average of these two values is 4.4 BG/yr or 8,300 gpm.
Information provided to the Public Utility Commission of Texas indicates that the annual
average water demand of the Tolk station ranged from 3.4 BG/yr (6,500 gpm) to 4.0 BG/yr
(7,600 gpm) and averaged 3.7 BG/yr (7,000 gpm) over the five year period 2009 to 2013.154
Thus, the average water demand for Plant X is about 1,300 gpm (8,300 - 7,000 = 1,300).
Assuming the upper end of the range, the water demand for Plant X is about 2,300 gpm (9,900 -
7,600 = 2,300). The lower end of the range is not reasonable to consider as it is substantially less
than the blowdown from Plant X's cooling tower (>3,000 gpm) that is sent to Tolk.
Estimated Water Demand for Plant X
Plant
Low End Annual
Water Usage
(gallons)
High End Annual
Water Usage
(gallons)
Average Annual
Water Usage
(gallons)
Plant X + Tolk
3.5 billion
5.2 billion
4.4 billion
Tolk
3.4 billion
4.0 billion
3.7 billion
Resulting for Plant X
N/A
1.2 billion
0.7 billion
Plant X consists of four gas-fired turbines. Below is the annual heat input data for these four
units from 2010 to 2014:155
Annual Heat Input of Plant X Gas-Fired Turbines
5 Yr Heat
Facility
Name
Unit ID
Year
Heat Input
(MMBtu)
Input Avg.
(MMBtu)
%of
total
Plant X
11 IB
2010
1,138,763
Plant X
11 IB
2011
1,473,226
Plant X
11 IB
2012
1,648,757
1,239,256
8.4
Plant X
11 IB
2013
1,708,798
Plant X
11 IB
2014
226,735
Plant X
112B
2010
2,932,116
Plant X
112B
2011
2,345,968
2,741,626
18.6
Plant X
112B
2012
2,756,134
Plant X
112B
2013
3,406,092
153 Email from Jeff West, Xcel Energy Inc. to Joe Kordzi, EPA, August 20, 2015.
154 SOAH Docket No. 473-14-1665, Docket No. 42004, Southwestern Public Service Company's Response to
Sierra Club's First Request for Information Question Nos. 1-1 through 1-49, p. 29.
155 Downloaded from our Air Markets Program at http://ampd.epa.gov/ampd/.
250
-------
Plant X
112B
2014
2,267,818
Plant X
113B
2010
2,208,577
2,695,015
18.3
Plant X
113B
2011
2,256,570
Plant X
113B
2012
3,179,326
Plant X
113B
2013
3,166,651
Plant X
113B
2014
2,663,949
Plant X
114B
2010
6,164,434
8,084,042
54.8
Plant X
114B
2011
7,949,946
Plant X
114B
2012
10,160,965
Plant X
114B
2013
8,008,612
Plant X
114B
2014
8,136,254
Totals
14,759,938
100.0
Based on the average CAMD data from 2010 to 2014, the heat input for Units 111A and 112B
slated for shutdown before scrubber startup, comprised about 27% of the total heat input for the
station (8.4 + 18.6 = 27.0). Thus, about 27% of Plant X's average water demand (1,300 gpm) or
about 351 gpm on average (0.27 x 1,300) or up to 621 gpm (0.27 x 2,300) would be freed up for
use by the Tolk scrubbers. This is more than enough to supply the additional 19 to 108 gpm
required to operate the Tolk scrubbers.156 Further, in 2020, when Unit 113B shuts down, an
additional 238 to 421 gpm of water will be freed up (0.183 x 1,300 & 0.183 x 2,300), which
could be used to offset any subsequent decline in the water table. Thus, we conclude there is
adequate water to continue to supply both Tolk and Plant X, while accommodating the addition
of dry scrubbers.
We believe any costs required to secure Tolk's existing water supply should not be passed off as
costs required for new scrubbers. As demonstrated above, the existing water supply, consisting
of boiler blowdown and excess water from Plant X unit retirements, is adequate to support the
proposed scrubbers. Additionally, a pipeline connecting these two facilities is already in place.
Further, studies done by consultants to SPS who costed scrubbers did not conclude that scrubbers
were infeasible due to lack of water. Rather, they assigned a cost to acquiring the water, but
amortized it over the life of the facility and reported it as an O&M cost in dollars per thousand
gallons. The unit cost of water that we assumed in our cost analysis is substantially higher than
these estimates. Burns & McDonald concluded: "Tolk water (new wells) - $0.75/kgal."157
Elsewhere, URS, another consultant to SPS, concluded: "Any new water consumption will be at
a cost of $0.11/1,000 gallons without considering the capital investment for the new well field
located 50-70 miles from the Tolk plant."158 Thus, as we assumed a water cost of $1.00/kgal,
well in excess of these estimates, we conclude that any additional water required to supply dry
scrubbers could also be obtained by drilling new wells, as assumed in our cost analyses. We
156 Note this calculation is conservative as we included the 2014 value for unit 11 IB in averaging its 5 year heat
input, when it is significantly below its historical value.
157 Burns & McDonnell, Draft Report Presentation, Harrington and Tolk Generating Stations, AQCS Engineering
Services (Phase 1) Study, September 21, 2011, TOLK 0000389.
158 12/12/13 URS, TOLK 0000150.
251
-------
discuss this below.
Based on information in its Securities and Exchange Commission Form 10-Q filing,159 Xcel has
apparently conceded that dry scrubbers are technically feasible at Tolk, and that the water issue
described above could be addressed:
In May 2014, the EPA issued a request for information under the CAA related to
SO2 control equipment at Tolk Units 1 and 2. In December 2014, the EPA
proposed to disapprove the reasonable progress portions of the SIP and instead
adopt a Federal Implementation Plan. The EPA proposed to require dry scrubbers
on both Tolk units to reduce SO2 emissions to help achieve reasonable progress
goals for Texas and Oklahoma national parks and wilderness areas. As proposed,
the dry scrubbers would need to be installed and operating within five years of the
EPA's final action, currently expected in December 2015. Whether dry scrubbers
are required is dependent on the EPA's final decision. If required, they would
cost approximately $600 million, with an annual operating cost of approximately
$10.4 million. SPS believes these costs would be recoverable through regulatory
mechanisms and therefore does not expect a material impact on results of
operations, financial position or cash flows.
We note from the above that Xcel has estimated the total cost of installing scrubbers to be
approximately $600 million. Annualized over 30 years at 7% interest, this results in an annual
cost of $48,351,842. Added to Xcel's annual operating cost of $10,400,000, results in a total
annualized cost of $ 58,751,842. It has been our experience that costs reported in this manner
include AFUDC and owner's costs, which, as explained in our responses to other comments, are
disallowed by the Control Cost Manual methodology we use. Our total proposed SDA costs for
both units, with these costs added back in, total $514,279,000. Annualized over 30 years at 7%
interest, results in an annual cost of $41,443,895. Our proposed operating costs total
$23,359,302. Adding these cost together results in a total annualized cost of $64,803,197, which
is significantly greater than an equivalent annualization of the costs reported to the SEC by Xcel.
We do not know to what degree the costs reported to the SEC are backed by engineering
analyses. However, we conclude this is additional evidence that our proposed costs were
reasonable and inclusive of all costs, including the cost of the water necessary for dry scrubber
operation.
In sum, Xcel will have adequate water to supply dry scrubbers at the Tolk plant, even assuming
its overestimated water demand of 1,046 gpm. Thus, there is no need to ".. .purchase significant
water rights from existing farmers along with a gathering system or look at other costly
alternatives."
Water Rights
Our conclusion that Xcel will definitely have, with the closure of the Plant X units, enough water
to supply dry scrubbers aside, we believe Xcel has added additional water reserves. We
159 United States Securities and Exchange Commission, Form 10-Q, available here: http://www.getfilings.com/sec-
filings/150803/SOUTH WESTERN-PUBLIC-SER VICE-CO10-Q/.
252
-------
contacted the High Plains Water District (HPWD) about Xcel's water rights in the area of Tolk
and Plant X, and were provided with the following map:
We were informed by the HPWD160 that this map was provided to it by Xcel and represents the
areal extent of land to which Xcel controls the water rights (note the numbers refer to metered
locations and not well locations). The boundary encloses an area of 47,188 acres. The HPWD
informed us that according to its rules, this allows Xcel to annually withdraw 18" for every acre
it controls, or in excess of 23 Billion gallons annually.
Xcel has supplied us with information161 that causes us to conclude that due to aquifer depletion
and well drawdown, the actual amount of water that can annually be withdrawn is less. Xcel
states that it has 73 wells that annually produce from 3.5 to 5.2 billion gallons (11,000 acre-feet
to 15,900 acre-feet). Xcel explained that it expects its well productivity to decrease due to
aquifer depletion. We find this to be a reasonable conclusion. However, according to Xcel's
own information, the percentage of Tolk's total water usage that would be devoted to the dry
scrubber is 9-12%. Consequently, we conclude that absent a switch to dry or hybrid cooling
(which we discuss below), the aquifer's depletion will be a limiting factor on the plant itself, not
on the operation of the scrubbers.
We believe Xcel has recognized the strategic need for additional water and has taken steps to
acquire additional water rights, and thus potentially limit its future water usage. For instance,
based on information we have reviewed, we believe that additional acreage has been added to the
above map. In testimony in the aforementioned rate making case, Mr. Davidson identified the
following capital projects for rate recovery that were completed or underway:
168 See email from Jed Leibbrandt to Joe Kordzi, dated 8/12/2015.
161 See emails from Jeff West to Joe Kordzi dated 8/20/2015 and 8/26/2015.
253
-------
• Tolk Station - Purchase Barrett Water Rights (Parent Work Order 11763413) - This
project consisted of purchasing additional water rights, wells, electrical infrastructure,
roads, and rights-of-way from the Barrett family that were contiguous with SPS's current
water rights. The additional water rights and wells will be used to offset production
declines from SPS's existing wells supplying water to Tolk Station and to Plant X.
• Tolk Station and Plant X - Water Wells Phase 6 (Parent Work Order 28 11488948) -
This project continued the development of new water wells to off-set production declines
from existing wells at the Tolk Station and Plant X. Three new wells, additional
gathering pipelines, electrical service to the 31 wells, roads, pumps, and controls were
added. To maintain reliability, it is critical to have adequate water supply available all
year, but especially in the summer. Decline of SPS's current well field production puts
Tolk Station and Plant X at risk for a water shortage.
• Tolk Station - Purchase Rose Water Rights (Parent Work Order 33 11649187) - This
project consists of purchasing additional water rights, 34 roads, and rights-of-way from
the Rose family that are contiguous with SPS's current water rights. The additional water
rights will be used to off-set 2 production declines from SPS's existing wells supplying
water to Tolk 3 Station and to Plant X.162
In response to our questions regarding these additional water rights and other related questions,
Xcel has provided the following map that depicts the current areal extent of its water rights, and
indicates the locations of its wells and pipelines:
162 Davidson Testimony, December I, 2014.
254
-------
The yellow boundary defines the areal extent of Xcel's water rights. It can be seen from a
comparison to the HP WD map that Xcel has added additional acreage to the southeast, which is
outlined in red. Possibly additional acreage has also been added in the western end.
Xcel has also supplied a map that indicates the saturated aquifer thickness.163 The saturated
thickness is a measure of the amount of recoverable water.164 The greater the saturated
thickness, the more water that can be recovered using conventional methods.
163 Email from Jeff West to Joe Kordzi dated 8/26/15. See Attachment 2.
164 Recoverable water = surface acreage x saturated thickness x specific yield (fraction of bulk aquifer volume that
will yield water).
255
-------
It can be seen from the above map that the additional area added to the southeast (and possibly
the west) lies within the areas containing some of the greatest saturated aquifer thickness. This
map also indicates that additional areas for potential future development and exploration are
available to Xcel for the drilling of additional wells. According to Xcel, individual wells have
achieved peak flows of up to 550 gpm.165 Even considering production decline, we conclude
that one or two additional wells located in these high production zones would be adequate to
supply the additional water needed for Tolk's scrubbers. Further, even areas that are currently
not economic to produce with a standard vertical well would likely be available for production
with alternative well completions.166 We therefore conclude that even discounting our finding
that Xcel will have more than adequate water available to run our proposed dry scrubbers with its
planned shutdown of the Plant X units coupled with its boiler blowdown currently sent to an
evaporation pond, it has more than adequate additional capacity within its well field to supply
these scrubbers.
165 Email from Jeff West to Joe Kordzi dated 8/26/15.
166 Letter Report from Tyler A. Davidson. LBG-Guyton Associates, to Amy Williite, Xcel Energy, Re: Tolk
Station/Plant X Water Rights Area Saturated Thickness and Static Volume, June 17, 2015, p. 2.
256
-------
As we noted above, the decline in the aquifer, which we acknowledge, is not a limiting factor on
the installation, but rather on the continued operation of the plant, since other plant processes use
much more water than would dry scrubbers. Xcel has recognized this basic limitation and has
taken steps to address it, beyond simply acquiring more water rights. For instance, much of a
typical coal fired power plant's total water budget goes to wet cooling. We contacted GEA Heat
Exchangers concerning the potential conversion of Tolk's wet cooling system to a dry or hybrid
system. We were informed by GEA that Xcel had already contacted them for an estimate to
convert their existing wet cooling system to a hybrid system in which 70% of the cooling needs
would be handled by dry cooling and 30% would be handled by wet cooling. GEA supplied us
with redacted and unredacted versions of its quote to Xcel, which covered a number of potential
options. We have placed the redacted version in our docket.167 Although we have concluded
that adequate water is available, it appears that a hybrid cooling system, such as that costed by
GEA, is technically feasible and could increase the water supply by as much as 232 gpm. We
encourage Xcel to consider adopting such a system in order to limit its water usage.
Comment: Baghouse- and waste-related costs were not adequately considered [at Tolk].
[Xcel Energy (0064) p. 31-32]
Xcel Energy stated that Tolk beneficially re-uses 100% of its fly ash. In the Technical Support
Document for the Cost of Controls Calculations for the Texas Regional Haze Federal
Implementation Plan ("Cost TSD"), EPA recognizes that it is important to retain consistent fly
ash specifications to be able to beneficially re-use fly ash. Cost TSD, at 9 (2014) (EPA Docket
ID EPA-R06-0AR-2014-0754-0008). Tolk currently has a baghouse to control particulate matter
emissions. If dry scrubbers were to be installed at Tolk, the plant would not be able to
beneficially re-use 100% of its flyash, which would result in two major issues that EPA failed to
consider. The first is the potential need for additional baghouses and the second is the need for a
landfill.
Xcel Energy stated that there are two options for a scrubber to be placed in the flue gas stream.
Option A would locate the scrubber upstream of the existing baghouse and Option B would
locate the scrubber downstream of the existing baghouse but it also would require an additional
baghouse to be built downstream of the scrubber.
Option A would end the re-use of the existing fly ash as all of the fly ash would be contaminated
with scrubber waste and, thus, would no longer be suitable for re-use. Option A is lower in
capital cost and aligns better with EPA's current estimate, which does not factor in additional
baghouse costs. The loss of revenue from the sales of the fly ash (approximately $2 million
annually), additional O&M for handling of fly ash and bottom ash (currently netted out of fly ash
revenues) and increased bag replacements would have to be included in the total cost estimates
for SO2 removal. It is Xcel Energy's opinion at this time that Option A has the most economic
benefit of the two options even though EPA's cost estimates are artificially low.
167 GEA, Investigation of Conversion of Wet to Dry/Wet Cooling System for Tolk Power Plant, Texas, USA,
September 19, 2012. The current cooling tower makeup demand is on p. 11 and the makeup for the various options
that were studied are found at pp. 18, 24, 32, 38, and 45. The best case option, p. 32, reduces makeup water
demand by (173,659,000 gal - 51,498,000 gal/yr)/[365 day/yr x24 hr/day x 60 min/hr] = 232 gpm.
257
-------
Option B would allow for the continued re-use of the existing fly ash, but would require
substantial additional capital cost estimated at $45-62 million for Tolk for a post scrubber
baghouse. This cost estimate is based on historical studies conducted by Bums & McDonnell
Engineering Company, Inc. ("B&M") for Xcel Energy's Harrington Generating Station, where
B&M estimated $38 million for a new baghouse. Harrington's coal-fired units are smaller in size
than Tolk's. Accordingly, for Tolk, Xcel Energy increased the $38 million estimate by 37% to
address the ratio between the size of the Tolk units and the Harrington units. Xcel Energy then
scaled down the Tolk estimate by 10% as the secondary baghouse would be smaller than a full
particulate baghouse but still would require a high cloth to air stream number. Finally, Xcel
Energy increased the Tolk estimate by 12% for overhead, engineering and management and 15%
for contingency for unknowns.
Xcel Energy stated that the EPA erroneously assumed this post scrubber baghouse was in place
and would allow Tolk to continue the re-use of ash. This is not the case for Tolk. It is Xcel
Energy's opinion that the fly ash revenue stream and reduced O&M expense do not cover the
additional cost of the baghouse and this option is not economically viable. If Option B were
chosen by EPA, then EPA would need to add substantial capital (with additional O&M) to its
cost estimates for emissions removal.
Xcel Energy stated that either scrubber arrangement option would require a new landfill at Tolk,
a cost that EPA also has not factored into its estimates. Tolk currently generates approximately
170,000 tons of fly ash on an annual basis. The addition of a scrubber would generate
approximately 80,000 tons of scrubber waste. There is variability to this number as some of the
existing ash is used in start-up and becomes part of the waste.
Xcel Energy stated that a new landfill would vary in size and capacity depending on which
option was chosen for the baghouse arrangement. If Option A were selected, then the total of the
ash and scrubber waste would need to be landfilled annually. If Option B were selected, then
only the scrubber waste would be landfilled annually. Under Option A, Xcel Energy estimates
the cost for the construction of a landfill for this ash would be an initial capital investment of $10
million with ongoing capital expenses every five years of approximately $5 million for new cell
construction. [See Appendix D to comment 0064, excerpt from Harrington Generating Station
Units 1, 2 & 3 and Tolk Generating Station Units 1 & 2 Phase I Air Quality Control Study
Engineering Services Project Study Prepared for Xcel Energy, B&M (Nov. 2011 ).] B&M
estimated the initial cost of a landfill to be approximately $1.1 to $5.2 million. There also are
ongoing capital costs for the addition of a new cell and capping an old cell every five years at a
cost of approximately $1.1 to $5.2 million. After reviewing these costs, Xcel Energy utilized
past internal experience for site work for ponds, estimates for using a lime landfill with minimal
liner and monitoring wells and concluded that the B&M costs were understated for the initial
landfill as they only considered the first cell opening and not the development of the whole site.
Therefore, Xcel Energy estimates that initial landfill development (site selection and first five-
year cell) would be approximately $10 million and then $5 million every five years for ongoing
cell development. These costs do not include the ongoing operational costs associated with ash
hauling placement and compaction in the landfill. The current EPA estimate fails to include
these additional capital costs. Additionally, the new CCR rule will impose significant new
258
-------
requirements for the construction of an ash landfill. These requirements include prescribed
composite liners, leachate collection, ground water monitoring and a host of other operating
costs that were not included in EPA's cost estimates. There also are O&M costs associated with
ash handling and placement in a new landfill that were not factored into EPA's cost analysis. See
Cost TSD, at 6.
Regardless of the arrangement of a scrubber in the flue gas stream, Xcel Energy noted that the
EPA failed to include all of the costs necessary to manage and landfill the wastes created by the
scrubber, which artificially lowers EPA's estimated cost per ton removed. EPA also failed to
properly analyze the cost benefit ratio of discontinuing the re-use of the fly ash on the
environment by forcing the creation of a landfill that would be attendant to the addition of
scrubbers at Tolk.
Response: We are aware that Tolk beneficially re-uses 100% of its fly ash, and we support this
re-use. This comment identifies two options for location of the scrubber in the pollution control
train: (1) Option A locates the scrubber upstream of the existing baghouse and (2) Option B
locates the scrubber downstream of the existing baghouse.
The EPA's cost estimate locates the scrubber downstream of the existing baghouse and includes
the costs of a new baghouse downstream of the scrubber (Option B in this comment), to protect
the beneficial use of flyash and enhance the SO2 removal efficiency of the dry scrubber. A
substantial amount of the SO2 removal achieved by a dry scrubber occurs in the tail-end
baghouse because the scrubber ash is rich in lime. Without the tail-end baghouse, the dry
scrubber would only achieve about 75% SO2 control. As the Integrated Planning Model (IPM)
algorithms that EPA used in its Tolk cost estimate are based on 95% SO2 control, the costs must
therefore include a tail-end baghouse.168
The IPM support document for the SDA dry scrubber,169 for example, states: "The curve fit was
set to represent...a "typical" SDA FGD retrofit for removal of 95% of the inlet sulfur."170 A 95%
SO2 control efficiency can only be achieved with a tail-end baghouse. The O&M costs for this
tail-end baghouse are included in EPA's cost analysis. The IPM support document, for example,
states: "The fixed maintenance materials and labor (includes bag replacement) is a direct
function of the process capital cost..."171 The baghouse costs are included in the "balance of plant
costs" in the IPM SDA model. We confirmed that the SDA cost model we relied on includes a
tail-end baghouse with the EPA Lead Reviewer of the IPM Project.172
This comment also argues that either scrubber option would require a new landfill at Tolk, which
it asserts is not included in EPA's cost estimate. The comment presents capital and O&M costs
for a new landfill for Option A, an upstream baghouse and landfilling of all ash and scrubber
168 Sargent & Lundy, White Bluff Station Units 1 and 2, Evaluation of Wet vs. Dry FGD Technologies, October
28, 2008, pp. 2-2 and 2-9.
169 Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC Technologies, SDA FGD Cost
Development Methodology, Final, March 2013 (3/13 S&L SDA).
170 3/13 S&L SDA, p. 1.
171 3/13 S&L SDA, p. 4.
172 E-mail from William Stevens, EPA, to Joe Kordzi, EPA, Re: Question on SDA IPM Cost Algorithm, August 13,
2015.
259
-------
wastes from a 2011 Burns & McDonnell report, but failed to report any cost information for
Option B, a tail-end baghouse, which is the option that EPA costed.
However, subsequent work by URS in 2013 for Xcel reviewed and revised the 2011 Burns &
McDonnell work cited in this comment. The subsequent URS study concludes with respect to
the landfill: "the B&Mc study included the capital cost for a new cell to be developed. In this
analysis, no additional capital development cost is included; the landfill costs for solid wastes are
dealt with as strictly an operating cost at the $/ton rate specified in Section 3. The URS study
further does not include costs for the TCEQ unlined landfill registration process as were included
in the B&Mc study."173 This 2013 URS study assumes disposal costs of $5/ton for calcium
solids and $10/ton for sodium solids disposal in 2013 dollars.174 In comparison, we assumed a
dry FGD solids disposal cost (sum of sodium and calcium solids) of $30/ton in 2012 dollars.
Thus, we substantially overestimated these costs and do not believe a revision is required.
The comment is correct that the new CCR rule175 may impose significant new requirements for
construction of landfill as these requirements apply to flue gas desulfurization materials. We
note that Xcel has stated in discussing its corporate costs (not just Tolk) that, "Based on our
initial review of the final rule, we believe our costs for the management and disposal of coal ash
will not significantly increase under the new rule."176
Regardless, a new ash landfill is not required for the small amount of calcium and sodium solids
that will be generated by the Tolk dry scrubbers. The EPA's cost analysis assumes these
materials will be sent for offsite disposal and assigns a per ton cost of $30/ton. As EPA's per ton
cost is substantially higher than estimated by SPS's consultants before the CCR rule was on the
horizon, we believe it has adequate margin to accommodate any future increase in cost to comply
with the CCR rule. However, even if the scrubber solids disposal cost were increased to $50/ton,
the cost effectiveness of the Tolk scrubbers would remain highly cost effective. At Tolk 171B,
cost effectiveness would increase from $3,178/ton to $3,285/ton and at Tolk 172B, from
$2,998/ton to $3,091/ton or by 3%. Thus, no change is required.
Below, we reproduce the Securities and Exchange Commission Form 10-Q filing we referenced
in another comment, in which SPS reported its estimated cost of proposed scrubbers at its Tolk
units:177
In May 2014, the EPA issued a request for information under the CAA related to
SO2 control equipment at Tolk Units 1 and 2. In December 2014, the EPA
proposed to disapprove the reasonable progress portions of the SIP and instead
173 12/12/13 URS, TOLK 0000115/116.
17412/12/13 URS, Table 4, TOLK 0000112.
175 http://www2.epa.gov/coalash/coal-ash-rule
176 https://www.xcelenergv.com/staticfiles/xe/Corporate/CRR2014/environment/environmental-policY.html. Xcel
Corporate Responsibility Report 2014, page 119, available at:
https ://www. google, com/url? sa=t&rct=j &q=&esrc=s&source=web&cd=2&ved=0 CCMQFj ABahUKEwj akMaO-
f7HAhUMGz4KHZdPCoE&url=https%3A%2F%2Fwww.xcelenergy.com%2Fstaticfiles%2Fxe%2FCorporate%2F
CRR2014%2Fcrr-2014 ,pdf&usg=AFQj CNGPdTb VnVS 1 ljNJE 1 nD JH8k0o 14Q&cad=ija
177 United States Securities and Exchange Commission, Form 10-Q, available here: http://www.getfilings.com/sec-
filings/150803/SOUTH WESTERN-PUBLIC-SER VICE-COIO-Q/
260
-------
adopt a Federal Implementation Plan. We proposed to require dry scrubbers on
both Tolk units to reduce SO2 emissions to help achieve reasonable progress goals
for Texas and Oklahoma national parks and wilderness areas. As proposed, the
dry scrubbers would need to be installed and operating within five years of the
EPA's final action, currently expected in December 2015. Whether dry scrubbers
are required is dependent on the EPA's final decision. If required, they would
cost approximately $600 million, with an annual operating cost of approximately
$10.4 million. SPS believes these costs would be recoverable through regulatory
mechanisms and therefore does not expect a material impact on results of
operations, financial position or cash flows."
While SPS's capital costs are higher than ours ($600 million vs. $445 million), likely due to the
use of the all-in costing method and the inclusion of invalid costs such as the costs discussed
above plus AFUDC and owners cost, its O&M costs are lower (10.4 million/yr vs. our 23.4
million/yr). The cost effectiveness of scrubbers at Tolk, based on SPS's Form 10-Q report is
then $3,059/ton removed178. This is approximately equal to our own calculations for scrubber
retrofits for Unit T171B ($3,178/ton) or T172B ($2,998/ton).
Comment: EPA did not adequately consider lower-cost options for compliance, such as dry
sorbent injection [at Tolk]. [Xcel Energy (0064) p. 32-33]
Xcel Energy argued that EPA too quickly eliminated the lower-cost option of dry sorbent
injection ("DSI") at Tolk without sufficient evidence for doing so in the proposal. In the Cost
TSD, EPA says, "we lack the site specific information, which we believe requires an individual
performance test, in order to be able to accurately determine the maximum SO2 removal
efficiency for the individual units .... "and EPA goes on to note that such site-specific
information has already shown infeasibility at more than one unit (i.e., Luminant's Big Brown
and Monticello units). Cost TSD, at 7. Yet EPA still evaluates "each unit at its maximum
recommended DSI performance level[:] ... 90% SO2 removal..." Cost TSD, at 8. Without the
site-specific information that EPA admits it needs and does not have, it is arbitrary to select the
maximum performance level for each unit under consideration. This lack of information does
not fully consider all appropriate options for controls at a facility. In turn, EPA has failed to
properly perform a four-factor analysis because EPA arbitrarily and unreasonably failed to
consider alternative, less expensive controls such as DSI, or to compare the incremental costs
and environmental benefits associated with such controls and scrubbers at particular units. EPA
has done this comparative cost incremental analysis for source-specific RPG controls in other
regional haze SIPs/PIPs. See Proposed North Dakota SIP Approval/Disapproval, 76 Fed. Reg.
at 58,631 (analyzing RPGs for a source in the North Dakota FIP using incremental cost
effectiveness analyses for various controls of NOx emissions).
Response: First, it is standard practice in cost estimating to use a performance range to bound
costs when a site-specific estimate is not available. The BART Guidelines explain this
procedure. When a range of performance levels are possible, as here, the Guidelines hold that:
178 Cost effectiveness of scrubbers at Tolk Units 174A and 174B, based on its June 30, 2015 10-K filing:
[(600,000,000)*0.0806 + 10,400,000]/[10,015+9,195 ton/yr] = $3,059/ton S02 removed.
261
-------
"2. In assessing the capability of the control alternative ... Without a showing of
differences between the source and other sources that have achieved more stringent
emissions limits, you should conclude that the level being achieved by those other
sources is representative of the achievable level for the source being analyzed.
3. You may encounter cases where you may wish to evaluate other levels of control in
addition to the most stringent level for a given device. While you must consider the most
stringent level as one of the control options, you may consider less stringent levels of
control as additional options... ,"179
This is the procedure we followed in selecting the range of 50% to 80% (ESP) to 90% (FF) for
DSI.180 This procedure requires that the upper end of the range be the level achieved by other
sources when there is no contrary information, as here, and that lower levels may also be
evaluated, so long as the most stringent level is considered. The IPM DSI methodology
identifies 80% (ESP) to 90% (FF) SO2 control as the upper end of the range and 50% as the
lower end.181 Thus, the evaluation of this range, in the absence of site specific tests, is not
arbitrary, but rather standard practice. In response to our request for information under Section
114 of the CAA, Xcel itself has provided to us DSI analysis that indicates it assumed control
efficiencies that fit comfortably with the range of efficiencies we analyzed.
Second, this comment incorrectly asserts that "EPA goes on to note that such site-specific
information has already shown infeasibility at more than one unit ().. .yet EPA still evaluates
"each unit at its maximum recommended DSI performance level[:]... 90%." This is incorrect.
Luminant disclosed that DSI had been tested at its Big Brown and Monticello units and was
found to be infeasible. Thus, while we presented DSI costs for these units for completeness, we
concluded that DSI was not feasible at these units, based on Luminant's site-specific
information. We evaluated all other units that are not currently controlled with a scrubber
assuming DSI at both 50% and 90% and invited comments on the feasibility of DSI on these
units.182
The performance of a DSI system depends on coal type, combustion conditions, ductwork,
temperature, other acid gases in the flue gas (i.e., HC1), and other air pollution control equipment
at a plant. Every boiler unit is unique. A reliable estimate of the actual performance of DSI
cannot be determined without an on-site demonstration test.183 However, it is feasible to evaluate
179 40 CFR 51, Appx. Y, p. 627.
180 See Cost TSD, pp. 6-8. A control level of 80% was used when the unit was equipped with electrostatic
precipitators (ESP) and a control level of 90% was used when the unit was equipped with fabric filters (FF).
181 Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC Technologies, Dry Sorbent Injection
for SO2 Control Cost Development Methodology, March 2013, pp. 1,3.
182 See Cost TDS at pp. 6-8; 76 FR 74,876. Table 32 (December 16, 2014).
183 See, for example, United Conveyor Corp., Dry Sorbent Injection for SO2, S03, Hg and HC1, Available at:
http://unitedconvevor.com/uploadedfiles/svstems/drv sorbent iniection/ucc-
040 dsibrochureusversion spreadsfinal.pdf: DCI-ACI Technology for MAT Compliance, Power Engineering:
("Testing is the first step for determining emission rates and ascertain if existing APC equipment can achieve
compliance on their own."), Available at: http://www.power-eng.com/articles/print/volume-118/issue-l/features/dsi-
262
-------
the range of expected performance, based on numerous tests conducted by vendors and CBI data
submitted in response to our Section 114 requests. Thus, we evaluated 50% and 80% (ESP) or
90% (FF) SO2 control using DSI at all unscrubbed units including the Luminant units, where DSI
was concluded to be infeasible based on site-specific tests.
As to Tolk, SPS's response to EPA's Section 114 request did not include any DSI performance
data. Thus, we evaluated the typical range of 50% to 90% for the Tolk units to bound
performance. Evaluating a control range to bracket performance is a standard method used in
cost effectiveness analysis.
In fact, the documents produced by SPS include a URS study commissioned by SPS that
concluded in 2013 that "[s]ite specific testing is the suggested way to determine the impacts that
might be expected at the Tolk and Harrington plants." TOLK 0000116. SPS did not produce
the test results recommended by its consultants. Thus, we have no basis to evaluate a control
efficiency outside of the widely accepted range of 50% (ESP) to 90% (FF) SO2 control and
specifically higher than 90%, as recommended by the IPM DSI documentation. It would be
arbitrary for us to assume a higher control efficiency than 90%, which is the upper bound
recommended by the IPM documentation, or a lower efficiency than 50%, which is the
recommended lower bound when a facility is equipped with a baghouse.
Further, a higher SO2 control efficiency would not alter our decision to select SDA rather than
DSI. In fact, it supports our decision to eliminate DSI. As we state on page 30 of our FIP TSD:
As we note in section 5, for all but one of the units we analyzed that currently
have no SO2 controls, even at the lower level of control of 50%, the cost-
effectiveness of DSI was worse (higher $/ton) than either SDA or wet FGD, even
with the latter options offering much greater levels of control and visibility
benefit. At the higher 80% or 90% level of control, the cost-effectiveness of DSI
was worse than either SDA or wet FGD in all cases. Consequently, we are not
proposing that DSI be installed at any unit.
Xcel has taken the statement, "we lack the site specific information, which we believe requires
an individual performance test, in order to be able to accurately determine the maximum SO2
removal efficiency for the individual units..." out of context. Our cure for the lack of site
specific data was to evaluate a range of control efficiencies, known to be achievable, based on
the IPM model documentation, the DSI literature, and confidential information submitted by
other utilities and invited comment on its applicability to specific units.
Our purpose in testing the cost effectiveness of DSI at its theoretical upper level of control was
to both acknowledge the problem of a lack of site specific data, but in so doing to also bound the
aci-technology-for-mats-compliance.html: Thomas A. Dunder, Stack Testing Technologies for DSI Evaluation
Studies ("DSI must be evaluated at each facility to determine optimal sorbents, injection rate, and injection points."),
Available at: http://www.slideshare.net/trc companies/euec-paper-c5-l-emissions-testing-for-dsi-evaluation-trc.:
Diane Fischer and Preston Tempero, Early Lessons Learned from Implementation of Dry Sorbent Injection Systems,
p. 5 ("Experience has shown that baseline testing and sorbent injection testing are critical steps in the design of DSI
systems."), Available at: http://bv.com/docs/reports-studies/early-lessons-learned-from-implementation-of-dsis.pd.
263
-------
range of DSI cost effectiveness. In other words, we wished to test whether any reasonable range
of DSI control was cost effective in comparison with scrubbers. As we discussed in our Cost
FIP: "...we believe that the maximum performance level for DSI can only be determined after an
onsite performance test. Therefore, we don't know whether a given unit is actually capable of
achieving these DSI control levels, and (2) we believe it is useful to evaluate lesser levels of DSI
control (and correspondingly lower costs). We therefore also evaluated all the units at a DSI SO2
control level of 50%, which we believe is likely achievable for any unit."184 The Tolk dry
scrubbers were evaluated at an SO2 control efficiency of 91.7%, while the maximum SO2 control
efficiency for DSI was evaluated at 90%. As we have demonstrated, the cost effectiveness of
DSI worsens (higher $/ton) as the control level increases. Thus, if the cost effectiveness of DSI
at 90% control is already worse than the cost effectiveness of SDA at 91.7% control, then as we
demonstrated, DSI cannot compete with SDA in cost effectiveness. This ignores the increased
visibility benefit of SDA due to its higher control level, increasing the attractiveness of SDA.
One consultant to SPS, Burns & McDonnell, evaluated DSI at 70% SO2 control, while another
evaluated the range of 50% to 80%,185 both without any demonstration test data.186 These ranges
are consistent with the range we used. The resulting cost effectiveness values for DSI at both
50%) and 90% are reported in the Cost TSD, Table 5, and in the proposed rulemaking, 79 FR
74,876, Table 32 (December 16, 2014).
Third, this comment claims we failed to properly perform a four-factor analysis because it
"arbitrarily and unreasonably failed to consider less expensive controls such as DSI..." This is
incorrect. As explained above, we considered DSI at two control levels, 50% and 80% (ESP) to
90%) (FF), which bracket the range of demonstrated performance at other similar facilities,
documented in the DSI literature and CBI data submitted by other applicants.
Fourth, this comment claims "EPA arbitrarily and unreasonably failed .. .to compare the
incremental costs and environmental benefits associated with such controls and scrubbers at
particular units." As explained in the BART Guidelines: "The incremental cost effectiveness
analysis compares the costs and performance level of a control option to those of the next most
stringent option..." Part51,Appx. Y, p. 629.
As an initial matter, our control cost and visibility analysis was performed under the reasonable
progress and long-term strategy requirements of the Regional Haze Rule, which do not require
the type of incremental cost analysis to which Xcel refers. That type of analysis is required
under the BART provisions of the Regional Haze Rule. That aside, the incremental cost
effectiveness is negative, meaning it costs less to reduce more SO2 with the most effective
control than with the next most efficient control. The cost of removing 167 ton/yr more SO2 at
Unit 17IB using SDA eliminates spending $3.2 million dollars per year for only 90% control
using DSI. This supports the selection of the SDA over DSI. Similarly, the cost of removing 84
ton/yr more SO2 at Unit 172B using SDA eliminates spending $1.9 million per year, compared to
the next most effective control, a 90% efficient DSI. Thus, the cost of removing additional SO2
is lower with an SDA than with a 90% efficient DSI and no incremental cost analysis is required.
184 FIP TSD, p. 11.
185 URS ccc
186 Burns & McDonnell, TOLK 00001092/1093.
264
-------
As noted above, we contrasted the cost effectiveness and visibility benefits of DSI with dry and
wet scrubbers and concluded that in all cases, scrubbers represented a better cost versus benefit
solution.
Lastly, this comment asserts we failed to consider the environmental benefits of DSI compared
to SDA. In the TSD FIP, p. 8, we concluded that DSI and scrubbers do not present any unusual
environmental impacts, with the exception of water supply at Tolk, which we addressed by
eliminating wet scrubbers. The only environmental benefit of DSI is that it does not use any
water. However, DSI has many adverse impacts that offset this benefit.
Elsewhere, Xcel disclosed that if lime sorbent for SDA were injected upstream of the existing
baghouse, it would render the fly ash unusable as Tolk beneficially re-uses 100% of its fly ash.187
While we did not propose this option, this disclosure by Xcel reveals a new adverse impact of
DSI that we did not consider that supports our decision to eliminate it.
In DSI, trona is injected upstream of the baghouse, which would contaminate the baghouse
flyash. This would result in additional costs, not included in our DSI cost analysis. Based on
Xcel's claims in -another comment, this could include: (1) lost revenue from the sale of the fly
ash (approximately $2 million annually), (2) increased capital cost for a new landfill ($10 million
for landfill development and then $5 million every 5 years for ongoing cell development); and
(3) additional O&M for handling of bottom ash, which is currently netted out of fly ash revenues.
The cost model we used included the resulting increase in O&M cost for fly ash handing and
increased bag replacements due to increased particulate loading from injected trona.
Additionally, the new CCR rule may impose significant new requirements for the construction of
an ash landfill. These requirements include prescribed composite liners, leachate collection,
ground water monitoring and a host of other operating costs that were not included in EPA's DSI
cost estimates. The adverse impact of DSI on the salability of Tolk's fly ash and attendant fly
ash waste disposal issues supports our decision to eliminate DSI.
Further, DSI injection upstream of the baghouse could result in other adverse environmental
impacts, including reduction in mercury capture and increase in NOx and particulate matter
emissions as well as sorbent buildup in ducts and particulate control devices, causing degraded
performance and frequent downtime.188
SPS recently installed mercury control at its Tolk plant, consisting of activated carbon injection
(ACI) upstream of the baghouse to comply with the mercury standards under the Mercury and
Air Toxics Standards (MATS) rule. 189 Sodium sorbents used in DSI, such as trona, generate
NO2 in the presence of carbon, which increases NOx emissions and reduces mercury capture by
carbon-based sorbents, limiting SO2 control to about 30% to 40% and requiring increased
187 Xcel Energy (0064) pp. 31-32.
188 Cost TSD, p. 8.
189 8/20 15 West Letter, p. 3; Letter from Jeff West, Xcel, to Joe Kordzi, EPA, Re: Proposed Approval/Disapproval
of Texas and Oklahoma Regional Haze State Implementation Plans; Federal Implementation Plan for Regional
Haze, August 28, 2015; See also: http://www.getfilings.com/sec-filings/150803/SQUTHWESTERN-PUBLIC-
SERVICE-CO 10-0/ and 2014 Form EIA-860 Data - Schedule 6B, 'Emission Standards and Control Strategies,"
Available here:http://www.eia. gov/electricitv/data/eia860/.
265
-------
amounts of carbon to sustain mercury reduction goals. A recent study concluded:
"Although choosing DSI for SO2 or HC1 trim may appear to be the appropriate
economic choice for many unscrubbed plants, challenges associated with
inadvertent impacts on mercury control due to N02, and impacts to ash and
landfill management may result in additional unplanned costs."190
A consultant to SPS reported similarly:
"Initial tests of the DSI on PRB fired units with ACI installed for Hg reduction
indicated some negative impacts on Hg emissions performance. This was
confirmed during recent testing at a Michigan generating plant where activated
carbon usage was seen to increase when used in conjunction with a trona DSI
system. It appeared that the catalytic reaction of NO in the presence of the
sodium compounds produced sufficient quantities of N02 to interfere with the
ACI adsorption of mercury (N02 competes for active sites on AC and inhibits
carbon capture of Hg). The AC feed rates at that facility had to be increased more
than 30-40% to maintain similar Hg performance when operating the trona DSI
system at SO2 removal efficiencies above 70%."191
DSI also results in adverse landfill impacts. Sodium in the trona is very soluble and sodium-
containing ash can mobilize heavy metals in ash. Discharge limits are of particular concern and
are identified in the proposed Effluent Limitation Guidelines. Sodium-containing ash entering
an ash pond or landfill may mobilize metals otherwise stable in the pond, such as selenium and
arsenic. Sodium-contaminated fly ash may also require special ash handling procedures to limit
sodium leaching, which can affect local vegetation. If large amounts of sodium are present in the
ash, landfill subsidence may be a concern if significant mass leaches out of the landfill, leaving
voids and thus affecting stability.192
Alternatively, if trona is injected downstream of the existing baghouses, thus protecting the
current sale of flyash, additional costs would be incurred for a tail-end, polishing baghouse plus
landfilling the spent trona. The SO2 control efficiency in the downstream position would be
lower than in the upstream position because control efficiency is much greater at the higher
temperatures upstream of the existing baghouses than downstream.193 However, this position
would not eliminate the solids disposal issues discussed above, as the waste trona itself must still
be disposed. These additional costs coupled with lower SO2 control efficiencies would render
190 C. Senior and S. Sjostrom, A Look into the Crystal Ball: Post-MATS Utility Environmental Challenges, August
19-21, 2014, pp. 6-9, Available at: http://www.adaes.com/wp-content/uploads/Megal4_116_Senior.pdf.
191 TOLK 000016.
192 Senior and Sjostrom, 2014, pp. 7-8; K. Baldrey, S. Sjostrom, and K. Ellison, Options to Stabilize and Utilize
Tomorrow's Fly Ash: Approaches and Intial Results, 2015 World of Coal Ash Conference, May 5-7, 2015,
Available at: http://www.flyash.info/2015/105-sjostrom-2015.pdf.
193 M. Wood and others, Solvay Chemicals, Dry Sorbent Injection of Trona or Sodium Bicarbonate for Air
Pollution Control and Corrosion Prevention, CoalPower, 2008, pdf 29 ("Flue gas temperature - minimum 275 F, the
higher, the better up to 800 F).
266
-------
the incremental cost effectiveness of SDA relative to DSI even less cost effective.
Finally, DSI injection anywhere along the pollution control train can increase particulate matter
emissions, adversely affect the performance of particulate control devices, and cause plugging
and caking in ducts, causing flow blockages.194 Thus, we believe for the units we examined, the
adverse impacts of DSI, far outweigh the adverse impacts of SDA for much smaller reductions in
SO2. For long-term, highly efficient SO2 control, DSI cannot match the performance or cost
effectiveness of SDA
Comment: EPA failed to consider non-air environmental impacts [at Tolk]
[Xcel Energy (0064) p. 33]
Xcel Energy stated that the EPA's four-factor reasonable progress analysis requires that potential
controls be evaluated for their non-air environmental impacts. As discussed above, the
installation of dry scrubbers at Tolk would require a significant increase in the use of precious
water resources and would result in the landfilling of all of Tolk's fly ash. EPA failed to give
any consideration to the environmental impact of increasing water usage in this extremely arid
region. The Ogallala aquifer is the main source of water for all of the farmers, residences and
industrial activities in this area. The Ogallala aquifer is famously over-utilized and stressed.
Operating dry scrubbers at Tolk would increase its water demand by 9% - 12%. EPA gave no
consideration to the impact of requiring dry scrubbers on the extremely scarce water resources on
this part of West Texas.
Similarly, Xcel Energy stated that the EPA failed to weigh the non-air environmental impact of
requiring Tolk to begin landfilling the fly ash from its operations. Currently, Tolk beneficially
utilizes 100% of its fly ash. Installing dry scrubbers would require SPS to cease selling its fly
ash and, instead, construct extensive landfills to dispose of the scrubber residue. This is not just
a significant cost, but also an unnecessary burden on the land and would result in additional
environmental and management risks associated with such landfills.
Response: We considered both water supply and waste disposal in our four factor analysis for
Tolk. As to water, see our response to another comment. As to flyash, the issue is irrelevant as
we selected the dry scrubber option located downstream of the baghouse, costed to include a tail-
end, polishing baghouse. This eliminates contamination of the fly ash from the primary
baghouse and allows Tolk to continue to beneficially reuse its flyash. Thus, landfilling of flyash
is not required. As there is no impact to fly ash under our proposal, there is no obligation to
evaluate flyash contamination as a non-air environmental impact. See our response to a related
comment for more information.
194 Dry Sorbents and Systems and Material Handling in Coal-fired Power Plants, Hot Tope Hour, June 7 & June 21,
2012, Available at:
http://www.mcilvainecompanY.com/Decision Tree/subscriber/Tree/DescriptionTextLinks/Drv%20Sorbent%20Hot
%20Topic%20June%207.%202012.htm Ron Sahu, Dry Sorbent Injection (DS() and Its Applicability to TVA's
Shawnee Fossil Plant (SHF), April 2013, Available at; http://www.cleanenergy.org/wp-
content/uploads/Final_Sahu_DSI_Report.pdf.
267
-------
Comment: EPA's amortization timeframe [at Tolk] is inappropriate.
[Xcel Energy (0064) p. 33-34]
Xcel Energy stated that the EPA's amortization period assumption of 30 years for the required
controls at Tolk is inappropriate and leads to underestimated costs of reduction per ton of
emissions for these controls. Typically, in public utility commission ratemaking processes,
utilities seek and receive cost recovery for 20 years of equipment life for emissions controls.
Further, emissions retrofit equipment does not last for 30 years without substantial new
investments, often in the face of increasingly stringent emissions requirements that may require
upgrades. The 30-year amortization period also is a questionable assumption for emissions units
that will be approaching 40 years old by the time the Proposal requires the scrubbers in 2020.12
Finally, EPA has previously accepted a 20-year amortization period for the life of these types of
controls. EPA, Air Pollution Cost Control Manual, at 3-33 (2002) (stating amortization over 20-
30 years is appropriate); Wyoming Regional Haze FIP, 79 Fed. Reg. 5032, 5064-65 (Jan. 30,
2014) (using a 20-year amortization period for the Dave Johnson and Naughton plants); Arizona
FIP, 79 Fed. Reg. 52,240, 52,459 (Sept. 3, 2014) (assuming a 20-year amortization period);
Montana FIP, 77 Fed. Reg. 57,864, 57,882 (Sept. 18, 2012) (using a 20-year amortization
period).
Figure 6 provided by Xcel Energy below shows that a more appropriate 20-year amortization
period for the scrubbers would increase the cost per ton of emissions controlled by 9%. The
table below in Figure 6 utilizes the EPA estimates in the Cost TSD, but only considers the
amortization factor and does NOT include the additional impacts of the costs to obtain water or
to construct and operate a new landfill.
Cost Estimate for Tolk Units 1 and 2 with Revised Amortization and Costs (Figure 6
provided by Xcel Energy)
Tolk Unit 1
$3,178
$3,505
9% (
Tolk Unit 2
$3,310
!
Commenter's References:
12 Tolk Unit 1 will have been in operation for 38 years in 2020, and Tolk Unit 2 for 35 years
Response: The lifetime used to estimate the capital recovery factor in a cost effectiveness
analysis is a site-specific determination, based on the design of the proposed control equipment
and the remaining useful life of the facility in which it is installed. Thus, our decisions in other
cases do not establish the lifetime of scrubbers at Tolk. We address the comment alleging our
use of a 20 year life in the consistency section of this document.
268
-------
We have also used a 30 year scrubber lifetime in other cases, including the Oklahoma FIP.195 As
explained in the FIP TSD for Texas, "We have used this 30 year lifetime approach in prior
actions and we therefore adopted the same scrubber lifetime in our present analysis. See 76 FR
52388 (Aug 22, 2011); 76 FR 81728 (Dec. 28, 2011); Oklahoma v. EPA, 723 F.3d 1201 (July
19, 2013), cert, denied (U.S. May 27, 2014)."
Scrubbers typically have a lifetime even longer than the 30 years assumed in EPA's analysis.
S&L, for example, estimated the lifetime of dry scrubbers retrofit to White Bluff Units 1 and 2 of
40 years.196
In a retrofit application, as here, the lifetime assumed in the cost analysis may be constrained by
the remaining useful life of the facility. If the remaining useful life of a facility clearly exceeds
the lifetime of the control, the remaining useful life has no effect on control costs and on the
BART determination process. However, where the remaining useful life is less than the lifetime
of the control, e.g., scrubbers, the shorter time period must be used in the cost calculations.197
Control cost studies prepared by two consultants to Xcel Energy, Burns & McDonnell and URS,
selected a site-specific lifetime for a dry scrubber installed on the Tolk units of 30 years. In
2011, Burns & McDonnell selected a lifetime of 30 years for each of the Tolk units.198 In
December 2012, URS, in a subsequent study for Xcel Energy, assumed a remaining plant life of
30 years199 in costing a dry scrubber and other control equipment.
Elsewhere, in proceedings before the Texas Public Utility Commission, SPS reported retirement
dates for these units as 2042 for Unit 1 and 2045 for Unit 2, as of June 30, 2013200 and remaining
useful lives, as of June 30, 2013 of 29.5 years for Unit 1 and 32.5 years for Unit 2.201 It would be
typical in utility practice to extend these retirement dates as they approach, if the units remain
economic to continue to operate.202
195 EPA, Response to Technical Comments for Sections E. through H. of the Federal Register Notice for the
Oklahoma Regional Haze and Visibility Transport Federal Implementation Plan, Docket No. EPA-R06-OAR-2010-
0190, December 13, 2011.
196 Sept. 2002, Sargent & Lundy, Table 1-1.
197 BART Guidelines, 40 CFR 51, p. 662-663 ("Where the remaining useful life is less than the time period for
amortizing costs, you should use this shorter time period in your cost calculations.").
198 Nov. 2011 B&M Study, Table 8.6, TOLK 00001063.
199 URS, Final Report: SPS AQCS Cost Refinement Study - Tolk and Harrington Generating Plants, Xcel Energy,
December 12, 2013, TOLK 0000097 - 208 (12/12/13 URS Report), pdf 16, Table 4, TOLKOOOOOH2 and
TOLK 0000146.
200 SOAH Docket No. 473-14-1665, Docket No. 42004, Southwestern Public Service Company's Response to
Sierra Club's First Request for Information, Question Nos. 1-1 through 1-49, pdf 28, Available at:
http://www.xcelenergv.eom/staticfiles/xe/Regulatorv/Regulatorv%20PDFs/TX-Rate-Case-
2014/Revenue%20Reauirment%20Testimonv/Dane-Watson-Direct-Testimonv.pdf. based on Direct Testimony of
Dane A. Watson on behalf of Southwestern Public Service Company, Docket No. 42004, Public Utility
Commission of Texas, Appendices C and D, Available at:
http://interchange.puc.state.tx.us/WebApp/Interchange/Documents/42004 333 785057.PDF.
201 Direct Testimony of Dane A. Watson on behalf of Southwestern Public Service Company, Docket No. 42004,
Public Utility Commission of Texas, Appendix D, pdf 50, Available at:
http://www.xcelenergy.com/staticfiles/xe/Regulatory/Regulatory%20PDFs/TX-Rate-Case-
2014/Revenue%20Requirment%20Testimony/Dane-Watson-Direct-Testimony.pdf.
202 See, e.g., Davidson Testimony for SPS, p. 46.
269
-------
Thus, the 30 year lifetime that we used in our cost analysis actually corresponds to the remaining
useful life of these facilities. We conclude that our original determination that the proposed
scrubbers for the two Tolk units would have at least a 30 year service life remains correct and
fully supported by SPS's fillings in other proceedings. If Xcel wishes to use a shorter lifetime,
taking into account the date the new scrubbers would come on line, it must accept an enforceable
shutdown date. As no such date has been proposed, and this change would not affect the cost
effectiveness of the scrubbers, no change is required.
Comment: Limestone scrubber upgrades would not be cost-effective. [NRG (0078) p. 13]
NRG stated that, in addition to being minuscule, the visibility improvements that EPA proposes
to require with scrubber upgrades at Limestone 1 and 2 would not be cost-effective. As
documented in the attached confidential report by Sargent & Lundy, EPA's cost-effectiveness
analysis greatly understates the annual capital and operating cost of the scrubber upgrades that
EPA is proposing to mandate. The estimated actual cost-effectiveness, based on a preliminary
analysis, is projected to be $2,579 per ton, or more than 16 times that projected by EPA.
[Sargent & Lundy Report at 17.] The real costs, which would need to be based on a more
extensive engineering analysis, could be far higher. Accordingly, EPA's cost-effectiveness
analysis is dramatically overstated as to Limestone, making the cost per deciview of visibility
improvement unreasonable.
Response: The items that NRG summarizes in this comment, and the information provided in
its attached Sargent and Lundy (S&L) report, regarding our scrubber upgrade costs are detailed
in a separate comment package submitted under the Confidential Business Information (CBI)
provisions of 40 C.F.R.§ 2.203(b). Within those CBI comments, S&L also provides its own cost
analyses for upgrading NRG's scrubbers. We are unable to respond to many of these
summarized scrubber upgrade comments here with any specificity, because doing so would
involve citing and discussing in detail items that NRG has claimed as CBI. Accordingly, many
of our responses to the scrubber upgrade comments are contained within a separate document
that is not a part of our posted docket, but will be available for review by NRG. The responses to
comments that do not contain CBI information are contained within this document.
With regard to our scrubber upgrade cost analysis, we generally disagree with NRG that our
analysis was flawed. We used NRG's own information, backed by independent contractors hired
by it, supplied by NRG in response to our Section 114 requests for information. This included
cost estimates from well-known and respected contracting firms with a history of many scrubber
upgrades. In any event, criticisms regarding our use of this information are moot, because S&L
has provided its own cost analysis (under the CBI protections), which it offers as a replacement
to our own cost analyses. We have reviewed the scrubber upgrade cost analyses performed by
S&L and adopted its methodology. However, we noted many errors and undocumented cost
figures in its analyses. We corrected these errors and rejected some of the undocumented
assertions and/or costs in S&L's cost analyses. Nevertheless, in order to produce a conservative
scrubber upgrade cost analysis and set many of the issues that NRG raises aside, we incorporated
many of NRG's cost items. The resulting costs for NRG's scrubber upgrades increased slightly,
270
-------
resulting in a range of $368/ton to 910/ton for all the scrubber upgrades, well within a range that
we believe is cost effective, given the significant visibility benefits that will result from the
installation of those controls.
Comment: Controls should not be required for the Parish plant. [NRG (0078) p. 13]
NRG supported EPA's proposed exclusion of the Parish plant from new SO2 limits under this
proposal for similar reasons as apply to Limestone:
Even if new scrubbers were installed at three units and the existing scrubber upgraded
at the Parish plant, these controls would only achieve a 0.361 deciview improvement
as compared to average natural conditions far less than the eye can detect. (79 Fed.
Reg. at 74,881 & Tbl. 34.)
The projected visibility improvement from new and upgraded SO2 controls at Parish
under EPA's projected 2018 emissions inventory would be significantly lower, at
0.071 deciview.
Actual costs of increasing SO2 control at the Parish plant are significantly higher than
EPA's estimates. While NRG did not commission an engineering analysis similar to
that performed on the Limestone scrubbers, we note that EPA's cost effectiveness
analysis for W. A. Parish was incomplete and omitted necessary capital and
operating and maintenance costs.
Thus, NRG disagreed that new and upgraded SO2 controls at Parish would be cost-effective.
Response: We agree with NRG that additional controls at the Parish facility are not appropriate
during this planning period, but Limestone warrants controls this planning period on the basis of
the technical record. That record, predicated on visibility benefits and cost effectiveness, is
distinct and distinguishable from that of the Parish plant. We urge the State of Texas to re-
examine the Parish facility with the submission of its next regional haze SIP.
Comment: Remaining Useful Life [GCLC (0063) p. 16]
GCLC stated that the EPA attempts to claim that the remaining useful life of the units indicates
that the costs of controls will not be significant, because the costs can be distributed over the
remaining life of the targeted EGUs. However, this is seriously misguided. Particularly for units
that EPA anticipates will have to have scrubber retrofits, there is a very high probability that the
unit will cease to operate. While there may be decades of useful life remaining, it would be
extremely difficult in Texas' competitive electricity market to justify the installation of controls
nearing $300 million for each unit; and those costs are based on EPA's estimates, when in reality,
the costs will likely be much higher.
Response: As we state in our responses to other comments, we believe a 30 year operational life
271
-------
is appropriate for our proposed scrubber upgrades and retrofits. Because none of the facilities
affected by our proposal have entered or offered to enter into an enforceable commitment to
guarantee a shorter operational life, we consider it appropriate to assume 30 years in our cost
analyses. The claim that any units would cease to operate has not been factually substantiated
with information provided by any commenter. See our response to similar issues concerning
grid reliability elsewhere.
Comment: EPA Assumed Too High of an Interest Rate in Determining the Annualized
Capital Costs of SO2 Controls Evaluated. [Earthjustice (0067) p. 37, Stamper (0068) p. 13]
Earthjustice et al., and Stamper noted that the EPA assumed a 7% social interest rate in
determining the annualized capital costs of SO2 controls evaluated. Cost TSD at 10. While the
Office of Management and Budget ("OMB") has, in the past, recommended a 7% social interest
rate, OMB's current recommendation is 1.4% for a 30-year period. Moreover, the Control Cost
Manual recommends that a source-specific interest rate be used for reasonable progress
determinations, rather than the social interest rate applied in promulgation of regulations.
Stamper stated that the EPA did not provide much justification for the 7% interest rate, other
than to state that a 3% interest rate and a 7% interest rate were used in determining cost
effectiveness for the BART Guidelines, and that a 7% interest rate is recommended by the Office
of Management and Budget. Cost TSD at 10.
Stamped noted that to determine the annualized capital costs of a control technology, the total
capital costs are multiplied by the cost recovery factor, which is determined based on an assumed
life of the pollution control equipment and an assumed interest rate. Specifically, the capital
recovery factor (CRF) is based on the following equation:
CRF = [i(l+i)n]/[(l+i)n - 1]
where "i" is the interest rate and "n" is the life of the pollution control equipment. In essence,
annualization establishes an annual payment sufficient to finance the capital investment for its
entire life.36
Stamper stated that the EPA cites to an out-of-date OMB circular for justifying a 7% interest rate
in its Cost TSD. The OMB circular cited to by EPA is dated September 17, 2003, and thus is
from almost 12 years ago. Further, the 2003 OMB circular indicates that a 7% interest rate is a
default interest rate, but that the real cost of capital or the "social" interest rate has averaged
around 3% in real terms on a pre-tax basis. Consequently, this 2003 OMB circular cited by EPA
recommends that, for regulatory analyses, one should utilize both a 3% and a 7% interest rate.
However, the docket for EPA's proposed reasonable progress controls only includes cost
effectiveness analyses based on Capital Recovery Factors determined with a 7% interest rate.
Stamper explained that the 7% "social interest rate" is used to estimate the cost to society of
taking an action.38 However, the Control Cost Manual states that this social interest rate "is
probably not appropriate for industry." 37 39 The Regional Haze Rule requires EPA to make case-
272
-------
by-case determinations of "the costs of compliance" for identified BART and reasonable
progress options, 42 U.S.C. § 7491(g)(1), (2)—i.e., the actual cost to the source of
implementing the studied alternatives. In such situations, where EPA, the state, or industry is
evaluating "the economic impact that [air pollution-control] equipment would have upon the
source," a source-specific interest rate is appropriate.40
Stamper noted that OMB updates interest rates yearly and the current social interest rate is 1.4%
for a 30-year period.41Therefore, even if EPA were correct in applying the social interest rate, it
should have used the current published OMB rate in accordance with the Control Cost Manual.
While the 7% interest rate assumed by EPA is much higher than any of these EGU owners will
be subject to, Stamper did not revise EPA's cost analyses to use these lower "real cost of capital"
interest rates or the current social interest rate. A consistently-used interest rate among the Texas
units at least provides for apples-to-apples cost effectiveness comparisons in this proceeding.
However, EPA should make clear that the annualized cost projections for the controls it has
evaluated do not reflect the true annual costs of these controls at the Texas EGUs. Use of more
realistic interest rates would significantly lower the annualized costs and cost effectiveness (i.e.,
make the controls more cost effective) for the controls evaluated by EPA.
Footnotes:
36 EPA Control Cost Manual, January 2002, Section 1, at 2-21.
37 See EPA's Control Cost Manual at 2-12 to 2-13.
38 Id. at 2-12 to 2-13.
39 Id. at 2-13.
40 Id.
41 See OMB Circular A-94, App. C (revised January 21, 2015) (Ex. 13).
Response: We agree with the commenter that our selection of an interest rate has a significant
impact on our cost effectiveness calculations, and that were we to adopt a 3% (or lower) interest
rate, the cost effectiveness of our proposed controls would become much more attractive (lower
$/ton). As we have noted in our responses to other comments, we believe that our proposed cost
analyses calculations are already cost effective, especially the scrubber upgrades. We have
retained the use of a 7-percent interest rate in calculating the capital recovery factor. For cost
analyses related to government regulations, an appropriate "social" interest (discount) rate
should be used, not the source's actual rate of borrowing. OMB Circular A-4, providing Federal
agencies guidance on developing regulatory analyses, and dated September 17, 2003,203
reiterates the guidance found in the earlier Circular A-94:204
As a default position, OMB Circular A-94 states that a real discount rate of 7
percent should be used as a base-case for regulatory analysis. The 7 percent rate
is an estimate of the average before-tax rate of return to private capital in the U.S.
economy, based on historical data. It is a broad measure that reflects the returns
to real estate and small business capital as well as corporate capital. It
approximates the opportunity cost of capital, and it is the appropriate discount rate
whenever the main effect of a regulation is to displace or alter the use of capital in
203 https://www.whitehouse.gov/sites/default/files/omb/assets/omb/circulars/a004/a-4.pdf
204 https://www.whitehouse.gov/sites/default/files/omb/assets/a94/a094.pdf
273
-------
the private sector.
The yearly updates to OMB Circular A-94 to which the commenter references, are to Appendix
C. However, as the letter that transmits this yearly updates explains:205 "The rates presented in
Appendix C do not apply to regulatory analysis or benefit-cost analysis of public investment.
They are to be used for lease-purchase and cost-effectiveness analysis, as specified in the
Circular." Note that the reference to "cost-effectiveness analysis" here does not refer to that used
in the regional haze program, but rather to the analysis of programs requiring the expenditure of
federal funds.
Comment: Remaining Useful Life (30-yr equipment life) [Stamper (0068) p.9]
Stamper stated that the EPA used a 30-year life for the pollution controls that it evaluated in its
SO2 control cost analyses. FIP TSD at 8-9. There is ample support for the assumption of a 30-
year life for a scrubber or a DSI system. There are numerous EGUs with wet scrubbers and
SDAs that have been in operation for at least 30 years.23 Many of the scrubbers in operation
today were installed in the late 1970's to early 1980's24, and thus have been in operation more
than 30 years. In addition to the BART determinations that EPA cited in which EPA assumed a
30-year life of the pollution controls being evaluated, Sargent & Lundy assumed a 40-year life
for retrofit SO2 scrubbers on the White Bluff facility in a BART analysis for that facility.25
Moreover, Stamper agreed that, if an EGU owner indicates that a shorter life of the SO2 controls
should be evaluated in EPA's cost effectiveness determination due to a planned shutdown of a
unit, then that shorter lifetime needs to be made into an enforceable requirement. This is
consistent with how EPA has considered shorter equipment lifetimes in cost analyses for BART
determinations, and there is no justification for a different approach for cost effectiveness
analyses done for reasonable progress requirements.
For example, Stamper noted that the owners of Dave Johnston Unit 3 in Wyoming informed
EPA that the unit would be shut down in 2027. 79 Fed.Reg. 5045 (January 30, 2014).
Consequently, EPA considered a shortened lifetime for the NOx controls it was evaluating for
the unit to meet BART. Although EPA had initially found that selective catalytic reduction
(SCR) was cost effective for Dave Johnston Unit 3, when EPA evaluated selective catalytic
reduction based on a remaining useful life of the unit ending in 2027, EPA found that SCR was
no longer cost effective.26 Id. Based on these analyses, EPA imposed two requirements, either
one of which could be met by the Dave Johnston Unit 3 owners: 1) that Dave Johnston Unit 3
meet a NOx limit of 0.07 lb/MMBtu based on SCR installation by March 4, 2019, or 2) that the
owners permanently cease operation of the unit on or before December 31, 2027. 79 Fed.Reg.
5221 (January 30, 2014). In other words, EPA only took into consideration a reduced remaining
useful life in its NOx BART cost effectiveness analysis when there was an enforceable limitation
on the unit's remaining useful life.
In summary, Stamper concluded that the EPA's assumption of a 30-year life for the SO2 controls
evaluated is based on real life scrubber operating experience and is fully supported. Further,
205 https://www.whitehouse.gov/sites/default/files/omb/memoranda/2015/m-15-05.pdf
274
-------
EPA is justified in requiring a federally enforceable restriction for a remaining useful life of less
than 30 years in its cost effectiveness evaluations, as is required by the BART Guidelines. FIP
TSD at 9.
Footnotes:
23 See Burns & McDonnell, Utility FGD Design Trends, which provides, among other things, the year each FGD
system at an EGU began operation. (Ex. 67).
24 Id.
25 Sargent & Lundy, White Bluff Station Units 1 and 2, Evaluation of Wet vs. Dry FGD Technologies, Prepared for
Entergy Arkansas, Inc., Rev. 3, October 28, 2008. (Ex. 20).
26 With a shorter period of time over which the capital expenditure of a pollution control is amortized, the
annualized costs and the overall cost effectiveness numbers will be higher.
Response: We agree with the commenter that a 30 year operational life is appropriate for our
cost analyses.
Comment: EPA Assumed Default Auxiliary Power Costs Rather than Site-Specific
Auxiliary Power Costs in its Cost Effectiveness Analyses.
[Earthjustice (0067) p. 37; Stamper p. 15]
Earthjustice et al., stated that EPA assumed higher default auxiliary power costs rather than site-
specific auxiliary power costs in its cost effectiveness analysis. And EPA's assumed level of
SO2 control in its cost effectiveness analyses for SO2 for wet FGD and SDA systems is likely
conservative and does not reflect the full extent of SO2 emission reductions that can be achieved
with these control technologies. That means EPA's cost effectiveness calculations should be
lower and the visibility improvements higher for several of the EGUs evaluated by EPA for
reasonable progress controls.
Stamper noted that the Sargent & Lundy IPM cost modules assume a cost of auxiliary power of
$0.06/kW.42 Although EPA acknowledged that the true auxiliary power cost for "most if not all
of the units" EPA analyzed is "considerably less" than this value, EPA conservatively used the
default value of $0.06/kWh in the cost evaluations. Cost TSD at 10. EPA indicated that it has
the actual cost of auxiliary power in the information obtained via its Section 114(a) requests
made to each company. Id. Given that EPA has site-specific auxiliary power cost information
for each EGU, EPA should use such actual auxiliary power cost in its SO2 control cost
effectiveness calculations.
Response: We agree with the commenter that our auxiliary power cost assumptions are likely
high and thus worsen our cost effectiveness calculations (higher $/ton). One of our intentions in
performing our cost analyses was to conservatively estimate many of the individual cost
parameters and demonstrate that even doing this, our proposed scrubber upgrade and scrubber
retrofit cost analyses were cost effective. We believe we have met that goal.
Comment: Energy and non-air quality environmental impacts. EPA should have
considered blending with low-sulfur coal, lignite drying, and a circulating dry scrubber
called Novel Integrated Desulfurization (NID™) technology. [Earthjustice (0067) p.36,
275
-------
Stamper (0068) p.8]
Stamper noted that the second factor in establishing controls measures and reasonable progress
goals for a Class I area is to consider the energy and non-air quality environmental impacts of
compliance with such control measures. EPA included in its cost effectiveness analyses the
costs of auxiliary power required to operate the SO2 controls evaluated (i.e., wet scrubber, spray
dryer absorber (SDA) dry scrubbers, and dry sorbent injection (DSI)) and also the costs of waste
disposal associated with these SO2 controls. FIP TSD at 7-8.
Earthjustice et al., and Stamper agreed that including the energy and waste impacts in the cost
analyses is an appropriate way to evaluate these impacts in the analysis of reasonable progress
controls, but they stated that there are additional factors that EPA should consider with respect to
the energy and non-air quality environmental impacts of SO2 scrubbers and DSI technology.
Specifically, Earthjustice et al., and Stamper stated that the EPA should have evaluated the
blending with lower sulfur coal for those EGUs that burn high sulfur coal, and/or EPA should
have considered the use of coal drying for the units that burn lignite coal. Blending with lower
sulfur coal, which essentially is a pollution prevention control, would decrease the power needs
of an SO2 scrubber and waste products of the SO2 controls due to lower concentrations of SO2 to
remove. Stamper explained that based on publicly available information, the costs of such
controls are admittedly difficult to calculate, because boilers are typically designed for a certain
type of coal and it is often unclear the impacts that coal blending will have on a unit without
access to unit-specific information or testing various coals. However, if feasible for a unit, coal
blending provides a viable option to not only lower the operational costs of SO2 controls (due to
less SO2 to be removed) but also decrease the unit's energy needs and waste generated.
Similarly, Stamper explained that lignite drying is a viable option to decrease the power and non-
air quality impacts of SO2 scrubber upgrades and scrubber retrofits. The Coal Creek power plant
which burns North Dakota lignite has been using low grade heat rejected from the steam
condenser and waste heat from the flue gas to evaporate some of the moisture in the lignite coal
in a fluidized bed dryer. Lignite coal is very high in moisture content.17 The reduction in
moisture content of the Coal Creek coal improves the heating value of the coal, which means
fewer tons of coal need to be burned to achieve the same amount of heat input.18 Further, the
decrease in lignite moisture content results in a higher boiler efficiency and lower flue gas
volume, which results in increased SO2 removal efficiency.19 Moreover, the lignite drying
system reduced SO2, NOx, carbon dioxide, and mercury emissions, reduced station power
consumption by 18%, and reduced water use by 2.5%.20 Thus, Stamper and Earthjustice et al.,
concluded the use of a lignite drying system in combination with a scrubber could reduce overall
energy needs and scrubber waste generated, as well as reduce the operational costs and possibly
even the capital costs of a scrubber.
Stamper and Earthjustice et al., stated that another control measure that EPA should have
evaluated to decrease energy and non-air quality impacts is a circulating dry scrubber. EPA
limited its evaluation of dry scrubbers to spray dryer absorbers (SDAs), which do use less water
than wet scrubbers but also typically do not achieve as high of SO2 removal rates as wet
scrubbers. EPA did not give any consideration to circulating dry scrubbers, which can achieve
276
-------
higher SO2 removal efficiencies (as high as 98%), with lower water use and waste. Specifically,
Stamper stated that a type of circulating dry scrubber called Novel Integrated Desulfurization
technology (NID™), because a baghouse is an integral part of these scrubbers, can achieve very
high SO2 removal efficiencies, as well as high PM removal, at similar costs to dry scrubbers.
According to Alstom, some of the benefits of a NID™ system are: low capital investment and
maintenance costs, low power consumption, high SO2 removal and high removal efficiencies of
S03 and hazardous air pollutants.21 Further, a NID™ system typically has lower water
consumption and lower reagent use, and thus is best for future National Pollution Discharge
Elimination System (NPDES) permit compliance.22
Stamper stated that the EPA should have considered these control methods as viable options to
reduce auxiliary power needs and/or reduce non-air quality impacts of SO2 controls. However,
irrespective of these additional considerations, Earthjustice et al., and Stamper stated that the
EPA appropriately addressed the costs of energy usage and of waste disposal in its cost
evaluation of scrubber upgrades. Further, as stated by EPA, the SO2 scrubber technologies
evaluated by EPA are widely used by coal-fired EGUs, and these SO2 controls do not pose any
unusual energy and non-air quality impacts. FIP TSD at 8.
Footnotes:
17 See, e.g., EPA AP-42, Chapter 1.7 Lignite Combustion.
1 x See December 12, 2007, Coal Creek Station Units 1 and 2, Best Available Retrofit Technology Analysis, at iv, 31
(Ex. 50).
19Id. at iv.
20 Id. at 31.
21 See Alstom Brochure, NID™ Flue Gas Desulfurization System for the Power Industry at 3-4 (Ex. 28).
22 See February 8, 2012 Direct Testimony of Christian T. Beam on behalf of Southwestern Electric Power
Company, In the Matter of Southwestern Electric Power Company's Petition for a Declaratory Order Finding that
Installation of Environmental Controls at the Flint Creek Power Plant is in the Public Interest, Before the Arkansas
Public Utilities Commission, Docket 12-008-U, at 19-21. (Ex. 29).
Response: First, this comment asserts that we should have evaluated blending with lower sulfur
coals for EGUs that burn high sulfur coal. However, we cannot dictate to facilities which coal
they should burn as a control measure.
Second, we disagree with Earthjustice that consideration of coal drying falls within the purview
of "the energy and non-air quality environmental impacts of compliance," under Section
51.308(d)(l)(i), which is within the reasonable progress portion of the Regional Haze Rule that
governs the subject cost analysis. Our Reasonable Progress Guidance considers this factor from
the standpoint of whether it would cause the reconsideration of the control in question due to
energy or non-air quality penalties.206 In other words, the energy and non-air quality
environmental impacts of compliance serves as a potential modifier to the particular control
being considered. Thus, in this context, coal drying is not in and of itself something that should
be considered from an energy and non-air quality standpoint. Rather it is an option for
improving the power plant's efficiency, which also offers modest improvements in the control of
a number of pollutants, including SO2. We agree that in some circumstances coal drying can be
a viable technology for improving boiler efficiency, and in the process, reduce emissions because
206 Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program, page 5-2.
277
-------
less coal is burned to achieve the same heat input to the boiler. However, we are not required to
consider every available technology under the reasonable progress and long-term strategy
provisions of the Regional Haze Rule, which applies to the analysis in question. We considered
both wet and dry scrubbing, and the next most promising SO2 removal control, DSI. We rejected
DSI as not being cost effective in comparison to both types of scrubbers. Were we to have
considered coal drying, it would have ranked below DSI in its ability to remove SO2. We agree
with Earthjustice that we appropriately addressed the costs of energy usage and of waste disposal
in our evaluation of scrubber technologies. We discuss Earthjustice's assertion that we should
have considered circulating dry scrubbers in our response to another comment.
Comment: Commenter 0054-45 noted that scrubbers take toxins out of power plant
smokestacks, but scrubber waste is dumped in the water or accumulate in huge piles such that
toxic scrubber waste will sooner or later be dispersed to poison air, water, and soil. The
commenter noted that the SWEPCO AEP Turk plant near Texarkana dumps its scrubber waste
into the Little River, which goes down to the Red and the Mississippi and the Gulf of Mexico to
poison oceans.
Response: We agree with the commenter that the burning of coal and the disposal of its waste
products, including waste products from the use of some types of scrubbers, can pose a variety of
environmental issues or hazards. Because our proposal seeks to address only one aftereffect of
coal combustion—regional haze—we are limited here in addressing only those comments that
relate to regional haze. We note that our cost evaluation for scrubbers does include the
consideration of costs for proper disposal or reuse of scrubber waste.
Comment: EPA's Cost Effectiveness Analyses for SO2 Scrubber/DSI Retrofits: Big Brown
Units 1 and 2, Monticello Units 1 and 2, Coleto Creek Unit 1, Welsh Units 1, 2, and 3, and
Parish Units 5, 6, and 7. [Stamper (0068) p. 21]
Stamper stated that the Big Brown power plant was shown to significantly contribute to visibility
impairment at several Class I areas, with its largest impacts being modeled at the Wichita
Mountains (contributing 1.590%). FIP TSD at A-51. The Big Brown power plant was also
shown to significantly contribute to visibility impairment at Caney Creek (contributing 0.626%),
Guadalupe Mountains (contributing 0.502%), Hercules Glades (contributing 0.363%), Big Bend
(contributing 0.435%), and its contribution to impairment at Salt Creek was very close to EPA's
0.3%) threshold.81 Big Brown was the second highest contributor to visibility impairment at
Wichita Mountains based on the modeling results presented by EPA.82 EPA adjusted the
modeled impacts at Wichita Mountains, Guadalupe Mountains, and Big Bend to reflect 2008-
2012 average emissions, which increased the Big Brown facility's contribution to visibility
impairment at these Class I areas to 2.060%, 0.651%>, and 0.564%, respectively.83 FIP TSD at A-
51. According to the data in EPA's Clean Air Markets Database, the Big Brown Units 1 and 2
were the largest emitters of SO2 of all of the EGUs in Texas based on 2013 emissions data.84
Despite this, neither Big Brown Unit 1 nor Unit 2 have SO2 controls. Thus, EPA has ample
justification for evaluation of the Big Brown plant for pollution controls.
278
-------
Stamper stated that the EPA presents the results of its SO2 control cost effectiveness evaluation
for Big Brown Units 1 and 2 in its FIP TSD at 14 (Table 4). For the reasons discussed in Section
I.A.l.d) (1) above, we have revised EPA's analyses to be based on a 5-year annual average
emissions baseline, 5-year annual average SO2 rate in lb/MMBtu, and 5-year average gross heat
rate and MW-hrs generated, based on actual operating data from 2009 to 2013.
With respect to DSI, Stamper stated that it was not appropriate to evaluate DSI at 80%/90%
control for any EGU without adequate support for such a high SO2 removal efficiency and
without accounting for an increase in particulate matter emissions. Further, for Big Brown Units
1 and 2, EPA stated in its Cost TSD that it determined that DSI was not technically feasible for
the Big Brown units. Cost TSD at 8. This was based on Luminant's study of DSI at its units in
2011, which found that a very high feed rate was required to achieve "modest SO2 removal" and
that sorbent buildup harmed the performance of the control equipment (presumably this was
referring to the baghouses). Cost TSD at 8.
Although EPA found that DSI was not a feasible alternative for the Luminant facilities including
Big Brown, Stamper noted that the EPA nonetheless evaluated the cost effectiveness of DSI at
Big Brown (as well as Monticello) at both 50% and 90% SO2 removal. It appears questionable
that even 50% removal is achievable without adverse impacts at Big Brown Units 1 and 2, based
on Luminant's testing as discussed in the Cost TSD. However, for the purpose of completeness,
we also revised EPA's DSI cost effectiveness analyses for DSI in the same manner discussed
above (i.e., based on 5-year average baseline, SO2 rate, gross heat rate and MW-hours generated
per year), but we only evaluated a 50% removal efficiency.
Last, Stamper presented a fourth control option for the Big Brown units of a NID™ circulating
dry scrubber, based on the SDAIPM cost module but assuming an SO2 control efficiency of
98% or an SO2 limit of 0.04 lb/MMBtu, whichever is more stringent. The SDA IPM cost
module likely overstates the costs of a NID™ circulating dry scrubber by 1-2%.
279
-------
Tnbif 4. Revised SO2 C onriol C'o'_TX_Scitr:ei:_P.ev;:-ei_\"S_!.i,.i 2~_-01xkv' \~D
CDS Co;t IPM IX_Soiffce:_*"SJ«!.ii Z? 201? xkx. ^SDA_Co;!jP!a_Tr{_SeaTcei_?>-,-.»cm\'S lis
rjBlJxbs/ mi DSI_C1'JM sl:x 'j:Eute 54 S.s&wdr
:¦? 'Jus icport
5m HP JSD it 14,
Stamper stated that all of these costs are reasonable, with the top two technologies in terms of
SO2 emission reductions - i.e., wet FGD and NID™ circulating dry scrubber - being the most
cost effective. As the above table shows, using the 5-year average baseline, SO2 rates, and
projections for heat rate and MW-hours generated provides an average cost effectiveness that is
lower than projected by EPA.
Stamper stated that Monticello Units 1 and 2 had some of the most significant contributions to
visibility impairment at Class I areas in Oklahoma and Texas of all of the power plants analyzed
by EPA. Specifically, the modeling of Monticello Station showed that it contributes 1.734% to
visibility impairment at the Wichita Mountains Class I area, greater than any of the other 37
sources modeled. FIP TSD at A-51. EPA adjusted the modeled impacts at Wichita Mountains to
reflect 2008-2012 average emissions, which increased the Monticello facility's contribution to
visibility impairment 1.834%. FIP TSD at A-51. On a cumulative basis, the Monticello Station
also contributed a greater percentage to visibility impairment at the 18 Class I areas in total
compared to the other sources modeled by EPA.87 In addition to Wichita Mountains, the
Monticello Station was shown to contribute 1.759%) to visibility impairment at Caney Creek,
0.827%) at Hercules Glades, 0.450%> at Upper Buffalo, 0.261%> at Salt Creek, 0.244%> at
Guadalupe Mountains, and other Class I areas.88 The Monticello Units have no SO2 controls.
Thus, EPA had ample justification for evaluating SO2 controls for the Monticello units.
Stamper stated that the EPA presents the results of its SO2 control cost effectiveness evaluation
for Monticello Units 1 and 2 in its FIP TSD at 16 (Table 6). For the reasons discussed above, we
have revised EPA's analyses to be based on a 5-year annual average emissions baseline, 5-year
annual average SO2 rate in lb/MMBtu, and 5-year average gross heat rate and MW-hrs
generated, based on actual operating data from 2009 to 2013.
Stamper stated that the EPA found that DSI was not feasible for Monticello Units 1 and 2 based
on testing done by Luminant. Cost TSD at 8. Although EPA found that DSI was not a feasible
280
-------
alternative for the Luminant facilities including Monticello, EPA nonetheless evaluated the cost
effectiveness of DSI at Big Brown (as well as Monticello) at both 50% and 90% SO2 removal. It
appears questionable that even 50% removal is achievable without adverse impacts at Monticello
Units 1 and 2, based on Luminant's testing as discussed in the Cost TSD. However, for the
purpose of completeness, we also revised EPA's DSI cost effectiveness analyses for DSI in the
same manner discussed above (i.e., based on 5-year average baseline, SO2 rate, gross heat rate
and MW-hours generated per year), and we only evaluated a 50% removal efficiency.
Last, Stamper presented a fourth control option of a NID™ circulating dry scrubber, based on the
SDAIPM cost module but assuming an SO2 control efficiency of 98% or an SO2 limit of 0.04
lb/MMBtu, whichever is more stringent. The SDA IPM cost module likely overstates the costs
of a NID™ circulating dry scrubber by 1-2%.
Tallin 5, Revised SOI C ontiol C 0^1 Efft>crii enw Evaluation foi MotinceUo liufs I and 2W
Moimcello
Till!
so:
C omiol
Tuf.il
Amumlized
Cents
Tvir Mi:
Removed
Cost
Effectiveness
EPA's
Oi isnn.il (' ost
Effeetn
1
We? FGD
$:8J at Wichita Mountains and 0.473%> at Big Bend Class I area. FIP TSD at
A-51. In 2013 and 2014, the Coleto Creek Unit 1 had the 8th highest SO2 emissions of all of the
41 coal-fired EGUs that reported emissions to EPA's Clean Air Markets Database in 2013.91
Coleto Creek Unit 1 has no SO2 controls, and thus EPA evaluated SO2 retrofit controls to reduce
this facility's impacts on visibility impairment.
281
-------
Stamper stated that the EPA presents the results of its SO2 control cost effectiveness evaluation
for Coleto Creek Unit 1 in its FIP TSD at 18 (Table 8). For the reasons discussed above, we
have revised EPA's analyses to be based on a 5-year annual average emissions baseline, 5-year
annual average SO2 rate in lb/MMBtu, and 5-year average gross heat rate and MW-hrs
generated, based on actual operating data from 2009 to 2013. We only evaluated DSI at 50%
SO2 removal efficiency because, as stated above, additional site-specific test data should be
submitted before EPA considers DSI to be able to achieve 90% control, especially without
significantly increasing particulate matter. Last, we present a 4th control option of a NID™
circulating dry scrubber, based on the SDAIPM cost module but assuming an SO2 control
efficiency of 98% or an SO2 limit of 0.04 lb/MMBtu, whichever is more stringent. As discussed
in Section I. A. 1 .d) (6) above, the SDA IPM cost module likely overstates the costs of a NID™
circulating dry scrubber by 1-2%.
Table 0. RevjUed SG2 C onnol C ost Effecrn ?»«•<;<. E\.iltunon foi C oieto C i«-k Unit l'*
C oleto
C reek 1'nir
S02
C oun ol
Total
Annualized
C 05tS
Tons SO2
Removed
Cost
Effectiveness
EPA's
Oi igiiMi C o;t
Effectiveness9"1
i
Wee FGD
S30.7OI.P03
15 T26tpv
SLP5S ion
s: ton
1
NID™ CDS
530-4W.52P
15-726 tpy
SI 95"ton
I
SDA
S3 0.3 51.540
15 257tpv
Sl.PSS ton
$2,356 ton
1
A
DSI at 50%
SID 0S5 740
S.333 tpv
SI 930 ton
S3 "?2 ton
Stamper stated, as the above table shows, using the 5-year average baseline, SO2 rates, and
projections for heat rate and MW-hours generated provides an average cost effectiveness that is
lower than projected by EPA. With respect to DSI, the use of the 5-year average MW-hrs
generated and SO2 emission rate resulted in much lower operating costs, which are the bulk of
the costs associated with DSI, in comparison to EPA's analysis which was based on the
maximum year of generation and the maximum monthly uncontrolled SO2 rate. As will be
discussed further below, all of these costs are reasonable, with the top two technologies in terms
of SO2 emission reductions - i.e., wet FGD and NID™ circulating dry scrubber - providing the
greatest emission reductions.
Stamper noted that the EPA's modeling showed that the Welsh Power Plant contributed
significantly to visibility impairment at Caney Creek and at Wichita Mountains, with
contributions of 0.595%) and 0.475%), respectively.110 The Welsh Power Plant also was shown to
contribute 0.226% to Hercules Glades Class I area.111 EPA adjusted the contribution at Wichita
Mountains up to 0.862%> based on 2008-2012 average emissions. FIP TSD at A-51. Had EPA
adjusted the Caney Creek and Hercules Glades impacts based on 2008-2012 average emissions,
those impacts would have increased significantly as well. The Welsh Power Plant emitted
almost 20,000 tons of SO2 in 2013 based on information submitted to EPA's Clean Air Markets
Database.112 None of the Welsh units have SO2 pollution controls. Thus, EPA evaluated SO2
retrofit controls to reduce the Welsh Power Plant's contribution to visibility impairment.
Stamper stated that the EPA presents the results of its SO2 control cost effectiveness evaluation
for Welsh Units 1, 2 and 3 in its FIP TSD at 23 (Table 12). For the reasons discussed above, we
have revised EPA's analyses to be based on a 5-year annual average emissions baseline, 5-year
282
-------
annual average SO2 rate in lb/MMBtu, and 5-year average gross heat rate and MW-hrs
generated, based on actual operating data from 2009 to 2013. We only evaluated DSI at 50%
SO2 removal efficiency because, as stated above, additional site-specific test data should be
submitted before EPA considers DSI to be able to achieve 90% control, especially without
significantly increasing particulate matter. Last, we present a fourth control option of a NID™
circulating dry scrubber, based on the SDAIPM cost module but assuming an SO2 control
efficiency of 98% or an SO2 limit of 0.04 lb/MMBtu, whichever is more stringent. As discussed
in Section I.A.l.d) (6) above, the SDA IPM cost module likely overstates the costs of a NID™
circulating dry scrubber by 1-2%.
Table Revised SOI C ontiol C oq EffeefivMie-s; Evaluation foe W ehii I'mrs, 1,1 and 3m
Wehli Unit
so:
C outrol
Total
AtilHi.llilMl
Cost?
Toils SOI
Removed
C ost
Effotriwtitm
EPA <
Oi igiii.il C on
EffectiveiM*«114
3
Wet FOD
7.245 tpv
S3.459 ton
S3,5OS ton
1
NID™ C DS
s:> si$ :k>
7.245 tpv
S3.2 S 7, ton
—
1
SDA
$23,716 772
6 S°0 tpv
S3 442 ton
S3,4S„^i
Ui 1
3
mi
0 ]
P j
¦>
Wet FOD
$25,236,325
7,355 tpv
*3,431 Ton
$3,454 ion
NID™ C DS
S23.PP6.05S
7,355 tpv
S3,262 ton
SDA
S23.SP5.470
6,9^5 tpv
S3 416 toil.
S3.43S toa
">
DSI at 50%
^TiTW!#
4,03$ tpy
S3,]86 ton
S3,611 ton
y
Wet FGD
S2 5,5 50,640
7,744 tpy
$3,2°° ton
S3.570 toa
5
NID™ CDS
$24,331,623
7.744 tpv
$3,142 ton
—
%
SDA
< ->4 -n}
7 371 tpv
S3 2$7wo
S3.36S toa
5
DSI at 50° 0
$13,301 043
4 245 tpy
S3.155 ton
S3 6P'0 ton
Stamper concluded based on the above table that using the 5-year average baseline, SO2 rates,
and projections for heat rate and MW-hours generated provides an average cost effectiveness
that is lower than projected by EPA. With respect to DSI, the use of the 5-year average MW-hrs
generated and SO2 emission rate resulted in much lower operating costs, which are the bulk of
the costs associated with DSI, in comparison to EPA's analysis which was based on the
maximum year of generation and the maximum monthly uncontrolled SO2 rate. All of these
costs are reasonable.
Stamper noted that the modeling of the W.A. Parish Station showed that it contributed 0.291%
to visibility impairment at Wichita Mountains, 0.181%) to Caney Creek, 0.125% to Big Bend,
and 0.089%) to Guadalupe Mountains.115 EPA adjusted the modeled impacts up significantly to
0.881%) at Wichita Mountains, 0.559%> at Big Bend, and 0.268%> at Guadalupe Mountains. FIP
TSD at A-51. Had EPA adjusted the modeled impacts at Caney Creek based on 2008-2012
average emissions, the W.A. Parish Station's impacts at Caney Creek would likely have
exceeded EPA's 0.3%> threshold. On a plantwide basis, the W.A. Parish Station had the third
highest emissions of SO2 in 2013 of the 20 coal-fired power plants in Texas that report to EPA's
Clean Air Markets Database.116 W.A. Parish Units 5, 6, and 7 do not have any SO2 controls.
EPA thus has ample justification for the evaluation of SO2 controls to reduce W.A. Parish
Station's impacts on visibility.
283
-------
Stamper stated that the EPA presents the results of its SO2 control cost effectiveness evaluation
for W.A. Parish Units WAP5, WAP6, and WAP7 in its FIP TSD at 25 (Table 13). For the
reasons discussed above, we have revised EPA's analyses to be based on a 5-year annual average
emissions baseline, 5-year annual average SO2 rate in lb/MMBtu, and 5-year average gross heat
rate and MW-hrs generated, based on actual operating data from 2009 to 2013. We only
evaluated DSI at 50% SO2 removal efficiency because, as stated above, additional site-specific
test data should be submitted before EPA considers DSI to be able to achieve 90% control,
especially without significantly increasing particulate matter. Last, we present a fourth control
option of a NID™ circulating dry scrubber, based on the SDAIPM cost module but assuming an
SO2 control efficiency of 98% or an SO2 limit of 0.04 lb/MMBtu, whichever is more stringent.
As discussed in Section I.A.l.d) (6) above, the SDA IPM cost module likely overstates the costs
of a NID™ circulating dry scrubber by 1-2%.
X *1 >f t V '* * %• *w' m u « V* • k IH ft.
WAP?. WAP6. and WAP"11"
W.A.
Paiish
I nit
so:
€ ouiiol
Total
Annualized
Costs
Ton? S02
Removed
C ost
Effectiveness
EPA's
Original Cost
Effectiveness11*
WAP 5
Wef FGD
$30 137 671
15.414 tpv
$2.24? ton
$2,389 v
WAP 5
NED™ C DS
S2P.695.Slo
13.414 tpv
$2,214 ton
WAP 5
SDA
$29,571,390
12.975 tpv
$2,279 ton
Oil
WAP 5
DSI a? 50°0
S14.973,555
7.14S tpv
$2,095 ton
$2.55° v
WAPo
Wet FGD
$50 S8S.503
14.461 tpv
$2.1 3d ton
$2,334 v
WAP6
NID™ CDS
$30.SOP.329
14,461 tpv
$2.110 ton
——~
WAP6
SDA
$30,380,079
14.004 tpv
$2,169 ton
$2,401."ton
WAPo
DSI at 50° &
S15.770.760
7 6SS tpv
$2,051 ton.
$2,699 ti
WAP?
Wei FGD
f J
O.
OO
' 0%
O.
11.407 tpv
S3
O '
rf\
^ f
$2,542
WAP 7
$25,666,913
11 407 tpv
$2,250 ton
WAP 7
SDA
$25 565,365
11 051 tpv
$2,313 ton
$2,559 :¦
"WAP7
DSI at 50° 0
S12.700.S75
6.060 tpv
$2,096 ion
$2,805 t
Stamper concluded based on the above table that using the 5-year average baseline, SO2 rates,
and projections for heat rate and MW-hours generated provides an average cost effectiveness
that is lower than projected by EPA. With respect to DSI, the use of the 5-year average MW-hrs
generated and SO2 emission rate resulted in much lower operating costs, which are the bulk of
the costs associated with DSI, in comparison to EPA's analysis which was based on the
maximum year of generation and the maximum monthly uncontrolled SO2 rate. All of these
costs are reasonable.
Footnotes:
115 See TX116-07-_29_Source_selection_analysis_TX_RH-l-31-14.xlsx at "All Class I areas data" tab.
116 See spreadsheet with TX EGUs 2013 CAMD Data Ranked for SO2, Ex. 69.
117 See spreadsheets with filenames "Wet_FGD_Cost_IPM_TX_Sources_Revised_VS_Mar 27 2015.xlsx," "NID
CDS_Cost_IPM_TX_Sources_VS_Mar 27 2015 .xlsx," " SDA_Cost_IPM_TX_Sources_Revised_VS_Mar
27_2015.xlsx," and "DSI_Cost_IPM_TX_Sources_Revised_VS_Mar 27_2015.xlsx," at Exhibits 34, 35, 36, and 37
to this report.
284
-------
118 See FIP TSD at 26.
Footnotes:
110 See TX116-07-_29_Source_selection_analysis_TX_RH-l-31-14.xlsx at "All Class I areas data" tab.
111 Id.
112 See spreadsheet with TX EGUs 2013 CAMD Data Ranked for SO2, Ex. 69.
113 See spreadsheets with filenames "Wet_FGD_Cost_IPM_TX_Sources_Revised_VS_Mar 27 2015.xlsx," "NID
CDS_Cost_IPM_TX_Sources_VS_Mar 27 2015.xlsx," "SDA_Cost_IPM_TX_Sources_Revised_VS_Mar
27_2015.xlsx," and "DSI_Cost_IPM_TX_Sources_Revised_VS_Mar 27_2015.xlsx," at Exhibits 34, 35, 36, and 37
to this report.
114 See FIP TSD at 20.
Footnotes:
91 See spreadsheet with TX EGUs 2013 CAMD Data Ranked for SO2, Ex. 69.
92 See spreadsheets with filenames "Wet_FGD_Cost_IPM_TX_Sources_Revised_VS_Mar 27_2015.xlsx," "NID
CDS_Cost_IPM_TX_Sources_VS_Mar 27 2015.xlsx," "SDA_Cost_IPM_TX_Sources_Revised_VS_Mar
27_2015.xlsx," and "DSI_Cost_IPM_TX_Sources_Revised_VS_Mar 27_2015.xlsx," at Exhibits 34, 35, 36, and 37
to this report.
93 See FIP TSD at 14.
Footnotes:
87 See TX116-07-_29_Source_selection_analysis_TX_RH-l-3 l-14.xlsx at "All Class I areas data" tab.
88 Note that it appears that EPA did not adjust the visibility impacts results for these other Class I areas based on
2008 to 2012 emissions.
89See spreadsheets with filenames "Wet_FGD_Cost_IPM_TX_Sources_Revised_VS_Mar27_2015.xlsx," "NID
CDS_Cost_IPM_TX_Sources_VS_Mar 27_2015.xlsx," "SDA_Cost_IPM_TX_Sources_Revised_VS_Mar
27_2015.xlsx," and "DSI_Cost_IPM_TX_Sources_Revised_VS_Mar 27_2015.xlsx," at Exhibits 34, 35, 36, and 37
to this report.
Footnotes:
81 See TX116-07-_29_Source_selection_analysis_TX_RH-l-31-14.xlsx at "All Class I areas data" tab. Note that
EPA's Cost TSD cites to this spreadsheet as well as TX116-07-30_Source_selection_analysis_TX_RH-es-l-31-14
in its FIP TSD for its data on percent of extinction on the worst 20% days contributed by a source and unit (in
Tables A.4-3 and A.4-4). We do not find any significant differences between these two spreadsheets.
82 Id.
83 Note that it appears that EPA did not adjust the visibility impacts results for any other Class I area.
84 See spreadsheet with TX EGUs 2013 CAMD Data Ranked for SO2, Ex. 69.
Response: This comment supports our decision to regulate SO2 emissions from the Big Brown,
Monticello, Coleto Creek units and presents revised analyses that suggest the control of SO2
from these units is even more cost effective than our analyses indicate.
These revised analyses involve the use of a different baseline, the reanalysis of our DSI cost
effectiveness, and the appropriateness of NID. See our response to other comments regarding
these issues. With regard to the Big Brown units, we agree that DSI is not feasible at these units,
based on Luminant's test data.
The selection of a NID, rather than WFGD would not change our proposed SO2 limit of 0.04
lb/MMBtu. The non-air quality environmental impacts of a NID and WFGD are similar and do
not warrant eliminating either technology. We proposed that the units in question meet certain
SO2 emission limits, but we did not mandate a specific control technology in doing so.
285
-------
Consequently, any unit, including the ones discussed herein, may elect to use a NID to achieve
our proposed SO2 emissions with no adverse non-air quality impacts.
Comment: EPA's Cost Effectiveness Analyses for SO2 Scrubber/DSI Retrofits: Tolk Units
171B and 172B [Stamper (0068) p. 25]
Stamper stated that the EPA's modeling showed that the Tolk Station contributed significantly to
visibility impairment at Salt Creek Class I area, with a contribution of 1.012%.94 The Tolk
Station also was shown to contribute to visibility impairment in excess of EPA's 0.3%
contribution threshold at White Mountains (contributing 0.657%), Guadalupe Mountains
(contributing 0.646%)95, and Bandelier National Monument (contributing 0.368%). On a
cumulative basis, the Tolk Station had the 4th high impacts in total (combined contributions to
visibility impairment at all 18 Class I areas modeled) among all of the 37 sources modeled by
EPA.96 Neither of the Tolk units have SO2 controls and thus EPA properly evaluated SO2 retrofit
controls to reduce the Tolk Station's contribution to visibility impairment.
Stamper stated that the EPA presents the results of its SO2 control cost effectiveness evaluation
for Tolk Units 17IB and 172B in its FIP TSD at 20 (Table 10). EPA states in its FIP TSD that it
limited its control evaluation for Tolk to SDA and DSI. It was assumed by EPA that well water
is used at the Tolk plant rather than surface water, and EPA apparently assumed use of well
water would have adverse energy and non-air quality impacts for use of a wet FGD which uses
more water than an SDA. FIP TSD at 8. A review of data submitted to the Energy Information
Administration shows that the source of water at the Tolk power plant is indeed well water.
Stamper stated that while many EGUs that rely solely on well water use dry scrubbing to remove
SO2 rather than wet scrubbing, there are some EGUs that rely on well water and use wet FGDs
for SO2 control. For example, the Coronado power plant in Arizona relies on well water for its
water supply97, and it recently installed wet scrubbers at both of its units pursuant to a Consent
Decree with EPA.98
Stamper stated that there are many EGUs in arid areas that rely on scrubbers for SO2 control. In
fact, there are 38 EGUs (including the Coronado Plant) located in areas that receive lower annual
precipitation than Lamb County where the Tolk Station is located, and yet these facilities all
utilize wet scrubbers for SO2 control.99 The location of the Tolk Plant, in Lamb County, Texas,
gets approximately 18.9 inches of precipitation per year, which is 46% greater than the mean
amount of precipitation in the locations of the 38 non-Xcel boilers with wet FGD systems of 12.9
inches per year.100 Indeed, the county that the Tolk plant is located in receives greater
precipitation on an annual average basis than all but 2 of the 38 EGUs with wet scrubbers in arid
areas.101 Thus, the record for EPA's FIP must include more documentation to justify the outright
elimination of wet FGD as an SO2 control option for the Tolk units.
Stamper contended that if water availability and costs are justified as a concern for the Tolk
units, then a circulating dry scrubber, which can achieve 98% SO2 removal efficiencies similar to
a wet FGD system, may be the best choice for SO2 control at these units, since circulating dry
scrubbers use 60% less water than wet FGD systems. Stamper noted that SWEPCO selected a
286
-------
NID™ system for SO2 control due to the lowest water consumption, in addition to lowest capital
and operation and maintenance costs on a 30-year cumulative present worth basis, lowest
auxiliary power usage, lowest reagent usage, smallest footprint, among other benefits.102
Below Stamper provided a cost effectiveness analysis for a NID™ circulating dry scrubber, as
well as revised cost effectiveness analyses for wet FGD, SDA, and DSI at 50% control. For the
reasons discussed above, we have revised EPA's analyses to be based on a 5-year annual average
emissions baseline, 5-year annual average SO2 rate in lb/MMBtu, and 5-year average gross heat
rate and MW-hrs generated, based on actual operating data from 2009 to 2013. And we have
used the same assumptions for the circulating dry scrubber analysis. As discussed in Section
I.A. 1 .d) (6) above, the SDA IPM cost module likely overstates the costs of a NID™ circulating
dry scrubber by 1-2%. We only evaluated DSI at 50% SO2 removal efficiency because, as stated
above, additional site-specific test data should be submitted before EPA considers DSI to be able
to achieve 90% control, especially without significantly increasing particulate matter.
pujjlOJ
Tolk I uir
SOI
C ©into!
Toral
Annualized
Cosfs
Tou> S02
Removed
Cost
Effectiveness
EPA <
Origin,il C osi
Effectiveness1"
I TIB
Wet FGD
S27.242.107
0 4^2 tpv
S2.S70 ton
S3.2-04 ton
171B
NED™ CDS
$25,826 757
9.492 tpv
$2,72! ton
I "IB
SDA
S25.723.o48
9,125 fpv
S2.S19 ton
S3.17S ion
I'IB
DSI ,11 50%
$11 355.03?
5.122 tpv
S 2.221 toil
S3.0S4 ton
I72B
Wet FGD
$28.,<501 038
10.092 tpv
$2,554 ton
S3.010 ton
172B
NID™ CDS
S27.1S9.344
10.GP2 fpv
S2 504 foil
172B
SDA
$27.076.P36
9.694 tpv
S2.7P3 ton
$2.90S ton
I72B
DSI at 50°0
SI 2 166.SS2
5.445 tpv
S2.235 ton
S2 828 ton
Stamper noted, as the above table demonstrates, use of annual average baseline emissions, SO2
emission rates, and MW-hours in the cost effectiveness analyses significantly reduces the cost
effectiveness values compared to EPA's analysis. Moreover, the use of a NID™ circulating dry
scrubber would result in the greatest SO2 emission reductions for the most reasonable costs, and
this is likely the best choice for the Tolk units due to the lower water usage expected with a
circulating dry scrubber. All of these costs are reasonable, as will be discussed further below.
Stamper noted that the EPA considered the costs to purchase water rights for operation of
scrubbers in its SO2 BART cost evaluation for the Gerald Gentleman Station, and EPA found
that when such costs were added in, installation of scrubbers (either wet or dry) was still cost
effective. 77 Fed. Reg. 40,150, 40,162 (July 6, 2012). With respect to water availability for the
Tolk units, it should be noted that in 2004 Xcel Energy began using blowdown water from its
Plant X generation station to use in Tolk's cooling towers, which reduced the combined plants
water consumption by 180 million gallons per year.105 That water savings would equate to most
of the water needs for operation of a scrubber at one of the Tolk units.106 It is not clear if Xcel
maintains those pre-existing water rights or has lost them. EPA should investigate Xcel's
287
-------
existing water rights further before taking into account the cost to purchase additional water
rights for the operation of scrubbers at Tolk Station.
In an attempt to address concerns that may be raised regarding the need to purchase land for the
additional water needs of scrubbers at the Tolk Station, Stamper analyzed the costs to purchase
land for water rights for all of the makeup water needed for operation of wet FGD, SDA, and a
NID™ CDS at Tolk Units 17IB and 172B, and recalculated cost effective numbers from Table 7
above. We used the same assumptions that EPA used for this analysis for the Gerald Gentleman
plant. Specifically, we assumed 0.176227 acre-feet of water is available per acre of land, and we
used EPA's high end land cost estimate of $10,000 per acre. 77 Fed. Reg. at 40,162. Table 8 of
comment 0068 contains Stamper's updated cost effectiveness analysis with the costs to purchase
land for water rights for wet FGD, SDA, and a NID™ CDS at Tolk Units 171B and 172B.
Consideration of Costs to Obtain Water Rights to Operate SO2 Scrubbers at Tolk Units
171B and 172B107 (provided by Stamper, Table 8 to comment 0068)
Tolk Unit 171B
Tolk Unit 172B
Wet FGD
SDA
NID™ CDS
Wet FGD
SDA
NID™ CDS
Acre-feet of water
887
653
653
966
713
713
for scrubber per
yearl08
Acres of land
5,013
3,707
3,707
5,483
4,047
4,047
required
Total costs for land
$50,314,371
$37,073,747
$37,073,747
$54,826,239
$40,466,986
$40,466,986
(@$10,000 per
acre)
Annualized costs
$4,055,338
$2,988,144
$2,988,144
$4,418,995
$3,261,639
$3,261,639
of obtaining land
for water rightsl09
($/year)
Annualized cost of
$27,242,107
$25,723,648
$25,826,757
$28,601,938
$27,076,936
$27,189,344
SO2 control
($/year)
Total Annualized
$31,297,445
$28,711,792
$28,814,901
$33,020,933
$30,338,575
$30,450,983
Cost of SO2
control +
water rights
($/year)
Tons SO2 reduced
9,492
9,125
9,492
10,092
9,694
10,092
(tpy)
Cost Effectiveness
$3,297/ton
$3,146/ton
$3,036/ton
$3,272/ton
$3,130/ton
$3,017/ton
($/ton)
Stamper argued that all of these costs, which include the potential added costs to obtain land for
water rights for the scrubbers, are reasonable, as will be shown further below in this report. The
NID™ CDS scrubber is the most cost effective and has the highest SO2 removal for the lowest
water use.
Footnotes:
94See TX116-07-_29_Source_selection_analysis_TX_RH-l-31-14.xlsx at "All Class I areas data" tab.
95 Note that EPA adjusted these impacts at Guadalupe Mountains downwards to 0.620% based on 2008-2012
288
-------
average emissions. FIP TSD at A-51.
96 Id.
97EIA-860.
98 See "Salt River Project Agriculture Improvement and Power District Settlement," at
http://www2.epa.gov/enforcement/salt-river-project-agriculture-improvement-and-power-district-settlement.
99 See spreadsheet with data on EGUs with wet scrubbers in arid areas (Ex. 32). The data in this spreadsheet are
from EPA's Clean Air Markets Database, based on 2013 data, and from the National Climate Data Center (NCDC).
100 Id. at Summary tab.
101 Id.
102 See February 8, 2012 Direct Testimony of Christian T. Beam on behalf of Southwestern Electric Power
Company, In the Matter of Southwestern Electric Power Company's Petition for a Declaratory Order Finding that
Installation of Environmental Controls at the Flint Creek Power Plant is in the Public Interest, Before the Arkansas
Public Utilities Commission, Docket 12-008-U, at 19-21. (Ex. 29).
103 See spreadsheets with filenames "Wet_FGD_Cost_IPM_TX_Sources_Revised_VS_Mar 27 2015.xlsx," "NID
CDS_Cost_IPM_TX_Sources_VS_Mar 27 2015.xlsx," "SDA_Cost_IPM_TX_Sources_Revised_VS_Mar
27_2015.xlsx," and "DSI_Cost_IPM_TX_Sources_Revised_VS_Mar 27_2015.xlsx," at Exhibits 34, 35, 36, and 37
to this report.
104 See FIP TSD at 20.
105 See, e.g., Xcel Energy, We are energized, Texas and New Mexico, downloaded from
https://www.xcelenergy.eom/staticfiles/xe/Corporate/Corporate%20PDFs/SPS_Collateral_2011_CRR.pdf, attached
as Ex. 33. See also Xcel Energy, Tolk Generating Station, Environmental Highlights, at
http://www.xcelenergy.com/Company/Operations/Power_Generation_Stations/Tolk_Generating_Station.
106 For example, at 2008-2013 average generation rates, the makeup water needs for a NID™ circulating dry
scrubber based on EPA's IPM SDA cost module would be approximately 213 million gallons per year for Tolk Unit
171B. See Ex. 38, Tolk Costs for Water Rights Purchase for SO2 Scrubbers.xlsx.
107 See Tolk Costs for Water Rights Purchase for SO2 Scrubbers.xlsx (Ex. 38).
108 Based on Makeup Water Rate calculated in the IPM cost modules for wet FGD and SDA. See Exs. 34, 35, and
36 at cell E17 in the tabs for the Tolk units (tabs "T 171B" and "T 172B").
109 As with the SO2 control cost analysis, a cost recovery factor was applied based on a 7% interest rate and a 30
year life of the pollution controls.
Response: This comment supports our decision to regulate SO2 emissions from the Tolk units
and presents revised analyses that suggest the control of SO2 from these units is even more cost
effective than our analyses indicate. These comments also urge the use of more efficient SO2
controls.
First, this comment argues for the use of a wet FGD, which achieves 98% SO2 control,
compared to the 95% efficient dry scrubber we proposed. We eliminated wet FGD as it uses
more water than any other scrubber technology. See -our response to other comments on this
issue. This comment notes that other similar units use wet FGD and use well water and/or are
located in arid areas with less annual precipitation. However, the use of well water and the
amount of precipitation are not key factors in our decision to eliminate wet FGD at Tolk. The
Tolk units use groundwater from a portion of the Ogallala aquifer which is among the most
severely overdrafted in the High Plains Aquifer, i.e., more water is withdrawn, mostly for
irrigation, than is replaced by precipitation.207 This comment fails to point to another similarly
situated facility. The local overdraft situation warrants selecting the scrubber technology with
207 USGS, Water-Level and Storage Changes in the High Plains Aquifer, Predevelopment to 2013 and 2011-2013,
http://ne.water.usgs.gov/ogw/hpwlms/files/HPAa WLC pd 2013 SIR 2014 5218 pubs brief.pdf: Water
Encyclopedia, Ogallala Aquifer, http://www.waterencvclopedia.com/Oc-Po/Ogallala-Aquifer.html: JasonJ. Gurdak
and Cassia D. Roe, Recharge Rates and Chemistry Beneath Playas of the High Plains Aquifer - A Literature
Review and Synthesis, USGS Circular 1333, 2009 http://pubs.usgs.gov/circ/1333/pdf/C1333.pdf.
289
-------
the lowest water demand, which is the SDA. We recognized this situation in our proposal and
discuss it in our response to another comment.
Second, this comment argues that if water availability and costs are justified, then a circulating
dry scrubber (NID) should be chosen as it can achieve 98% SO2 control and uses 60% less water
than wet system. However, as we explain in response to another comment, there is currently not
adequate performance data to support basing a reasonable progress limit on the NID. Further, a
study-level analysis done by AEP for Flint Creek Unit 1 indicates that the SDA that we selected
uses less water than the NID or any other scrubber technology.
Third, this comment presents revised cost analyses that use a different baseline. See our
response to another comment on this issue.
Fourth, this comment presents an analysis of the cost to obtain additional water rights, based on
an analysis for Gerald Gentleman. The acquisition of water rights involves site-specific
considerations such as the local economy and the local cost of land and its beneficial uses. We
discuss this in detail in our response to a comment from Xcel.
Comment: The Costs of Scrubber Retrofits at All of the EGUs Evaluated by EPA Are
Reasonable. [Stamper (0068) p. 32]
Stamper noted, as shown in the FIP TSD and in the above tables, the EPA's calculated costs of
scrubber retrofits range from $l,255/ton to $3,500/ton of SO2 removed. As discussed above, the
EPA's costs for SO2 controls represent a worst case cost estimate, since EPA's costs are based
on highest uncontrolled sulfur content and maximum MW-hours generated per year. Our revised
cost analyses for scrubber retrofits, which follow a methodology consistent with prior EPA
analyses, are all lower than EPA's cost estimates, and range from $l,103/ton to $3,459/ton.
Stamper stated that all of these costs are reasonable, in that other similar sources have had to bear
similar costs for pollution control to address regional haze. For example, EPA approved
installation of a dry FGD at the Colorado Springs Nixon Unit 1 as a reasonable progress measure
at a cost effectiveness of $3,744/ton.119 In addition, EPA proposed a FIP of reasonable progress
measures based on switching to a lower sulfur fuel oil at the fuel oil-fired boilers at the
Kanoelehua Hill Power Plant, the Puna Power Plant, and the Shipman Power Plant at a cost
effectiveness of approximately $5,600/ton.120 EPA also approved Wyoming's adoption of
reasonable progress requirements for each of the Jim Bridger units to install SCR at a cost
effectiveness ranging from $2,743/ton to $3,403/ton in 2013 dollars.121
Stamper stated that these Texas EGU control costs are also reasonable compared to the costs
incurred by other EGUs to meet SO2 BART. Data compiled by the National Park Service of
State SO2 BART determinations shows that the costs of SO2 controls to meet BART at EGUs
ranges from $l,571/ton to $7,309/ton.122 Colorado based its SO2 BART determination for Martin
Drake Units 6 and 7 based on installation of dry FGDs at a cost effectiveness of $2,808/ton and
$2,345/ton, respectively.123 To address regional haze requirements for SO2, the state of Wyoming
found that a new dry scrubber and baghouse at Dave Johnston Unit 4 was cost effective at $5,028
290
-------
per ton of SO2 removed.124 Wyoming also found that the costs of SO2 retrofit controls at
Naughton Units land 2 were cost effective at $1,622 to $2,654/ton.125 Several EGUs were
required to be retrofitted with scrubbers as part of the North Dakota regional haze plan, with cost
effectiveness ranging from $463/ton to $2,006/ton.l26 In the regional haze FIP for Oklahoma
sources, EPA found that dry scrubbers were reasonable to meet BART at cost effectiveness
numbers ranging from $l,291/ton to $l,544/ton in 2009 dollars, or $l,446/ton to $l,729/ton in
2012 dollars. 76 Fed. Reg. 16,183 (March 22, 2011).
Stamper stated that the projected costs for new scrubbers at the Texas EGUs are also comparable
to costs for NOx controls that EPA and states have found to be reasonable to meet BART. EPA
Region IX has required SCR as BART for the Four Corners Units 1 - 5 to meet a NOx limit of
0.11 lb/MMBtu at a cost effectiveness of $2,515/ton to $3,163/ton in 2008 dollars,127 or
$2,555/ton to $3,214/ton in 2012 dollars. In its FIP for Montana, EPA found that the cost
effectiveness of SCR controls for Colstrip Units 1 and 2 of approximately $3,200/ton per unit (in
2010 dollars), or $3,396/ton in 2012 dollars, was reasonable.128 In its FIP for Arizona regional
haze, EPA required SCR along with combustion controls to meet BART at the BART-subject
coal-fired units at Apache, Cholla, and Coronado power plants at cost effectiveness values
ranging from $2,275/ton to $3,472/ton.129 EPA Region VIII proposed SCR as BART at the
Leland Olds Unit 1, Milton R. Young Unit 1, and Milton R. Young Unit 2 at costs of
$l,800/ton, $2,600/ton, and $2,700/ton in 2009 dollars (or $2,016/ton, $2,912/ton, and
$3,024/ton in 2012 dollars), respectively.130 In its final action on the Wyoming regional haze
plan, EPA found that costs for SCR plus low NOx burners and overfire air ranging from
$2,635/ton to $4,461/ton (2008 dollars) are reasonable to require SCR as BART at Naughton
Unit 3, Dave Johnston Unit 3, and at Laramie River Units 1, 2, and 3.131 These costs range from
$2,677/ton to $4,532/ton in 2012 dollars. Moreover, the state of Arizona has stated that a cost
effectiveness value of more than $4,489/ton of NOx removed is cost effective.132
Stamper stated that the costs of scrubber upgrades proposed by EPA in the Texas regional haze
FIP are quite reasonable to justify these controls to meet regional haze requirements, whether
based on EPA's conservative cost estimates or the revised cost estimates provided in this report.
The costs of SO2 control, which range from $l,255/ton to $3,500/ton of SO2 removed based on
EPA's cost estimates, are within the range that other similar sources have had to bear to meet
regional haze requirements.133
Footnotes:
119 77 Fed. Reg. 18052, 18082 (March 26, 2012).
120 77 Fed. Reg. 61478, 61490 (October 9, 2012; see also 77 Fed.Reg. 31692, 31711-2 (May 29, 2012).
121 See 79 Fed. Reg. 5032 , 5040-41, 5046 (January 30, 2014).
122 See March 2011 National Park Service spreadsheet "EGUs with Proposed BART Controls." (Ex. 39).
123 77 Fed. Reg. 18052, 18070-1 (March 26, 2012)
124 See May 28, 2009 Wyoming Department of Environmental Quality BART Application Analysis, Dave Johnston
Plant, at 23 (Ex. 40).
125 See May 28, 2009 Wyoming Department of Environmental Quality BART Application Analysis, Naughton
Plant, at 26-28 (Ex. 41).
126 76 Fed. Reg. 58570, 58586-7, 58589, 58594 (September 21, 2011).
127 See 75 Fed.Reg. 64227 (October 19, 2010). See also 77 Fed.Reg. 51620, 51621-2 (August 24, 2012).
128 See 11 Fed. Reg. 24026-7, 24034-5 (April 20, 2012).
129 77 Fed. Reg. 42857, 42860, 42862 (July 20, 2012).
130 See 76 Fed.Reg. 58599 (September 21, 2011). While EPA did not ultimately require SCR as BART for these
291
-------
North Dakota EGUs in its final regional haze action, EPA's decision was based on whether SCR was technically
feasible for these lignite-fired units and not based on costs of controls. See 77 Fed.Reg. 20897-8 (April 6, 2012).
131 79 Fed. Reg. 5032-5222, at 5039-5043 (January 30, 2014).
132 Letter from Arizona Department of Environmental Quality to Steve Fry, EPA Region IX, Re: Consultation
Regarding Best Available Retrofit Technology Analyses for the Four Corners Power Plant and Navajo Generating
Station, May 12, 2008. Ex. 42.
133 While we do not agree that TCEQ's cost effectiveness threshold of $2,700/ton has been justified or is reasonable
for defining cost effective controls, it should be noted that the revised cost effectiveness of scrubber retrofits
provided in this report show costs at $2,700/ton or well below $2,700/ton for all EGUs evaluated with the exception
of the Welsh units.
We agree with the commenter that our cost effectiveness calculations were based on
conservative inputs. We also agree that the cost effectiveness of the controls we proposed are
well within a range that have been found to be cost effective by us in previous FIPs and by states
in many BART analyses for similar power plant units.
ll.d. Control Level
Comment: [Sargent & Lundy (0061) p. ES-2] S&L stated that EPA overestimated SO2 removal
that would be continuously achievable on a long-term basis for the scrubber upgrades due to the
following errors:
• EPA's calculation of baseline SO2 emissions from the existing units fails to provide a
realistic depiction of anticipated emissions, especially for sources that implemented SO2
emission reduction strategies during the 2009-2013 baseline period.
• EPA's calculations of controlled SO2 emissions are based on: (1) incorrect assumptions
regarding design conditions and variation in operation; (2) limitations on future fuel
flexibility; and (3) failure to differentiate between design/guarantee performance and
long-term performance.
[Sargent & Lundy (0061) p. ES-3] S&L stated that EPA overestimated SO2 emission reductions
associated with the proposed retrofit scrubber technologies due to the following:
• Inaccurate estimate of baseline SO2 emissions.
• EPA's calculations of controlled SO2 emissions are overstated and based on incorrect
assumptions regarding the actual performance and operation of wet FGD technology.
EPA's selection of a controlled emission rate of 0.04 lb/MMBtu is more consistent with
the lowest achievable emission rate (LAER) applicable to newly-constructed units, rather
than an emission rate achievable with retrofit controls installed on existing units. In fact,
a controlled emission rate of 0.04 lb/MMBtu is significantly lower than the most
aggressive Best Available Retrofit Technology (BART) SO2 emission limits imposed by
EPA on BART-eligible sources throughout the U.S.
• A controlled SO2 limit of 0.04 lb/MMBtu is not a realistic or sustainable value to
maintain on a long-term basis when considering the normal variation in operating
parameters that occur at all coal-fueled facilities.
Response: The SO2 removal efficiency and some of the other items that Luminant summarizes
292
-------
in this comment, and the information provided in its attached Sargent and Lundy (S&L) report,
regarding our scrubber upgrade costs are detailed in a separate comment package submitted
under the Confidential Business Information (CBI) provisions of 40 C.F.R.§ 2.203(b). Within
those CBI comments, S&L also provides its own cost analyses for upgrading Luminant's
scrubbers. We are unable to respond to many of these summarized scrubber upgrade comments
here with any specificity, because doing so would involve citing and discussing in detail items
that Luminant has claimed as CBI. Accordingly, many of our responses to the scrubber upgrade
comments are contained within a separate document that is not a part of our posted docket, but
will be available for review by Luminant. The responses to comments that do not contain CBI
information are contained within this document.
With regard to our scrubber upgrade cost analysis, we generally disagree with Luminant that our
analysis was flawed. We used Luminant's own information, backed by independent contractors
hired by it, supplied by Luminant in response to our Section 114 requests for information. This
included cost estimates from well-known and respected contracting firms with a history of many
scrubber upgrades. In any event, criticisms regarding our use of this information are moot,
because S&L has provided its own cost analysis (under the CBI protections), which it offers as a
replacement to our own cost analyses. We have reviewed the scrubber upgrade cost analyses
performed by S&L and adopted its methodology. However, we noted many errors and
undocumented cost figures in its analyses. We corrected these errors and rejected some of the
undocumented assertions and/or costs in S&L's cost analyses. Nevertheless, in order to produce
a conservative scrubber upgrade cost analysis and set many of the issues that Luminant raises
aside, we incorporated many of Luminant's cost items. The resulting costs for Luminant's
scrubber upgrades increased slightly, resulting in a range of $368/ton to 910/ton for all of the
scrubber upgrades, well within a range that we believe is cost effective, given the significant
visibility benefits that will result from the installation of those controls.
We address S&L's comments related to our calculation of baseline emissions elsewhere in our
response to comments. Regarding S&L's assertion that our proposed emission limits are more
consistent with LAER, we note that BART stands for "Best Available Retrofit Technology." As
we state in the BART rule,208 "You should be sure to consider the level of control that is
currently best achievable at the time that you are conducting your BART analysis." Therefore,
whether a particular emission limit could be viewed as BACT, LAER, or some other regulatory
designation has no relevance here. BART requires consideration of the best performance the
scrubber is able to achieve. As we note in our COST TSD, we believe that our proposed
emission limits are conservative and are continuously achievable on a long term basis. See our
response to another comment in which we present current monitoring information for scrubber
retrofits that demonstrates this fact.
Regarding S&L's assertion of the need for fuel flexibility, we can only perform scrubber cost
analysis on the basis of the sulfur content of coal that has historically been burned. Absent
information from the facilities (and appropriate enforceable commitments when warranted) we
cannot anticipate what the sulfur content of the fuels that facility may burn in the future.
208 70 FR 39171.
293
-------
Comment: EPA overestimated annual SO2 emission reductions achievable with scrubber
upgrades - biased methodology for estimating baseline emissions [Sargent & Lundy (0061) p.
10]
S&L stated that EPA also overstated the amount of additional SO2 that would be removed were
the upgrades performed. EPA calculated baseline SO2 emissions by averaging the annual SO2
emissions reported in the Air Markets Program Data from the previous five years (2009-2013),
excluding the maximum and minimum annual values. EPA asserted that this was a "reasonable
compromise between simply selecting the maximum value from 2009 - 2013 or using the
average of the values from 2009 - 2013."28 However, the approach used by EPA is entirely
arbitrary, and may not provide a realistic depiction of anticipated emissions from the existing
sources.
According to S&L, the BART Guidelines (40 CFR Part 51 Appendix W) state that baseline
emissions from exiting sources "should represent a realistic depiction of anticipated annual
emissions for the source."29 In general, for the existing sources, facilities should estimate the
anticipated annual emissions based upon actual emissions from a baseline period.30 However,
EPA provides no explanation or analysis to demonstrate that the approach taken results in a
realistic depiction of anticipated annual emissions from the existing sources. In addition, there is
no basis for concluding that EPA's approach of excluding actual emissions data more accurately
represents the actual operation of the units. This would be especially true on existing units that
have implemented SO2 control strategies during EPA's 2009-2013 baseline period. Finally, to
our knowledge, this approach has not been used previously by EPA as a methodology for
evaluating baseline emissions in other evaluations (and even if EPA had done so, it is not
justified here).
The following table shows a comparison between the baseline emissions as established using
EPA's approach, baseline emissions calculated as a straight average of the 2009-2013 data,
baseline emissions calculated as a 3-year average of actual emissions from 2011-2013, and
baseline emissions calculated as a 5-year average of actual emissions from 2010-2014.
Table 1: Comparison of Baseline SO2 Emissions for Existing Scrubbers
Unit
EPA
5 Year
3 Year
5 Year
Approach
Average
Average
Average
3 Year
2009-2013
2011-2013
2010-2014
Average*
(tons)
(tons)
(tons)
(tons)
Martin Lake 1
24,495
22,292
20,524
20,631
Martin Lake 2
21,580
21,128
19,512
18,861
Martin Lake 3
19,940
20,807
18,217
19,170
Monticello 3
13,857
13,375
10,429
11,455
Sandow 4
22,289
21,765
22,289
19,979
With the exception of Martin Lake Unit 3, EPA's approach of eliminating the maximum and
minimum values results in higher baseline SO2 emissions compared to averaging the entire 5-
year period, and in all cases EPA's approach results in higher baseline SO2 emissions compared
294
-------
to the 3-year average emission rate achieved for 2011-2013. Moreover, by using the most recent
data from 2010 to 2014 to calculate the 5-year average, SO2 baseline emissions in all cases are
lower than the 5-year average using data from 2009 to 2013. Overestimating the baseline SO2
emissions results in overestimating the amount of SO2 that would be removed and overstating the
cost-effectiveness of the scrubber upgrades.
Footnotes:
28 Technical Support Document for the Cost of Controls Calculations for the Texas Regional Haze Federal
Implementation Plan, November 2014, page 11.
29 79 FR 39167.
30 Id.
Response: We disagree with S&L that we erred in the procedure we used in estimating baseline
emissions for our scrubber upgrade cost analyses. As we note in our proposal, we used the
BART Guidelines for some aspects of our analysis. Regarding the calculation of baseline
emissions, the BART Guidelines state209:
How do I calculate baseline emissions?
1. The baseline emissions rate should represent a realistic depiction of anticipated
annual emissions for the source. In general, for the existing sources subject to
BART, you will estimate the anticipated annual emissions based upon actual
emissions from a baseline period.
2. When you project that future operating parameters (e.g., limited hours of
operation or capacity utilization, type of fuel, raw materials or product mix or
type) will differ from past practice, and if this projection has a deciding effect in
the BART determination, then you must make these parameters or assumptions
into enforceable limitations. In the absence of enforceable limitations, you
calculate baseline emissions based upon continuation of past practice.
We believe the procedure we outline below is in compliance with this language. We calculated
our baseline SO2 emissions by first acquiring the 2009 to 2013 emissions as reported to us the
facilities in question.210 This is reflective of the actual emissions due to the underperforming
scrubber systems installed at the units in question. We then calculated the uncontrolled SO2
emissions by acquiring EIA coal usage data. This considers the amount of each type of coal
burned and the sulfur content of each type of coal that the units in question reported to the EIA.
We used these two figures to calculate the level of control for each year. The following is a
summary of that information for Martin Lake Unit l211:
209 70 FR 39167.
210 http://ampd.epa.gov/ampd/
211 See the file, "Coal vs CEM data 2009-2013.xlsx," which was in the docket for our proposal.
295
-------
2009-2013
3 yr Average
(eliminate
max and
Average
2009
2010
2011
2012
2013
min)
Annual Emissions CEM
22,292.3
23,922.0
25,966.2
24,728.0
12,010.3
24,834.8
24,494.9
(tpy)
Annual Emissions Coal data
72,380.7
82,293.4
79,832.5
69,206.9
57,281.4
73,289.4
74,109.6
(tpy)
Annual Heat Input Coal Data
60,825,717.0
61,512,078.8
67,858,114.7
64,876,594.8
46,059,863.8
63,821,933.0
65,518,880.8
(MMBtu)
Average estimated percent
emitted
30.8%
29.1%
32.5%
35.7%
21.0%
33.9%
33.1%
Average estimated percent
69.2%
70.9%
67.5%
64.3%
79.0%
66.1%
66.9%
not emitted
CEM data emission rate
0.76
0.81
0.79
0.77
0.56
0.82
(lb/MMBtu)
Coal data emission rate
2.38
2.68
2.35
2.13
2.49
2.30
2.26
(lb/MMBtu)
95% removal from coal data
3,619.0
4,114.7
3,991.6
3,460.3
2,864.1
3,664.5
3,705.5
(emissions tpy)
95%eff. Removal rate
0.12
0.13
0.12
0.11
0.12
0.11
0.12
(lb/MMBtu coal)
Additional tons of SO2
removed from annual
20,789.4
emissions at 95% control
We eliminated the high low values from the 2009-2013 emission to better address the issues
S&L raises in its other comments (variations in coal sulfur data, capacity usage, upgrades in
equipment, etc.) and make the baseline more representative of plant operations. The difference
between our baseline calculations and a straight 2009-2013 average is small and would not
change our proposed conclusion that the scrubber upgrades we proposed are very cost effective.
We direct S&L to our response to a related comment concerning our scrubber upgrade baseline
emissions calculations which demonstrates that our methodology was likely inherently
conservative.
Comment: Incorrect assumptions for estimating achievable SO2 emission rates
[Sargent & Lundy (0061) p. 11]
In its 2006 studies for Luminant's predecessor, S&L identified the necessary upgrades and
associated costs to eliminate flue gas bypass on each of Luminant's existing scrubbers while
maintaining a high level of SO2 removal. For each of these units and the associated upgrades,
S&L identified the achievable performance based on the design conditions. The achievable
performance associated with these upgrades identified by S&L is as follows:
Table 2: Summary of FGD Upgrade Performance from S&L Studies
296
-------
Design Inlet
Unit
Sulfur
SO2
SO2 Emission
(lb/MMBtu)
Rem ova
Rate (lb/MMBtu)
Monticello Unit 3
1.84
93.0
0.13
Sandow Unit 4
3.90
93.0
0.27
Martin Lake Units 1-3
3.88
95.0
0.20
The S&L report predicted 95% removal may be achievable in the FGD systems with the bypass
still in operation, but further predicted that eliminating the bypass would result in 93% removal
in the scrubbers at Monticello and Sandow. This performance reduction is due to a higher
velocity in the absorber associated with treating more flue gas; this in turn reduces the residence
time available for the chemical reactions to occur. EPA, however, wrongly assumes 95%
removal efficiency could be achieved with the FGD upgrades identified in S&L's report in
conjunction with dibasic acid (DBA) addition. This is not a proper reading of our reports and
results in an overstatement of the amount of SO2 removal that can be achieved. This error is
even more apparent when the removal of the trays is taken into consideration. Trays provide
additional contact between the liquid and the gas which facilitates the scrubbing of SO2 from the
flue gas at lower liquid to gas ratios but also introduces significant pressure drop. Luminant
removed the trays in an effort to accommodate additional flue gas in the absorbers while limiting
the impact on the existing fans. When trays are removed the overall removal efficiency of the
system is reduced; therefore, the stated performance in the reports cannot be achieved without
additional improvements beyond those identified in the S&L reports such as higher L/G ratios
and the use of dibasic acid (DBA) or replacement of the removed trays. Even if it were possible
in theory to achieve 95% SO2 removal with DBA addition and the other upgrades (a fact that
EPA's proposal does not establish), EPA failed to account for all of the additional costs
associated with achieving and maintaining that removal efficiency.
S&L stated that EPA further erred by applying the 95% SO2 removal to the inlet sulfur
concentrations developed from the Energy Information Administration (EIA) coal database.
This approach is based on several flawed assumptions including:
• The historical coal sulfur levels will remain constant
• The SO2 removal identified can be achieved at lower inlet sulfur concentrations
• The SO2 removal is sustainable over time and through varying operating conditions
By taking this approach, EPA assumes that the current coal characteristics will not vary in the
future, which is not a reasonable assumption. S&L's report estimated that the expected
performance was achievable while firing the design fuel. This estimate was based on the design
coal, the original equipment design, and the proposed upgrades. The inlet sulfur concentrations
derived from the EIA coal data are much lower than the design inlet sulfur used as the basis of
S&L's analysis.
The S&L report did not indicate that the estimated removal efficiency could be achieved
consistently at these lower fuel sulfur levels. SO2 removal efficiencies cannot be universally
applied across any range of coal sulfur levels, especially when considering upgrades to existing
equipment. Existing scrubbers have limitations based on their original equipment design. While
upgrades, such as those proposed in the reports, can improve the performance of existing
297
-------
scrubbers, it is not possible for these existing scrubbers to achieve the same flexibility and
performance as new FGD systems. The S&L reports do not establish the achievable
performance at conditions other than the specified design parameters. Therefore, it is not
appropriate for EPA to assume that 95% SO2 removal can be achieved at the lower inlet sulfur
concentrations derived from the EIA coal database.
Further, the SO2 emission rates associated with the achievable performance identified in the S&L
report are design values for the upgrades which represent guarantee or initial performance at the
design conditions. Guarantee performance testing is short-term testing that is conducted under
ideal operating conditions just after the new equipment is installed and performing its best.
These values do not account for variability in the operating conditions nor normal equipment
performance degradation between maintenance cycles. Although these emission rates are
acceptable design targets for the upgrades, they do not represent emission rates that can be
achieved on a long-term basis under all normal operating conditions. It is customary to include
reasonable margin between the design target and the anticipated long-term actual controlled
emission rate to allow for normal fluctuations in the controlled emission rate and, especially in
the case of FGD upgrades, periods of equipment maintenance.
According to S&L, EPA's approach to estimating controlled SO2 emissions has several
significant flaws. EPA is overestimating the capabilities of the technology, overestimating the
tons of SO2 removed, and thus overestimating the cost-effectiveness of the proposed upgrades.
To accurately reflect S&L's determinations and findings in it reports, and in light of the above
considerations, EPA should revise the tons of SO2 removed to be based on a future emission rate
of 0.13 lb/MMBtu for Monticello Unit 3, 0.27 lb/MMBtu for Sandow Unit 4, and 0.20
lb/MMBtu for the Martin Lake units.
[Sargent & Lundy (0061) p. 13] The cumulative effect of EPA's errors is to substantially
overstate the cost-effectiveness of the scrubber upgrades that it would require. Although we
believe the starting estimates are not sufficient to properly quantify the 2015 costs associated
with the scrubber upgrades at the Luminant units, we have calculated more accurate cost-
effectiveness numbers by correcting the errors discussed above starting from the original
estimates. Even just correcting these limited errors by EPA shows that, for all three of
Luminant's plants with existing scrubbers, EPA has overstated by at least 2-3 times the cost-
effectiveness of the upgrades EPA is proposing. In other words, our corrected $/ton values are 2-
3 times higher than those relied on by EPA to justify its proposal. We believe that the adjusted
cost-effectiveness numbers are conservative when considering that the starting estimates
insufficiently capture the 2015 costs to upgrade the scrubbers. The specific values and
calculations are provided in the confidential appendix to this report.
[Stamper (0068) p. 10; Earthjustice (0067) p.37] Stamper stated that there are several aspects of
EPA's SO2 retrofit cost effectiveness analyses in which EPA applied overly conservative
assumptions. Also, it appears that EPA used a different methodology in its wet FGD analysis
than in its SDA cost analysis.
Specifically, Stamper stated that in the wet FGD analysis, EPA used the following information:
First, EPA used a period of baseline emissions based on the three year average of annual
298
-------
emissions from 2009 to 2013,28 excluding the years of maximum and minimum emissions. EPA
also calculated gross heat rate from EPA's Clean Air Markets Database data based on a three
year average excluding the maximum and minimum heat rates from 2009 to 2013. Second, EPA
assumed an uncontrolled SO2 emission rate based on the maximum calculated monthly SO2
lb/MMBtu emission rate over 2009 to 2013.29 Third, EPA used a 3-year average of gross load
(MW-hours) over 2009 to 2013, excluding the maximum and minimum years of MW-hours
generated.
In the SDA and DSI analyses, Stamper stated that the EPA used somewhat different information.
While EPA continued to use a period of baseline emissions based on the three year average of
annual emissions from 2009 to 2013 excluding the years of maximum and minimum emissions,
EPA used a gross heat rate based on the maximum annual value from 2009 to 2013. Similar to
the wet FGD analysis, EPA used the maximum monthly SO2 rate over 2009 to 2013 to reflect
uncontrolled emissions. However, EPA used the maximum annual gross load over 2009 to 2013
to project operation and maintenance costs.
Stamper stated that the EPA should be consistent in its methodologies for estimating costs and
emission reductions between control technologies being evaluated for a source. Indeed, EPA
should be consistent in its methodologies for estimating costs and emission reductions in all of
its regional haze cost analyses. However, the methodologies that EPA applied in the Texas SO2
control cost analyses are not consistent with how EPA has typically determined the cost
effectiveness of pollution control technologies for retrofit controls.
Stamper stated that the EPA typically has determined cost effectiveness based on long term
averages of emissions, uncontrolled emission rates, and capacity factors. For example, in
evaluating SO2 retrofit controls for the Corette Power Plant in Montana, EPA used an annual
average actual SO2 lb/MMBtu emission rate over three years to reflect uncontrolled SO2
emissions for determining the capital and operations & maintenance costs of SO2 controls.30 EPA
also determined the SO2 removal efficiency based on the three year annual average SO2 emission
rate in lb/MMBtu (based on CAMD data) and based on an assumed SO2 emission rate for the
control technology being evaluated.31 EPA then calculated the SO2 emissions reduced from the
three year average of SO2 emissions (tons per year) multiplied by the SO2 removal efficiency,
which in turn was based on the three year average SO2 rate in lb/MMBtu and the proposed SO2
emission limit for the control technology being evaluated. EPA has evaluated NOx control cost
effectiveness in the same manner, using the three year average of annual lb/MMBtu emission
rates and the three year average of annual tons of NOx emitted in determining cost effectiveness
of NOx controls.32
By using the highest monthly average SO2 emissions rate over five years, Stamper stated that the
EPA's methodology applied for these Texas reasonable progress controls would tend to
overestimate the SO2 control technology costs, especially the operations & maintenance costs, by
overstating the typical uncontrolled SO2 emission rate. That is because EPA's methodology
determines the operational costs for a scrubber based on historical worst case sulfur content.
With higher sulfur content coal, more reagent is needed to remove SO2 and more scrubber waste
is generated. For example, EPA's cost analysis for Monticello Unit 1 was based on an
uncontrolled SO2 rate of l.33 lb/MMBtu, whereas EPA's cost analysis for Big Brown Unit 1 was
299
-------
based on an uncontrolled SO2 rate of 2.0 lb/MMBtu.33 Although these are similarly sized units
(Monticello Unit 1 is 562.9 MW and Big Brown Unit 1 is 572.9 MW), the costs for limestone
reagent and waste disposal are much lower at Monticello Unit 1 with its lower uncontrolled SO2
rate than at Big Brown Unit 1. Specifically, costs for limestone reagent and waste disposal at
Monticello Unit 1 were projected to be $0.36/MWh and $0.65/MWh respectively, whereas the
costs for limestone reagent and waste disposal at Big Brown Unit 1 with higher sulfur coal were
$0.52/MWh and $0.94/MWh respectively.34 Further, a review of the average monthly SO2
emission rates for Big Brown Unit 1 shows that EPA's assumed 2.0 lb/MMBtu uncontrolled SO2
emission rate only occurred one month out of five years and, in fact, the average uncontrolled
monthly SO2 rate (i.e., average of all monthly averages over five years) was 1.44 lb/MMBtu.35
EPA's approach of using the highest monthly SO2 rate for determining the costs of the SO2
controls thus overestimates the operational costs of the control by designing for a worst case
coal.
Earthjustice et al., and Stamper stated that by using a historical maximum MW-hours generated,
EPA overstated the operational costs. That is because the IPM cost module used by EPA
estimates operational costs in terms of dollars per MW-hours generated. When EPA uses a
higher than typical MW-hours generated in its costs analyses, the annual operational expenses
will be higher than if based on a longer term annual average expected generation rate.
Stamper stated that the EPA's methods for estimating annual emission reductions in its cost
effectiveness are based on historical data that are not internally consistent. For example, for the
SDA and DSI analyses, EPA used a maximum annual gross heat rate from the years 2009 to
2013. Heat rate indicates the amount of British Thermal Units (Btus) of energy needed from the
fuel burned at a specific unit to produce 1 kilowatt-hour, and is essentially a measure of the
efficiency of the unit. The gross heat rate is, in turn, used to project the annual heat input to
produce the projected annual level of generation (MW-hours). The annual heat input is then
used to project future controlled SO2 emissions, based on the assumed controlled SO2 emission
rate. By using a historical maximum annual gross heat rate, EPA thus overstates the annual heat
input at the unit and, consequently, overstates the controlled tons of SO2 emitted per year. Then,
to determine the tons per year of SO2 removed with the SO2 control, EPA subtracted baseline
emissions based on a 3-year average over 2009 to 2013 excluding the maximum and minimum
years. Depending on how variable a unit's heat rate and generation rate has been over 2009 to
2013, this approach overstated the emission reductions with a particular control for some units,
and it underestimated emission reductions for other units in comparison to determining the
emission reductions from a control simply based on long term average heat rate, generation rate,
and emissions.
For the wet FGD cost analyses, Stamper stated that the EPA used more of an apples-to-apples
methodology, with all data used to project emission reductions with a control based on a 3-year
average over 2009 to 2013 excluding the maximum and minimum years. However, EPA has not
justified excluding the maximum and minimum years from a projection of heat rate, annual MW-
hours generated, or baseline SO2 emissions. There can be many different reasons for a
maximum or minimum year of heat rate, generation, and/or SO2 emissions at a unit. For
example, heat rate is usually highest in the year or two before a turbine overhaul, which typically
occurs every 5 to 10 years. SO2 baseline emissions can be higher due to the increased heat rate
300
-------
and also due to increases in coal sulfur content. Generation can be higher one year over another
due to an adjoining unit having an extended outage. EPA should account for those maximums
and minimums in projecting emission reductions from a control by simply using a long term
average of this data, as EPA has done in other control cost evaluations such as for the Corette
power plant BART determination discussed above.
In summary, Earthjustice et al., and Stamper stated that EPA should use the same methodology
in determining cost effectiveness of reasonable progress controls as it has used for BART cost
effectiveness analyses (that is, in essence, basing all costs and emission reductions on long term
average emission rates and capacity factors/generation rates). As will be shown further below,
use of the EPA's historically-applied methodology shows that the retrofit SO2 controls evaluated
by EPA are even more cost effective than shown by EPA.
Footnotes:
28 Note that, in EPA's Wet FGD Cost spreadsheet, SDA Cost spreadsheet, and DSI Cost spreadsheet, the "Annual
Emissions" and "Monthly Emissions" tabs incorrectly indicate that SO2, gross load, and gross heat rate are from
2009 to 2012. The formulas in the spreadsheets encompass 2009 to 2013, so the headings in these tabs should be
corrected to state that the averages and maximums are from 2009 to 2013.
29 Note that EPA's Spreadsheet of Wet FGC Costs (TX166-008-092_Costing_-Wet FGD_Cost_IPM_5-
13_TX_Sources_Ver 2.xlsx) incorrectly states that the uncontrolled SO2 rate is the "Avg of months with full
operating time." EPA's spreadsheets for SDA costs and DSI costs make clear that EPA assumed the maximum
monthly value from 2009 - 2013, and it is clear from a review of the emissions data that EPA also used the
maximum monthly SO2 rate in its wet FGD cost calculations.
30 Copies of EPA's cost effectiveness spreadsheets for SDA, wet FGD, and DSI at the Corette Power Plant are
attached as Exs. 10A, B, and C.
31 It must be noted that EPA assumed too low of SO2 removal efficiencies in its evaluation of cost effectiveness of
SO2 controls for the Corette Power Plant, as was discussed in detail in the Technical Support Document to
Comments of Conservation Organizations, Proposed Montana Regional Haze FIP - June 15, 2012, at 94 to 101 (Ex.
30).
32 See, e.g., EPA's NOx Cost Effectiveness Analyses for Jim Bridger Power Plant, EPA-R08-OAR-2012-0026-0085,
attached as Ex. 12.
33 See TX166-008-092_Costing_-_Wet_FGD_Cost_IPM_5-13_TX_Sources_ver_2.xlsx (in Docket ID EPA-R06-
OAR-2014-0754-0008) at BB1 tab and Ml tab, cell E8 in each tab.
34 Id. at cells C69 and C70 in each tab.
35Id. at "Monthly Emissions tab."
[Stamper (0068) p. 37] Stamper stated that the EPA determined the current SO2 removal efficiencies
of the wet scrubbers in operation at the 9 EGUs analyzed by EPA by first determining the uncontrolled
SO2 rate based on the percent sulfur in the coal and the types and amount of coal burned at each EGU.
Cost TSD at 29. However, EPA's analysis assumes all of the sulfur in the coal is converted to SO2, when
typically about 2-5% of the sulfur in the coal falls out in the bottom ash in the boiler.151 Further, as shown
in EPA's analysis comparing its theoretical SO2 emissions to monitored SO2 emissions for unscrubbed
units, EPA demonstrated that its estimation of uncontrolled SO2 emissions actually overstates the
uncontrolled SO2 emission rate. Cost TSD at 30 (Table 15). To determine SO2 removal efficiency of the
existing scrubbers, EPA compared uncontrolled SO2 emissions based on sulfur content and amount of
coal burned to actual SO2 emissions from CAMD. However, because EPA has likely overestimated the
uncontrolled SO2 emissions for the scrubbed units, that means EPA also likely overstated the current SO2
removal efficiency of the scrubbers at the existing EGUs.
301
-------
Stamper stated that the EPA acknowledged this issue, along with other potential issues, in its
methodology to determine the SO2 removal efficiencies of the existing wet scrubbers at the 9
EGUs evaluated by EPA. FIP Cost TSD at 31. All of this means it is likely that EPA has
overestimated the current SO2 removal efficiency of the existing scrubbers at the 9 EGUs
evaluated for scrubber upgrades.
Footnotes:
151 U.S. EPA AP-42, Section 1.1 Bituminous and Subbituminous Coal Combustion, at 1.1-3.
Response: Although we cannot provide details here on many aspects of S&L scrubber upgrade
comments for the reasons we outline below, we disagree with S&L that we misread their reports.
In addition, S&L apparently implies we relied solely on its reports in concluding that the units in
question should be controlled to the level we proposed. In fact the S&L reports were not the
primary source of information we relied upon in proposing the control levels for the units in
question, which can be discerned by reading our Cost TSD.212
The items that Luminant and S&L summarizes in this comment, and the information provided in
its attached Sargent and Lundy (S&L) report, regarding our scrubber upgrade costs are detailed
in a separate comment package submitted under the Confidential Business Information (CBI)
provisions of 40 C.F.R.§ 2.203(b). Within those CBI comments, S&L also provides its own cost
analyses for upgrading Luminant's scrubbers. We are unable to respond to many of these
summarized scrubber upgrade comments here with any specificity, because doing so would
involve citing and discussing in detail items that Luminant has claimed as CBI. Accordingly,
many of our responses to the scrubber upgrade comments are contained within a separate
document that is not a part of our posted docket, but will be available for review by Luminant.
The responses to comments that do not contain CBI information are contained within this
document.
With regard to our scrubber upgrade cost analysis, we generally disagree with Luminant that our
analysis was flawed. We used Luminant's own information, backed by independent contractors
hired by it, supplied by Luminant in response to our Section 114 requests for information. This
included cost estimates from well-known and respected contracting firms with a history of many
scrubber upgrades. In any event, criticisms regarding our use of this information are moot,
because S&L has provided its own cost analysis (under the CBI protections), which it offers as a
replacement to our own cost analyses. We have reviewed the scrubber upgrade cost analyses
performed by S&L and adopted its methodology. However, we noted many errors and
undocumented cost figures in its analyses. We corrected these errors and rejected some of the
undocumented assertions and/or costs in S&L's cost analyses. Nevertheless, in order to produce
a conservative scrubber upgrade cost analysis and set many of the issues that Luminant raises
aside, we incorporated many of Luminant's cost items. The resulting costs for Luminant's
scrubber upgrades increased slightly, resulting in a range of $368/ton to 910/ton for all of the
scrubber upgrades, well within a range that we believe is cost effective, given the significant
visibility benefits that will result from the installation of those controls.
212 See our Cost TSD, particularly Sections 6 and 7, which is heavily footnoted with the references we used to
justify our proposed control levels.
302
-------
We agree with Ms. Stamper that we employed conservative assumptions and data analyses
techniques in our cost analyses. However, we disagree that we used different heat rates and
gross loads in our wet FGD and versus our DSI and SDA cost analyses. This can be verified
from an examination of the cost spreadsheets.213 We also disagree that we deviated significantly
from our previous cost effectiveness methodologies. We agree that in some cases we (and the
states) have used the more simple method of establishing the historical uncontrolled emission
rate of simply averaging annual averages of emissions. However, in the present case, we wanted
to employ a more refined approach that would allow us to account for variations in emissions
that could occur due to changes in coal blending, sulfur content, and capacity usage. This is
especially important when designing a scrubber system, as it must be capable of addressing
reasonably worst case operating conditions.214 These types of considerations are not always
necessary, especially for units with rather static operations and fuels. We believe that employing
the more direct method the commenter suggests in our proposal would have led to
unrepresentative emission profiles in some cases and potentially under designed scrubbers. The
commenter's example for averaging the SO2 emission rate for Big Brown would have had us
design the scrubber based on an emission rate of 1.44 lbs/MMBtu fuel, as opposed to our use of a
2.0 lb/MMBtu value. This could result in the unit in question unable to comply with our SO2
emission rate for the one month that resulted in that 2.0 lb/MMBtu emission spike. Hence, we
established our uncontrolled emission rate by selecting the maximum monthly SO2 emission
rates. We acknowledge this methodology may result in a worse cost effectiveness calculation
(higher $/ton).
We also acknowledge that we employed a maximum gross heat rate in all of our SDA, DSI, and
wet FGD cost calculations and this had the effect, as the commenter notes, of maximizing the
operational costs. We disagree with the commenter that our SO2 baseline emission calculation
methodology "overstated the emission reductions with a particular control for some units, and it
underestimated emission reductions for other units in comparison to determining the emission
reductions from a control simply based on long term average heat rate, generation rate, and
emissions." As we discuss above, we do not believe that using input average values was the
most appropriate option.
We agree with Ms. Stamper that some of our assumptions may appear to be internally
inconsistent. However, our overall goal was to demonstrate that all of our proposed controls
were cost effective, even if they were performed using conservative parameters and assumptions.
We believe we have accomplished that goal.
We also agree with Ms. Stamper that in all of the cases in which we tested our calculation of the
theoretical uncontrolled SO2 calculation, the result was higher than the actual measured SO2
emissions. We discuss this in our Cost TSD215 and outline the possible reasons for this, which
include those the commenter lists. This means that our calculation of the additional SO2 tons
removed when the existing scrubbers are upgraded to operate at 95% control are likely low.
213 See the files, "Wet FGD Cost IPM 5-13 TX Sources ver 2.xlsx," "SDA FGD Cost IPM 5-13 TX Sources ver
2.xlsx," "DSI FGD Cost IPM 5-13 TX Sources ver 2.xlsx,"
214 As we note in other comments, unless an operator provides information to the contrary, we must base our
emission baseline on historical data.
215 See Table 15 of our Cost TSD and the text immediately before and after it.
303
-------
Because this figure appears in the denominator of our cost effectiveness metric ($/ton), it means
the resulting fraction is likely high, or overstated. In other words, were we to have the actual
uncontrolled SO2 emissions instead of having to rely on our theoretical calculations, we believe
the cost effectiveness of our scrubber upgrades would actually be more attractive (lower $/ton).
This is another area in which we strove to be conservative and demonstrate that even with many
conservative cost assumptions, our analyses indicate that the scrubber upgrades we proposed are
very cost effective.
Regarding S&L's assertions that we erred in using the EIA data to calculate our proposed
scrubber upgrade control efficiency, we make the following points:
S&L states that we cannot assume that the coal sulfur content of the coals the facilities burn will
remain constant into the future. We agree that we cannot predict what coals the facilities in
question will burn, and it is not our intention to dictate that to them. Furthermore, as the quote
from our BART Guidelines concerning the calculation of the emission baseline that we
reproduced in a related comment indicates, the emission baseline calculations are based on actual
emissions from the baseline period. The "baseline emissions rate should represent a realistic
depiction of anticipated annual emissions." Lacking any commitment from the facilities in
question concerning what coals they may choose to burn in the future, we believe it is
appropriate to use the historical emissions as an indicator of future operations.
As our response above to Ms. Stamper indicates, we believe any error in our scrubber upgrade
baseline SO2 emissions due to our use of the EIA coal data will likely serve to worsen our cost
effectiveness (higher $/ton) calculation. Thus, our use of the EIA coal data was likely
conservative.
Regarding S&L's assertions that our proposed control level is too high, we note that our Cost
TSD references many industry sources that conclude that a 95% control level for a scrubber
upgrade is in fact a conservative assumption and that many scrubber upgrades have performed
above this level while burning a variety of coals. We note in our proposal that we believe a 95%
control assumption provides an adequate margin of error for any of the units for which we have
proposed scrubber upgrades, such that they should be able to comfortably attain the emission
limits we have proposed. However, the conservative nature of our SO2 baseline calculations
aside, for the operator of any unit that disagrees with us on this point, we proposed the following:
(1) The affected unit should comment why it believes it cannot attain the SO2 emission limit we
have proposed, based on a scrubber upgrade that includes the kinds of improvements (e.g.,
elimination of bypass, wet stack conversion, installation of trays or rings, upgraded spray
headers, upgraded ID fans, using all recycle pumps, etc.) typically included in a scrubber
upgrade.
(2) After considering those comments, and responding to all relevant comments in a final
rulemaking action, should we still require a scrubber upgrade in our final decision making
action we will provide the company the following option to seek a revised emission limit
after taking the following steps:
304
-------
(a) Install a CEMS at the inlet to the scrubber.
(b) Pre-approval of a scrubber upgrade plan conducted by a third party engineering firm that
considers the kinds of improvements (e.g., elimination of bypass, wet stack conversion,
installation of trays or rings, upgraded spray headers, upgraded ID fans, using all recycle
pumps, etc.) typically performed during a scrubber upgrade. The goal of this plan will be
to maximize the unit's overall SO2 removal efficiency.
(c) Installation of the scrubber upgrades.
(d) Pre-approval of a performance testing plan, followed by the performance testing itself.
(e) A pre-approved schedule for 2.a through 2.d.
(f) Should we determine that a revision of the SO2 emission limit is appropriate, we will have
to propose a modification to our decision making to do so. It should be noted that any
proposal to modify the SO2 emission limit will be based largely on the performance
testing and may result in a proposed increase or decrease of that value.
Although we conclude that our proposed control efficiencies remain conservative, we believe the
above procedure will adequately address any errors, which we expect to be small, in our
proposed control efficiency calculations. For the above reasons, we disagree with S&L that our
proposed control efficiency is in error.
Comment: EPA over-estimation of SO2 emission reductions - biased methodology for
estimating baseline emissions [Sargent & Lundy (0061) p. 21]
In the report prepared for Luminant, S&L stated that the proposed approach by EPA to establish
baseline emissions may not result in a realistic depiction of anticipated emissions from the
existing sources, and is inconsistent with previous methodologies used by EPA for evaluating
baseline emissions in other evaluations. The following table shows a comparison between the
baseline emissions under EPA's approach (which uses the highest 3 of 5 years), baseline
emissions calculated as a straight average of the 2009-2013 emissions data, baseline emissions
calculated as a 3-year average of actual emissions from 2011-2013, and baseline emission
calculated as a 5-year average of actual emissions from 2010 to 2014.
Table 3: Comparison of Baseline SO2 Emissions for Un-Scrubbed Units
EPA
Approach
5 Year
3 Year
5 Year
3 Year
Average
Average
Average
Average*
(2009-2013)
2011-2013
2010-2014
Unit
(tons)
(tons)
(tons)
(tons)
Big Brown 1
30,667
30,606
30,990
29,179
Big Brown 2
30,814
30,638
31,467
30,033
Monticello 1
17,865
16,435
14,168
13,714
Monticello 2
16,429
15,458
12,163
12,548
S&L stated that EPA's approach using the highest 3 of 5 years fails to account for SO2 emission
reduction strategies that may have been implemented at an existing facility during the 2009-2013
305
-------
baseline period. This is clearly the result for Monticello Units 1 & 2. In addition, by using the
most recent data from 2010 to 2014 to calculate the 5-year average, SO2 baseline emissions in all
cases are lower than the 5-year average using data from 2009 to 2013. This illustrates that
EPA's approach results in significantly higher baseline SO2 emissions and does not provide a
realistic depiction of anticipated emissions from the sources, contributing to EPA's overstating
the cost-effectiveness of the new scrubbers.
Response: As in our scrubber upgrade baseline calculations, we eliminated the high low values
from the 2009-2013 emission to better address the issues S&L raises in its other comments
(variations in coal sulfur data, capacity usage, upgrades in equipment, etc.) and make the
baseline more representative of plant operations: We believe our methodology for calculating
the SO2 baseline emissions for our proposed scrubber retrofits is in compliance with the Regional
Haze Rule for the same reasons we outlined with regard to our scrubber upgrade baseline
calculations. Regardless, the difference between our baseline calculations and a straight 2009-
2013 average is small and would not change our proposed conclusion that the scrubber retrofits
we proposed are cost effective.
Comment: Incorrect assumptions for estimating achievable SO2 emission rates
[Sargent & Lundy (0061) p. 22]
S&L stated that EPA calculated SO2 emission reductions based on the assumption that a retrofit
wet FGD will achieve 98% reduction or a controlled SO2 emission rate of 0.04 lb/MMBtu. In
our experience, this assumption is unrealistic and cannot be sustained on a continuous, long-term
basis. As a basis for its assumption, EPA cites the IPM wet FGD cost development document,
which states: "The least squares curve fit of the data was defined as a 'typical' wet FGD retrofit
for removal of 98% of the inlet sulfur. It should be noted that the lowest available SO2 emission
guarantees, from the original equipment manufacturers of wet FGD systems, are 0.04
lb/MMBtu."63 However, EPA misinterprets this statement when it uses the guarantee emission
rate to calculate long-term SO2 emissions.
S&L asserted that EPA's proposal is far too stringent to be achievable with the retrofit of an
existing unit. EPA's selection of a controlled emission rate of 0.04 lb/MMBtu is more consistent
with the lowest achievable emission rate (LAER) applicable to newly-constructed units, rather
than an emission rate achievable with retrofit controls installed on existing units. In fact, a
controlled emission rate of 0.04 lb/MMBtu is significantly lower than the most aggressive BART
SO2 emission limits imposed by EPA on BART-eligible sources throughout the U.S.
According to S&L, a controlled SO2 limit of 0.04 lb/MMBtu is not a realistic or sustainable
value to maintain on a long-term basis when considering the normal variation in operation that
occurs at all coal-fueled facilities. As noted in the IPM wet FGD document, the 0.04 lb/MMBtu
emission rate corresponds to the lowest available SO2 emission guarantees from wet FGD
suppliers. Compliance with a vendor's guarantee value is typically demonstrated during very
short term testing conducted at ideal operating conditions. Vendor guarantees do not reflect
controlled emission rates that may be achievable on a consistent long-term basis as the unit
operation varies from design conditions.
306
-------
S&L noted that wet FGD control systems, like all large air pollution control systems, are not
steady state control systems, and controlled SO2 emissions will continually fluctuate in response
to changing operating parameters. Operating parameters that may affect SO2 emissions include
the fuel sulfur content, boiler load, load changes, flue gas flow rate, and flue gas temperatures,
all of which continually change during normal operation of the boiler. Figure 1 shows the hourly
emissions data for Spruce Unit 1. Spruce Unit 1 is a 550 MW subbituminous coal-fired boiler
equipped with wet limestone FGD. The FGD control system was installed in 1992, and
upgraded in 2009. Figure 1 shows the normal fluctuation in controlled hourly SO2 emissions
from Spruce Unit 1, which is typical for all wet FGD control systems.
Figure 1: SO2 Emission Permit Limits for Recent Wet FGD Projects [Figure not excerpted]
According to S&L, between January 1, 2012 and December 31, 2013, Spruce Unit 1 achieved an
average controlled SO2 emission rate of 0.055 lb/MMBtu (2-year average). During that same
time period, the standard deviation in its controlled 30-day average SO2 emission rate was 0.10
lb/MMBtu, approximately 20% greater than the long-term average, and the controlled SO2
emission rate achieved at a 95% confidence level was 0.075 lb/MMBtu (0.02 lb/MMBtu above
the long-term average). Based on our experience with other wet FGD control systems,
fluctuations seen in the controlled SO2 emission rate at Spruce Unit 1 are typical.
S&L stated that projecting future emissions using the anticipated control system vendor
guarantee (i.e., 0.04 lb/MMBtu) as EPA did is overly aggressive and provides no margin for
normal operating conditions or long-term operation. A reasonable margin between the vendor
guarantee value or design target, and the projected actual long-term achievable emission rate is
needed to allow for normal fluctuations in the controlled emissions. In S&L's opinion, an
operating margin of at least 0.02 lb/MMBtu between the vendor guarantee and projected long-
term emission rate is reasonable. In addition, it would be necessary to contact FGD suppliers to
provide bids, on a case-by-case basis, that identify what outlet emission rates would be
guaranteed on a 30-day rolling average to determine adequate outlet emission rates for
compliance, rather than applying the lowest available emission rate guarantee in all cases.
S&L concluded that EPA's approach to estimating controlled SO2 emissions is incorrect and
based on a misunderstanding of the actual performance and operation of wet FGD technology.
By using this approach, EPA is overestimating the tons of SO2 removed and thus overstating the
cost-effectiveness of the retrofit FGD control systems.
[Stamper (0068) p. 15] Stamper stated that the EPA assumed SO2 removal efficiencies of up to
98%) control in its wet FGD analyses and up to 95%> control in its SDA analyses. However, in no
case did EPA assume a controlled wet FGD SO2 rate of less than 0.04 lb/MMBtu or a controlled
SDA SO2 rate of less than 0.06 lb/MMBtu in its cost effectiveness analyses. EPA's rationale
was that these were the lowest emission guarantees and this methodology was consistent with
EPA's cost effectiveness calculations for the Oklahoma regional haze FIP. Cost TSD at 16, 21-
22.
Stamper stated that the EPA's assumptions limiting the SO2 removal efficiency based on the
307
-------
guaranteed permitted limits of 0.06 lb/MMBtu for an SDA and 0.04 lb/MMBtu for a wet FGD
underestimated the actual SO2 emission reductions that would likely occur with these SO2
scrubbers. This is primarily a concern for those EGUs that burn or blend with low sulfur Powder
River Basin coal (i.e., Coleto Creek 1, Tolk 171B and 172B, WA Parrish 5, 6, and 7, and Welsh
1, 2, and 3). Further, wet FGD systems can achieve greater than 99% SO2 removal, and dry
scrubbers can achieve greater than 95% SO2 removal efficiency, as discussed further below.
According to Stamper, wet scrubbers are the most effective SO2 control technology available.
State-of-the-art wet scrubbers can achieve 98-99% removal efficiency. A prime example is the
Chiyoda CT-121 FGD. Vendor information for this technology indicates that this scrubber has
achieved 98-99% SO2 removal even with low sulfur coal.43 For example, the Chiyoda's bubbling
jet reactor has consistently achieved >99% SO2 removal during long-term operation at the
Shinko-Kobe power plant in Japan. This facility consists of two 700-MW coal-fired utility
boilers. The wet FGD was designed to achieve 0.014 lb SCh/MMBtu (9 ppmv at 3% oxygen) on
an instantaneous basis and has consistently exceeded this level of control while treating gases
with inlet SO2 concentrations of 1.78 lb/MMBtu.44 This technology has been guaranteed by
Chiyoda to achieve 99% SO2 removal on three coal-fired boilers in Japan.45
Stamper stated that the Chiyoda CT-121 FGD has been demonstrated in the U.S. at the
University of Illinois's Abbott power plant, Georgia Power's Plant Yates46, Dayton Power &
Light's Killen Unit 2,47 and Plant Bowen Unit 3.48 It has been licensed for installation on several
additional units in the U.S., including the other three units at Plant Bowen in Georgia, the other
units at Dayton Power & Light's Killen plant, Dayton Power & Light's Stuart plant, and AEP's
Big Sandy Unit 2, Conesville Unit 4, Cardinal Units 1 and 2, and Kyger Creek, among others 49
Black & Veatch and Southern Company are both U.S. licensees. Further, this technology also
has shown to be very effective in removing fine particulates, oxidized and elemental mercury,
and acid gases, and the technology uses less energy compared to traditional wet scrubbers.
Stamper noted that Mitsubishi, a vendor of scrubber systems, reports it has guaranteed SO2
removal efficiencies up to 99.8 percent, including for coal-fired boilers.50'51'52
Stamper stated that Sargent & Lundy has indicated that the lowest achievable SO2 emission rate
with low sulfur Powder River Basin coal for a limestone forced oxidation (LSFO) wet scrubber
would be 0.03-0.06 lb/MMBtu.53
Stamper stated that these SO2 removal efficacy claims for wet scrubbers at coal-fired boilers
burning lower sulfur coal have been demonstrated in practice. SO2 emissions data was compiled
for 2009 to 2013 to reflect newly operating and recently retrofitted wet FGD systems at EGUs
that burn low sulfur coal. The results are in Table 3 below. All of these units burn Powder River
Basin coal and have recently been retrofitted with wet FGD systems. As the table below
demonstrates, wet FGDs at EGUs burning Powder River Basin coal can achieve much lower SO2
emission rates than 0.04 lb/MMBtu.
308
-------
Tabled v V-. f ... . i ..I I * •• •
fliaf Bui i» Low Silftii* Coal54
Pi.itir
Unit
Annual Av« age
SOI Rate,
lb MM Bui
Iatan
1
0 007
lataii
t
0.007
Bowell Eiierjrv Center
5
0
Muscatine
0
0 015
Pleasant Prairie
1
0 02!
Pleasant Prairie
•j
O.O:?
Thus, Stamper concluded that the EPA's assumption that a wet FGD could not achieve SO2 rates
lower than 0.04 lb/MMBtu is overly conservative, as demonstrated by the data presented above.
In addition, Stamper noted that recently constructed EGUs that burn Powder River Basin coal
and that are equipped with SDAs have achieved lower SO2 emission rates than the minimum
0.06 lb/MMBtu SO2 rate assumed by EPA in its cost effectiveness analyses. The Newmont
Nevada power plant (aka TS Power Plant), equipped with a dry lime FGD system, has achieved
an annual average SO2 rate of 0.037 lb/MMBtu over 2009 to 20 1 3.55 The Wygen II is also
equipped with a dry lime scrubber and burns low sulfur coal, and is achieving annual average
SO2 rates of 0.047 lb/MMBtu.56 And the Dry Fork Station which began operation in 2011 and is
equipped with a dry lime scrubber is achieving an annual average SO2 rate of 0.044 lb/MMBtu.57
Further, Stamper stated that a circulating dry scrubber can achieve even higher levels of SO2
removal than an SDA and lower SO2 emission rates. As discussed elsewhere in this document,
EPA should have evaluated circulating dry scrubbers in its analyses of reasonable progress
measures for the Texas EGUs.
In summary, Stamper contended that the EPA's assumed level of SO2 control in its cost
effectiveness analyses for SO2 for wet FGD and SDA systems is likely conservative and does not
reflect the full extent of SO2 emission reductions that can be achieved with these control
technologies. That means EPA's cost effectiveness calculations should be lower and the
visibility improvements higher for several of the EGUs evaluated by EPA for reasonable
progress controls.
Footnotes:
42 See, e.g., Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC Technologies, Dry Sorbent
Injection for SO2 Control Cost Development Methodology, March 2013, at 6.
43See Black & Veatch vendor brochure on CT-121, (Ex. 14).
44 Yasuhiko Shimogama, Hirokazu Yasuda, Naohiro Kaji, Fumiaki Tanaka, and David K. Harris, Commercial
Experience of the CT-121 FGD Plant for 700 MW Shinko-Kobe Electric Power Plant, Paper No. 27, presented at
MEGA Symposium, Air & Waste Management Association, May 19-22, 2003, (Ex. 15).
45 CT-121 FGD Process - Jet Bubbling Reactor, http://www.bwe.dk/fgd-ctl21.html, (Ex. 16).
46 Emission-control Technologies Continue to Clear the Air, Power, May/June 2002.
47 See Black & Veatch, First Black&Veatch/Chiyoda Wet Flue Gas Desulfurization System in North America
309
-------
Successfully Goes Operational.
48 Blankinship, Steve, Go Take a Bath, Power Engineering, October 2008, available at
http://pepei.pennnet.eom/display_article/342997/6/ARTCL/none/none/l/Go-Take-a-Bath/.
49 Chiyoda Licenses Its Flue Gas Desulfurization Technology in USA Newly for 5 Coal-Fired Generation Units,
Press Release, May 2, 2005; Chiyoda Licenses its Flue Gas Desulfurization Process in USA for Georgia Power
Owned 4 FGD Units, January 26, 2005.
50 Jonas S. Klingspor, Kiyoshi Okazoe, Tetsu Ushiku, and George Munson, High Efficiency Double Contact Flow
Scrubber for the U.S. FGD Market, Paper No. 135 presented at MEGA Symposium, Air & Waste Management
Association, May 19-22, 2003, p.8, Table 4, (Ex. 17).
51 Yoshio Nakayama, Tetsu Ushiku, and Takeo Shinoda, Commercial Experience and Actual-Plant-Scale Test
Facility of MHI Single Tower FGD, (Ex. 18).
52 Mitsubishi High SO2 Removal Experience, (Ex. 19).
53See White Bluff Station Units 1 and 2 Evaluation of Wet vs Dry FGD Technologies, Rev. 3, October 28, 2008,
prepared by Sargent & Lundy at 3-7. (Ex. 20).
Footnotes:
63 Sargent & Lundy LLC, IPMModel - Revisions to Cost and Performance forAPC Technologies, Wet FGD Cost
Development Methodology, March 2013.
Response: All of the claims in this comment have been previously addressed in confidential
expert reports in the Cinergy case216, including achievable FGD SO2 removal efficiency and the
reliability of vendor guarantees for setting emission limits, based on CEMS data, published
works, and deposition testimony of five major scrubber vendors.217 These reports and SO2
scrubber vendor depositions conclude that 98% to 99% SO2 control has been achieved and is
routinely achievable. These reports and depositions also conclude that vendor guarantees are a
reliable basis for establishing emission limits. We cannot include these expert reports and
depositions in our docket, because they are held confidential under a protective court order.
However, we do include the testimony of Dr. Phyllis Fox, one of the government expert
witnesses, in which she testifies, based on her expert report with Hal Taylor, that the removal
efficiency of wet scrubbers can reach 99%. 218 These conclusions were upheld in court rulings in
United States v. Cinergy Corp., 618 F. Supp. 2d 942, 947 and 961-962 (S.D. Ind. 2009).
216 United States and Plaintiff-Intervenors States of New York, New Jersey, and Connecticut, Hoosier
Environmental Council, and Ohio Environmental Council, Civil Action No. IP99-1693 C-M/S, United States, et al.,
v. Cinergy Corp., et al., In the United States District Court for the Southern District of Indiana, Indianapolis
Division.
217 Phyllis Fox and Hal Taylor, Expert Report of Phyllis Fox and Hal W. Taylor. Prepared on Behalf of Plaintiff
United States and Plaintiff-Intervenors States of New York, New Jersey, and Connecticut, Hoosier Environmental
Council, and Ohio Environmental Council, Civil Action No. IP99-1693 C-M/S, United States, et al., v. Cinergy
Corp., et al., In the United States District Court for the Southern District of Indiana, Indianapolis Division, August
29, 2008; Phyllis Fox and Hal Taylor, Expert Rebuttal Report of Phyllis Fox and Hal W. Taylor, Prepared on
Behalf of Plaintiff United States and Plaintiff-Intervenors States of New York, New Jersey, and Connecticut,
Hoosier Environmental Council, and Ohio Environmental Council, Civil Action No. IP99-1693 C-M/S, United
States, et al., v. Cinergy Corp., et al., In the United States District Court for the Southern District of Indiana,
Indianapolis Division, October 14, 2008, 2008.
218 United States and Plaintiff-Intervenors States of New York, New Jersey, and Connecticut, Hoosier
Environmental Council, and Ohio Environmental Council, Civil Action No. IP99-1693 C-M/S, United States, et al.,
v. Cinergy Corp., et al., In the United States District Court for the Southern District of Indiana, Indianapolis
Division. Trial Proceedings, Volume 2. Testimony of Dr. Phyllis Fox, beginning on vol. 2-302.
310
-------
First, this comment argues that a "controlled SO2 limit of 0.04 lb/MMBtu is not a realistic or
sustainable value to maintain on a long term basis when considering the normal variation in
operation that occurs at all coal-fired facilities." This is misleading as the proposed limit for
retrofits is not "0.04 lb/MMBtu" on an instantaneous basis, as suggested in this comment, but
rather is based on a "30-boiler-operating-day" (30-BOD) average.219 A boiler operating day is
any 24-hour period between 12:00 midnight and the following midnight during which any fuel is
combusted at any time at the steam generating unit."220 As explained in our FIP TSD (p. 31)
and the proposed rulemaking, days are skipped when the unit is down, as for maintenance. This
provides a margin of safety by eliminating spikes that occur at the beginning and end of outages,
when the scrubber is not operating.221 Regardless, all major scrubber vendors guarantee 99%
SO2 control and expect routine performance at this level. This level of control is confirmed
under vendor guarantees by making simultaneous measurements of SO2 at the inlet and outlet of
the scrubber.
The emission limit that we have proposed for all wet scrubbed units, 0.04 lb/MMBtu, is based on
98% control, unless the resulting controlled limit fell below an outlet SO2 level of 0.04
lb/MMBtu, in which case, we would have assumed the percentage control equal to 0.04
lb/MMBtu.222 All of the calculated 98% control levels, except for Big Brown Unit 1, were lower
than 0.04 lb/MMBtu, as shown in the following table. Thus, the limits we propose have built in
margin.
Calculation of Wet Scrubber Limits
Proposed
Baseline
Control
Margin
SO2 Limit
Level
Unit
(lb/MMBtu)
(lb/MMBtu)
(%)
(lb/MMBtu)
Big Brown 1
0.04
0.04
98
0
Big Brown 2
0.04
0.0392
97.91
0.00084
Monticello 1
0.04
0.0388
96.99
0.0012
Monticello 2
0.04
0.0387
96.83
0.0013
Coleto Creek 1
0.04
0.0383
95.65
0.0014
219 79 FR 74,822, Table 1; 74,884 (Dec. 16, 2014).
220 70 FR 39,172 (July 6, 2005).
221 79 FR 74,884 (Dec. 16, 2014).
222 Cost TSD, p. 21; 79 FR 74,884 (Dec. 16,2014). The proposed control level was calculated in the wet
FGD spreadsheet in cell E13. First, we tested to see if 98% of the uncontrolled SO2 rate would result in a value less
than 0.04 lb/MMBtu. If it did, then a control level was calculated based on a reduction to 0.04. If it didn't, then the
control level was set to 98%. The uncontrolled SO2 rate itself was calculated based on the maximum monthly SO2
rate from 2009-2014. This was calculated in the "Monthly Emissions" tab of the wet FGD spreadsheet. The
maximum monthly value was used to ensure the scrubber was designed to address the dirtiest coal burned in the
recent past.
311
-------
Recent operating experience reported by Hitachi confirms that 99% SO2 control is achievable.223
Hitachi has guaranteed many units at 99% SO2 control efficiency.224 Hitachi has explained that
99% control on an "instantaneous basis" (not even a rolling average) is routinely achieved in
Japan, where rolling averages are not allowed, and has stated that Hitachi is "comfortable with
98%>" for PRB coals.225 See also Stamper comment 0068, pp. 15-17. Alstom has guaranteed
units at 99%, concluding the new Trimble County Unit 2 "can reliably achieve a high efficiency.
99.8%) SOx removal could be expected."226 Advatech, a joint venture of URS and Mitsubishi,
likewise has built scrubbers guaranteed at 98%> to 99%>.227 Advatech reports its highest SO2
guarantee is 99.8%. Their experience list includes many units in Japan guaranteed at this
level,228 which must be met on an instantaneous basis.
Second, this comment asserts that vendor guarantees do not reflect controlled emission rates that
may be achievable on a consistent long-term basis as the unit operation varies from design
conditions." A survey of owners of FGD systems found that "[w]ith regard to FGD reliability, in
every case except for one SDA system, performance overall met or exceeded owner expectation.
Therefore, there is reason to believe that the FGD systems will perform as expected (albeit, with
some specific components that may perform more or less reliably than other components.)"229
Major scrubber vendors have testified in confidential depositions that they expect their scrubbers
to meet guarantees, which typically have a built-in vendor margin of safety.
Further, vendors offer "make-right" guarantees that cover the life of the FGD and all operating
conditions. Make-right guarantees restore a malfunctioning FGD to its design basis. Regardless,
the vendor must protect its reputation. Scrubber vendors have a vested interest in maintaining
their market position and customer base. We are not aware of any vendor that would walk away
from an FGD that did not perform according to its guarantee (even after the guarantee period has
ended).
Third, this comment argues that using the vendor guaranteed level of 0.04 lb/MMBtu is "overly
aggressive and provides no margin for normal operating conditions or long-term operation."
Variations in scrubber performance due to changes in coal quality, reagent quality, startups,
shutdowns, and malfunctions are normally accommodated in permitting by specifying a
sufficiently long averaging time, in this case, the 30-day BOD, which is specifically designed to
223 B.C. Studley and others, Recent Operating Results of the Five New Wet FGD Installations for Ameren
Corporation, Available at:
http://www.psa. mhps.com/supportingdocs/forbus/hpsa/technical_papers/Recent%200perating%20Results%20of%2
0the%20Five%20New%20Wet%20FGD%20Installations%20for%20Ameren%202011 .pdf.
224 Hitachi Wet FGD Sample of Recent Installations (showing 99% guarantees at Boswell #3, Coffeen #1,2, Duck
Creek #1, Sioux 1,2).
225 E-mail from Joseph Barba, Hitachi, to Phyllis Fox, January 31, 2007.
226 Alstom Responses to Questions, Trimble County Unit 2, October 25, 2004, BV-00000171.
227 Advatech, Double Contact Flow Scrubber, Available at: http://www.advatechllc.com/wp-
content/uploads/2013/10/advatech-double-contact-flow-scrubber-brochure.pdf.
228 Advatech, Advanced Flue Gas Desulfurization, 2005.
229 James E. Staudt, Sikander R. Khan, and Manuel J. Oliva, Reliability of Selective Catalytic Reduction (SCR)
and Flue Gas Desulfurization (FGD) Systems for High Pollutant Removal Efficiencies on Coal Fired Utility Boilers,
MEGA Symposium, 2004, p. 17.
312
-------
average out spikes. In general, averaging smooths out fluctuations in data.230 We built further
margin into the 0.04 lb/MMBtu limits, which are based on less than 98% control, except at Big
Brown Unit 1, as summarized in -the above table. Finally, margin is built into the guarantee by
the vendor.
Fourth, this comment attempts to support these claims by presenting seven years of SO2
emission data for a single facility, Spruce Unit 1. Spruce Unit 1 is a 585-MW tangentially-fired
Combustion Engineering boiler, designed to burn PRB coal, which started up in December
1992.231
The original SO2 scrubber, designed for 70% SO2 control, was started up in 1992 when the boiler
came online. The scrubber was upgraded in 2009 by Alstom to achieve 95% SO2 control at a
cost of about $9 million.232 However, the SO2 emission limit of 0.04 lb/MMBtu that this
comment is critiquing is proposed only for retrofit scrubbers, i.e., new scrubbers on existing
boilers, not upgraded scrubbers. Thus, the Spruce data is misleading, as the performance of a
scrubber upgrade may not be representative of the performance that can be achieved by a new
scrubber retrofit on an existing unit. In fact, we assumed 95% control for our scrubber upgrade
analyses, versus 98% control (not to go below 0. 04 lbs/MMBtu) for our scrubber retrofit
analyses.
Fifth, the upgraded Spruce Unit 1 scrubber came on line in early 2009, just as the SO2 allowance
market it would have benefited from collapsed, removing the incentive to operate it at its design
SO2 efficiency.233 As scrubbers are costly to operate (see O&M costs in our Cost TSD), and this
unit has very high SO2 permit limits,234 CPS Energy has no incentive to operate the upgraded
scrubber at its design basis.
The apparent poor performance of Spruce Unit 1 is not due to issues implied in this comment,
but rather the lack of any requirement or incentive to do better. The 30-day BOD data
demonstrate that in spite of these disincentives, this upgraded scrubber achieved a 30-day BOD
emission rate of 0.07 lb/MMBtu, except when the scrubber was offline, from July 16, 2011 to
August 26, 2011. The Spruce permit exempts "maintenance, startup, and shutdown periods"
from compliance with emission limits.235 In other words, the boiler can legally continue to
230 Thad Godish. Air Quality. Lewis Publishers. 2nd Ed. ,1991, p. 216, Figure 7.1; Richard W. Boubel, Donald L.
Fox, Bruce Turner, and Arthur C. Stern, Fundamentals of Air Pollution. Academic Press, 3rd Ed., 1994, pp. 41 -
43.
231 R. Peltier, J.K. Spruce Power Plant, Unit 1, San Antonio, Texas, POWER, October 15, 2008, Available at:
http://www.powermag.com/j-k-spruce-power-plant-unit-l-san-antonio-texas/7pagenunFl.
232 CPS, Corporate Sustainability Report 2010, p. 57, Available at:
https://www.cpsenergy.com/content/dam/corporate/en/Documents/Sustainability_Report_2010.pdf.
233 C. Hitaj and A. Stocking, Market Efficiency and the U.S. Market for Sulfur Dioxide Allowances, Working
Paper 2014-01, Congressional Budget Office, January, 2014. See especially Figure 1. Available at:
https ://www. google, com/url? sa=t&rct=j &q=&esrc=s&source=web&cd= 1 &ved=0CCAQFj AAahUKEwi 1 w9Hws JD
IAhVIJB4KHRAvB2Y&url=https%3A%2F%2Fwww.cbo.gov%2Fsites%2Fdefault%2Ffiles%2F113th-congress-
2013-2014%2Fworkingpapei%2F45044-SO2MarketAnalysis_l.pdf&usg=AFQj CNHYiy aL31TMimbRvluvI3DVa-
T18w&cad=ija.
234 Spruce Unit 1 Permit, Permit Numbers 18426 and PSDTX742M1, Special Conditions, Condition 8.
235 Spruce Unit 1 Permit, p. 7, Condition 15.
313
-------
operate while the control equipment is shutdown. The 0.07 Ib/MMBtu achieved at Spruce Unit 1
when the scrubber was operating is well within the range of our proposed SO2 limits for scrubber
upgrades.
We are not proposing an SO2 emission rate of 0.04 lb/MMBtu for any scrubber upgrade, as they
may be less efficient than a new retrofit scrubber. Rather, we proposed SO2 emission rates of
0.06 to 0.60 lb/MMBtu for scrubber upgrades. The 0.04 Ib/MMBtu limit that is the subject of
this comment was only proposed for new scrubbers on existing boilers or "retrofits".236 Thus,
the performance of Spruce Unit 1 is irrelevant for supporting the arguments made in this
comment.
Sixth, we note that this comment visually distorts the performance of Spruce 1 by plotting
conventional 30-day moving averages, rather than 30-day BOD averages, as proposed in the
rulemaking. Further, the graph is plotted on a fine scale to accentuate fluctuations. We rep lotted
it using a 30-day BOD average and a scale that includes the uncontrolled spike.2'' This analysis
is shown in the figure below.
• BODSQ2 Rate [Ibs/MMBtu)
30BODSO2Avg. (Ihs/MMBtu)
Spruce 1 S02 Emissions
This plot shows that Spruce Unit 1 is achieving a 30-day BOD of 0.07 lb/MMBtu after the
upgraded scrubber stabilized, except for the period from July 16, 2011 to August 26, 2011, when
the scrubber was off-line, but the boiler continued to operate. This is allowed under Spruce's
current permit to operate during "routine maintenance, startup, and shutdown". The EPA
recently issued a final rule barring these SSM exemptions in state permits.238
Finally, Sargent & Lundy omits that there are units that demonstrate our proposed limits and
control efficiencies have been and are being achieved on a routine basis. To demonstrate that the
proposed wet scrubber SO2 emission limits (0.04 lb/MMBtu) are achievable, we reviewed our
236 79 FR 74 X22. Table 1 (Dec. 16, 2014).
liJ See "Selected BOD S02 Averages-3.xlsx," Tab "Sprace 1" in our docket.
23g http://www.epa.gov/airqualitv/urbanair/sipstatus/emissions.html
314
-------
Clean Air Market Database (CAMD) to identify similar coal-fired boilers that have met a a 30
BOD average SO2 emission rate of 0.04 lb/MMBtu or better continuously for at least 1 year.
First, we downloaded all annual SO2 data from 2007-2014, inclusive. This annual data was
filtered to select only the units that met the following criteria:
• Electric utility only
• Coal burning only
• Avg. annual SO2 rate <0.045 lb/MMBtu
• Wet FGD only
• Operating time > 4,800 hours
• 12 months of reported data only
• Retrofit or upgraded scrubber
We next researched the subset of wet FGDs that met these criteria and eliminated units that used
lime wet FGD, as it can achieve a higher SO2 removal efficiency, but is generally more
expensive due to the cost of lime. We only retained one lime FGD, Milton Young 1, as it is the
only unit in our dataset that fires lignite coal. We found many units coded as using lime FGD in
the CAMD database that in fact used limestone FGD. We also eliminated scrubbers installed on
newly built units at the time of initial construction. Most of the scrubbers that met these criteria
were retrofits, the target of our analysis, but some upgrades were included. We retained these to
compare retrofit with upgrade experience.
We adopted this conservative criteria and methodology in order to quickly and easily
demonstrate that in fact there are a number of similar units operating wet scrubbers comfortably
below our proposed SO2 emission limit. In so doing, we note there are a number of additional
similarly configured units (e.g., Scherer 1) that have demonstrated their ability to operate for
extended periods of time below our proposed control efficiency, but do not meet these criteria.
We do not advocate that our criteria and methodology are the only way to demonstrate the
operational effectiveness of any particular control, scrubbers or otherwise.
We used EIA information to determine the coal type burned at each facility and eliminated all
units that did not burn PRB, lignite, or a blend of PRB or lignite with another coal type, which
are the predominate fuels burned by the subject Texas plants. We also researched each plant to
confirm the scrubber type, scrubber install date, design control efficiency, and permit limits,
where available. The resulting list of plants is:
• Iatan Unit l239
• Boswell Energy Center Unit 3 240
239 Robert Peltier, Plant of the Year: KCP&L's Iatan 2 Earns POWER'S Highest Honor, POWER, August 1, 2011,
Available at: http://www.powermag.com/plant-of-the-Yearkcpls-iatan-2-earns-powers-highest-honor/.
240 Retrofitting Boswell Energy Center, Power Engineering, September 1, 2010, Available at: http://www.power-
eng.com/articles/print/volume-114/issue-9/Features/retrofitting-boswell-energy-center.html. See also:
http://www2.epa.gov/enforcement/minnesota-power-settlement (requiring SO2 emission rate of no greater than
315
-------
• James H. Miller Units 1, 2, 3, and 4241
• Jeffrey Energy Center Units 1, 2, and 3 242
• Milton Young Unit B1243
• Pleasant Prairie Unit 1 and 2244
• Sam Seymore (Fayette) Unit 1, 2,245and 3246
• Scherer Units 2, 3, and4247
We prepared graphs of these units showing the 30-day BOD SO2 data superimposed on the
measured daily data in lb/MMBtu. These graphs are included in our docket in the file "Selected
BOD SO2 Averages-3.xlsx." This file contains tabs corresponding to each of the above units,
and a "Read Me" tab that explains our methodology.
0.030 lb/MMBtu at Boswell Units 3 and 4) and Consent Decree at http://www2.epa.gov/sites/production/files/2014-
07/documents/minnesotapower-cd.pdf ("Minnesota Power shall continuously operate an FGD device at Boswell
Unit 3 such that the unit achieves and maintains a 30-day rolling average emission rate for S02 no greater than 0.03
lb/MMBtu." and "No later than May 31, 2016, Minnesota Power shall continuously operate a new FGD device at
Boswell Unit 4 such that the Unit achieves and maintains a 30-day rolling average emission rate for SO2 no greater
than 0.03 lb/mmBTU."
241 ADVATECH, Plant Miller - Wet FGD Retrofit Project, Available at: http://www.advatechllc.com/proiects/air-
aualitv-control-solutions/acid-gas-control/acid-gas-proiect-descriptions/southern-companv-miller/. Title V Permit
at: http://www.icdh.org/EH/AnR/AnR13.aspx. Consent Decree at: http://www2.epa.gov/enforcement/alabama-
power-company-clean-air-act-settlement.
242 Westar Energy Sulfur Dioxide Removal Project on First of Jeffrey Energy Center's Generating Units Fully
Operation, August 11, 2008, Available at:
http://www.energvcentral.com/generationstorage/fossilandbiomass/news/vpr/5341/WESTAR-ENERGY-SULFUR-
DIOXIDE-REMOVAL-PROJECT -ON-FIRST-OF-JEFFREY-ENERGY-CENTER-S-GENERATING-UNITS-
FULLY-OPERATIONAL: Gail Reitenbach, WATER AWARDE: Jeffrey Energy Center's Constructed Wetland
Treatment System, POWER, August 1, 2014, Available at: http://www.powermag.com/ieffrev-energy-centers-
constructed-wetland-treatment-svstem/?pagenum= 1: and Westar Energy, Jeffrey Energy Center Wetlands,
Available at: https://www.westarenergy.com/jeffrey-energy-center-wetlands.
243 Burns & McDonnell, Milton R. Young Units 1 & 2, 2015, Available at:
http://www.burnsmcd.com/Resource /Proiect/2234/ProiectPdf/Milton-R-Young-Units-l-amp-2.pdf: MET, Wet
FGD, Available at: http://www.met.net/wet-fgd-technologies-lime.aspx.
244 Steven Gebhart and others, Pleasant Prairie Power Plant Air Quality Control Upgrade Project, Pleasant Prairie,
Wisconsin, POWER, October 15, 2007, Available at: http://www.powermag.com/pleasant-prairie-power-plant-air-
quality-control-upgrade-project-pleasant-prairie-wisconsin/ See also:
http://www.pleasantprairieonline.com/services/communitydevelopment/powerplant.asp
245 C. Frazer and others, Fayette Power Project Unit 3 FGD Upgrade: Design and Performance for More Cost-
Effective S02 Reduction, Mega Symposium, 2010, Available at: http://www.babcock.com/library/documents/br-
1845 .pdf; LCRA Adds Scrubbers to Clean Sulfur Dioxide from Plant Emissions, August 1, 2011, Available at:
http://www.statesman.com/news/news/local/lcra-adds-scrubbers-to-clean-sulfur-dioxide-from-l/nRc8M/.
246 Austin City Council, Audit Report, Fayette Power Project Scrubber Costs, October 3, 2006, Available at:
Powerplant Upgrades Test New Technology, ENR Texas & Louisiana, April 18, 2011, Available at:
https://austintexas. gov/sites/default/files/files/Auditor/au06113 .pdf:
http://texas.construction.com/texas construction proiects/2011/0418 powerplantupgrades.asp: Fayette Power
Project, Available at: http://www.lcra.org/energy/electric-power/facilities/fayette-power-
project/Documents/fay ette_power_project_2013_03_22.pdf.
247 ADVATECH, Plant Scherer - Wet FGD Retrofit Project, Available at: http://www.advatechllc.com/proiects/air-
aualitv-control-solutions/acid-gas-control/acid-gas-proiect-descriptions/southern-companv-scherer/: ADVATECH,
Double Contact Flow Scrubber, 2011, Available at: http://www.aecomprocesstechnologies.com/wp-
content/uploads/2012/03/advatech-dcfs-brochure.pdf
316
-------
The performance of each scrubber in our data set is summarized in the file, "Selected scrubber
retrofit efficiencies.xlsx." The "Efficiencies" tab within this file reports vendor guaranteed
control efficiency, achieved control efficiency, and the maximum 30-day BOD, excluding
shakedown periods following initial startup of scrubber retrofits, and other items. This
information supports our proposed SO2 limits for both retrofits and upgrades. The 13 retrofit
units have guaranteed control efficiencies of 95% to 99%, with eight of them guaranteed at 98%
to 99%. These eight units are achieving 98% to 99% SO2 control, with the exception of Scherer
Unit 4, when calculated using our very conservative method explained in the "Read Me" tab.
We note that compliance with the vendor guaranteed SO2 control efficiency is typically
determined by simultaneously measuring scrubber inlet and outlet SO2 emissions. As CAMD
does not report and facilities generally do not measure inlet SO2 CEMS data, our estimated
control efficiencies are based on pre-scrubber retrofit SO2 and post-scrubber retrofit 30-day BOD
data, as explained in the "ReadMe" tab noted above. None of the eight units that are achieving
98%) to 99% SO2 control have permit limits that require operation at this level. However, several
have entered into consent decrees that do require control efficiencies approaching these values.
The maximum 30-day BOD, over the period of record when the scrubbers were operating,
excluding scrubber "shakedown" periods, for retrofit FGDs designed to achieve 98% to 99% SO2
control range from 0.02 lb/MMBtu at Iatan Unit 1 to 0.08 lb/MMBtu at Scherer Unit 3. Three of
the units have achieved a maximum 30-day BOD equal to or less than our proposed SO2
emission limit for scrubber retrofits of 0.04 lb/MMBtu:
• Scherer Unit 2: 0.01 lb/MMBtu based on 485 data points
• Iatan Unit 1: 0.02 lb/MMBtu based on 2,004 data points
• Boswell Energy Center: 0.03 lb/MMBtu based on 1,881 data points
All of these units burn PRB coal and are operating at these levels without permit limits that
require compliance at achieved levels. The best performers are discussed below.
Scherer Unit 2
Plant Scherer consists of four 818 MW, Combustion Engineering boilers that became operational
in 1982, 1984, 1987, and 1989. All of the units burn PRB coal. All four units were retrofitted
with wet scrubbers that became operational in late 2010 to 2014. All of the units have
continuously operated below 0.01 lbs/MMBtu for months at a time. The maximum 30-day BOD
at Unit 2, after scrubber shakedown, was 0.01 lb/MMBtu, with the majority of the 30-day BOD
data points below this level. The compliant data set includes 423 continuous 30-day BOD data
points, between March 2014 and June 2015, all less than or equal to 0.01 lb/MMBtu. See the
figures below for Unit 2. The second figure shows the period just after the scrubber became
operational, plotted on a finer scale to better display the scrubber's performance:
317
-------
Seherer 2 S02 Emissions
5
2
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
BOD S02 Rate (fbs/MMBlu)
-30 BOD S02 Avg. {Ibs/MMBtu}
. * *~ \ *
§o§ooooo§oS§2SS22S222rtHrtrtrtrt«
OrtfSmrtrtNOHNfflOiHNOrtNrilOrtMOrtNMOHN
*4-40000000<-
^rH^r»ifN»rvir-jfsifNirjr^rsirrtrrirrirripr»iriirifrimm*t^*J-^^*»"rtTr'?iALr>tnuOLnu-i»/,itn
5c!CIS9 CC15.CClCl$
3" »-t r:v?£'pH*{55'o3'o3'e5"{r!'
-------
emissions and as a requirement for the air permit for Iatan 2, under a settlement agreement with
the Sierra Club. 248
The design of the air quality control system for both existing Unit 1 and new Unit 2 was very
similar, resulting in similar performance. "The entire post-combustion AQCS [air quality control
system] for Iatan 1 is virtually identical to that of the new Iatan 2... .The wet FGD systems, using
the same size vessel on both units, routinely operate at 99% removal, producing SO2 levels in the
exhaust gas well below the 0.06 lb/MMBtu permit limits. Each vessel is equipped with five
recycle pumps but usually runs only three or four, depending on operating conditions." Thus,
while the Unit 1 scrubber is a retrofit, its performance is nearly identical to scrubber on new Unit
2.249
This is the best performing wet scrubber retrofit in the data set based on 6 years of continuous
operation. The maximum 30-day BOD at this unit, after scrubber shakedown, was 0.02
lb/MMBtu, with the majority of the 30-day BOD data points far below this level. The compliant
data set includes 2,004 continuous 30-day BOD data points, between April 2009 and June 2015,
all less than or equal to 0.02 lb/MMBtu. See the figures below. The second figure shows the
period just after the scrubber became operational, plotted on a finer scale to better display the
scrubber's performance. This is much better performance than is required by Iatan Unit l's
operating permit, which limits SO2 emissions to 0.07 lb/MMBtu, based on a conventional 30-day
rolling average.250
2
2
"ltT
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Iatan 1 S02 Emissions
BOO S02 Rate (Ibs/MMBtu}
-30 BOD S02 Avg. (tbs/MMBtu)
ft
Mm**
OfflMMKKOBfflffimfflfflOmmfflOOOOOOOOOHiHr
5" oS135" oS1 oS"o" o1o1 ^ ^ ^ o& 55 e5" 53" oS"o? ^ ^^ rJ ?? r? r? ^ ^ ^ oS1 o3" of g> S" o^" r? rJ ^ ^ r*? iS1 iB1 S
(IfOHNMr
248 Empire District Electric Co -
Emissions." Available at:
'10-Q' for 3/31/12 - R13, "Compliance Plan" and "SO2
http://www.secinfo.com/dl lMXs.plZm2. Is.htm: State of Kansas, Air Quality State Implementation Plan, Regional
Haze, October 26, 2009, Available at: http://www.kdlieks.gov/bar/reghaze/KDHERegHaze.pdf.
249 Robert Peltier, Plant of the Year: KCP&L's Iatan 2 Earns POWER'S Highest Honor, POWER, August 1, 2011,
Available at: http://www.powennag.com/plant-of-the-Yearkcpls-iatan-2-earns-powers-liighest-honor/.
2511 Kansas City Power & Light, Iatan Generating Station, Permit No. 012006-019B.
319
-------
latan 1 SO2 Emissions
5
BOD S02 Rate (Ibs/MMBtu)
-30 BOD $02 Avg. flbs/MMBtu)
OOCQWCICIOClO
o o o o e» o o o
r? ©J *?" fN SfuS" t-I
SSSS&5SS.
— -H-IOOOOOC
¦OOO0 0»-i^C'OOOOOOO--t'-iOCiOOCiOOO^i'~iOOOOCiOC»0
Of note, the 30-day BOD SO2 data for latan Unit 1 does not include the spikes found at some of
the other units we reviewed. latan's operating permit does not include any exemptions from
permit limits for startups, shutdowns, and malfunctions (SSM), while plants with frequent spikes
include this exemption. The latan Unit 1 permit limits for all pollutants (except opacity)
explicitly "include|ing] periods of start-up, shutdown and malfunction.. ,"251 Other plants with
SSM exemptions in their permits have no obligation to operate their scrubbers during SSM
events, resulting in the boiler continuing to run while the scrubber is offline for maintenance,
creating spikes that are not related to scrubber performance.
latan 2, a new boiler and scrubber, displays a similar level of performance. Because it's scrubber
pre-scrubber data is not available:
251 latan Generating Station Permit Number: 012006-019A, July 13,2007, p. 5, Note to Special Condition 2.E.
320
-------
• BOD SO? fiats (lbs/MMBtu|
30 BOD 502 Avg. flhs/MMBOi)
latan 2 S02 Emissions
©0©OOrH^r*r*r*^H^fSrsirJrjrjrjrJryc4rJpjrgr^rgmrniTOmi*iirorrt««ri>rrtrflmrt^"tf*f'»*TTr^r<,f*i*rTr*3-*t*TLrsu"i!''rvi/iini'/vu"iu"ti"n
drtHOOONNlSrtW-(00©N(NrjHHHOOO(SNNHHrtOOmNNiNHrtrlOOrtNNHHrtOOO(n(NNNHHOOONNNNr(HOOON
Oo S? *H"r? «?
-------
The Consent Decree limit (0.03 lb/MMBtu 30-day rolling average) is much lower than the SO2
limit proposed for wet FGD retrofits (0.04 lb/MMBtu 30-day BOD average).256 A 30-day rolling
average is more conservative (i.e., harder to meet) than the 30-day BOD average proposed in this
rulemaking. Further, 0.03 is lower than 0.04. Thus, the Boswell Energy Center Consent Decree
requires an SO2 emission limit that is more aggressive than any of the limits proposed by EPA to
satisfy reasonable progress. This 0.03 lb/MMBtu 30-day rolling average limit is currently being
achieved.
The SO2 CEMS data for Boswell Unit 3 is shown in the figures below. The second figure shows
the period just after the scrubber became operational, plotted on a finer scale to better display the
scrubber's performance. These figures demonstrate that the proposed 0.04 lb/MMBtu 30-day
BOD for retrofit scrubbers under the Texas FIP has been met with a significant margin at this
unit for over five years or for 1,881 individual continuous 30-day BOD data points.
256 79 FR 74,822 (Dec. 16, 2014).
322
-------
Boswell 3 S02 Emissions
5
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
• I
• BOD 502 Rate (Ibs/MMBtii)
. *fc* * "t
4.
At
30 6 00 SOJ Aug. [1 hs/M MBlu)
til' '•&*
MLiM
J ft
*P\
•r >
W- lw&ri
** *, *. j
1
*
* 1 **
!•
< M N (S r
Nmwwrtimiflmn
f^r^rwiurbvftfcrtiflvnin
5^; _
QfHr-flOOOOOOi—*«-
«NHNOMOH(flHrj"iMONO«rtrtNHrSpNOrtOHNHNONOrtOf? w" r?f ^ t£? «&¦ ^ tS IB1 oo 3v ^ M'fN' n? IS12? ^
-------
S02 prior to and after scrubber installation. This is consistent with its 99% SO2 guarantee.257
The maximum 30-day BOD is 0.03 lb/MMBtu since stable operation of the scrubber started in
November 2009. This is less than required for retrofit units under the proposed Texas FIP. The
majority of the BOD values are substantially lower than this maximum.
Jeffrey Energy Center (Scrubber Upgrade)
Jeffrey Energy Center consists of three 720 to 750 MW, pulverized coal, dry bottom ABB
Combustion Engineering boilers that are permitted to combust either liquid and/or solid fossil
fuel. The existing wet scrubbers on Jeffrey Energy Center Units 1-3258 were rebuilt and
upgraded in 2008 to 2009 by Burns & McDonnell to achieve a minimum of 95% SO2 control.
The project included replacing the module internals including mist eliminators; adding forced
oxidation; eliminating flue gas bypass; increasing limestone grinding capacity; and adding a
gypsum dewatering system. Induced draft fans were upgraded and ductwork was replaced. The
existing stack was modified for wet-stack operation.259
These units are operating under a Consent Decree260 that requires that they achieve 97% SO2
reduction and an SO2 emission limit of 0.07 lb/MMBtu, both on a 30-day rolling average basis.
The operating permit also includes an SO2 limit of 0.8 lb/MMBtu for each unit, no averaging
time specified, and no SSM exemptions. As of January 1, 2014, they also must meet a SO2
emission limit of 0.15 lb/MMBtu as a 30-day rolling average, excluding periods of startup,
shutdown and malfunction under a Regional Haze Agreement between Westar Energy, Inc. and
KDHE.261 They are currently meeting the percent reduction requirement.
Nearly five years of continuous operating data demonstrate that these three units have
consistently exceeded their 95% SO2 removal guarantee, achieving 97% to 98% SO2 control
relative to the original scrubber baseline.
The highest 30-day rolling BOD at these units since the upgraded scrubbers achieved stable
operation in 2008 to 2009 is 0.06, 0.05, and 0.09 lb/MMBtu, at Units 1, 2, and 3 respectively.
257 The guaranteed SO2 efficiency is calculated from the scrubber inlet and outlet SO2 concentrations. The CAMD
CEMS data is only for the outlet. Thus, we estimated the control efficiency from the average pre-scrubber and post-
scrubber emissions, as explained in the "ReadMe" in the file, "Selected scrubber retrofit efficiencies.xlsx" in our
docket
258 Units 1, 2, and 3 are 720 to 750 MW pulverized coal dry bottom ABB Combustion Engineering boilers that
started operations in 1978, 1980, and 1983, respectively. They burn PRB coal and were equipped with 60%
efficient Combustion Engineering SO2 scrubbers on startup.
259 Burns & McDonnell, Jeffrey Energy Center Units 1, 2 & 3 Upgrade Project, 2015, Available at:
http://www.burnsmcd.com/Resource /Proiect/1244/ProiectPdf/Jeffrev-Energv-Center-Units-l-2-3-Upgrade-
Proiectpdf.
260 Consent Decree, United States of America and State of Kansas v. Westar Energy, Inc., U.S. District Court for the
District of Kansas, Civil Action No. 09-CV-2059 JAR/DJW, Lodged with federal court: 75 FR 4,847 (January 15,
2010). Available here: http://www2.epa.gov/enforcement/consent-decree-westar-energy-inc-civil-action-no-09-cv-
2059-jardjw.
261 Kansas Department of Health and Environment (KDHE), Air Emission Source Class I Operating Permit, Jeffrey
Energy Center, January 28, 2002, p. 25, Condition 6 and p. 28, Condition 14.
324
-------
This performance is generally better than the proposed SO2 BART limits for scrubber upgrades
(0.06-0.60 lb/MMBtu).262
Startup, Shutdown, and Maintenance Exemptions
Many of the units that we did not specifically discuss, including all of the Miller units, and
Scherer units 3 and 4, have large spikes that correspond to periods when the boiler was
operating, but the scrubber was off-line. See, for example, "Selected BOD SO2 Averages-
3.xlsx," Tab: Scherer 4, the period 8/15/12 to 9/20/12. These (and other) plants with similar
spikes have permits that exempt periods of startup, shutdown, and malfunction (SSM), allowing
the boiler to continue to operate when the control equipment is offline.
The Title V Permit for the Scherer units263 exempts many operating conditions, allowing the
control equipment to shutdown while the boilers continue to operate. The Scherer permit limits
SO2 to 95% control on a 30-day rolling average basis, except it allows the boilers to operate
without scrubbers during: (1) restarts; (2) periods of startup; (3) periods of shutdown; (4) periods
of scheduled and/or preventative maintenance of control technology equipment if such
maintenance cannot reasonably be performed during a scheduled outage; (5) periods of
malfunction of the EGU or control technology; and (6) during certain testing and research
periods. The scrubber on Scherer Unit 2 came on line in January 2014 and has only operated
post-shakedown for 485 days. Its performance thus far is spectacular (maximum 30-day BOD =
0.01 lb/MMBtu), indicative of what the other longer-operating Scherer scrubbers could achieve,
but for the SSM exemption periods when the scrubbers go offline, and the boilers continue to
operate.
The spikes that diminish performance are generally not due to limitations of the scrubbers, but
rather the failure of the facility to properly schedule adequate and timely maintenance to limit
SO2 emissions. All scrubbers require periodic maintenance. A facility without SSM exemptions
will plan maintenance to avoid scrubber shutdowns during boiler operation, as demonstrated by
the performance of Iatan Unit 1, discussed above.
Similarly, the Title V permit for Miller Units 1 - 4264 limits SO2 emissions to 1.2 lb/MMBtu (a
very high limit, roughly double to its coal sulfur content265), with compliance only by stack test.
The SO2 permit condition includes this exception: "The FGD may be placed in bypass mode,
provided all SO2 emissions limitations are met." In other words, it can be turned off at will, as
262 79 FR 74,822, Table 1 (Dec. 16, 2014).
263 Scherer Steam-Electric Generating Plant, Part 70 Operating Permit, Permit Number: 4911-207-0008-V-03-0,
May 8, 2012, Condition 3.4.14.
264 Alabama Power Company, J. H. Miller, Jr. Steam-Electric Generating Plant, Major Source Operating Permit,
For Units 1 & 2, p. 16, Condition 2: "The FGD may be placed in bypass mode, provided all SO2 emissions
limitations are met." For Units 3 & 4, p. 23, Condition 2: "The FGD may be placed in bypass mode, provided all
SO2 emissions limitations are met." Permit Available at: http://www.icdh.org/EH/AnR/AnR13.aspx.
265 FromEIA-923, available here: http://www.eia.gov/electricity/data/eia923/
The 2014 sulfur content ranges from 0.23 to 0.31% and the coal heat content is approx. 17.7 MMBtu/ton.
Converting to uncontrolled SO2: (0.31% sulfur) X (1 ton/17.7 MMBtu) X (2,000 lbs/ton) X (2 molecules SO2/I
molecule S) X (1/100%) = 0.7 lbs/MMBtu.
325
-------
long as it burns low sulfur coal. As the typical coal that is burned at this facility contains about
0.7 lb/MMBtu of SO2, this condition effectively allows the scrubbers to shutdown while the
boilers continue to operate. This pattern is seen in the CEMS data for all four of the Miller units,
which have similar SSM exemptions. Thus, we do not report the maximum 30-day BOD for any
of these units in the "Efficiencies" tab of "Selected scrubber retrofit efficiencies.xlsx" document,
as the CEMS data are not representative of scrubber performance.
The scrubbers on Miller Units 3 and 4 additionally have to meet a 30-day rolling average SO2
control efficiency of 95%, which drops to 90% if coal with more than 1% sulfur is burned, to
comply with a consent decree. However, the 30-day rolling average removal efficiency "shall
exclude all emissions that occur during any period of malfunction.. ,".266 Further the Miller
scrubbers are designed to achieve 98% SO2 control, while the permit only requires that they
achieve 90% to 95% SO2 control, depending upon coal quality. The CEMs data includes many
spikes when the boiler is operating but the scrubbers were not. Thus, the apparent poor
performance of these scrubbers is not due to poor scrubber performance, but the failure of the
permit to adequately limit emissions. The data summarized above indicates that similar
scrubbers achieve much better performance when there are no SSM exemptions.
Finally, some of the spikes appear to be due to unspecified errors. For instance, in the Miller 1
data, the daily SO2 average for 6/22/14 is 3.25 lb/MMBtu. This is likely a bad data point, as it is
way above the SO2 baseline from the PRB coal the Miller units have historically burned. Other
graphs have similar spikes that exceed the sulfur content of coal quality.
Thus, we disagree with S&L that our proposed scrubber retrofit SO2 emission limits are not
realistic or maintainable on a long-term basis. We agree with Ms. Stamper that it may be
possible that many of the scrubber retrofit units can achieve greater control efficiencies than we
proposed. Greater control efficiencies would result in a more favorable cost effectiveness (lower
$/ton) and more visibility improvement. This is another area in which we strove to be
conservative and demonstrate that even with many conservative cost assumptions, our analyses
indicate that the scrubber retrofits we proposed are cost effective.
Comment: [Sargent & Lundy (0061) p. 24] S&L reviewed the approach EPA takes in its
proposed disapproval of Texas's regional haze SIP and proposed FIP, including EPA's
determination of costs for retrofit FGD scrubbers and scrubber upgrades, and EPA's evaluation
of annual SO2 emission reductions. Our analysis identifies several areas where EPA overstates
the cost-effectiveness ($/ton of SO2 removed) of the scrubber upgrades and control technology
retrofits that EPA would require in its FIP. As discussed in this analysis, cost-effectiveness is
influenced by two variables: the total annualized cost of the upgrades or retrofit controls ($/yr)
and the corresponding reduction in annual SO2 emissions (tons per year "tpy"). EPA's approach
does not accurately calculate either variable. In its approach, EPA understated the annualized
cost of the control systems and overstated the tons of SO2 that would be removed by its FIP-
imposed scrubber retrofits and upgrades.
Response: For the reasons described in our individual responses to S&L's and Luminant's
266 Miller Permit, pp. 22-23, Conditions 2 and 3.
326
-------
specific comments, we disagree that our cost analyses were significantly flawed. We have made
some changes in response to these and other comments we have received. We find that the cost
effectiveness of the scrubber upgrades is validated and the technical record demonstrates that the
controls are very cost effective in comparison to typical SO2 controls we have approved or
required in FIPs. We find that the scrubber and retrofits are cost effective and lie well within a
range we have found to be cost effective in previous approvals and other FIPs we have required.
The issues raised generally by S&L are addressed more specifically in our responses to other
comments.
Comment: Actual visibility changes associated with scrubber upgrades at Limestone will
be lower than EPA's projections. [NRG (0078) p. 12]
NRG argued that even the minuscule visibility changes described in EPA's proposal are
overstated, as they rely on an assumed emission rate of 0.08 lbs SCh/MMBtu from each of the
Limestone units. As documented in the attached confidential report by Sargent & Lundy, it is
not reasonable for EPA to require the units to achieve the proposed 0.08 rate on the basis of a
scrubber upgrade. Any visibility improvement attributed to the scrubber upgrade would need to
be based on a higher limit that reflects the constraints imposed by the current design and
arrangement of the existing scrubbers and expected increases in the sulfur content of the plant's
fuel supply, as described in the attached confidential report by Sargent & Lundy.41 Thus,
scrubber improvements at Limestone would achieve far lesser reductions than EPA has
estimated, further diminishing the already-imperceptible visibility changes that EPA has
modeled in support of the proposed new limits.
Footnote:
41 Sargent & Lundy LLC, S&L Comments to EPA Assessment of Limestone FGD Capability (Apr. 15, 2015)
("Sargent & Lundy Report"), at 9.
Response: The items that NRG summarizes in this comment, and the information provided in
its attached Sargent and Lundy (S&L) report, regarding our scrubber upgrade costs are detailed
in a separate comment package submitted under the Confidential Business Information (CBI)
provisions of 40 C.F.R.§ 2.203(b). Within those CBI comments, S&L also provides its own cost
analyses for upgrading NRG's scrubbers. We are unable to respond to many of these
summarized scrubber upgrade comments here with any specificity, because doing so would
involve citing and discussing in detail items that Luminant has claimed as CBI. Accordingly,
many of our responses to the scrubber upgrade comments are contained within a separate
document that is not a part of our posted docket, but will be available for review by NRG. The
responses to comments that do not contain CBI information are contained within this document.
We disagree with NRG's assertion that scrubber upgrades at the Limestone facility are not
capable of achieving the SO2 emission limits we proposed. Our reasoning is detailed within
these same CBI response to comments.
Comment: There are significant technical issues with the proposed emissions limits.
[EEI (0076) p. 5-7]
327
-------
EEI stated that the unit specific limits in EPA's proposed FIP also present significant technical
concerns and may not be achievable in all cases, or would require more significant operational
and engineering challenges than EPA has considered.
According to EEI, for example, EPA is proposing that Southwestern Public Service's Tolk
Facility retrofit dry flue gas desulfurization (FGD) equipment in order to meet a proposed sulfur
dioxide (SO2) emissions limitation of 0.06 lbs/MMBtu on a 30-day rolling average. Id. at
74,822. The scrubber retrofits would not be in place by the end of the first planning period in
2018. Further, it also is not clear that the Tolk facility will be able to achieve a SO2 emissions
limitation of 0.06 lbs/MMBtu. The installation of a dry FGD system at the Tolk Facility will
stress the already scarce local water resources of western Texas, as dry FGD systems (in contrast
to their name) require significant amounts of water in order to run effectively and operate
efficiently. The Tolk facility is in an area that already is stressed for water use; adding a dry
FGD system will introduce a significant new draw on the water resources near the facility that
might not be sustainable. EEI urges that EPA seriously consider the significant technical
comments filed by Southwestern Public Service regarding issues and concerns about the impacts
of the proposal on its Tolk facility and the surrounding community.
EEI noted that EPA also is requiring that Luminant retrofit several facilities with FGD systems
and upgrade other FGD systems in order to meet more stringent SO2 limits at Big Brown 1-2,
Martin Lake 1-3, Monticello 1-3 and Sandow 4 units. Id. Similar to Southwestern Public
Service's Tolk facility, the retrofitted equipment cannot be completed within the first planning
period. Moreover, the stringent SO2 limits that EPA is proposing for Luminant's units are neither
consistent with "reasonable progress" nor reflective of the state's consideration of several factors,
including available technology and costs that are specified in EPA regulations. See 40 C.F.R. §
51.301. Specifically, EPA's proposed emissions rate of 0.041bs. SCh/MMBtu for many of
Luminant's existing units is as stringent as the lowest achievable emission rate (LAER) for new
units, rather than values typically applied to existing units, and is more stringent than Best
Available Retrofit Technology (BART) limits that EPA has recently approved for existing
EGUs. EEI urges EPA to consider the significant technical comments filed by Luminant
regarding issues and concerns about the impacts of the proposal on their facilities.
Response: This comment includes a number of issues raised in more detail elsewhere. Please
see our responses to those more specific comments. As to the achievability of a 0.06 lb/MMBtu
SO2 emission limit on a 30-day rolling average by the Tolk dry scrubbers, see our response to
similar comments in our previous Oklahoma FIP.267 In that FIP, we finalized the same emission
limit of 0.06 lbs/MMBtu on a 30 BOD average for 6 coal fired units that burned coal that is very
similar to Tolk's and were of a similar size and configuration. We justified those limits based on
the same SDA technology (although Xcel is free to choose any technology that meets our
proposed SO2 limit), using a combination of industry publications and real world monitoring
data. Much of that information is summarized in our response to a comment to that action268 and
in our TSD.
267 76 FR 81728.
268 Response to Technical Comments for Sections E. through H. of the Federal Register Notice for the Oklahoma
Regional Haze and Visibility Transport Federal Implementation Plan, Docket No. EPA-R06-OAR-2010-0190,
12/13/2011. See comment and response beginning on page 91.
328
-------
Comment: Commenter 0054-54 stated that scrubbers are 99 percent efficient and should have
SO2 limits more stringent than 0.04 to 0.06. The commenter stated that 0.2 for Shadow 4 and 0.6
for San Miguel is unacceptable. The commenter suggested with these changes the annual SO2
reduction would probably increase to 500,000 tpy which would benefit three states (Texas,
Oklahoma, and Arkansas).
[Earthjustice (0067) p.25] Earthjustice et al., stated that Texas underestimated the effectiveness
of scrubbers. Texas's analysis assumed that scrubbers would reduce a source's sulfur dioxide
pollution by 90%. But as EPA shows, scrubbers can achieve 95%-98% control efficiencies. Id.
at 74,840. As a result, Texas overestimated the costs and underestimated the visibility benefits
of scrubbers.
Response: We agree with the commenters that in some cases, SO2 scrubbers are capable of a
higher control efficiency. For the reasons we have outlined in our responses to specific
comments we received on our proposed control level, and the need for the units in question to
attain our proposed emission limits on the basis of a 30 day boiler operating average, we believe
that our proposed level of control is appropriate.
Comment: EPA Assumed Too High of SO2 Removal Efficiencies with DSL
[Stamper (0068) p. 17]
Stamper stated that the EPA evaluated DSI at control effectiveness levels of 50%, at 80% for
those EGUs equipped with electrostatic precipitators (ESPs), and at 90% for those EGUs
equipped with baghouses. FIP TSD at 12. The example Complete Cost Estimate for a DSI
System in the Sargent & Lundy IPM DSI Cost Development Methodology does indicate
maximum SO2 removal targets of 80% using milled trona for units with ESPs and 90% using
milled trona for units with baghouses.58 However, the Sargent & Lundy documentation also
identifies several potential issues that can affect the SO2 removal efficiency of DSI, including the
particulate control technology (with baghouses achieving higher SO2 removal efficiencies "by
virtue of the filter cake on the bags, which allows for a longer reaction time between the sorbent
solids and flue gas"), the temperature of the flue gas, the unit size, and the uncontrolled SO2
rate.59 Moreover, the Sargent & Lundy documentation states that, with an ESP, 40% to 50% SO2
control is typically achieved without an increase in PM emissions, and 10-15% control is
typically achieved in a baghouse without an increase in PM emissions.60
In fact, Stamper stated that it appears that Sargent & Lundy suggests the use of the higher SO2
removal efficiencies in projecting the costs for DSI to account for all of the variables that impact
the SO2 removal efficiency of a DSI system, because the sorbent feed rate is the "bulk" of the
DSI system costs.61 However, as stated in the Sargent & Lundy documentation, a coal-fired
boiler will not likely achieve 80% to 90% SO2 removal efficiencies with the use of DSI without
an increase in PM emissions. Such an increase in PM emissions would likely trigger new source
review permitting requirements and best available control technology (BACT), which would
require the replacement of ESPs with baghouses or, for those EGUs with baghouses, upgrades to
existing baghouses, to meet BACT. It does not appear that EPA took into account the costs for
329
-------
increased particulate matter control in its evaluation of DSI at 80-90% SO2 removal efficiencies.
Thus, the true cost effectiveness of DSI to achieve 80-90% should be based on higher capital and
operational costs.
Stamper stated that the EPA has acknowledged these issues in determining the cost effectiveness
of DSI technology to control SO2 emissions, and has requested further information from the
owners of EGUs that have performed DSI testing. Cost TSD at 4-7. In fact, Luminant has found
that a very high sorbent injection rate was necessary to achieve "modest" levels of SO2 removal,
and the sorbent injection had negative impacts on plant operation. Cost TSD at 8. As a result,
EPA determined that DSI was not a feasible alternative for the Luminant facilities (Big Brown
and Monticello). Id. Yet, in spite of these unknowns, EPA evaluated DSI at 80-90% SO2
removal efficiencies which EPA is not sure can be achieved at these units, and EPA also failed to
account for the costs of improved PM controls that would likely be necessary under the NSR
permit requirements triggered with the increased particulate emissions resulting from use of DSI
to achieve 80-90% SO2 control.
According to Stamper, even the amount of sorbent to be injected to meet 50% SO2 removal
efficiency could cause particulate emission increases and opacity problems at EGUs, especially
those with ESPs such as at the Welsh units.62 Thus, it is questionable whether 50% removal can
be achieved without causing other operational problems or increases in PM emissions at EGUs
with ESPs.
Stamper stated that the other potential impacts from a DSI system include the impacts on the fly
ash handling system, which may need modifications or to be operated more frequently to handle
the increased fly ash.63 While EPA included costs for increases in fly ash disposal (Cost TSD at
8-9), it is not clear that EPA also included costs for upgrades to fly ash handling systems to
handle increases in fly ash.
Further, as stated by EPA, Stamper noted that the addition of trona in the fly ash increases the
water solubility of the waste, and an upgraded landfill may be required as a result. Cost TSD at
10. Indeed, a recent study of the physical and chemical characteristics of fly ash produced from
plants with trona injection found that trona injection increased fly ash solubility, pH, and
leachability of anionic elements such as fluoride, sulfate, and chloride, as well as arsenic and
selenium.64 EPA should evaluate this issue further and include the costs for any necessary
landfill upgrades in its DSI cost analyses.
According to Stamper, given all of the unknowns with the costs and efficacy of DSI to reduce
SO2 at the unscrubbed Texas EGUs evaluated by EPA, it is imperative that EPA adequately
justify its assumptions about the technical feasibility of achieving the assumed 80-90% control
efficiency with DSI. It is not clear that DSI could reliably achieve 80-90% removal (or even
50% SO2 removal) without upgrades to PM controls, fly ash handling, and/or landfills. Without
further documentation, EPA should not have evaluated DSI at an SO2 removal efficiency any
higher than 50%.
Footnotes:
58 Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC Technologies, Dry Sorbent Injection
for SO2 Control Cost Development Methodology, March 2013, at 7. (Ex. 23).
330
-------
59 Id. at 1-2.
60 Id. at 2.
61 Id. at 2.
62 Not only can the sorbent injection increase particulate emissions due to increased particulate loading, but sorbent
injection can changes the resistivity of the particulate emissions which could alter the ESP's ability to remove
particulate emissions. See Fischer, Diane and Preston Tempero, Black&Veatch, Early Lessons Learned from
Implementation of Dry Sorbent Injection Systems, 2012, at 4 (Ex. 24).
63 Id. at 4.
64 Jianmin Wang, et. al., Leaching Behavior of Coal Combustion Products and the Environmental Implication in
Road Construction, A National University Transportation Center at Missouri University of Science and Technology,
NUTCR214, April 2011. (Ex. 68).
Response: We agree with the commenter that we may have assumed DSI control efficiencies
that are greater that can actually be achieved. We have seen individual DSI control efficiencies
that are in the range we have assumed. However, these high levels of control are highly site
specific and depend on conditions (e.g., long duct runs) that may not be present at the units in
question. Our assumption of conservatively high control efficiencies maximized the favorability
of the DSI cost effectiveness (lower $/ton) and maximized DSI visibility improvement, but we
did so mindful that the technical record for our proposal was pointing us to different control
options with superior, yet still cost effective control efficiencies. This is another area in which
we strove to be conservative and demonstrate that even with many conservative cost
assumptions, our analyses indicate that the scrubber retrofits we proposed are more cost effective
and a demonstrably more favorable control option than DSI.
Comment: EPA Should Have Evaluated a Circulating Dry Scrubber NID™ Control
Option. [Earthjustice (0067) p.38, Stamper (0068) p. 19]
Earthjustice et al., and Stamper stated that the EPA should have evaluated the cost effectiveness
of a NID™ system at 98% control using the SDA cost module in its evaluation of SO2 control
technologies. Earthjustice noted that Stamper provides analyses of application of the NID™
system for several of the Texas EGUs, demonstrating that it is the most cost-effective option in
many instances. Earthjustice et al., and Stamper stated that the EPA should have also conducted
an evaluation of circulating dry scrubbers, which can meet SO2 removal efficiencies of 98% or
greater over a wide range of uncontrolled SO2.
Stamper noted that the EPA evaluated only DSI, SDAs, and wet scrubbers for SO2 pollution
controls. EPA should have also conducted an evaluation of circulating dry scrubbers, which
Sargent & Lundy have indicated can meet SO2 removal efficiencies of 98% or greater over a
wide range of uncontrolled SO2 rates.65 Sargent & Lundy reported in their August 2010 SDA
FGD Cost Development Methodology that the lowest SO2 emission guarantees for a circulating
dry scrubber are 0.04 lb/MMBtu.66
Stamper stated that the Alstom Novel Integrated Desulfurization system (NID™), which is based
on the J-reactor,67 has been selected as the most cost effective scrubber option when compared to
other technologies in several recent evaluations including Flint Creek (558 MW)68, Homer City
(2 x 660 MW)69, Indian River (440 MW, operating)70, Brayton Point Unit 3 (630 MW)71, and
Boswell Unit 4 (535 MW)72 It is also currently in use as a polishing scrubber on three
331
-------
circulating fluidized bed (CFB) boilers: Seward Units 1 and 2 (2 x 285 MW), Gilbert Unit 3 (300
MW), and Spurlock Unit 4 (300 MW).73
Stamper stated that the NID™ system, like the conventional DFGD, integrates a baghouse with
the absorber.74 According to Alstom, some of the benefits of a NID™ system are: low capital
investment and maintenance costs, low power consumption, high SO2 removal and high removal
efficiencies of S03 and hazardous air pollutants.75 Alstom has installed more than 60 NID™
systems worldwide, and is in the process of installing NID™ systems at three coal-fired power
plants in the U.S. which include the Brayton Point Unit 3, Indian River Unit 4, and for two units
at Homer City power plant.76
Stamper stated that the Southwestern Electric Power Company (SWEPCO) has proposed to
install a NID™ system at the Flint Creek Power Plant in Arkansas. Flint Creek is a 528 MW
unit that burns low sulfur Powder River Basin coal of 0.8 lb/MMBtu uncontrolled SO2 rates.77
After evaluating several SO2 control systems, SWEPCO selected a NID™ system for SO2
control for the following benefits of a NID™ system: lowest capital and operation and
maintenance costs on a 30-year cumulative present worth basis, lowest water consumption,
lowest auxiliary power usage, lowest reagent usage, smallest footprint, best for mercury
reduction with activated carbon injection, best for S03 removal, and best for future National
Pollution Discharge Elimination System (NPDES) permit compliance.78
Stamper noted that, in comments submitted by several conservation organizations on EPA's
proposed Montana regional haze FIP, extensive analysis and documentation was provided to
show that the annual costs of a NID™ circulating dry scrubber system would be about 1-2%
lower than the annual costs of an SDA.79 Thus, the SDAIPM cost module provides a
conservative estimate of the costs of a NID™ system, but should be used with an assumed 98%
SO2 removal efficiency.
Stamper stated that not only is a NID™ circulating dry scrubber less costly than an SDA and as
efficient as a wet scrubber for SO2 removal, but there are numerous other benefits to a circulating
dry scrubber system like a NID™ system as previously stated. Those include smaller footprint,
easier installation, integrated baghouse for improved control of particulate emissions (especially
fine particulate) and numerous other pollutants including mercury and S03, low power
consumption, lowest reagent use, and lower water use.80
Footnotes:
65 See, e.g., Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC Technologies, SDA FGD
Cost Development Methodology, Final March 2013, at 1 (Ex. 25).
66 Id.
67 Lawrence Gatton, Alstom Power, Next Generation NID for PC Market, Coal-Gen, August 17-19, 2011. (Ex. 26).
68 See February 8, 2012 Direct Testimony of Christian T. Beam on behalf of Southwestern Electric Power
Company, In the Matter of Southwestern Electric Power Company's Petition for a Declaratory Order Finding that
Installation of Environmental Controls at the Flint Creek Power Plant is in the Public Interest, Before the Arkansas
Public Utilities Commission, Docket 12-008-U (Ex. 29).
69 See "Alstom to supply NID™ emission control system for the Homer City Generating Station," 4/13/12 Alstom
press release, available at http://www.alstom.com/press-centre/2012/4/alstom-to-supply-nidtm-emission-control-
system-for-the-homer-city-generating-station/.
70 Id.
71 Id.
72 See "Alstom emission control system to cut environmental footprint of Minnesota Power's largest power plant,"
332
-------
available at http://www.alstom.com/press-centre/2013/8/alstom-emission-control-system-to-cut-environmental-
footprint-of-minnesota-powers-largest-power-plant/
73 Black & Veatch, LG&E/KU - Mill Creek Station, Phase II Air Quality Control Study, Air Quality Control
Validation Report, March 4, 2011, Revision D - Issued for Project Use, p. 5-16 (Ex. 27).
74 See Alstom Brochure, NID™ Flue Gas Desulfurization System for the Power Industry at 3 (Ex. 28).
75 Id. at 4.
76 See "Alstom to supply NID™ emission control system for the Homer City Generating Station," 4/13/12 Alstom
press release, available at http://www.alstom.com/press-centre/2012/4/alstom-to-supply-nidtm-emission-control-
system-for-the-homer-city-generating-station/.
77 See February 8, 2012 Direct Testimony of Christian T. Beam on behalf of Southwestern Electric Power
Company, In the Matter of Southwestern Electric Power Company's Petition for a Declaratory Order Finding that
Installation of Environmental Controls at the Flint Creek Power Plant is in the Public Interest, Before the Arkansas
Public Utilities Commission, Docket 12-008-U, at 5, 18. (Ex. 29).
78 Id. at 19-21.
79 See Technical Support Document to Comments of Conservation Organizations, Proposed Montana Regional Haze
FIP - June 15, 2012, at 59-65 (Ex. 30). See also Sargent & Lundy, White Bluff Station Units 1 and 2, Evaluation of
Wet vs. Dry FGD Technologies, Prepared for Entergy Arkansas, Inc., Rev. 3, October 28, 2008 (Ex. 20); Sargent
& Lundy, Big Sandy Plant Unit 2, Order-of-Magnitude FGD Cost Estimate, Volume 1 - Summary Report,
September 29, 2010 (Ex. 31).
80 See. e.g., Alstom Brochure, NID™ Flue Gas Desulfurization System for the Power Industry at 3 (Ex. 28). See
also February 8, 2012 Direct Testimony of Christian T. Beam on behalf of Southwestern Electric Power Company,
In the Matter of Southwestern Electric Power Company's Petition for a Declaratory Order Finding that Installation
of Environmental Controls at the Flint Creek Power Plant is in the Public Interest, Before the Arkansas Public
Utilities Commission, Docket 12-008-U, at 5, 18. (Ex. 29).
Response: We agree that the Alstom NID circulating dry scrubber is a promising SO2 control
option. We reviewed the NID in our preliminary work but ultimately decided not to evaluate it
as a reasonable progress control because we had no relevant operating data and no method to
estimate costs. This comment cites only to vendor literature and fails to provide any cost or
operating data similar to what we analyzed in response to other comments to validate the ability
of the wet FGD to meet our proposed SO2 emission limits.
The comment cites to proposed NIDs at Flint Creek (2016269), Homer City 1 & 2 (2016), Indian
River Unit 4 (2011), Brayton Point Unit 3 (2013), and (Clay) Boswell Unit 4 (2016). We
researched each one of these facilities and learned that only Indian River Unit 4 and Brayton
Point Unit 3 are currently operating.
We download the SO2 CEMS data for these two facilities from CAMD, computed 30-day BOD
averages, achieved SO2 control efficiencies, and determined the maximum 30-day BODs. This
analysis is present in our final docket.270 Neither of these units is performing at the level of a wet
FGD - 98% SO2 control or 0.04 lb/MMBtu.
Indian River Unit 4 has achieved only 86% SO2 removal based on a design level of 96%, rather
than our target of 98% using wet FGD.271 The maximum 30-day BOD is 0.94 lb/MMBtu and the
269 Parenthetical dates are projected or actual startup dates, as reported in the Mcllvaine Utility Tracking System or
in CAMD.
270 See "Selected NID BOD SO2 Averages.xlsx."
271 Burns & McDonnell, Indian River Unit 4, http://www.burnsmcd.com/Resource_/Project/1302/ProjectPdf/Indian-
River-Unit-4.pdf.
333
-------
lowest 30-day BOD, representing its best performance, is 0.07 lb/MMBtu. Brayton Point Unit 3
has achieved 94% SO2 removal and the maximum 30-day BOD is 0.16 lb/MMBtu. Brayton
Point Unit 3 is scheduled for shutdown in 2017.272 Further, these facilities are not similar to the
Texas units as they burn bituminous coal with higher sulfur contents than most of the units in the
Texas FIP.273
Thus, the limited available SO2 CEMs data for these two NIDs indicate that they are not yet
achieving the same performance as a conventional wet FGD or the dry FGD that we evaluated
for Tolk. However, this technology is promising, and we will continue to monitor NID
performance for consideration in the next round of SIPs, when the technology has more relevant
operating history.
The other NIDs cited in this comment are "polishing" scrubbers on circulating fluidized bed
boilers, which are not similar boiler types. Further, none of them burn PRB coal. Seward burns
high sulfur waste coal; Gilbert Unit 3 burns high sulfur bituminous coal; and Spurlock Unit 4
burns high sulfur bituminous coal and switchgrass.274
Finally, we note that as NID technology is theoretically capable of achieving the same SO2
control efficiency as the wet FGD scrubber retrofits that we evaluated, the facilities in question
are free to choose NID or any other technology to satisfy their emission limits. Thus, if the
NID is as claimed in this comment — low capital investment and maintenance costs, low power
consumption, and high removal of hazardous pollutants for about the same cost— plant owners
will select the NID over wet FGD. Thus, including NID in our evaluation would not have
changed our proposed SO2 emission limits for any facility except potentially Tolk.
In the case of Tolk, a NID scrubber could theoretically yield a lower SO2 emission limit for a
similar cost and thus would potentially be more cost effective than a dry FGD. However, as we
note above, we do not have any relevant SO2 CEMS data for a similar facility, i.e., SO2 emission
data for a pulverized coal fired boiler burning low sulfur PRB coal, to demonstrate that a higher
control efficiency than the 95% we proposed assuming SDA is achievable using NID at Tolk.
The only available CEMs data so far demonstrates that the SDA is able to achieve lower SO2
emissions than the NID. Thus, we have no basis for proposing a NID at Tolk.
Comment: EPA's Assumed 95% Control with Scrubber Upgrades is Reasonable, and the
Control Level Could Be Higher for Some EGUs. [Stamper (0068) p. 35]
Stamper stated that the EPA assigned a "presumptive scrubber upgrade target" of 95% SO2
control. Cost TSD at 28. Given that wet scrubbers can achieve 98-99% SO2 control, a
272 Dorothy Davis Ballard, Brayton Point Coal-Fired Power Plant Confirmed for Shut Dow by 2017, Power, January
29, 2014, http://www.pennenergy.com/articles/pennenergy/2014/01/brayton-point-coal-fired-power-plant-
confirmed-for-shut-down-by-2017.html.
273
http://www.google.com/url?sa=t&rct=i&a=&esrc=s&source=web&cd=8&ved=OCFEOFiAHahUKEwifqrrXvc HA
hWFiw0KHOL9Cc8&url=http%3A%2F%2Fpennwell.websds.net%2F2013%2Fbangkok%2Fpga%2Fpapers%2FT2
S4Q1 -paper.pdf&ei=GkXiVZ-1 Go WfNoL6p gM&usg=AFOi CNFWrX13 9kOm6Ya 4YnvwtYZItZDOO&cad=ri a
274 http://www.ekpc.coop/pressreleases/2009%20press%20releases/2009-04-01_Spurlock4_commercial_start.pdf
334
-------
presumptive 95% control target for the EGUs with existing wet scrubbers is more than
reasonable.
First, Stamper stated that it must be noted that EPA's proposed SO2 emission limits are based on
95% control assuming all of the sulfur in the coal is emitted as SO2. As discussed above, 2-5%
of the sulfur in the coal typically falls out in the bottom ash of a coal-fired boiler. Thus, EPA's
proposed SO2 emission limits reflect somewhat less than 95% control.
Second, Stamper stated that the with all of the available options to improve SO2 removal with
physical and/or operational changes to the existing wet scrubbers at the Texas EGUs evaluated
by EPA for scrubber upgrades, 95% control or better should be achievable. Indeed, the
technology used in modern wet FGD systems that can achieve 98%-99% control can be
incorporated into older wet FGD systems.135 Many of the improvements in state-of-the-art wet
FGDs are based on improving the liquid-to-gas contact and residence time, and thus adding wall
rings and/or scrubber trays and new designs of spray headers that ensure more complete contact
with the flue gas136 can be often be readily incorporated into existing wet FGD systems. Further,
chemical additives can be added to the scrubber slurry to enhance SO2 removal. Older
generation mist eliminators can be replaced with modern mist eliminator designs that more
effectively wash the mist eliminators and prevent solids deposition, which ultimately makes the
wet scrubber work more efficiently.137 Further, existing wet FGDs can be converted to the
limestone forced oxidation system that is currently the most common system for wet FGDs.138
Thus, the technology of today's wet FGD systems can in many cases be incorporated into older
scrubbers to raise SO2 removal efficiencies to the control levels expected with a new FGD
system.
Stamper suggested, for example, the wet FGD scrubber at Unit 3 of the Fayette Power plant was
upgraded to achieve an emission limit that was reflective of 95.5% SO2 removal without any
scrubber bypass while burning 1% sulfur Powder River Basin coal139. The original wet scrubber
was designed to achieve 90% control with up to 20% bypass, while burning high sulfur lignite
coal.140 Prior to the upgrade, the unit was achieving approximately 81% to 84% SO2 removal.141
To meet an SO2 limit reflective of 95.5% control and no bypass, it was decided that the most cost
effective solution for Fayette Power Unit 3 was to install one tray in each FGD absorber,
increase the number of spray nozzles for each spray level, replace the original turning veins, and
convert the chimney to wet stack operation.142 The performance testing results of the upgraded
wet scrubber greatly exceeded the 95.5% SO2 removal efficiency target, achieving on average
99.2% SO2 removal efficiency (based on scrubber inlet and scrubber outlet testing).143
Stamper noted that Babcock Power provided evaluations of two sets of upgrades to the Mill
Creek Unit 1, 2, and 4 wet FGD systems (which Unit 3 would be routed to): 1) to enable the wet
FGD to achieve 96% SO2 removal efficiency and 2) to achieve 98% SO2 removal efficiency144.
It was first determined that Units 1 and 2 FGD scrubbers could achieve a maximum of 88% SO2
removal and the Unit 3 scrubber could be expected to achieve a maximum of 91% SO2 removal
without modifications.145 It was next determined that these three scrubbers could achieve 96%
SO2 removal efficiency with changes to spray nozzles, the addition of wall baffles, removing
existing spargers, an increase in recycle pump capacity to handle an increase in liquid-to-gas
ratio, and installation of agitators and oxidation air lancers on reactor tanks.146 Last, it was
335
-------
determined that these scrubbers could achieve 98% SO2 removal with additional modifications.
In general, Stamper noted that those modifications included changes in spacing of spray levels,
changes in the types of spray level nozzles, changes in the angles of spray nozzles, increasing
number of spray nozzles, staggering of spray header layout, adding wall baffles, along with
increasing the liquid-to-gas ratio and recycle pump capacity and other modifications.147 Babcock
Power predicted SO2 removal efficiencies in excess of 98% with these modifications.148 It was
estimated that the wet FGD upgrades to achieve 96% removal would cost $10.5 to $14 million
per unit, and upgrades to achieve 98+% removal would cost $20 to $33 million per unit.149 This
reflects a range of installation costs of $32/kW to $75/kW. Ultimately, it appears that the owners
of the Mill Creek units are opting to install new scrubbers at Units 1 and 2 and just upgrade the
Unit 3 scrubber.150
For all of these reasons, Stamper concluded that the EPA's presumption that those EGUs with
existing scrubbers can meet 95% SO2 removal with scrubber upgrades is a reasonable
assumption.
Footnotes:
135 Moretti, Albert L., State-of-the-Art Upgrades to Existing Wet FGD Systems to Improve SO2 Removal, Reduce
Operating Costs and Improve Reliability, Presented to Power-Gen Europe, Cologne, Germany, June 3-5, 2014, at 1-
2 (Ex. 45).
136 Id. at 5-6.
137 Id. at 6-1.
138 Id. at 7-8.
139 Frazer, C., A. Jayaprakash, S.M. Katzberger, Y.J. Lee, B.R. Tielsch, Fayette Power Project Unit 3 FGD
Upgrade: Design and Performance for More Cost-Effective SO2 Reduction, presented to EPRI Power Plant Air
Pollutant Control Mega Symposium, August 30 - September 2, 2010, Baltimore, MD, at 1 (Ex. 46).
140 Id.
141 Id. at 2.
142 Id. at 3.
143Id. at 6.
144See February 2011, Babcock Power, LG&E Services Company Contract No. 501654, Mill Creek FGD
Performance Upgrade Study, Assess the feasibility of upgrading the Mill Creek Units 1 & 2 FGD's and upgrading
the existing Mill Cree 4 FGD and utilizing it for Mill Creek Unit 3 (Ex. 47).
145 Id. at 2.
146 Id. at 3-6.
141 Id. at 6-10.
148 Id. at 11.
149Id. at 6 and 10.
150See https://lge-ku.com/our-company/community/neighbor-neighbor/mill-creek-generating-station.
Response: We agree with the commenter that it may be possible, with the exception of San
Miguel, that many of the scrubber upgrade units can achieve greater control efficiencies than we
proposed. Greater control efficiencies would result in a more favorable cost effectiveness (lower
$/ton) and more visibility improvement. This is another area in which we strove to be
conservative and demonstrate that even with many conservative cost assumptions, our analyses
indicate that the scrubber upgrades we proposed are very cost effective.
336
-------
ll.e. Use of IPM cost algorithms
Comment: [Sargent & Lundy (0061) p. ES-2] For the proposed retrofit FGD control systems,
S&L reviewed the Technical Support Document (TSD) developed by EPA to support the
proposed FIP, as well as EPA's reliance on its IPM model to estimate the capital costs, cost-
effectiveness, and controlled SO2 emissions associated with the proposed control technologies.
Our review identified several areas where EPA's analysis is inconsistent with its own guidance
and deviates from the approach taken in previous reasonable progress evaluations. EPA's
approach does not provide an accurate representation of the actual costs that would be incurred
by Luminant to install and operate the retrofit FGD control systems, and does not provide an
accurate accounting of the SO2 emission reductions achievable with the controls. The following
items were identified by S&L as significant flaws in EPA's approach to establishing the control
technology costs and achievable emission rates proposed in the FIP.
1. S&L stated that EPA failed to follow its own guidelines and the cost estimating methodology
described in the Office of Air Quality Planning and Standards (OAQPS) Control Cost Manual to
develop capital costs for the retrofit technologies:
• Although EPA claims that it relied on the methods and principals contained within the
Control Cost Manual in developing its individual control technology cost estimates for
the FGD retrofit units, in fact, EPA relied on the IPM cost algorithms to calculate capital
costs. The IPM cost algorithms, which were developed by S&L for EPA, do not provide
unit-level cost estimates. Moreover, reliance on the IPM cost algorithms is inconsistent
with the Regional Haze cost impact assessment guidelines in 40 CFR Part 51 Appendix
W ("Guidelines for BART Determinations Under the Regional Haze Rule"), which EPA
claims to rely on for its FIP.
2. S&L stated that, by relying on the IPM cost algorithms, EPA did not adequately evaluate and
account for the total capital investment associated with retrofitting FGD technology at the
Luminant units. The IPM cost modules were developed to provide generic order-of-magnitude
costs for various air quality control technologies that can be applied to the electric power
generating industry on a system-wide basis and can significantly underestimate capital costs on a
unit-specific basis due to the following limitations:
• The IPM model does not account for site-specific location constraints or variation in
operating parameters such as flue gas temperature, which is applicable at Big Brown and
Monticello, both of which would impact the retrofit costs of FGD at these stations.
• The IPM cost algorithms do not address regional labor productivity, taking into account
project location, local workforce characteristics, local unemployment and labor
availability, project complexity, local climate, and working conditions.
• The IPM cost modules only include costs for "minor physical and chemical wastewater
treatment." Future regulatory standards proposed by EPA will likely require more
significant waste water treatment than assumed by IPM.
• Indirect capital costs included in the IPM cost modules do not account for all project-
related indirect costs a facility would incur to install a retrofit control and do not include
all indirect capital costs allowed by the Control Cost Manual such as project contingency.
337
-------
• EPA excluded Owner's Costs and AFUDC from its estimates. However, Owner's Costs
and AFUDC are real costs that the owner will incur and should be included in EPA's
estimate of total capital investment.
Response: This comment is a summary of detailed comments presented elsewhere. Please see
our responses to those comments.
Comment: EPA's approach to evaluating cost-effectiveness of retrofit proposed FGDs -
EPA's flawed reliance on the IPM cost algorithms [Sargent & Lundy (0061) p. 14]
S&L reviewed the Technical Support Document (TSD) developed by EPA in support of its
proposed FIP as well as EPA's use of the IPM model to develop capital costs and cost
effectiveness estimates for the proposed retrofit FGD control technology. Our review identified
several areas where EPA deviated from its approach in previous regional haze evaluations. In
addition, S&L reviewed EPA's use of the IPM model and cost algorithms used by EPA to
estimate capital costs and cost-effectiveness of the retrofit FGD controls. EPA's analysis is in
conflict with its own guidelines, and is inconsistent with the approach EPA took to develop
capital costs in other regional haze evaluations.
According to S&L, in its proposal, EPA claims that it "relied on the methods and principals
contained within the EPA Air Pollution Control Cost Manual (the Control Cost Manual)31 in
developing cost estimates for the individual units that EPA is targeting in this proposal.
The cost estimating procedure described in the Control Cost Manual consists of five steps:
(1) obtaining the facility parameters and regulatory options for a given facility;
(2) roughing out the control system design;
(3) sizing the control system components;
(4) estimating the cost of these individual components; and
(5) estimating the costs (capital and annual) of the entire system.32
S&L stated that the Control Cost Manual describes the equipment and other direct costs that are
typically included in an estimate of the Total Capital Investment (TCI) required to install a given
control technology. The Manual defines TCI to include all costs required to purchase equipment
needed for the control system (purchased equipment cost), the costs of labor and materials for
installing that equipment (direct installation costs), costs for site preparation and buildings, and
certain other costs (indirect installation costs).33 TCI also includes costs for working capital and
off-site facilities. 34 Direct installation costs include costs for foundations and supports, erecting
and handling the equipment, electrical work, piping, insulation, and painting. Indirect
installation costs include costs such as engineering costs; construction and field expenses (i.e.,
cost for construction supervisory personnel, office personnel, rental of temporary offices, etc.);
contractor fees (for construction and engineering firms involved in the project); start-up and
performance test costs (to get the control system running and to verify that it meets performance
guarantees); and contingencies.35
338
-------
S&L stated that the manual includes specific chapters for a number of air pollution control
systems. Each chapter includes a process description, sizing or design procedures, procedures
for estimating capital and annual O&M costs, and cost estimating example problems.36 However,
the Control Cost Manual states that it "does not directly address the controls needed to control air
pollution at electrical generating units (EGUs) because of the differences in accounting for utility
sources."37 In fact, the Manual specifically states that for certain control systems (e.g.,
specifically SCR reactors and FGD units) it deviates from its standard approach of providing
study level costs and, instead, provides a description of the factors that influence TCI for the
analyst to consider.38 The Manual takes this approach because "the control in question is either
so large or so site-specific in design that suppliers design, fabricate, and construct each control
according to the specific needs of the facility." 39
According to S&L, although EPA claims in its proposal that it relied on the methods and
principals contained within Control Cost Manual in developing its individual control technology
cost estimates, in the supporting Cost TSD EPA stated that "[i]n order to estimate the costs for
DSI, SDA scrubbers, and wet FGD scrubbers, we programmed the DSI, SDA, and wet FGD cost
algorithms, as employed in version 5.13 for our IPM model."40
S&L stated that the IPM model and the Control Cost Manual provide two entirely different
approaches to calculating control system capital costs. IPM is described by EPA as a multi-
regional, dynamic, deterministic linear programming model used by EPA to analyze system-
wide impacts of air emissions policies on the U.S. electric power sector in the 48 contiguous
states and the District of Columbia.41 The model has been used by EPA to analyze impacts
associated with proposed regulatory programs such as the Clean Air Interstate Rule (CAIR) and
Mercury and Air Toxics Standard (MATS). The primary purpose of the model is to provide
forecasts of least-cost capacity expansion, electricity dispatch and emission control strategies for
meeting energy demand and environmental, transmission, dispatch and reliability constraints.
The model includes cost modules for various air quality control technologies, and, as previously
discussed, S&L developed the cost algorithms used in the IPM model to estimate costs
associated with DSI, SDA, and wet FGD control systems.42 The IPM model is not referred to in
either the Control Cost Manual or the BART Guidelines as an acceptable tool to develop site
specific estimates.
S&L stated that the cost algorithms in the IPM model were developed based on a statistical
evaluation of cost data available from various industry publications, and do not take into
consideration site-specific cost or constructability issues.43 The cost modules are intended to
provide "typical" retrofit costs for a unit located within 500 feet of sea level.44 The cost modules
provide a capital, engineering, and construction cost (CECC) subtotal, which includes the
purchased equipment costs, installation costs, and certain indirect capital costs (e.g., engineering
and construction management, labor adjustment, and contractor profit and fees). Additional
costs and financing expenditures, including Owner's home office costs and AFUDC, are
included, as applicable, to calculate the Total Project Cost (TPC) or TCI.
According to S&L, the primary purpose of the IPM cost modules is to provide generic order-of-
magnitude costs for various air quality control technologies that can be applied to the electric
power generating industry on a system-wide basis, not on an individual unit basis. By necessity,
339
-------
the cost algorithms were designed to require minimal site-specific information available from
publicly available sources. Inputs to the IPM DSI, SDA, and WFGD capital cost algorithms are
limited to gross unit size (MW), fuel type, unit heat input or heat rate, and an SO2 removal
efficiency. In addition, a subjective retrofit factor can be applied for control technologies
installed on existing units.
S&L noted, because of the limited number of site-specific inputs, the IPM cost algorithms
provide order-of-magnitude control system cost estimates, but do not provide case-by-case
project-specific cost estimates meeting the requirements of the BART Guidelines, nor do the
IPM equations incorporate the cost estimating methodology described in the Control Cost
Manual.
S&L concluded that EPA's use of IPM is thus not an appropriate choice for the unit-specific
analysis it claims to be conducting in its proposal. By relying on the IPM cost modules to
calculate DSI, SDA, and WFGD costs, EPA did not adequately evaluate and account for
potential project-specific site constraints or balance-of-plant (BOP) costs that Luminant would
incur to install and successfully operate the FGD control systems EPA is proposing. In addition,
using the IPM cost algorithms to calculate FGD control system capital costs is inconsistent with
the case-by-case BART cost analysis described in the BART Guidelines for at least two
reasons.45 First, the IPM model does not account for unit-specific design and operating
parameters that can affect control system design and costs. Second, the IPM cost equations do
not take into consideration site-specific conditions that could affect the BOP costs that a facility
would incur to install and operate the control system.
Footnotes:
31 Technical Support Document for the Cost of Controls Calculations for the Texas Regional Haze Federal
Implementation Plan, "Cost TSD", Section 2, page 2.
32 Control Cost Manual, Section 1, Chapter 2, page 2-23.
33 Id. at page 2-5.
34 Id.
35 Id.
36 Control Cost Manual, Section 1, Chapter 1, page 1-5.
37 Control Cost Manual, Section 1, Chapter 1, page 1-3,
38 Control Cost Manual, page 2-27.
39 Id.
40 Cost TSD, Section 2, page 3.
41 See, EPA website: www.epa.gov/airmarkt/progsregs/epa-ipm/
42 See, e.g., IPM Model- Updates to Cost and Performance for APC Technologies Wet FGD Cost Development
Methodology, Sargent & Lundy LLC, March 2013 ("IPM-Wet FGD Cost Methodology")
43 IPM-Wet FGD Cost Methodology, pg. 1
44 Id., atpg. 2.
45 S&L does not endorse EPA's use of the BART Guidelines for its analysis for Texas's reasonable progress SIP
submission, but is merely pointing out the internal inconsistencies in how EPA conducted its analysis—claiming at
times to follow the BART Guidelines, while at times ignoring them.
Response: S&L accuses us of not following our Control Cost Manual. On the contrary, as we
stated in our Cost TSD, we "relied on the methods and principals contained within the EPA Air
Pollution Control Cost Manual (the Cost Control Manual, or Manual)..." Our Cost TSD goes on
to explain that by this, it means the general costing methodology, namely the use of the overnight
340
-------
costing method.275 We did not assert that we followed the five-step cost estimating procedure
cited in this comment. This five step procedure is simply a way of organizing the cost. The
same information can be organized in different, equally legitimate ways. Further, this
organizational method is not required under the BART Guidelines. In fact, the BART
Guidelines themselves are not required, as we are not conducting a BART determination, but
rather a reasonable progress analysis.
The BART Guidelines state: "In order to maintain and improve consistency, cost estimates
should be based on the OAQPS Control Cost Manual, where possible."276 This statement is
footnoted to indicate: "You should include documentation for any additional information you
used for cost calculations, including any information supplied by vendors that affects your
assumptions regarding purchased equipment costs, equipment life, replacement of major
components, and any other element of the calculation that differs from the Control Cost
Manual." The Cost TSD and the Section 114 information satisfy this requirement.
The five-step general cost estimating procedure cited in this comment is not a rigid method that
must be used to determine costs, but rather just a general costing procedure. The specific steps
vary with the control technology, which can be ascertained by inspecting subsequent technology
chapters in the Control Cost Manual. See, for example, the chapter on SCR, which does not
follow this five step procedure.
Regardless, the Control Cost Manual does not include any method for estimating the costs of any
of the SO2 control methods evaluated in this action. In fact, the Table of Contents of the Control
Cost Manual indicates that the chapter on "Wet and Dry Scrubbers for SO2," two of the three
controls EPA costed, is a "planned chapter." This chapter was never developed.277 The Table of
Contents does not even include a section on DSI, because DSI was not widely used when the
Control Cost Manual was last updated in 2002. Thus, it is not possible to use the Control Cost
Manual to cost DSI and dry and wet scrubbers, except through the overnight cost methodology
and the other general principles contained therein. As a consequence, faced with having to
perform cost analysis on applicable control technology, we elected to use the only publically
available current SO2 costing models, the cost models developed by S&L for use in our IPM.
This is similar to the strategy the states themselves employed in the development of their own
SIPs. For instance, as explained in the Texas SIP278:
The TCEQ used the control strategy analysis completed by the CENRAP as the
starting point for the analysis of additional controls. The CENRAP analysis used
the EPA AirControlNET tool to develop cost per ton estimates for the relevant
pollutants. The TCEQ reviewed this information and made changes based on
275 Cost TSD, p. 2.
276 70 FR 39166.
277 Control Cost Manual, Table of Contents, Section 5.2, Chapter 2, footnote 2.
278 Revisions To The State Implementation Plan (SIP), Concerning Regional Haze, Texas Commission On
Environmental Quality, P.O. Box 13087, Austin, Texas 78711-3087, Project No. 2007-016-SIP-NR, Adopted,
February 25, 2009 (Texas Regional Haze SIP). Page 10-4.
341
-------
knowledge of the particular facilities and agency experience with implementing
ozone control strategies. The analysis focused on moderate cost controls for
sources that were likely to contribute to visibility impairment at Class I areas.
And further:279
CENRAP used the latest revised version of the EPA AirControlNET model
(Alpine 2007) to analyze potential add-on control device strategies.
AirControlNET is a control technology analysis tool developed to support the
EPA in analyses of air pollution policies and regulations. The tool provides data
on emission sources, potential pollution control measures and emission
reductions, and the costs of implementing those controls. Every available SO2
and NOX control strategy in AirControlNET was run against the electric
generating units (EGUs) and non-EGU point source inventories to develop a
master list of available incremental control strategies for the entire CENRAP 36
kilometer domain.
The CENRAP states, including Texas, used the results of the AirControlNET model to estimate
the costs of the same types of controls as we did, and like us, used the results to inform their
decisions of whether controls were warranted for specific units. Texas summarized the results of
that effort for selected facilities in Tables 6 and 7 of Appendix 10-1 of its Regional Haze SIP.
We have not maintained the AirControlNET model for a number of years and it was not
available to use in the development of our proposal.
Thus, the only relevant portion of the Control Cost Manual to our present work is the general
methodology chapter, Chapter 2: Cost Estimation: Concepts and Methodology. We did adopt
the Control Cost Manual general cost methodology, the overnight method, as we did in all of our
previous FIPs. As explained in the Cost TSD, we modified the IPM cost models to use the
overnight cost method by removing AFUDC and owners costs and to additionally calculate cost
effectiveness, rather than just total capital and O&M costs.
This comment asserts that the IPM model is not cited in either the CCM or the BART Guidelines
as evidence that it is not "an acceptable tool to develop site specific costs." These models were
developed by Sargent & Lundy, under contract to EPA.280 They were first published in 2010,281
after the last Control Cost Manual modification (2002) and after the BART Guidelines were
published in 2004.282 Thus, it is illogical to expect the IPM cost algorithms to be cited in either
the Control Cost Manual or the BART Guidelines. Regardless, the absence of a specific cite to a
279 Texas Regional Haze SIP. Page 10-7.
OOQ
Section 5. Emission Control Technologies, Explaining: "EPA Base Case v.4.10 includes a major update of
emission control technology assumptions. For this base case EPA contracted with engineering firm Sargent and
Lundy to perform a complete bottom-up engineering reassessment of the cost and performance assumptions for
sulfur dioxide (SO2) and nitrogen oxides (NOX) emission controls."
Available at: http://www.epa.gov/airmarkets/documents/ipm/Chapter5.pdf.
281 See Chapter 5 at http://www.epa.gov/airmarkets/programs/ipm/BaseCasev410.html.
282 40 CFR Part 51: Regional Haze Regulations and Guidelines for Best Available Retrofit Technology (BART)
Determinations; Proposed Rule, 69 FR 25,184 (May 5, 2004).
342
-------
model in the Control Cost Manual is not evidence that the model is not acceptable. We have
discretion to select the best tool for the job.
We have used IPM cost models to estimate BART costs in other similar rulemakings including
our Arizona Regional Haze Federal Implementation Plans283 the Wyoming Regional Haze
Federal Implementation Plan,284 and to supplement our analysis in the Oklahoma Federal
Implementation Plan.285 As explained in the Arizona rulemaking: "EPA's Clean Air Markets
Division contracted with engineering firm Sargent and Lundy to perform a complete bottom-up
engineering reassessment of the cost and performance assumptions for SO2 and nitrogen oxides
NOx emission controls."286
This comment also confuses "the IPM model" and the cost models that are inputs to the IPM
model. The comment asserts that the IPM model and the Control Cost Manual provide two
entirely different approaches to calculating control system capital costs. In supporting this
argument, it states: "IPM is described by EPA as a multi-regional, dynamic, deterministic linear
programming model used by EPA to analyze system-wide impacts of air emissions policies on
the U.S. electric power sector in the 48 contiguous states and the District of Columbia." This is
an accurate description of "IPM model", but not of the cost models that generate input to the
IPM.287 The models used to generate these cost inputs are not "multi-regional, dynamic,
deterministic linear programming models" but rather conventional costing models that use the
same general methods set out in the Control Cost Manual to estimate total capital and O&M
costs based on various design parameters. We only used the control technology cost models, not
the IPM model itself.
Accuracy of IPM Cost Algorithms for Site-Specific Costs Estimates
This comment argues that the IPM model only provides ".. ."typical" retrofit costs for a unit
located within 500 feet of sea level" and "generic order-of-magnitude costs," implying they are
not acceptable for unit-specific cost estimates for the individual plants included in this
rulemaking. However, this comment contradicts itself, by first stating the IPM model provides
typical retrofit costs for a "unit" and then asserting it doesn't project "unit" costs. We believe
these cost models are adequate for projecting unit-level control costs for the purpose of
conducting the type of cost-benefit analysis we have done in our proposal. For instance, S&L's
own documentation for the wet FGD cost model states the following:
The 2004 to 2006 industry cost estimates for wet FGD units from the "Analysis of
MOG and LADCO's FGD and SCR Capacity and Cost Assumptions in the
Evaluation of Proposed EGU 1 and EGU 2 Emission Controls" prepared for
Midwest Ozone Group (MOG) were used by Sargent & Lundy LLC (S&L) to
develop the wet FGD cost model. In addition, S&L included data from "Current
283 77 FR 42852 (July 20, 2012).
284 Memorandum from Jim Staudt to Dough Grano, EPA, Re: Review of Estimated Compliance Costs for Wyoming
Electricity Generating Units (EGUs) - Revision of Previous Memo, February 7, 2013, EPA-R08-OAR-2012-0026-
0086_Feb7 2013 .
285 76 FR 81728 (December 28, 2011).
286 77 FR 42852 (July 20, 2012)
287 http://www.epa.gov/airmarkets/programs/ipm/index.html.
343
-------
Capital Cost and Cost-effectiveness of Power Plant Emissions Control
Technologies" prepared by J. E. Cichanowicz for the Utility Air Regulatory
Group (UARG) in 2010. The cost increases reported from 2007 to 2008 by G.
W. Sharp in "Update: What's That Scrubber Going to Cost?" published in Power
Magazine, March 2009 were also considered. The published data was
significantly augmented by the S&L in-house database of recent wet FGD
projects. Cost data from the various sources showed similar trends versus
generating capacity. Escalation based on the CEPI was deemed acceptable. All
data sources were combined so as to provide a representative wet FGD cost basis.
The data was converted to 2012 dollars based on the Chemical Engineering Plant
Index (CEPI) data. Additional proprietary S&L in-house data from 2007 to 2012
were included to confirm the index validity. Finally, the cost estimation tool was
benchmarked against recent wet FGD projects to confirm the applicability to the
current market conditions.
As discussed above, S&L used real world cost data to construct its cost algorithms and confirm
its validity. These cost models have been updated and maintained since their introduction in
2010 and have been continuously used by EPA since that time. These control costs are based on
databases of actual control project costs and account for project specifics such as unit size, coal
type, gross heat rate, retrofit factor, and require unit specific inputs such as reagent cost, waste
disposal cost, auxiliary power cost, labor cost, gross load, and emission information. We believe
that the IPM cost models provide sufficiently accurate, study-level, unit-specific costs for
regulatory cost analysis such as required for BACT, BART, and reasonable progress.
Case-by-Case Analyses
This comment argues that the use of IPM algorithms is "inconsistent with the case-by-case
BART cost analysis described in the BART Guidelines" because: (1) it does not account for unit-
specific design and operating parameters and (2) it does not consider site-specific conditions that
could affect balance of plant costs.
First, we did not conduct a BART analysis—our cost analyses were conducted under the
reasonable progress and long-term strategy provisions of the Regional Haze Rule. We looked to
our BART Guidelines for assistance in certain aspects of our analysis:288"We are therefore
relying on our BART Guidelines for assistance in interpreting those reasonable progress factors,
as applicable." Many aspects of a BART analysis are not applicable. Also, this comment
redefines "case-by-case" as used in "case-by-case BART". BART means "an emission
limitation based on the degree of reduction achievable through the application of the best system
of continuous emission reduction for each pollutant which is emitted by ... [a BART-eligible
source]. The emission limitation must be established, on a case-by-case basis, taking into
consideration the technology available, the costs of compliance, the energy and non-air quality
environmental impacts of compliance, any pollution control equipment in use or in existence at
the source, the remaining useful life of the source, and the degree of improvement in
visibility... ,"289 In this context, case-by-case refers to each pollutant at each emission unit. We
288 FIP TSD page 6.
289 70 FR 39163.
344
-------
applied the IPM cost models for SO2 at each emission unit using unit-specific inputs. Thus, the
"case-by-case" requirement of the definition of BART is satisfied.
Second, we accounted for "unit-specific design and operating parameters" by using unit- and
site-specific data as we discussed in our COST TSD.290 Every input parameter to our DSI, SDA
and wet FGD was presented and discussed. In many instances, the sensitivity of the cost model
to particular inputs was also reviewed. For certain key parameters such as trona, lime and
limestone costs, we presented information that documented our selections. We requested site-
specific information from each facility in our Clean Air Action Section 114 Information
Requests. Further, we invited feedback from the affected facilities as to the selection of these
input parameters, but received little response.
Accuracy of IPM Cost Estimates
This comment asserts that the IPM cost models provide only "order of magnitude" control cost
estimates. However, as the Control Cost Manual explains, the sole input required for making an
"order of magnitude" estimate is "the control system's capacity (often measured by the
maximum volumetric flow rate of the gas passing through the system),"291 Such an estimate, for
example, could be obtained from the cost reported in dollars per megawatt ($/MW) or dollars per
million BTUs fired ($/MMBtu), metrics that are widely reported in the literature. A quick look
at the many inputs to the IPM cost models292 demonstrates that they provide more accurate
estimates than "order of magnitude." The Control Cost Manual indicates that "the costs and
estimating methodology in this Manual are directed toward the "study" estimate with a nominal
accuracy of +/-30% percent."293 This is the long-standing rule of thumb for cost estimate
accuracy used by the EPA for regulatory cost effectiveness analyses.294
Site-Specific Inputs
This comment also argues: "the cost algorithms were designed to require minimal site-specific
information available from publicly available sources. Inputs... .are limited to gross unit size
(MW), fuel types, unit heat input or heat rate, and an SO2 removal efficiency. In addition, a
subjective retrofit factor can be applied for control technologies installed on existing units."
Comment--, p. 16.
This list of site-specific inputs is incomplete as we note above. The SO2 cost models also
include other site-specific inputs including: type of coal, coal factor, lime rate, waste rate,
auxiliary power, makeup water rate, reagent cost, waste disposal cost, auxiliary power cost,
makeup water cost, operating labor rate, elevation, interest rate, equipment lifetime, and DSI
reagent, among others. See site-specific input values highlighted in yellow in the Cost TSD,
290 See discussion beginning in Section 2, where we present our cost analyses methodology and in Sections 3, 4, and
5, where we go over every input value.
291 Control Cost Manual, p. 2-3.
292 See Cost TSD, Sections 3, 4, and 5 and Tables 6 (DSI), 6 (SDA), and 10 (WFGD).
293 Control Cost Manual, p. 2-3.
294 EPA, New Source Review Workshop Manual, Draft, October 1990, p. B.35.
345
-------
Table 2 (DSI inputs), Table 6 (SDA inputs), and Table 10 (Wet FGD inputs). This large number
of site-specific inputs is a clear demonstration that these cost models provide more accurate
estimates than "order-of-magnitude,"295 which only require a single parameter as the input. We
specifically solicited comments on the inputs we used to our cost analyses. We received few
comments objecting to our inputs and we have addressed these in our responses to other
comments.
Lastly, we have enough confidence in the basic methodology behind the S&L cost algorithms
that in our recent proposal for updating the SCR chapter of the Control Cost Manual,296 we
presented an example costing methodology that is based on the IPM S&L SCR algorithms,
which were developed using a similar methodology to the wet FGD, SDA, and DSI cost
algorithms discussed herein For the reasons we describe above, we disagree with S&L that our
modifications to and use of the cost algorithms it has supplied us are inappropriate for use in
regional haze control cost analysis.
Comment: Site-specific conditions affecting FGD retrofit costs
[Sargent & Lundy (0061) p. 17]
As described in the report prepared for Luminant, S&L stated that the IPM cost algorithms
generate order-of-magnitude costs for atypical control system installation project; however, the
IPM cost modules are not set-up to account for site-specific retrofit challenges and unique BOP
impacts, design conditions that vary from the standard conditions used as the basis of the cost
algorithms, and regional impacts on labor costs. Examples of site-specific issues that would
affect the cost of a retrofit FGD control system include the quantity of existing equipment and
duct work that must be demolished and relocated to provide space for the FGD control system;
demolition and relocation of existing electrical equipment, water piping, and underground
equipment and utilities; modifications that may be required to the existing ash handling systems;
replacement of the existing induced draft (ID) fan or booster fan modifications; and
modifications/upgrades to the existing auxiliary power system. All of these issues must be
evaluated on a site-by-site basis, and can significantly affect the cost of a retrofit control system
project. By using IPM, EPA's cost estimates do not account for these items.
S&L stated that the IPM cost modules attempt to account for retrofit difficulty by applying a
subjective retrofit factor to the capital cost estimate.46 To establish a unit-specific retrofit factor,
the analyst should, at a minimum, review the existing site layout, develop a conceptual level
general arrangement for the retrofit control system, and identify potential site constraints. EPA
did not conduct this type of analysis when it developed its cost estimates for the Texas
generating units; rather, EPA simply assigned a retrofit factor of 1.0 to all retrofit SO2 control
technologies at all plants, noting that "[bjecause we are not aware of any significant retrofit
issues at any of the facilities we evaluated, we adopted the default retrofit value of 1.0, which
represents an average retrofit difficulty, for all the units we evaluated."47 EPA provides no sound
basis for making this assumption for all of the units at issue.
295 Control Cost Manual, p. 2-3.
296 80 FR 33515.
346
-------
According to S&L, the BART Guidelines state that the cost analysis should take into account any
site-specific design or other conditions identified above that affect the cost (70 FR 39166). By
using the IPM cost algorithms, and using the default retrofit factor of 1.0, EPA failed to account
for any site-specific conditions that may affect the control system costs.
S&L noted that the Big Brown Units 1 & 2 and Monticello Units 1 & 2 have site-specific
constraints that affect the cost of a retrofit FGD control system. For example, at Big Brown the
new FGD systems would have to be located north or east of the existing generating units because
of constraints towards the south (cooling water intake) and west (exiting switchyard). The
relatively remote location of the retrofit FGD control system would require significantly more
duct work than the typical FGD arrangement, increasing the cost of the project. Similarly, there
are site constraints at Monticello given the location of the cooling lake (west of power block),
existing ash ponds (east of power block), and the existing coal pile. The impacts of these site
constraints on the cost of a retrofit FGD control system would have to be determined during
preliminary engineering and development of a conceptual level control system general
arrangement. EPA's analysis fails to take this into account.
In addition to site constraints and constructability issues, S&L noted that the type of fuel fired
and the unit-specific boiler operating conditions will affect the cost of a retrofit FGD control
system. These design parameters affect the volume of flue gas at the scrubber inlet and the
sizing of the scrubber vessels. The IPM cost modules include a "coal factor" for the type of fuel
fired, but are not designed to account for unit-specific boiler operating conditions. The default
coal factor for lignite-fueled units is set at 1.07, which is intended to account for the approximate
7% increase in flue gas volume on lignite-fueled units compared to bituminous coal-fueled units.
However, flue gas volume is also a function of boiler operation and flue gas temperatures at the
scrubber inlet. FGD control systems on units with higher flue gas temperatures must be designed
to handle higher flue gas volumes.
Based on S&L's review of operating data from Big Brown Units 1 & 2 and Monticello Units 1 &
2, flue gas temperatures at the scrubber inlets can range from 360°F to 370°F. The flue gas
temperatures are significantly higher than the generic design value of 300°F, which the IPM
model applies to all units. An increase of 60 to 70°F at the scrubber inlet would increase flue gas
volume by approximately 10%.48 This unit-specific operating parameter is not captured by the
IPM algorithms, but would significantly affect scrubber sizing and costs.
S&L stated that another example of a project-specific cost that is not adequately accounted for in
the IPM cost equations is labor productivity. In addition to the labor adjustment in the IPM cost
modules (to account for 6 x 10 hours days and per diem costs), the capital cost estimates must
account for labor productivity in the area of the proposed project. Labor productivity is an
important component of any large multi-year construction project, and is designed to account for
the project location, local workforce characteristics, local unemployment and labor availability,
project complexity, local climate and working conditions. It is common practice on large
construction projects to apply a productivity factor to account for local worker productivity and
construction site conditions. Industry experience also shows that retrofit projects and congested
site conditions can adversely affect labor productivity. Labor productivity must be evaluated on
a project-by-project basis, and can have a significant impact on overall project costs, increasing
347
-------
labor costs for the project by 15% or more.
Footnotes:
46IPM Model - Updated to Cost and Performance for APC Technologies, Wet FGD Cost Development
Methodology,
Sargent & Lundy, March 2013, page 4.
1 Cost TSD. Section 3.1. page 5.
4j Increase = [—7^] 1U0 where r>=!\ according to the ideal gas law when P]=P;
These equations can be combined 'net -:asf = [~ -1 ] H'tO where T ts in degrees Rankuse 11 ~ °F ~ 4w11
Response: This comment asserts that "IPM cost modules are not set-up to account for site-
specific retrofit challenges and unique BOP [balance of plant] impacts, design conditions that
vary from standard conditions used as the basis of the cost algorithms, and regional impacts on
labor costs."
However, retrofit issues are built into the IPM cost models, as they are based on retrofits of
control equipment on existing facilities,297 all of which have retrofit challenges relative to a
similar installation on a new facility. Thus, these models already incorporate increased costs due
to site constraints, in many cases, much greater than exist at the subject Texas plants, which have
no reported retrofit issues. The subject Texas units, for example, are generally not located in
congested urban areas and have adequate space to locate add-on scrubber technology, which is
located at the tail-end of the plant, outside of the congested interior. Thus, retrofit constraints are
very minor. Thus, we adopted the average retrofit factor of one, based on our analysis of CBI
information and aerial photographs. In fact, an average retrofit factor of one represents an
overestimate of actual costs for some units as demonstrated by comparisons of site-specific cost
estimates prepared for specific plants with IPM cost model estimates.
The Tolk units, for example, have no retrofit issues, based on field visits by Xcel consultants, yet
EPA selected a retrofit factor of 1, which assumes an average level of retrofit difficulty. The
response to the Section 114 request for the Tolk units included the results of site visits that did
not identify any retrofit issues. TOLK 0000148/150. These site visits conclude: "The area
available for retrofit of emissions control equipment is sufficient for all systems under
consideration. There are no overhead obstructions, one electrical duct bank underground that
may have to be relocated." TOLK 0000150. B&M concluded similarly. TOLK 0000373.
Thus, the use of a retrofit factor of one for these units overestimates cost, as they have no retrofit
issues.
We evaluated confidential site-specific information submitted in response to our Section 114
requests to verify these inputs. Our cost analysis evaluation of scrubber retrofits was necessarily
very site specific and required a close reliance on CBI data. However, in performing our
scrubber retrofit cost analysis, we were able to employ a model that was in the public domain,
yet could be configured to produce reasonably conservative results. As we discuss below, we are
aware that a number of our input values are high in comparison to their CBI counterparts,
297 S&L March 2013, SDA Cost Model, p. 1 (based on "typical" SDA FGD retrofit), p. 2 (cost could increase
significantly for "congested sites"); S&L March 2013, Wet FGD Cost Model, p. 1 (based on "typical" wet FGD
retrofit), p. 2 (cost could increase significantly for "congested sites").
348
-------
resulting in conservative analyses.
Affected facilities were free to provide comments on all aspects of our proposed decision. Any
affected facility could have provided us with site-specific and documented information to
address any errors and/or omissions they believed existed in our proposed rulemaking.
However, the comments we received mostly fail to provide any site-specific information that
would enable us to conclude our proposed cost estimates contain errors and to revise our cost
estimates. Rather, the comments allege general deficiencies without providing site-specific cost
information. The exceptions are addressed elsewhere in these comments.
S&L, for example, commenting on behalf of Luminant, provided general comments such as: (1)
the Big Brown FGDs "would require significantly more duct work than the typical FGD
arrangement" (without stating how much more, where the ductwork is located, or its cost,
compared to the assumption in the cost models); (2) the Monticello units have site constraints
(which were not identified); and (3) labor productivity "could" increase labor costs by "15% or
more" (without indicating whether it actually would increase at the Luminant or any other units).
We note that labor rates disclosed in the CBI materials were generally much lower than the
default value we used in the IPM cost models, eliminating any concerns about underestimates
due to labor productivity.
S&L also provided a list of general examples of retrofit issues, including space for the FGD
system; demolition and relocation of existing electrical equipment; water piping and
underground equipment and utilities; modifications to the ash handling system; replacement of
induced draft fans, etc. However, S&L failed to provide any site-specific information to allow us
to address these items, such as estimated cost or length of duct work that would allow us to
revise cost estimates, a revised retrofit factor to address site congestion, and site specific labor
productivity estimates. Also, in our analysis of the material submitted to us as part of our
Section 114 request, we reviewed several scrubber retrofit reports by S&L and other engineering
firms. In no case, were any unusual retrofit issues identified. In contrast, we did receive
specific, actionable information regarding our proposed scrubber upgrade costs. We reviewed
that information and revised those costs accordingly.
Regardless, these types of issues result in small increases in costs that are well within +/-30%
accuracy and do not affect cost effectiveness conclusions due to the conservative nature of our
estimates, as demonstrated elsewhere in these responses. They also are not required for study
level estimates, which are based on "rough sketches," "preliminary sizes", "approximate sizes,"
"rough estimates," etc.298
Further, our cost estimates are upper-bound estimates that encompass these site-specific issues
because we intentionally and consistently made assumptions throughout our analyses to assure
conservatively high cost estimates and cost effectiveness values, including:
• We selected maximum inputs from 2009 to 2013 for the inlet SO2 rate (Cost TSD, p.
11), gross heat rate, gross load (Cost TSD, p. 10), and percentage lignite, assumed to
298 Control Cost Manual, p. 2-3.
349
-------
be 100% (Cost TSD, p. 16), even if these values didn't appear in the same year of
data. Cost TSD, p. 2.
• We backstopped SO2 control efficiency by limiting the SO2 emission rate to reported
vendor guarantees. However, CEMs data presented in our response to other
comments indicates that many units are consistently achieving SO2 emission rates
much lower than vendor guarantees.
• The costs are reported in 2012 dollars, rather than 2013 dollars, which underestimates
all costs by about 3%, because the CEPCI index for 2013 (567.3) is less than the
index for 2012 (584.6). Cost TSD, p. 3.
• We used the default auxiliary power cost of $0.06/kWh, even though the CBI
materials indicated the true power cost for most units was considerably less. Cost
TSD, p. 10.
• We used default labor rates of $60/hr, even though the CBI materials indicated actual
labor rates were typically lower. Cost TSD, p. 10.
• We used water costs that were higher than typically reported in the CBI information.
• For units that burned blended fuels, our scrubber design was based on 100% lignite,
which results in a conservative estimate, since a scrubber based on burning lignite is
more expensive than one based on burning PRB.
• Our costs were based on single unit scrubber retrofits. However, in all cases we
reviewed in the CBI materials in which two or more units underwent scrubber retrofit
cost analysis, substantial savings were achieved by sharing reagent preparation areas.
We did not assume these cost savings.
Finally, S&L asserts operating data from Big Brown Units 1 & 2 and Monticello Units 1 & 2
indicate that the flue gas volume would be 10% greater than assumed in the IPM cost model,
without supporting this estimate by providing its data and calculations. The comment further
asserts that the model applies a generic design value of 300 F to all units.
First, we do not believe that the IPM cost model estimate costs using a flue gas volume estimated
based on a generic design temperature of 300 F, but rather it estimates costs from regression
equations based on actual completed projects that include retrofit difficulties. The
documentation for the IPM model contains no evidence that the IPM model applies a "generic
design value of 300 F" to all units.
Second, even assuming this were true, there are a number of factors affect the volume of gas
flow that passes through a scrubber system, other than temperature. These include the amount of
in-leakage in the system (which often increases due to inefficient or worn seals in the air
preheater) and the type and characteristics of the coal that is being burned.
350
-------
We re-examined two of the scrubber retrofit reports for Big Brown we received in response to
our Section 114 requests,299 which were issued by S&L in 2004 and 2007. The 2004 report
indicated that the design flue gas flow rate at the scrubber inlet was approximately 19.7% less
than that in the 2007 report. Both reports indicated that the reference temperature at the inlet was
370°F—the same temperature S&L references in its comment, and both were at the same
pressure. Thus, even if the IPM model assumes a lower than actual scrubber inlet temperature,
its own reports show a change of almost twice the effect it cites at the same inlet temperature for
the same unit.
We note that although these two scrubber retrofit reports assume lignite/PRB blends, the
resulting coal characteristics are slightly different, which undoubtedly accounts for some of this
difference. However, the HHV of the coal in the 2004 report was much lower than that in the
2007 report, which all else being equal, should have resulted in more of it being burned to supply
the same heat to the boiler, and in turn producing a higher flue gas flow rate. However, the flue
gas volume of the 2004 report was much lower. Consequently, Luminant's own data apparently
contradicts with the correlation it has presented. We conclude that S&L's temperature
comparison is too simple to properly characterize the situation, and we do not believe that S&L
has established that we erred in our cost analysis due to an under estimation of the flue gas
volume. Nevertheless, we will address S&L's concern:
S&L states, "The default coal factor for lignite-fueled units is set at 1.07, which is intended to
account for the approximate 7% increase in flue gas volume on lignite-fueled units compared to
bituminous coal-fueled units. However, the gas volume is also a function of boiler operation and
flue gas temperature at the scrubber inlet. FGD control systems on units with higher flue gas
temperatures must be designed to handle higher flue gas volumes." As we note in our Cost TSD
and above, we were conservative in the selection of many of our cost model inputs. One of these
conservative inputs was our decision to use assume a lignite coal factor of 1.07 for the units that
blended PRB coal with lignite (such as the ones in question), even though a lesser value would
have been justified. Thus, we have already at least partially addressed S&L's concern here.
Nevertheless, going one step further and even increasing this already conservative coal factor by
an additional 10% would increase the cost effectiveness for a wet FGD at Big Brown Unit 1
from $l,255/ton to $l,302/ton and at Big Brown Unit 2 from $l,257/ton to $l,304/ton, an
insignificant 4% to 6% increase,300 well within the +/- 30% accuracy for a regulatory cost
analysis.
S&L further asserts labor productivity could increase labor costs by 15% or more. Labor
productivity is built into the cost models. S&L provides no evidence that the labor productivity
at the Big Brown and Monticello sites would be different from the assumptions in the cost
models. Furthermore, we have no reason to suspect that site conditions would be any worse or
that laborers in Texas would be less productive than elsewhere. However, again assuming for
sake of argument that labor productivity is 30% lower in Texas than at all the other facilities
299 LUMINANT_000277496.pdf and LUMINANT_REGHAZ_1-000001183 to -000001257.pdf.
300 The increase in cost effectiveness was calculated by increasing the coal factorfrom 1.07 to 1.07x1.1 = 1.18, but
otherwise using all of the other wet FGD inputs in the file "Wet FGD Cost IPM 5-13 TX Sources ver 2.xlsx," which
is present in our docket.
351
-------
incorporated in the cost models, this, again would have a de minimus impact on cost
effectiveness. This would increase the cost effectiveness for a wet FGD at Big Brown Unit 1
from $l,255/ton to $l,271/ton and at Big Brown Unit 2, from $l,257/ton to $l,273/ton, a 1% to
4% increase301.
Assuming both changes together, which as we conclude above is not warranted, the increase in
flow rate and the decrease in labor productivity, the cost effectiveness of a wet scrubber at Big
Brown Unit 1 would increase from $l,255/ton to $1,318/ton and at Big Brown Unit 2, from
$l,257/ton to $l,321/ton, or by 5% to 8%. These increases are very small, well within the +/-
30% accuracy required for a regulatory cost analysis and do not change our proposed decision
making. The cost effectiveness values remain highly cost effective. In fact, we believe that our
cost estimates compare quite favorably to a number of Luminant SDA and wet FGD scrubber
retrofit studies Luminant has claimed as CBI that we have reviewed as a result of our Section
114 requests.
Comment: Wastewater treatment costs associated with WFGD
[Sargent & Lundy (0061) p. 18]
In the report prepared for Luminant, S&L stated that The Control Cost Manual defines "Total
Capital Investment" as including "all costs required to purchase equipment needed for the
control system." 49 In general, supporting facilities are assumed to already be installed at an
existing facility; however, costs for supporting facilities should be included in the cost estimate if
additional or unique facilities are needed to support the operation of the control system and are
not already installed at the facility.50
Examples of supporting facilities that would be required to successfully operate a wet FGD
control system include wastewater treatment systems and solid waste disposal facilities.51 The
extent of wastewater treatment required will be a function of the characteristics of the wastewater
and the wastewater discharge limits applicable to the specific facility and receiving water. The
IPM cost module includes costs only for "minor physical and chemical wastewater treatment."52
However, wastewater treatment standards proposed by EPA, and anticipated to be published as a
final rule in 2015, will likely require significantly more advanced treatment of FGD wastewaters.
As proposed by EPA, these regulations would require both physical and biological treatment of
FGD wastewaters to meet the proposed standards. Advanced FGD wastewater treatment could
add $30-$40 million to the cost of a retrofit WFGD control systems. Because dedicated
advanced wastewater treatment will be required to meet the proposed effluent guidelines for
FGD wastewaters, these costs should be included in an evaluation of the cost-effectiveness of a
retrofit WFGD control systems.
Footnotes:
49 CCM, Section 1, Chapter 2, pg. 2-5.
50 Id., at pg. 2-5 and 2-6.
51 For example, on page 2-32 of the CCM EPA states: "Though often overlooked, there can be a significant cost
associated with treating and/or disposing of waste material captured by a control system that neither can be sold nor
301 The increase in cost effectiveness was calculated by increasing the labor rate from $60/hr to $60/hr x 1.3 =
$78/hr, but otherwise using all of the other wet FGD inputs in the file, "Wet FGD IPM 5-13 TX Sources ver 2.xlsx."
352
-------
recycled to the process. Liquid waste streams, such as the effluent from a gas absorber, are usually processed before
being released to surface waters. The type and extent of this processing will, of course, depend on the
characteristics of the effluent. For example, the waste can first be sent to one (or more) clarifiers, for coagulation
and removal of suspended solids. The precipitate from the clarifier is then conveyed to a rotary filter, where most of
the liquid is removed. The resulting filter cake is then disposed of, via landfilling, for example."
52IPM - Wet FGD Cost Methodology, pg. 2.
Response: A wet FGD generates a wastewater stream that must be either evaporated, recovered,
or sent to the existing wastewater treatment plant. S&L indicates that the wet FGD cost module
only includes costs for "minor physical and chemical wastewater treatment." S&L indicates that
proposed wastewater treatment standards "will likely require significantly more advanced
treatment of FGD wastewaters." These potential future costs, which are currently unknown and
undocumented, cannot be included in our cost analyses. Further, other options are available,
including zero liquid discharge systems and the selection of a SO2 control technology that are
capable of achieving our proposed limits without generating a wastewater stream, such as the
NID, which we believe is capable of achieving our proposed emission limits, and has been
selected in some recent evaluations.
As we note in a response to another comment, the IPM cost algorithms we have incorporated
into our cost model include databases of actual control project costs. S&L states in the
documentation for its wet FGD cost documentation:
The 2004 to 2006 industry cost estimates for wet FGD units from the "Analysis of
MOG and LADCO's FGD and SCR Capacity and Cost Assumptions in the
Evaluation of Proposed EGU 1 and EGU 2 Emission Controls" prepared for
Midwest Ozone Group (MOG) were used by Sargent & Lundy LLC (S&L) to
develop the wet FGD cost model. In addition, S&L included data from "Current
Capital Cost and Cost-effectiveness of Power Plant Emissions Control
Technologies" prepared by J. E. Cichanowicz for the Utility Air Regulatory
Group (UARG) in 2010. The cost increases reported from 2007 to 2008 by G.
W. Sharp in "Update: What's That Scrubber Going to Cost?" published in Power
Magazine, March 2009 were also considered. The published data was
significantly augmented by the S&L in-house database of recent wet FGD
projects.
In the article S&L references above, "Update: What's That Scrubber Going to Cost?" the author
states that the following regarding the survey used to gather the scrubber information:
The survey instructions requested that respondents define project costs for their
just-completed, current, under construction, and/or planned FGD projects and that
they include the following line items, at a minimum, so that comparable data were
reported that could be compared across projects:
• Project design costs
• New stack and ductwork costs
• Reagent prep method and costs
• Absorber island/reactor technology costs
353
-------
• Site prep costs
• Wastewater treatment costs
• Balance-of-plant costs
• Other direct costs (such as engineering and project management)
• Associated boiler work (such as boiler modifications and draft fans)
As above, one of the surveys that fed cost data into S&L's algorithms appears to have
incorporated any applicable wastewater treatment costs. The other articles cited by S&L do not
state the degree to which wastewater treatment may have been incorporated into the reported
costs.
We specifically requested input from affected facilities on the cost of wastewater treatment for
wet FGD systems. However, S&L failed to demonstrate that a supplemental wastewater
treatment facility was needed, and provided no documentation of the costs it cites.
In addition, S&L failed to disclose that there are other, less costly options. A zero liquid
discharge treatment system is another viable option that has been selected as the least cost
alternative by other facilities.302
We do not believe that S&L has established that we erred in our cost analysis by not including an
additional wastewater treatment allowance, other than what is built into the model. We are not
even clear if the costs S&L cites to would apply to the facility or to the units individually.
Nevertheless, even assuming each Luminant unit would incur an additional capital cost of $40
million for a new wastewater treatment system, the cost effectiveness of the Big Brown units
would only increase by about 8%303 and the Monticello units by 10%. The revised cost
effectiveness values remain highly cost effective, including when combined with a 10% increase
in flue gas and a 30% increase in labor costs. Again, even assuming these unproven,
undocumented costs, our proposed decision making regarding the units in question would remain
unchanged.
Comment: Indirect capital costs [Sargent & Lundy (0061) p. 19]
In the report prepared for Luminant, S&L stated that total capital costs are defined in the Control
Cost Manual to include all costs required to purchase and install equipment needed for the
control system, as well as certain other indirect capital costs.53 EPA calculated indirect capital
costs based on factors provided in the IPM cost modules. Indirect capital costs included in the
IPM equations include:
302 Aquatech, Project Profile: Aquatech Supplies Zero Liquid Discharge Treatment for FGD System at the Iatan
Generating Station, Available at: http://www.aauatech.com/wp-content/uploads/66.-KCPL-ZLD.pdf: Aquatech,
Project Profile, Aquatech Provides Wastewater Treatment Solution for New Hampshire's Largest Coal Power Plant,
Available at: http://www.aauatech.com/wp-content/uploads/81.New-Hampshire-Merrimack.pdf. See also: Water
and Wastewater Treatment Technology Update, March 1, 2015, Available at: http://www.powermag.com/water-and-
wastewater-treatment-technology-update/?pagenum= 1.
303 The cost effectiveness at Big Brown Unit 1 would increase from $l,255/ton to $l,362/ton and at Big Brown 2
from $l,257/ton to $l,363/ton. The cost effectiveness at Monticello 1 would increase from $l,937/ton to $2,123/ton
and at Monticello 2 from $2,170/ton to $2,373/ton.
354
-------
• Engineering & Construction Management: 10% of the Base Module Cost
• Labor Adjustment: 10% of the Base Module Cost
• Contractor Profit/Fees: 10% of the Base Module Cost
• Owner's Costs: 5% of the Total CECC
• Allowance for Funds During Construction (AFUDC): 10% of the TPC
S&L noted that indirect capital costs included in the IPM cost modules do not account for all
project-related indirect costs a facility would incur to install a retrofit control, and do not include
all indirect capital costs allowed by the Control Cost Manual. For example, the IPM model does
not estimate, and EPA did not include, any project contingency.
S&L stated that indirect capital costs are defined in the Control Cost Manual to include "costs
such as; construction and contractor fees, startup and testing, inventory capital, and any process
and project contingency costs."54 Indirect installation costs include engineering costs;
construction and field expenses (i.e., cost for construction supervisory personnel, office
personnel, rental of temporary offices, etc.); contractor fees (for construction and engineering
firms involved in the project); start-up and performance test costs; contingencies; and costs for
working capital and offsite facilities.55
According to S&L, cost estimating examples included in the Control Cost Manual for all major
air pollution control systems include contingency as an indirect capital cost. For example, the
selective catalytic reduction (SCR) cost example includes a process contingency calculated at 5%
of direct costs and a project contingency calculated at 15% of total direct and indirect costs.56
Contingency is intended to represent unforeseeable elements of cost, particularly in fixed
investment estimates, which previous experience has shown to be statistically likely to occur,
and should be applied to those items that could incur a reasonable but unanticipated increase in
projects costs.57 The Control Cost Manual allows the analyst to include both process and project
contingencies, as applicable, in an estimate of the indirect capital costs. By excluding
contingency from its calculation of total capital costs, EPA failed to follow the methodology
described in the Control Cost Manual, and failed to include an important line-item in the capital
cost calculations.
S&L noted that EPA excluded Owner's Costs from the IPM cost modules based on its
conclusion that these costs "are disallowed by the 'overnight' cost method used in the Control
Cost Manual."58 The U.S. Energy Information Administration (EIA) defines overnight cost as
"an estimate of the cost at which a plant could be constructed assuming that the entire process
from planning through construction could be completed in a single day." EPA argues that the
concept is useful to avoid any impact of financing issues and assumptions on estimated costs;
however, Owner's costs do not represent costs associated with project financing. In fact, the EIA
explicitly includes Owner's Costs in its capital cost estimates for electricity generating plants.59
As discussed previously, Owner's Costs are real costs that the owner will incur during the project
and are typically included in cost estimates prepared for large air pollution control retrofit
projects.
Based on S&L's experience on large air pollution control system projects and given the nature of
355
-------
the work being proposed by EPA, it is reasonable to estimate Owner's Costs using a factor of 5%
of the total direct costs, and it is incorrect to exclude Owner's Costs from the cost estimate. By
excluding Owner's Costs from its calculation of total capital costs, EPA failed to follow the
methodology described in the Control Cost Manual (even though EPA said it was following the
Manual), and failed to include an important line-item in the capital cost calculation.
According to S&L, AFUDC accounts for the time value of money associated with the
distribution of construction cash flows over the construction period, which for an FGD system
could be spread over a construction period of 36 months or more. Total Capital Investment, as
defined in the Control Cost Manual, includes all costs required to purchase equipment needed for
the control system (purchased equipment costs), the costs of labor and materials for installing
that equipment (direct installation costs), costs for site preparation and building, working capital,
and off-site facilities.60 The EIA notes that overnight capital costs "serve as a starting point for
developing the total cost of new generating capacity" and that "other parameters also play a key
role in determining the total capital costs."61 Lead time is identified by the EIA as one of the
most notable parameters affecting total capital costs, as "[pjrojects with longer lead times
increase financing costs."62 Although the EIA starts with overnight cost estimates, other
parameters, including financing, lead time, and inflation of material and construction costs play a
key role in determining total capital costs.
As S&L discussed previously, AFUDC can represent a significant cost on large construction
projects with long project durations. Thus, including AFUDC in the capital cost estimate is
consistent with the constant dollar approach described in the Control Cost Manual, and excluding
AFUDC from the cost estimate will underestimate the total capital requirement for capital
intensive projects with extended project durations. By excluding AFUDC from its calculation of
total capital costs, EPA failed to follow the methodology described in the Control Cost Manual,
and failed to include an important line-item in the capital cost calculation.
Footnotes:
53 Control Cost Manual, Section 1, Chapter 2, page 2-5.
54 See, Control Cost Manual, Section 4.2, Chapter 1, page 1-32; and Chapter 2, page 2-43.
55 Id.
56 See, Control Cost Manual, Section 4.2, Chapter 2, page 2-44.
57 See, e.g., Control Cost Manual, Section 1, Chapter 2, page 2-30.
58 Cost TSD, Section 3.2, page 11, f/n 23.
59 EIA, "Updated Capital Cost Estimates for Utility Scale Electricity Generating Plants," April 2013, pg. 2.
60 Control Cost Manual, Section 1, Section 2, page 2-5.
61 EIA, "Updated Capital Cost Estimates for Utility Scale Electricity Generating Plants," April 2013, pg. 3
62 Id.
Response: This comment discusses in general the various factors included in indirect costs in
the IPM cost models and in the Control Cost Manual, but only specifically identifies project
contingency, AFUCD, and owners costs as omitted from the IPM cost models but included in the
Control Cost Manual.
Project Contingency
This comment claims that the Control Cost Manual requires project contingency, citing an
356
-------
example from the SCR chapter, which it asserts is excluded from the IPM cost models. The
Control Cost Manual does not "require" a contingency but rather states: "Due to the uncertain
nature of many estimates, analysts may want to add an additional contingency (i.e., uncertainty)
factor to their estimate. However, the retrofit factor is a kind of contingency factor and the cost
analyst must be careful to not impose a double penalty on the system for the same unforeseen
conditions."304 As the subject controls are all retrofits, an "average" retrofit contingency is built
into the estimates produced by the IPM cost models. As discussed elsewhere, EPA's costs are
based on worst-case assumptions. Thus, no further contingency is required.
Further, under the overnight cost estimating method in the Control Cost Manual, contingency has
a specific definition: "A contingency factor should be reserved (and applied to) only those items
that could incur a reasonable but unanticipated increase but are not directly related to the
demolition, fabrication, and installation of the system. For example, a hundred year flood may
postpone delivery of materials, but their arrival at the job site is not a problem unique to a retrofit
situation."305
In contrast, S&L urges a definition of "contingency" based on a different concept used in so-
called "all-in" costing, which is not used for regulatory cost analyses. Sargent and Lundy urges
this definition: "unforeseeable elements of cost, particularly in fixed investment estimates, which
previous experience has shown to be statistically likely to occur, and should be applied to those
items that could incur a reasonable but unanticipated increase in project costs." This definition is
inappropriately cited to the Control Cost Manual, Chapter 1, p. 2-30, which is actually the
definition for the overnight method, but is inaccurately cited by S&L. The overnight cost
methodology in the Manual does not contemplate contingency for changes in fixed investment
estimates that are statistically likely to occur because the costs are based on current prices.
The Control Cost Manual specifically excludes unanticipated increases due to demolition,
fabrication, and installation of the system. In contrast, S&L claims the Control Cost Manual
mandates a contingency to all project and process costs, including demolition, fabrication, and
installation, in direct contradiction to the fundamental definition in the Manual.
Thus, we believe that no change is required.
AFUCD and Owner's Cost
AFUDC and owner's costs are not valid costs under the Control Cost Manual. Please see our
other responses to comments elsewhere concerning these issues for more details.
Comment: [Earthjustice (0067) p. 37; Stamper (0068) p. 10] Earthjustice et al., and Stamper
stated that the EPA provided much of the details on its SO2 upgrade and retrofit cost
effectiveness analyses in its Technical Support Document for the Cost of Controls Calculations
for the Texas Regional Haze Federal Implementation Plan (Cost TSD), EPA-R06-OAR-2014-
0754-0008. For those units without a scrubber, EPA used version 5.13 of its Integrated Planning
304 Cost Control Manual, Sec. 2.5.4, p. 2-30.
305 Control Cost Manual, Sec. 2.5.4, p. 2-30.
357
-------
Model (IPM) to determine the capital and operations & maintenance costs for the retrofit SO2
controls of DSI, SDA, and wet FGD. Cost TSD at 3. EPA then determined cost effectiveness
based on the methods and principles of the EPA Air Pollution Control Cost Manual. The IPM
control costs are based on databases of actual pollution control project costs.27 Thus, the IPM
cost modules provide a solid basis for estimating control costs in a cost effectiveness analyses.
However, Earthjustice et al., stated that the EPA was overly conservative in its SO2 cost
effectiveness analyses in projecting emission reductions. EPA's approach of using the highest
monthly SO2 rate for determining the costs of the SO2 controls overestimates the operational
costs of the control by designing for a worst case coal. With higher sulfur content coal, more
reagent is needed to remove SO2, and more scrubber waste is generated.
Footnote:
27 See, e.g., Sargent & Lundy, IPM Model - Updates to Cost and Performance for APC Technologies, Wet FGD
Cost Development Methodology, Final March 2013, at 1 (Ex. 9).
Response: We agree with Earthjustice that our scrubber retrofit cost methodology basis was
sound. We also agree, for the reasons we have detailed in our responses to other comments, that
we made many conservative assumptions in our cost effectiveness calculations.
ll.f. $/deciview
Comment: The Costs of EPA's Proposal Are Extreme and Unjustified.
[Luminant (0061) p. 3]
Luminant stated that the EPA disregards the law of diminishing returns. The costs that EPA's
proposal would impose are staggering and well out of proportion to the imperceptible "visibility
benefits" that EPA projects. EPA calculates that its proposed emission limits would require just
four Texas companies to spend $2 billion for so-called "visibility improvements" that no person
will be able to detect.13 The vast majority of this cost that EPA projects—over $1 billion—would
be borne by Luminant alone under EPA's proposal. This is patently unreasonable and contrary
to any notion of "reasonable progress." EPA's own regional haze guidance says that "a dollar-
per-deciview calculation" is the more "meaningful" way to determine reasonable progress,14 yet,
tellingly, EPA refuses to provide that calculation here. In our comments, we provide the dollar-
per-deciview values that EPA failed to provide, and they demonstrate that EPA's proposal goes
well past the point of efficient and reasonable improvements in visibility and is substantially
more costly than measures that EPA has found to be not required in other states for reasonable
progress.
Footnotes:
13 See EPA, Technical Support Document for the Cost of Controls Calculations for the Texas Regional Haze Federal
Implementation Plan (Cost TSD) 24-5 (Nov. 2014) ("Cost TSD") (scrubber retrofit capital costs); id. at 55
(scrubber upgrade costs). See also 79 Fed. Reg. at 74,876-77 (same).
14 EPA, Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program 5-2 (June 1, 2007)
("EPA Reasonable Progress Guidance"), available at http://tinyurl.com/EPARPguidance.
[Luminant (0061) p. 134] Luminant noted, in its decision, EPA claims that its proposal "reduced
down to an analysis of whether, in light of the balance between the cost of control and visibility
358
-------
benefits of control at each source, additional SO2 controls should be installed on each of certain
large coal fired EGUs in Texas in order to improve the visibility at these Class I areas."828 EPA
further claims that "[t]he modeled benefits that would result from the installation of those
controls are reviewed, and the cost of controls are weighed against their projected visibility
benefits at a number of Class I areas."829
Yet, Luminant noted that nowhere in its proposal does EPA actually balance or compare the
costs it would impose with the visibility benefits it believes would be achieved. There is no
actual cost-benefit analysis to be found, and thus no basis for EPA's claim that the controls it
would require are cost-effective. There is a simple way to do this—instead of assessing the costs
on a dollar-per-ton basis, EPA could have evaluated costs on a dollar-per-deciview basis. As
EPA explains in its own reasonable progress guidance: "Therefore, in assessing additional
emissions reduction strategies for source categories or individual, large scale sources, simple cost
effectiveness estimates based on a dollar-per-ton calculation may not be as meaningful as a
dollar-per-deciview calculation, especially if the strategies reduce different groups of
pollutants."830
According to Luminant, EPA, however, does not conduct a cost-per-deciview analysis to support
its proposal. EPA thus fails to follow its own guidance and fails to explain why it has departed
from its guidance on this point. Indeed, EPA does exactly what it warned states against—it has
provided "simple cost effectiveness estimates based on a dollar-per-ton calculation" instead of
the more "meaningful" "dollar-per-deciview calculation."831 EPA looks at costs (in its Costs
TSD), and it separately looks at deciviews (in its FIP TSD), but it never joins the two, at least not
in any document in the record. Nor does EPA support its claim of cost-effective visibility
improvement with any data or benchmarks. EPA's only analysis of costs and benefits is
superficial and impossible to validate. For example, EPA says "scrubber retrofits for the Big
Brown units" are "cost effective" and the projected visibility benefits are "significant."832 Yet, it
never looks at costs in relation to benefits or attempts to quantify a cost/benefit ratio, much less
make comparisons of that metric as between units or control equipment. Indeed, the data that
EPA presents is incomplete and does not allow for any meaningful comparison among control
options.833 Amazingly, EPA "specifically solicits] comments on the appropriateness of. . .
scrubber retrofits" at the Monticello, Coleto Creek, and Tolk units "[i]n recognition of their
lesser cost/benefit ratio,"834 but EPA fails to explain or provide what EPA believes is the
cost/benefit ratio for those units (or any other units). EPA has thus denied the public the
opportunity to provide meaningful comment on what is apparently a cornerstone of its proposal.
According to Luminant, that EPA scrupulously avoids revealing the "cost/benefit ratio" that it
has apparently calculated, but failed to reveal, is telling.835 Although it is impossible to know
what "cost/benefit ratio" EPA is using to make its decision on which units to regulate and which
not, a basic comparison between costs and visibility benefit shows that EPA's proposed costs are
patently unreasonable. This is evident from simply comparing EPA's own estimates of cost and
visibility benefits. Table 14 compares the capital costs of the scrubber retrofits that EPA would
impose on Luminant units versus the visibility benefit as calculated by EPA.
359
-------
TABLE 14- f-PA'S ESTIMATED COSTS VERSUS ADJUSTED MODELED VISlRllllY tMPRQVEMENl FROM
ADDITIONAL SO CONTROLS A1 LtJMINANl UNITS AGAINST 2018 BACKGROUND (DLCIVIEWS, 2U°o
WORST DAYS. HIGH CONTROLS)
:5T^
[PA's
Estimated
Cost (Capital
Cost Only)
BiRE
Modeled Visibility
, 1 ' Am| t , K
GUMO
Modeled Visibility j
t" i f.t
ffp V¥Q] is
WfMO
Mkxislsd Visibility
Improvement'
(deeiviews, average
it")
—
irt 2
sr)
$259,141,000
i ¦ i ;¦;
0,0267 i
0.0883
1 i
$250,80-1000
0.0055
0,006.8 |
i
i
0,0507
2
$254, i 77,0(30
0,0465
Luminant noted that, as seen in Table 14, the capital costs alone that EPA estimates are in the
hundreds of millions of dollars for individual units, whereas the projected benefits are in the
hundredths of a deciview, and in some cases thousandths of a deciview. When annual operating
costs are included, the cost numbers go even higher. Nowhere does EPA cite, and we have been
unable to locate, any instance in which EPA or a state has determined that such miniscule
visibility benefits (not health benefits) justify such staggering costs. Indeed, EPA has recently
determined that visibility improvements of 0.04 deciview are "minimal" and do not justify more
stringent controls even at a small additional incremental cost.837 Here, EPA does not even assert
in its proposal that these negligible benefits justify the cost. They do not. EPA's proposal is
unreasonable on its face and arbitrary and capricious.
Luminant asserted, unlike EPA, Texas did consider costs in relation to visibility benefit and
explained its analysis. Using cost figures and estimated visibility benefits very similar to EPA '.v,
Texas determined that the massive costs were not reasonable for the imperceptible benefit.838
EPA does not take issue with Texas's cost or visibility improvement estimates. Instead, EPA
would simply reach a different conclusion from essentially the same data about what is
"reasonable." But this is not EPA's role under the statute and regulations. As EPA itself has
explained, "[tjhere is no particular threshold for determining significance of visibility benefit in
the regional haze rule."839 Further, "States have latitude to determine these thresholds," and "[a]s
long as this evaluation is done adequately and the states provide a reasoned basis for their
decisions, EPA will defer to the state "84° EPA has forgotten its own practices and standards
here. Texas fully explained and supported its cost effectiveness analysis, pursuant to EPA's
regulations and guidance, and EPA does not find otherwise. EPA must approve Texas's
submission.
Footnotes:
828 FIP TSD at 2
829 Id. at 11.
830 Reasonable Progress Guidance at 5-2. More recently, EPA has affirmed that "dollars per deciview is one of
several metrics that can be used to analyze cost of visibility improvement. . . ." 77 Fed. Reg. at 40,156.
831 Reasonable Progress Guidance at 5-2.
360
-------
832 FIP TSD at 31.
833 For example, EPA apparently has modeled, but it fails to consider or report, the visibility benefits from less
costly scrubber upgrades that it is requiring. 79 Fed. Reg. at 74,883 (reporting estimated deciview improvement
from only 95% control efficiency upgrades, but not less costly upgrades EPA modeled). Thus, EPA failed to
consider, for example, whether 90% SO2 removal at Martin Lake would yield similar deciview improvement or
would have a similar cost/benefit ratio.
834 FIP TSD at 31 (emphasis added).
835 For example, in creating Table A.6-4 for its FIP TSD. , which EPA says provides the deciview improvement for
the controls it is requiring, EPA removed (without explanation) the "$/ton" column from the docket spreadsheet
TX116-007-_33_Vis_modeling_summary, which would have presented both cost and "benefits" together and
allowed for calculation of a dollar-per-deciview figure.
836 Source for visibility figures: 79 Fed. Reg. at 74,881; FIP TSD at A-68 to A-70; TX116-007-
_33_Vis_modeling_summary (("WFGD" tab, baseline set as "3yr average 2009-2013 (eliminate max and min),"
"2018 environ (PSAT run)"). Source for cost figures: Cost TSD at 24.
837 80 Fed. Reg. 18,944, 18,994 (Apr. 8,2015).
838 2009 Texas SIP Narrative at 10-6 to 10-7.
839 77 Fed. Reg. 40,150, 40,156 (July 6, 2012) (approval of Nebraska SIP).
840 Id. (emphasis added).
[Luminant (0061) p. 137] Luminant stated, given EPA's failure to provide cost-per-deciview
calculations to back up its so-called "cost/benefit" analysis, we asked NERA Economic
Consulting to calculate the cost-per-deciview of EPA's proposal for each Texas unit that EPA
proposes to regulate in its FIP.841 Consistent with EPA's prior practice, NERA conducted its
analysis using annualized cost per deciview considering the Class I area that EPA contends
would see the most visibility improvement (Wichita Mountains). NERA also considered the
cumulative improvements that EPA contends would occur at the other Class I areas at issue.
Viewed in either manner, the cost-per-deciview values for EPA's proposed FIP are enormous
and patently unreasonable. As calculated by NERA (based on EPA's data and assumptions), for
the proposed scrubber upgrades (which are the only measures that EPA concedes can be in place
by the end of this planning period), the cost-per-deciview of EPA's proposal ranges from $129
million per deciview to $651 million per deciview, depending on the unit, considering the
visibility improvement that ENVIRON modeled at Wichita Mountains.842 The numbers are
significantly larger when considering the smaller improvements ENVIRON modeled at Big
Bend and Guadalupe Mountains.843 As we explain in Section 14, because, as EPA concedes, the
scrubber retrofits cannot be in place during this planning period, EPA has no authority to impose
them in a FIP. Thus, the benefits of those retrofits cannot justify EPA's proposal. But even if
the benefits of the scrubber retrofits could be considered, the hypothetical cost-effectiveness of
EPA's proposal is worse when the retrofits are included. As calculated by NERA (again, based
on EPA's own cost and benefit data), the cost-per-deciview of EPA's proposal ranges from $429
million per deciview to $1.3 billion per deciview considering the visibility improvement that
ENVIRON modeled at Wichita Mountains.844 Again, the numbers are significantly larger when
considering the smaller improvements ENVIRON modeled at Big Bend and Guadalupe
Mountains.845 NERA's analysis and calculations are set out fully in their report, a copy of which
is attached to these comments and incorporated herein by reference.
As these enormous cost-per-deciview values further illustrate, Luminant asserted that EPA's
proposal is unreasonable and would impose costs that are well out of proportion to the alleged
benefits. Indeed, as we explain elsewhere in these comments, EPA has understated the costs and
overstated the benefits of its proposal. Thus, the real cost-per-deciview of EPA's proposal would
361
-------
be substantially higher than calculated by NERA, which used EPA's claimed values.
Footnotes:
841 See NERA Economic Consulting, Technical Comments on Economic Issues in EPA's Proposed Regional Haze
Federal Implementation Plans for Oklahoma and Texas (April 20, 2015) ("NERA Report").
842 NERA Report at 6.
843 Id. at 8.
844 Id. at 11.
845 Id. at 11-14.
[NERA (0061) p. 4] In the report prepared for Luminant, NERA stated that EPA's cost
effectiveness analysis in its proposed FIP is based entirely on an evaluation of the estimated
dollars per ton of reduction in haze-forming emissions.6 Repeatedly, in EPA's FIP technical
support documents (TSD), the unit "$/ton" is noted in brackets after the use of the term "cost
effectiveness,"7 and it is the unit reported in tables for cost effectiveness.8 EPA is clearly
considering costs, not in relation to the visibility benefit produced, but in relation to the tons of
SO2 reduced.9
However, NERA noted that an economically-correct concept of "effectiveness" is the extent to
which people will benefit from additional controls. When the regulatory goal is visibility
improvement in Class I areas, benefits should be considered based on the improvement in
visibility per dollar spent, not the reduction in emissions. Although reductions in haze-forming
emissions do have some connection to improvement in visibility in Class I areas, not every ton of
emission has the same impact on visibility, and thus not every ton of emission has the same
impact on benefit. In addition to having different visibility impairment impacts per unit of
atmospheric concentration due to their different physical and chemical properties, the visibility
effectiveness of each ton reduced depends on the location of the emission source with respect to
the Class I area, the characteristics of that source, and prevailing atmospheric conditions that
disperse those emissions in different directions, among other variables.
According to NERA, in other words, not every ton is equally "effective" in reducing visibility
impairment, and thus dollars per ton reduced is not a reliable indicator of cost effectiveness. To
measure cost effectiveness, one should look at the cost of visibility improvement per deciview
(dv) rather than per ton of emission reduction. This distinction is made by EPA itself in its
Guidance for how to set RPGs:
Therefore, in assessing additional emissions reduction strategies for source
categories or individual, large scale sources, simple cost effectiveness estimates
based on a dollar-per-ton calculation may not be as meaningful as a dollar-per-
deciview calculation...1 "(emphasis added)
However, NERA asserted that EPA has not provided a dollar per deciview calculation or any
discussion of it in its proposed FIP for Texas. Nevertheless, the data necessary to make these
calculations are all available in the proposed FIP and associated TSDs, and we make those
calculations in this report. EPA has not explained how it would calculate a dollar per deciview
value for purposes of its proposal, and thus we are unable to evaluate or comment on EPA's
methodology.
362
-------
According to NERA, for the rest of this report, we will use the term "cost effectiveness" to refer
specifically to dollars per deciview. Except where otherwise noted, it will be computed as the
annualized cost of a proposed control (or set of controls) divided by the amount of deciview
improvement that it will provide on the average of the 20% worst-visibility days, at the Class I
area that EPA contends is most impacted {i.e., WIMO in this case).11
NERA noted, because the record contains the necessary information to compute dollars per
deciview, there is no reason for EPA to use the unreliable metric of dollars per ton to assess cost
effectiveness of alternative potential controls under the reasonable progress analysis. In the next
section, we explain how the information EPA has put into the record can be used to perform an
economically-valid cost-effectiveness analysis in lieu of the unreliable dollars per ton criterion.
EPA could have and should have calculated the economically-correct measure of cost
effectiveness but inexplicably EPA did not.
Footnotes:
6 79 Fed. Reg. at 74,884.
7 See for example, FIP TSD, pp. 12 and 30.
8 FIP TSD, Tables 4, 6, 8, 10, 12, and 14.
9 Multiple different types of emissions can contribute to visibility reduction. Some types reduce visibility in the
chemical form in which they are originally emitted, while other types of emissions first must be converted in the
atmosphere into new compounds that then reduce visibility, and contribute to regional haze. The different types of
haze-forming emissions result in different levels of impact on visibility impairment per unit of atmospheric
concentration due to their different physical and chemical properties. The proposed FIP focuses on potential
controls on emissions of SO2, which reduce visibility only after transformation in the atmosphere into sulfates.
Other haze-forming emissions include NOx dust, fly-ash, soot (elemental carbon), and some volatile organics. The
proposed FIP does not propose controls on any of these other haze-forming emissions.
10 EPA, Guidance for Setting Reasonable Progress Goals under the Regional Haze Program, Office of Air Quality
Planning and Standards, Research Triangle Park, NC, June 1, 2007 (rev). More recently, EPA has affirmed that
"dollars per deciview is one of several metrics that can be used to analyze cost of visibility improvement.. . ." 77
Fed. Reg. at 40,156.
11 Sometimes the cost per deciview may be computed as the dollars divided by the sum of all deciview changes at
every Class I area at issue. The "cumulative" dollars per deciview will always be smaller than the cost per deciview
at the most-impacted area. The important point is that comparisons of costs per deciview should always be
computed using the same approach.
[NERA (0061) p. 5] In the report prepared for Luminant, NERA first addressed the cost-
effectiveness calculations for the set of upgrades that represent the only controls described in the
proposed FIP that could legally be mandated by EPA during the first regional haze planning
period. This is done using (1) the air quality modeling results that were developed by
ENVIRON for CENRAP (and which were relied on by Oklahoma and Texas for their SIP
submissions and used by ENVIRON in its modeling for EPA), and (2) the revised visibility
benefits that EPA developed in its review. As we show, even if EPA's revised estimates of air
quality contributions were to be used, the proposed reasonable progress control requirements
remain profoundly inconsistent with the economic cost-effectiveness thresholds EPA has utilized
and relied upon in other states.
[NERA (0061) p. 6] In the report prepared for Luminant, NERA noted that Table 1 contains
information for each of the seven upgrade projects in the proposed FIP. Data on the SO2
reductions and the change in deciviews in each Class I area (for the average of the 20% worst
363
-------
visibility days at each respective Class I area) come from EPA TSDs and supporting
spreadsheets. Estimates of the annualized total costs are also based on EPA information.13 The
estimates in Table 1 are based on the CENRAP modeling that was originally performed for the
Texas SIP using source emissions projected to occur in 2018 and which EPA uses to calculate its
proposed RPGs for 2018 for these three areas.
NERA noted, using EPA's data, the dollars per deciview for each individual control does vary
from unit to unit, even though the estimated dollars per ton is presumed the same in every case
(i.e., $500/ton). The cost effectiveness, based on benefits at the most affected Class I area,
WIMO, ranges from $129 million per deciview to $651 million per deciview, depending on the
unit. In the bottom row, the aggregated cost of all the upgrades and their aggregated cost
effectiveness is shown to be $364 million per deciview of improvement at WIMO.
Table 1: The max and cumulative cost effectiveness ($/dv) of the proposed FIP scrubber
upgrades based on EPA's proposed Reasonable Progress Goal (CENRAP modeled 2018
emissions) [Table not excerpted]
NERA noted that the deciview impact at WIMO for each upgrade in Table 1 is between 0.008 dv
and 0.033 dv, depending on the unit. The aggregate deciview improvement of all these upgrades
at the maximally-impacted area on the 20% worst-visibility days of the year is 0.141 dv, as can
be seen in the bottom row of the table, and this is the value that EPA proposes for setting the
RPG for Wichita Mountains.14 Thus, Table 1 represents the true cost-effectiveness of EPA's
proposal, and it also shows the specific deciview improvement EPA relies on to justify its RPGs.
The specific deciview impact for each of the individual upgrades is never reported in the
proposed rule or supporting TSDs but can be obtained by careful examination of EPA's
supporting spreadsheets.15
NERA noted that even the aggregate amount of change in visibility at the maximally-impacted
area from all seven upgrades is not physiologically possible to perceive,16 which means that the
actual economic value of this change in the RPG is likely zero. Although EPA predicts that two
other Class I areas will potentially benefit from the controls it would require, Table 1 shows that
the other two Class I areas are so distant that their coincidental visibility improvement from these
same controls would be miniscule: less than 0.04 dv in aggregate at either BIBE or GUMO.
These benefits most certainly could not be viewed as justifying any of these upgrades over the
case at WIMO, as the economic cost effectiveness for any single one of those upgrades is much
worse: ranging from $369 million per deciview to $3.7 billion per deciview, depending on the
unit.
NERA also noted that the last column of Table 1 provides the cost per deciview if computed for
the cumulative impact to all three Class I areas of interest in the proposed FIP. As is
mathematically necessary, it is smaller than the cost effectiveness for the maximally-impacted
area; the estimate for the aggregate of all seven upgrades is $249 million per cumulative
deciview.
Footnotes:
13 Note these are just estimates and not the values calculated by EPA using confidential business information. We
did not review the confidential business information or EPA's unit-specific calculations. Per the FIP TSD, EPA
364
-------
states that the "cost effectiveness was less than $600/ton" (FIP TSD at page 27). Further, EPA has used $500/ton as
an "approximate cost/placeholdef' based on confidential business information data (see cells D61 to D69 in the
worksheet labeled "summary table" in "TX116-007- 33 Vis modeling summary.xlsx" in EPA Docket Folder
(http://www regulations.gov/#!documentDetail;D=EPA-R06-OAR-2014-0754-0007)). Thus, like EPA, we used
$500/ton as an approximate figure. The fact that we use EPA's estimates to demonstrate the lack of cost
effectiveness of those proposed controls does not imply that we are endorsing them.
14 See cell D8 in the worksheet labeled "proposal total vis 2018" in "TX116-007- 33 Vis modeling summary.xlsx" in
EPA Docket Folder (http://www.regulations.gov/#!documentDetail;D=EPA-R06-OAR-2014-0754-0007).
15 Specifically, when one opens the EPA docket file cited in the prior footnote, the tab called "dv summary" shows
dv changes based only on "Estimated deciview improvement from actual emissions (3-yr average annual emissions
2009-2013 eliminating min and max year)." However, the values in the tables shown in that tab are linked to more
detailed results in other tabs called "SDA," "WFGD" and "WFGD upgrade." In these other tabs, there is a
dropdown box to select the baseline of emissions. The version of the file in the docket has those drop-down boxes
set on the option "3yr average 2009-2013 (eliminate max and min)." However, a user can change the option to
"2018" to reflect the modeling results provided to EPA by ENVIRON prior to EPA's adjustments. When "2018" is
selected in each of those tabs, the deciview impacts are recalculated using the CENRAP projected 2018 emissions,
and one can then observe the unit-specific upgrade deciview impacts that are in Table 1 and which add up to the
0.141 aggregate value that EPA is using as the basis for its proposed RPG for Wichita Mountains.
16 The Regional Haze Rule "established] the deciview (dv) as the principal metric for measuring visibility," which
is "a useful measure for tracking progress in improving visibility, because each deciview change is an equal
incremental change in visibility perceived by the human eye." 77 Fed. Reg. at 30,249.
[NERA (0061) p. 8] NERA, in its report prepared for Luminant, stated that EPA also provides
alternative visibility impact estimates based on adjusted emission levels, even though these
estimates do not form the basis of EPA's proposed FIP. Even using EPA's visibility estimates
based on adjusted emissions {i.e., EPA uses the 3-year average annual tons per year based on
2009-2013 actual emissions, after eliminating the two years with the highest and lowest reported
emissions), the dollars per deciview remains extremely high {i.e., is extremely cost-ineffective).
Those calculations are provided in Table 2. Note that the unit-specific deciview impacts in Table
2 are the ones that EPA is using to justify these upgrades {i.e., they are the values that appear in
the Federal Register17) but that these are not consistent with the calculation of the 0.141 dv value
that EPA proposes for setting the RPG for WIMO based on the scrubber upgrades.18 These unit-
specific deciview impacts aggregate to a 0.281 dv impact shown on the bottom row of Table 2.
According to NERA, using these revised deciview impact estimates, which EPA apparently
believes is the best case for its proposal, the cost effectiveness based on benefits at the highest-
impacted Class I area (WIMO) ranges from $142 million per deciview to $223 million per
deciview, depending on the unit. In the bottom row, the aggregated cost of all the upgrades and
their aggregated cost effectiveness is shown to be $183 million per deciview of improvement at
WIMO.
NERA stated, even in this case, the upgrades would provide only 0.03 dv to 0.06 dv of impact at
the maximally-affected Class I area. In aggregate, all of the upgrades combined produce a
maximum of 0.28 dv of improvement. Thus, despite EPA's re-analyses to increase the purported
visibility benefit of its proposal, this is still not a humanly perceptible amount of change19 and,
again, the actual economic value of such a change in the RPG is likely zero.
NERA noted that although EPA predicts that two other Class I areas will potentially benefit from
the controls it would require, the table shows that the other two Class I areas are so distant that
their coincidental visibility improvement from these same controls would be miniscule: less than
365
-------
0.07 dv in aggregate at either BIBE or GUMO even using EPA's higher emissions assumption.
These benefits most certainly could not be viewed as justifying any of these upgrades over the
case at WIMO, as the economic cost effectiveness for any single one of those upgrades is much
worse: ranging from $338 million per deciview to $1.6 billion per deciview.
NERA noted that the last column of Table 2 provides the cost per deciview if computed for the
cumulative impact to all three Class I areas of interest in the proposed FIP. As is mathematically
necessary, it is smaller than the cost effectiveness for the maximally impacted area; the estimate
for the aggregate of all seven upgrades is $123 million per cumulative deciview if using EPA's
adjusted emission levels.
Table 2: The max and cumulative cost effectiveness ($/dv) of the proposed FIP scrubber
upgrades based on EPA's alternative approach (3-year average annual tons per year (2009-2013),
eliminating min and max years) [Table not excerpted]
Footnotes:
17 Tables 34, 35 and 36 in 79 Fed. Reg. at 74,881-2.
18 Tables 43 and 44 in 79 Fed. Reg. at 74,887.
19 See footnote 16.
[Associations (0059) p. 17-19] The Associations stated that the EPA's proposed FIP for Texas is
also arbitrary and capricious because the costs of the emissions controls that EPA would require
are excessive in comparison to the minimal visibility benefits that they would provide. In
evaluating the costs of the proposed emission control requirements, EPA disregards its own
guidance and appears to apply only a cost per ton methodology that fails to account for important
differences in the emissions that impair visibility as well as the role that location and other
facility-specific factors can play in determining the effect that emissions controls will have on
visibility in the Class I areas that are the focus of the regional haze provisions. As a result, EPA
singles out a handful of sources that would be required to spend billions of dollars to install
emissions controls that would result, at most, in miniscule improvements in visibility in Class I
areas that would be imperceptible to observers.
The Associations noted, unlike generally applicable emissions standards that are designed to
improve air quality everywhere, the regional haze provisions are focused specifically on a
limited number of Class I areas throughout the country. As a result, not all emissions reductions
will have the same impacts on visibility in Class I areas. Key factors such as the type of
pollutant at issue, distance from Class I areas, and prevailing winds can all affect the degree to
which certain emissions will contribute to visibility impairment and, as a result, the visibility
benefits that will be produced by reducing those emissions. In other words, not every ton of
emissions reductions is the same. Recognizing this fact, EPA explains in guidance that "in
assessing emission reduction strategies for source categories or individual, large scale sources,
simple cost effectiveness estimates based on a dollar-per-ton calculation may not be as
meaningful as a dollar-per-deciview calculation, especially if the strategies reduce different
groups of pollutants." EPA, Guidance for Setting Reasonable Progress Goals Under the
Regional Haze Program 5-2 (June 1, 2007). EPA has reaffirmed that view in subsequent SIP
reviews, stating its belief "that dollars per deciview is one of several metrics that can be used to
366
-------
analyze cost of visibility improvement." 77 Fed. Reg. 40,150, 40,156 (July 6, 2012). That is,
because of distance, wind patterns and other relevant meteorological factors, even emissions that
might be quite inexpensive to reduce may have no meaningful impact on downwind visibility.
The Associations stated that the EPA fails to apply its own guidance in evaluating the cost
effectiveness of its proposed FIP. While EPA provides an evaluation of dollars per ton of
emissions reduction, 79 Fed. Reg. at 74,876-77, T. 32, it makes no effort to connect those costs
to actual visibility benefits. Instead, EPA simply asserts, without providing a detailed
cost/benefit comparison that emissions controls at individual units are "cost effective" and that
visibility benefits are "significant." See EPA, FIP Technical Support Document at 31. In fact,
while EPA separately evaluates the cost of the proposed FIP and the visibility benefits, it never
compares them directly in an attempt to evaluate the cost effectiveness of the proposed controls.
See generally EPA, FIP Technical Support Document; EPA, Cost Technical Support Document.
Instead, EPA asserts that the Unfunded Mandates Reform Act, which requires a written cost
benefit analysis, does not apply. 79 Fed. Reg. at 74,889-90. By failing to include the relevant
cost benefit analysis in the record, it is impossible to verify EPA's assertion that in preparing the
FIP, "the cost of controls [were] weighted against their projected visibility benefits at a number
of Class I areas." EPA, FIP Technical Support Document at 11.
According to the Associations, had EPA conducted a proper dollars per deciview cost benefit
analysis, it would be readily apparent that the emission controls in the proposed FIP are not cost
effective. In total, EPA projects that installing the proposed emission controls at six facilities
would cost approximately $1.8 billion. &e79Fed. Reg. at 74,876-77, T32. However, the
effect of those emission controls on nearby Class I areas would be imperceptible to the human
eye. In fact, in 2018 the visibility conditions would improve at Big Bend from 16.6 to 16.57
deciviews and at Guadalupe Mountains from 16.3 to 16.26 deciviews. See id. at 74,887. In
other words, EPA proposes to impose enormous costs to improve visibility by a few hundredths
of a deciview. Even by EPA's own metrics these costs are excessive in comparison to the
benefits. Indeed, the human eye cannot detect changes in visibility of less than one deciview
and, under EPA's own statistical standards, these 2018 "improvements" would be treated as
nonexistent.4 That is why, in another FIP proposal, EPA recently concluded that a similar
incremental visibility improvement was minimal and could not justify the much smaller
difference in cost between wet and dry SO2 scrubbers. EPA, Arkansas FIP Proposal,
Prepublication Version 160-61 (Mar. 6, 2015). Imposition of such massive costs without
achieving any cognizable visibility benefit during the interim planning period cannot be
considered cost effective. Thus, in light of the statutory obligation to consider the costs of
compliance, 42 U.S.C. § 7491(g)(1), EPA's proposed FIP is unreasonable, arbitrary, and
capricious.
Commenter's Reference:
4 U.S. EPA, Technical Support Document for Demonstration of the Transport Rule as a BART Alternative 24, n.24
(Dec. 2011), Docket ID No. EPA-HQ-OAR-2011-0729-0014 ("All differences that are < 0.05 [deciviews] were
rounded down to 0.0 and are considered to be no degradation").
Response: At various points in their comments the commenters either wrongly imply we used
the $/ton metric as the sole indicator of whether a particular control was justified, or that we did
not consider the visibility improvement from those controls. We carefully compared the cost
367
-------
effectiveness of our proposed controls to their modeled visibility benefits. We disagree that the
$/dv metric is more meaningful than our use of the $/ton metric in conjunction with our
consideration of the visibility benefit from the installation of controls. As we note in our
Oklahoma FIP:306
[T]he BART Guidelines require that cost effectiveness be calculated in terms of
annualized dollars per ton of pollutant removed, or $/ton.307 OG&E provided a
$/deciview analysis for its units and comparable BART determination performed
by us. In our analysis for our BART FIP for OG&E and AEP/PSO, we did not
evaluate $/deciview. We explain that the BART Guidelines list the $/deciview
metric as an optional cost effectiveness measure that can be employed along with
the required $/ton metric for use in a BART evaluation. The metric can be useful
in comparing control strategies or as additional information in the BART
determination process; however, due to the complexity of the technical issues
surrounding regional haze, we have never recommended the use of this metric as
a cutpoint in making BART determinations. We note that to use the $/deciview
metric as the main determining factor would most likely require the development
of thresholds of acceptable costs per deciview of improvement for BART
determinations for both single and multiple Class I analyses. We have not
developed such thresholds for use in BART determination made by us. As
OG&E acknowledges, EPA did not use this metric as part of its proposed BART
determinations for either the Four Corners Power Plant FIP in AZ, or the San Juan
Generating Station FIP in NM. Generally speaking, while the metric can be
useful if thoughtfully applied, we view the use of the $/deciview metric as
suggesting a level of precision in the calculation of visibility impacts that is not
justified in many cases. While we did not use a $/deciview metric, we did,
however, consider the visibility benefits and costs of control together, as noted
above by weighing the costs in light of the predicted visibility improvement.
Our decision was reviewed and upheld in Oklahoma v. EPA, 723 F.3d 1201 by the Tenth Circuit
which ruled:
Oklahoma first suggests EPA should not have rejected the visibility analysis it
conducted in the SIP, which used the dollar-per-deciview method. This argument
is misguided. The EPA rejected the SIP because of the flawed cost estimates.
When promulgating its own implementation plan, it did not need to use the same
metric as Oklahoma. The guidelines merely permit the BART-determining
authority to use dollar per deciview as an optional method of evaluating cost
effectiveness. See 40 C.F.R. pt. 51 app. Y(IV)(E)(1).308
306 Response to Technical Comments for Sections E. through H. of the Federal Register Notice for the Oklahoma
Regional Haze and Visibility Transport Federal Implementation Plan, Docket No. EPA-R06-OAR-2010-0190,
12/13/2011, pdf 116.
307 70 FR 39167.
308 We note, however, that in both its final rule and in its brief the EPA asserts that the guidelines require the use of
the dollar-per-ton metric in evaluating cost effectiveness. The guidelines themselves are a bit unclear. In the section
on cost effectiveness, the guidelines mention only the dollar-per-ton metric. 40 C.F.R. pt. 51 app. Y(IV)(D)(4)(c).
368
-------
And in the final rule, the EPA explained why it did not use the dollar-per-
deciview metric used by Oklahoma. "Generally speaking, while the metric can be
useful if thoughtfully applied, we view the use of the $/deciview metric as
suggesting a level of precision in the calculation of visibility impacts that is not
justified in many cases." 76 Fed.Reg. at 81,747. The EPA has never mandated the
use of this metric, and has not developed "thresholds of acceptable costs per
deciview improvement." Id. While the federal land managers have developed
thresholds, these thresholds were apparently developed without input from the
EPA and without notice-and-comment review. EPA Br. at 54 n. 13. In light of
this, we do not find it arbitrary or capricious that the EPA chose not to use the
dollar-per-deciview metric in evaluating BART options in creating the FIP. We
therefore also conclude that any argument by the petitioners that the dollar-per-
deciview measurement proves the scrubbers are not cost effective lacks merit.
See Pet. Reply Br. at 16.
We see no reason, despite the fact that the facilities we evaluated in our proposed TX/OK FIPs
were done under the reasonable progress and long-term strategy sections of the Regional Haze
Rule, to deviate from our view of the dollar per deciview metric here. We also note that the use
of the dollar per deciview metric is further complicated in the present case due to our use of
CAMx modeling. As we discuss in our proposal and elsewhere in our response to comments,
there is no way to directly compare the CAMx modeling we used in our proposed TX/OK FIPs
with previous CALPUFF modeling results used for BART because of differences in the models,
model inputs, and metrics used.309 Many of these differences result in CAMx modeled visibility
impacts and benefits to be much lower than CALPUFF modeled visibility impacts and benefits
relied on in other actions. For example, one difference between the two model analyses is that
CALPUFF modeling is focused on the maximum impact from the modeled source, whereas the
CAMx modeling is focused on the average impact over the 20% worst days as determined by the
monitored data during the baseline. Consequently, even if we were to use the dollar per
deciview metric in our TX/OK FIPs, we would be unable to effectively compare the results
based on our CAMx modeling against other modeling and cost analyses, the vast majority of
which employed CALPUFF. The commenter's estimates do not consider these differences.
We disagree with Luminant and other commenters that we did not balance the cost of our
proposed controls with their modeled visibility benefits. This information is prominently
discussed in both our FIP TSD and in our Federal Register notice. For instance, Appendix A of
our FIP TSD very thoroughly develops our modeling strategy and concludes with a comparison
of the visibility improvement we expected to result from the controls we proposed. We review
this information in the in Section 7 of the main body of our FIP TSD, and in Sections 8 and 9, we
compare that visibility benefit to its cost effectiveness in proposing our control set.
However, the guidelines later state that in evaluating alternatives, "we recommend you develop a chart (or charts)
displaying for each of the alternatives" that includes, among other factors, the cost of compliance defined as
"compliance — total annualized costs ($), cost effectiveness ($/ton), and incremental cost effectiveness ($/ton),
and/or any other cost-effectiveness measures (such as $/deciview)." Id. app. Y(IV)(E)(1) (emphasis added).
309 FIP TSD at A-3 5
369
-------
Luminant and other commenters have made several comparisons to our previous actions. We
address those comments in the consistency section of this document.
Additionally, EPA was not required under Title II of the Unfunded Mandates Reform Act of
1995 (UMRA), Pub. L. 104-4, to provide a written cost-benefit analysis because, as a threshold
matter, the proposal is not a "rule" for purposes of Title II of UMRA. Therefore, the
requirements of UMRA do not apply to the proposal. Specifically, 2U.S.C. Section 1502(1)
provides that all terms in Title II of UMRA have the meanings set forth in 2 U.S.C. Section 658,
which in turn further provides that the terms "regulation" and "rule" have the meanings set
forth in 5 U.S.C. Section 601(2). Under 5 U.S.C. Section 601(2), "the term 'rule' does not
include a rule of particular applicability relating to. . . facilities." Because this proposed rule is a
rule of particular applicability relating to eight named facilities, EPA has determined that it is not
a "rule" for the purposes of Title II of UMRA.
In summary, we reject the commenters' assertions that we did not properly compare the cost
effectiveness of our proposed controls to their modeled visibility benefits. We also reject the
commenters' assertions that this type of comparison requires the $/dv metric. Even if we had
placed emphatic focus on the $/dv metric, we would have had no ability to compare its results to
previous actions by either us or the states, because of the inability to reliably compare deciviews
from our CAMx modeling to CALPUFF modeling, which was the modeling platform most often
used in previous BART and reasonable progress/long-term strategy analyses.
12. Cost versus visibility benefit
Comment: The proposed scrubber retrofits are even less cost effective than the upgrades
[NERA (0061) p. 11]
In its report prepared for Luminant, NERA stated that the proposed FIP also includes seven
scrubber retrofits, although the visibility benefits from these cannot be included as justification
for EPA's FIP, which covers only the first regional haze planning period that ends in 2018 and
EPA acknowledges that none of them can be operational before the end of 2018. These controls
therefore fail under the statutory factor to consider the time necessary for compliance.
Nevertheless, because EPA seems to justify its proposal based on estimated benefits from these
controls, we have examined their cost-effectiveness as well. NERA found that the retrofits also
cannot be justified because they entail an even higher dollar per deciview cost.
NERA noted that table 3 presents EPA's estimates of the costs of each of these seven proposed
retrofits, and their modeled impact on deciviews at each of the three Class I areas discussed in
the proposed FIP. EPA justifies five of the retrofits based on their contributions to haze at
WIMO, and those have cost effectiveness estimated between $559 million per deciview and
$868 million per deciview with respect to their maximally-impacted area, WIMO. EPA justifies
two of the retrofits (the spray dryer absorbers, or SDAs, at the Tolk units) based on their
contributions to haze at GUMO, and those cost $1.1 to $1.3 billion per deciview with respect to
that maximally-impacted area. The aggregate visibility improvement on the 20% worst days at
WIMO from these seven retrofits is projected to be 0.304 dv, which is also imperceptible. The
370
-------
aggregate deciview impact at the other two Class I areas is about one-third of the impact at
WIMO.
NERA stated, even if EPA's alternative emission values were to be used, the cost effectiveness
of the proposed retrofits with respect to their highest-impact areas ranges from $429 million per
to over $1.3 billion per deciview, as shown in Table 4. Note that the unit-specific deciview
impacts in Table 4 are the ones that EPA is using to justify these retrofits {i.e., they are the values
that appear in the Federal Register20), but that these are not consistent with the calculation of the
0.304 dv value that EPA uses to calculate its proposed RPG for Wichita Mountains "[ajssuming
all controls in place by 2018," which EPA itself admits cannot be accomplished.21 These unit-
specific deciview impacts aggregate to a 0.331 dv impact shown on the bottom row of Table 4.
Again, NERA noted that the visibility improvement is less than humanly perceptible: a combined
benefit of 0.331 dv for the average of the 20% worst days at WIMO, and about one-third that at
either of BIBE or GUMO.
According to NERA, the last columns of Tables 3 and 4 provide the cost per deciview if
computed for the cumulative impact to all three Class I areas of interest in the proposed FIP. As
is mathematically necessary, these are smaller than the cost effectiveness for the maximally
impacted area. The estimates for the aggregate of all seven retrofits are $467 million and $430
million per cumulative deciview, using the CENRAP 2018 projected emissions or EPA's
adjusted unit emission levels, respectively. Again, the fact that we use EPA's estimates to
demonstrate the lack of cost effectiveness of those proposed controls does not imply that we are
endorsing them. We simply find that, even under EPA's best case, the proposed controls are
cost-ineffective.
Table 3: The max and cumulative cost effectiveness ($/dv) of the proposed FIP scrubber retrofits
(CENRAP modeled 2018 emissions) [Table not excerpted]
Table 4: The max and cumulative cost effectiveness ($/dv) of the proposed FIP scrubber retrofits
based on EPA's alternative approach (3-year average annual tons per year (2009-2013),
eliminating min and max years) [Table not excerpted]
Footnotes:
20 Tables 34, 35, and 36 in 79 Fed. Reg. at 74,881-2.
21 Table 44 in 79 Fed. Reg. at 74,887.
Response: We agree with NERA that our proposed scrubber retrofits are less cost effective
(higher $/ton) than our proposed scrubber upgrades. This is completely normal and expected,
since a scrubber upgrade uses most of the existing scrubber system.
We believe all of our proposed controls are cost effective, which is particularly demonstrated by
our close examination of the applicable values for the cost per ton of emission reduced. We
disagree with NERA that our proposal to require the operation of controls beyond 2018 violates
either the Clean Air Act or our Regional Haze Rule and we address that issue in responding to a
separate comment. We also disagree with the commenter that the Regional Haze Rule requires
that controls result in perceptible visibility improvement. Also, we did not use the metric NERA
371
-------
uses—dollar per deciview—because we do not believe it is an appropriate point of focus when
there are, as in this case, demonstrated impacts and controls demonstrated to be cost effective on
a cost/ton basis. We direct NERA to our responses to other comments in which we respond to
these specific issues in more detail.
Comment: CCP stated that the EPA rejected TCEQ's cumulative approach in favor of a unit-
specific approach. As above, this rejection was improper and TCEQ's cumulative approach is
actually more consistent with the statute, regulations, and even EPA's guidance. But even under
a unit-specific approach, EPA's analysis and conclusions are deficient. In the Proposed Rule,
EPA "propose[s] to find . . . that a scrubber installation on the CCP unit would also yield
significant visibility benefits." 79 Fed. Reg. 74,882. CCP disagrees with this proposed
determination. EPA's model significantly overstates the visibility improvements on which EPA
is basing its proposal. When properly adjusted to account for this recognized over-prediction
bias, the impacts from Coleto Creek Unit 1 are insufficient to justify installation of controls.
CCP's own modeling further demonstrates the lack of impact of emissions from Coleto Creek
Unit 1 on visibility conditions and the lack of an improvement in visibility that can be attributed
to the proposed controls.
Response: We describe why we proposed to disapprove the TCEQ's four factor analysis in our
proposal310:
The TCEQ constructed a large potential control set consisting of a mix of large
and small sources, located at various distances from Class I areas, with a large
geographical distribution. Because of the variation in size, type, and location of
these sources, the potential to impact visibility and potential benefit from controls
at a given Class I area can vary greatly between the identified sources. This
potential control set identified by the TCEQ included controls on some sources
that would likely result in significant visibility benefits, but also included controls
on many sources with much less anticipated visibility benefits. Because it only
estimated the visibility benefit of all the controls together, the TCEQ was not able
to assess the potential benefit of controlling individual sources with significant,
and potentially cost-effective, visibility benefits. Also, we believe that individual
benefits were masked by the inclusion of those controls with little visibility
benefit that only served to increase the total cost figures. For example, the TCEQ
identified SO2 controls at Big Brown to be approximately $l,500/ton,
significantly less than its $2,700/ton threshold. These controls were estimated to
achieve greater than 40,000 tpy SO2 emission reductions. Despite this evidence in
the record of an identified cost-effective control that results in large emission
reductions, and source apportionment modeling identifying large impacts from
EGU sources in northeast Texas, the TCEQ did not separately evaluate the
visibility benefit from the implementation of this control, or appropriately weigh
the four reasonable progress factors in determining the reasonableness of this
individual control.
310 79 FR 74838.
372
-------
For the reasons described above, we believe that the TCEQ's four factor analysis was flawed.
We disagree that the TCEQ's approach can be approved or that it is consistent with the CAA, our
guidance, or our regulations. The comment that our model overestimates the anticipated
visibility improvement from controls and that the commenter's modeling demonstrates a lack of
visibility impacts from the source are addressed in a separate comment response. In brief
response here, we disagree with the comment and believe we used the appropriate model and
model results in our determination that controls on Coleto Creek unit 1 will yield significant
visibility benefits at Wichita Mountains and Big Bend.
Comment: Luminant provided a summary of EPA's analysis of the costs of controls for the
"smaller subset" of Texas sources that it selected through its visibility analysis. [Luminant
(0061) p. 45]
Luminant explained that for the 21 coal-fired generating units that EPA identified at Step 5 of its
visibility analysis (plus San Miguel), EPA subsequently conducted a cost analysis for the
additional SO2 controls that it had determined yielded significant visibility benefits.316 For the 10
units equipped with scrubbers, EPA "calculate[d] the costs of upgrading that scrubbing
system."317 For the 11 units without scrubbers, EPA "calculate[d] the costs for DSI [dry sorbent
injection], dry scrubbing, and wet scrubbing."318 In performing these calculations, EPA says it
"relied on the methods and principles contained within the EPA Air Pollution Control Cost
Manual."319 However, as discussed below, EPA's analysis deviated significantly from the
Control Cost Manual that it claimed to be relying on, and its cost estimates substantially
understate the costs and overstate the potential for SO2 reductions.
For the 10 units with scrubber systems already installed and in operation, EPA's analysis
assumed: 1) that the existing system had already been upgraded to an SO2 removal efficiency of
at least 95%, or could be upgraded to that level with "proven equipment and techniques"; 2) that
scrubber bypasses could be "eliminated" and the additional flue gas could be treated by system at
95% removal efficiency or greater; and 3) that additional changes necessary to eliminate bypass
(including adding fan capacity, upgrading the electrical distribution system, and conversion to a
wet stack) could be performed using "proven equipment and techniques."320 With these
assumptions—which, as we discuss below, EPA does not support—EPA calculated the costs and
concluded that "in all cases, the cost effectiveness was less than $600/ton" of SO2 removed.321
EPA conceded that its "calculations of scrubber efficiency may contain some error" that required
"adjustments" to its cost-per-ton estimates, but it explained that regardless of the "error in
calculating the true tons of SO2 removed," "we would still propose to upgrade these SO2
scrubbers."322 In other words, the actual cost-per-ton was ultimately irrelevant to EPA's decision.
For the 11 units without scrubbers, EPA calculated the costs of DSI (at both 50% and 90%
removal), dry scrubbing, and wet scrubbing. The SO2 removal efficiencies that EPA analyzed
for dry scrubbing and wet scrubbing varied from unit to unit, without explanation.323 EPA used
the results of its costs estimates to compare the different types of controls to each other.324 EPA
concluded that, for all but one of the units, "the cost-effectiveness of DSI was worse (higher
$/ton) than either SDA or wet FGD" and that "[a]t the higher 80% or 90% level of control, the
cost-effectiveness of DSI was worse than either SDA or wet FGD in all cases."325
373
-------
"Consequently," EPA explained, "we are not proposing that DSI be installed at any unit."326 As
between dry scrubbing (SDA) and wet scrubbing (FGD), EPA found that "[t]he capital cost of
wet FGD is higher than SDA by approximately 8-13%."327 But EPA found that "the cost-
effectiveness ($/ton) of wet FGD was better than SDA in all cases except for the Tolk and Welsh
units,"328 "mainly due to the greater level of control (98% maximum versus 95% maximum) of
wet FGD over SDA... "329
Ultimately, EPA concluded that "[g]iven the greater visibility improvement of wet FGD over
SDA, we propose to base our cost/benefit reasonable progress and long-term strategy
determination on wet FGD, except for the Tolk units, due to their potential water issue."330 For
example, with respect to Luminant's units, EPA found that the capital cost for a wet scrubber at
Big Brown Unit 1 would be $256,032,000, and the capital cost for dry scrubbing would be
$226,656,000.331 The so-called "greater visibility improvement," according to EPA, from the
additional $29.4 million for the wet scrubber at Big Brown Unit 1 was projected to be 0.001
deciview at Big Bend and Guadalupe Mountains and 0.003 deciview at Wichita Mountains
(2018, 20% worst days).332 For Monticello Unit 2, the capital cost difference between dry
scrubbing and wet scrubbing was approximately $26 million,333 and the so-called "greater
visibility improvement" would be zero at Big Bend and Guadalupe Mountains and 0.001
deciview at Wichita Mountains (2018, 20% worst days).334 However, EPA does not actually
compare the costs for the projected visibility improvement between the different controls at the
different units.
Footnotes:
316 Cost TSD at 1.
317 Id.
318 Id.
319 Id. at 2.
320 Id. at 55.
321 Id.
322 79 Fed. Reg. at 74,885.
323 Id. at 74,876.
324 Id. at 74,884. EPA explained that it followed the Control Cost Manual and used the "overnight" method of cost
estimation for this reason. Cost TSD at 2.
325 79 Fed. Reg. at 74,884.
326 Id.
327 Cost TSD at 25.
328 79 Fed. Reg. at 74,884.
329 Cost TSD at 25.
330 79 Fed. Reg. at 74,884 (emphasis added).
331 Cost TSD at 24.
332 79 Fed. Reg. at 74,881-82.
333 Cost TSD at 24.
334 79 Fed. Reg. at 74881-82.
Response: We take no position on Luminant's narrative of our proposal. Regarding the
remainder of Luminant's comments, as we discuss in our FIP TSD311:
We present a reasonable progress and long-term strategy cost analyses for those
311 FIP TSD, page 11.
374
-------
units being analyzed for DSI or scrubber retrofits in which we assess the cost of
DSI, SDA, and wet FGD. The modeled benefits that would result from the
installation of those controls are reviewed, and the cost of the controls are
weighed against their projected visibility benefits at a number of Class I areas.
We then propose which units should install SO2 control equipment and the control
level those units should achieve. Please see our Cost TSD for more detail on how
we performed the cost analysis and Appendix A to this TSD for more details
about how we conducted our visibility analysis.
We note that elsewhere we have addressed Luminant's objection to our consideration of
visibility in proposing the reasonable progress and long-term strategy aspects of our FIP. As we
explain elsewhere in our responses to comments, we believe that we properly considered
visibility in proposing whether additional controls should be installed by the affected facilities.
Nevertheless, as we also noted in our FIP TSD, our proposal to required controls was heavily
influenced by the generally greater cost effectiveness offered by wet FGD over SDA:312
With the exception of Tolk, all of the scrubber retrofits were analyzed on the basis
of both SDA and wet scrubbers. The SDA level of control was assumed to be a
maximum of 95% not to go below 0.06 lbs/MMBtu. The wet FGD level of
control was assumed to be a maximum of 98% not to go below 0.04 lbs/MMBtu.
As we discuss in our Cost TSD, the cost-effectiveness ($/ton) of wet FGD was
better than SDA in all cases except for the Tolk and Welsh units, which burn
Power River Basin (PRB) coal. However, even in those cases, the cost-
effectiveness of wet FGD was only 0.5 to 0.8% greater than SDA. Given the
greater visibility improvement of wet FGD over SDA, we propose to base our
cost/benefit reasonable progress and long-term strategy determination on wet
FGD, except for the Tolk units, due to their potential water issue.
As we indicate above, we examined cost in the form of cost effectiveness ($/ton) which with
visibility are the familiar metrics we have long used to assess BART. The cost-effectiveness
calculations for the controls are based on the tons reduced and the estimated annualized cost of
controls, which not only considers capital costs, but also operational expenses. The asserted cost
differences provided by the comment do not consider the additional costs of operating and
maintaining the control equipment. The annualized costs of operating SDA was greater than
WFGD for all the evaluated units. Thus, the capital costs of wet FGD may be higher as
Luminant notes, but the cost effectiveness was generally more favorable (lower $/ton) due to the
greater control (more tons reduced) and lower operating costs of wet FGD. Furthermore, as we
indicate above, the visibility benefits from wet FGD were greater than those from SDA. In
summary, wet FGD was more cost-effective than SDA in terms of $/ton and provided for more
visibility benefit at all the units evaluated for scrubber retrofits at Big Brown, Monticello, Coleto
Creek, and Parish. At Tolk and Welsh, wet FGD is only slightly less cost-effective (0.3 to 0.8%
greater) than SDA, but produces visibility benefits 3 to 4% larger. We therefore disagree with
Luminant that we did not properly consider cost of controls at each unit.
Luminant is incorrect in stating that "the SO2 removal efficiencies that EPA analyzed for dry
312 FIP TSD page 30.
375
-------
scrubbing and wet scrubbing varied from unit to unit, without explanation." We explained that
the SDA level of control was assumed to be a maximum of 95% not to go below 0.06
lbs/MMBtu and the wet FGD level of control was assumed to be a maximum of 98% not to go
below 0.04 lbs/MMBtu.313
As to the comment that the actual cost-per-ton for scrubber upgrades was ultimately irrelevant to
our decision, we disagree with the comment and address it fully in a separate response to
comment.
Comment: [NERA (0061) p. 3] In a report prepared for Luminant, NERA stated that:
In any reasonable decision process, but especially in light of EPA's regional consistency
regulations,5 measures that involve exceptionally cost-ineffective controls in some States when
compared to those required in other States should have other extenuating benefits, such as highly
absolute perceptible benefits, if they are still to be required.
• No such extenuating benefits have been or can be identified for the proposed reasonable
progress controls in Texas.
• Even when considered in aggregate, neither the seven FGD upgrades nor the seven
scrubber retrofits are projected to produce a visibility improvement at any Class I area
that nears 0.5 deciviews.
[NERA (0061) p. 19] Furthermore, we find they are inconsistent with the cost-to-benefit ratio
that EPA has utilized in other States, and there are no discernible differences in the benefits that
they might provide to Class I areas that Texas sources are projected to affect. Such
inconsistency in how regulations are devised and applied to different states is arbitrary and
unlawful. This inconsistent application imposes an unfair social cost burden on certain
populations of the country, while providing no discernable difference in visibility. It is also in
conflict with EPA's own regional consistency regulations.
In sum, NERA concluded that EPA's justifications for its proposed incremental control measures
are logically flawed, legally inappropriate, and also - as documented in these comments -
unjustified under standard economics principles for policy evaluation.
Footnotes:
5 40 C.F.R. § 56.5(a)(2) (requiring that EPA regional offices "shall assure that actions taken under the act. .. [a]re
as consistent as reasonably possible with the activities of other Regional Offices")
[Xcel Energy (0064) p. 1] SPS owns and operates the Tolk Generating Station near Muleshoe,
Texas, which EPA proposes to regulate under the FIP. The Tolk Generating Station operates
two coal-fired, steam-electric generating units with a total power production capacity of 1,067
megawatts. EPA's Proposal concludes that retrofitting both Tolk units with dry scrubbers to
control emissions of SO2 is an appropriate control application to demonstrate reasonable
progress, despite imperceptible visibility improvements in the Class I areas analyzed in this
313 79 FR 74876 (December 16, 2014)
376
-------
Proposal.
Response: For the reasons we have discussed in our responses to other comments, we disagree
with NERA that our proposal included cost-ineffective controls or that we are in conflict with
our consistency regulations. We also disagree that NERA has properly characterized Section
56.5, which states:
§56.5 Mechanisms for fairness and uniformity—Responsibilities of Regional
Office employees.
(a) Each responsible official in a Regional Office, including the Regional
Administrator, shall assure that actions taken under the act:
(1) Are carried out fairly and in a manner that is consistent with the Act and
Agency policy as set forth in the Agency rules and program directives,
(2) Are as consistent as reasonably possible with the activities of other Regional
Offices, and
(3) Comply with the mechanisms developed under §56.4 of this part.
(b) A responsible official in a Regional Office shall seek concurrence from the
appropriate EPA Headquarters office on any interpretation of the Act, or rule,
regulation, or program directive when such interpretation may result in
inconsistent application among the Regional Offices of the act or rule, regulation,
or program directive.
(c) In reviewing State Implementation Plans, the Regional Office shall follow the
provisions of the guideline, revisions to State Implementation Plans—Procedures
for Approval/Disapproval Actions, OAQPS No. 1.2-005A, or revision thereof.
Where regulatory actions may involve inconsistent application of the
requirements of the act, the Regional Offices shall classify such actions as special
actions.
As can be seen, there is no reference to "should have other extenuating benefits, such as highly
absolute perceptible benefits." Also, for the reasons we have discussed in a number of other
responses to comments, we disagree that our proposal is inconsistent with the cost-to-benefit
ratio we have utilized in other states. As we have noted and discussed in separate response to
comment, the results of the CAMx modeling we have utilized in our proposal cannot be directly
compared to the results of CALPUFF modeling, which has been utilized in the vast majority of
other BART and reasonable progress/long-term strategy actions because of differences in the
models, model inputs, and metrics used.314 Many of these differences result in CAMx model
results to be much lower than CALPUFF modeling results relied on in other actions. Thus, we
disagree with NERA that our "justifications for [our] proposed incremental control measures are
logically flawed, legally inappropriate." Lastly, NERA has not provided any reference for it
314 FIP TSD at A-35
377
-------
assertion that our proposal is "unjustified under standard economics principles for policy
evaluation." We believe our proposal adheres to all applicable law, our regulations, and our
guidance.
Comment: EPA's Proposal Would Achieve No Detectable Change in Visibility.
[Luminant (0061), p. iii, 3]
[Luminant (0061), p. 3] Luminant stated that the overall change in visibility that EPA would
mandate in 2018 (which, again, has already been achieved) is miniscule and provides no
reasonable basis for EPA's proposed emission controls. The human eye can generally only
detect a change of 1.0 deciview or more. Yet, EPA projects that its proposal would, at most,
improve visibility in 2018 (the interim date at issue) at Big Bend by 0.03 of a deciview, at
Guadalupe Mountains by 0.04 of a deciview, and at Wichita Mountains by 0.14 of a deciview.
(79 FR 74887) Not only is this level imperceptible to any visitor at these Class I areas—and well
out of proportion to the massive projected costs—the required "improvements" at Big Bend and
Guadalupe Mountains round to zero, under EPA's own convention of rounding to the nearest
tenth of a deciview.
[Luminant (0061), p. iii] Luminant stated that perhaps the most glaring problem with EPA's
alternative is that it would produce no perceptible changes in visibility beyond what Texas's plan
achieves in this planning period, but at a cost of approximately $2 billion more. The vast
majority of these costs, EPA concedes, are for measures that cannot even be implemented by the
interim goal of 2018 and are thus outside the agency's authority to impose in the first place.
Luminant stated that the regional haze program is not a health-based program that requires
emission reductions at any cost and on any schedule conceived by EPA, but instead is a program
designed to achieve reasonable incremental improvements in visibility at specific areas over a
long-term horizon. Yet EPA's replacement proposal for Texas would not achieve any noticeable
visibility improvements in the three Class I areas in Texas and Oklahoma.
Luminant noted that visibility is measured in deciviews. Anything less than one deciview
generally can't be perceived by the human eye. As the table below shows, EPA's alternative
would, under EPA's own estimate, result in imperceptible changes in visibility measured in
tenths and hundredths of a deciview.
Summary of Visibility Improvement Provided by Luminant (0061)
378
-------
EPA'S PROPOSAL ACHIEVES NO iSCEKIMK IMPROVEMENT
Class 1 Area
State Established Goal
(2010) (20* worst
days)
trn rroposeo VjOai
(2011) pox wont
days)
Difference Between
State Goal and EPA
V*
i 6 ..2 I? dV
Wicfirta Mountains
21.47 dv
2133 dv
0.14*
In other words, EPA would force $2 billion in costs on a few Texas companies in hopes of
achieving imperceptible improvements in visibility.
Luminant stated that although no one will discern the difference from EPA's proposal, every
person in Texas will be exposed to increased costs of electricity and risks to reliability. There is
no form of cost-benefit analysis that could justify such an approach, and, unsurprisingly, EPA
does not even try to offer one.
[Luminant (0061) p. 60] Luminant stated, in other words, the basis upon which EPA would
second-guess Texas's reasonable progress analysis—and for which it would impose billions of
dollars in emission control measures on a handful of select Texas sources—is stated in the
hundredths of a deciview: 0.03 deciview improvement at Big Bend and 0.04 deciview
improvement at Guadalupe Mountains. And in terms of achieving the national goal of natural
visibility conditions, EPA's proposal would have no meaningful benefit, under even EPA's
calculations. EPA projects that its RPGs would achieve natural visibility at Big Bend in the year
2198 (compared to 2206 with no further controls); at Guadalupe Mountains in the year 2163
(compared to 2169 with no controls); and at Wichita Mountains in the year 2095 (compared to
2101 with no controls)429
Footnote:
429 See TX116-007-_33_Vis_modeling_summary ( "2018 RPG calcs").
[Associations (0059) p. 17-19] Indeed, the human eye cannot detect changes in visibility of less
than one deciview and, under EPA's own statistical standards, these 2018 "improvements"
would be treated as nonexistent.4
Commenter's Reference:
4 U.S. EPA, Technical Support Document for Demonstration of the Transport Rule as a BART Alternative 24, n.24
(Dec. 2011), Docket ID No. EPA-HQ-OAR-2011-0729-0014 ("All differences that are < 0.05 [deciviews] were
rounded down to 0.0 and are considered to be no degradation").
[NERA (0061) p. 19] NERA stated that EPA's proposed FIP for Texas includes upgrades for
existing FGDs as well as retrofits of SO2 scrubbers on electricity generating units in Texas.
However, even in aggregate, EPA projects these upgrades/retrofits would cause no perceptible
improvement on visibility in Class I areas in the region. The actions in the proposed FIP for
Texas are either infeasible to implement within the first planning period (i.e., the scrubber
retrofits), and/or not cost effective from an economic perspective of producing actual benefits to
379
-------
people who live near or visit Class I areas.
The TCEQ argued that reductions to sources that do not have any perceptible impact are not
effective regardless of their cost. The regional haze program is designed to improve visibility.
The analysis approach completed by the TCEQ was to determine potential, cost -effective
controls that would have a perceptible impact on visibility at a Class I area. The program was
not designed to make reductions because reductions were possible, nor is that required by either
the CAA or the RHR.
[Texas Governor (0066) p. 1-2] The Texas Governor stated that this dispute boils down to a fight
over so-called "decisiveness, more accurately, a fight over fractions of a "deciview." "A
deciview is a haze index derived from calculated light extinction, such that uniform changes in
haziness correspond to uniform incremental changes in perception across the entire range of
conditions, from pristine to highly impaired." 40 C.F.R. section 51.301. The human eye only
can detect a change in haziness of 1.0 or more deciviews. E.g., 79 Fed. Reg. 58,302, 58,303.
EPA nonetheless proposed to FIP the State of Texas because it wanted to reduce haziness at Big
Bend by merely 0.12 deciviews and at the Guadalupe Mountains by merely 0.15 deciviews—
reductions that fall dramatically below the threshold of visibility. 79 Fed. Reg. 74,818, 74,887
tbl. 44.
According to the Texas Governor, EPA's actions are unlawful on its own terms. The Clean Air
Act gives EPA authority only over the "impairment of visibility." 42 U.S.C. section 7491(a)(1)..
And "visibility," of course, extends only to the things that humans can see with their naked eyes,
e.g., Webster's Third New International Dictionary 2557 (1981) ("visible" means "capable of
being seen"; "visibility" means "the degree or extent to which something is visible . . . [by] the
observer's eye unaided by special optical devices"). The statute obviously does not give EPA
authority to regulate invisible haze, which falls far below the 1.0-deciview threshold.
AECT stated that each of the projected total visibility improvements is much less than 0.5 dv,
which, it is critical to note, EPA has previously determined is the appropriate visibility
improvement threshold against which to compare the predicted visibility improvement due to an
individual source installing additional emissions control for reasonable progress purposes.22
Footnotes:
22 77 Fed. Reg. at 30464
NRG stated that even if EPA's modeling and emissions inventories were accepted, additional
controls would be unjustified because there would be no change to visibility as a result of EPA's
proposed control measures. At most, EPA projects that its proposed controls would result in a
0.284 deciview visibility improvement associated with NRG's facilities.1 This is a visibility
change that is less than one-third of what the human eye can detect.
NRG stated that scrubber upgrades at NRG's Limestone plant would not meaningfully improve
visibility even if EPA's analysis were taken at face value. EPA's modeling results suggest that
upgrading the Limestone units' scrubbers to achieve 95% control would result in a visibility
change of no more than 0.284 deciview compared to what EPA has described as average natural
conditions, and no more than 0.057 deciview compared to EPA's projection for 2018. See 79
380
-------
Fed. Reg. at 74,883. Such deciview changes are imperceptible to the human eye, as a single
(1.0) deciview is the lowest level of visibility change that the human eye can detect.
Response: We disagree with Luminant and others that the Regional Haze Rule requires that
controls on a source or group of sources result in perceptible visibility improvement.315 As we
noted in our TSDs, we derived much of our approach to analysis of control costs and visibility
impacts from the BART Guidelines. In a situation where the installation of BART may not
result in a perceptible improvement in visibility, the visibility benefit may still be significant, as
explained by the Regional Haze Rule:316
Even though the visibility improvement from an individual source may not be
perceptible, it should still be considered in setting BART because the contribution
to haze may be significant relative to other source contributions in the Class I
area. Thus, we disagree that the degree of improvement should be contingent
upon perceptibility.
We accordingly disagree that selection of control measures should be contingent upon
perceptible visibility improvement. As we stated in our previous rulemaking addressing
the BART determinations in Oklahoma:317
Given that sources are subject to BART based on a contribution threshold of no
greater than 0.5 deciviews, it would be inconsistent to automatically rule out
additional controls where the improvement in visibility may be less than 1.0
deciview or even 0.5 deciviews. A perceptible visibility improvement is not a
requirement of the BART determination because visibility improvements that are
not perceptible may still be determined to be significant.
The Regional Haze Rule recognizes this on a basic level when it requires that a source with a 0.5
dv impact at a Class I area "contributes" to visibility impairment and must be analyzed for
BART controls.318 Because it is generally recognized that a 1.0 dv change in visibility is just
barely perceptible, controlling such a source will not result in a perceptible visibility change.
Consequently, even though the visibility improvement from an individual source may not be
perceptible, it should still be considered because the contribution to haze may be significant
relative to other source contributions in the Class I area. Thus, in our visibility improvement
analysis, we have not considered perceptibility as a threshold criterion for considering
improvements in visibility to be meaningful. Rather, we have considered visibility improvement
in a holistic manner, taking into account all reasonably anticipated improvements in visibility,
and the fact that, in the aggregate, improvements from controls on multiple sources will
contribute to visibility progress towards the goal of natural visibility conditions. Visibility
impacts below the thresholds of perceptibility cannot be ignored because regional haze is
produced by a multitude of sources and activities which are located across a broad geographic
area. In this action, as discussed below, we found that the required cost-effective controls reduce
315 It is generally recognized that a change invisibility of 1.0 deciview is humanly perceptible.
316 70 FR 39104, 39129
317 76 FR 81739.
318 Moreover, BART determining authorities are free to establish thresholds less than 0.5 dv.
381
-------
visibility impairment from those sources with the largest visibility impacts and result in
meaningful visibility benefits towards the goal of natural visibility conditions. Furthermore, as
we have noted and discussed in separate response to comment, the results of the CAMx
modeling we have utilized in our proposal cannot be directly compared to the results of
CALPUFF modeling, which has been utilized in the vast majority of other BART and reasonable
progress/long-term strategy actions, because of differences in the models, model inputs, and
metrics used.319 Many of these differences result in CAMx modeled visibility impacts and
benefits that are much lower than the CALPUFF modeled visibility impacts and benefits relied
on in other actions.
We also note that Luminant provides incorrect estimates of cost with respect to visibility. The
commenter presents estimated visibility benefits of only the scrubber upgrades but then contrasts
them to the total cost of scrubber upgrades and scrubber retrofits combined. The scrubber
upgrades are only a fraction of the total cost (less than 20% of the total capital costs) and the
visibility benefits due to the scrubber upgrades are only a fraction of the total anticipated
visibility benefit from all required controls.
We disagree with the commenters that the projected visibility improvements at Big Bend of 0.03
dv and Guadalupe Mountains of 0.04 dv should round to zero. The commenter refers to the air
quality modeling analysis performed in support of the determination that the trading programs in
CSAPR achieve greater reasonable progress towards the national goal of achieving natural
visibility conditions in Class I areas than source-specific Best Available Retrofit Technology
(BART) in the 28 states covered by CSAPR.320
For the specific purposes of the CSAPR analysis, we determined it was appropriate in
considering if visibility conditions degraded between projected visibility conditions for a future
year baseline and projected visibility conditions under CSAPR at a Class I area, that a difference
of less than 0.05 dv should be considered no degradation. The emission levels and the
geographic scope of these two scenarios are vastly different. The modeled SO2 reductions from
the CSAPR were -3.8 million tons per year and NOx reductions were -125,000 tons per year.
The analysis showed a large visibility benefit of up to 5.7 dv at Class I areas located close to the
where the emissions reductions occurred, in the Eastern U.S. However, the CSAPR analysis also
examined visibility impacts at all of the Class I areas in the continental U.S. In the Western U.S.
(hundreds of miles from where the emissions reductions occurred), there were very small
differences in visibility between the two scenarios, some of which were in the 0-0.05 dv range.
Given the very large differences in emissions between the CSAPR scenarios and the large
distances between these Class I areas and the regions with the largest anticipated emission
reductions, we determined it was appropriate to consider a difference between the scenarios of
less than 0.05 dv as no degradation for the purposes of demonstrating that visibility does not
decline in any Class I area under CSAPR.
In contrast, in the Texas FIP, we are focused on assessing the visibility benefit of emission
reductions from a small group of sources at the nearest Class I areas. The emissions reductions
319 FIPTSDat A-35.
320 U.S. EPA, Technical Support Document for Demonstration of the Transport Rule as a BART Alternative (Dec.
2011), Docket ID No. EPA-HQ-OAR-2011-0729-0014.
382
-------
in this case are 100,000 tons of SO2 from scrubber upgrades and 130,000 tons of SO2 from 7
scrubber retrofits. Because the number of sources and the emission reductions are much smaller,
a different threshold in determining significance of visibility benefits is appropriate. A visibility
difference due to reductions of 3.8 million tons of SO2 and 125,000 tons of NOx spread over
thousands of facilities is very different than a reduction of 100,000 tons spread over 7 units or a
total of 230,000 tons spread across 14 units. Beyond what has been applied in practice in various
SIPs and FIPs, there is no visibility deciview significance threshold that exists in regulation or
guidance. As a result, we interpret the visibility improvement from the modeling results on a
case-by-case basis. We believe that it is appropriate to apply different interpretations of
rounding conventions based on the circumstances of the emissions reductions scenario(s) and
geographic scope of the analysis.
Furthermore, the commenter presents estimated visibility benefits of only a subset of required
controls, the scrubber upgrades, and not the total visibility benefit of all required controls. The
estimated benefit of all required controls is 0.12 dv at Big Bend and 0.12 at Guadalupe
Mountains. We also note that our estimates for the amount of visibility benefit from the required
controls based on consideration of recent actual emission levels rather than the 2018 CENRAP
projected emissions is 0.07 dv due to the scrubber upgrades at both Big Bend and Guadalupe
Mountains, and 0.17 dv (Big Bend) and 0.20 dv (Guadalupe Mountains) when considering all
required controls. Therefore, even though we do not agree with the commenter that the CSAPR
rounding convention is appropriate, the estimated visibility benefits in this case exceed 0.05 dv.
We disagree with commenters that the visibility benefits from controls are miniscule. We
observe that several comments that are critical of the extent of the visibility benefits have cited
only to benefits from the scrubber upgrades, omitting the total anticipated visibility benefit from
all required controls. As we discuss in the FIP TSD and in separate response to comments, we
believe it is necessary to consider visibility benefits based on "clean" natural background
conditions to assess the full potential for visibility benefits from controls. For example, we
estimated that the required controls provide for over 3 dv improvement at Wichita Mountains
when estimated using a "clean" background. On a "dirty" background approach based on 2018
CENRAP projected visibility conditions, we estimate all required controls would improve
projected visibility conditions in 2018 by 0.45 dv over the visibility conditions projected by
CENRAP and Texas and an estimated 0.62 dv improvement in visibility when considering recent
actual emissions (values are for the 20% worst days). The required controls result in a greater
than 5% improvement in overall visibility conditions at Wichita Mountains on the 20% worst
days. We also estimate that the required controls are a significant acceleration of the final goal,
reducing the number of years to meet natural visibility by 25 to 30 years at Big Bend and
Guadalupe Mountain.
The CENRAP modeling showed that Texas sources have significant visibility impacts at
Wichita Mountains and the Texas Class I areas. Our analysis identified those point sources with
the largest contributions to visibility impairment at these Class I areas and the required controls
reduce visibility impairment from those sources with the largest impacts where controls were
determined to be available and reasonable for this first planning period. For example, the
Monticello and Big Brown facilities are projected to contribute approximately 1.3 Mm"1 and 1.2
Mm"1, respectively to visibility impairment on the 20% worst days at Wichita Mountains in 2018
383
-------
based on the CENRAP 2018 projected emissions for these facilities. This is 1.7% and 1.5% of
the total visibility impairment at Wichita Mountains. In our FIP TSD we noted that Texas used
an impact extinction level threshold of 0.5 Mm"1 (a level less than half of the estimated impact
from the Monticello or Big Brown facilities) from all sources in a state as a threshold for inviting
a state to consult. Oklahoma selected a threshold of 1.0 Mm"1 to determine which states should
consult in analyzing visibility impairment at Wichita Mountains. We also noted that the largest
projected contribution from all point sources within a state at Wichita Mountains after Texas
(14%>) is Oklahoma at 3.9%. In other words, elimination of all point sources in Oklahoma would
result in less visibility benefit (3.9%>) than the required controls (greater than 5%>). As these facts
demonstrate, the identified facilities have significant impacts on visibility conditions. Our
technical record makes it equally plain that the required controls reduce impacts from these
sources and result in meaningful visibility benefits towards the goal of natural visibility
conditions.
We disagree with Luminant that our proposal will result in reliability issues and address that
comment in more detail elsewhere in our responses to comments.
Comment: [Luminant (0061) p. 63] Luminant stated that Texas's consideration of visibility
benefits, in relation to cost, in its four-factor analysis was reasonable. Texas's reference to a 0.5
dv threshold (also used in the BART context) was a reasonable choice. As EPA itself has
explained, "States have wide latitude to determine [the] thresholds" "for determining
significance of visibility benefit in the regional haze rule."447 Indeed, EPA has said that states
may consider a threshold cut-off as high as 1.0 deciview improvement for an individual source,
below which the source would not be regulated.448
According to Luminant, Texas's choice of 0.5 deciview as a benchmark for total visibility
improvement (from all sources) to use in its four-factor analysis was reasonable and consistent
with EPA guidelines. Under the BART Guidelines, a source "contributes to any visibility
impairment," and thus becomes subject to BART, if it has an impact greater than 0.5 deciview at
any Class I area.449 It is thus logical that a level of visibility improvement at a single Class I area
that is less than the threshold at which a source becomes subject to BART in the first place
would be deemed insignificant for all sources. Indeed, in other regional haze actions, EPA has
"defer[red]" to states' consideration of the 0.5 deciview threshold.450 And given Congress's
special emphasis on BART sources,451 Texas's reference to the BART 0.5 deciview threshold to
evaluate reasonable progress for the first planning period was conservative, and Texas could
reasonably determine that total visibility benefits below the BART threshold for an individual
source should be deferred until a later planning period for reasonable progress. Moreover, even
absent reference to the BART threshold, it was eminently reasonable for Texas to reject as
unreasonable the enormous costs that would be required to achieve what EPA itself concedes
would be an imperceptible change in visibility.452
Footnotes:
446 Id. at 9.
447 77 Fed. Reg. at 40,156.
448 Id. at 40,156 n.14.
449 See 70 Fed. Reg. at 39,161.
384
-------
450 77 Fed. Reg. at 40,156.
451 42 U.S.C. § 7491(b)(2)(B) (carefully defining the universe of sources that must install BART and requiring
compliance with BART "as expeditiously as practicable").
452 Further, even if Texas's use of 0.5 deciview as a reference point were not appropriate, that does not justify EPA's
use of other thresholds and metrics, found nowhere in the regulations or guidance (as discussed in Section VII.C), to
decide which sources to regulate in its FIP and by how much.
Response: The first quote Luminant references is only partially reproduced and is inaccurate.
The full and actual quote from our guidance is:
There is no particular threshold for determining significance of visibility benefit
in the regional haze rule. Significance is a source- and Class I-specific
evaluation, meaning that it depends on how much visibility improvement is
needed at the Class I area(s), how much a specific source impacts the Class I
area(s), and the cost effectiveness and potential visibility improvement of
available control options. States have latitude to determine these thresholds,321
providing support and a reasonable and adequate basis for why they selected the
thresholds, and to determine BART and reasonable progress controls, in
consultation with other impacted states. As long as this evaluation is done
adequately and the states provide a reasoned basis for their decisions, EPA will
defer to the state.
As can be seen from the full and accurate representation of our guidance, states have some
latitude to determine thresholds of visibility significance. However, states also have an
obligation to, "[provide] support and a reasonable and adequate basis for why they selected the
thresholds, and to determine BART and reasonable progress controls, in consultation with other
impacted states." In our proposal, we describe why we believe the TCEQ's four factor analysis,
which includes the consideration of visibility benefits, was flawed322:
The TCEQ constructed a large potential control set consisting of a mix of large
and small sources, located at various distances from Class I areas, with a large
geographical distribution. Because of the variation in size, type, and location of
these sources, the potential to impact visibility and potential benefit from controls
at a given Class I area can vary greatly between the identified sources. This
potential control set identified by the TCEQ included controls on some sources
that would likely result in significant visibility benefits, but also included controls
on many sources with much less anticipated visibility benefits. Because it only
estimated the visibility benefit of all the controls together, the TCEQ was not able
to assess the potential benefit of controlling individual sources with significant,
and potentially cost-effective, visibility benefits. Also, we believe that individual
benefits were masked by the inclusion of those controls with little visibility
321 14 BART guidelines at 70 FR 39170: However, we believe the States have flexibility in setting absolute
thresholds, target levels of improvement, or de minimis levels since the deciview improvement must be weighed
among the five factors, and States are free to determine the weight and significance to be assigned to each factor.
For example, a 0.3, 0.5, or even 1.0 deciview improvement may merit stronger weighting in one case versus another,
so one "bright line" may not be appropriate.
322 79 FR 74838.
385
-------
benefit that only served to increase the total cost figures. For example, the TCEQ
identified SO2 controls at Big Brown to be approximately $l,500/ton,
significantly less than its $2,700/ton threshold. These controls were estimated to
achieve greater than 40,000 tpy SO2 emission reductions. Despite this evidence in
the record of an identified cost-effective control that results in large emission
reductions, and source apportionment modeling identifying large impacts from
EGU sources in northeast Texas, the TCEQ did not separately evaluate the
visibility benefit from the implementation of this control, or appropriately weigh
the four reasonable progress factors in determining the reasonableness of this
individual control.
We believe that the TCEQ's consideration of visibility benefits was flawed because the visibility
benefits of controls on individual sources or the group of sources with the largest visibility
impacts controls was masked by the TCEQ's methodology. Following our guidance, we are
unable to defer to the state in this case. Texas' evaluation was not done adequately, and the
evaluation did not contain a reasonable consideration of costs and visibility, leading to an
unreasoned basis for the decisions made with that analysis.
We further disagree with Luminant that Texas' choice of a 0.5 dv visibility threshold, including
the manner in which it was applied, was proper in its analysis. First, the quote Luminant
reproduces from our BART Guidelines was based on CALPUFF modeling and not CAMx
modeling. Texas extrapolated results from CAMx modeling to estimate the visibility
improvement due to all the identified controls in their analysis and then compared it to a
threshold developed for CALPUFF modeling. As we state in the FIP TSD and discuss in detail
in our response to comments, "[a] common metric used in BART visibility modeling using
CALPUFF is the BART screening level of 0.5 del-dv used by most states for screening out
facilities from further BART consideration. However, there are a number of factors that make
the two analyses uniquely different and not comparable, invalidating the use of the BART
screening metric, or other such comparisons with modeled visibility impacts for RP with CAMx
or CMAQ." In the FIP TSD and in separate responses to comments we discuss the differences in
the models, model inputs, and metrics used.323 Many of these differences contribute to CAMx
modeled visibility impacts and benefits for reasonable progress being much lower than the
CALPUFF modeled visibility impacts and benefits for BART relied on in other actions. As
detailed in the FIP TSD, these differences include the emission rates modeled, the metrics used
and whether the deciview impacts are calculated based on "clean" natural background conditions
or a "dirty" background based on degraded visibility conditions projected for 2018. The
CALPUFF emissions modeled for BART are representative of maximum emission rates and are
therefore usually significantly larger (often in the range of double) than average emission rates
used in CAMx modeling for a reasonable progress analysis. One of the main metric differences
is that the CALPUFF analysis for BART utilizes a clean background and compares the 8th
highest daily maximum impact from the specific source modeled to compare against a 0.5 dv
threshold to indicate significant impacts while the visibility benefit that was estimated by Texas
to assess the benefit of additional controls for reasonable progress was based on a "dirty" or
degraded background and average benefits over the 20% worst days observed by the monitor at
the Class I area which may or may not be inclusive of the highest impact days from the specific
323 FIP TSD at A-35 and modeling section of the this document
386
-------
source modeled with CALPUFF for BART. As we discuss in detail in the FIP TSD, because the
deciview metric is a logarithmic function of extinction, visibility impacts and improvement
calculated based on "dirty" conditions are substantially lower than those calculated based on
natural "clean" conditions.324 These differences were not considered in Texas' selection of
threshold. We note that Texas did calculate visibility impacts compared to natural visibility
conditions and focused on the maximum impact from the modeled sources in their BART
visibility analysis, which also relied on CAMx photochemical modeling, to determine the
significance of visibility impacts from BART sources for BART screening purposes. However,
in assessing the benefit of additional controls for reasonable progress, Texas only considered
visibility benefits averaged over the 20% worst days based on a "dirty" or degraded background.
The difference between comparing visibility improvement on a "clean" and "dirty" background
is analogous to comparing the change in sound volume that would occur if one person stopped
singing loudly in an empty room (clean background) to the change that would occur if one
person stops singing loudly in a room crowded with a 100 people singing loudly (dirty
background). In both cases, to return the room to natural background sound level, the individual
singers must be addressed, but there will be little or no perceptible difference in volume when
one singer in the crowded room stops singing. To carry the analogy further, our analysis was
designed to identify the Texas sources with the greatest visibility impact (the loudest singers) and
address them in this first planning period.
Second, the 0.5 dv threshold in the context of BART is used to assess the maximum total
visibility impact from all BART units at a facility. If the impact from all the BART sources at a
facility is above the threshold, then each BART unit must be evaluated for controls, and
therefore the visibility improvement anticipated from controls would be less than 0.5 dv on a
facility basis, and much less than 0.5 dv on a unit specific basis for BART sources with multiple
BART units. For these reasons, the BART threshold of 0.5 dv has no relation to the analysis
Texas performed and is inappropriate. We also note that we discuss in the preamble to the final
Regional Haze Rule and Guidelines for BART Determinations that a threshold less than 0.5 dv
may be appropriate.325
Even setting aside Texas' approach of aggregating sources with varying impacts on visibility, the
use of a 0.5 dv threshold as applied by Texas for determining the significance of visibility
benefits of all controls combined would have ensured that little visibility improvement would
occur during this planning period. Texas and Oklahoma acknowledged in their SIP submittals
that sources in Texas have a large impact on visibility at the Wichita Mountains; indeed, the
visibility impacts at this Class I area from Texas point sources are several times greater than the
324 FIP TSD at A-38. "For example, see Figure A.3-5 which shows the del-dv change due to a 10 (1/Mm) change at
both the 2018 projected extinction level ["dirty background"] and the 2064 natural visibility conditions ["clean
background"] extinction level for the Wichita Mountains. In the 'dirty background' case the 10 (1/Mm) yields a
1.26 del-dv, whereas in the 'cleanbackground' case the same 10 (1/Mm) yields a 3.86 del-dv improvement. In this
example, the 'clean background' situation yields a del-dv improvement 3 times greater than the 'dirty background'
for the same level of extinction improvement.
325 "..., if there were 100 sources each changing visibility by 0.1 deciviews, the total impact would be a 10-deciview
change invisibility. In this hypothetical example, all 100 sources would be contributing, in equal amounts, to
substantial visibility impairment..,." 70 FR 39121.
387
-------
impacts from Oklahoma's own point sources. Based on CENRAP 2018 modeling, all point
sources in Texas combined have a visibility impact in terms of light extinction of 10.58 Mm-1 at
the Wichita Mountains, which based on "dirty" 2018 CENRAP projected background conditions
equals a 1.34 dv impact for the 20% worst days. Therefore, adopting the 0.5 dv threshold, using
Texas' approach to assessing reasonable progress measures, would require the identification of a
control set large enough (and with a correspondingly large total cost) to address over one-third of
the total impacts from all Texas point sources, before the visibility benefit would be considered
significant. To put this into context, achieving the national goal at the Texas Class I areas will
require just over ten deciviews of improvement (approximately a reduction in light extinction of
35 Mm-1), a task that EPA has estimated could reasonably take until 2064. Given that the
Regional Haze Rule recognizes that improving visibility is an iterative process that will take
many years, declining to establish any additional measures to ensure reasonable progress until
Texas could identify a combined set of cost-effective and affordable controls that could achieve
0.5 dv or more improvement is unreasonable, especially when there are cost-effective and
affordable controls that result in meaningful visibility improvements towards the goal of natural
conditions. We also note that delaying even incremental action during this first planning period
pushes out the likely date of achieving natural conditions well past 2064.
We respond to comments concerning the metrics and thresholds used in our analysis in
elsewhere in a separate response to comment where we address comments on modeling.
Comment: [Luminant (0061) p. 69] According to Luminant, EPA's asserted justification for
requiring a source-specific analysis of potential visibility benefits of individual controls at a
small number of Texas units is fundamentally flawed. EPA claims that, in TCEQ's analysis,
"individual benefits were masked by the inclusion of those controls with little visibility benefit
that only served to increase the total cost figures."490 As noted, visibility benefit is not one of the
statutory factors that TCEQ was required to consider, as EPA concedes,491 and thus may not
form the basis of EPA's disapproval.492 Further, EPA's speculation is not supported by the
record. As its sole example of alleged "masking," EPA cites potential controls at Luminant's
Big Brown Plant, which it says could reduce SO2 emissions by more than 40,000 tons per year at
a cost of $1,500 per ton 493 EPA suggests, but does not assert, that reducing 40,000 tons per year
would have a meaningful impact on visibility at Texas's two Class I areas (the only two areas for
which Texas was required to perform a reasonable progress analysis) and that the benefit of
additional controls on Big Brown's units were masked by Texas's source category analysis. Yet,
EPA's own analysis demonstrates that the visibility impacts of Big Brown's SO2 emissions at
Big Bend and Guadalupe Mountains are negligible and in fact fall below EPA's own regulatory
"threshold" of 0.3% extinction.494 Thus, EPA's own analysis shows that the "masking effect"
that EPA cites as the justification for disapproving Texas's four factor analysis and RPGs for Big
Bend and Guadalupe Mountains (the only Class I areas for which Texas was required to conduct
a reasonable progress analysis) simply does not exist.
Footnotes:
490 79 Fed. Reg. at 74,838.
491 TX SIP TSD at 18 ("[Visibility is not an explicitly listed factor to consider when determining whether additional
controls are reasonable . . . .").
492 Luminant Generation, 675 F.3d at 925.
493 79 Fed. Reg. at 74,838.
494 FIP TSD at A-49 to A-52, tbl.A.4-4. Indeed, this is true for all of Luminant's units that EPA is proposing to
388
-------
regulate. Id.
Response: We disagree with the comment. As we discuss in a separate response to comment,
we do not believe that a source-specific analysis is required and our disapproval of the Texas
reasonable progress analysis is not based on a failure to complete a source-specific analysis. We
also discuss how we and Texas considered visibility benefits in the determination of reasonable
controls in a separate response to comment. We believe that the manner that Texas considered
visibility benefits and determined that no controls were reasonable was unreasonable. In a
separate response to comment we discuss the flaws in the Texas analysis and our reasons for
disapproving portions of the Texas regional haze SIP regarding reasonable progress and long-
term strategy.
In our proposal we explain that one of the flaws in the Texas analysis is how the four factors
were weighed. Regarding that we state:
The TCEQ constructed a large potential control set consisting of a mix of large
and small sources, located at various distances from Class I areas, with a large
geographical distribution. Because of the variation in size, type, and location of
these sources, the potential to impact visibility and potential benefit from controls
at a given Class I area can vary greatly between the identified sources. This
potential control set identified by the TCEQ included controls on some sources
that would likely result in significant visibility benefits, but also included controls
on many sources with much less anticipated visibility benefits.
Because the TCEQ only estimated the visibility benefit of all the controls together, it was not
able to assess the potential benefit of controlling those individual sources or groups of sources
with significant, and potentially cost-effective, visibility benefits. We provide the example of
Big Brown to illustrate that based on the available information from CENRAP and Texas, it was
clear that a subset of the sources identified by Texas deserved additional scrutiny to determine if
reasonable controls were available. However, this is not the "sole example" in which the
contribution of individual sources were masked.326 As we discuss in our proposal and elsewhere
in our response to comments, Texas and we agreed that it was reasonable to focus on impacts
from point sources for this planning period. The CENRAP source apportionment modeling
shows that visibility impairment from EGU sources in northeast Texas (this includes Big Brown)
are a large portion of the total visibility from point sources. At the Wichita Mountains, 46.7% of
the total visibility impairment due to all point sources in Texas is due to just EGUs in the
northeast portion of the state. This is 6.53% of the total visibility impairment at Wichita
Mountains. At Big Bend and Guadalupe Mountains, 43.6% and 31.0% of the total visibility
impairment from all Texas point sources are due to EGUs in the northeast portion of the state.
The control cost analysis presented in the Texas regional haze SIP shows that reductions of
42,105 tpy of SO2 from an EGU in northeast Texas (Big Brown) can be achieved at a cost of
approximately $l,500/ton, a cost much lower than the $2,700/ton threshold selected by Texas to
identify cost-effective controls. A reduction on 42,000 tpy of SO2 would be a reduction of 12%
of the total projected emissions from all EGUs. We also note that reductions of 35,000 tpy of
SO2 from another EGU in northeast Texas (Monticello units 1 and 2) can be achieved at a cost of
326 See Tables 6 and 7 of Appendix 10-1 of the Texas Regional Haze SIP for other examples.
389
-------
approximately $l,800/ton. Therefore, the available information demonstrates that cost-effective
controls for specific sources that would result in very large emission reductions (>20% of the
total EGU emission in the state) are available for the source type (EGUs) and region (northeast
Texas) that are shown to be responsible for a significant portion of the visibility impairment at
impacted class I areas. Our determination at this point within our review of the Texas analysis
was that reducing 40,000 tons per year at an EGU in northeast Texas may have a meaningful
impact on visibility conditions and that the potential benefit of additional controls on Big
Brown's units were masked by Texas's analysis that only estimated the benefit of all controls
within a large source set that included a wide range of source types, locations, and potential for
visibility benefits, and compared this total visibility benefit to the total costs of controls.
Our additional analysis reveals that impacts from the sources we identified as having the largest
visibility impacts, including Big Brown, are significant. Our technical record makes it equally
plain that the required controls reduce impacts from these sources and result in meaningful
visibility benefits towards the goal of natural visibility conditions. As discussed in the FIP TSD
and in response to comments elsewhere in this document, we evaluated visibility benefits at
Wichita Mountains and determined that the visibility improvements for controls at Big Brown,
Monticello and Coleto Creek were significant. We also concluded that scrubber installations on
Big Brown 1 and 2 would also yield significant benefits at the Guadalupe Mountains, and that a
scrubber installation on the Coleto Creek unit would also yield significant visibility benefits at
Big Bend.327 The visibility impairment from Texas point sources is significant, and as our
analysis shows, a significant portion of this impairment can be addressed by controlling a small
number of sources. For example, controls on just four units at Tolk and Big Brown are estimated
to reduce visibility impairment due to all Texas point sources at the Guadalupe Mountains by
approximately 13%. All required controls combined are estimated to reduce visibility
impairment at the Guadalupe Mountains from all Texas point sources by approximately 22%.
The identified facilities have significant impacts on visibility conditions.
We note that an RPG analysis, even though it is used to make a demonstration and establish the
RPGs for in-state class I areas, specifically, is part of the technical foundation for consultation
and other aspects of what is needed to meet LTS requirements. As such, the flawed RPG
analysis undermined both state's SIPs. Big Brown is subject to control under the FIP both
because those controls would have reasonably been required under a proper RPG analysis but
also because those controls would have been expected in an appropriately-informed consultation
setting. We note that, while flawed, the Texas analysis did consider visibility impacts and
potential benefits at the Wichita Mountains and other Class I areas in other states in evaluating
the sources for additional controls. In our FIP, we appropriately considered visibility impacts not
only at the Texas Class I areas, but also at the Wichita Mountains. We address comments
concerning the need to evaluate controls to address visibility impairment in downwind states in a
separate response to comment.
Furthermore, Luminant is incorrect in stating that the unit level contributions to extinction fall
below the 0.3% threshold we established to identify sources for additional control analysis. The
impacts of Big Brown units at recent emission levels (2008-2012) exceed 1% of the total
327 FIP TSD at page 28.
390
-------
visibility impairment at Wichita Mountains and are 0.326% at Big Bend. Impacts at Guadalupe
Mountains are 0.282%. We disagree with the commenter that units with impacts below this
threshold are "negligible." As discussed in a separate response to comment, we established this
threshold to identify those units with the largest visibility impacts for additional control analysis.
We also considered additional information to determine whether or not sources near this
threshold should be included in our analysis.
We address comments regarding our use of visibility in the legal section of this document where
we respond to comments concerning reasonable progress and long-term strategy.
Comment: [TCEQ/PUCT (0056) p. 3-4] The TCEQ stated that the projected visibility
improvement from the proposed FIP requirements are imperceptible at all three Class I areas.
The EPA's modeling analysis projects that the combined effect of all the proposed scrubber
upgrades (for seven individual units at four sites) will achieve at most only an imperceptible
improvement of 0.14 deciviews at Wichita Mountains. Even smaller improvements are projected
for Big Bend and Guadalupe Mountains, 0.03 and 0.04 deciviews, respectively. Tables 44 and
45 in the preamble exaggerate the potential benefits of the EPA's proposed FIP and are irrelevant
to the approvability of the 2009 RH SIP.
The TCEQ stated that both Table 44: CalculatedRPGs for 20% Worst Days ... and Table 45:
Anticipated Visibility Benefit... should be removed from the final action because they tabulate
calculated benefits that will not occur by 2018, the only year that is appropriate for evaluating
the visibility impacts of proposed controls. The 2018 visibility conditions that the 2009 RH SIP
will produce are the appropriate starting points for evaluating the effects of the EPA's proposed
FIP. Table 45 misleads a reader to believe that the EPA's proposed FIP action would produce a
0.62 deciview improvement in visibility at Wichita Mountains. Instead of calculating a benefit
from the air quality that the 2009 RH SIP would produce in 2018, Table 45 misleads the reader
by calculating "benefits" from 2011 through 2013 emissions, long before the 2009 RH SIP is
fully effective instead of from 2018.
The TCEQ stated that Table 43 in the Preamble presents the calculated benefits in 2018 that
could result from the EPA's proposed FIP. However, the potential 0.14 deciview improvement at
Wichita Mountains is almost certainly an overstatement of the incremental benefit from the
proposed FIP in 2018 because SO, emission reductions are occurring due to other requirements,
and the actual SO, emissions will likely be lower than those in the CENRAP 2018 emissions
projections.
The TCEQ explained that typically, a person can perceive a one (1.0) deciview change in
visibility impairment. Visibility differences of 0.14, 0.04, and 0.03 deciview are imperceptible.
391
-------
Tabic 1. Visibility Data
(in Deciviews)-
Big Bend
Guadalupe Mountains
Wifhila Mountains
Baseline Visibility
Impairment
17.30
17.19
23.81
2000 11004
State-oslablisliixi IPG
for 2018
16.60
16.30
21.47
lncremi'Lital 2018
Improvement from
EPA's Proposed MP
0.03
0.04
0,14
Scrublxn* Upgrades
EPA-proposeil KPCs
for 2018
16,57
16.26
21,33
Currcnf Visibility
1.6.30
15.30
21.20
20OC) - 2013
Also, the TCEQ stated that the potential improvement from the proposed FIP is 2% or less of the
total impairment projected to exist in 2018 on the most impaired 20% days and even that is likely
an overestimate of the FIP's potential benefit because the EPA's analysis does not consider the
reductions that will occur from other federal programs, such as the Mercury and Air Toxics
Standards (MATS) rule and the implementation of the sulfur dioxide (SO2) National Ambient
Air Quality Standard (NAAQS).
The TCEQ stated that the actual effects of the EPA's proposed FIP are correctly represented in
Table 43, which includes the only controls that could be in place by the end of 2018, which is the
end of the first regional haze planning period established by the RUR.
The TCEQ asserted that the EPA does not have an appropriate basis for adopting the proposed
FIP given that current monitored visibility is better than the EPA calculates the proposed FIP
would achieve in 2018 and that the potential visibility improvements from the proposed FIP are
both small and uncertain.
Commenter's Reference:
3 From Table 43, (79 FR84887), and the Western Regional Air Partnership-Technical Support System (WRAP-TSS)
Response: As discussed in depth in a separate response to comment, the Regional Haze Rule
does not require that controls result in perceptible visibility improvement. As we address in
response to comments above, the potential visibility improvements from the FIP are neither
small nor uncertain. Instead, the controls will assure large reductions of haze-causing pollutants
at those sources identified as having the largest visibility impacts and result in measureable and
important visibility improvements from a State having a large number of sources and ample
opportunities for making progress by improving controls at lesser-controlled sources. For
example, all required controls are estimated to reduce visibility impairment by greater than 5% at
Wichita Mountains. We address the TCEQ's comments concerning Tables 44 and 45,
consideration of recent actual emissions, and anticipated future emission reductions due to other
requirements in separate responses to comments where we address comments on modeling. We
392
-------
disagree with the comment that Tables 43, 44, and 45 in the proposal exaggerated, or otherwise
misled readers; these tables present available modeling data and related information, using
appropriate available inputs. There are potentially different methods of displaying the benefits
that would be obtained, but Tables 43, 44, and 45, were useful for our study of the issues and
were provided as part of the information made available for public comment. This information
should be examined with a correct understanding of their contributing data set and context.
As discussed elsewhere where we address comments concerning controls that are anticipated to
implemented beyond 2018, we disagree with commenters who consider it inappropriate for
controls to be required after the planning period because corresponding visibility benefits may
not be realized during the planning period. The fact that benefits of such controls may not be
realized within the first planning period does not affect our determination that the controls are
necessary nor deprive us of our authority to impose the requirements. We cannot agree that 2018
is the only year that is the appropriate year for evaluating the visibility impacts of proposed
controls. The comment appears to suggest that the effects of the FIP can only be correctly
represented by showing controls that could be in place by the end of 2018. This is flatly wrong.
It suggests that we "represent" our action in a way that ignores the thoroughly studied benefits
that are achievable through the implementation of FIP controls. Our analysis, including the
analysis of the visibility benefits of scrubber retrofits, showed that cost-effective controls were in
fact available that resulted in significant visibility benefits and provided for meaningful
improvements in visibility conditions at the Class I areas of interest. Because these retrofits are
not anticipated to occur before 2018, we evaluated the visibility benefit from them separately,
however this does not impact the fact that they are reasonable controls.
We address comments concerning recent monitoring data elsewhere in this response to
comments document.
Comment: [TCEQ/PUCT (0056) p. 11] The TCEQ disagreed with the EPA's approach of
requiring emissions reductions at certain sites, not necessarily because the reduction had any
perceptible improvement in visibility at a Class I area, but because emissions from that source
may be significant when compared to other sources.
The TCEQ noted that Texas analyzed emissions reductions using four factor analysis, as required
by the EPA's RHR (64 FR 35766). Emissions reductions were estimated for sources with the
potential suite of controls selected using a $2,700 per ton threshold. A four factor analysis was
performed on this group of sources; no perceptible visibility improvement was determined. The
goal of the regional haze program is to focus on reasonable progress towards visibility
improvement at each Class I area, not to target reductions at specific sources. The EPA appears
to have performed its control analysis in the proposed FIP in a reverse-logic form. It targeted
reductions at larger-emitting sources, only because they are larger emitting, not through an
application of the reasonable progress four factor analysis on potential controls when considering
perceptible progress towards achieving natural visibility.
Response: The TCEQ appears to have concluded that perceptibility of visibility benefit is a
necessary criterion for control. Please see our response to other comments in which we disagree
393
-------
with this premise. A goal of the regional haze program is that the state's SIP for each planning
period make reasonable progress toward the national goal of a return to natural visibility
conditions in 2064. This requires that large sources of visibility impairing pollution which have
available cost effective controls that result in significant visibility improvement—like the ones
we proposed to control—are in fact controlled. We disagree with the TCEQ that we targeted
larger emitting sources because they are larger emitting. We identified those sources with the
greatest potential to impact visibility conditions at Class I areas. We evaluated all point sources
considering the distance those sources were from Class I areas and the level of emissions to
identify those sources close enough to a Class I area or with large enough emissions to
potentially impact visibility at a Class I area, just as Texas did in their analysis. We did that in
order to identify those sources with the greatest potential to impact visibility from the thousands
of potential sources in Texas. We then performed visibility modeling on that smaller set, and
performed a four factor analysis on the subset of sources that exhibited the largest visibility
impacts. We believe our proposed approach of identifying the largest impacting sources for
additional control analysis is logical and consistent with the CAA and the Regional Haze Rule.
Furthermore, as we discuss elsewhere, the Texas analysis that is flawed because it includes
emission reductions at certain sites that produce very little visibility benefit. Inclusion of these
controls serve only to increase the total cost of controls in the Texas analysis. We discuss our
methodology to identify those sources with the largest visibility impacts in depth in the FIP TSD
and in response to comments on modeling.
Comment: [Texas Governor (0066) p. 1] The Texas Governor explained that the EPA's
proposed decision to partially disapprove Texas's state implementation plan ("SIP") and to
promulgate a federal implementation plan ("FIP") would do nothing to improve visibility in Big
Bend or the Guadalupe Mountains. Moreover, EPA's proposed actions would impose more than
$2 billion in compliance costs on Texans. Whatever EPA's motivation, the results of the
"regional-haze" rule are absurd, arbitrary, capricious, and contrary to law.
[Texas Governor (0066) p. 1-2] The Texas Governor explained that cost alone renders the FIP
unlawful. EPA has a statutory obligation to "take [] into consideration the costs of compliance."
42 U.S.C. section 7491(g)(1). Yet EPA's FIP makes no mention of how much its additional
controls will cost. EPA staff have confirmed that those controls will cost at least $2 billion - all
for reductions in haziness that are l/8th the magnitude that would be visible to the naked eye.
EPA cannot comply with Section 7491(g)(1) by asking its staff to make informal, back-of-the-
envelope guesstimates. Nor can it comply with the statute by dictating such unreasonably large
expenditures for invisibly small benefits.
Response: We disagree with the Governor of Texas that our proposal would do nothing to
improve the visibility of the Texas Class I Areas. Our objective is to improve the visibility at
Texas and other Class I areas, and this accords with the Congressional purpose embodied in the
CAA and the Regional Haze Rule. In finalizing our action, we necessarily disagree that our
action is arbitrary, capricious, and contrary to law. We further disagree that we did not mention
how much our proposed controls will cost. The costs for our proposed scrubber retrofits are
listed in our Cost TSD and are summarized in our proposal. As we explained in our Cost TSD
and our proposal, we could not list the costs for our proposed scrubber upgrades because they
394
-------
were based on information claimed by the respective companies as being Confidential Business
Information. As discussed in depth in a separate response to comment, we disagree with the
commenter that the Regional Haze Rule requires that controls result in perceptible visibility
improvement
Comment: The proposed FIP is arbitrary, capricious, and an abuse of discretion because,
based on EPA's analyses, it would have no perceptible or meaningful effect on visibility
conditions in any Class I area. p. 28-30]
UARG argued that for EPA's proposed rule to justify the enormous costs that compliance with
its scrubber upgrade and retrofit requirements would impose, the rule must produce a significant
improvement in visibility conditions in the Class I areas that are the subject of the regional haze
program. The proposed rule demonstrates, however, that its visibility benefits (if any) will be
negligible. Those vanishingly small improvements do not support the emission control
requirements that EPA seeks to impose at great cost.
UARG stated that the EPA's source apportionment modeling results demonstrate the minimal
effects of emissions from the units EPA assesses on visibility in Big Bend, Guadalupe, and
Wichita Mountains. See 79 Fed. Reg. at 74,839. The highest percentage contribution to total
visibility impairment for the 20 percent worst days at Big Bend is the Sommers Deely Spruce
facility (also known as the Calaveras Plant) at 0.57 percent. Id. at 74,839, Table 12. That
facility would not be subject to control requirements under the proposed rule, but the facility
with the second\ highest percentage contribution to the 20 percent worst days at Big Bend (i.e.,
Coleto Creek) would be subject to control requirements under the proposed rule even though its
contribution, as modeled by EPA, is less than one-half of one percent - a mere 0.44 percent. Id.
And the highest percentage contribution to total visibility impairment for the 20 percent worst
days at Guadalupe is Tolk Station at 0.65 percent. Id. at 74,839, Table 13. That these minuscule
contributions should be targeted by EPA for regulation is irrational on its face.
UARG noted that this conclusion is also supported by EPA's calculation, in deciviews, of the
visibility improvement that would result from installation or upgrading of scrubbers at the
various facilities it has chosen to assess. EPA proposes to find that scrubber installations are
appropriate for Big Brown units 1 and 2, Coleto Creek, Tolk units 171B and 172B, and
Monticello units 1 and 2. 79 Fed. Reg. at 74,881-82 & Tables 34-36. Installation of scrubbers
at these units does not result in perceptible visibility improvement at any Class I area. On a unit-
by-unit basis, estimated visibility improvements are less than 0.5 deciview, EPA's criterion for
contribution to visibility impairment. See id. Even on a facility-wide basis - and even when
measured against natural conditions - the vast majority of the visibility improvements that EPA
projects come nowhere near EPA's 1.0-deciview threshold of perceptibility. See id. The same is
true for the visibility effects of scrubber upgrades that EPA has modeled, which fall far below
the 0.5- deciview contribution benchmark with respect to every unit evaluated. Id. at 74,883 &
Table 37. EPA's proposed conclusion that these visibility improvements justify new emission
controls is unreasonable.
Given the negligible visibility benefit that EPA projects to result from the proposed rule's
395
-------
requirements, and the evidence showing that those requirements will not - and are not needed to
- achieve visibility benefits at any Class I area, the huge costs that this rule would impose cannot
be justified. EPA should, therefore, withdraw the proposed rule and should instead approve the
Texas and Oklahoma regional haze SIP provisions in full.
Response: As discussed in depth in a separate response to comment, we disagree with UARG
that the Regional Haze Rule requires that controls result in perceptible visibility improvement.
We also discuss in a separate response to comment that the use of a 0.5 dv threshold, a threshold
applied to BART single-source visibility impact analysis using CALPUFF modeling, is an
inappropriate comparison or threshold to apply to visibility improvement estimated by CAMx
modeling due to differences in models, model inputs, and metrics used.
We disagree that a contribution to total visibility impairment of 0.44% or 0.65% on the 20%
worst days from a single source is minimal. As discussed elsewhere in this document, we
identified a threshold of 0.3% contribution to total visibility impairment on a unit-level basis to
identify the sources with the largest visibility impacts. Visibility impairment from the single unit
at Coleto Creek accounts for approximately 2% of the total impact from all Texas sources and
over 6% of the total visibility impact from the over 1,600 Texas point sources at Big Bend. Of
the group of sources identified by Texas and us to evaluate for controls due to their impacts,
point sources of SO2 and NOx, controlling one unit at Coleto Creek addresses 6% of this
visibility impact. Controls on Big Brown's two units addresses another 6%. The recommended
controls address 23.4% of impact from all Texas point sources at Big Bend (based on 2018
CENRAP projected emission levels). We agree that the Sommers-Deely-Spruce facility is
modeled to have larger visibility impacts at Big Bend. However, the two units responsible for
roughly 80% of these visibility impacts are scheduled to shutdown in 2018. Emissions for the 2
units at Tolk account for nearly 8% of the total Texas point source impact at Guadalupe
Mountains. The recommended controls 25.74%) of impact from all Texas point sources at
Guadalupe Mountains (based on 2018 CENRAP projected emission levels).
Coleto Creek's impact in initial modeling is 0.216 Mm"1 on BIBE and Tolk is 0.302 Mm"1 on
Guadalupe Mountains. We compared this to the threshold Texas used of 0.5 Mm"1 to determine
which states to invite for consultation due to their impacts on Big Bend and Guadalupe
Mountains. In other words, if all emission sources from all source categories within a
neighboring state impacted the Class I areas by more than 0.5 Mm-1, that state was invited to
consult in order to discuss if reasonable controls were available to reduce this impact. The
impacts from these individual sources in Texas, approach that threshold level, and there is an
obligation to address the impacts with sensible controls. As we discuss in detail where we
address modeling comments, we identified those sources with the largest visibility impacts to
evaluate for additional controls. As these facts demonstrate, the identified facilities have
significant impacts on visibility conditions. Our technical record makes it equally plain that the
required controls reduce impacts from these sources and result in meaningful visibility benefits
towards the goal of natural visibility conditions.
Comment: Costs of Compliance and Visibility Improvements [GCLC (0063) p. 14-15]
396
-------
GCLC stated, the costs of compliance, especially for units required to retrofit with new
scrubbers, are astronomical and could likely force the shutdown of the impacted EGUs. BART
standards, which EPA uses as guidance, were never intended to be a vehicle to shut down an
existing source with decades of remaining useful life. A technology is not a cost-compliant
"available retrofit" if it forces the shutdown of the unit.
According to GCLC, regarding scrubber upgrades, EPA believes that "any reasonable amount of
visibility improvement due to their installation justifies their cost."66 In other words, EPA does
not even consider cost-effectiveness as a factor in its analysis. But this is not the purpose or
intent of regional haze rules, which clearly require a more measured approach to assessing cost
impacts. As demonstrated by EPA's own data, the net effect on visibility is virtually
imperceptible statewide - there is no "reasonable amount of visibility improvement." This is
compared to costs that total in the aggregate of over $2 billion. 67
GCLC suggested, considering the "degree of improvement in visibility," compared to the costs
of compliance, EPA should avoid imposing the limitations contemplated in the FIP. The
simplest cost-benefit analysis would demonstrate this. However, one of the key reasons EPA
failed to appropriately consider the costs of compliance is because, apparently, EPA never
actually attempted to balance or compare the costs of the rule with the alleged visibility impacts.
EPA's analysis was merely superficial, reviewing costs in one TSD (i.e. the Cost TSD) and
visibility improvements (i.e., the FIP TSD) without making any real comparison. If EPA had, it
would realize that there is no way EPA could claim that the controls are cost effective.
Footnotes:
66 Proposed FIP, 79 Fed. Reg. at 74884.
67 TCEQ, TCEQ Response to EPA's Proposed Action on Regional Haze Plan, Nov. 24, 2014. Available at:
http://www.tceq.state.tx.us/news/releases/ll-14-eparegionhaze. (Last modified: Nov. 24, 2014).
Response: We have not seen any credible evidence thus far from any commenter, including the
affected facilities, that our proposal would force any unit to shut down. However, faced with the
expense of installing controls and other business considerations, facilities could conceivably
decide to retire units instead of installing the controls. Such a situation does not, in and of itself,
conflict with the CAA or our Regional Haze Rule. Our proposed controls were evaluated
according to the criteria in Section 308(d)(l)(i)(A) which charges us to "consider the costs of
compliance, the time necessary for compliance, the energy and non-air quality environmental
impacts of compliance, and the remaining useful life of any potentially affected sources, and
include a demonstration showing how these factors were taken into consideration in selecting the
goal." We did exactly that. We concluded that cost effective controls were available for 14
uncontrolled or under controlled units, and that these controls would result in significant
visibility benefits at Class I Areas. These same controls have been installed at dozens of similar
coal fired power plant units across the United States. We disagree with GCLC that we did not
balance the cost of the controls against their visibility benefits. Our proposal summarizes the
information present in our Cost and FIP TSDs and contrasts the costs to their modeled visibility
benefits.328 As discussed in depth in a separate response to comment, we also disagree with the
commenter that the Regional Haze Rule requires that controls result in perceptible visibility
328 See our proposal beginning on page 74881.
397
-------
improvement.
As to the comment that the actual cost-per-ton for scrubber upgrades was ultimately irrelevant to
our decision, we disagree with the comment and address it fully in a separate response to
comment.
Comment: The Visibility Impacts Do Not Justify the Proposed Control Levels and the
Proposal Will Require Significant Costs to Consumers. [EEI (0076) p. 8]
EEI noted that EPA's own analysis in the proposed rule shows that the units EPA has identified
as required to install additional controls or upgrade existing controls as part of the FIP have a
very minimal impact of far less than one deciview on the target areas' visibility See 79 Fed. Reg.
at 74,887, Table 45.
[EEI (0076) p. 9] EEI stated that the Agency's proposed FIP- per EPA's own estimate-will
require the investment of approximately $2 billion dollars in retrofits and upgrades, resulting in
increased consumer costs for Texas energy consumers. See 79 Fed. Reg. 74,876-77. The
retrofit cost estimates EPA used for Southwestern Public Service's Tolk facility are inaccurately
low and not based on Tolk's site-specific circumstances. Specifically, the proposal does not
reflect the significant capital cost of water upgrades that would be necessary to operate proposed
dry scrubbers at this location.6 EPA's proposed FIP will impose far greater costs on utilities and
customers than estimated in the Agency's proposal.
Footnotes:
6 Comments submitted by Luminant explain the many ways in which EPA has understated the costs, and overstated
the benefits, of the controls that EPA proposes for Luminant units.
Response: We disagree with EEI that our proposal will not result in significant visibility
benefits. As discussed in depth in a separate response to comment, we disagree with the
commenter that the Regional Haze Rule requires that controls result in perceptible visibility
improvement. We believe the visibility benefits we have proposed are necessary in order for
Texas and other Class I areas to make reasonable progress toward the national goal of a return to
natural visibility by 2064. This goal requires that states make steady progress with each SIP
submittal. We address our cost analysis for Tolk in another response.
Comment: [Xcel Energy (0064) p. 6, 7] Xcel Energy expressed multiple concerns with the FIP
that EPA has proposed for Texas to address regional haze ("Proposal"). Fundamentally, the
Proposal partially rejecting the Texas SIP is contrary to the clear allocation of authorities and
obligations under the CAA, and is arbitrary and capricious. The Proposal will produce no
visibility benefits beyond what Texas provided in its SIP. Further, because of EPA's refusal to
acknowledge and address international emission sources, EPA's FIP will never result in natural
background visibility levels, as required by the CAA. Even if every stationary source in Texas
were to cease operating tomorrow, Texas' Class I areas would be, and remain, above the uniform
rate of progress under either the Texas SIP or EPA's Proposal.
398
-------
Xcel Energy requested that EPA adopt a final rule that approves the Texas Regional Haze SIP's
reasonable progress analysis and conclusions, or at least concludes that installation of dry
scrubbers at Tolk should not be required in light of the extremely low modeled deciview benefit
and the high cost of installing entirely new scrubbers.
Response: Please see our more detailed responses to these general comments in other responses.
In general, we disagree with Xcel that our proposal is contrary to the clear allocation of
authorities and obligations under the CAA, and is arbitrary and capricious. We also disagree that
our proposal will not produce visibility benefits beyond what was proposed in the Texas SIP.
We do agree that international emissions are a significant source of visibility impairment in
Texas. However our proposal does not in any way expect Texas to make up for those emissions.
Our action requires the control of particular Texas sources that due to their own emissions were
impacting the visibility at Texas and other state Class I areas. For the reasons we have outlined
in our proposal, we cannot fully approve the Texas regional haze SIP. Please see our responses
to other specific comments regarding the control of the Tolk power plant.
Comment: The costs of compliance of EPA's proposal are unreasonable [AECT (0074) p. 6-
7]
AECT stated that the CAA, EPA's Regional Haze rules, and EPA's guidance support evaluation
of the reasonable progress costs of compliance factor by considering the total cost of the
additional emissions controls at the identified EGUs relative to the visibility improvement that
modeling predicts would result from such additional controls. Evaluating the cost of compliance
of EPA's Proposal in that manner shows that its cost of compliance is clearly unreasonable, and,
thus, EPA's Proposal is not supportable.
According to AECT, EPA calculates that its Proposal would require just four Texas companies
to spend approximately $1.8 billion for additional SO2 emissions controls for their EGUs.19
AECT, however, believes that the costs those companies would have to spend for such
additional SO2 emissions controls would be greater than $2 billion. In addition, EPA's projected
total visibility improvements that would result by 2018 in each of the three Class I areas due to
the addition of such SO2 emissions controls are only 0.03 dv for Big Bend, 0.04 dv for
Guadalupe Mountains, and 0.14 dv for Wichita Mountains.20 Each of those projected total
visibility improvements is much less than 1.0 dv, which is the degree of visibility improvement
that can be detected by the eyes of most humans.21
AECT noted, based on the foregoing, regardless of whether the total cost for additional SO2
emissions controls for the identified EGUs is $1.8 billion or greater than $2 billion, the cost of
compliance of EPA's Proposal is clearly unreasonable, and, thus, EPA's Proposal is not
supportable.
Footnotes:
19 See EPA, Technical Support Document for the Cost of Controls Calculations for the Texas Regional Haze Federal
Implementation Plan at 24-25 (Nov. 2014) (scrubber retrofit capital costs); Id. at 55 (scrubber upgrade costs)
20 79 Fed. Reg. at 74887, Table 43
21 77 Fed. Reg. at 30250
22 77 Fed. Reg. at 30464
399
-------
Response: We have adjusted these costs slightly in response to comments we have received.
However, for the reasons we have discussed in response to more specific comments, we disagree
that our proposed costs of compliance are unreasonable. As discussed in depth in a separate
response to comment, we disagree with the commenter that the Regional Haze Rule requires that
controls result in perceptible visibility improvement. As presented in our proposed action, FIP
TSD and in separate responses to comments, we believe these controls result in significant
visibility benefits and provide for progress towards meeting the goal of natural visibility
conditions at the impacted Class I areas.
We note that AECT provides incorrect estimates of cost with respect to visibility. The
commenter presents estimated visibility benefits of only the scrubber upgrades but then contrasts
them to the total cost of scrubber upgrades and scrubber retrofits combined. The scrubber
upgrades are only a fraction of the total cost (less than 20% of the total capital costs) and the
visibility benefits due to the scrubber upgrades are only a fraction of the total anticipated
visibility benefit from all required controls. Furthermore, as we discuss in the FIP TSD and in
separate response to comments, we believe it is necessary to consider visibility benefits based on
"clean" natural background conditions to assess the full potential for visibility benefits from
controls.329 We estimated that the required controls provide for over 3 dv improvement at
Wichita Mountains when estimated using a "clean" background and result in improving
projected visibility conditions in 2018 by 0.45 dv over the visibility conditions projected by
CENRAP and Texas and an estimated 0.62 dv improvement in visibility in 2018 when
considering recent actual emissions. The required controls result in a greater than 5%
improvement in overall visibility conditions at Wichita Mountains on the 20% worst days. We
also estimate that the required controls are a significant acceleration of the final goal, reducing
the number of years to meet natural visibility by 25 to 30 years at Big Bend and Guadalupe
Mountain.
Comment: EPA's proposed FIP does no better at achieving the URP and visibility
improvement than Texas' SIP yet imposes billions of dollars in costs.
[Xcel Energy (0064) p. 21]
Xcel Energy suggested that it would be one thing for EPA to reject the Texas SIP for failing to
meet the RPG if EPA has a plan that would lead to natural visibility more quickly. But EPA's
plan does no better than the Texas SIP. Under the Texas SIP, the visibility impairment is
projected to be at 16.36 dv on the 20% worst days at Guadalupe Mountains in 2018. See Figure
1 above. EPA's FIP purports to lower this number to 16.21 dv; a not perceptible 0.09 dv
difference. !d. Both Texas and EPA would leave Guadalupe Mountains significantly above the
URP.
Xcel Energy argued that EPA's FIP also does nothing to accelerate the final goal of natural
visibility levels. EPA's FIP purports to reach natural visibility levels at Guadalupe Mountains
only after 141 years and, at Big Bend, only after 173 years. 79 Fed. Reg. at 74,887, Table 44. It
is absurd and clearly arbitrary for EPA to reject the Texas SIP for setting inadequate RPGs and
329 FIP TSD at A-37
400
-------
then impose on the state and its sources costly control requirements that, at best, theoretically
achieve the statutory goal 141 to 173 years in the future. In fact, unless EPA addresses the
emissions from Mexico, EPA's theoretical, long-term improvements are not possible. The entire
state of Texas could shut down and move away and visibility in these Class I areas would not
improve.
Response: Our FIP addresses the emissions from uncontrolled and under controlled sources in
Texas that are impacting Class I areas. The cost effective control of these emissions does not
demand shutdowns in Texas or relocation from Texas. Having determined that controls would
be cost effective, the FIP controls are of a type and magnitude that are demonstrably critical for
making reasonable progress. We agree that impacts from Mexico also contribute to the haze
problem in Texas Class I areas. Nothing in the regional haze rule or our FIP is calculated to hold
Texas accountable for emissions from Mexico. We agree those emissions need to be addressed
to achieve natural visibility, but our agreement on this point does not in any way relieve Texas of
the obligation to make reasonable progress, including through controls, and particularly through
the emissions addressed with controls through our FIP.
Comment: EPA's Proposal to Require Installation of Scrubbers at Tolk Would Produce
Miniscule Visibility Benefits. The predicted improvement would produce no discernible
visibility benefit. [Xcel Energy (0064) p. 6-7, 28; 0053-24 and 0054-4]
[Xcel Energy (0064) p. 6-7] Xcel Energy referred to the proposed SO2 scrubbers for Tolk
(approximately $400 million in capital), and stated that the visibility benefits from the proposed
scrubbers at Tolk are miniscule at best while EPA is imposing these costs immediately on Xcel
Energy and its customers. For instance, at Guadalupe Mountains, the required dry scrubber on
Tolk 1 would produce a modeled visibility benefit of approximately 0.022 deciviews for the 20%
worst days, roughly 1/50 the level for human perception. At the same time, international sources
of visibility impairment, which vastly exceed the contributions of Tolk, and indeed all Texas
point sources in combination, are left untouched such that there will be no discernable visibility
benefit resulting from the entire FIP.
[Xcel Energy (0064) p. 28] Xcel Energy noted, in the FIP TSD, EPA has estimated a total
deciview improvement of 0.763 over the average natural conditions clean background for the
Guadalupe Mountains National Park based on CAMx modeling. FIP TSD, at A-76, Table A.6-5
- Net benefit of Proposed Controls on 2018 Visibility Projections. This estimated deciview
improvement is the cumulative improvement that would result from EPA's proposed controls,
consisting of scrubber upgrades on seven units and scrubber retrofits for seven units. This
analysis is the main reason the Tolk units are proposed to be controlled in the proposed FIP.
EPA is proposing to require facilities in Texas to spend billions of dollars through the addition
and/or upgrade of control devices when the collective visibility improvement from such facilities
will not be perceptible.
Xcel Energy stated that the deciview scale was selected by EPA as the measure of visibility
improvement in the RHR specifically because by definition each deciview reflects "perceptible
changes" in visibility. See Proposed RHR 62 Fed. Reg. 41,138, 41,145 (July 31, 1997) ("A one
401
-------
deciview change in haziness is a small but noticeable change in haziness under most
circumstances when viewing scenes in mandatory Class I Federal areas."). Accordingly, by
EPA's own standard, a total deciview improvement at the Guadalupe Mountains of 0.763 dv
from the installation of controls at the selected Texas facilities would not be perceptible to the
human eye. Further, studies have since demonstrated that not only is the deciview scale not
uniform in perception over a wide range of visibility conditions, but a 1-deciview change in
visibility is not even perceptible to the human eye for observation. See Appendix A to comment
0064, "Just-Noticeable Differences in Atmospheric Haze," Ronald C. Henry, Department of
Civil and Environmental Engineering, University of Southern California, Los Angeles, Air &
Waste Manage. Assoc. (2002).
In addition, Xcel Energy stated that observations and air modeling for the haziest days suggest
that there is not a clear downward or upward trend in regional haze for the Guadalupe
Mountains, Big Bend, or the Wichita Mountains Class I areas. See Appendix B to comment
0064. For the Guadalupe Mountains, the clearest days show a downward trend of -0.09 dv/yr
(reduction). Overall, there is variation of ± 2 to 3 deciviews over the last 21 years. Therefore,
EPA's estimated deciview improvement for average natural conditions at Guadalupe Mountains
associated with the addition of dry scrubbers on the Tolk units (0.182), while clearly
imperceptible, also is insignificant compared to the natural variation in background regional haze
over the last 20 years. See 79 Fed. Reg. at 74,882, Table 36.
Response: We disagree with the commenter that the visibility benefits from emission reductions
at Tolk are miniscule. The commenter states that visibility benefit from controls on Tolk unit 1
are 0.022 dv for the 20% worst days at Guadalupe Mountains, but this value is based on a "dirty"
2018 background. As we discuss in detail elsewhere, a "dirty" background approach
underestimates the visibility benefits of controls and a "clean" background approach is necessary
to fully assess the potential benefit from controls. Based on consideration of a "clean"
background, we estimated the benefit at Guadalupe Mountains from SDA controls on Tolk's
units 1 and 2 to be 0.087 dv and 0.095 dv, respectively on the 20% worst days. The visibility
benefit from controlling these two units is almost equal to the anticipated benefit of controls on
Big Brown units 1 and 2 of 0.105 dv each at Guadalupe Mountains. Required controls at Tolk
and Big Brown address the largest individual visibility impacts from Texas point sources at
Guadalupe Mountains. Controls on Tolk alone address approximately 8% of the total impact
from Texas point sources at Guadalupe Mountains. Based on our evaluation of the four factors,
including comparison of cost to the anticipated visibility benefit, we determined that these
controls are reasonable and provide for significant improvement towards the goal of natural
visibility conditions. We also note that the estimated visibility impact330 from the two Tolk units
based on recent actual emissions is estimated to be 0.281 Mm"1, compared to a threshold of 0.5
Mm"1 applied to all emission sources in a state selected by Texas to determine which states
would be invited to consult due to their impacts on Texas Class I areas.
We agree with the commenter that annual monitored visibility conditions can vary and can be
impacted by events such as dust storms and wildfires that vary in frequency and intensity from
year to year. Therefore, we focus on a 5-yr average in assessing current or baseline visibility
330 Based on 2009-2013 annual emission levels See "No Control" worksheet in Vis modeling summary.xlsx
available in the docket for this action.
402
-------
conditions. We disagree with the commenter's apparent assertion that variability in monitored
visibility conditions somehow relieves the state from the requirements of implementing
reasonable controls towards the goal of natural visibility conditions.
We address comments concerning no perceptible visibility benefits and visibility impairment
from international emissions in separate responses to comments.
Comment: The highest cost controls are proposed for Tolk for the least amount of visibility
improvement. [Xcel Energy (0064) p. 34]
Xcel Energy stated that even using EPA's costs estimates for installation of dry scrubbers at
Tolk, the Proposal would impose on Tolk the highest cost of the scrubber retrofits contemplated
for the lowest visibility benefit modeled for any unit for which EPA proposes controls. The Tolk
units were modeled at barely over EPA's declared 0.3% threshold for light extinction, and yet
would have the most expensive scrubber installation. Other units for which EPA proposes
controls under a 0.5% modeled light extinction impact are upgrades to existing scrubbers, which
are many times less expensive than building entirely new scrubbers. In addition, EPA did not
propose controls for units which modeled barely under EPA's chosen screening level, even
though much less costly scrubber upgrades were an available option. These points are illustrated
in the chart in Appendix E.
Given the extremely low modeled deciview impact and the high cost of installing entirely new
scrubbers, Xcel Energy stated that the EPA's proposal to require scrubbers on Tolk cannot be
considered "reasonable" and will not result in any perceptible "progress." The emission
reductions achieved would be simply overwhelmed by international contributions and natural
dust, and diluted by Tolk's substantial distance (over 330 km) well northeast of the Guadalupe
Mountains National Park. Additionally, as discussed above, given that EPA underestimated the
costs of dry scrubbers for Tolk, the Proposal is even more unreasonable than the chart in
Appendix E to comment 0064 demonstrates.
Response: Please see our responses to other comments concerning elements of our analysis of
Tolk's scrubber cost. We believe that even considering the revisions we have made to Tolk's
scrubber cost analysis, the significant visibility benefits justify the cost of these controls. See our
response to other comments where we disagree that perceptibility is a threshold requirement of
visibility improvement and address comments concerning international emissions.
Comment: [CCP (0075) p. 2] The Proposed Rule provisions that specifically target Coleto
Creek Unit 1 are even more unwarranted. Coleto Creek Unit 1 is over 550 kilometers from the
nearest Class I area. The actual impact of the unit's emissions on haze in those areas, considering
the well-known bias of EPA models to over-predict impacts at such large distances, is negligible
and the controls required under the proposed FIP will not result in any improvement in visibility
by the end of the current planning period in 2018 nor any appreciable visibility improvement
thereafter. Still, the EPA proposes approximately $300 million in controls at Coleto Creek Unit
1 based on flawed models and generic assumptions that fail to consider the legally required unit-
403
-------
specific factors and time for compliance. Time and time again EPA has declined to impose
controls under its regional haze authorities on units that have actual impacts on Class I areas less
than those of Coleto Creek Unit 1.
Response: Please see our responses to other comments concerning elements of our analysis of
Coleto Creek's scrubber cost, and estimated visibility improvement. CCP does not specify here
the actions for which it alleges inconsistency. However, as we discuss in our proposal and
elsewhere in our response to comments, we disagree that our action is inconsistent with previous
determinations made under regional haze. Furthermore, there is no way to directly compare the
CAMx modeling we used in our proposed TX/OK FIPs with previous CALPUFF modeling
results due to differences in models, model inputs, and metrics. Therefore, comparisons to
previous actions that depended on CALPUFF modeling is not valid. We note that Coleto Creek
did provide CALPUFF modeling results as part of their comments. We address this modeling
analysis in a separate response to comment.
Comment: Big Bend, Guadalupe Mountains, and Other Impacted Class I Areas.
[Earthjustice (0067) p. 11]
Earthjustice et al., stated that in finalizing the proposed FIP for Texas and Oklahoma, EPA has a
unique opportunity—and an obligation—to protect air quality by reducing emissions from
Texas's numerous facilities which are responsible for visibility impacts and other air quality
degradation in Class I areas in Texas, Oklahoma, Arkansas, New Mexico, Louisiana, Colorado,
and Missouri.6 Emissions from Texas sources impact two in-state Class I areas: Big Bend
National Park and Guadalupe Mountains National Park. Both national parks are located in west
Texas and contain spectacular scenic views that draw visitors from across the United States and
around the world. "Big Bend National Park is known for its scenic beauty, which ranges from
stark seemingly barren wastelands to majestic forested mountains to gigantic canyons."7
Guadalupe Mountains National Park is an "internationally significant" park, in part, because of
its "[sjpectacular scenery," which is a "major attraction for visitors."8 Guadalupe Mountains is
home to the highest summit in Texas, as well as the culturally significant peak, El Capitan,
which has been used as signal peak by travelers and settlers for hundreds—perhaps thousands—
of years 9
Earthjustice et al., stated that air pollution from Texas sources mars the unique scenic views at
both Big Bend and Guadalupe Mountains. The National Park Service has acknowledged that
"[t]he scenic beauty of Big Bend National Park is often spoiled by haze that obscures its many
vistas."10 This haze is primarily caused by nitrogen oxides (NOx), sulfur dioxide (SO2), and
particulate matter (PM) pollution from power plants and other anthropogenic sources. See, e.g.,
EPA, Federal Implementation Plan Technical Support Document at A-17 [hereinafter "FIP
TSD"]. For example, at Big Bend, baseline visibility impairment on the most impaired days is
17.3 deciviews (dv). Id. at 34; see also 79 Fed. Reg. at 74,832. These baseline visibility
conditions are far worse than natural visibility conditions at Big Bend, which are 7.16 dv,
according to EPA. FIP TSD at 34, Table 19. Similarly, according to EPA, at Guadalupe
Mountains, baseline visibility impairment on the most impaired days is 17.19 dv, while natural
visibility is 6.65 dv. Id.
404
-------
Earthjustice et al., stated that emissions from Texas sources also impair visibility at Class I areas
in other states. As EPA explains in its proposed FIP, Texas sources cause significant visibility
impairment at the Wichita Mountains Wilderness Area in Oklahoma that are "several times
greater than the impact from Oklahoma's own point sources." 79 Fed. Reg. at 74,822. Texas
sources also cause visibility impairment at Caney Creek Wilderness Area in Arkansas and many
other out-of-state Class I areas, including, but not limited to, Carlsbad Caverns National Park,
Bandelier National Monument, and the Salt Creek and White Mountain Wilderness Areas in
New Mexico; Great Sand Dunes, Rocky Mountain, and Mesa Verde National Parks in Colorado;
Upper Buffalo Wilderness Area in Arkansas; Hercules-Glades and Mingo Wilderness Areas in
Missouri; and Breton Wilderness Area in Louisiana. Id. at 74,830; Texas SIP at 1-5, 11-7 to 11-
28.
Footnotes:
6 Sierra Club also submits the following two reports for background. These reports rely on EPA emissions data to
detail the disproportionate level of harmful SO2 pollution released by Texas power plants covered by this rule in
comparison to other sources in other states. Benchmarking Air Emissions of the 100 Largest Electric Power
Producers in the United States (May 2014), available at
http://www.nrdc.org/air/pollution/benchmarking/files/benchmarking-2014.pdf; Daniel Cohan, Addressing pollution
from legacy coal power plants in Texas (June 2013), available at http://www.dallas-cms.org/news/coalplants.pdf.
7 Nat'l Park Serv. (NPS), Big Bend National Park General Management Plan 103 (2004), available at
http://www.nps.gov/bibe/parkmgmt/gmp.htm. [Documents are attached to comment 0067 - Items 9 and 10]
8 NPS, Guadalupe Mountains National Park Draft General Management Plan 152 (2008), available at
http://parkplanning.nps.gov/documentsList.cfm?parkID=69&projectID=11120.
9Id. at 5.
10 NPS, Understanding Haze in Big Bend National Park, available at
http://www.nps.gov/bibe/learn/nature/upload/Bravo Fact Sheet.pdf.
Response: We agree that the Regional Haze Rule requires that states and we take certain
prescribed steps to protect and improve visibility at Class I areas. We also agree that the
visibility impairment at Texas' Class I areas is partly due to power plant and other anthropogenic
sources. Lastly, we agree that power plants in Texas impair the visibility of Class I areas not
only in Texas, but in other states as well.
Comment: The Regional Haze Program's Public Health, Welfare, and Economic Benefits.
[Earthjustice (0067) p. 12], et al
Earthjustice et al., explained that in addition to improving visibility, pollution reductions under
EPA's regional haze plan for Texas will yield significant public health, economic, and other
environmental benefits.
Earthjustice et al., explained that EPA's proposed FIP would result in billions of dollars in public
health benefits. According to Earthjustice et al., the same pollutants that cause visibility
impairment also cause significant public health impacts. Nitrogen oxides are precursors to
ground level ozone, which is associated with respiratory diseases, asthma attacks, and decreased
lung function. Similarly, sulfur dioxide increases asthma symptoms, leads to increased hospital
visits, and can form particulates that aggravate respiratory and heart diseases and cause
premature death.11 Both NOx and SO2 react with ammonia, moisture, and other compounds to
405
-------
form fine particulate matter that can cause and worsen respiratory diseases, aggravate heart
disease, and lead to premature death.12 PM can penetrate deep into the lungs and cause a host of
health problems, such as aggravated asthma, chronic bronchitis, and heart attacks.13 In 2005,
EPA valued the regional haze program's BART health benefits nationally at $8.4 to $9.8 billion
annually.14
Earthjustice et al., stated that Dr. George D. Thurston, a professor of Environmental Health at
the New York University School of Medicine, and a nationally-renowned expert in the field,
conducted a health risk analysis of EPA's FIP. Dr. Thurston found that EPA's proposed SO2
emissions reductions will significantly reduce the serious public health toll imposed by Texas
coal-burning power plants on residents of Texas and Oklahoma, as well as residents of Arkansas,
Colorado, Illinois, Kansas, Louisiana, Mississippi, Missouri, and New Mexico. [See the Written
Report of George D. Thurston Regarding the Public Health Benefits of EPA's Proposed
Rulemaking Regarding Texas and Oklahoma Regional Haze (Apr. 18, 2015) submitted as
comment 0071], Thurston estimates that EPA's proposed SO2 reductions for the 14 EGUs in
Texas will save at least 316 lives each year, and prevent thousands of asthma-related or
cardiovascular events and hospitalizations every year. Id. at Tables 1-2 provided by commenters
0067 and 0071. Dr. Thurston "conservatively estimate[s]" the total public health-based
economic benefits associated with these reductions will be at least $3 billion each year. Id. 24-
26.15
Earthjustice et al., stated that these are annual benefits, meaning that "ten years from the
compliance date, the health benefits and valuations of the proposed controls will be roughly ten
times" that estimate, before adjustment for a discount rate, as appropriate. Id. 26.
Additionally, Earthjustice et al., stated that if EPA requires SO2 emission reductions at W.A.
Parish and Welsh, as they the agency to do, the health-related economic benefits of the proposed
FIP increase by another $1.1 billion each year. Id. at 17, Table 3 provided by commenters 0067
and 0071. The health-related monetary benefits associated with SO2reductions at Texas's three
largest sources of SO2 pollution—Big Brown, Martin Lake, and Monticello—will alone be more
than $2 billion across the ten states most affected by Texas pollution. Id.
According to Earthjustice et al., Dr. Thurston arrived at these estimates by relying on air quality
modeling conducted by Dr. H. Andrew Gray, an expert in air modeling, who used EPA's
standard air quality modeling methodology to (1) "estimate the incremental PM2.5 concentrations
attributable to each source unit affected by this EPA action," (2) "model[] gridded annual
average PM2.5 concentrations for the high [i.e., current actual] and low emission [i.e., EPA
proposed SO2 emission limit] scenarios," and the (3) "estimate[d] the incremental PM2.5
concentration at each county in Texas and its surrounding states for the control scenarios."
Thurston Decl. 23. (See comment 0070 p.25 summarized below for details on Dr. Gray's
modeling of incremental PM2.5 concentrations.) Dr. Thurston then used the EPA-approved
health risk modeling program, called BenMAP, to translate these pollution reductions into
human health outcomes. Thurston Decl. at 23-24. As Dr. Thurston summarized in the
following tables, even when applying the most conservative estimate of mortality—one
comparable to the estimate EPA relied upon in 2005, to evaluate the regulatory impacts of the
Regional Haze Rule—the public health benefits and avoided medical costs associated with
406
-------
EPA's Regional Haze FIP are considerable.
Tables 1, 2 and 3 provided by commenters 0067 and 0071
Table 1. Health Benefit From EPA < Proposed Federal Flan1'
Health Endpoutt
Expected
Number Per
Year Avoided*
Total DoBar
Valuation
C010S)**
Respiratoiy Hospital Adiuissions (Kloog et at, 2012;
Zanobetti et al, 2009)
591
S1.S69.000
Cardiovascular Hospital Adnusfton* i Bell et al,
2008 Pens et at. 2901; Peng et a!., 2009; Zanobetii
etal.,2009)
5S'1
S2.210.000
Acute BroicMii^»Docker/ et ai 1PP61
639
$307,000
Acute Myocardial Infection. Noiifatai {lope et al,
2006: Suilivan et aL 2005; Zanctei et al,.. 2009;
Zijsotxti & Schwartz 20ut>)
ir
$4,732,000'
fejiiefgffltYKoom visits (Wisef si, zuii. Maret
al, 2010; Similiter et aL,2005)
1S7-
S80.000*
Asthma Exacerbation Symptoms (Mai ei al. 2004;
Ostro ct at. 20011
12,021* 1 $694,000
Upper Respiratory Symptoms (Pope et aL 1991)
11606 i $386,000
l^wer Respiratory Symptoms (Schwartz .\iicl Xem
2000)
s 140 srtw:»
Minor Restricted Activirv Bays (0«ro & Rotlucittld.
1989)
;<:c so i s:0 669 000
W«t Days Los! (Ostro et al IPSTi
51221 | 5? 034 mX:
Chronic Bronchitis. sAbbev et al I9P5i
251 si rxocun.coe:
Mortalitv All Caatei iKrewski et al 2009:-
514 S3 021 IPO 000
Mortafir;, All C auie> iLepeule et al, 2012)
714 1 S6.869.213.000
Monahrv. All C ausei, -Laden et al. 20071
PT| sgTssm®"
* Rounded to nearest whole number.
• *Roundedtoneare:,t Si000
a Pooled effects with averaging approach a* per EPA BenMa$ default setting.
b Pooled effects witli random/fixed effects approach, as per EPA BeuMap default setting,
c Pooled effects with summation approach., a; per EPA BenMap default setting.
407
-------
Table 1. State-By Stare Total Valuation of Annual Health Benefits of EPA Proposed F1P
Applied to Seven Power Plaits At Issue*
AR
HO-
O-
KS
TI-
MS
W
KM
OK
Total
i
Applying Krev.-ita et ,il 2009 for mortality)
(20io$y
ITFmoow
$gji2:wo;
TT:lis5 ooo
inrj^soSM
rrr-s
isssjm.
SI I 314 000
S327JOLOOO
kc Brown. Cotao Geek Lia»"?:ne MartmLake. Mcu3;e-o SanJow ?o&
"Rookfad to newest $1000.
HMfJfLW
Table 3. Plant-By Plant Total Valuation of Annual Health Benefits of EPA Proposed IIP
(Applying Krewski et at., 2009 for i
iiioitalit'fl
-Tt
Bis Brown
C oleto Creek
limestone
Matt in Lake
Moaicello
Sandow
Tolk
Parish*
Welsh*
Parish and Welsh power plane an presented s
but ant nor included lit Tables 1 or 2,.
*' P„<>aiM to nearer $1000.
(2010S)
s**
7.440,000
$253,765,000
$248,613,000
$690,304,000
$681,602,000
5278,718,000
$89,501,000
$748,913,000
$346,853,000
m:h rw or/;«r "pkurs bmd
Earthjustice et al., stated that the figures above are conservative estimates of the health and
economic benefits of EPA's proposed FIP. For the reasons explained in Dr. Thurston's analysis,
these projections likely underestimate the benefits of EPA's rule. Applying a realistic adjusted
mortality estimate, EPA's proposed FIP for Texas and Oklahoma could save as many as 893
lives each year, and up to $8.5 billion annually in public health-based economic costs associated
with the treatment of the thousands of respiratory and cardiovascular events and hospitalizations
that occur each year across the ten states affected by the SO2 emissions from these 14 units in
408
-------
Texas. Id. at Table 1.
Earthjustice et al., stated that the results of Dr. Thurston's analysis of EPA's proposed FIP are
hardly surprising given the extensive and ever-growing scientific literature showing strong
correlations between exposure to air pollution from coal-fired power plants and adverse health
impacts to human beings. Dr. Thurston's findings show that EPA's proposed FIP will result in
significant public health-related economic benefits. Contrary to arguments advanced by several
of the affected facilities, EPA's FIP does not simply protect aesthetic values. The rule will result
in significant, quantifiable public health benefits across a large swath of the central United
States.
Dr. Gray (0070) performed modeling to provide inputs to the health impacts analysis by Dr.
Thurston (0071).
Dr. Gray noted that to estimate the Class I area visibility impacts associated with Texas point
source emissions, EPA Region 6 relied on the results of CAMx dispersion modeling.28 The EPA
developed two emission scenarios for the 21 Texas source units, corresponding to high and low
control levels, which were modeled using projected 2018 emission estimates for all other
sources. For each of the high and low emissions scenarios, each individual source unit was
removed ("zeroed out") one at a time, and the model was re-run. The difference between the
modeled concentrations from all sources and from all sources minus each zeroed out source unit
provided an estimate of the incremental contribution of each individual source. The modeled
PM species concentrations that were contributed by each individual source unit were then used
by EPA to estimate the visibility impacts associated with each source unit (using the most recent
IMPROVE equation to compute extinction), for both the high and low control levels.
The model results for the high and low control levels for each source were used by EPA to
estimate the visibility impacts under various emission scenarios, including 2018 baseline (no
additional controls), and also for EPA's proposed control strategy (consisting of WFGD retrofits
on five units, SDA retrofits on two units, upgraded WFGD on seven units; see FIP TSD Section
7). The modeled concentration impacts (and extinction impacts) were observed to vary linearly
with the SO2 emissions level for each source. EPA performed a linear regression analysis using
the modeled impacts for high and low control levels on each source unit along with earlier 2018
baseline facility-wide model results. The regression slopes (change in extinction per ton of SO2
emissions) were then used to interpolate or extrapolate the model results in order to estimate
impacts for different emission levels at each source unit.
Dr. Gray noted that a similar scaling approach was used to estimate the incremental PM2.5
concentrations attributable to each source unit in his assessment of the health effects associated
with Texas point source emissions. The modeled gridded annual average PM2.5 concentrations
for the high and low emission scenarios, together with the zeroed-out model results, were used to
estimate the incremental PM2.5 concentration at each county centroid in Texas and the
surrounding states. The first step was to extract the modeled PM2.5 concentrations for each grid
cell from the CAMx derived Modeled Attainment Test Software (MATS) output files. The
MATS files contain the simulated concentrations of sulfate (SO4), nitrate (NO3), ammonium
(NH4), elemental carbon (EC), organic compounds (OC), and crustal material (CRUSTAL).
PM2.5 concentrations were calculated as the sum of these components. A list of unit/source
409
-------
numbers was provided (see comment 0070). Concentrations were provided for the following
29 30
scenarios '
• 2018LowBase: 2018 simulation with low controls on 21 units
• 2018Low.zoxx (where xx is 02 to 22): 2018 low control scenario without source 02, 03,
04... 22
• 2018HighBase: 2018 simulation with high controls on 21 units
• 2018High.zoxx (where xx is 02 to 22): 2018 high control scenario without source 02, 03,
04 ... 22
A total of 88 files were prepared (44 for the 12- km grid, and 44 for the 36-km grid). Each file
contains the grid cell number (i, j location) and the modeled annual average PM2.5 concentration
for that grid cell and scenario.
The next step was subtracting the "zo"31 from the "all source" PM2.5 concentrations to determine
each source's contribution, for both high and low control scenarios. This was done separately
for the fine (12-km) and coarse (36-km) gridded receptors.32 Then the fine and coarse grid PM2.5
concentrations were linearly scaled to EPA's baseline emissions levels (EPA estimated the
"current" baseline as the average of actual 2009-2013 CEM emissions, eliminating the maximum
and minimum average years) and also to EPA's proposed control levels. EPA's proposed
control strategy included controls on units 02 through 15 (i.e., not including Parish and Welsh).
For this health assessment, the benefits of WFGD retrofits on Parish and Welsh were also
included.
The resulting gridded PM2.5 concentrations for the baseline and control scenarios were then
interpolated to each county centroid, using a 1/R2 interpolation of the four closest fine or coarse
grid modeled receptors. The PM2.5 concentrations contributed by each facility at each county
centroid were tabulated for the baseline and control scenarios. These PM2.5 data were
subsequently used to estimate the health effects corresponding to each emission scenario. The
modeled low and high emission levels, as well as the baseline emissions and proposed SO2
emission reductions for each modeled source unit is shown in Table 1. Note that EPA's
proposed control plan does not include controls on Parish and Welsh; the SO2 emission rates for
these two facilities under EPA's proposed control plan would reflect no emission reductions and
would therefore be equal to the baseline levels shown in the third column of the table. The
controlled emissions for Parish and Welsh (assuming WFGD scrubber retrofits on Parish Unit 5,
6, and 7 and Welsh Units 1, 2, and 3; and a WFGD upgrade on Parish Unit 8) were evaluated to
determine the additional health effects associated with including retrofit controls on these two
facilities.
SO2 Emission Rates from Dr. Gray's Analysis
(Table 7 from comment 0070)
410
-------
SQ,
.ipvi
C onn o
I Scto.tm
Scut cc I'uit
§tt|I§lilsifC
';.v
BigJBnmnJ
: • i-r 'f
!>• : 1 -
WFGD
Q QfiA
0.979
645.69
Co>to_C:«k_:
„ 1
WFGD
0.957
698.53
JjsneilBiie lim 1
7,422.61
1474.20
10,912,51
WGBifpiie
0.950
2.466.51
52299*
2.614.97
11,946.36
WJrlfJL# l^3gjasie
0.950
2,615.36
Mntm tale f
19.277.73
3.855.55
24.494.93
WFGDqgrade
0.950
3.705.93
Mala. Lake 2
11,652.05
3.884.02
21.580.42
WFGD wade
0.950
3.663.42
lAffeJjieJ
7,444.15
3,722.08
19.940.02
WFGD iferide
0.950
3,551.02
Mioatieello I
13.685.89
1355.30
omit
WFGD
0.970
537.07
IV ,'C efwitlHi
9,200.89
1.145.64
16.429.35
WFGD
0.968
521.90
\ B.M—af f A *5
M«®cett0 J
3.53C. SI
1,851.27
13.856.61
WFtiDifgnde
0.950
1.570.61
Wide* -~
22,978,12
4,595.62
21289.21
WFGD upgrade
0.950
-M121
Tolk rib
7.450.19
U09.35
10.031.39
SDA
0.917
836.31
Tol"l?2%
4520.10
1.103.07
11.033.65
SPA
0.908
1.018.17
Piiish WAP5
10.943.82
1,396.72
14,157.29
1TGD
0.950
708.14
Par.Ji WAP6
7.422.94
1.419.24
15.306.80
WFGD
0.954
703.57
Pan -Ji WAP~
8.107.70
1,24401
12.334.97
WFGD
0.951
601.68
P.ir.sh WAPS
1.79038
1.371.38
1585,18
WFGD fgml
0.950
835.88
Welsh 1
5.892.87
1.110.16
8,013.91
WFGD
0.925
610.12
Welsh, 2
3,973.62
1.117.20
1255,51
WFGD
0.922
647,04
tekli '
5,014.24
:.i:3.85
S 608 30
WFGD
0.925
^<4
Footnotes:
28 Snyder, E., Feldman, M., and Kordzi, J., Technical Support Document for the Oklahoma and Texas Regional
Haze Federal Implementation Plans (FIP TSD), U.S. Environmental Protection Agency Region 6, Dallas, TX,
November 2014.
29 Environ. 2014. "2018 Low-control and High-control CAMx Simulations, Texas Regional Haze Evaluation."
Memorandum to Ellen Belk, EPA Region 6. Prepared by Uarporn Nopmongcol, Greg Yarwood, and Tanarit
Sakulyanotvittaya, Environ International Corporation, Novato, California. August 12, 2014. Electronic file
included in the docket as "Memo_TXHAZE_20181ow_highControls_CAMx_12Ausgl4.docx"
30 High and low control levels were (arbitrarily) selected by EPA corresponding to differing levels of SO2 emission
reduction for each modeled unit.
31 "zo" refers to the set of model results with each individual source removed, or "zeroed out".
32 See Figure 1 of Environ. 2014. "2018 Base Case CAMx Simulation, Texas Regional Haze Evaluation."
Memorandum to Ellen Belk, EPA Region 6. Prepared by Uarporn Nopmongcol and Greg Yarwood, Environ
International Corporation, Novato, California. September 7, 2013. Electronic file included in the docket as
"Memo_TXHAZE_2018CAMx.7Septl 3 .docx"
Footnotes:
11 EPA, Health - Sulfur Dioxide, available at http://www.epa.gov/air/sulfurdioxide/health.html.
12 EPA, Health - Nitrogen Dioxide, available at http://www.epa.gov/air/nitrogenoxides/health.html.
13 EPA, Health - Particulate Matter, available at http://www.epa.gov/air/particlepollution/health.html.
14 EPA, Fact Sheet - Final Amendments to the Regional Haze Rule and BART Guidelines, available at
http://www.epa.gov/visibility/fs_2005_6_15.html. It is important to note that EPA's study focused primarily on the
BART component of the Regional Haze Rule. EPA, Regulatory Impact Analysis for Final Clean Air Visibility Rule
of the Guidelines for Best Available Retrofit Technology (BART) Determinations Under the Regional Haze
411
-------
Regulations, EPA-452/R-05-004 (June 2005), available at
http://www.epa.gov/oar/visibility/pdfs/bart_ria_2005_6_15.pdf. Moreover, EPA's study "assumes that BART-
eligible EGUs affected by the Clean Air Interstate Rule (CAIR) (March 2005) have met the requirements of this
rule. Thus, no additional controls for EGUs beyond CAIR are anticipated or modeled for the 28 State plus District
of Columbia CAIR region." Id. at 2-6. Similarly, the model assumed that "no additional SO2 controls for sources
located in States of Arizona, Utah, Oregon, Wyoming, and New Mexico or Tribal lands located in these States due
to agreements made with the Western Regional Air Partnership (WRAP)." Finally, EPA's analysis assumed that
SO2 BART emission limits of only .15 lbs/MMBtu, as opposed to the lower limits in the proposed rule, and also
assumed that no facilities with existing scrubbers would need further reductions. Id. at 7-2, Table 7-1. As a result,
EPA's 2005 study likely underestimates significantly the health benefits of the regional haze program. 15 Thurston
derived this estimate using an EPA-approved health risk modeling program, in which mortality change is the
product of the projected change in air pollution, exposed population, incidence of mortality, and a "mortality effect
estimate." Id. at 12-13. The mortality effect estimate is an estimate of the percentage change in mortality due to a
one unit change in ambient air pollution. Epidemiological studies are a well-established source for such estimates.
In this analysis, Thurston relied primarily upon a mortality effects estimate that is consistent with the estimate used
by EPA in the agency's 2005 nationwide analysis of the health benefits of Guidelines for Best Available Retrofit
Technology (BART) Determinations Under the Regional Haze Regulations. As Thurston explains, this estimate is a
conservative (i.e., low) mortality effect estimate because it only accounts only for reductions in fine particulate
matter and not the co-benefits associated with reduced SO2 and other pollutants.
16 Reproduced from the Thurston Report, Table 1, Annual Multi-State Human Health Effects and Monetary
Valuations Associated With the PM2.5 Air Pollution Avoided by Applying the Federal Implementation Plan for
Regional Haze and Interstate Transport of Pollution Affecting Visibility in Oklahoma and Texas.
Multiple public hearing commenters ( 0053-8, 0053-41, 0053-61, 0053-62, 0054-1, 0054-2,
0054-9, and 0054-46) described health effects and co-benefits.
Commenter 0053-8 (Public Citizen) supported strengthening the rules because it will improve
the health of residents statewide and support tourism to national parks in Texas. Much of the
haze that tourists encounter in Big Bend National Park comes from NOx, SO2, and PM from
coal-fired power plants that crosses state lines and makes Texans sick. According to statistics
from the Clean Air Task Force, coal-fired generation costs Texans over $2.5 billion a year in
hospital admissions, treatment for conditions such as chronic bronchitis and asthma, and costs
related to premature death resulting from exposure to coal pollution. According to the same
study, 330 people in Texas will die a year from exposure to this toxic soup of chemicals, but this
is an improvement from previous years as in 2000 there were 369 deaths in the Dallas/Fort
Worth area alone from coal-fired pollution.
Commenters 0053-8, 0053-61, and 0054-2 noted that the EPA has estimated that implementation
of the regional haze plan would prevent 1,600 premature deaths, 2,200 nonfatal heart attacks, and
over 1 million lost school/work days due to pollution-related illnesses.
Commenter 0053-8 noted that constituents found in haze are detrimental to the health of Texans,
including children, people with asthma, and the elderly. Exposure to these constituents can
contribute to increased respiratory illnesses, decreased lung function, increased risk of cardiac
arrest, and premature death. The sulfate found in haze also contributes to acid rain formation.
Fine particles that are found in haze are especially damaging to lungs and circulatory systems
and are even carcinogenic. Long-term exposure to fine particles are associated with the
development of chronic bronchitis and the worsening of symptoms in people who have COPD,
coronary artery disease, asthma, and diabetes.
412
-------
Commenter 0053-41 supported strong rules to end air pollution and improve visibility in national
parks. The most critical impacts of allowing our air to be polluted with PM, SO2, and NOx are
the health problems, especially respiratory illnesses, which these pollutants cause. Since the
early 1980s scientists have linked urban haze in the Dallas/Fort Worth area to east Texas coal
plants. The cost of this pollution comes in the form of medical bills, lost productivity, lost
educational opportunity, and sometimes even loss of life.
Commenter 0053-62 noted that coal-plant haze is an esthetic problem and SO2 poses health
effects. The commenter noted that Monticello, Martin Lake, and Big Brown are on the list of
affected coal plants (Luminant's dirtiest old technology), and contended that emissions from
those coal plants are particularly bad for people's health. The commenter stated that, through
their work on the Beyond Coal campaign, they met people living around Texas coal plants who
suffer different ailments.
Commenter 0054-1 supported the EPA's recommendations and stated that cleaning up haze and
SO2 emissions will help clean the air in DFW. It has been proven that Monticello, Martin Lake,
and Big Brown affect DFW in air quality and in turn the health of our families and all those who
live there. Retrofitting plants with scrubbers and other pollution controls are likely to not only
affect the levels of NOx to the tune of 228,000 tons but will also likely reduce NOx and VOCs,
which cause smog, CO2, and mercury.
Commenter 0054-2 stated that children are at greatest risk from air pollution because they're
more likely to be outside. Their lungs are still developing. Asthma attacks one out of every ten
school children in the U.S., and this is the number one health issue that causes kids to miss
school.
Commenter 0054-9 stated that Dallas/Fort Worth is the second highest region for ozone pollution
in the United States, and 1200 people die from ozone poisoning every year in Texas. Five Texas
power plants are major contributors to SO2 and NOx and are several times higher than number
six in the whole country. The commenter supported EPA in adopting tighter haze standards to
protect our health.
Commenter 0054-46 supported the rule and encouraged the EPA not to capitulate to the power
corporations in consideration of people's health, children's health, wildlife, and the views.
[Jeannie McDaniel, OK House of Representatives (0080) p. 2] OK Representative McDaniel
stated that there are important public health co-benefits of the EPA proposal. For example, the
14 coal boilers covered by the regional haze rule emit 35% of the ozone-causing NOx pollution
from all the 121 power plants in Texas. In addition, the 14 coal boilers covered by the regional
haze proposal emit 26% of the carbon emissions from all the power plants in Texas. Just
replacing these 14 coal boilers with clean energy will get the state 2/3 of the way to meeting the
EPA's target in the Clean Power Plan for carbon emissions from Texas.
Commenter 0053-11 (Air Alliance Houston) supported the proposed rule and stated that the
welfare benefits of the rule are not trivial. The national parks and wilderness areas bring billions
of dollars in commerce (tourism) every year and provide psychological and health benefits to
people. The commenter suggested that that all benefits be considered in the cost-benefit analysis
413
-------
for the rulemaking.
Commenter 0053-62 stated that pecan farms across the street from Austin's LCRA Fayette power
plant were reunited by SO2 from that plant, ruining people's livlihoods. The SO2 deposits on the
top of the leaves and also comes in to the leaves from the stomata underneath, and over about 20
years it killed entire pecan groves.
Commenter 0054-2 stated that the haze pollution in wilderness areas is far more serious than just
providing a beautiful view of our national parks and wildlife areas because the poisonous haze
kills flora and fauna.
Response: We appreciate the commenters' concerns regarding the potential adverse health effects
and resulting benefits of air pollution controls to improve air quality In Class I areas. We generally
agree that the same emissions that cause visibility impairment can also cause health related
problems, such as respiratory ones. Although our action addresses visibility impairment, we note
that there is the potential for improvements in human health through reductions in regional
concentrations of visibility impairing pollutants. We generally agree with Representative
McDaniel, Dr. Thurston, Earthjustice and the other commenters that in addition to improving
visibility, pollution reductions under our proposal will yield significant public health, economic,
welfare, and other environmental benefits. Thus, because our FIP will lead to emissions
reductions, there will be co-benefits for public health. However, for purposes of this action, we
are not specifically considering these health based benefits under this visibility program.
Therefore to the extent that the focus of Dr. Thurston's report was to address the health effects of
our proposal, we do not specifically respond to it as our Regional Haze program targets visibility
impairment only.
More specifically, EPA also recognizes the role that protecting visibility in national parks and
wilderness areas has to tourism throughout the state. Reducing regional haze will help ensure that
views in these parks and wilderness areas are preserved, and will continue to support tourism.
Further, we appreciate the comment regarding the negative ecosystem impacts of visibility
impairing emissions. We agree that visibility impairing pollutants can have negative impacts on
ecosystems, however, for purposes of this Regional Haze action, we are not authorized to consider
these ecosystem impacts. Therefore, while we note the potential for co-benefits to ecosystem
health resulting from our action today, we have not taken these potential benefits specifically into
account in this action. Lastly, we acknowledge that today's action may have positive economic
impacts, as described by Commenter. We agree that our Class I areas and other national parks and
wilderness areas are a source of jobs and contribute to the economies of their respective states.
We also agree the visibility at Class I areas must be improved according to our instructions from
Congress as codified in the Clean Air Act but we cannot specifically take these types of potential
economic benefits into account in our action.
Comment: Wilderness areas and State parks. Commenter 0053-32 pointed out that we are
not only dealing with Class I properties here but wilderness areas. There is designated
wilderness at Guadalupe Mountains National Park and proposed wilderness at Big Bend National
Park. Part of wilderness is clean air and far vistas. The National Park Service cannot manage
414
-------
these areas as wilderness unless the EPA acts and requires action which the TCEQ will not
require.
Commenter 0053-32 also pointed out that in addition to national Class I and wilderness areas,
State parks will benefit (e.g., Big Bend Ranch State Park, Davis Mountains State Park, and
Chinati Mountain State Park).
Response: We understand that Class I areas also include wilderness areas. We agree that
although the regional haze program targets Class I areas, other scenic areas such as state parks
will likely co-benefit as well.
Comment: The Aggregate Visibility Benefits of Controls Texas Considered Are
Significant. [Earthjustice (0067) p.22]
Earthjustice et al., noted that Texas concluded that no reasonable progress controls are necessary
in part because of Texas's conclusion that the visibility benefits would be too low. But Texas's
analysis is both unsupported and inaccurate. The analysis is unsupported because the record is
devoid of any analysis that Texas may have undertaken to determine whether the projected
visibility improvement would be significant relative to the costs. Instead, Texas merely asserts
that the projected benefits are too low to justify the costs of controls. EPA cannot approve a
state plan that is neither supported by the record nor well-reasoned. See Oklahoma, 723 F.3d at
1206-12; North Dakota, 730 F.3d at 761.
Earthjustice et al., stated that Texas references a threshold of "perceptibility" and also the
threshold of 0.5 dv sometimes used to determine whether a source is subject to BART. Neither
threshold justifies Texas's failure to consider cost-effective controls. As EPA appropriately
explains, "perceptibility" is not a determining factor in establishing the significance of visibility
impacts or benefits. 79 Fed. Reg. 74,840.
According to Earthjustice et al., the implied use of a 0.5 deciview threshold, taken from the
BART Guidelines, is likewise inappropriate. First, the BART Guidelines identify 0.5 dv as the
highest threshold to use to determine whether a source should be analyzed for BART controls.
The BART Guidelines do not suggest using 0.5 deciviews as a threshold for determining when
visibility improvement is significant enough to warrant requiring controls. Indeed, both states
and EPA have required controls that provide benefits of less than 0.5 dv. See, e.g., FIP TSD at A-
75 (mentioning controls required in Wyoming and Arizona that would provide a visibility benefit
less than 0.5 deciviews). Second, even in the context of determining whether a source is subject
to BART, EPA guidelines urge states to consider thresholds lower than 0.5 dv, especially where
there are a large number of sources having impacts on a given Class I area, 40 C.F.R. pt 51 App.
Y § III(A), as is clearly the case with Texas's 1600+ point sources.
Moreover, Earthjustice et al., stated that Texas never acknowledged that the magnitude of
visibility improvement from reasonable progress controls will always appear much lower than
they do in the BART context because of the differences between the modeling process used to
evaluate reasonable progress and modeling used to evaluate whether a source is subject to
415
-------
BART. As discussed in the attached Gray Report (comment 0070), the modeling differs in three
fundamental ways: (1) the background visibility conditions against which improvement is
measured; (2) whether the impact on the worst twenty percent of days is measured versus the
maximum impact; and (3) whether the emissions input into the model are based on annual
average emissions or on the maximum 24- hour emissions. Texas appears to have judged the
benefits from reasonable progress controls to be too small without ever having adjusted for these
fundamental differences between modeling the visibility improvement from reasonable progress
versus BART controls.
In the table below (from comment 0067), Earthjustice et al., noted that EPA translated its
reasonable progress modeling results from a "dirty" background (one that considers the impacts
of all sources together, thereby lowering the impact of any one source) into results from a
"clean" background (one that considers each source individually against a natural background).
This example addresses only one of the three main differences described above between Texas's
reasonable progress modeling and typical BART modeling that can make reasonable progress
controls appear to provide a smaller benefit.
Table 4 from comment 0067
I . ¦: !:; ¦ ¦ :i.. I : ; ¦ "
i !!;i : ¦' ¦ ji - '-I:' .!s > !¦. •! * *'¦ "i; I, f i.i
Mountain'
21)18 "Dim " Avei agt Nafui al
"C Iran" omul
Proposed .984
Scrubber Retrofits
and Upgrades
31 EPA, FIP TSD at A-76.
When translated in this way, Earthjustice et al., stated that the benefits Texas evaluated are on
the same order of magnitude as EPA's. 79 Fed. Reg. 74,837. EPA's table reproduced above
shows that when compared to natural background visibility conditions, the benefits Texas
evaluated in the aggregate would have been above its implied threshold of 0.5 dv, even without
accounting for the differences in metrics and emission rates between Texas's modeling and
typical BART analyses. The use of a 0.5 dv threshold is not appropriate in this context, but the
benefits that Texas dismissed were significant even by that flawed standard. Moreover, when the
visibility improvement is translated from the results based on a dirty background to a clean
background, the results are nearly five times as large. Texas, however, appears to have
considered only the visibility improvement modeled against a dirty background.
Earthjustice et al., summarized that Texas's analysis is unreasonable because the thresholds
416
-------
advanced lack substantiation or merit, and the benefits dismissed are significant. To the extent
that Texas relied on a threshold of 0.5 dv, its analysis was flawed because the threshold itself is
inappropriate; the BART Guidelines mention 0.5 deciviews for evaluating whether baseline
impacts are significant enough to warrant an analysis of potential BART controls, not to evaluate
whether the benefits of controls are significant enough to require the controls, Moreover, the
BART Guidelines suggest a maximum threshold of 0.5 deciviews for use with a different type of
modeling than is used for modeling reasonable progress controls. Even if 0.5 deciviews were an
appropriate threshold, the benefits of the controls Texas considered would be at least 0.5
deciviews when accounting for the differences between its reasonable progress modeling and
typical BART modeling. Finally, Texas's analysis was flawed because it inappropriately
dismissed visibility benefits as being too low to be "perceptible" when in fact perceptibility is
not an acceptable threshold.
Response: We agree with Earthjustice that many areas of Texas' regional haze SIP were
inadequate. We agree, as we discuss in another response to comments, that a threshold of
perceptibility for visibility improvement is not appropriate for the acceptability of a control. We
similarly agree that the BART threshold of 0.5 is not appropriate for use in our proposal for all of
the reasons Earthjustice cites. As we discuss in our proposal and elsewhere in our response to
comments, there is no way to directly compare the CAMx modeling we used in our proposed
TX/OK FIPs and used by Texas in its SIP to estimate the visibility benefit from controls with
previous CALPUFF modeling results and thresholds established with respect to those results due
to differences in the model, model inputs, and metrics used. Therefore, comparisons to
CALPUFF-derived thresholds are not appropriate.
Comment: EPA Should Disapprove Texas's Determination that No Source is Subject to
BART. [Earthjustice (0067) p.57]
Earthjustice et al., stated that Texas identified 126 sources as BART-eligible or potentially
BART eligible. See SIP at 9-2 to 9-4; 79 Fed. Reg. at 74,845-47. Yet Texas ultimately
concluded that no BART eligible source is subject to BART. SIP at 9-10.66 Texas's
determination is based in part on the unsupported selection of 0.5 deciviews as the threshold for
contribution to visibility impairment. In addition, Texas's analysis for the ExxonMobil
Beaumont refinery unlawfully uses future, projected emissions as the baseline rather than actual
emissions. For these reasons, EPA must disapprove Texas's determination as to the sources
subject to BART.
Earthjustice et al., stated that Texas adopted 0.5 deciviews as the threshold for "contribution" to
visibility impairment. SIP Volume I at 10-4. If a source's 98th percentile, baseline impact was
below 0.5 deciviews at each affected Class I area, the source was deemed not subject to BART.
But Texas provided no justification for using a 0.5 deciview threshold. There is no
documentation in the record as to how or why Texas selected this threshold, and there is no legal
support for such threshold. EPA's BART Guidelines do not authorize states automatically to use
a 0.5 deciview contribution threshold. Instead, the BART Guidelines state only that "any
threshold that you use for determining whether a source 'contributes' to visibility impairment
should not be higher than 0.5 deciviews." 40 C.F.R. pt. 51, App. Y, § 111(A)(1). In the next
417
-------
sentence, the Guidelines instruct each state that it "should consider the number of emissions
sources affecting the Class I areas at issue and the magnitude of the individual sources' impacts."
Id. There is no evidence in the record that Texas ever conducted this analysis.
Furthermore, Earthjustice et al., noted that the Guidelines conclude that "a larger number of
sources causing impacts in a Class I area may warrant a lower contribution threshold." Id. As
Texas's list of 126 BART eligible sources indicates, a large number of sources impact the Class I
areas in Texas and in neighboring states. Indeed, the subset of sources that screened out of
BART based on individual modeling have a combined, baseline impact of nearly 10 deciviews.67
Thus, the situation in Texas is exactly what EPA had in mind when it noted that a contribution
threshold lower than 0.5 deciviews may be appropriate. Had Texas followed the BART
Guidelines, it may well have selected a threshold lower than 0.5 deciviews.
Earthjustice et al., stated that using a lower contribution threshold would change Texas's
conclusion as to which sources are subject to BART because there are sources with a baseline
impact just below 0.5 deciviews. For example, the Ash Grove Cement plant's 3-year average of
the 98th percentile visibility impact ranges from .0325 to 0.431 at Wichita Mountains, based on
the modeling approach used. SIP Appendix 9.8, BART Exemption Modeling Report, Ash Grove
Cement Company, Midlothian, Texas at 13-14. ExxonMobil's Beaumont refinery has a 3-year
average, 98th percentile impact of .422 deciviews at Caney Creek. SIP Appendix 9.8, BART
Modeling Analysis ExxonMobil Beaumont Refinery at 16. The International Paper, Texarkana
Mill has a 3-year average, 98th percentile impact of .296 deciviews at Caney Creek. SIP
Appendix 9.8, International Paper Demonstration of Exemption from BART Requirements at 28.
Earthjustice et al., stated that the EPA has a statutory responsibility to ensure that a SIP meets all
applicable Clean Air Act requirements and is supported by the record. See 42 U.S.C. §
7410(k)(3), (1); North Dakota v. EPA, 730 F.3d at 761 (EPA must examine the substance of a
state's BART determinations in order to determine whether the state's decision and plan is based
upon "reasonable analysis" and is "reasonably moored" to the Clean Air Act's requirements);
Oklahoma v. EPA, 723 F.3d at 1208-10 (EPA reasonably disapproved BART determinations that
did not comply with EPA regulations and guidance). Here, Texas's use of a 0.5 deciview
threshold has two fatal flaws: it is not based on the analysis prescribed by the BART Guidelines,
and it is not supported by any analysis whatsoever in the record. Therefore, EPA must
disapprove Texas's conclusions that sources are not subject to BART, where Texas screened out
sources because of a visibility impact below 0.5 deciviews. EPA must then develop an
appropriate contribution threshold and determine which sources have impacts above the
contribution threshold EPA develops and are therefore subject to BART.
In addition to using an unsupported 0.5 deciview threshold, Earthjustice et al., stated that Texas
used unlawful assumptions to conclude that ExxonMobil's Beaumont facility is not subject to
BART. Modeling showed that using the Beaumont facility's actual emissions between 2002 and
2004, the subject had a maximum impact greater than 0.5 deciviews at 7 Class I areas. SIP
Appendix 9.8, BART Modeling Analysis ExxonMobil Beaumont Refinery at 14. Instead of then
examining the 98th percentile impacts at these Class I areas, Texas relied on additional modeling
based on projected emissions reductions based on anticipated controls; the additional modeling
showed impacts lower than 0.5 deciviews, which led the State to conclude that the facility is not
418
-------
subject to BART. Id. at 16.
Earthjustice et al., stated that the BART Guidelines instruct states that the "baseline emissions
rate should represent a realistic depiction of anticipated annual emissions for the source." 40
C.F.R. pt 51 App. Y § IV(D)(4)(d). "In general," baseline emissions should be "based upon
actual emissions from a baseline period." Id. While baseline emissions can be based on future
changes at the source, the overriding mandate is that baseline emissions be a "realistic depiction
of anticipated annual emissions." Id.
Here, it is unclear whether the pollution controls that were expected to reduce emissions at the
Beaumont facility have actually been used consistently since they were installed. In 2008, EPA
fined ExxonMobil for widespread violations of emissions limits and monitoring requirements at
its refineries, including the Beaumont facility. EPA's finding in 2008 that ExxonMobil had
violated the consent decree governing operation of pollution controls at its refineries followed a
2005 consent decree resolving EPA's allegations that the refineries were in violation of multiple
Clean Air Act requirements. EPA, ExxonMobil Refinery Settlement, available at
http://www2.epa.gOv/enforcement/exxonmobil-refinery-settlement#violations.
Earthjustice et al., stated that there is no evidence in the record that the baseline emissions Texas
used for the Beaumont refinery represent a "realistic depiction of anticipated annual emissions,"
as required by the BART Guidelines. 40 C.F.R. pt 51 App. Y § IV(D)(4)(d). At a minimum,
EPA should examine and document in the record whether the Beaumont refinery has actually
achieved the emissions rates that Texas used in its future emissions scenario. If the Beaumont
refinery's emissions have exceeded the emissions rates that Texas modeled, EPA should
disapprove Texas's conclusion that ExxonMobil Beaumont is not subject to BART, conduct its
own analysis of whether the facility is subject to BART, and conduct a five-factor analysis if the
facility is subject to BART.
Footnotes:
66 It is unclear whether Texas found that certain EGUs are subject to BART, but that CAIR could substitute for SO2
and NOx BART controls, or whether Texas found that it need not even determine whether EGUs subject to CAIR
are subject to BART. Regardless, Texas did not conduct SO2 and NOx BART analyses for any EGU.
67 See EPA, BART TSD-Texas (Nov. 2014) at 74-75 (29 sources that screened out based on CALPUFF modeling
had a baseline impact of approximately 8.1 deciviews); Id. at 77 (6 sources that screened out based on CAMx
modeling had a baseline impact of 1.3 deciviews). For the CALPUFF modeling, we avoided double-counting the 2
results reported for Ash Grove. We acknowledge the differences in the CALPUFF and CAMx models, and further
acknowledge that the CALFPUFF modeling was conducted differently for different sources, based on either the
CENRAP or refined modeling approaches. Thus, aggregating the impacts reported from the different modeling
should be viewed as an approximation; the results would likely be slightly different if all sources had used the same
modeling approach.
Response: We disagree with the comment. We address the screening model impacts for
individual BART source using CAMx elsewhere in this document. We note that this modeling
based on short-term permitted emission levels were 0.42 dv or less and would have been even
lower if based on actual emission levels. For other sources that relied on CAMx modeling for
BART screening, maximum modeled impacts for groups of sources were compared to the 0.5 dv
threshold, and therefore any individual source's impact would be only a portion of the group's
impacts that fell below 0.5 dv. Those sources that screened out based on CALPUFF modeling
419
-------
relied on a 0.5 dv threshold to demonstrate that they did not cause or contribute to visibility
impairment at any Class I area. Overall EPA did not find any issues/concerns with the
CALPUFF and CAMx modeling evaluations because they followed modeling protocols
established in consultation with EPA and FLM representatives, including the choice of threshold.
EPA's BART Guidelines allow states conducting source-by-source BART determinations to
exempt sources with visibility impacts as high as 0.5 dv.331 While we agree that a state may
choose to use a lower threshold, this should be based on consideration of not only the number of
sources, but the proximity to the Class I area and the potential combined visibility impacts from
a group of sources. Therefore, if a group of sources are located such that visibility impairing
pollution from that group of sources are anticipated to be transported together and
simultaneously impact visibility at a Class I area, a lower threshold may be appropriate to
address the impacts from the sources within that group. The cumulative visibility impacts cited
to by the commenter (e.g., 29 sources that screened out based on CALPUFF modeling had a
baseline impact of approximately 8.1 deciviews) combines the maximum visibility impacts from
each facility without any consideration of the location of the source or even the Class I area that
is being impacted. The commenter's approach overstates the combined impact at a given Class I
area and does not contemplate if sources are located near each other and would likely impact a
Class I area at the same time.
With regards to comments concerning ExxonMobil Beaumont, as the commenter states, the
facility was subject to a consent decree in 2005, prior to Texas submitting its SIP. Therefore at
the time of the SIP submittal, future emission levels tied to enforceable emission reductions were
identified for the source and considered as a realistic depiction of anticipated future emissions in
the BART screening.332 Violations of the consent decree and permitted emission limits are an
enforcement matter and outside of the scope of this action.
Finally, we note that all Texas point sources, including those BART sources that screened out,
were included in our Q/d analysis for RP and subsequent modeling to identify those sources with
the largest impacts on visibility conditions on the 20% worst days.
With regard to BART-eligible EGUs, as we discuss in our proposal,333 we have already issued a
limited disapproval of the Texas regional haze SIP for its reliance on CAIR. However, we
determined that CSAPR provides for greater reasonable progress towards the national goal than
would BART and Texas is included in CSAPR for NOX and SO2. Therefore, our proposed FIP
to replace reliance on CAIR with reliance on the trading programs of CSAPR as an alternative to
BART included a FIP to replace Texas' reliance on CAIR in 30 TAC 116.1510(d) with reliance
on CSAPR. We proposed to approve the remainder of the provisions in the Texas BART rules
and Texas' application of the BART rules regarding the identification of all BART eligible
sources within the state and the screening of BART sources from full BART analysis.
On July 28, 2015, the D.C. Circuit Court's issued its decision in EME Homer City Generation v.
331 40 CFR part 51 Appendix Y Section III. A. 1
We note that emissions inventory data for 2009-2011 show annual emissions from the facility range from about
500 to 750 tpy of SO2.
333 79 FR 74854.
420
-------
EPA, 795 F.3d 118 (D.C. Cir 2015), upholding CSAPR but remanding without vacating a
number of the Rule's state emissions budgets. Specifically, the court invalidated a number of the
Phase 2 ozone season NOX budgets and found that the SO2 budgets as to four states resulted in
over control for purposes of section 110(a)(2)(D)(i) of the CAA. Texas' ozone season NOx
budget and its SO2 budget are both implicated in this remand. We are in the process of acting on
the Court's remand. As a result, at this time we cannot ensure that CSAPR will continue to be an
appropriate alternative to BART for Texas EGUs. Given the uncertainty arising from the
remand of some of the state CSAPR budgets, we have decided not to finalize that portion of our
FIP relying on CSAPR as an alternative to U and NOx BART for EGUs in Texas.
As the question of how best to address the BART requirements for these significant sources of
emissions of visibility impairing pollutants remains undecided, we have also concluded that our
proposed portion of the FIP to address the requirements of section 110(a)(2)(D)(i)(II) with
respect to visibility should also be addressed in a future rulemaking. We will address the
question of appropriate SO2 and NOX BART limits for EGUs in Texas and the remaining issues
in a future rulemaking once EPA has determined how best to respond to the remand of some of
the CSAPR state budgets. We note that a few of the sources for which we are finalizing SO2
controls as part of the Texas long-term strategy are also BART-eligible. Should we determine in
the future that it is necessary to perform source-specific BART determinations for these sources
instead of relying on CSAPR, we anticipate that the SO2 controls we are finalizing today, which
are currently the most stringent available, will also be sufficient to satisfy the BART
requirement.
Comment: TCEQ properly relied on aggregate visibility benefits analysis to determine that
available controls were not reasonable. [NRG (0078) p. 5]
NRG stated that the EPA has proposed to disregard Texas' determination that available controls
to reduce regional haze were not reasonable because Texas considered aggregate costs and
visibility benefits. "We propose to find that the TCEQ's analysis is insufficient to determine the
visibility benefit.... TCEQ's reliance on an aggregate analysis materially affected its
conclusion that existing and scheduled controls would achieve reasonable progress. " 79 Fed.
Reg. at 74,841.
NRG stated that the Texas approach is consistent with "option 3" identified in EPA's regulations
for demonstrating that no sources in a state should be subject to BART, which states:
You may also submit to EPA a demonstration based on an analysis of overall visibility
impacts that emissions from BART-eligible sources in your State, considered together,
are not reasonably anticipated to cause or contribute to any visibility impairment in a
Class I area, and thus no source should be subject to BART. 40 C.F.R. Pt. 51, Appx. Y,
§111. A.3, Option 3.
NRG stated that the EPA's underlying analysis makes clear that BART was the agency's template
for doing its analysis. Among the numerous comparisons and references to BART rules, one of
EPA's technical support documents states:
421
-------
Our Reasonable Progress Guidance notes the similarity between some of the reasonable
progress factors and the BART factors contained in section 51.308(e)(l)(ii)(A), and
suggests that the BART Guidelines be consulted regarding cost, energy and non-air
quality environmental impacts, and remaining useful life. We are therefore relying on
our BART Guidelines for assistance in interpreting those reasonable progress factors as
applicable.
.... Also, similar to a BART analysis, we are considering the projected visibility benefit
in our analysis.20
NRG stated that not only was the Texas approach consistent with other EPA regional haze
regulations, it also resulted in the objectively reasonable conclusion that substantial new controls
with imperceptible visibility impacts are unreasonable. Texas determined what visibility benefits
might accrue from new SO2 control strategies, and that the maximum projected visibility benefit
was only 0.36 deciview, which could not be perceived by the human eye.21
Thus, NRG disagreed with EPA's proposed rejection of Texas' reliance on an aggregate cost and
visibility benefit analysis.
Footnotes:
20 EPA, Technical Support Document for the Oklahoma and Texas Regional Haze Federal Implementation Plans
(FJP TSD) (Nov. 2014), at 6.
21 EPA, Technical Support Document for the Texas Regional Haze State Implementation Plans (TX TSD) (Nov.
2014), at 15.
Response: It appears that NRG assumes that our reasonable progress and long-term strategy
was completely borrowed from the BART Guidelines, which is incorrect. As we state in our FIP
TSD and as NRG reproduces, "We are therefore relying on our BART Guidelines for assistance
in interpreting those reasonable progress factors as applicable [emphasis added]." Certainly,
there are many aspects of the BART analysis that are not applicable, such as the limitations of
which sources are BART-eligible (some of the sources we proposed to control are not BART-
eligible sources), the basic methodology of whether a source is subject to BART, etc.
NRG reproduces a part of our BART Guidelines and proposes that Texas's approach falls under
the BART option in which the state's overall visibility impact can be analyzed in order to
demonstrate that it is so small that individual source impacts are inconsequential. The full
portion of the quote that NRG references and asserts to be applicable is:
You may also submit to EPA a demonstration based on an analysis of overall
visibility impacts that emissions from BART-eligible sources in your State,
considered together, are not reasonably anticipated to cause or contribute to any
visibility impairment in a Class I area, and thus no source should be subject to
BART. You may do this on a pollutant by pollutant basis or for all visibility-
impairing pollutants to determine if emissions from these sources contribute to
visibility impairment.
For example, emissions of SO2 from your BART-eligible sources may clearly
422
-------
cause or contribute to visibility impairment while direct emissions of PM2.5 from
these sources may not contribute to impairment. If you can make such a
demonstrati on, then you may reasonably conclude that none of your BART-
eligible sources are subject to BART for a particular pollutant or pollutants. As
noted above, your demonstration should take into account the interactions among
pollutants and their resulting impacts on visibility before making any pollutant-
specific determinations.
In fact, SO2 emissions from Texas do clearly impact the visibility impairment at a number of
Class I areas both inside and outside of Texas. For example, the following chart was compiled
from information present in the TX regional haze SIP:334
Texas Impacts on Wichita Mountains, 2018
Percent Extinction
- TX Non-EGU point, 6.3%
-TX EGU point, 7.7%
Tx point source contribution
Is 3.6 times larger than OK's
Source: CENRAP-2007
m to t Co M»«
¦SC*»
¦ SOU
¦ NcoLCU Ptlt r« trfft
¦ UiKtfH (Tk OtJtf
¦ Atm
¦ low-iavvl lc«it
.. .1 1
I.II
Clearly, Texas's contribution to the visibility at the Wichita Mountains alone is much greater
than that due to Oklahoma itself, in which the Wichita Mountains Class I Area is located. As
Oklahoma noted in its SIP, "Texas contributes more to visibility impairment at the Wichita
Mountains than Oklahoma or any other state does."335 Consequently, even if Texas had desired
to model its reasonable progress and long-term strategy after the above cited provision of the
334 Appendix 8-1, "Technical Support Document for CENRAP Emissions and Air Quality Modeling to Support
Regional I-Iaze SIP."
335 Regional Haze Implementation Plan Revision State of Oklahoma, Department of Environmental Quality.
February 2,2010. Page 107.
423
-------
BART Guidelines, it would not have been able to make this type of demonstration. Furthermore,
NRG confuses the estimated visibility improvement from controls estimated by TCEQ and the
visibility impacts anticipated from Texas emissions. We consequently disagree with NRG's
assertion that Texas' approach was reasonable and we disagree specifically that Texas's
approach of aggregating the visibility impacts and costs was appropriate.
13. Natural Conditions
Comment: Earthjustice et al., provided background on baseline and natural conditions.
[Earthjustice (0067) p.7]
Earthjustice et al., stated that for each Class I area within its borders, the state must calculate
baseline and natural visibility conditions. 40 C.F.R. § 51.308(d)(2). Both baseline and natural
visibility conditions must be calculated for the twenty percent most impaired days and the twenty
percent least impaired days. Id. § 51.308(e)(3). Given the statutory goal of eliminating all haze
in Class I areas caused by "manmade" air pollution, 42 U.S.C. § 7491(a)(1), the calculation of
natural visibility conditions establishes the ultimate goal for each Class I area. Each haze plan
must then make reasonable progress toward achieving natural visibility conditions. Id. §
7491(b)(2). Thus, it is critical for a state to properly determine natural visibility conditions.
Response: The EPA acknowledges the background information provided by the commenter.
We take no position with respect to the commenter's synopsis.
Comment: Luminant provided background information on Texas's determination of the
uniform rate of progress for Big Bend and Guadalupe Mountains. [Luminant (0061) p. 18]
Luminant noted that, to set the URPs, Texas first calculated baseline, natural, and current
visibility conditions at Big Bend and the Guadalupe Mountains. Using the Visibility Information
Exchange Webs System ("VIEWS") and the IMPROVE algorithm, Texas calculated baseline
conditions using data from 2001 to 2004.137 At Big Bend, baseline visibility was calculated at
5.78 deciviews for the best 20 percent days and 17.30 deciviews for the worst 20 percent days.138
At the Guadalupe Mountains, baseline visibility was calculated at 5.95 deciviews for the best 20
percent days and 17.19 deciviews for the worst 20 percent days.139 For natural conditions, Texas
estimated all fine soil and coarse mass to be naturally occurring.140 Based on the assumption of
coarse mass and fine soil as 100 percent natural, Texas calculated natural visibility at Big Bend
at 10.09 deciviews for the worst 20 percent days and at 2.19 deciviews for the best 20 percent
days.141 Texas calculated natural visibility at the Guadalupe Mountains at 12.26 deciviews for
the worst 20 percent days and at 2.10 deciviews for the best 20 percent days.142 Texas also
provided a comparison with 80 percent fine soil and coarse mass as natural.143 When assuming
80 percent fine soil and course mass as natural, the estimated natural visibility would be 9.2
deciviews at Big Bend for the worst 20 percent days and would be 11.0 deciviews at the
Guadalupe Mountains for the worst 20 percent days.144
424
-------
Luminant noted that Texas calculated the uniform rate of progress to achieve natural conditions
by 2064 by using the calculated baseline conditions and natural visibility conditions that rely on
100 percent coarse mass and soil as natural.145 As shown in Table 1 provided by Luminant, under
the uniform rate of progress, visibility conditions in 2018 would be 16.0 deciviews at Guadalupe
Mountains and 15.6 deciviews at Big Bend.146
Table 1 from Luminant Comment (0061): Texas's 2018 URPs for Big Bend and Guadalupe
Mountains
Class ! Area
State f stat> kshed URP (20 8)
(20% wool days)
Big Bend
! ..
G^ack.upe Mountains
16.0 cfvs
Response: The EPA acknowledges the background information provided by the commenter.
We take no position with respect to the commenter's synopsis.
Comment: Luminant provided a summary of EPA's review of Texas's URPs for Big Bend
and Guadalupe Mountains. [Luminant (0061) p. 48]
Luminant noted that the EPA proposes to approve Texas's determination of baseline visibility
conditions for Big Bend and Guadalupe Mountains.346 EPA proposes to disapprove Texas's
determination of natural visibility conditions. In calculating natural visibility conditions, Texas
used EPA's default values for all components of visibility impairment, except fine soil and
coarse mass, which it treated as 100% attributable to natural conditions.347 As discussed in
correspondence, several of the Federal Land Managers recognized that EPA's regulations at "40
CFR 51.308 gives the State [the] right" "to recalculate natural conditions for Big Bend NP and
Guadalupe Mountains NP" and concurred "that the basic approach used [by Texas] to adjust
natural conditions is reasonable, provided that the Proposed SIP address the uncertainty of the
assumption that all of the coarse mass and fine soil fraction on the worst 20 percent days is
natural."348
Luminant stated that the EPA disagrees with both Texas and the Federal Land Managers and
would instead use a default value that does not capture actual natural conditions at these areas.
EPA bases its proposed disapproval, not on any actual data that refutes TCEQ's determination,
but on the assertion that "TCEQ has not adequately demonstrated that all coarse mass and fine
soil measured in the baseline period can be attributed to 100% natural sources."349 EPA,
however, does not address or refute the substantial body of technical data and research upon
which TCEQ based its determination, nor does EPA point out any particular flaws in TCEQ's
analysis.350
In lieu of Texas's natural visibility calculations, EPA proposes instead a FIP that would use
EPA's "default" value for these components of the calculations, yet EPA "agree[s] that dust
425
-------
storms and other blown dust from deserts are a significant contributor to visibility impairment at
the Texas Class I areas that may not be captured accurately by our default method,"351 Nor
does EPA's proposal take into account natural forest fires, but instead relies on an outdated
default based on a time period of intense fire suppression.
Based on its proposal to disapprove Texas's calculation of natural visibility conditions, EPA also
proposes to disapprove Texas's URPs for these two areas.352 While EPA finds that "TCEQ . . .
correctly followed the procedures for analyzing and determining the rate of progress needed to
attain natural visibility conditions by the year 2064," because the natural visibility conditions is
an element of the calculation of the URP, EPA also proposes to disapprove Texas's URPs.353
EPA's proposed FIP would "reset the amount of natural visibility impairment for these Class I
areas" and establish different URPs.354 Yet, EPA concedes that it does "not have the information
necessary to determine how much [coarse mass and soil] should be attributable to natural
sources" at Guadalupe Mountains and Big Bend and "solicit[s] comment on the acceptability of
alternative estimates in the range between our default estimates and the Texas estimates."355
[Luminant (0061) p. 72] Luminant asserted that EPA provides no reasoned basis for
disapproving Texas's calculation of natural visibility conditions and the resulting uniform rate of
progress for Big Bend and Guadalupe Mountains. Under the regional haze regulations, "natural
conditions" "includes naturally occurring phenomena that reduce visibility," such as windblown
dust and fire.509 As part of their reasonable progress analysis, states must calculate natural
visibility conditions "by estimating the degree of visibility impairment existing under natural
conditions for the most impaired and least impaired days, based on available monitoring
information and appropriate data analysis techniques . . . ,"510
According to Luminant, in accordance with the regulations, Texas calculated natural visibility
conditions at Big Bend and Guadalupe Mountains at 10.09 deciviews and 12.26 deciviews,
respectively, for the 20% worst days.511 Texas made its determination based on a substantial
body of data and technical analysis that clearly demonstrate that haze conditions at these two
areas are heavily influenced by natural dust storms and other natural events.512 Specifically,
Texas found that haze conditions at these two areas "are heavily impacted by large long range
dust storms, that originate from dry desert and dry lake bed areas with little or no human activity,
almost all of which are situated in the Chihuahuan Desert."513 Texas shared its approach with
several of the Federal Land Managers, who concurred that Texas's approach was
"reasonable."514
Luminant stated that EPA disagrees with both Texas and the Federal Land Managers and would
instead use a default value for natural conditions, which even EPA concedes does not
"accurately" reflect natural conditions for these two Texas areas.515 Yet, EPA provides no
support for its disagreement or its proposed default value. In dismissing Texas's analysis, EPA
finds no error in Texas's analytical approach or the technical support that Texas relied on.
Indeed, EPA provides no data that would dispute Texas's approach and provides no additional
analysis or rationale whatsoever.
Luminant stated that EPA's use of the default value does not support either its disapproval of
Texas's calculation or its proposed replacement FIP. EPA's "default" value was never intended
426
-------
to be the required "norm" for all states to follow in the development of their regional haze SIP.
In its regional haze guidance, EPA explains that states need not adopt EPA's default value and
cautions that use of the default value does not "guarantee[]" that a state's analysis will be
approved.516 Rather, states "are welcome to adopt the default values for natural visibility
conditions or, with sufficient technical justification, to propose alternatives to the basic approach
or to generate refined estimates."517
[TCEQ/PUCT (0056) p. 7-8] The TCEQ contended that the natural conditions estimates that the
EPA proposes are not technically supportable and should be withdrawn. The EPA failed to
meaningfully address Texas'justification for its RPG and natural visibility condition analysis.
The TCEQ urges the EPA to approve Texas' estimation that 100% of the coarse mass and fine
soil observed at Big Bend and Guadalupe Mountains is the best estimation available.
The TCEQ stated that the EPA's proposal to use the Natural Conditions II (NCII) Committee
estimations of natural conditions for coarse mass, i.e., dust, and fine soil, ignores the site-specific
evidence and analysis presented on page 5-4 of the 2009 RH SIP. Further information and
evidence is presented clearly in the appendices and in peer-reviewed scientific publications that
are cited.7
The TCEQ stated that the technical evidence submitted in the 2009 RH SIP demonstrates that, on
the most impaired 20% of days, the suspended soil (coarse mass and fine soil) at Guadalupe
Mountains and Big Bend is best estimated by calculating that 100% of the soil is natural. The
TCEQ asks the EPA to take note of the following conclusion in Chapter 5, page 5-4, the second
paragraph of the 2009 RHSIP:
The times when human-caused dust is likely to be more important at these sites are on
days with less visibility than on the worst dust impaired days, since the most dust
impaired days are dominated by dust storms and other blowing dust from the surrounding
desert landscapes.
According to the TCEQ, in the proposal, the EPA correctly states:
We note that with any of the methodologies for calculating natural conditions discussed
above, Texas' Class I areas are not projected to meet the URP in 2018 according to the
CENRAP modeling and are not projected to meet the goal of natural visibility conditions
by 2064 (79 FR 74832).
Importantly however, the TCEQ stated that the EPA failed to note that, since over 50% of the
visibility impairment at Big Bend on the most impaired 20% days comes from outside the U.S.
and since there is no basis for projecting a reduction in that impact, the goal of reaching natural
conditions at Big Bend is unrealistic, as is the implied goal of attaining the URP at any time. A
more appropriate goal would be to achieve an appropriate reduction of the visibility impairment
caused by anthropogenic emissions in Texas and the rest of the U.S.
The TCEQ asserted that they correctly calculated natural visibility conditions at Big Bend and
Guadalupe Mountains in accordance with §51.308(d)(2)(iii) and EPA guidance. The use of a
427
-------
refined estimate is allowed under the rule and guidance. The EPA's determination that this
refined approach to estimating natural visibility conditions is "not adequately demonstrated" is
improper. Such a basis for review is not found in rule, statute or guidance. The EPA cites
"uncertainty" in the TCEQ's assumptions yet the EPA's proposed disapproval and use of the
default NCII values is contrary to the evidence presented in the 2009 RH SIP and is unjustified.
The EPA admits that dust storms and blown dust from deserts, in a very arid region, are
significant contributors to impairment in Big Bend and Guadalupe Mountains. The EPA's
preference for the default estimates is equally unjustified. It is reasonable to assume coarse mass
and dust as 100% naturally sourced for the natural visibility estimate for these areas that are
located in a desert environment and close to sources of wind-blown dust. The EPA has not
demonstrated that the TCEQ's estimate violates the rule or runs afoul of guidance, or is more
uncertain than using the default values. Just because everyone else used the default is not a valid
basis for disapproval given that the EPA's rules allow such a refined approach.
[TCEQ/PUCT (0056) p. 8-9] The TCEQ urged the EPA to accept the use of 100% natural dust
as the most reasonable estimate for calculating natural conditions. The EPA's proposal presents
no evidence that human activity contributes to the coarse mass or fine soil (dust) at Guadalupe
Mountains or Big Bend.
The TCEQ asserted that the EPA did not do what the rule requires to calculate natural conditions
"by estimating the degree of visibility impairment existing under natural conditions for the most
impaired and least impaired days, based on available monitoring information and appropriate
data analysis techniques." [See 40 CFR 51.308(d)(2)(iii)] Since the Texas 2009 RH SIP did
present substantial evidence that natural blowing dust is the cause of the coarse mass and fine
soil at both parks on the 20% of days with the most visibility impairment, the TCEQ strongly
urges the EPA to accept the use of the 100% approximation.
[CCP (0075) p. 2] CCP stated that the EPA's proposed use of default values that do not reflect
the substantial contribution of coarse matter, soils, and organic mass on natural conditions in the
affected Class I areas is arbitrary. The TCEQ regional haze SIP and the rulemaking record
support adjustments that further indicate that the rate of progress towards realistic natural
conditions is sufficient for the 2018 planning period.
[Xcel Energy (0064) p. 6] Xcel Energy asserted that the EPA's rejection of Texas' attempt to
account for the significant visibility impacts resulting from natural, wind-borne particulate matter
is arbitrary and capricious. EPA's insistence on relying on its "default" methodology for natural
conditions, which EPA acknowledges is inapplicable and inaccurate in an arid, semi-desert
region such as West Texas, is manifestly improper and illegal.
[Xcel Energy (0064) p. 21-22] Xcel Energy stated that in developing its SIP, Texas undertook
extensive, "refined" analysis of the role played by large particle particulate matter ("PM") on
visibility conditions. Texas SIP, at ES-1. Both Big Bend and Guadalupe Mountains are in desert
or semi-desert areas and are susceptible to high levels of wind-blown dust. To account for this in
its analysis and modeling, Texas assumed that all of the large particle PM was naturally
occurring dust. Texas SIP, at 11-1.
428
-------
Xcel Energy stated that the EPA proposes to disapprove Texas' calculation of natural visibility
conditions, stating that "[ajnthropogenic sources of coarse mass and fine soil in the baseline
period could have included emissions associated with paved and unpaved roads, agricultural
activity, and construction activities." 79 Fed. Reg. at 74,831. That is to say, some portion of the
dust could come from human activity, although not from stationary sources. However, EPA did
not undertake any analysis to determine whether, or to what extent, its assertion was correct.
The mere possibility that other anthropogenic emissions "could have contributed" to the dust in
these Class I areas is not a reasoned basis for rejecting TCEQ's technical assumptions. Further,
because of the extremely sparse populations near these Class I areas, the likelihood is that
virtually none, or very little, of the dust is human generated.
Rather than deferring to Texas' extensive analysis and conclusions about natural versus
manmade dust in these remote locations, Xcel Energy stated that the EPA inexplicably resorts to
a default value for the Proposal, despite acknowledging that its default methodology is
manifestly not suited to the West Texas desert. "[D]ust storms and other blown dust from deserts
are a significant contributor to visibility impairment at the Texas Class I areas that may not be
captured accurately by our default method." Id. In addition, EPA noted that the impact of dust is
less certain in Big Bend than at Guadalupe Mountains and noted that a different assumption
"may be appropriate in estimating natural conditions there." EPA, Technical Support Document
for the Texas Regional Haze State Implementation Plans, at 41 (2014) ("Texas TSD") (EPA
Docket ID EPA-R06-0AR-2014-0754-0005). Nonetheless, EPA applied its default value to
Guadalupe Mountains anyway. 7
Xcel Energy stated that the EPA should defer to Texas' analysis of naturally occurring dust in the
two Class I areas as it clearly is more representative and more accurate than EPA's default value,
which was developed for non-arid and non-desert areas. While using any of the methodologies
will not allow Texas to meet the URP in 2018 or the natural visibility goal in 2064 due to the
overwhelming impact of Mexico's sources, assuming that all dust in the two Class I areas is
naturally occurring would at least allow for realistic RPGs and a more appropriate natural
visibility target.
[NRG (0078) p. 11] NRG stated that the EPA has proposed to disapprove Texas' findings on the
natural visibility conditions for Big Bend and Guadalupe Mountains on the basis of "significant
uncertainty in the assumptions used in the Texas methodology and the demonstrated sensitivity
to the assumption of 100% natural versus 80% soil and coarse mass from natural sources." 79
Fed. Reg. at 74,831-32. Because the natural visibility conditions are a factor in the natural
visibility impairment and uniform rate of progress determinations, EPA has also proposed to
disapprove Texas' findings on those two metrics.34
Instead, NRG stated that the EPA proposes to recalculate the natural visibility conditions using a
default value. This results in a lower estimate of visibility impairment under natural visibility
conditions, which in turn tends to require faster improvements in visibility through 2064. A
comparison of the effect of EPA's proposal can be seen in Figures 2 and 3 in the attached report
by Alpine Geophysics, LLC. Alpine Report at 8-9.
429
-------
NRG disagrees with EPA's proposal to override Texas' findings on natural visibility conditions
and the associated metrics for Big Bend and the Guadalupe Mountains, for the reasons stated
below.
NRG stated that the EPA's only stated basis for disapproving the Texas findings on natural
visibility_conditions and related metrics is "uncertainty" relating to what amount of "soil and
coarse mass" detected at air monitoring sites is attributable to natural sources. 79 Fed. Reg. at
74,831. At the same time, EPA notes in one of its supporting technical analysis that:
The TCEQ has provided data that supports the conclusion that a large portion of
dust impacting visibility at its Class I areas is likely due to natural sources. We
agree that dust storms and other blown dust from deserts are a significant
contributor to visibility impairment at the Texas Class I areas that may not be
captured accurately by our default method. However, we do not believe, as the
TCEQ asserts, that all coarse mass and soil can be attributable to 100% natural
sources. FIP TSD at 32.
NRG stated that the EPA's alternative is a default metric that assumes that none of the soil and
coarse mass is attributable to natural sources. However, this default approach is unreasonable
because it conflicts with EPA's acknowledgment, quoted above, that natural sources contribute
substantially to the observed visibility impairment at Big Bend and the Guadalupe Mountains.
Further, NRG stated that the EPA has recognized that Texas already addressed the concern that
its approach might overestimate visibility impacts from natural sources, because "to the extent its
assumption that 100% of coarse mass and fine soil is natural is an overestimate, it [TCEQ]
expects that its low organic carbon estimate will more than compensate for any errors In this
assumption at this time." 79 Fed. Reg. at 74,831. EPA's method by contrast would not
appropriately compensate for natural contributions to visibility impairment.
In summary, NRG stated that the EPA has rejected a reasoned and logical approach to
calculating natural visibility conditions by Texas, substituting for it a default value that EPA
admits will have the effect of requiring Texas to improve visibility beyond natural conditions.
GCLC stated that Texas calculated natural visibility conditions for Big Bend at 10.09 dv and the
Guadalupe Mountains at 12.26 dv, for the 20% worst days. 36 Texas' determination was based on
a review of data and technical analyses for these two specific Class I areas. Rather than finding
any fault with Texas' area-specific analysis, EPA has proposed to use a default value for natural
conditions, even though EPA itself believes that this does not accurately reflect natural
conditions for these two areas. 37 Without a sound basis for this disapproval, EPA's action to
disapprove Texas' submission regarding natural visibility, and the resulting uniform rate of
progress, is an arbitrary and capricious act.
AECT stated that EPA has proposed to disapprove the natural visibility conditions in Texas' SIP
and to set more stringent natural visibility conditions, which would make Texas' Uniform Rate of
Progress glide path more stringent (i.e., require faster improvement in visibility). EPA's stated
basis for these proposed actions is its belief that Texas' approach for establishing the natural
430
-------
visibility conditions has "significant uncertainty".28 that is an insufficient basis for EPA to use to
support those proposed actions, especially since EPA has not shown that its own approach for
establishing the natural visibility conditions has greater certainty.
In addition, AECT asserted that EPA has not adequately supported its proposed position that it is
inappropriate for Texas to establish natural visibility conditions by assuming that most or all of
the soil and coarse mass identified in the monitoring data can be attributed to natural sources,
such as wildfires and dust storms, or EPA's proposed default method that does not account for
natural sources. Moreover, EPA's proposed position to not allow Texas to consider the
emissions from natural sources, such as wildfires and dust storms, in establishing natural
visibility conditions is inconsistent with its approval of the New Mexico Regional Haze SIP.
EPA concurred with New Mexico that it was appropriate to consider emissions from natural
sources, such as wildfires and dust storms, because New Mexico "has limited ability to control"
such sources of visibility impairment and they will continue to "limit the visibility improvement
[that is] achievable".29
For the foregoing reasons, AECT believes that Texas has fully supported its determination of
natural visibility conditions in its Regional Haze SIP, and AECT requests that EPA approve
those natural visibility conditions.
Footnotes:
28 79 Fed. Reg. at 74831
29 77 Fed. Reg. 70693 (Nov. 27, 2012)
36 Revisions to the State Implementation Plan (SIP) Concerning Regional Haze, TCEQ, 5-3 (Feb. 25, 2009) ("2009
Texas SIP Narrative").
37 FIP TSD at 32.
34 See 79 Fed. Reg. at 74,832-33 ("We have reviewed the TCEQ's estimates of the natural visibility impairment at
Big Bend and the Guadalupe Mountains and we propose to disapprove these estimates because this calculation
depends on the TCEQ's calculations for natural visibility conditions. ... [W]e propose to find the TCEQ has
calculate this [uniform] rate of progress on the basis of, and compared baseline visibility conditions to, a flawed
estimation of natural visibility conditions for the Big Bend and Guadalupe Mountains, as we describe above.
Therefore, we propose to disapprove the TCEQ's calculation of the URP needed to attain natural visibility
conditions by 2064.").
Commenter's References:
7 In response to Federal Land Manager comments, Texas also had calculated natural visibility conditions assuming
that 80% of coarse mass and soil should be attributed to natural sources. 79 Fed. Reg. at 74,831. EPA also fails to
address why the 80% estimate is inappropriate.
Commenter's Reference:
7 See Appendix 5-1: Discussion of the Original and Revised Interagency Monitoring of Protected Visual
Environments (IMPROVE) Algorithms; Appendix 5-2: Estimate of Natural Visibility Conditions; Appendix 5-2a:
Natural Events: Dust Storms in West Texas; Appendix 5-2b: Estimating Natural Conditions Based on Revised
IMPROVE Algorithm; Appendix 5-2c: Texas Natural Conditions SAS Program File and Data; see under References
- Gillet. al. 2005; Kavouras et. al. 2006, 2007.
Footnotes:
509 40 C.F.R. §51.301.
510 Id. § 51.308(d)(2)(iii).
511 2009 Texas SIP Narrative at 5-3.
512 See id. at app. 5-2.
431
-------
513 Id. app. 5-2 at 5.2.
514 See U.S. Fish & Wildlife Serv. & Nat'l Park Serv., Comments on Texas Proposed Regional Haze Rule State
Implementation Plan 2, 3 (Jan. 11, 2008).
sis FIp TSD at 32
516 EPA, Guidance for Estimating Natural Visibility Conditions Under the Regional Haze Program 1.2 to 1.3 (Sept.
2003).
517 Id. at 1.2 n.2.
Footnotes:
346 79 Fed. Reg. at 74,832.
347 Id. at 74,831.
348 See U.S. Fish & Wildlife Serv. & Nat'l Park Serv., Comments on Texas Proposed Regional Haze Rule State
Implementation Plan 2, 3 (Jan. 11, 2008), available at http://tinyurl.com/FLMLetter.
349 79 Fed. Reg. at 74,831.
350 See generally 2009 Texas SIP Narrative app. 5.2.
351 79 Fed. Reg. at 74,831 (emphasis added).
352 Id. at 74,833.
353 Id.
354 Id. at 74,885.
355 Id.
Response: We agree with commenters that the Regional Haze Rule and our guidance336 do
allow states to develop an alternate approach to estimate natural visibility conditions. The fact
that the states have the option of calculating their own natural visibility conditions is not at issue.
However, that approach must be fully supported and documented. As we state in our guidance,
states are "free to develop alternative approaches that will provide natural visibility conditions
estimates that are technically and scientifically supportable. Any refined approach should be
based on accurate, complete, and unbiased information and should be developed using a high
degree of scientific rigor."337 Consistent with what was stated in our proposal, we find the
TCEQ did not provide a technically and scientifically supportable approach, specifically by not
adequately supporting the assumptions used in calculating "refined" estimates of natural
visibility conditions. We disagree with Luminant's statement (and other similar statements made
by commenters), that "[EPA] does not address or refute the substantial body of technical data
and research upon which TCEQ based its determination, nor does EPA point out any particular
flaws in TCEQ's analysis." The TCEQ's analysis and our own observations in our proposal do
support a conclusion that much of the contribution of coarse mass and fine soil to the visibility
impairment at the Guadalupe Mountains and Big Bend is due to natural sources. They do not
demonstrate that 100% of this contribution is due to natural sources.
First, we disagree that Texas's natural visibility calculations can be viewed as a "substantial
body of technical data and research." As we discuss below, the Texas natural visibility
calculations depend on a key assumption that Texas itself concludes that it cannot verify: that
100% of the coarse mass and fine soil contribution to the natural visibility impairment at Big
Bend and the Guadalupe Mountains is due to natural sources. The TCEQ points to "the site-
specific evidence and analysis presented on page 5-4 of the 2009 RH SIP ... [and] information
and evidence is presented clearly in the appendices and in peer-reviewed scientific publications
336 Guidance for Estimating Natural Visibility Conditions Under the Regional Haze Rule, EPA, September 2003.
337 Idatp 1-11.
432
-------
that are cited." However, these references do not scientifically validate the absolute assumption
of 100% adopted by the TCEQ.
We also disagree with Luminant that we did not refute Texas' natural visibility calculation. In
our proposed disapproval, we noted significant uncertainty in the estimates developed by Texas
and the demonstrated sensitivity of the estimates to this assumption. We also noted that a
different assumption for Big Bend was appropriate because the observed impacts from dust are
different from Guadalupe Mountains.338 In our proposal, we specifically solicited comments on
the acceptability of alternate estimates in the range between the EPA default estimates and
Texas' estimates.
AECT is incorrect that we did not allow Texas to consider the emissions from natural sources,
such as wildfires and dust storms, in establishing natural visibility conditions. As we discuss
below, both the Texas natural visibility calculation and the new IMPROVE calculation we
employed contain parameters that address these types of natural sources. Our disagreement with
the position relied upon in the Texas SIP is mainly with the undocumented assertion that all
coarse mass and fine soil is natural.
Luminant states that the FLMs concurred "that the basic approach used [by Texas] to adjust
natural conditions is reasonable, provided that the Proposed SIP address the uncertainty of the
assumption that all of the coarse mass and fine soil fraction on the worst 20 percent days is
natural." However, this quote is not fully reproduced, leading to an inaccurate characterization of
the comment from the FLMs. The full quote is:
We have reviewed Appendix 5.2 and find that the basic approach used to adjust
natural conditions is reasonable, provided that the Proposed SIP address the
uncertainty of the assumption that all of the coarse mass and fine soil fraction on
the worst 20 percent days is natural. Since there is human activity in the
region, the State should provide a rationale for what fraction of soil and
coarse mass is natural, and present an alternative where that fraction of the
coarse mass and fine soil concentrations are assumed to be natural within the
SIP narrative. In addition, we request that the SIP narrative include the
default EPA predictions of natural conditions so that the reader can better
understand the scope of changes Texas has chosen to make and can judge the
effect of the State's choice on efforts to assess reasonable progress later in the
SIP. Default values of natural conditions should be included in Table 5-2. It
would also help the reader to summarize how the refinement affects the revised
natural condition if the State included a chart showing the breakdown of each
basic pollutant component. This would give a non-technical reader a simple
reference about which components in the haze calculation were changed and by
how much339 [emphasis added].
338 The "refined" approach that the TCEQ took for Big Bend is also questionable because the state's claimed
reliance on "site-specific analysis and evidence" principally connects to Guadalupe Mountains conditions but it was
uncritically adopted for Big Bend, as well.
339 U.S. Fish & Wildlife Serv. & Nat'l Park Serv., Comments on Texas Proposed Regional Haze Rule State
433
-------
In response to this comment from the FLMs, the TCEQ held additional discussions with the
FLMs concerning the assumptions in their natural conditions estimate. Like us, the FLMs did
not agree with the assumption that 100% of the coarse mass and soil was natural, and stated that
there is human activity in the region. The FLMs "suggested that the commission could
judiciously use 80 percent as the natural source of coarse and fine dust and 20 percent of coarse
and fine dust due to human activity."340 Thus, Luminant's statement that, "the EPA disagrees
with both Texas and the Federal Land Managers" mischaracterizes the situation: Both we and
the FLMs concluded that Texas did not substantiate its assumption that 100% of the coarse mass
and soil contribution to the natural visibility conditions at the Texas Class I areas is natural.
In fact, even the TCEQ acknowledged that the information they cite to in their SIP does not
quantify the percentage of anthropogenic or natural contributions to total coarse mass and fine
dust, and that some portion must be from human activity, as demonstrated by the following quote
from the Texas Regional Haze SIP:341
Unfortunately, while the Gill et al., and Kavouras et al., works suggest that the
relative contribution of agricultural and/or other anthropogenic activities to
suspended dust are only a very small portion of the source points/regions, they are
not able to quantify the actual percentage of source load, let alone receptor
concentrations. However, while some dust (CM and Soil) at both of Texas' Class
I areas must be from some human activity, the times when human caused dust is
likely to be more important at these sites are on days with less visibility
impairment than on the worst dust impaired days, since the most dust impaired
days are dominated by dust storms and other blowing dust from the surrounding
desert landscape. So the TCEQ has chosen, for the sake of the most and least
impaired natural visibility estimates, to treat 100 percent of the CM and Soil
concentrations measured at each of its Class I areas as natural.
However, to the extent to which the 100 percent natural CM and Soil estimate is
an over estimate, the TCEQ expects this low OC estimate to more than
compensate, at this time.
We therefore disagree with Luminant and others and reaffirm our proposed position that Texas
has not demonstrated that 100% of the coarse mass and fine soil contributions to the natural
conditions for Big Bend and the Guadalupe Mountains are due to natural conditions.
We agree with Luminant that in lieu of Texas's natural visibility calculations, we proposed a FIP
that relied on a modified version of our "default" natural visibility conditions calculations.342
These calculations are derived from the guidance we provided to the states,343 which included
Implementation Plan 2, 3 (Jan. 11, 2008), emphasis from NPS stating: "The comments which are highlighted in
bold face discuss what we consider major shortcomings of the proposed SIP that we believe warrant additional
consultation prior to final adoption of the Texas Regional Haze Plan. "
340 Appendix 2-2 of the Texas Regional Haze SIP.
341 Appendix 5-2 of the Texas Regional Haze SIP.
342 See the discussion on this in our proposal, beginning on 79 FR 74830.
343 Guidance for Estimating Natural Visibility Conditions Under the Regional Haze Rule, EPA-454/B-03-005,
September 2003.
434
-------
natural visibility conditions based on the IMPROVE equation. This guidance allows states to
use a "refined" approach or alternative approaches to the guidance defaults to estimate the
values that characterize the natural visibility conditions of their Class I areas. Another option
that we noted in our proposal that was open to the states, and the one we used in proposing the
natural conditions for the Texas Class I areas in our FIP, was the "new IMPROVE equation" that
was adopted for use by the IMPROVE Steering Committee in December 2005. This refined
version of the IMPROVE equation provided more accurate estimates of some of the factors that
affect the calculation of light extinction. NRG incorrectly characterizes this equation as
"assuming] that none of the soil and coarse mass is attributable to natural sources." The new
IMPROVE equation which we used in our FIP most certainly does contain parameters for coarse
mass and fine soil, as indicated by Appendix 5-1 of the Texas SIP.
The TCEQ used this refined version of the IMPROVE equation, but altered the parameters
concerning the concentrations of coarse mass and fine soil, without adequate documentation.
We found that the TCEQ's documentation was flawed, but we are under no obligation to follow
in the TCEQ's footsteps and make whole its methodology, when we had already provided
guidance with default natural visibility conditions, which were further refined by the 2005
IMPROVE Steering Committee. We disagree with Xcel that we owe the TCEQ any deference
other than to have provided it with an opportunity to adequately document its own natural
visibility calculations. Instead, the burden was on the TCEQ to demonstrate its natural visibility
conditions were more appropriate, which it failed to do. As we note in our proposal, the TCEQ
stated within its SIP that it will continue to evaluate data, modeling, and any other sources of
information in order to further improve its estimates. Furthermore, the TCEQ plans to work with
us and the federal land managers to improve natural conditions estimates for future regional haze
SIP revisions. We encourage these efforts, but for present purposes, we are unable to approve
the calculation of natural visibility conditions based on the technically indefensible assumption
that there is 0% dust (CM and soil) from human activity when the state rightly concedes that
some impairment "must be from some human activity."
Even as we are disapproving Texas' natural conditions estimates, we conclude that our
determinations for emissions limitations for EGUs in the FIP for the first planning period would
be justified on the basis of natural conditions estimates at either levels in the SIP or the levels in
the FIP, given the level of visibility impairment at each Class I area above the different estimates
for natural conditions and the availability of cost-effective controls at those sources with the
largest visibility impacts that result in meaningful progress towards the natural visibility goal.
Furthermore, as we noted in our proposal, based on both our recalculated natural conditions and
the Texas natural condition estimates that we are disapproving, Texas' Class I areas are not
projected to meet the uniform rate of progress in 2018 according to the CENRAP modeling and
are not projected to meet the goal of natural visibility conditions by 2064.344
We address the impacts of international emissions in our responses to more detailed comments
on that issue.
344 79 FR 74832
435
-------
Comment: Adjustments to Default Natural Haze Conditions [Luminant (0061) p. 73,
AECOM (0061/0075) p. 3-1], [AECOM (0061/0075) p. 1-5]
AECOM stated that the EPA's use of default values for coarse matter, soils, and organic mass as
its basis for disapproving Texas' calculation of natural conditions is not justified. Adjustments
to these values and to the natural conditions end point in the establishment of URPs are justified,
as EPA recognizes in the proposal. While Texas' approach has merit, in response to EPA's
request, we have calculated an alternative adjustment to reflect natural conditions at GUMO and
B1BE. These adjustments further indicate that the rate of progress toward realistic natural
conditions is sufficient for the 2018 URP interim goals at GUMO and BIBE, without the
additional SO2 controls that EPA is proposing.
AECOM noted that in order to calculate a URP, it is necessary to determine natural visibility
conditions (i.e., visibility as it would be without any anthropogenic impacts). To do so, EPA has
defined natural haze conditions for annual average, 20% best, and 20% worst days.53 These
estimates have been updated56 with the use of a revised IMPROVE equation,57 which Texas
adopted for its regional haze SIP.
Luminant and AECOM noted that in developing its SIP, Texas found that some of the haziest
days at its two Class I areas (BIBE and GUMO) are the result of uncontrollable natural
conditions such as windblown dust and wildfire emissions. These events clearly show
themselves in haze composition plots, which show the constituents of visibility impairment. "'8.
For example, Exhibit 1 provided by AECOM (Figure 3-1 in 0061/0075) and Luminant (Figure 1
in 0061) is a haze composition plot for 2003 at Big Bend. Some of the 20% worst haze days are
labeled with "W" and circled in red. These days show predominant percentages of haze due to
four types of naturally-occurring haze-forming particles: organic mass (carbon) ("OMC"), coarse
matter (large dust particles) ("CM"), "soils" (fine dust), and elemental carbon (e.g., soot from
wildfires).
Exhibit 1. Haze Composition Plot for Big Bend (2003) provided by AECOM and Luminant
BIBE1
I Area - Big Bend NP, TX
I
A VA-1,.'- Nrt„| s 'lih %h
¦_
Jan *03 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
| arrmW03t_bext ammS041_bext CMJjext
|OMCf bexl SeaSa* bext: ¦SOU bext:
IECIbext
436
-------
Luminant asserted that Texas's approach correctly accounts for these natural conditions. EPA,
on the other hand, presents no data or analysis to dispute Texas's findings, nor does it find any
error in Texas's approach. Texas did exactly what it was required to do under the regulations—it
"estimate[ed] the degree of visibility impairment existing under natural conditions for the most
impaired and least impaired days, based on available monitoring information and appropriate
data analysis techniques."519 Nothing in the regulations requires Texas to adopt, or even
consider, EPA's "default" value. Indeed, the default value that EPA applies is out-of-date and
underestimates natural haze levels because it is based on a period of intense natural fire
suppression (pre-1988) that does not occur today.520 Thus, EPA's proposed disapproval—the
only basis for which is EPA's bare reference to an out-of-date "default" value—is both contrary
to the regulations and arbitrary and capricious.
Luminant stated that one thing is crystal clear—the default values that EPA applies in its
proposed FIP are unreasonable and unsupportable. In issuing its FIP, "EPA steps into the State's
shoes, and must meet the same requirements" as the state.521 Here, EPA has not followed the
regulations in estimating natural conditions but has simply applied the "default" with no
justification or support for doing so. And the record is undisputed that natural conditions at Big
Bend and Guadalupe Mountains are not accurately reflected in the "default" value. Not only
does EPA's default value understate natural fire levels as discussed above, EPA "agree[s]" with
Texas that "dust storms and other blown dust from deserts are a significant contributor to
visibility impairment at the Texas Class I areas," and EPA concedes those natural conditions
"may not be captured accurately by our default method."522
Recognizing that its default value is not adequate, Luminant noted that the EPA "solicit[s]
comment on the acceptability of alternate estimates in the range between the EPA default
estimates and Texas' estimates."523 We believe that Texas's estimates are fully supported based
on the record before EPA and must be approved. We also asked AECOM to review EPA's
proposal and the available data and to determine if it was possible to develop an alternative
natural conditions value for each of Texas's two Class I areas that, consistent with EPA's
regulations, reflects adjustments to days that are dominated by natural events. AECOM's
methodology and rationale are discussed in its full report, which is attached to and incorporated
in full into these comments. AECOM's approach adopts what Luminant asserts to be EPA's
recent "conclusion" that "removing (or possibly adjusting) days that are dominated by natural
events is consistent with the existing Regional Haze Rule (RHR)."524
According to Luminant, the results of AECOM's analysis are alternative values for natural
conditions at Big Bend and Guadalupe Mountains that, unlike the default value EPA has
proposed, take into account natural events and are an additional estimate of natural conditions.
The resulting values—9.25 dv for Big Bend and 12.12 dv for Guadalupe Mountains—fall in
between Texas's approach and EPA's default value. Applying these refined values for natural
conditions results in a 2018 URP for Big Bend of 15.42 and a 2018 URP for Guadalupe
Mountains of 16.01, which more accurately reflect natural conditions than EPA's default value.
Based on the most recent five-year observed conditions, both areas will meet their refined URP
in 2018, without any further controls.
437
-------
Luminant stated that AECOM's analysis thus further confirms Texas's determination that its SIP
revision results in reasonable progress by 2018 at these two Class I areas. AECOM's analysis
also confirms EPA's conclusion that the default value do not "accurately" capture the natural
conditions of these two areas.525 Luminant concluded that the EPA's reliance on the default
value is thus arbitrary and capricious and cannot form the basis for any action by EPA on
Texas's submission.
[AECOM 0061/0075 p. 3-2] AECOM stated that for several of the 20% worst days, it is clear
from looking over several years of composition plots for BIBE and GUMO that there are days
that are dominated by the effects of naturally-caused haze, which is underrepresented in EPA's
default estimates of natural conditions. The naturally-caused haze can be caused by wildfire
emissions as well as natural windblown dust that may be characterized as exceptional events.
AECOM stated that this condition is well-known and documented for these areas. To paraphrase
from the Forest History Society, throughout history except for recent decades, fire has been used
to clear land, vary plant and tree species, and maintain habitat, among other purposes.59 Native
Americans used fire as a method to maintain land or to improve habitats. Although early settlers
often used fire in the same way as the Native Americans, major wildfires on public lands were
largely ignored and were frequently seen as an opportunity to open forestland for grazing.60
AECOM stated that many large fires erupted in North America in the 1800's to early 1900's and
the life-threatening impact of these fires was slowly becoming apparent to the public. In
response, the federal government began attempting to control forest fires in the 1890's by
employing General Land Office rangers during the fire season. In 1905, the forest reserves
(presently, the national forests) were transferred to the Forest Service. This agency became an
authority for forming professional standards for firefighting, including hiring more rangers and
locals to help with their efforts.61 Since the beginning of the 20th century, fire suppression has
resulted in a buildup of vegetative "fuels" and catastrophic wildfires. Recent estimates of
background visual range, such as Trijonis62 (1990) have underestimated the role of managed fire
on regional haze. Since about 1990, various government agencies have increased prescribed
burning to reduce the threat of dangerous wildfires, and the increased haze due to these fires is
sometimes more of an impairment to visibility than industrial sources.
AECOM noted that the National Park Service has explained that, "[f]or most of the 20th
Century, wildfires were extinguished immediately with the assumption that doing so would
protect lives, property, and natural areas. However, following the unusually intense fire season
of 1988, agencies including the National Park Service began to rethink their policies."63 The data
examined by Trijonis and adopted by EPA in their natural conditions default values precedes this
change in thinking and underestimates natural haze due to wildfires, which causes elevated
emissions of organic matter, dust, and soot. The frequency of wildfires in the western U.S. is
shown in Figure 3-2 in AECOM comment 0061/0075.64 Wildfire frequency was generally quite
low prior to 1988 while frequency substantially increased following 1988.
AECOM noted that the default natural background assumed by EPA in their 2003 guidance
document65 and the corresponding updated default natural background using the revised
IMPROVE algorithm are not realistic, and are unattainable due to these natural events. One
438
-------
important aspect of the uncontrollable haze, wildfires, is affected by the biased quantification of
its contribution to natural haze due to suppression of wildfires during the 20th century. In
addition, localized sources of naturally-caused windblown dust are another area underestimated
by EPA for these Class I areas.
According to AECOM, the definition of natural conditions that can be reasonably attained for a
reasonable application of EPA's regional haze rule should be revised for Texas and other states
affected by naturally-caused haze. In fact, EPA's 2003 "Guidance for Estimating Natural
Visibility Conditions Under the Regional Haze Program" acknowledges that wildfires are a
contributor to natural visibility conditions, but the data used in estimates of natural conditions
were taken during a period of artificial fire suppression so that the true impact of natural
wildfires is understated. The report notes that "data should be available for EPA and States to
develop improved estimates of the contribution of fire emissions to natural visibility conditions
in mandatory Federal Class I areas over time."
Moreover, AECOM stated that achieving true natural conditions will require the elimination of
all anthropogenic sources of emissions, including those from other countries, such as Mexico,
where the U.S. has no jurisdiction. Estimating the number of years needed to attain natural
conditions requires accurate predictions about future energy sources, technology improvements
for emission sources, and every aspect of human behavior that causes visibility impairing
emissions in the U.S. and other countries.66
AECOM stated that since Texas cannot control emissions from Mexico, these sources of haze
must be factored into the end point goal for these two Class I areas. Doing so demonstrates
further that additional reductions from Texas sources leading up the 2018 interim goal would
have no measureable effect on visibility conditions, and EPA's proposal focuses upon a very
minor component of haze while other much more important sources of haze are not addressed by
EPA.
AECOM stated that the Federal Land Managers (FLM) support the development of refined
estimates of natural conditions for the Texas Class I areas. Comments provided by the National
Park Service on the Texas 2014 Five-Year Regional Haze SIP Revision stated, "Hopefully EPA
will be working with states and FLM on improving estimates of natural conditions for the 2018
SIPs." 67 Nevertheless, EPA proposes to reject the efforts of Texas to provide more realistic
estimates of natural conditions by providing an adjustment that accounts for naturally-caused
haze because of a concern they have that the state did not provide sufficient technical support for
the revised methodology used in the development of the SIP. Even so, EPA recognizes that it
has no basis for imposing default values, as it proposes to do. EPA "agree[s] [with Texas] that
dust storms and other blown dust from deserts are a significant contributor to visibility
impairment at the Texas Class I areas that may not be captured accurately by our default
method."68 EPA has thus solicited comment on this issue, and we provide here a more reasonable
alternative method to EPA's default value for quantifying the effects of natural haze due to
wildfires and windblown dust that accounts for days of predominant influences of naturally-
caused dust. The alternative refined approach (TX ALT NC) described below is conservative
(that is, it likely understates the adjustment needed) in that it does not account for adjustments
due to international emissions of sulfates and nitrates, nor days with somewhat modest
439
-------
components of naturally-caused haze. The adjusted particulate species for this alternative
approach are limited to OMC, CM, and soils.
The method applied by AECOM here was to review the visibility extinction composition of each
of the 20% worst days during the period of 2004-2013 (10 years) with IMPROVE data at the
Texas Class I areas. Aerosol extinction composition over the 10-year period for BIBE and
GUMO are shown in Figures 3-3 and 3-4 in AECOM comment 0061/0075. During 2011-2013,
there is a greater fractional quantity of CM, OMC, and soil extinction. Drought conditions in
Texas and other surrounding states likely caused greater impacts from windblown dust and
wildfires during this period. Also, over time, the sulfate contribution to total extinction at BIBE
and GUMO has steadily decreased, as expected from the emission trends noted above.
To develop more realistic estimates of natural conditions than assumed by EPA, a daily threshold
percentage of total aerosol extinction caused by CM, OMC, and soil species was determined by
AECOM for each Texas Class I area. This threshold was developed by examining histograms of
the 20% worst days for a noticeable step-up in frequency of higher contributions of CM, OMC,
and soil (i.e., from right to left). As shown in Figure 3-5, the resulting frequency thresholds of
the total aerosol extinction caused by CM, OMC, and soil species for days of exceptional natural
cases can be assigned as 40% for both BIBE and GUMO.
AECOM noted that in developing its SIP, TCEQ found that "while some dust (CM and Soil) at
both Texas' Class I areas must be from some human activity, the times when human caused dust
is likely to be more important at these sites are on days with less visibility impairment than on
the worst dust impaired days, since the most dust impaired days are dominated by dust storms
and other blowing dust from the surrounding desert landscape."69 The frequency threshold
percentages developed above can be used to identify those days where haze was caused by CM,
OMC, and soil species from natural causes, and thus those conditions can be properly attributed
to natural conditions.
Further, these frequency thresholds can be used to develop alternative approaches for
establishing natural haze conditions at these two areas. The fraction of the OMC, CM, and soils
aerosol extinction for the specific days above the frequency threshold, relative to that for all of
the 20% worst days, was computed as a reasonable estimate of the uncontrollable fraction for
these particulate species, due to natural causes. These uncontrollable fraction estimates are
presented in Table 3-1 provided by AECOM (0061/0075).
440
-------
fail# S-1: Uneoritfolabto Fraction of Baaattna Extinction for Cllt CMC, and &
Uncontrollable Fractions of Baseline Extinction
for Throe Spc cios
i filter
i r.rv^r *1»
CM
OMC
"«iS
j
:.4W
UU?V!L»
>
< tv4
0.878
These estimates of the uncontrollable extinction were then applied (added) to EPA's default 20%
worst days natural conditions extinction estimate to result in a more realistic end point for natural
conditions for BIBE and GUMO than assumed by EPA (i.e., the NCII default). The species'
extinctions were then summed for estimating the 5-year total extinction and converted to
deciviews. Table 3-2 provided by AECOM (0061/0075) compares our refined estimate of
natural conditions, as compared to the default value proposed by EPA and Texas' assumption of
100% (TX NC). The alternative method developed here (TX ALT NC), similar to TX NC,
assumed 100% of CM, OMC, and soil species were natural on specific days. However, this
assumption was applied only for those days where these species' combined extinction was equal
to or greater than the frequency threshold percentages indicated above for the Texas Class I areas
of the aerosol extinction to develop the uncontrollable fraction. Given the overwhelming amount
of data and information indicating substantial impact from natural causes on the worst days, this
is a more reasonable approach to estimating natural conditions than using the default values as
EPA proposes. Our estimates are also likely conservative because they do not make any
adjustments for days that are not dominated by naturally-caused haze.
Table 3-2
Ntfumt Concflions (
-------
and Figures 3-6 and 3-7 from AECOM 0061/0075) show the resulting uniform glide paths for
the two Class I areas using the TX ALT NC results for natural conditions. (Figure 4 from
Luminant 0061 and Figures 3-8 from AECOM 0061/0075) is an alternate figure shown for BIBE
without the anomalous 2011 information included in the regression line. With the TX ALT NC
values, the new URPs for 2018 become 16.01 dv for GUMO and 15.42 dv for BIBE. Using
these adjusted natural condition end points and the associated glide paths, it is apparent that the
2018 URP goals will be met for GUMO based on the current IMPROVE measurements' linear
regression. BIBE will also meet the revised glide path, and this does not factor in uncontrollable
international emissions from Mexico (more discussion on this issue is provided below). Thus,
the weight of evidence indicates that with appropriate adjustments and based on the current
visibility data, the 2018 URPs will be met for all three Class I areas under consideration in
EPA's proposed rule.
GUMO Glide Path with Alternative Natural Conditions
(Luminant (0061) Figure 2 and AECOM (0061/0075) Figure 3-6)
Refined Uniform Rate of Progress for GUMO
20
18
16
14
5
12
u
10
Ol
O
8
6
4
2
0
. ~ 17.2
f
~
~
~
— » - X
^ 16.0
15.3
0
^ ~
14.4
•
TX ALT NC
~
Observations 2000-13
X
5-Year Obs. 2000-2004, Obs. 2009-13
Linear (TX ALT NC)
Linear (Observations 2000-13)
i i i i i
I
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
BIBE Glide Path with Alternative Natural Conditions
(Luminant (0061) Figure 3 and AECOM (0061/0075) Figure 3-7)
442
-------
BIBE Glide Path with Alternative Natural Conditions
(TX ALT NC) Excluding 2011 Outliers
(Luminant (0061) Figure 4 and AECOM (0061/0075) Figure 3-8)
20
IB -I »
Refined Uniform Rate of Progress for BIBE
Excluding 2011 Outliers
¦v~ -V— * 4 16.3
~ Y 1542
16 -- ~ "" t ——
14
3 12
m
> 10
V • TX ALT NC
Q o
14.8
~ Obs. 2001-13 No 2011 Outliers
X 5-Year Obs. 2001-04, 2009-13
•Linear (TXALT NC)
Linear (Obs. 2001-13 No 2011 Outliers)
1 i 1 1 1 1 i 1 1 1
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
Luminant Footnotes:
443
-------
519 40 C.F.R. § 51.308(d)(2)(ii).
520 AECOM Report at 3-3 to 3-4.
521 77 Fed. Reg. at 40,164.
522 FIP TSD at 32.
523 TX SIP TSD at 42.
524 EPA, Pre-Meeting Materials for the EPA-FLM-RPO-States-Tribes Meeting on the Future of the Regional Haze
Program, Topic III: The URP and RPG Framework 2 (Feb. 3, 2015).
525 FIP TSD at 32.
AECOM Footnotes:
55 http://www.epa.gov/ttn/oarpg/tl/memoranda/rh_envcurhr_gd.pdf.
56 http://www.nature.nps.gov/air/Pubs/pdf/flag/FLAG_2010.pdf.
57http://vista. cira.colostate.edu/improve/Publications/GrayLit/016_IMPROVEeqReview/IMPRO VEeqReview.htm.
This equation relates measured concentrations of various particulate species to visibility extinction, which is a
measure of the particle loading per unit length of viewing that causes visibility impairment, or haze.
58 http://views.cira.colostate.edu/web/Composition/.
59 http://www.foresthistory.org/ASPNET/Publications/first_century/sec3.htm.
60 http://www.foresthistorv.org/ASPNET/Publications/first centurv/sec3.htm.
61 http://www.foresthistory.org/ASPNET/Publications/first_century/sec3.htm.
62 Trijonis, J. C, 1990. Characterization of Natural Background Aerosol Concentrations. Appendix A in Acidic
Deposition: State of Science and Technology. Report 24. Visibility: Existing and Historical Conditions - Causes
and Effects. J. C. Trijonis, lead author. National Acid Precipitation Assessment Program: Washington, DC.
63 http://www.nps.gov/thro/parkmgmt/firemanagement.htm.
64 Westerling, A. L., H. G. Hidalgo, D. R. Cayan, andT. W. Swetnam, 2006. Warming and Earlier Spring
Increase
Western U.S. Forest Wildfire Activity. Science, 313, 940-943.
65 http://www.epa.gov/ttn/oarpg/tl/memoranda/rh envcurhr gd.pdf
66 https://www.ndhealth.gov/AQ/RegionalHaze/Regional%20Haze%20Link%20Documents/Main%20SIP%20Secti
ons%20 l-12.pdf at 196-197.
67 http://www.tceq.texas.gov/assets/public/implementation/air/sip/haze/14aAppB_rtc.pdf at 1.
68 79 Fed. Reg. 74,831 (emphasis added).
69 http://www.tcea.texas.gov/assets/public/implementation/air/sip/haze/App5 2.pdf at 4.
70 79 Fed. Reg. 74,831.
71 79 Fed. Reg. 74,831.
Response: We disagree with Luminant that we did not find any error in Texas' approach to its
calculation of the natural visibility calculations for its Class I areas, as is detailed in our
proposal345 and in our response to other comments. We agree with Luminant that Texas was not
required to adopt our default values for natural visibility. However, as we address in our
responses to other comments, Texas did have an obligation to properly document the
assumptions it made in calculating its own natural visibility conditions, and it did not do so. We
found that a key aspect of Texas' calculation—that it assumed 100% of the coarse mass and fine
soil visibility impairment at its Class I areas was due to natural sources—was not supported. We
further found evidence (that Luminant's own analyst has confirmed in this comment) that there
are legitimate reasons to conclude that different assumptions regarding the coarse mass and fine
soil contributions to visibility impairment are appropriate for Big Bend and the Guadalupe
Mountains. Further, after having found that the TCEQ's documentation was flawed, we were
under no obligation to follow in the TCEQ's footsteps and make whole its methodology, when
we had already provided guidance with default natural visibility conditions, which were further
345 See discussion beginning on 79 FR 74830.
444
-------
refined by the 2005 IMPROVE Steering Committee. On the contrary, the burden to substantiate
and support the assumption was on the TCEQ.
Luminant also incorrectly ties our solicitation of comments on alternative estimates of natural
visibility calculations to a recognition that our default values were inadequate. Our natural
visibility calculations were based on estimates of natural levels of visibility-impairing pollutants
and the IMPROVE equation present in our Natural Visibility Guidance,346 as modified by the
"new IMPROVE equation" that was adopted for use by the IMPROVE Steering Committee in
December 2005. This refined version of the IMPROVE equation provided more accurate
estimates of some of the factors that affect the calculation of light extinction. These estimates,
while they may stand further refinement, were fully vetted and adopted by all of the states except
for Texas. Thus, we have confidence that it represents good science and is adequate for the
purposes for which it was intended. As we indicate in our Reasonable Progress Guidance, Texas
was free to base its natural visibility calculations on an alternative methodology, but in doing so
it carried with it the burden of supporting and substantiating its assumptions, which was not met.
We agree with Luminant that we stated that, "dust storms and other blown dust from deserts are
a significant contributor to visibility impairment at the Texas Class I areas," which "may not be
captured accurately by our default method." However, Luminant fails to note in its comment
that we followed that statement with the following:
However, we do not believe, as the TCEQ asserts, that all coarse mass and soil can be
attributable to 100% natural sources. Although we believe that some coarse mass and
soil should be attributable to natural sources, we do not have the information necessary to
determine how much should be attributable to natural sources. We therefore
acknowledge that like the TCEQ, we cannot accurately reset the natural conditions for the
Guadalupe Mountains and Big Bend by using the TCEQ's methodology, which depends
on this information. In lieu of this, we proposed to rely on the adjusted default estimates
for the new IMPROVE equation from the Natural Conditions II committee,347 which was
the starting point for the Texas natural visibility calculations, but solicited comments on
the acceptability of alternate estimates in the range between our default estimates and the
Texas estimates. Again, we are under no obligation to accept the TCEQ's methodology
and "fill in the blanks" in its flawed calculation of natural visibility conditions.
Luminant's contractor, AECOM, analyzed monitoring data from 2001 to 2013 for Big Bend and
Guadalupe Mountains in an attempt to establish alternative natural conditions for Big Bend
(BIBE) and the Guadalupe Mountains (GUMO). Luminant states that AECOM's approach
adopts our recent conclusion that "removing (or possibly adjusting) days that are dominated by
natural events is consistent with the existing Regional Haze Rule (RHR)." Here, Luminant
346 Guidance for Estimating Natural Visibility Conditions Under the Regional Haze Rule, EPA-454/B-03-005,
September 2003.
347 Regional Haze Rule Natural Level Estimates Using the Revised IMPROVE Aerosol Reconstructed Light
Extinction Algorithm, Copeland, S. A., et al, Final Paper # 48, available in our docket.; NCII, or new IMPROVE
natural visibility conditions are available at: http://vista.cira.colostate.edu/Docs/
IMPROVE/Aerosol/NaturalConditions/NaturalConditionsII_Format2_v2.xls, for which we have filtered the data
for Texas Class I areas and which is also available in our docket.
445
-------
references a document348 we distributed to the states, tribes, FLMs, and RPOs that served to
establish a framework of issues that we were considering for inclusion within an update to
Regional Haze Rule Guidance. Luminant is correct that we have been considering methods for
improving the calculation of the natural visibility conditions. These include "removing (or
possibly adjusting) days that are dominated by natural events is consistent with the existing
Regional Haze Rule (RHR)." However, the mere fact that we solicited input on these issues does
not mean that the method the TCEQ and Luminant (AECOM) chose to address them is correct or
acceptable. In fact, we believe AECOM's effort is flawed in a number of key areas which we
outline below.
Assumptions Underpinning AECOM's Methodology
We agree with AECOM that fire has a natural role in the ecosystem and that historical fire
suppression has resulted in a buildup of vegetative fuels that has likely contributed to
catastrophic forest fires. AECOM states that this rethinking of the role of natural fire influenced
the FLMs and after 1988, wildfire frequency increased. AECOM then asserts that the Trijonis
estimates that form the basis of our original natural conditions estimates predate this change in
fire management strategy and therefore the contribution of fire is biased low in our default
natural visibility calculations. However, AECOM does not present any documentation for this
assertion. While AECOM presents information on increased frequency of wildfires, they do not
discuss the intensity of the fires, the area burned, fuel loading or estimated emissions and
impacts from these fires. In addition, there is considerable uncertainty in what prescribed fires
are considered natural. Furthermore, while fire management strategies may have changed in the
United States, AECOM does not address the impact from international wildfires. In addition,
as we state above, our natural visibility calculations used the revised IMPROVE equation from
the Natural Conditions II committee, which, was developed in 2005. Natural conditions
calculated by the NC II committee using the revised IMPROVE equation was fitted to aerosol
data from 2000-2004. Thus, we disagree that AECOM has adequately linked these facts to an
underrepresentation of fire in our default natural conditions calculation. Similarly, AECOM
merely presents the statement, "In addition, localized sources of naturally-caused windblown
dust are another area underestimated by EPA for these Class I areas," but provides no
documentation to support it. Further, we agree with AECOM that Texas should not have to
compensate for the impact of international emissions from Mexico, and we do not ask Texas to
do so. Luminant appears to summarize AECOM's report by concluding that windblown dust,
fire, and emissions from Mexico constitute for so much of the visibility impact at its Class I
areas, that Texas should be relieved of its responsibility to control its own sources of visibility
impairing pollution. As we have detailed in other response to comments, our FIP is limited to
controlling emission sources in Texas that impact visibility at Texas' two Class I areas and the
Wichita Mountains.
Although AECOM restricts its assumption to specific days, it nevertheless assumes that all
Coarse Mass (CM), Organic Mass Carbon (OMC), and soil visibility impacts at BIBE and
GUMO are 100% due to natural causes. AECOM provides no documentation to support this
348 Guidance Topics Topic III: The URP and RPG Framework. DRAFT February 3, 2015, Pre-Meeting Materials
for the EPA-FLM-RPO-States-Tribes Meeting, on the Future of the Regional Haze Program. This document has
been added to our docket.
446
-------
conclusion. Although we agree that much of those species contributions are due to natural
sources, we do not believe that all of these contributions are due to natural sources. Fires,
windblown CM and soil do have both anthropogenic and natural origins.
AECOM's Methodology
AECOM downloaded visibility monitoring data for BIBE and GUMO from 2001 to 2013 from
the Federal Land Managers Database.349 AECOM then extracted the data corresponding to the
20% worst days and then calculated a parameter "bCOS/abext,''' defined as the extinction due to
the CM, OMC, and the soil divided by the total aerosol extinction. AECOM then constructed
histograms that display the frequency in which its calculated parameter, bCOS/abext, occurred
within particular ranges, from 2004 to 2013.330 We reproduce those histograms below:
50
45
40
35
g 30
a ->C
3 25
O"
£ 20
LL
15
10
5
0
Histogram BIBE
/
HI.!-
Frequency
¦Cumulative %
100%
90%
80%
70%
60%
50%
40%
30%
20%
I- 10%
0%
0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95
Bin
349 http://views.cira.colostate.edu/fed/
350 Throughout this discussion, we reference AECOM's spreadsheet in which it performed the natural visibility
calculations, "Copy_of_AECOM_improve_2000-13.xlsx."
447
-------
a>
v
0
Histogram GUMO
40
35
30
25
20
£ 15
10
5
/
Frequency
Cumulative %
r 100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95
Bin
In the above histograms, AECOM defined ranges in which it counted the occurrences of its
parameter, bCOS/abext, which appear in the horizontal axis as "Bin." For instance with regard
to BIBE, AECOM noted that one occurrence of bCOS/abext occurred within the range of 0 to
0.05, and 7 occurrences of bCOS/abext occurred within the range of 0.05 to 0.10. Through
visual inspection of these histograms, AECOM then looked for a "noticeable step-up in
frequency of higher contributions of CM, OMC, and soil (i.e., from right to left)."351 AECOM
noted the "resulting frequency thresholds of the total aerosol extinction caused by CM, OMC,
and soil species for days of exceptional natural cases can be assigned as 40% for both BIBE and
GUMO."352 We believe that AECOM concluded that a break in the frequency of the
bCOS/abext ratio occurred in the bCOS/abext range of 0.35 to 0.40 (note the cumulative
percentage plotted on the right vertical axis is not used in AECOM's analysis). AECOM
asserted this break for both BIBE and GUMO. AECOM then further concluded that any days in
which the bCOS/abext parameter exceeded 0.40 represented days in which the "haze was caused
by CM, OMC, and soil species from natural causes, and thus those conditions can be properly
attributed to natural conditions." Having identified those days, AECOM then assumed all CM,
OMC, and soil extinctions for those days were due to natural causes. AECOM then summed the
individual CM, OMC, and soil extinctions for those days that occurred from 2004 to 2013 and
351 AECOM's discussion of its methodology* is taken from its report to Luminant entitled, "Analysis of the U.S.
EPA's Proposed Rule on the Texas and Oklahoma Regional Haze State Implementation Plans," which is attached as
Appendix B to Luminant's comments, contained in the file,
"Comments_of_Luminant_Generation_Company_LLC_-_Docket_EPA-RO6-OAR-2014-07_4.pdf."
351 As we discuss below, we have reservations as to whether any such "frequency threshold" can be defined for
GUMO. and we question the legitimacy of AECOM's method for choosing a 40% value for either BIBE or GUMO.
448
-------
divided them by the total CM, OMC, and soil extinctions for all monitored days from 2004 to
2013. AECOM then presented the resulting summary of that information:
Uncontrollable Fractions of 0as#lln« Extinction
for Three Species
Site
Frequency
"Wt'Stmlti
CM
OMC
Soil
BIBE
GUMO
40* 0T.5 0.484 0 053
40* 0 M0U 0.624 0 67H
AECOM then calculated the visibility impairment of these CM, OMC, and soil extinctions and
added that impairment to our default natural visibility conditions. For instance with regard to
CM, AECOM subtracted the natural conditions average CM value for the 20% worst days from
the baseline average CM value for the 20% worst days, then multiplied the result by the CM
cCOS/abext parameter it calculated above. This resulting value, which AECOM views as its
adjustment for CM extinction over and above our default extinction contribution, was then added
to the same. The calculation is presented below:
CM adjustment = CM ng90 + (CM C90-CM ng90) X CM uf
where:
CM ng90 = Natural Conditions II average of 20% worst days for CM
CM C90 = Baseline average of 20% worst days for CM
CM uf = AECOM's uncontrollable fraction of baseline extinction for CM, or 0.635
AECOM makes similar adjustments for OMC and soil and sums up the resulting adjusted and
unadjusted extinctions into a new total aerosol extinction. It then uses those values to calculate a
revised natural visibility condition value using the standard haze equation. We present that
information below:
Natural Conditions (dv)
Class 1 Area
NCI!89
TX NC70
IX ALT NC
BIBE
7 16
10.09
GUMO
8.65
12.26
1 "M
1 £*>, i
Flaws in AECOM's methodology
449
-------
As an initial matter, we believe that AECOM erred in assembling its histograms. This can be
readily detected by ordering the bCOS/abext values in AECOM's spreadsheet,3-3 and counting
the number of occurrences within the ranges AECOM defined. For instance, there is one value
(0.044) in the 0 to 0.05 range, two values (0.084 and 0.086) in the 0.05 to 0.10 range, and 15
values (0.103, 0.106, 0.113, 0.116, 0.117, 0.120, 0.121, 0.128, 0.129, 0.130, 0.135, 0.138, 0.144,
0.145, and 0.148) in the 0.10 to 0.15 range. We have constructed corrected histograms and
reproduce them below:354
35
30
25
U 20
c
at
D
cr
2 15
U.
10
5
0
EPA Histogram BIBE
/
/
/
iFrequency
-Cumulative %
III ill
50%
40%
30%
20%
10%
0%
0.05 0.15 0.25 0.35 0.4S 0.55 0.65 0.75 0.85 0.95
Bin
355 Column AD ill tab "NCThreshd" for BIBE.
354 Our corrected histograms appear in tab "EPA Histograms" in our reworked version of AECOM's spreadsheet,
entitled, "Copv_ol_ AECOM_iniprovc_2000-13_EPA.\ls\." which is in our final docket.
450
-------
It is readily apparent that our histograms differ significantly from AECOM's histograms. In
general, we believe that the "noticeable step-up in frequency of higher contributions of CM,
OMC, and soil (i.e., from right to left)" that AECOM points to in its report is now more muted
for both Class I Areas when the histograms are assembled correctly, to the point it is essentially
absent for the Guadalupe Mountains. Although we believe AECOM's assertion that the
identification of such a step-up in frequency was questionable for GUMO, it does not appear to
exist in the case of our corrected histogram for GUMO. Thus, we must reject AECOM's
methodology for GUMO just on this basis. In the case of BIBE, we believe that AECOM's
choice of a 40% value, based on visual inspection of its histograms, was somewhat arbitrary. It
is now difficult to see where such a break would occur. Nevertheless, setting aside this
reservation and other observations we discuss below concerning AECOM's methodology, we
believe it is instructive to test the sensitivity of AECOM's methodology to the choice of a
"frequency threshold." Choosing a value of 50% for BIBE, which we believe is minimally
appropriate, revises AECOM's natural visibility calculation for the 20% worst days for BIBE
from 9.25 to 8.79.355 Thus, AECOM's methodology is not robust and is sensitive to the
threshold chosen, which itself is subjective.
AECOM and Luminant construct revised URPs for BIBE and GUMO based on AECOM's
revised natural conditions calculations. AECOM concludes that because recent monitoring data
indicates that the Wichita Mountains (WIMO) and GUMO are projected to meet our FIP URP,
and with AECOM's revised URP BIBE will do the same, no further controls are necessary. As
we note above, we must reject AECOM's natural visibility calculations due to flaws in its
methodology. Because these calculations were subsequently used in AECOM's URP revisions
they are similarly flawed. Moreover, under the Regional Haze Rule, even if it were concluded
that the uniform rate of progress will be met for Big Bend and the Guadalupe Mountains, this
355 This was accomplished by changing the formulae in Column T from =IF(S3>=0.4,1,0) to =IF(S3>=0.5,1,0), in
tab "20wBB." of AECOM's afore mentioned spreadsheet.
451
-------
does not change the requirement that the reasonable progress goals be selected based on proper
evaluation of the four factors. As discussed in the proposal and this document, the uniform rate
of progress is not a "safe harbor" under the Regional Haze Rule. We address AECOM and
Luminant's assertion that because WIMO and GUMO are meeting our FIP URPs no further
controls are needed in our response to other comments, particularly our section responding to
assertions that RPGs are already being met.
Comment: [TCEQ/PUCT (0056) p. 8, 9] The TCEQ stated that if the EPA does not approve the
TCEQ natural conditions estimation that 100% of the soil dust at Big Bend and Guadalupe
Mountains on the 20% most impaired days is natural, it should choose an estimate between the
80%) natural estimate and 100%> approximation. The TCEQ urged the EPA to choose an estimate
that the dust is between 80%> and 100%> natural if the EPA chooses not to accept that estimate or
to withdraw its proposed partial SIP disapproval and FIP.
The TCEQ noted that the FLMs commented that 80%> would be more reasonable, but they did
not present evidence to support this suggestion. However, the TCEQ considers that 100% is well
supported in the 2009 RH SIP. The TCEQ contended that the 2009 RH SIP submittal presented
strong, peer-reviewed publication evidence that, on the most impaired 20% of days, essentially
all the coarse mass and fine soil at Guadalupe Mountains National Park is natural. It also
presented evidence assembled by six scientists, including the chairman of the IMPROVE
steering committee, that the dust impacts at Big Bend are largely from locally windblown dust.
Because of the strong National Park Service restrictions on human activity in Big Bend and the
fact that the IMPROVE monitor in Big Bend is surrounded in all directions by 10 or more miles
of the park, the conclusion is that naturally eroded soil contributes all or nearly all the coarse
mass and fine soil at Big Bend on the 20% of days with the most impaired visibility.
Response: We disagree with the TCEQ that we should "choose" an estimate between the 80%
and 100%) coarse mass and fine soil estimates. As we discuss above, we find that the TCEQ's
100%) estimate was not supported. The documentation to which the TCEQ here cites, "dust
impacts at Big Bend are largely from locally windblown dust" and "naturally eroded soil
contributes all or nearly all the coarse mass and fine soil at Big Bend"[emphasis added] supports
this conclusion. We have seen no documentation to support the FLM's 80%> estimate that the
TCEQ ultimately rejected. Consequently, we have no basis to conclude the 80 to 100%> range
itself even bounds the problem, let alone any insight as to which endpoint may be closer to the
truth.
Comment: [TCEQ/PUCT (0056) p. 8] The TCEQ agreed with the proposed EPA finding that
the TCEQ's estimate of baseline visibility conditions at Big Bend and Guadalupe Mountains
have satisfied the requirements of §51.308(d)(2)(i).
Response: We agree with the TCEQ's comment.
452
-------
Comment: TCEQ/PUCT (0056) p. 8] The TCEQ stated that, in Section V. B. 3 of the
preamble, the EPA has mischaracterized the requirement for states to calculate natural visibility
impairment beyond natural conditions. Table 3: Natural Visibility Impairment on page 74832 of
the proposal is an incorrect and misleading characterization of Chapter 5, Table 5-2: Visibility
Metrics for the Class I Areas in Texas, page 5-4 of the 2009 SIP. The TCEQ disagreed with the
EPA's assessment of compliance with this requirement and urges the EPA to approve TCEQ's
appropriate and technically defensible estimates of natural conditions, such as those used in the
2009 RH SIP. Section 51.308(d)(2)(iv)(A) of the RHR says:
For the first implementation plan addressing the requirements of paragraphs (d) and (e) of
this section, the number of deciviews by which baseline conditions exceed natural
visibility conditions for the most impaired and least impaired days ... [underline added]
The TCEQ stated that, although the EPA appropriately proposes to find that the 2009 RH SIP
correctly stated the baseline conditions at Big Bend and Guadalupe Mountains, the subsection
just cited requires that the natural visibility conditions for the most and least impaired days at
each Class I area be subtracted from the baseline conditions for the most and least impaired days
to determine the number of deciviews by which baseline conditions exceed natural conditions on
the respective sets of days.
Response: The table in our proposal that the TCEQ cites is merely a reformatted version of the
same table the TCEQ itself presents on page 5-4 of its SIP. In this section of our proposal, we
were summarizing the TCEQ's own information. Thus, the TCEQ's assertions that we somehow
erred in following our own regulations is misplaced. We rechecked the information in our table
and conclude that it correctly reproduced the TCEQ's information. As we discuss in our
proposal, we are disapproving the determination of the number of deciviews by which baseline
conditions exceed natural conditions because this calculation depends on the TCEQ's
calculations for natural visibility conditions, which we are also disapproving. We disagree with
the TCEQ's assertion that we "mischaracterized the requirement for states to calculate natural
visibility impairment beyond natural conditions," or that this table constitutes "an incorrect and
misleading characterization of [the TCEQ's] Chapter 5, Table 5-2."
Comment: [TCEQ/PUCT (0056) p. 9] The TCEQ disagreed with the EPA's proposed URP and
natural conditions for both the Texas Class I areas. Once a final, technically supportable
estimate of natural conditions has been selected, the URP can be calculated by straight-line
interpolation from the baseline visibility conditions (2000 - 2004) to the estimated natural
conditions in 2064 for each of the Texas Class I areas.
The TCEQ stated that the EPA failed to note that, since over 50% of the visibility impairment at
Big Bend on the most impaired 20% days comes from outside the U.S. and since there is no basis
for projecting a reduction in that impact, the goal of reaching natural conditions at Big Bend is
unrealistic, as is the implied goal of attaining the URP at any time.9 A more appropriate goal
would be to achieve an appropriate reduction of the visibility impairment caused by
anthropogenic emissions from Texas and the rest of the U.S. Later in the first full paragraph on
page 79 FR 74843, the EPA correctly concluded that "it is not reasonable to meet the URP for
453
-------
the Texas Class I areas for this planning period." The EPA also recognized that "emissions and
transport from Mexico and other international sources will limit the rate of progress achievable
on the 20% worst days ..."
Commenter's Reference:
9 See the EPA's approval of Arizona's natural conditions goal of767 years out for Saguaro East in 79 FR 52469.
Response: We address our calculation of natural visibility conditions in our responses to other
comments. Also, in other comments, the TCEQ states that it agrees with our calculation of its
baseline conditions. Because as the TCEQ notes, the URP is merely a straight line connecting
the baseline and natural conditions, we cannot offer any further information concerning these
issues. We address the impact of international emissions in our responses to more detailed
comments on that subject.
Comment: [TCEQ/PUCT (0056) p. 17] As part of their comments on the EPA's proposed FIP,
the TCEQ disagreed with the EPA proposal to calculate visibility impairment, (i.e., baseline
visibility conditions minus natural visibility conditions) using the EPA's proposed substitute
natural visibility conditions for Big Bend and Guadalupe Mountains instead of the natural
visibility conditions calculated by Texas for its two Class I areas.
The TCEQ stated that the EPA should accept Texas' calculation of natural visibility conditions at
Big Bend and Guadalupe Mountains. These calculations followed the requirements of 40 CFR
51.308(d)(2)(iii) using data and analyses specific to each of the Class I areas. The EPA's
proposed substitute estimates of natural conditions were developed by a committee working on
national estimates rather than using site specific scientific studies. The EPA did use the correct
Baseline Visibility Conditions, 2000-2004, in Table 40.
Response: These issues have been addressed in our responses to other comments.
Comment: [TCEQ/PUCT (0056) p. 17-18] As part of their comments on the EPA's proposed
FIP, the TCEQ supported the EPA's proposal to find that it is not reasonable to provide for rates
of progress at Wichita Mountains, Big Bend, or Guadalupe Mountains that would attain natural
visibility conditions by 2064 and to use the baseline conditions calculated by Texas in
establishing the URP at Big Bend and Guadalupe Mountains.
The TCEQ noted that once technically supportable natural conditions estimates are selected for
these two Class I areas, the URP can be established for them. However, the TCEQ disagreed
with the EPA's proposal regarding the natural conditions estimates.
Response: These issues have been addressed in our responses to other comments.
Comment: [UARG (0065) p. 20-21] As part of their argument that EPA does not provide any
lawful basis for disapproving the RPGs for Big Bend and Guadalupe or the Texas LTS, UARG
noted that EPA states that it proposes to disapprove Texas's URP analysis because "we do not
454
-------
believe that the rate of improvement the TECQ [sic] has selected is reasonable, because we
disagree with its four factor analysis and the analysis of emission measures needed to meet the
URP." 79 Fed. Reg. at 74,843. This is not a lawful basis for a SIP disapproval. Texas has
complied with the regional haze rule's requirements with respect to a URP analysis, and, for the
reasons stated above, EPA has not provided a valid justification for disapproving Texas's
reasonable progress analysis.
UARG stated that the EPA proposes to disapprove Texas's calculation of the URP as a result of
its proposal to disapprove the state's calculation of natural visibility conditions at Big Bend and
Guadalupe. Id. at 74,822. The primary reason for this proposed disapproval is Texas's
assumption that fine soil and coarse mass concentrations, and their resulting light extinction
effects, is entirely attributable to natural sources. Id. at 74,831. Texas provided a reasoned
justification for its decision in this regard, explaining that "to the extent its assumption that 100%
of coarse mass and fine soil is natural is an overestimate, it expects that its low organic carbon
estimate will more than compensate for any errors in this assumption at this time." Id. Despite
the reasonableness of this finding, Texas, at the FLMs' request, performed a supplemental
analysis assuming that only 80 percent of this particulate matter was from natural sources. Id. In
the final analysis, the FLMs recognized that EPA's regulations at "40 CFR 51.308 give[] the
State [the] right" "to recalculate natural conditions for Big Bend [National Park] and Guadalupe
Mountains [National Park]" and agreed that "the basic approach used [by Texas] to adjust natural
conditions is reasonable, provided that the Proposed SIP address the uncertainty of the
assumption that all of the coarse mass and fine soil fraction on the worst 20 percent days is
natural."5 Texas's final SIP submission provides the basis for the assumption, as the FLMs
requested.6 In contrast, EPA, without any meaningful analysis, would impose use of default
values for this particulate matter provided for in the IMPROVE equation and proposes
disapproval of the Texas SIP on that basis. Id. at 74,832. EPA provides no basis for rejecting
Texas's reasoned explanation. Particularly in light of the FLM comments and Texas's rationale\
for using 100 percent, EPA has a duty to fully explain and to provide an adequate rationale for
selecting the default values over Texas's determination. EPA failed to discharge that duty, and
its proposed action is improper for that reason as well.
Commenter's References:
5 U.S. Fish & Wildlife Serv. & Nat'l Park Serv., Comments on Texas Proposed Regional Haze Rule State
Implementation Plan at 2, 3 (Jan. 11, 2008), Doc. ID No. EPA-R06-OAR-2014-0754-0002, TX166-002-03.
6 See generally TCEQ, Revisions To The State Implementation Plan (SIP) Concerning Regional Haze at Appendix.
5.2 (adopted Feb. 25, 2009), Doc. ID No. EPA-R06-OAR-2014-0754-0002, TX166-002-05 ("2009 Texas SIP").
Response: These issues have been addressed in our responses to other comments.
Comment: EPA unjustifiably proposes disapproval of TCEQ's URPs for Big Bend and
Guadalupe Mountains in favor of URPs that reflect default values with no site-specific
adjustment for natural haze conditions [CCP (0075) p. 4]
CCP stated that the URP is the rate of visibility improvement (expressed in deciviews) needed to
attain natural visibility conditions in a Class I area by 2064. 40 C.F.R. § 51.308(d)(1). For
purposes of the first regional haze planning period, the relevant URP is the amount of
improvement needed by 2018 to be on track to attain natural visibility conditions in 2064. In
455
-------
order to calculate a URP for a Class I area, it is first necessary to determine natural visibility
conditions in that area. EPA regulations provide that the ultimate responsibility for calculating
natural conditions lies with the state. 40 C.F.R. §51.308(d)(2)(iii). Even though the EPA
"Guidance for Estimating Natural Visibility Conditions Under the Regional Haze Program"
provides "default" estimates of natural visibility, it, too, emphasizes the state's right to derive
"refined" estimates.
Consistent with EPA guidance, CCP noted that TCEQ found that natural visibility conditions at
Big Bend and Guadalupe Mountains were significantly influenced by particulate matter from
natural conditions such as dust storms and fires as would be expected in this region. Therefore,
TCEQ adjusted its natural visibility conditions in these areas to reflect only the visibility
impairment caused by man-made emissions sources that are the potential targets for control
under the CAA. If anthropogenic sources are not the cause of haze in these Class I areas, the
CAA does not require those sources to be controlled since those controls would not result in
improvements in visibility. See 40 C.F.R. § 51.308(d)(3)(iv) (requiring states to identify "all
anthropogenic sources of visibility impairment").
CCP stated that, in the Proposed Rule, EPA agrees that "dust storms and other blown dust from
deserts are a significant contributor to visibility impairment at the Texas Class I areas that may
not be captured accurately by our default method." 79 Fed. Reg. 74,831. The record and
supporting materials clearly show the contributions of natural events on visibility conditions,
including data on wildfires, exceptional dust storms, droughts, and composition plots showing
the substantial percentage of naturally-occurring haze-forming particles on the worst visibility
days. Indeed, modeling shows that during one of the worst droughts on record in 2011, some of
the worst haze days in Big Bend corresponded to events flagged as "Fire - Mexico/Central
America."
CCP stated that the EPA arbitrarily ignores the observed contribution of natural dust on visibility
conditions and rejects TCEQ's adjustments in favor default values for these areas that. EPA
acted arbitrarily in rejecting TCEQ's adjustments in favor of its own default values, speculating
that "[ajnthropogenic sources of coarse mass and fine soil in the baseline period could have
included emissions associated with paved and unpaved roads, agricultural activity, and
construction activities." 79 Fed. Reg. 74,831. The mere possibility that other anthropogenic
emissions "could have contributed" to coarse mass is not a sufficient rationale for using
inappropriate default factors and rejecting "reasonable" site-specific adjustments.
Although EPA erred in not proposing adjustments supported by the available information in the
record, CCP stated that the EPA did seek comment on its use of a default value and alternatives
to TCEQ's and EPA's approaches. The attached comments offer an adjustment approach based
on the available data that satisfies EPA guidance and should be used by EPA in the event that
EPA fails to withdraw the Proposed Rule.
Response: These issues have been addressed in our responses to other comments.
456
-------
Comment: EPA's Disapproval of Texas's Calculation of Natural Visibility Conditions and
Associated Calculations of Natural Visibility Impairment and the Uniform Rate of
Progress Is Justified. [Earthjustice (0067) p.20]
Earthjustice et al., stated that the EPA correctly proposes to disapprove Texas's determination of
natural visibility conditions at its two Class I areas, Big Bend and Guadalupe Mountains. 79
Fed. Reg. at 74,830- 32. Every state except for Texas calculated natural visibility at their Class I
areas in accordance with EPA's standard NCII methodology. Id. at 74,832. Texas, however,
chose to calculate natural visibility at Big Bend and Guadalupe Mountains based on a different
approach that departed from EPA's standard methodology. Id. at 74,831. The result is an
artificially inflated calculation of "natural visibility" conditions. If Texas's calculation is used to
define the goal to meet, true natural visibility at Big Bend and Guadalupe Mountains would
never be restored.
For example, according to EPA's standard method, natural visibility at Big Bend on the 20%
worst days is 7.16 dv. Id. But Texas calculated Big Bend's natural visibility to be 10.09 dv—
which is 2.93 dv greater than natural visibility conditions under EPA's method. Id. The
disparity between Texas's approach and EPA's standard method is even more pronounced at
Guadalupe Mountains, where natural visibility is 6.65 dv on the 20% worst days according to
EPA's method and 12.26 dv under Texas's approach. Id. As described below, the Texas
approach significantly skews its analysis towards a result that does not require additional
emissions reductions from Texas sources.
Earthjustice et al., stated that the primary cause of this wide disparity between the Texas
approach and EPA's standard methodology is the state's decision to attribute 100% of coarse
mass and fine soil concentrations to natural causes. Id. However, it is extremely unlikely that
natural dust storms cause 100% of the coarse mass and fine soil concentrations at Texas Class I
areas. As EPA explains, coarse mass and fine soil pollution is often caused by dust from paved
and unpaved roads, agricultural activity, and construction activities—all of which occur in
Texas. Id. Indeed, the ubiquity of emissions from construction underlies the requirement for all
state long-term strategies to consider "measures to mitigate the impacts of construction
activities." 40 C.F.R. § 51.308(d)(3)(v)(B).
Earthjustice et al., stated that Texas admits that there is significant uncertainty associated with its
estimates. Calculations of natural conditions should be based on supportable science, including
"available monitoring information and appropriate data analysis techniques." 40 C.F.R. §
51.308(d)(2)(iii). The state's natural visibility estimates were not. Its estimates of natural
conditions were highly influenced by and very sensitive to its attribution of 100% of coarse mass
and fine soil concentrations to natural causes—a decision that is both highly implausible and
unsupported. Therefore, EPA's disapproval - and the continued use of EPA's default values - is
appropriate. The approach Texas took significantly overestimates the level of natural visibility
impairment at Big Bend and Guadalupe Mountains. As a result, the state's approach
fundamentally undermines the haze program's core purpose, which is to restore natural visibility
conditions at our nation's most treasured public lands. Because the regional haze program's goal
is to attain natural visibility conditions at every Class I area, the natural visibility calculation sets
the ultimate target for each Class I area. See 64 Fed. Reg. 35,714, 35,729 (July 1, 1999)
457
-------
(explaining that the "[estimates of natural visibility conditions are needed to aid all interested
parties, including the general public, in understanding how 'close' or 'far' a particular Class I
area is in relation to the ultimate goal of the program"). The use of a higher natural visibility
impairment skews the slope of the uniform rate of progress and makes it appear that a much
lower rate of progress is sufficient. This biases the evaluation of what progress is "reasonable"
in favor of doing less to reduce anthropogenic impairment. Because both are based on the state's
flawed estimate of natural visibility conditions, EPA's disapproval of the Texas estimate of
natural visibility impairment and the uniform rate of progress was appropriate.
Earthjustice et al., stated that Texas's unorthodox calculation of natural visibility conditions also
leads to inconsistent and anomalous results. Guadalupe Mountains is located just 40 miles from
Carlsbad Caverns National Park in New Mexico, and both national parks share an IMPROVE
monitor. See id. at 74,843. Texas calculated natural visibility conditions at Guadalupe
Mountains to be 12.26 dv. Id. at 74,831. But New Mexico—which used EPA's standard
method—calculated natural visibility conditions at Carlsbad Caverns to be 6.65 dv. New Mexico
309(g) SIP at 31 (Mar. 31, 2011); see also 77 Fed. Reg. 36,044, 36,068 (June 12, 2012)
(proposed approval of New Mexico haze plan); 77 Fed. Reg. 70,693 (Nov. 27, 2012) (final rule
approving New Mexico haze plan). Given the two national parks' proximity and the fact that
they share an IMPROVE monitor, Texas's inflation of natural visibility impairment at
Guadalupe Mountains is problematic when compared to Carlsbad Caverns, for which national
conditions are based on EPA's standard methodology. Departing from this methodological norm
is unjustified. Both national parks are entitled to the same protections under the regional haze
program.
Earthjustice et al., stated that supported EPA's proposal to use the NCII default values for
determining natural visibility conditions at the Texas Class I areas. The NCII default values
were used for every other Class I area in the country, and EPA should not revise its proposal to
set a less stringent natural visibility target. Because the Federal Land Managers commented
during the development of the Texas regional haze SIP that it is more reasonable to assume that
80% of coarse mass at Texas Class I areas is natural (and not 100% as Texas assumed in its
natural visibility calculations), EPA has asked for comments on the acceptability of natural
conditions estimates between EPA's defaults and Texas's values. We support the use of EPA's
default values for several reasons.
First, as EPA notes, there is significant uncertainty associated with the assumptions used by
Texas. This uncertainty would persist in the use of any estimates between EPA's default and
Texas's values. Second, as noted above, all other states have used EPA's default values. They
should be used here in the interest of both national consistency and also local consistency
between Guadalupe Mountains and Carlsbad Caverns National Parks. Third, EPA's default
values are based on the most recent science. Any deviations or refinements should be based on
supportable science, including "available monitoring information and appropriate data analysis
techniques." 40 C.F.R. § 51.308(d)(2)(iii). Texas's natural visibility estimates were not.
Therefore, EPA's use of its default values in its FIP is appropriate.
Response: We agree with Earthjustice that our disapproval of Texas's natural visibility
calculations and replacement with our own calculations was appropriate.
458
-------
Comment: EPA Must Disapprove Texas's RPGs For Big Bend and Guadalupe
Mountains as Unlawful. [Earthjustice (0067) p.26]
Earthjustice et al., stated that the EPA rightly proposes to disapprove the progress goals that
Texas established for Big Bend and Guadalupe Mountains because the RPGs do not provide for
reasonable progress based on the four factors that a state is required to consider. Texas adopted
the CENRAP modeled 2018 visibility conditions as the RPGs for Big Bend and Guadalupe
Mountains. To set RPGs for 2018, Texas relied on improvements in visibility that are
anticipated to result from federal, state, and local control programs and narrowed the scope of its
control analysis to point sources of NOx and SO2. Under the reasonable progress goal proposed
by Texas, natural visibility conditions as calculated by Texas would not be reached at Big Bend
until 2155 and the Guadalupe Mountains until 2081. Using the NCII default values for natural
visibility conditions and Texas's RPG, natural visibility conditions would not be attained at Big
Bend until 2215 and Guadalupe Mountains until 2167. Texas's proposed rate of progress is
significantly slower than the URP. Texas found that its reasonable progress goal was reasonable
because contributions from Mexico and other international sources allegedly prevent Texas from
achieving a faster rate of progress. However, as discussed below, the contribution from Mexico
and other international sources is not a justification for such a slow rate of progress where cost-
effective controls are available for Texas sources.
Response: We agree with Earthjustice that we are correct to disapprove Texas' RPGs.
Comment: Natural Background Calculation Methods [Alpine (0078) p. 5, 10] Alpine stated that
the TCEQ's method for calculating natural background visibility is reasonable. Additionally,
EPA has not adequately demonstrated that the 80% natural source composition default value is
more appropriate than TCEQ's 100% value at these specific Class I areas.
Alpine noted that when including the alternate uniform rate of progress slope calculation made
by TCEQ in its SIP submittal, a calculation that uses a refined methodology to estimate natural
visibility conditions, we also find these observations are even more favorable at Big Bend and
Guadalupe Mountains than when compared to EPA's calculated uniform rate of progress line.
The TCEQ constructed their URP by plotting a straight graphical line from the baseline level of
visibility impairment to the level of visibility conditions representing no anthropogenic
impairment in 2064 for both Big Bend and the Guadalupe Mountains. No alternate calculation
was made by Texas for the Wichita Mountains Wilderness Class I area. This revised calculation
is represented by the green dotted line in Alpine Figures 2 and 3 and represents a natural
background visibility estimate that assumes 100% of the amount of natural fine soil and coarse
mass in 2064 is natural.
Alpine stated that Title 40 CFR §51.308(d)(2)(iii) states that the ultimate responsibility for
calculating natural conditions lies with the state. Even though the EPA "Guidance for
Estimating Natural Visibility Conditions under the Regional Haze Program" provides "default"
estimates of natural visibility, it, too, emphasizes the state's right to derive "refined" estimates.
459
-------
Alpine stated that if we were to define "natural conditions" as EPA does in its guidance, we
would state that this level is the result of visibility conditions that would be experienced in the
absence of human-caused impairment. Using this definition, it is reasonable to agree with
TCEQ's 2064 natural background visibility estimate that 100% of the amount of natural fine soil
and coarse mass are natural. EPA's argument includes the fact that their review of the baseline
visibility estimate, an average of recent observational conditions, and a level that TCEQ has
agreed is the same as EPA's calculation and that EPA approved in their review of the SIP, does
not take into account that 100% of the coarse mass and fine soil measured are attributable to
natural sources. In reality, it is not the current year average baseline value that EPA should be
concerned. This baseline value is the starting point for the uniform rate of progress line that
initiates with current condition calculations for the W20% days and ends at 2064 natural
conditions; a level absent of manmade influence.
Alpine stated that in calculating the uniform rate of progress slope, TCEQ's assumption of 100%
natural contribution to coarse mass and fine soil in 2064 meets the true definition of natural
background visibility, the end point of URP slope. EPA notes that anthropogenic sources of
coarse mass and fine soil in the recent baseline period could have included emissions associated
with vehicle perturbed paved and unpaved road dust, agricultural activity, and construction
activity. However, EPA does not make the same statement for the 2064 natural background
calculation, a visibility calculation purposefully absent the influence of these manmade
interactions with natural emissions, as TCEQ notes in its refined calculation.
Alpine noted that, in its disapproval of this area of the SIP, EPA states that because "we find that
the TCEQ has not adequately demonstrated that all coarse mass and fine soil measured in the
baseline period can be attributed to 100% natural sources", TCEQ does not meet the calculation
of the natural visibility conditions for the Big Bend and Guadalupe Mountain Class I areas under
§51.308(d)(2)(iii). However, EPA's judgment on this point only refers to the baseline, or
current, levels of the URP, not the natural background endpoint, the value that TCEQ uses its
right to calculate using an alternate method.
According to Alpine, since EPA's own definition of natural visibility conditions supports
TCEQ's usage of 100% natural source composition of coarse mass and fine soil, it can be
concluded that the TCEQ approach to estimating natural visibility conditions is reasonable.
Additionally, EPA has not adequately demonstrated that the 80% natural source composition
default value is more appropriate than TCEQ's 100% value at these specific Class I areas.
Response: We agree with the commenter that the Regional Haze Rule does allow states to
develop an alternate approach to estimate natural visibility conditions. The fact that the states
have the option of calculating their own natural visibility conditions is not at issue. The
commenter is incorrect in their explanation of how TCEQ estimated natural visibility conditions
for the 2064 goal at the two Texas Class I areas. Contrary to the commenter's assertion that
Texas estimated that course mass and fine soil for the 2064 to be 100% from natural sources,
Texas evaluated the baseline extinction from course mass and fine soil, assumed that the baseline
extinction was due to 100% natural sources, and then carried this baseline concentration forward
as an estimate of the 2064 natural conditions. As we state in the Texas TSD and discuss further
460
-------
in a separate response to comment, Texas did not adequately support the determination that the
baseline extinction was 100% from natural sources. The commenter confuses the EPA default
values with the FLM suggested assumption of 80% natural source composition that TCEQ used
for some additional calculations.
14. Consistency with Our Other Regional Haze Actions
The following comments allege specific instances of inconsistency with our other SIP and FIP
actions. In some instances these comments constitute the entirety of the comment, and in other
instances they are only a portion of the comment. We have collected these comments in this
separate document so that our responses to allegations of inconsistencies can be more easily
reviewed.
We also received comments that many of these cited instances are contrary to our regional
consistency regulations at 40 CFR § 56.5. We explain the applicability of our regional
consistency rules in detail in the first response to comments within this section below. We
recognize that we have a duty to ensure our regional haze actions are carried out fairly, are
consistent with the CAA, and are "as consistent as is reasonably possible" with other regional
haze actions. 356 Further, as evidenced by our response to allegations that we have been
inconsistent, we disagree that we are acting inconsistent with reasonable progress requirements
or prior SIP actions in taking this final action for Texas and Oklahoma. Rather, we believe we
have been consistent in our review and analysis of the Texas SIP given the specific facts
presented in Texas and Oklahoma.
The Regional Haze Program is one of our more complicated programs, by virtue of our desire to
extend flexibility to our states. We funded Regional Planning Organizations so that states could
have forums to discuss approaches to regional haze that were best suited to their individual
states. The Regional Haze Rule was designed to offer states multiple pathways to achieving the
national goal of a return to natural visibility. States have flexibility in meeting our regulations,
but our role as reviewer remains. Nevertheless, when we determine that a state's SIP is not
approvable and we must replace it with a federal plan, we too are afforded some flexibility, as
long as our federal plan complies with the CAA and our regulations.
What some commenters allege is inconsistency, is in fact the exercise of our judgment, based on
the specific facts at hand. Because this is a SIP review action, we believe that we are not only
authorized, but required to exercise independent technical judgment in evaluating the adequacy
of the State's regional haze SIP, just as we must exercise such judgment in evaluating other SIPs.
In evaluating other SIPs, we are constantly exercising judgment about SIP adequacy, not just to
meet and maintain the NAAQS, but also to meet other requirements that do not have a numeric
value. In this case, Congress did not establish NAAQS by which to measure visibility
improvement; instead, it established a reasonable progress standard and required that EPA assure
that such progress be achieved. Here, we are exercising judgment within the parameters laid out
in the CAA and our regulations.
356 40 CFR § 56.5(a)(2).
461
-------
Comment: EPA's disapproval is contrary to the agency's regional consistency rules
[Luminant (0061) p. 68]
Luminant stated that, because EPA's proposal applies a wholly different standard to Texas' SIP,
it is unlawful. Under EPA's Clean Air Act regulations, "[e]ach responsible official in a Regional
Office, including the Regional Administrator, shall assure that actions taken under the [CAA]:
(1) Are carried out fairly and in a manner that is consistent with the CAA and Agency policy as
set forth in the Agency rules and program directives [and] (2) Are as consistent as reasonably
possible with the activities of other Regional Offices . . . ,"484 These regulations "strongly
articulate EPA's firm commitment to national uniformity" in its CAA actions, and EPA actions
that violate these regulations are "contrary to law."485 Here, by applying a new and singularly
different standard to Texas, and prohibiting the state from using a source category approach,
EPA is acting inconsistent with both its own reasonable progress policy guidance and prior
actions of many other EPA regional offices. EPA's proposal violates the agency's own regional
consistency regulations and is contrary to law.486
[UARG (0065) p. 28] UARG stated that, in previous regional haze rulemakings, EPA has touted
national consistency in analytical approach and consistent results across states and regions as not
only a goal but a requirement of the CAA and the regional haze rules. See, e.g., 77 Fed. Reg.
72,512, 72,518 (Dec. 5, 2012) (rejecting reliance on site-specific BART compliance costs in
order to promote what EPA characterized as consistent costing methodologies across all states).
Even in this proposed rule, EPA notes that the regional haze rule imposes certain requirements
"to ensure that states use a common analytical framework and to provide an informed and
equitable decision making process." 79 Fed. Reg. at 74,834. EPA's proposed action here,
however, is anything but consistent with its past actions. Not only is this differential treatment
unjustified and inconsistent with the regional haze rule; it also violates EPA's regional
consistency regulations. Those regulations require that EPA "shall assure that actions taken
under the [CAA]: (1) Are carried out fairly and in a manner that is consistent with the CAA and
Agency policy as set forth in the Agency rules and program directives [and] (2) Are as consistent
as reasonably possible with the activities of other Regional Offices." 40 C.F.R. § 56.5(a).
EPA's proposed rule does not pass these basic tests of rational and consistent Agency decision-
making.
[Associations (0059) p. 15] The Associations stated that by applying a wholly different standard
in its evaluation of Texas' reasonable progress goals, EPA is violating its strict uniformity rule.
EPA regulations state that "[e]ach responsible official in a Regional Office, including the
Regional Administrator, shall assure that actions taken under the act: (1) Are carried out fairly
and in a manner that is consistent with the CAA and Agency policy as set forth in the Agency
rules and program directives [and] (2) Are as consistent as reasonably possible with the activities
of other Regional Offices ...." 40 C.F.R. § 56.5(a). This regulation reflects an agency-wide
commitment to uniformity in interpreting and applying the Clean Air Act, and agency actions
that violate these regulations are "contrary to law." See National Environmental Development
Association's Clean Air Project v. EPA, 752 F. 3d 999 (D.C. Cir. 2014). In this proposal, EPA
unlawfully applies a wholly different standard of review to Texas' reasonable progress goals
462
-------
than it has in prior reviews of reasonable progress goals submitted by other States that have
likewise relied on source category-based analyses. To satisfy its own uniformity rule, EPA must
treat Texas like any other State and approve its use of a source category-based analysis in setting
reasonable progress goals.
GCLC stated that EPA's Proposed FIP imposes unprecedented limitations on Texas through a
completely novel means. This includes conducting an "additional analysis," creating a new fifth
visibility factor, and imposing source-specific SO2 emissions limitations on 15 separate BART-
compliant and BART-exempt Texas EGUs a type of analysis or limitation that EPA has not
imposed on any other state through the regional haze SIP /FIP process. This is also a
contradiction of EPA's own rulemaking that CSAPR (and previously CAIR) serves as better-than
BART for BART sources, and this goes far beyond any other type of limitations considered for
non-BART sources in the statute, regulations, guidance documents, and every other review of
other states' SIP submission. Some of the specific examples where EPA has acted inconsistently
with prior actions have already been referenced in our comments above (e.g., EPA's actions on
the Idaho and Nebraska SIP submissions) but extend far beyond those. This includes EPA's
recent approvals of reasonable progress evaluations, where states have taken the same approach
and reached the same results as Texas, and EPA has approved them. 70
GCLC noted, by EPA's own admission, its FIP required a "thorough technical and policy
analysis" in order to "ensure compliance with the Regional Haze Rule," because EPA is
proposing the rule "without the benefit of prior precedent to streamline the process." 71 This is
despite Texas' FIP being proposed after years of EPA's regional haze SIP/FIP determinations for
numerous states. By taking this inconsistent and much-more burdensome path for Texas and
Texas EGUs, EPA's Proposed FIP is in clear contravention to the regional consistency
regulations found at 40 CFR, Part 56 - Regional Consistency. It is also in clear contravention of
the recent D.C. Circuit holding in National Environmental Development Association's Clean Air
Project v. EPA, which held that directives contravening the regional consistency regulations are
a violation of law.
Footnotes:
70 See Approval and Promulgation of Implementation Plans; Region 4 States; Visibility Protection Infrastructure
Requirements for the 1997 and 2006 Fine Particulate Matter National Ambient Air Quality Standards, Final Rule, 79
Fed. Reg. 26,143,26,145-46 (May 7, 2014), which included: "Regarding the reasonable progress evaluations, each
state at issue focused its reasonable progress analysis on SO2 emissions based on the conclusion that sulfate
particles account for the greatest portion of the regional haze affecting Class I areas in these states. Each state then
established areas of influence and contribution thresholds to determine which of its sources should be evaluated for
reasonable progress control. EPA approved each state's methodology for identifying units for reasonable progress
evaluation and each state's reasonable progress determinations in the respective regional haze SIP actions and
provided a detailed discussion of the methodology and the rationale for approval in the Federal Register notices
associated with those actions. Contrary to the Commenter's assertions, Alabama, Georgia, Kentucky, North
Carolina, and South Carolina did not "exempt [CAIR] sources ... that would otherwise be subject to reasonable
progress review." Each of these states considered the four statutory reasonable progress factors in evaluating
whether CAIR would satisfy reasonable progress requirements for the state's EGU sector and determined that no
additional controls beyond CAIR were reasonable for SO2 during the first planning period. As discussed in EPA's
Reasonable Progress Guidance, states may evaluate the need for reasonable progress controls on a source category
basis, rather than through a unit-specific analysis, and have wide latitude to determine additional control
requirements for ensuring reasonable progress. The guidance also notes that states may consider emissions
reductions from cap-and-trade programs such as CAIR in addition to source-specific controls."
463
-------
71 Declaration of Sam Coleman, Nat 'I Parks Conservation Ass 'nv. McCarthy, No. 11-01548, at 5 (D.D.C. 2014).
(emphasis added).
Footnotes:
484 40 C.F.R. § 56.5(a). These requirements also apply to EPA Headquarters officials "who are responsible for
developing the policies governing the implementation and enforcement of the CAA." Nat'l Envt'l Dev. Ass'n's
Clean Air Project v. EPA, 752 F.3d 999, 1009 (D.C. Cir. 2014).
485 Nat'l Envt'l Dev. Ass'n's Clean Air Project, 752 F.3d at 1010-11.
486 Id. (holding that EPA action contrary to its regional consistency regulations was contrary to law); see also 40
C.F.R. § 56.5(a)(2) (officials in EPA regional offices "shall assure that actions taken under the act. . . [a]re as
consistent as reasonably possible with the activities of other Regional Offices" (emphasis added)).
Response: We recognize that we have a duty, under our regional consistency regulations, to
ensure our regional haze actions are "consistent with the [Clean Air] Act and Agency policy"
and are "as consistent as reasonably possible" with other regional haze actions under 40 CFR §
56.5(a)(1) and (2). As explained below and elsewhere throughout this response to comments, we
believe we are meeting the goals of these regulations in light of the specific facts presented in
this action. We disagree that we are acting inconsistent with reasonable progress requirements or
prior SIP actions in taking this final action for Texas and Oklahoma. Moreover, while EPA's
regional consistency regulations and policies require us to carry out our actions pursuant to the
CAA in a consistent manner across Regions as reasonably as possible, they do not require
uniformity between those actions in all circumstances and instead "allow for some variation" in
actions taken in different regions. 357As explained below, in this action EPA is acting consistent
with the CAA and our regional haze policies in taking these specific actions for Texas, and our
final action is "as consistent as reasonably possible" with other actions given the specific facts
presented in Texas and Oklahoma.
We believe we have been consistent in our review and analysis of the Texas SIP. As explained
in more detail elsewhere in this Response to Comments, we must first review the SIP under the
requirements of the CAA and federal regulations. Congress crafted the CAA to provide for
states to take the lead in developing implementation plans, but balanced that decision by
requiring us to review the plans to determine whether a SIP meets the requirements of the CAA.
We have the authority to disapprove a SIP if it does not meet minimum CAA and/or regulatory
requirements. Our action today is consistent with the statute.
Further, each Regional Haze SIP requires that we review it on an individual, case by case basis,
based on the specific facts presented. Regional haze problems can fundamentally differ from
state to state and region to region and thus require a case by case review. The Regional Haze
Program is one of our more complicated programs, by virtue of our desire to extend flexibility to
our states. We funded Regional Planning Organizations so that states could have forums to
discuss approaches to regional haze that were best suited to their individual states. The Regional
Haze Rule was designed to offer states multiple pathways to achieving the national goal of a
return to natural visibility. States have flexibility in meeting our regulations, but our role as
reviewer remains - we must review state regional haze plans to determine whether they meet the
requirements of the CAA. Regardless of the inherent flexibility in the regional haze program, if
we determine that a state's SIP is not approvable, we must replace it with a federal plan. And in
357 80 FR 50258.
464
-------
devising that federal plan, we too are afforded some flexibility, as long as our federal plan
complies with the CAA and our regulations.
We disagree with Commenter's assertion that we are applying a new and singularly different
standard to Texas, and we further disagree that we have prohibited Texas from using a source
category approach. As we explain in other comments, Texas partially analyzed its sources on a
source-by-source approach, but did so in an incomplete and flawed manner. Our proposed
disapproval was not based on the fact that Texas also used a category approach, but rather that
this analysis was also flawed and did not comply with our regulations. While we believe we are
consistent in our review here, we address allegations that we have been inconsistent with our
other actions in our responses to other more specific comments
With regard to the commenter's general assertion that our action here is "unlawful," we disagree.
In support of their claims, the comment relies heavily on the case National Environmental
Development Association's Clean Air Project v. EPA (NEDA CAP), No. 13-1035 (DC Cir., May
30, 2014). NEDA CAP involved a specific December 2012 memorandum from EPA
headquarters to the EPA regions regarding the limited scope of a court decision issued by the
Sixth Circuit Court of Appeals based upon the doctrine of intercircuit nonacquiescence (see
"Memorandum from Stephen D. Page, Director of the EPA's Office of Air Quality Planning and
Standards, to Regional Air Division Directors, titled Applicability of the Summit Decision to the
EPA Title V andNSR Source Determinations (December 21, 2012")). 358 While that case
discussed application of EPA's regional consistency regulations, the court was focusing on the
issue of whether such a memo addressing intercircuit nonacquiescence was allowed under the
regulations. However, we are not dealing with such a question in this rulemaking action.
Instead, we are dealing with fact specific analyses regarding whether and how the CAA's
regional haze requirements are met in Texas, and our actions in this case are consistent with the
CAA and the regional haze requirements, as well as the regional consistency regulations.
While comments allege that some principle of national uniformity requires different actions in
this case, we are reasonably and consistently relying on our authority to exercise our judgment,
based on the specific facts at hand, in reviewing SIP actions. Here, we are exercising judgment
within the parameters laid out in the CAA and our regulations. Because this is a SIP review
action, we believe that we are not only authorized, but required to exercise independent technical
judgment in evaluating the adequacy of the State's regional haze SIP, just as we must exercise
such judgment in evaluating other SIPs. In evaluating other SIPs, we are constantly exercising
judgment about SIP adequacy, not just to meet and maintain the NAAQS, but also to meet other
requirements that do not have a numeric value. In the case of regional haze, Congress did not
establish NAAQS-like numeric standards by which to measure visibility improvement; instead, it
established a reasonable progress standard and required that we assure that such progress be
achieved. Here, we are exercising judgment within the parameters laid out in the CAA and our
regulations.
The SIP process anticipates a degree of flexibility in application of controls and requirements to
allow each State to choose its own methods by which to comply with the requirements of the
CAA and regulations. Likewise, when we must disapprove a SIP and issue a FIP, that same
358 We have recently proposed revisions to the regional consistency rules at "Amendments to Regional Consistency
Regulations" to address the situation over intercircuit nonacquiescence. (80 FR 50250, August 19, 2015).
465
-------
flexibility applies to allow for a program tailored to the particular State. In this action, we
proposed to approved sections of the SIP and, for necessary portions of the SIP that were
disapproved or missing, we proposed FIP requirements necessary to bring Texas into compliance
with the regional haze requirements. In doing so, we applied the same processes and general
knowledge used in other FIPs and followed the regional haze regulations consistently in
proceeding with our action. Commenter's concerns arise not from inconsistent application or
interpretation of regulations, but rather, Commenter objects to our state-specific decisions that
we consider necessary to ensure compliance with the regional haze program. The regional haze
program is not a "one size fits all" program, and the actions taken here are "as consistent as
reasonably possible" in terms of how we reviewed Texas' SIP and proposed our FIP while taking
into account the needs for flexibility in building a program designed upon state-specific needs.
However, contrary to the commenter's assertions, even if our fact-specific determinations
regarding the Texas SIP and FIP in this matter were found to be inconsistent with other regional
haze actions, this action would not be in violation of EPA's regional consistency regulations.
Those regulations actually allow for some variation between different regional actions.
Specifically, 40 CFR 56.5(b) provides that regional officials can seek concurrence from the EPA
headquarters with respect to any interpretations of the CAA, rule, regulation, or guidance in an
individual action that "may result in inconsistent application among the regional Offices."
Officials and staff from Region 6 have worked closely with EPA headquarters throughout the
proposed and final actions regarding the Texas and Oklahoma regional haze requirements,
including in the analysis and conclusions contained in the SIP and FIP determinations included
in this final rule. Moreover, headquarters' concurrence in the decisions contained in this action
is inherent in that fact that this final action is being issued by headquarters and signed by the
EPA Administrator.
We address comments alleging inconsistency with our Idaho and Nebraska actions elsewhere
within this section. We disagree that our action is in contravention of National Environmental
Development Association's Clean Air Project v. EPA as explained in our response above to the
NEDA CAP case. The Sam Coleman Declaration is discussed in the consideration of visibility
section of our responses.
Comment: Luminant stated that there is no requirement in the statute, regulations, or guidance
that Texas consider the visibility benefit from the implementation of individual controls in the
manner EPA would—or even to consider visibility at all in its four-factor analysis. Indeed, EPA
has approved other states' four-factor analyses, noting specifically that they did not perform this
type of visibility analysis.443
CCP stated that visibility is not a specific statutory factor to consider prior to the establishment
ofRPGs under CAA Section 169A. See 77 Fed. Reg. 20,894, 20,934 (Apr. 6, 2012) ("Nevada
SIP Approval") ("As we have noted, our regulations require consideration of four factors in
reasonable progress determinations; visibility improvement is not one of the specified factors.").
EPA previously rejected similar cumulative visibility improvements of 0.254 dv and 0.273 dv in
the New York SIP as too "small" to justify controls. 77 Fed. Reg. 24,818.
466
-------
GCLC noted, even if EPA did have the ability to impose a fifth "visibility factor," Texas' choice
of a 0.5 deciview ("dv") threshold as a benchmark for total visibility improvement was entirely
reasonable. For example, in recently reviewing and approving Idaho's reasonable progress goals,
EPA "independently evaluated whether there are reasonable control measures available for
sources located within Idaho's regulatory jurisdiction" and concluded that facilities with visibility
impacts of 0.5 dv or less at the nearest Class I area were "relatively small."26 Therefore, EPA
ultimately concluded in Idaho that additional controls for "reasonable progress purposes [were]
not reasonable at [that] time, because even though there [were] cost effective controls identified,
visibility improvement [was] anticipated to be relatively small."27 Luminant submitted similar
comments and added that, in finalizing its approval and responding to comments, EPA again
confirmed that, even though "several of the Idaho stationary sources have visibility impacts
between 0.3-1.3 deciviews (dv)," those impacts were not a "significant contribution to visibility
impairment" that warranted reasonable progress controls.705 EPA has used this same threshold in
other states to conclude that "reasonable progress controls" are not warranted 706
CCP stated that visibility may be appropriately considered on a cumulative basis, as TCEQ did
for all sources that are candidates for control. Using a cumulative approach, Texas appropriately
concluded there were insignificant cumulative visibility benefits, measured in deciviews, from
requiring additional controls. See 76 Fed. Reg. 74,387 Table 10 (identifying estimated
deciview improvements ranging from 0.16 dv in Big Bend to 0.36 dv in Wichita Mountains).
Footnotes:
443 See, e.g., 78 Fed. Reg. 10,546, 10,553 (Feb. 14, 2013) (approving Alaska's reasonable progress goals and
recognizing in response to comments that "the SIP submission does not specifically identify the contribution of coal-
combustion sources to visibility impairment inDenali National Park . .. ."); 77 Fed. Reg. 70,693, 70,702 (Nov. 27,
2012) (approving New Mexico's reasonable progress analysis that did not evaluate the contribution from individual
EGUs). See also WildEarth Guardians v. EPA, 770 F.3d at 944 (affirming EPA's approval of New Mexico's
reasonable progress analysis and holding: "Neither the Clean Air Act nor the Regional Haze Rule requires source-
specific analysis in the determination of reasonable progress.").
26 Idaho SIP Approval Proposal, 77 Fed. Reg. at 30256.
27 Id.
705 77 Fed. Reg. at 66,930-31.
706 See, e.g., 77 Fed. Reg. 30.454, 30,464 (May 23, 2012) (Oregon).
Response: As we discuss in our proposal,359 Texas basically employed the following approach
assessing reasonable progress controls:
• Texas narrowed the scope of the control analysis to point sources of NOx and SO2, and
developed a list of potential controls and costs associated with those controls to inform
their four factor analysis.
• It used the control strategy analysis developed by CENRAP as the starting point for its
analysis and further eliminated potential sources based on a series of screening thresholds
(e.g., cost, distance from a Class I Area, Q/d, previous analysis, etc.).
• It calculated the total cost of the controls that could be applied to the remaining sources
that were within specific areas of influence of individual Class I Areas.
359 79 FR 74835.
467
-------
• It constructed a subset of SO2 and NOx controls for particular facilities and calculated the
total annualized cost of those controls.
• Texas next weighed the four statutory factors in determining the reasonableness of
additional controls, using cost as a key factor.
• Texas then essentially compared the total annualized cost of the SO2 and NOx controls,
their total projected emission reductions and their total visibility impacts at a number of
Class I areas and determined that no controls were reasonable.
Also as we discuss in our proposal, we noted a number of flaws in Texas' analysis. Perhaps the
most critical flaw in Texas' analysis, was as we described in our proposal:360
The TCEQ constructed a large potential control set consisting of a mix of large
and small sources, located at various distances from Class I areas, with a large
geographical distribution. Because of the variation in size, type, and location of
these sources, the potential to impact visibility and potential benefit from controls
at a given Class I area can vary greatly between the identified sources. This
potential control set identified by the TCEQ included controls on some sources
that would likely result in significant visibility benefits, but also included controls
on many sources with much less anticipated visibility benefits. Because it only
estimated the visibility benefit of all the controls together, the TCEQ was not able
to assess the potential benefit of controlling individual sources with significant,
and potentially cost-effective, visibility benefits. Also, we believe that individual
benefits were masked by the inclusion of those controls with little visibility
benefit that only served to increase the total cost figures.
We address the consideration of visibility in a state or federal reasonable progress analysis in
another response to comment within this document.
When considering visibility, it is quite possible that a state with low overall impacts could assess
the totality of the visibility impacts of all of the sources within its borders, in consultation with
other states, and conclude that no controls are necessary for a specific planning period. It is also
possible that a state could proceed in this assessment by analyzing the individual source
contributions to visibility impairment from its sources on Class I areas. Other approaches are
also possible, but regardless of the approach taken, the state must engage in some rational
method for making this assessment that complies with the requirements in the regional haze rule.
As we discuss above, Texas' approach was highly flawed. Texas' cumulative approach
effectively masked the effect of controlling those sources or group of sources with the largest
visibility impacts. As we demonstrated in our proposal, when Texas facilities were analyzed
separately, cost effective controls were identified. This flaw, considered with the other flaws we
identified in Texas' four-factor reasonable progress analysis, caused us to conclude that Texas'
reasonable progress demonstration under Section 51.308(d)(l)(i)(A) was not approvable.
Luminant states that we have approved other states' four-factor analyses, noting specifically that
they did not perform this type of visibility analysis [individual source assessment] and cites to
360 79 FR 74838.
468
-------
the Alaska and New Mexico SIPs. With regard to the Alaska SIP, Luminant references the
quote, "the SIP submission does not specifically identify the contribution of coal-combustion
sources to visibility impairment in Denali National Park . . . The full quote is reproduced
below:
So while the SIP submission does not specifically identify the contribution of coal
combustion sources to visibility impairment in Denali National Park, it does
demonstrate that wildfires are the major source of PM2.5 in the State, that
wildfires have the greatest potential to impact visibility in Denali, and that
wildfires are the major source of OMC on the worst visibility days in Denali
National Park [emphasis added].
As the above quote indicates, the regional haze problem in Alaska is fundamentally different
than in Texas. As we discuss in our proposal, the CENRAP modeling, monitoring data and other
technical analyses demonstrated thatNOx and SO2 are the primary causes of haze at the Wichita
Mountains with SO2 from point sources being the predominant driver. It also showed that SO2
point sources in Texas were a significant contributor to the haze at the Wichita Mountains.361
Consequently, the Alaska approach, which was based on the recognition that wildfire was having
the greatest potential to impact the Denali Class I area, would not be appropriate. In addition, the
Alaska SIP does include the BART analysis and control of the Healey Unit 1.
CCP stated that visibility is not a specific statutory factor to consider prior to the establishment
of RPGs under CAA Section 169A and cited to 77 Fed. Reg. 20,894, 20,934 (Apr. 6, 2012)
which it refers to as the "Nevada SIP Approval," quoting, "As we have noted, our regulations
require consideration of four factors in reasonable progress determinations; visibility
improvement is not one of the specified factors." The quote that CCP reproduces comes from
our final action on the North Dakota regional haze SIP and the full quote is as follows:362
As we have noted, our regulations require consideration of four factors in
reasonable progress determinations; visibility improvement is not one of the
specified factors. As we have indicated, when a state considers visibility
improvement as an additional factor in evaluating single-source control
options, that consideration must be reasonable in light of the explicit goals
established by Congress in CAA section 169A [emphasis added].
Thus, our statement in our North Dakota action actually supports the consideration of
visibility.363 Furthermore, we note that statement was made in the context of evaluating source-
specific controls (precisely the methodology we have employed in evaluating the Texas regional
haze SIP), as long as that evaluation is reasonable in light of the goals of the regional haze
program. Texas itself considered visibility impact and visibility benefit in its reasonable
361 79 FR 74871
362 77 FR 20934.
363 We were subsequently upheld on this point by the Eighth Circuit Court in North Dakota v. EPA. 730 F.3d 750,
766 (8th Cir. 2013). The Eighth Circuit acknowledged in. North Dakota v. EPA, States can take visibility
improvement into account when evaluating reasonable progress controls so long as they do so in a reasonable way
469
-------
progress and RPG four-factor analysis but as we note above, its approach was highly flawed and
consequently not approvable.
Like Texas, we considered visibility in our reasonable progress analysis. We believe that states
(or EPA when promulgating a FIP) can consider visibility when determining reasonable progress
in at least two ways. States can consider the visibility impacts of sources when determining what
sources to analyze under the four-factor framework. As such, states can develop screening
metrics that target those sources with the greatest visibility impacts for further analysis. Our
2007 guidance advocated this approach, and nearly all states, including Texas, used metrics like
Q/d to consider the potential visibility impacts of their sources and screen out those sources with
low visibility impacts.364 We followed this same approach in our FIP by using both Q/d and a
second metric based on a source's modeled percent contribution to total visibility impairment at
impacted Class I areas. If states or we could not consider visibility impacts as a way of
identifying which sources should be considered for additional controls, then states would have no
rational way to differentiate between hundreds of sources that vary in distance from Class I
areas, emit different visibility impairing pollutants in varying amounts, and are subject to diverse
meteorological conditions that affect the transport of visibility-impairing pollutants. The result
would be a cumbersome analysis encompassing hundreds of sources (or in the case of Texas,
well over a thousand), many of which may have little if any impact on visibility in Class I areas.
Congress could not have intended such an incongruous result. Second, once a universe of
sources has been identified for analysis, we believe that States have the option of considering the
visibility improvement that will result from potential control options when weighing the four
statutory factors. In summary, if States were not permitted to consider visibility improvement
when conducting their control determinations, then States would have to require all cost-
effective controls (assuming no limiting energy or non-air quality environmental impacts)
regardless of whether some of those controls would be more beneficial than others.
GCLC stated that even if we did have the ability to impose a fifth "visibility factor," Texas'
choice of a 0.5 deciview ("dv") threshold as a benchmark for total visibility improvement was
entirely reasonable. As we discuss in our proposal:365
In evaluating and dismissing the estimated visibility benefit from the control set
identified by the TCEQ, the TCEQ states that the estimated benefit is not
perceptible (less than 1 dv) and that it is less than 0.5 dv, the threshold used under
BART requirements used to determine if a facility contributes to visibility
impairment. The 0.5 dv BART threshold referred to applies to the maximum
anticipated visibility impact on a single day due to the short-term maximum actual
baseline emissions from a single facility, compared to clean background
364 For example, in VISTAS states, to select the specific point sources that would be considered for each Class I
area, VISTAS first identified the geographic area that was most likely to influence visibility in each Class I area and
then identified the major SO2 point sources in that geographic area. The distance-weighted point source SO2
emissions (Q/d) were combined with the gridded extinction-weighted back-trajectory residence times. The distance-
weighted (Q/d) gridded point source S02 emissions were then multiplied by the total extinction-weighted back-
trajectory residence times on a cell-by-cell basis and then normalized. VISTAS Area of Influence Analyses, 2007,
is available in the docket for this action.
365 79 FR 74840.
470
-------
conditions. The reasonable progress analysis presented by the TCEQ
contemplates the visibility benefit anticipated for an average tpy emission
reduction (as opposed to the impact from the total short-term maximum emissions
from the sources) averaged across the 20% worst days, which would be
anticipated to be significantly lower. See our FIP TSD for a detailed discussion
of the different metrics and modeling typically used for BART and reasonable
progress analyses. Furthermore, in a situation where the installation of BART
may not result in a perceptible improvement in visibility, the visibility benefit
may still be significant, as explained by the Regional Haze Rule:366
Even though the visibility improvement from an individual source
may not be perceptible, it should still be considered in setting
BART because the contribution to haze may be significant relative
to other source contributions in the Class I area. Thus, we disagree
that the degree of improvement should be contingent upon
perceptibility.
As we stated in our Oklahoma final decision:367
Given that sources are subject to BART based on a contribution
threshold of no greater than 0.5 deciviews, it would be inconsistent
to automatically rule out additional controls where the
improvement in visibility may be less than 1.0 deciview or even
0.5 deciviews. A perceptible visibility improvement is not a
requirement of the BART determination because visibility
improvements that are not perceptible may still be determined to
be significant.
We further discuss in depth why Texas' choice of a 0.5 dv visibility threshold, including the
manner in which it was applied, was not appropriate in our response to comments on
consideration of cost versus visibility.
In support of its assertion, GCLC cites to our proposal to approve portions of Idaho's regional
haze SIP. GCLC states that therein, we stated that we "independently evaluated whether there
are reasonable control measures available for sources located within Idaho's regulatory
jurisdiction" and concluded that facilities with visibility impacts of 0.5 dv or less at the nearest
Class I area were "relatively small." Therefore, concludes GLCC, we concluded in Idaho that
additional controls for "reasonable progress purposes [were] not reasonable at [that] time,
because even though there [were] cost effective controls identified, visibility improvement [was]
anticipated to be relatively small." For Idaho, we conducted a Q/d analysis (as we did for
Texas). However, in Idaho's case, we relied on CALPUFF modeling results to assess visibility
impacts of the identified sources. In assessing visibility benefits from potential controls, Texas
extrapolated results from CAMx modeling to estimate the visibility improvement due to all the
366 70 FR 39130.
367 76 FR 81739.
471
-------
identified controls in their analysis.368 As we have discussed in our FIP TSD and separate
response to comments, the results of our CAMx modeling (or Texas' extrapolated results based
on CAMx modeling) cannot be directly compared to the results of CALPUFF modeling, which
was used in the vast majority of other BART determinations and some reasonable progress
determinations, because of differences between the models, model inputs, and metrics used.369
Many of these differences result in CAMx modeled visibility impacts and benefits that are much
lower than the CALPUFF modeled visibility impacts and benefits relied on in other actions.
Consequently, we do not believe our proposed action in Idaho conflicts with our proposed
Texas/Oklahoma FIP. We address Luminant's footnote to our Oregon action in our response to
another comment.
CCP alleges inconsistency with our New York action. This reference, as with other similar
references made by commenters alleging actions in which we rejected visibility results as being
too small to justify the expense of controls, used CALPUFF modeling. As we explain in our FIP
TSD370 and elsewhere in our response to comments, technical issues related to the applicability
of CALPUFF barred us from using it in our proposed action. We also explained why the results
from our proposed model CAMx, cannot be compared to CALPUFF results. As a consequence,
we disagree with CCP.
In light of the above, we do not agree that Texas' choice of a 0.5 dv threshold as a benchmark for
total visibility improvement was reasonable.
Comment: Luminant stated that EPA's second guessing of the consultation between Oklahoma
and Texas is in stark contrast to how EPA has treated other states. For example, EPA's action on
Mississippi's regional haze SIP illustrates the proper deference owed to an agreement between
states as to the level of reductions necessary. Mississippi considered potential controls on Plant
Watson due to possible impacts at the Breton Wilderness area in Louisiana.534 Following the
consultation process between Louisiana and Mississippi, Louisiana asked for no further
reductions from Mississippi.535 EPA found that because the two states came to the agreement
that no further reductions were required, it was proper for Mississippi to not impose further
controls on Plant Watson.536 In the final rule, EPA stated: "MDEQ has met its obligations with
regard to obtaining emissions reductions since no additional control measures specific to
Mississippi were identified by the Louisiana reasonable progress analysis."537 Further, "[sjince
Breton is in Louisiana, EPA believes that Mississippi appropriately relied on Louisiana's
368 See our extended discussion of the difference between CALPUFF and CAMx modeling in our FIP TSD,
beginning on page A-35 and our response to comments on modeling.
369 See the Modeling section of the this document and our FIP TSD, beginning on page A-35, in which we explain
why key differences in CALPUFF for BART and CAMx modeling for RP preclude the comparison of their
respective results. Some of the major differences are: (1) CALPUFF uses maximum 24-hour emission rates, while
CAMx uses annual average emission rates; (2) CALPUFF focuses on the day with the 98th percentile highest
visibility impact from the source being evaluated, whereas CAMx focuses on the average visibility impacts across
the 20% worst days regardless of whether the impacts from a specific facility are large or small; and (3) CAMx
models all sources of emissions in the modeling domain, which includes all of the continental U.S., whereas
CALPUFF only models the impact of emissions from one facility without explicit chemical interaction with other
sources' emissions.
370 FIP TSD, Appendix A. See discussion beginning on page A-35.
472
-------
determination of which sources to prioritize for reasonable progress control evaluation during
this implementation period."538
Response: Each review of a regional haze SIP is fact specific, making it difficult to apply the
type of broad generalizations made by Luminant in its comment. Luminant is correct that EPA
accepted the assessments of Louisiana and Mississippi that no additional emissions reductions
were necessary from sources in Mississippi to ensure reasonable progress at Breton Wilderness
area in Louisiana. That EPA deferred to the States' judgment as to the necessary measures in the
Mississippi SIP, however, does not mean that the Agency's disapproval of the consultation
between Texas and Oklahoma is improper.
One aspect of the consultation between Mississippi and Louisiana worth noting is that the two
states were members of two different RPOs. Mississippi participated in VISTAS and used for its
reasonable progress analysis the VISTAS' area of influence (AOI) methodology to identify those
sources with potentially significant visibility impacts on Class I areas in other states. Plant
Watson was one of two facilities that met the AOI criteria for potential visibility impacts at
Breton NWA in Louisiana. Mississippi evaluated Plant Watson for potential controls but found
that Plant Watson was subject to CAIR and indicated in its SIP that controls for NOx and SO2
were planned for the larger of its two units.371 Mississippi evaluated these projected reductions
and the analysis developed in promulgating CAIR when considering the four reasonable progress
factors. Mississippi also considered and provided information on the four factors for the other
identified facility, Dupont DeLisle. Mississippi notified Louisiana (via consultation) about its
findings. Louisiana, a member of CENRAP, relied on a different modeling and methodology to
identify possible impacts from other states on Breton. Specifically, Louisiana relied on the
source apportionment results from the CENRAP modeling and the Alpine Geophysics report
developed for CENRAP. CENRAP source apportionment modeling showed that emissions from
Mississippi were projected to contribute 5.26% of the total visibility impairment at Breton in
2018 compared to 24.7% from Louisiana emissions. The CENRAP modeling used by Louisiana
in establishing the RPGs for Breton assumed the anticipated SO2 and NOx controls at Plant
Watson and projected significant visibility improvement at Breton.
During consultation, Louisiana and Mississippi discussed reductions at BART facilities.
Mississippi identified two subject to BART sources (Mississippi Phosphates and Chevron
refinery - Pascagoula) within 50 km of Breton. Due to BART and consent decree requirements
on these sources, significant reductions in visibility impairment due to these nearby facilities was
anticipated at Breton. Louisiana informed Mississippi (again via the consultation process) that
regional modeling and other findings based on existing and proposed controls arising from local,
state, and federal requirements indicated that Breton Island is expected to meet the uniform rate
of progress goal and that it did not need additional reductions from Mississippi beyond those
reductions anticipated due to BART, CAIR and other on-the-book requirements in order to
ensure reasonable progress at Breton for the first planning period. Mississippi agreed that no
additional controls were identified or needed for making reasonable progress at Breton for this
planning period.
371 In contrast, as Texas itself has noted in its SIP, the IPM model analysis used by CENRAP predicts that by 2018
EGUs in Texas will purchase approximately 125,000 tpy of emissions allowances from out of state. See Texas
Regional Haze SIP, page 10-9.
473
-------
There are key differences in the factual background informing the consultations between
Louisiana and Mississippi and those of Oklahoma and Texas. As explained more fully
elsewhere in our final action and this document, information made available by CENRAP
demonstrated the significant impact of Texas emissions, particularly Texas point sources, at the
Wichita Mountains and the potential for cost-effective controls at some of these sources.
Recognizing this, Texas used the CENRAP analysis as a starting point, and performed
supplemental analyses for both its reasonable progress and long-term strategy demonstrations.
Visibility impacts from Texas sources on Wichita Mountains in Oklahoma were projected by
CENRAP to be a significant portion of the total visibility impairment (28%). And there was, at
the very least, a partial remedy in the form of cost effective controls on those Texas sources with
the largest impacts on visibility.
Although Oklahoma and Texas were aware that Texas sources significantly impact the visibility
at the Wichita Mountains and that cost-effective controls resulting in large emission reductions
were likely available on some of these sources, Texas determined that no additional controls
were reasonable based upon its technical analysis and Oklahoma did not pursue this point in its
consultations by either asking Texas to further investigate these sources or request additional
reductions from Texas, despite noting that additional reductions in Texas would be needed to
make progress towards the natural visibility goal. The analysis and information relied on by
Texas and Oklahoma during consultations does not support the determination that no additional
controls in Texas were reasonable or support Oklahoma's determination that the established
reasonable progress goal is proper.
As discussed in our proposed action, because Oklahoma adopted a reasonable progress goal that
provides for a slower rate of visibility improvement than the uniform rate of progress,
51.308(d)(l)(ii) requires a demonstration based on the four factors that the established goal is
reasonable and that meeting the uniform rate of progress is not reasonable. Oklahoma must also
consider the uniform rate of progress and the emission reductions measures necessary to achieve
it for this planning period.372 Given the significant impact from Texas sources, and EGUs in
particular, Oklahoma could not reasonably consider all the emission reductions needed to meet
or approach the URP without considering emission reduction measures available for those
sources in Texas that contribute the most to visibility impairment at Wichita Mountains.
Recognizing that the information made available by CENRAP indicated the significant impact of
Texas emissions and potential for cost-effective controls, Texas used the CENRAP analysis as a
starting point, and performed supplemental analysis for both its reasonable progress and long-
term strategy demonstrations. However, that additional technical analysis performed by Texas
was flawed and therefore did not provide the type of information necessary to fully evaluate the
reasonableness of controls at Texas sources with the largest potential to impact visibility at its
own Class I areas and the Wichita Mountains. Allowing this lack of adequate information to
continue was a critical misstep for ODEQ in setting its reasonable progress goals, and a critical
misstep for Texas when determining its fair share of emissions reductions under the long-term
strategy requirement. The plain language of the CAA requires that states consider the four
factors used in determining reasonable progress in developing the technical basis for the
372 40 CFR 51.308(d)(l)(i)(B)
474
-------
reasonable progress goals both in their own Class I areas and downwind Class I areas. Such
documentation is necessary so that interstate consultations can proceed on an informed basis, and
so that downwind states can properly assess whether any additional upwind emissions reductions
are necessary to achieve reasonable progress at their Class I areas. Therefore, Texas had an
obligation to provide appropriate information to Oklahoma so it could establish a proper progress
goal for the Wichita Mountains. Further, Texas had an obligation to conduct an appropriate
technical analysis, and demonstrate through that analysis (required under (d)(3)(ii)), that it
provided its fair share of emissions reductions to Oklahoma. In the review of the Texas and
Oklahoma SIPs, we were compelled to disapprove the Texas technical analysis because of its
flaws and consequently perform our own technical analysis. In summary, Texas was required
through the consultation process to provide Oklahoma the information it needed to establish its
reasonable progress goals for the Wichita Mountains, and it failed to do so.
In contrast, Louisiana, concluded that Breton Island was projected to be making reasonable
progress towards the goal of natural visibility conditions considering all on-the-book controls,
BART controls at Mississippi sources near Breton Island were anticipated to provide additional
reduction in visibility impairment, and the majority of the visibility impairment at Breton was
due to impacts from Louisiana sources. In consideration of all this information, Louisiana's
determination that no additional controls were necessary at this time to address the impact from
Mississippi sources was not unreasonable. Similarly, Mississippi's agreement with this
assessment was also not unreasonable.
Comment: CCP stated that the EPA is requiring more from Oklahoma and Texas regarding
consultation than it has previously required via rule, guidance and other SIP approvals. For
example, in Michigan, EPA approved interstate consultation efforts even when Michigan did not
offer additional controls for a Class I area not meeting its glide path until 2209. See Michigan
SIP Approval, 77 Fed. Reg. 46,917 (Aug. 6, 2012) ("By coordinating with the MRPO and
other RPOs, Michigan has worked to ensure that it achieves its fair share of overall emission
reductions"). In Kentucky, EPA concluded that Kentucky adequately addressed the consultation
requirements by determining that sources were meeting more stringent requirements than
regional MANE-VU recommendations. Kentucky SIP, 76 Fed. Reg. at 78,213. As previously
mentioned, EPA found that a level of 0.2 dv improvement was too low to justify additional
controls in Arkansas. See 77 Fed. Reg. 14,604, 14,625 (March 12, 2012). To the extent EPA is
inconsistently requiring more stringent requirements in Texas and Oklahoma than allowed
elsewhere, this is a violation of CAA regional consistency requirements. See 40 C.F.R. §
56.3(a) and (b); National Environmental Development Ass'n's Clean Air Project v. EPA, 752
F.3d 999 (D.C. Circ. 2013).
Response: We agree that EPA approved Michigan's interstate consultation efforts with
Minnesota and other states even though natural conditions would not be achieved for some time
at certain Class I areas in those states.373 CCP's comment appears to specifically reference
373 We determined in considering this comment that an error was made in the Federal Register notice addressing
EPA's review of the Michigan regional haze SIP cited by the commenter. The Federal Register notice states that
based on the rate of progress anticipated for Voyageurs National Park located in Minnesota, natural conditions on
475
-------
Voyageurs National Park, a Class I area in Minnesota. With regards to Michigan's impact on
visibility conditions at this Class I area, Minnesota determined (and Michigan agreed) that
Michigan sources were not a significant source of visibility impairing emissions for this planning
period. Emissions from Michigan accounted for less than 3% of the total visibility impairment at
either of the Minnesota Class I areas, and both states agreed that no additional reductions at
Michigan sources were necessary for reasonable progress at the Minnesota Class I areas.
During the regional haze planning process, Michigan also consulted with several northeastern
states in the MANE-VU RPO but ultimately determined that emissions from Michigan sources
contributed very little to the visibility impairment at Class I areas in Maine, New Hampshire,
New Jersey, and Vermont. Photochemical modeling showed that Michigan contributed no more
than 5% of the total visibility impairment at any other state's Class I area. The State also
concluded that significant reductions were anticipated from BART, and therefore additional
controls beyond anticipated reductions from CAIR, BART and other on-the-book measures were
not needed to address Michigan's impacts on other states. We note that EPA disapproved
Michigan's BART determinations at two facilities and promulgated a FIP requiring additional
reductions in emissions from these sources. This is in direct contrast to Oklahoma and Texas,
where the CENRAP analyses show significant impacts from Texas sources are impeding
progress towards natural visibility conditions in Oklahoma and that visibility impacts from Texas
sources on Wichita Mountains in Oklahoma were projected to be a very significant portion of the
total visibility impairment (28%) at Wichita Mountains, even after anticipated reductions due to
CAIR compliance and other "on the book" requirements.374 Texas, acknowledging this
information, performed its technical analysis to identify reasonable controls and in developing its
long-term strategy.
With respect to Kentucky's SIP, our approval of the consultation elements was consistent with
our other actions. The MANE-VU states of Maine, New Hampshire, New Jersey, and Vermont
identified Kentucky as a contributor to visibility impairment in their states and requested that
Kentucky reduce its SO2 emissions. In particular, these states requested that Kentucky require
90% control on 14 of its EGU stacks. Kentucky, a member of the VISTAS RPO, evaluated its
sources for impacts on other state's class I areas and determined that no emission units had an
impact over the threshold selected by Kentucky and many of the VISTAS states to identify units
for additional analyses. Although Kentucky did not agree that its sources had significant impacts
on visibility in the northeast, in response to the MANE-VU states' request, Kentucky noted that
all but one of the 14 identified EGU stacks either had existing SO2 controls or would be
controlled by 2015. Again, this is in contrast to the circumstances and decisions reached by
Oklahoma and Texas.
Regarding the comment that EPA found that a level of 0.2 dv improvement was too low to
justify additional controls in Arkansas, CCP misrepresents or fails to understand EPA's decision,
the 20% worst days would occur by 2209. See 77 FR 46,912, 46,919. EPA intended to state, however, that Seney
Wilderness Area, a Class I area in Michigan would attain natural conditions by 2209 if the improvement in visibility
at this area over the period of the first Michigan regional haze SIP were projected forward. The error, however, is
not significant in the context of the comment above as natural conditions on the 20% worst days at Voyageurs
National Park is not projected to occur until 2177
374 As discussed elsewhere, there are no additional reductions due to BART on non-EGU sources in Texas.
476
-------
and its comment does not provide the proper context about the decision as it relates to
appropriate interstate consultation. We note that the estimated visibility impairment contribution
from all sources in Arkansas on Wichita Mountains was 2.3% of the total extinction. For proper
context, compare this with the estimated 1.5% total contribution from just a sinsle source (i.e.
Big Brown) at Wichita Mountains. Modeling results showed that complete removal of
Arkansas' contribution would result in the 0.2 dv improvement in 2018 at Wichita Mountains.
Both Arkansas and Oklahoma agreed, through consultation, that additional reductions from
sources in Arkansas for reasonable progress at Wichita Mountains was unnecessary, and EPA
agreed that the consultation was satisfactory. The same conclusion could not be made regarding
the consultation between Oklahoma and Texas as the magnitude of emission reductions and
associated visibility benefits to Wichita Mountains due to controls on Texas' sources was much
more significant relative to Arkansas' impact on Wichita Mountains.
We accordingly do not agree that EPA has been inconsistent in its review of the regional haze
SIPs or that we have violated the federal regional consistency rule requirements. We discuss
National Environmental Development Ass'n's Clean Air Project v. EPA above.
Comment: The TCEQ stated that the EPA has viewed similar consultations in other state SIPs,
using the same CENRAP information, as meeting the RHR requirements for long-term strategy
consultations. A case in point is Arkansas's regional haze plan. The CENRAP modeling that the
EPA now finds lacking for Texas and Oklahoma's consultation was perfectly fine for Arkansas.
It demonstrated that visibility impairment from Arkansas sources at Hercules Glades in Missouri
was projected to increase during 2002-2018. In consultations with Missouri, Arkansas made no
commitment for additional controls beyond those already factored into CENRAP's modeling for
2018. All states agreed with this determination, including Missouri. Yet, with no further
explanation, the EPA approved Arkansas' consultation and its determination that no additional
controls were necessary, as consistent with the RHR, even though the data that was clearly
available to everyone showed impairment at Hercules Glades due to Arkansas' sources would
increase (76 FR 64186, 64216).
Response: The comment by TCEQ does not accurately reflect our proposed action on the
Arkansas regional haze and does not take into account the context of our action on the interstate
consultation between Arkansas and Missouri. Primarily, we note that the analyses done by
CENRAP showed that the visibility impairment at Hercules Glades in Missouri attributable to
emissions from Arkansas' sources was projected to decrease during the 2002-2018 time period.
(See Figures E-6c and E-6d in AR020.0121 CENRAP-Appdx-A-F-TSD-Draft3_e51.pdf
available in docket EPA-R06-QAR-2015-0189). However, as shown in Table 12 of the
proposed action that the commenter is referencing, Arkansas' percentage contribution to the total
impairment at Hercules-Glades was projected to increase. This is because, in a relative sense,
the visibility impairment contribution from other geographic areas is decreasing faster than that
of Arkansas, and thus, in percentage terms, the contribution from Arkansas is increasing. Both
Missouri and Arkansas appropriately factored the visibility improvement resulting from on-the-
book requirements and anticipated BART reductions, as demonstrated in the CENRAP modeling
analysis, into their consultation process resulting in their mutual agreement that Arkansas was
achieving its share of reductions at Hercules-Glades and that no additional controls were needed
477
-------
on Arkansas' sources. Missouri and Arkansas determined that their Class I areas are making
reasonable progress towards the goal of natural visibility during the first planning period and that
it was reasonable to not request reductions from other states at this time.
Comment: Xcel Energy stated that the EPA is requiring significantly more from Oklahoma and
Texas with respect to the consultation process than it has previously required via rule, guidance
and other SIP approvals. In adopting its regional haze regulatory guidelines, EPA noted that, if
States determine that no further controls are needed in a particular planning period, States must
merely document "any consultations with other States in support of their conclusions." RHR, 64
Fed. Reg. at 35,721-22. Texas and Oklahoma did this. See Texas SIP, at App. 4-2.
Xcel Energy noted that the EPA also appears to be arbitrarily taking a harder line in reviewing
Texas' and Oklahoma's consultation efforts than it has taken with other states. In other recent
regional haze SIP actions, EPA concluded that interstate consultation requirements were met,
even though there was substantially less consultation than the discussions between Oklahoma
and Texas:
• In Michigan, EPA found adequate consultation even when Michigan did not offer
additional controls for a Class I area not meeting its glide path until 2209. See Proposed
Michigan SIP Approval, 77 Fed. Reg. at 46,917 ("By coordinating with the MRPO and
other RPOs, Michigan has worked to ensure that it achieves its fair share of overall
emission reductions").
• In Arkansas, EPA concluded that Arkansas met consultation requirements based on three
calls with states and concurrence in the conclusion that controls in other states are not
necessary. Proposed Arkansas SIP Approval/Disapproval, 76 Fed. Reg. 64,186,64,196
(Oct. 17,2011).
• In Kentucky, EPA found that Kentucky adequately addressed the consultation
requirements by determining that sources were meeting more stringent requirements than
regional MANE-VU recommendations. Kentucky SIP Approval, 76 Fed. Reg. at
78,213.
Xcel Energy argued that the EPA cannot make inconsistent conclusions on the adequacy of the
consultation process between Oklahoma and Texas as compared to other States without any
reference to its rules, guidance and prior SIP approvals.
Footnotes:
9 Although EPA stated in 2006 that it planned to issue more specific protocols for state consultation, EPA never released
such guidance. See EPA, Additional Regional Haze Questions, at 11 (2006).
Response: Xcel cites to two pages from the 1999 Regional Haze Rule that contain EPA's
responses to comments related to EPA's proposal to require all States to submit regional haze
SIPs, including those without Class I areas. After explaining the statutory and technical basis for
its final decision that all States contain sources whose emissions are reasonably anticipated to
478
-------
contribute to visibility impairment and must therefore submit regional haze SIPs, we summarized
the overall import of that decision:
As noted in the proposal, EPA is not specifying in this final rule what specific
control measures a State must implement in its initial SIP for regional haze. That
determination can only be made by a State once it has conducted the necessary
technical analyses of emissions, air quality, and the other factors that go into
determining reasonable progress. The EPA also recommends the coordination of
resulting strategies for regional haze with strategies needed to attain the PM2.5
NAAQS. The EPA anticipates that as a result of the more refined analyses
required by this rule, some States may conclude that control strategies specifically
for protection of visibility are not needed at this time because the analyses may
show that existing measures are sufficient to meet reasonable progress goals. The
EPA is requiring States to document their analyses, including any consultations
with other States in support of their conclusions that further controls are not
needed at this time. The EPA believes that there is more than sufficient evidence
to support our conclusion that emissions from each of the 48 contiguous States
may be reasonably anticipated to cause or contribute to visibility impairment in a
Class I area.
64 FR at 35721. On the following page, in response to comments that we had sufficient
information to exclude all or part of some States from the regional haze program, we stated:
[T]he EPA believes that a State wishing to demonstrate that it does not contribute
to visibility impairment in any Class I area will need to provide information
showing that it has consulted with other potentially affected States to assist EPA in
assuring that the State's demonstration is not contradicted by evidence presented
by other States
64 FR at 35722. Read in context, there is nothing in the preamble language cited by Xcel to
suggest that EPA viewed its regulations as requiring States to "merely document' the fact that
consultation took place to satisfy the requirements that each state obtains it share of emission
reductions needed for reasonable progress. As the first excerpted paragraph makes clear, states
that conclude that existing control measures are sufficient to ensure reasonable progress in a
downwind Class I area must document their analyses supporting this conclusion, in addition to
documenting the consultation process. As to the second paragraph, the preamble language
addresses the possibility that a State may have information demonstrating that it does not, in fact,
contribute to visibility impairment in a neighboring State's Class I area. Again, nothing in this
language suggests that all that is required is evidence that the State consulted with the
neighboring state.
We have addressed Xcel's comments regarding Michigan and Kentucky above and largely
addressed its comments regarding Arkansas. With respect to Xcel's comment implying that
because other states did not have as many consultative discussions as did Oklahoma and Texas,
there was "less" consultation among those other states, we disagree that the interstate
consultation requirements are measured by or satisfied by the number of separate discussions or
479
-------
the time spent in discussions among states. To satisfy the interstate consultation requirements,
two (or more states) must reach appropriate conclusions about what constitutes reasonable
progress at a particular Class I area based on careful consideration of the facts. As the
commenter indicates, we did indeed take a hard look at the consultation efforts between Texas
and Oklahoma; however, we did so because both states — while acknowledging the very
significant visibility impact from Texas sources at Wichita Mountains, and the lack of progress
towards natural visibility conditions, and the availability of cost-effective controls on some of
these sources— nevertheless determined that no additional controls were necessary. Their
analysis did not yield enough information to support this conclusion. The number of discussions
between Texas and Oklahoma was not relevant to our proposed decision.
Comment: Luminant stated that EPA's actions with respect to Nebraska and South Dakota
unravel EPA's new theory of the need and justification for "simultaneous" action on Texas's and
Oklahoma's regional haze plans, and its interpretation of 40 C.F.R. § 51.308(d)(3)(ii) that
underlies the theory (discussed below). Commenters on EPA's action for Nebraska asserted that,
where one state's RPG is determined to be "not sufficient" (i.e., not "approvable"), "each state
participating in the regional planning process for the applicable Class I area [must] be required to
re-evaluate their LTS and make appropriate revisions to ensure they met their apportionment of
emission reduction obligations necessary for achieving reasonable progress."565 EPA disagreed
with this comment, and explained that the process works in a fundamentally different way. EPA
explained that states look to "air quality modeling performed by the RPOs" "[t]o set RPGs."566
EPA further explained that there is "an inherent amount of uncertainty in the assumed emissions
from all sources" and that when a state's final action "deviate[s] from what was included in the
modeling," the remedy is for affected states to "consider asking [the contributing state] for
additional emission reductions" "during subsequent periodic progress reports and regional haze
SIP revisions."567
Luminant noted that EPA's proposed action here cannot be squared with its actions in connection
with the Nebraska and South Dakota SIPs. EPA claims here that "[t]o properly assess whether
Oklahoma had satisfied the reasonable progress requirements," it "had to review and evaluate
Texas' regional haze SIP before proposing action on Oklahoma's RPGs."568 But this is
fundamentally at odds with EPA's explanation in its Nebraska/South Dakota action that states
fulfill their statutory obligations by consulting and making assumptions together to develop their
regional haze SIPs, and then make adjustments in future planning periods as necessary. As
discussed above, EPA's prior explanation of the correct process is consistent with its regional
haze regulations, and its current and novel interpretation is not. Further, while EPA now claims
that "[i]n order to address these intricately intertwined issues between Oklahoma and Texas, it is
appropriate to review them simultaneously,"569 Oklahoma and Texas are no more "intertwined"
than any other two states that must consult over out-of-state impacts. Indeed, the long distances
between Texas sources and Oklahoma's Class I area—and the negligible visibility impacts
involved—make Texas and Oklahoma less "intertwined" than most other states. There is no
justification for EPA's approach here. And if EPA wants to require states to follow its new
approach to regional haze planning, it must amend its regulations to establish a new consultation
process, but it cannot impose one by fiat and use it to retroactively judge the Texas and
Oklahoma submissions.
480
-------
Luminant stated that EPA's proposed disapproval of Texas's long-term strategy is fundamentally
inconsistent with EPA's prior action on Nebraska's long-term strategy, underscoring the error in
EPA's proposal. Nebraska, like Texas, is a CENRAP state and participated with neighboring
states in the CENRAP regional planning process.54
GCLC noted that EPA's analysis directly conflicts with its recent decision on Nebraska's regional
haze SIP, where it found that Nebraska's SIP complied with the LTS requirements. In the SIP
development process, South Dakota (the state with Class I areas potentially impacted by
Nebraska) had the opportunity to comment on Nebraska's SIP, but "did not ask for additional
reductions from Nebraska."55 Given South Dakota's action, EPA found that Nebraska had
"demonstrate[d] that it has included all measures necessary to obtain its share of the emission
reductions needed to meet the RPGs for Class I areas where it causes or contributes to
impairment" and had met its LTS obligations to South Dakota.56 According to GCLC, an
important factual distinction exists between the Nebraska-South Dakota cooperation, compared
to the Texas-Oklahoma cooperation, further underscoring that EPA's proposal here is unjustified.
In the Nebraska SIP process, Nebraska represented to South Dakota that it would impose sulfur
dioxide ("SO2") emission limits on a specific BART emissions source that it ultimately did not
adopt.57 EPA, rather than disapproving of Nebraska's LTS for failing to include this measure that
South Dakota relied on, recommended addressing this issue in the second planning period.58
With regard to the Texas SIP, Texas' long-term strategy includes all of the actions it agreed with
Oklahoma to take, and thus there is no basis for EPA's proposed additional controls. But even if
there were, those issues should be addressed in the next planning period, consistent with EPA's
action on the Nebraska SIP. This lack of basis for EPA's action, as well as the inequality
between the different SIP review processes, simply does not make sense and is arbitrary and
capricious.
Luminant stated, that even where it is later determined that the contributing state's long-term
strategy is not, in fact, adequate to meet the established goal, that is "not grounds for
disapproving either [states'] SIP" and issuing a FIP, as EPA itself found in approving Nebraska's
Sip 616 Rather, the proper course is for the states "to consider whether other reasonable control
measures are appropriate to ensure reasonable progress during subsequent periodic progress
reports and regional haze SIP revisions. "617 Thus, EPA's attempt here to "simultaneously
conduct[] reasonable progress and long-term strategy analyses" is fundamentally at odds with the
regulations and EPA's prior application of those regulations.618
Footnotes:
547 Id. at 40,155.
55 Nebraska SIP Final Rule, 77 Fed. Reg. at 40155.
56-58 Id
614 64 Fed. Reg. at 35,735 ("The EPA expects that much of the consultation, apportionment demonstrations, and
technical documentation will be facilitated and developed by regional planning organizations. We expect, and
encourage, these efforts to develop a common technical basis and apportionment for long-term strategies that could
be approved by individual State participants, and translated into regional haze SIPs for submission to EPA.").
615 77 Fed. Reg. at 40,155.
616 Id.
617 Id. (emphasis added).
ei8 pip TSD at 5.
481
-------
619 See EPA, Pre-Meeting Materials for the EPA-FLM-RPO-States-Tribes Meeting on the Future of the Regional
Haze Program (Feb. 3,2015).
565 77 Fed. Reg. at 40,155.
566 Id.
567 Id. at 40,155-56 (emphasis added).
568 79 Fed. Reg. at 74,821.
569 Id. at 74,822.
Response: Luminant and GCLC comment that EPA's actions on Nebraska's and South
Dakota's regional haze SIPs are inconsistent with EPA's proposed action on Texas and
Oklahoma. Nebraska, like Texas, was found to impact visibility in a Class I area outside its
borders. In particular, although Nebraska does not have a Class I area within its borders, it
consulted with South Dakota during the regional haze planning process because of its potential
impacts on Class I areas within South Dakota. During the RPO process, Nebraska indicated that
it would require one of its large BART sources - Gerald Gentleman Station (GGS) - to install
scrubbers to reduce emissions of SO2 to meet the BART requirements. The modeling used to set
the RPGs for the two Class I areas in South Dakota assumed that GGS would reduce its SO2
emissions, in line with Nebraska's assumed BART determination.375 Thus, South Dakota
implicitly relied on SO2 emission reductions from GGS in setting its RPGs.
Nebraska ultimately decided to not require GGS to install SO2 controls for BART, but EPA, in
reviewing the Nebraska regional haze SIP, concluded that the State had made significant errors
in its BART determination for GGS and disapproved the State's BART determination for GGS.
EPA also disapproved Nebraska's long-term strategy to the extent that it relied on its deficient
SO2 BART for GGS to achieve its share of emission reductions needed to meet the reasonable
progress goals of other states' Class I areas. In addressing the deficiencies in Nebraska's
regional haze SIP, however, EPA did not require SO2 controls at GGS but rather relied on
Nebraska's participation in CSPAR to address these deficiencies.
EPA is currently reconsidering our action on Nebraska's long-term strategy. The Eighth Circuit
granted our request for a voluntary remand of our action on the Nebraska long-term strategy on
March 19, 2015.376 In our motion for a voluntary remand, we explained that "EPA is concerned
that its present explanation could potentially be construed in a manner that is inconsistent with
EPA's interpretation of the relevant statutory and regulatory requirements. Remand is therefore
appropriate so that EPA has the opportunity to amend or further explain its rationale for
declining to require additional controls as part of the FIP's long-term strategy, to more fully
respond to comments submitted by the public, and to take further action if necessary." Therefore,
the statements relied upon by the commenter are no longer relevant. We are currently
reconsidering the Nebraska long-term strategy as it relates to the South Dakota reasonable
progress goals and will take appropriate action in the future.
Comment: Luminant stated that, given EPA's regional haze guidance, it is not surprising that
EPA has—prior to this proposal—repeatedly approved reasonable progress SIPs from other
375 77 FR. 12,770, 12776 (Mar. 2,2012).
376 See Respondent's Motion for Partial Voluntary Remand, Nebraska, v. US EPA (No. 12-3084) and Petition for
Review of an Order of the EPA, Nebraska v. US EPA (No. 12-3084).
482
-------
states that apply the four statutory factors on a source category basis as Texas did. For example,
EPA recently approved of Idaho's reasonable progress goals for the Class I areas based on the
state's "general level of review for the major source categories." Based on that review, EPA
agreed with Idaho that "it is not reasonable to achieve the UPR [sic] in 2018" and approved the
state's RPGs.478 There are many other examples as well where EPA approved reasonable
progress goals on a source-category basis. Some of these include:
• Alaska: "For this first Regional Haze Plan, ADEC believes that given the level of
improvement needed to reach natural conditions and the level of technical tools available
to demonstrate source specific impacts, it is reasonable to conduct the four-factor analysis
on the general source categories rather than on individual sources."479
• Oregon: "The Department looked at key pollutants and certain source categories and the
magnitude of their emissions in applying the four factors."480
• Washington: "[The state] decided to focus its four-factor analyses on the resulting set of
10 specific industries and emission source categories."481
• Alabama: "The preceding section [of the SIP narrative] (Section 7.6) discusses a general
four factor analysis for coal fired units."482
According to Luminant, indeed, of particular relevance here, EPA has previously approved of
Nebraska's reliance on CENRAP's source category analysis—the same source category analysis
that Texas relies on—in partially approving that state's long-term strategy and rejecting calls to
require further reductions for the benefit of out-of-state Class I areas.483
Footnotes:
478 77 Fed. Reg. at 30,256; 77 Fed. Reg. at 66,929.
479 Alaska 2011 SIP Narrative at 9-9 (emphasis added); 77 Fed. Reg. 11,022, 11035 (Feb. 24, 2012) (proposed
approval); 78 Fed. Reg. 10,546, 10,553 (Feb. 14, 2013) (final approval).
480 Oregon 2010 SIP Narrative at 163 (emphasis added); 77 Fed. Reg. 30,454, 30,461 (May 23, 2012) (proposed
approval); 77 Fed. Reg. 50,611, 50,612 (Aug. 22, 2012) (final approval).
481 Washington 2010 SIP Narrative at 9-5 to 9-7 (emphasis added); 77 Fed. Reg. 76,174, 76,203 (Dec. 26,2012)
(proposed approval); 79 Fed. Reg. 33,438, 33,438 (June 11, 2014) (final approval).
482 Alabama 2008 SIP Narrative at 79 (emphasis added); 77 Fed. Reg. 11,937, 11,949 (Feb. 28, 2012) (proposed
approval); 77 Fed. Reg. 38, 515, 38,519 (June 28, 2012) (final approval).
483 77 Fed. Reg. 40,150, 40,156 (July 6, 2012) ("The CENRAP modeling demonstration provided by the State [of
Nebraska] considered emissions of all anthropogenic source categories including major and minor stationary
sources, mobile sources, and area sources in developing its strategy. With the exception of the SO2 component of
the BART requirements as described elsewhere in this notice, the State has successfully demonstrated compliance
with all other remaining elements of the long-term strategy requirements.").
The Associations state that EPA fails to provide a reasoned explanation for disapproving Texas'
reasonable progress goals based on Texas' failure to conduct a source-by-source analysis of
emission controls when EPA has never required an individual source-based approach in the past.
In its prior reviews of State reasonable progress goals EPA has uniformly approved States'
reliance on source category-based analyses, even in the face of public comments urging a source-
based approach. For example, EPA approved Alaska's regional haze SIP in which the State
asserted that "it is reasonable to conduct the four-factor analysis on the general source categories
483
-------
rather than on individual sources." Alaska, SIP Narrative 9-9 (2011); 78 Fed. Reg. 10,546 (Feb.
14, 2013) (approving Alaska SIP). Likewise, EPA approved Oregon's reasonable progress goals
after the State explained that it "looked at key pollutants and certain source categories and the
magnitude of their emission in applying the four factors." Oregon, SIP Narrative 163 (2011); 77
Fed. Reg. 50,611 (Aug. 22, 2012) (approving Oregon SIP). EPA also approved Washington's
reasonable progress goals after the "state decided to focus its four-factor analyses on ... 10
specific industries and emission source categories." Washington, SIP Narrative at 9-5 to 9-7
(2010); 79 Fed. Reg. 33.439 (June 11, 2014) (approving Washington SIP).
According to the Associations, it is a well-established tenet of administrative law that
"[rjeasoned decision making ... necessarily requires the agency to acknowledge and provide an
adequate explanation for its departure from established precedent." Dillmon v. NTSB, 588 F.3d
1085, 1089-90 (D.C. Cir. 2009) (citingFCC vFox Television Stations, Inc., 129 S. Ct. 1800,
1811 (2009)). Indeed, given that the submitting States relied upon EPA's established guidance
and precedents in crafting their SIPs, EPA is required to provide a "more substantial
justification" for its disapprovals. Perez v. Mortgage Bankers Asssupra, slip op. at 13. In its
proposal to disapprove Texas' reasonable progress goals, EPA fails to even acknowledge, let
alone provide a reasoned explanation for, its departure from past precedent where it has
approved multiple SIPs based on analyses that were, in all relevant respects, identical to that
conducted by Texas.
EPA's adoption of an individual source-based approach is inconsistent with EPA's past practice
and with EPA's strict uniformity rule for regional offices. EPA fails to provide a reasoned
explanation for disapproving Texas' reasonable progress goals based on Texas' failure to
conduct a source-by-source analysis of emission controls when EPA has never required an
individual source-based approach in the past. In its prior reviews of State reasonable progress
goals EPA has uniformly approved States' reliance on source category-based analyses, even in
the face of public comments urging a source-based approach. For example, EPA approved
Alaska's regional haze SIP in which the State asserted that "it is reasonable to conduct the four-
factor analysis on the general source categories rather than on individual sources." Alaska, SIP
Narrative 9-9 (2011); 78 Fed. Reg. 10,546 (Feb. 14, 2013) (approving Alaska SIP). Likewise,
EPA approved Oregon's reasonable progress goals after the State explained that it "looked at key
pollutants and certain source categories and the magnitude of their emission in applying the four
factors." Oregon, SIP Narrative 163 (2011); 77 Fed. Reg. 50,611 (Aug. 22, 2012) (approving
Oregon SIP). EPA also approved Washington's reasonable progress goals after the "state
decided to focus its four-factor analyses on ... 10 specific industries and emission source
categories." Washington, SIP Narrative at 9-5 to 9-7 (2010); 79 Fed. Reg. 33.439 (June 11,
2014) (approving Washington SIP).
GCLC stated that EPA has also, on numerous recent occasions, approved of SIPs that did not
require source-specific requirements, but rather, looked more broadly at source categories. For
example, EPA recently approved the reasonable progress goals submitted by Idaho based on the
state's "general level of review for the major source categories."34
According to the TCEQ, neither CAA §169A, the RHR, nor the guidance available in 2009
required a unit-by-unit four factor analysis even where the state's RPGs would improve visibility
484
-------
less than the URP. The EPA itself supported the non-source specific four factor analysis
approach in reviewing New Mexico's regional haze plan. In a challenge to New Mexico's plan,
the EPA "points out that [§51.308(d)(l)(i)(A)] does not require a source-specific analysis."11
NRG stated that source-specific analysis also has not been applied by EPA in practice in the
reasonable progress context:
• EPA recently defended a state's reasonable progress goals that did not rely on a source-
specific analysis, and prevailed before the court: "[W]e reject the environmental groups'
argument that the EPA had to engage in a source-specific analysis for a reasonable-
progress determination. Nothing in the Regional Haze Rule or the Clean Air CAA
required New Mexico to conduct a four-factor analysis of the Escalante plant." WildEarth
Guardians v. EPA, 770 F.3d 919, 944.
Footnotes:
34 See Idaho SIP Approval Proposal, 77 Fed. Reg. at 30256; see also Approval and Promulgation of Implementation
Plans; State of Idaho; Regional Haze State Implementation Plan, 77 Fed. Reg. at 66929, 66929 (Nov. 8, 2012).
Footnotes:
tceq: n gee wildcarth Guardians v. EPA, 770 F3d 919, 944
mG: 11 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77 Fed.
Reg. 19,098 (March 30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,185
(June 27, 2012) (North Carolina); 77 Fed. Reg. 38,509 (June 28, 2012) (South Carolina); 77 Fed. Reg. 24,392
(April 24, 2012) (Tennessee); 77 Fed. Reg. 16,937 (March 23, 2012) West Virginia).
12 79 Fed. Reg. at 74,844.
Response: Please see our responses to other comments regarding our obligations under our
regional consistency rules.
Developing solutions to the complex problem of regional haze requires effective consultation
among states. During the first planning period, the States worked together through regional
planning organizations (RPOs) to help develop their regional haze SIPs. To assist in this effort,
we provided tens of millions of dollars to the RPOs following the issuance of the 1999 regional
haze rule to fund the development of the technical tools and analyses necessary to address
regional haze and to facilitate consultation among the States.
The States set up 5 RPOs to address visibility impairment from a regional perspective. The
technical analyses done by the RPOs for the first round of regional haze SIPs greatly increased
the understanding of the problem of visibility impairment at the federal Class I areas, including
that of the specific contribution of different species of pollutants.
We disagree with the fundamental premise of these commenters that our proposal to disapprove
portions of the Texas Regional Haze SIP was based on Texas' use of a source category-based
analysis in lieu of a source-by-source approach. First, Texas' approach to the reasonable
progress and long-term strategy provisions of the Regional Haze Rule cannot be described as a
being based on a "source-category" strategy. For instance, Texas did not examine all of the coal-
fired power plants (or refineries, cement plants, etc.) in the state and devise an approach that
485
-------
treated them all similarly. Instead, as we describe in our proposal,377 Texas did, in fact, partially
evaluate controls for certain individual sources. In evaluating these controls, Texas employed a
large, superficially refined control set consisting of a mix of large and small sources from a
number of different source categories located within varying distances of Class I areas.378 Texas
estimated the cost of controls on a variety of source types (albeit in a flawed manner as we
describe elsewhere) but abandoned that approach in favor of simply comparing the combined
cost of controls on all sources (regardless of their respective categories) against Texas flawed
estimate of their combined visibility benefit. This strategy is neither based on a source category
approach, nor is it based on a source-by-source approach. Therefore, comparisons of Texas'
approach to reasonable progress and long-term strategy against those of states that may have
used a category approach are not valid. We understand many of these comments arose because
our proposal included a statement that "individual sources were not considered by the TCEQ."
This statement was not offered to propose a basis for disapproval, but we understand it is
susceptible to being taken out of context (particularly in consideration of the comments
received). It is perhaps more plain to state that individual sources were not effectively considered
by the TCEQ.
Second, whatever its label, we proposed to disapprove Texas' reasonable progress analysis
because it was flawed in several specific ways. A primary flaw was that Texas' potential control
set was over-inclusive. It included controls on sources that served to increase the total cost, but
which conferred little visibility benefit. As was noted in our proposal,379 Texas adopted this
approach despite evidence in the record of identified source-specific, cost-effective controls that
would have resulted in large emission reductions on certain EGUs, and despite source
apportionment modeling that identified large impacts from EGU sources in northeast Texas. Our
proposal explained that this approach obscured benefits that might be obtained from individual
sources and only considered aggregated costs. As we also explained, the submitted analysis
failed to study or consider scrubber upgrade candidates. It was accordingly under-inclusive of
large, highly cost-effective emissions reductions that would lead to significant improvements in
visibility. These points are validated by the technical record for this FIP.
Therefore, whether Texas' analysis is labelled a source category analysis, an analysis of multiple
individual sources, or some hybrid, we conclude that it contained serious deficiencies that would
materially affect the outcome of the state's SIP process. As a result, we conclude this component
of the SIP requires disapproval.
Given the regional differences in the degree of visibility impairment, the pollutants of concerns,
and the impacts of fire and international emissions, we did not prescribe a one size fits all
approach to reasonable progress. The RPOs accordingly adopted somewhat different approaches
to recommending potential measures to ensure reasonable progress. That said, the RPOs and the
377 See discussion beginning on 79 FR 74838. See also the Texas SIP, Appendix 10-1: Analysis of Control
Strategies and Determination of Reasonable Progress Goals: Texas identified individual sources as possible
candidates for controls under reasonable progress, and did not limit the consideration of the statutory factors to
source categories.
378 For example, the control set evaluated by Texas included specific EGUs, carbon black plants, inorganic chemical
plants, paper mills and other sources.
379 79 FR 74838 ("[W]e believe that individual benefits were masked by the inclusion of those controls with little
visibility benefit that only served to increase the total cost figures.")
486
-------
States all agreed that large stationary sources of SO2 are the typically the primary cause or one of
the primary causes of anthropogenic visibility impairment at this time. In addition, in some
regions of the country, the RPOs and the States also recognized NOx as a similarly important
cause of visibility impairment.
In our review of the regional haze SIPs, we have attempted to take into account the differences
between states in assessing the reasonableness of each state's SIP submittal. For example, in
reviewing Alaska's regional haze plan, we took into account the following facts. First, Alaska
has relatively pristine air quality, and unlike Texas and Oklahoma, its four Class I areas suffer
from only a few deciviews of impairment. In Alaska's Denali National Park, for example, the
20% worst days have only 2.6 dv of impairment and to achieve the uniform rate of progress for
the first planning period, Alaska needed only 0.4 dv improvement.380 In Oklahoma, in contrast,
Wichita Mountains has impairment of over 16 dv and requires 3.8 dv of improvement to meet
the uniform rate of progress for this first planning period.381 We also note that, unlike Texas,
Alaska was not identified as contributing to visibility impairment at Class I areas in "nearby"
states. Beyond the differences in the respective scales of the overall problem, Alaska also
demonstrated that according to its 2002 emission inventory,382 fire accounts for more than half of
the of NOx emissions, about 70% of the SO2 emissions, and over 90% of the PM 2.5 emissions
in the state—far greater percentages than in Texas. Thus, the basic facts preclude a direct
comparison between what constitutes an acceptable reasonable progress strategy for Alaska
versus Texas and/or Oklahoma.
Nevertheless, setting these basic differences aside, the commenters have quoted a sentence from
the Alaska SIP narrative that does not provide the complete context of the analysis. Alaska did,
in fact, include information from individual sources in its analysis. As discussed in the Alaska
SIP narrative at 9-5, for example, the emissions from individual sources were applied to
supplement the modeling analysis used, which was based on the weighted emissions potential
(WEP) of source categories with potentially significant visibility impacts. The SIP states: "To
provide this insight the percent distribution of emissions from individual sources was organized into
common categories within the point and stationary area source categories (the two anthropogenic
categories that may be significantly impacting the Class I sites). The percent distribution of their
emissions within each source category, borough and year was applied to the corresponding WEP
value for those boroughs shown as potentially having a significant impact at each site (emphasis
added)." In addition, throughout the SIP, Alaska points out the significant emission reductions
expected from individual sources and how those reductions will ensure reasonable progress.
Of the remaining anthropogenic emission sources in Alaska, a significant fraction comes from
marine and aviation emissions, two categories that Alaska concluded it lacked authority to
regulate under the CAA. As a result, anthropogenic point source emissions constitute a much
smaller fraction of the problem for Alaska's Class I areas than for other states. Of those
anthropogenic point sources, Alaska determined that one was subject to BART and undertook a
source specific analysis that required the installation of controls. Alaska also noted that another
380 77 FR 11022, 11027-28.
381 76 FR at 16,176.
382 77 FR 11030.
487
-------
large point source of emissions would be reducing its emissions pursuant to a consent order, and
that another had stopped operating.383 Given those circumstances, that fact that the Class I Areas
are predicted to attain their respective URPs in 2018, and the very low overall degree of visibility
impairment at Alaska's Class I areas, we conclude that it was reasonable for Alaska to assess the
remaining industrial sources by category. The Texas SIP, in comparison, did not adequately ensure
reasonable progress would be achieved at Wichita Mountains.
In considering comments on consistency, we also note that we disagree with assertions by
Luminant and others that we are being inconsistent with respect to source category analysis in
States in the Pacific Northwest. In Idaho, the State reached the conclusion that no additional
controls beyond BART were needed to ensure reasonable progress, based on an evaluation of
possible cost-effective controls for certain categories, but the commenters fails to note that we
disagreed that the information relied on by Idaho in reaching that conclusion was sufficient. We
explained that "the Idaho SIP submittal did not contain sufficient analysis to support [the
conclusion that no additional control measures were needed].384 We conducted our own
independent screening analysis of individual sources in Idaho. In the end, our analysis supported
Idaho's conclusions that additional measures were not necessary in this planning period, and
because our ultimate conclusion was the same as Idaho's, we did not disapprove the SIP. The
process we followed for Idaho was consistent with the process we followed for Texas, although
the resulting conclusions were different.
For Oregon, as in Idaho, EPA took the same approach of supplementing the State's analysis in
considering the reasonableness of the State's determination that no additional controls beyond
BART were needed to ensure reasonable progress. Specifically, we did our own independent
screening of individual sources, and in so doing, we reached the same conclusion as Oregon did
that no additional controls were necessary for this planning period.385 We did not endorse the
methodology used by Oregon and therefore, we disagree with the commenter's assertion that we
approved the reasonable progress elements of the SIP based solely on an analysis by the State on
a source category basis.
Luminant cites to our Washington decision stating that Washington's demonstration was, like
Texas' based on a source-category demonstration. As above, we disagree that Texas' approach
can be strictly characterized as a source-category approach, and regardless, the type of approach
Texas used was not fundamentally at issue. We further disagree that the situations are similar.
First, as we note in our proposal,386 in establishing its RPGs, Washington did not take credit for a
number of reductions including almost 10,000 tons of SO2 reductions that occurred in the 2003
to 2005 timeframe from implementation of various control technologies from the Tesoro,
ConocoPhillips, and Shell refineries. Tesoro installed wet FGD on the CO Boiler (Fluidized
Catalyst Cracker) in 2005 for a reduction of 4,740 tons/yr SO2 and is considered BART in
Washington's BART determination. Conoco-Phillips installed wet-FGD on its CO boiler for a
reduction of 2,041 tons/yr SO2 which was not included in the WRAP modeling for RPGs. Shell
Puget Sound Refining installed wet-FGD on their CO boiler for a reduction of 3,045 t/y SO2
383 77 FR at 11035.
384 77 FR at 66931.
385 77 FR at 30464
386 77 FR 76204.
488
-------
which was not included in the WRAP modeling. Thus, Washington's reductions in visibility
impairing pollution were substantially underestimated in the modeling used to establish the
RPGs. Second, we see no similarity between Washington's approach to satisfying the
reasonable progress requirements and Texas' approach, other than in both instances the states'
RPGs were above their respective URPs. Washington did not adopt Texas' approach to RP/LTS,
in which individual benefits were masked by the inclusion of those controls with little visibility
benefit that only served to increase the total cost figures.387 For Washington, we stated that
considering the additional anticipated visibility improvements due to BART and additional
reductions on refineries, additional controls on point sources or other source categories at this
time are not likely to result in substantial visibility improvement in the first planning period due
to the significant contribution from emissions from natural fire, the Pacific offshore, Canada, and
outside the modeling domain.388 As we demonstrated in our FIP, however, substantial visibility
benefits were available from Texas sources.
We also note the state of Washington took individual sources into account when evaluating the
four statutory factors. For example, the state considered emission controls under its Reasonably
Available Control Technology (RACT) permitting requirements in reaching its conclusions for
reasonable progress: "The process in state law called Reasonably Available Control Technology
(RACT) requires a detailed valuation of the characteristics of each existing source covered by
the rule process along with an evaluation of the efficacy of installation of various control
equipment. The result of the process is a rule requiring all units of the defined source category
to achieve a set of defined emission limitations (emphasis added)."389
In general, many western states, including Washington, Oregon, Idaho, and Alaska, use and
reference analyses provided by the Western Regional Air Partnership (WRAP), including the
WRAP'S consideration of significant source categories for emission inventory development,
source apportionment modeling, modeling to determine the categories most likely to impair
visibility in Class I areas, and costs of controls. EPA has not taken issue with those types of
analyses and such analyses can be useful in identifying appropriate controls to ensure reasonable
progress. However, in cases where those analyses may not satisfactorily provide sufficient
information to identify whether or not there are additional reasonable controls available for
individual sources under reasonable progress, EPA has acted consistently. Specifically, we have
either conducted our own independent analysis or confirmed that the state adequately considered
emissions and visibility impacts from individual sources in addition to using the information
from the WRAP (or other regional planning body) analyses, based on the underlying evaluations
of the sources of visibility impairment in that state. In those instances where we reached the
same conclusion as the state, we approved the SIP.
Luminant also points to our action on the Alabama regional haze SIP as another example of EPA
approving reasonable progress goals on a source category basis. We note that the language
quoted by Luminant in its comment regarding Alabama appears to be drawn from the Alabama
SIP narrative rather than from our EPA's assessment of the state's plan. With respect to
387 79 FR 74838.
388 77 FR 76204.
389 This "RACT" requirement applies under the State's approved SIP distinct from RACT as may be required CAA
Section 172(c)("Nonattainment plan provisions").
489
-------
Alabama's reasonable progress analysis for its coal fired EGUs in the state, EPA explained how
participation in CAIR was expected to result in a 70% reduction in SO2 emissions from
Alabama. 77 FR 38,515, 38519 (June 28, 2012). Given these substantial reductions in SO2
emissions, we agreed with Alabama that CAIR ensured sufficient progress for coal-fired EGUs
for the first planning period. If the commenter is referring to non-EGU coal fired units, there is
extensive discussion in the notice regarding Alabama's approach to identifying sources for
potential reasonable progress controls through an "area of influence" methodology and the
state's approach to consideration of controls for specific sources.390
Luminant also points to Nebraska's reliance on CENRAP's source category analysis and our
partial approval of Nebraska's long-term strategy. We note that in Nebraska's regional haze SIP,
the State determined that EGUs account for almost all potentially controllable emissions of SO2
and NOx in the State.391 Three EGUs, in turn, account for over 80% of these emissions. [Id. at
27, Table 8.4] As each of these sources had units constructed between 1962 and 1977, Nebraska
considered each in turn to determine whether the facility was subject to BART. It determined
that the two largest had impacts that exceeded 0.5 dv and then undertook source specific BART
analyses to determine whether controls should be required. Although we disagreed with
Nebraska's analyses of these two sources, we do note that the State did not rely on CENRAP's
source category analysis in considering controls on these two EGUs that together comprise a
significant proportion of the State's controllable SO2 and NOx emissions, but rather relied on
their BART analysis to inform their decision on appropriate controls for these units.392 To the
extent that Nebraska relied on the CENRAP modeling to assess the remaining stationary sources
of emissions that are not-BART eligible in the state for reasonable progress, we do not consider
that analogous to the situation at issue in Texas, given the differences between the two states in
the number and size of sources of visibility impairing pollutants.393
TCEQ's and NRG's comments regarding EPA's approval of New Mexico's reasonable progress
assessment also do not undermine our action in Texas. As in other states, the question of an
appropriate reasonable progress analysis depends on the specific facts. In a petition for review of
EPA's approval of New Mexico's regional haze plan, environmental litigants argued that New
Mexico had erred by failing to consider the need for reasonable progress controls at Escalante
Generating Station. As EPA had noted in its review of New Mexico's SIP, "... NOx emissions
from the only subject-to-BART source in New Mexico (evaluated for controls under the BART
requirements) are greater than the next 20 largest NOx sources in the State combined... ,"394 And
as explained in its brief, EPA reasonably approved New Mexico's SO2 program which covered
Escalante. WildEarth Guardians v. EPA, No. 12-9596 (10th Cir.), EPA Respondent's Brief at
59-60. In other words, it is not EPA's position that a State must undertake a source specific
390 See e.g., 77 FR 35820.
391 Final Nebraska Regional Haze SIP at 20, Table 8.1.
392 In addition to requiring BART controls at subject facilities, the state also evaluated whether other sources need to
be addressed at this time to reach reasonable progress goals. Nebraska regional haze SIP at p5
393 In addition, the Eighth Circuit granted our request for a voluntary remand of our action on the Nebraska long-
term strategy on March 19, 2015. See Respondent's Motion for Partial Voluntary Remand, Nebraska, v. US EPA
(No. 12-3084) and Petition for Review of an Order of the EPA, Nebraska v. US EPA (No. 12-3084). We are
currently reconsidering the Nebraska long-term strategy as it relates to the South Dakota reasonable progress goals
and will take appropriate action in the future.
394 77 FR 70702, footnote 7.
490
-------
analysis of each and every coal-fired power plant within its borders to meet the reasonable
progress requirements. What it must do, however, is to reasonably assess the sources of
visibility impairing pollutants within the state and potential control strategies.
Most importantly, we note that the commenters have essentially set up a straw man to attack.
We did not propose to disapprove the Texas regional haze SIP because the state failed to
evaluate the four statutory factors on a source-by-source basis. In fact, as we pointed out in the
proposal, Texas did partially evaluate the potential for additional controls for individual sources.
However, the analysis was flawed because it did not go far enough to support the state's
conclusion that there were no reasonable controls available for reasonable progress. This
conclusion contrasted with the fact that Texas had identified very large individual emission
sources with cost effective options available to control those emissions. Our decision to require
additional controls was supported in part by analyzing the improvement in visibility that was
expected to result from controlling specific emission units, which is an element that was lacking
in the Texas submittal. As discussed above, there were other states that also reached a
conclusion that there were no reasonable additional controls, and similarly to Texas, EPA could
not confirm that decision with the available information. In those cases, we conducted our own
independent analyses and because our conclusions matched those of the affected states, we
approved their SIPs. In reviewing all of the regional haze SIPs, we consistently examined
whether or not states rejected potential reasonable progress measures that would have had a
meaningful impact on visibility in their Class I areas in ensuring that reasonable progress was
achieved during the first planning period. Texas did reject such controls which led to our
disapproval of the SIP. Our decisions were not contingent upon whether or not a state evaluated
the statutory factors on a source by source basis, but rather, whether or not the conclusion
reached by the state was appropriate and that the supporting information was adequate.
Comment: AECT stated that EPA's Proposal evaluates Texas EGUs and other emissions
sources using different and more stringent standards than it has used for emissions sources in
other states with respect to their Regional Haze SIPs. By way of example, contrary to how
EPA's Proposal evaluates Texas EGUs and other emissions sources, for the Regional Haze SIPs
for other states, EPA: (i) did not require that determinations of reasonable progress be made on a
source-by-source basis; 10 (ii) used a different visibility model;11 (iii) determined that CSAPR is
sufficient to meet reasonable progress requirements, and, thus, did not require any source that
complies with CSAPR to install additional emissions control to meet reasonable progress;12 and
(iv) determined that a predicted impact on a Class I area of 0.5 deciview ("dv") was "relatively
small" and that a source with a predicted impact on a Class I area at or below 0.5 dv should not
be required to install additional emission controls to meet reasonable progress13.
AECT asserted that EPA's evaluation of Texas EGUs and other emissions sources using different
and more stringent standards than it has used for emissions sources in other states with respect to
their Regional Haze SIPs is contrary to EPA's regional consistency rules in 40 CFR 56.3(a) and
(b). Those rules require EPA to uniformly apply the criteria, procedures, policies, and rules that
it employs in implementing the CAA, and to identify and correct regional inconsistencies by
standardizing such criteria, procedures, policies, and rules. In a recent decision, the D.C. Circuit
491
-------
Court of Appeals based its ruling that EPA had acted contrary to law on the requirements of 40
CFR 56.3(a) and (b).14
Based on the foregoing, AECT requested that EPA re-evaluate Texas' Regional Haze SIP using
the same standards than it has used for other states' Regional Haze SIPs.
GCLC states that EPA has acted inconsistently with prior actions have already been referenced
in our comments above (e.g., EPA's actions on the Idaho and Nebraska SIP submissions) but
extend far beyond those. This includes EPA's recent approvals of reasonable progress
evaluations, where states have taken the same approach and reached the same results as Texas,
and EPA has approved them. 70
Footnotes:
10 See, e.g., Alaska 2011SIP Narrative at 9-9 and 78 Fed. Reg. 10546, 10553 (Feb. 14, 2013) (final approval);
Oregon 2011 SIP Narrative at 163 and 77 Fed. Reg. 50611, 50612 (Aug. 22, 2012) (final approval); Washington
2010 SIP Narrative at 9-5 to 9-7 and 79 Fed. Reg. 33438 (June 11, 2014) (final approval); Alabama 2008 SIP
Narrative at 79 and 77 Fed. Reg. 38515, 38519 (June 28, 2012) (final approval)
11 Declaration of Sam Coleman, Nat'/Parks Conservation Ass'n v. McCarthy, No. 11-01548, at 5 (D.D.C. 2014).
12 See 77 Fed. Reg. 38515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38501(June 28, 2012) (Georgia); 77 Fed.
Reg. 34218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19098 (March
30, 2012} (Kentucky); 77 Fed. Reg. 71533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38007 (June 26, 2012)
(Missouri); 77 Fed. Reg. 38185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39177 (July 2, 2012) (Ohio); 79
Fed. Reg. 24340 (April 30, 2014) (Pennsylvania); 77 Fed. Reg. 38509 (June 28, 2012) (South Carolina); 77 Fed.
Reg. 24392 (April 24, 2012) (Tennessee); 77 Fed. Reg. 35287 (June 13, 2012) (Virginia); 77 Fed. Reg. 16937
(March 23, 2012) (West Virginia)
13 77 Fed. Reg. 30248, 30256 (May 22, 2012) (Idaho); 77 Fed. Reg. 30454, 30461, 30464 (May 23, 2012)
(Oregon)
14 National Environmental Development Ass'n's Clean Air Project v. EPA, 752 F.3d 999 (D.C. Cir. 2013)
70 See Approval and Promulgation of Implementation Plans; Region 4 States; Visibility Protection Infrastructure
Requirements for the 1997 and 2006 Fine Particulate Matter National Ambient Air Quality Standards, Final Rule, 79
Fed. Reg. 26,143,26,145-46 (May 7, 2014), which included: "Regarding the reasonable progress evaluations, each
state at issue focused its reasonable progress analysis on SO2 emissions based on the conclusion that sulfate
particles account for the greatest portion of the regional haze affecting Class I areas in these states. Each state then
established areas of influence and contribution thresholds to determine which of its sources should be evaluated for
reasonable progress control. EPA approved each state's methodology for identifying units for reasonable progress
evaluation and each state's reasonable progress determinations in the respective regional haze SIP actions and
provided a detailed discussion of the methodology and the rationale for approval in the Federal Register notices
associated with those actions. Contrary to the Commenter's assertions, Alabama, Georgia, Kentucky, North
Carolina, and South Carolina did not "exempt [CAIR] sources ... that would otherwise be subject to reasonable
progress review." Each of these states considered the four statutory reasonable progress factors in evaluating
whether CAIR would satisfy reasonable progress requirements for the state's EGU sector and determined that no
additional controls beyond CAIR were reasonable for SO2 during the first planning period. As discussed in EPA's
Reasonable Progress Guidance, states may evaluate the need for reasonable progress controls on a source category
basis, rather than through a unit-specific analysis, and have wide latitude to determine additional control
requirements for ensuring reasonable progress. The guidance also notes that states may consider emissions
reductions from cap-and-trade programs such as CAIR in addition to source-specific controls."
71 Declaration of Sam Coleman, Nat 'I Parks Conservation Ass 'nv. McCarthy, No. 11-01548, at 5 (D.D.C. 2014).
(emphasis added).
Response: We disagree with AECT that we evaluated Texas EGUs and other emissions sources
using different and more stringent standards than we used for emissions sources in other states.
See our responses to other comments in this section regarding allegations that we have been
492
-------
inconsistent with regard to Texas' supposed source-category approach to reasonable progress
and long-term strategy.
We address comments regarding our selection of the photochemical grid model CAMx over
CALPUFF to assess visibility impacts and benefits for Texas sources in a separate response to
comment. We note that Texas utilized CAMx modeling for BART screening purposes and
extrapolated CAMx model results to estimate visibility benefits of additional controls in the
Texas regional haze SIP.
Also, as we discuss elsewhere, our Regional Haze Rule requires a state to consider the four
statutory factors for reasonable progress even if the state has participated in CAIR or CSAPR. In
some cases, a state may have demonstrated, through the appropriate four-factor analysis, that
emission reductions and visibility improvement expected from CAIR or CSAPR and other
required measures, such as reductions required under BART for non-EGU sources, were
sufficient to achieve reasonable progress for the first planning period without imposing
additional controls. For those states, EPA approved the SIPs. The four-factor analysis for Texas
did not support the state's conclusion that it was achieving reasonable progress without requiring
additional controls. That flaw along with others as detailed in our final action was the basis of
our FIP, which was not related to a more stringent evaluation for Texas compared to other states.
The commenter also implies that our evaluation of the Texas SIP was different and more
stringent compared to other states because, for other states, EPA determined that "a predicted
impact on a Class I area of 0.5 deciview ("dv") was "relatively small" and that a source with a
predicted impact on a Class I area at or below 0.5 dv should not be required to install additional
emission controls to meet reasonable progress." The inference from the commenter that a
modeled impact of 0.5 dv is a "bright line" for reasonable progress decision has been improperly
taken out of context. The actual passage that the commenter cites is:
Since all of these sources have Q/d values below 20, EPA believes that their
impacts on nearby Class I areas are expected to be less than 0.5 dv. Thus, EPA
agrees with Oregon's conclusion that additional controls of non-BART point
sources for reasonable progress purposes are not reasonable in the first planning
period, because even though there are cost effective controls identified, visibility
improvement is anticipated to be relatively small.
The analysis that EPA used for Oregon and in other states cited by the commenter was
conceptually different, including the use of a different visibility model and metrics, than that
used for evaluating TX's sources. The visibility analysis approach that we used for Oregon is
different than the approach in Texas, and thus, drawing conclusions for the Texas sources based
on the metrics and technical analysis we used for Oregon is also not appropriate. We discuss the
differences in visibility models, model inputs, and metrics used elsewhere in a separate response
to comment. We further discuss in depth why Texas' choice of a 0.5 dv visibility threshold,
including the manner in which it was applied, was not appropriate in our response to comments
on consideration of cost versus visibility.
493
-------
For all regional haze SIPs, EPA assessed 1) whether or not the state appropriately considered the
four statutory factors and adequately demonstrated that it was achieving reasonable progress at
its Class I areas, and 2) whether or not the state, after appropriately consulting with other
downwind states for which it was contributing to visibility impairment, obtained its share of
emission reductions to ensure that the Class I areas of the downwind states were also achieving
reasonable progress. For some SIPs, the appropriate supporting technical tools, including the
models used, could differ in scope and complexity for a variety of reasons. However, criteria #1
and #2 above were uniformly applied in all cases. Please see our more detailed responses to
comments regarding uniformity, but in summary, we disagree with the commenter that we
contradicted the requirements of 40 CFR 56.3(a) and (b).
We disagree with AECT that we have violated our regional consistency rules and discuss that
allegation in detail in the beginning of this section of our response to comments.
We are unsure of GLCC's point regarding its general comment on Idaho and Nebraska. Specific
comments regarding previous actions in Nebraska and Idaho are addressed elsewhere. However
as discussed in our responses above, we believe comments regarding Nebraska are likely moot,
as we have taken a voluntary remand of our action on the Nebraska long-term strategy.395 To the
extent GLCC is referring to the issues of controls beyond CAIR/CSAPR or source-by-source
versus category approaches to reasonable progress, we have addressed those issues in our
response to other comments.
Comment: Luminant states as long as the state performs the required analysis (as Texas did
here), EPA must respect the state's choices and approve its reasonable progress goals—even
where EPA would have come to a different conclusion than the state in considering the
factors.432 And even though EPA believes that more reductions from Texas sources are
"feasible,"433 such a standard was specifically rejected by Congress in enacting the regional haze
program and is not a lawful basis for EPA's disapproval.434 EPA's proposal thus applied the
wrong legal standard by which to judge Texas's SIP revision.
Further, Luminant stated that there was no error in Texas's analysis. Texas's analysis followed
the regulations and EPA's own guidance in considering potential additional controls and
assessing the four statutory factors. And, as discussed elsewhere in our comments, Texas
followed the same approach as other states, whose SIPs EPA approved without reservation 435
Footnotes:
432 North Dakota, 730 F.3d at 768 ("[T]he CAA requires only that a state establish reasonable progress, not the most
reasonable progress."); 79 Fed. Reg. at 5205 (approving Wyoming's decision not to impose additional controls on
oil and gas source category because EPA concluded that the costs "were not so low that EPA could find it
necessarily unreasonable for the State to not have adopted them").
433 Technical Support Document for the Oklahoma Regional Haze State Implementation Plan and Federal
Implementation Plan (OK TSD) 29 ("Oklahoma TSD") (Nov. 2014).
434 In the conference committee that reconciled the House and Senate versions of the 1977 Clean Air Act
amendments, the term "maximum feasible progress" was specifically changed to "reasonable progress" in the final
395 See Respondent's Motion for Partial Voluntary Remand, Nebraska, v. US EPA (No. 12-3084) and Petition for
Review of an Order of the EPA, Nebraska v. US EPA (No. 12-3084).
494
-------
legislation. See 1 Legislative History of the Clean Air Act Amendments 1977 Pub. L. No. 95-95 155 (1977) ("The
term 'maximum feasible progress' is changed to read 'reasonable progress' whenever it appears in the section.").
435 For example, EPA approved, without question, Washington's decision that the URP was not reasonably
achievable and that additional controls were not necessary during the first planning period, based on the same
rationale Texas used in its SIP. 77 Fed. 76,174 (Dec. 26,2012). As EPA explained: "Additional controls on point
sources or other source categories at this time is not likely to result in substantial visibility improvement in the first
planning period due to the significant contribution from emissions from natural fire, the Pacific offshore, Canada,
and outside the modeling domain." Id. at 76,204.
Response: We disagree with the commenters. Although we agree that Texas conducted an
evaluation of the four reasonable progress factors, we determined that the state's evaluation was
flawed. As explained in our final action and elsewhere in our responses to comments, this
compelled us to conduct our own analysis and ultimately promulgate a FIP as required under the
CAA. In conducting our analysis, had we reached the same conclusion as Texas with respect to
additional controls necessary to achieve reasonable progress (as we did when we evaluated the
Idaho SIP), we would have approved the reasonable progress elements of the SIP.
As for the cited North Dakota court decision, we first note that the Court here reaffirmed the
determination in Oklahoma v. EPA,396 that we have more than a ministerial duty in reviewing
regional haze SIPs:
Although the CAA grants states the primary role of determining the appropriate
pollution controls within their borders, EPA is left with more than the ministerial
task of routinely approving SIP submissions. The Tenth Circuit recently
concluded that EPA acted within its power under § 169A in rejecting a BART
determination on the basis that the state "did not properly take into consideration
the costs of compliance when it relied on cost estimates that greatly overestimated
the costs of dry and wet scrubbing to conclude these controls were not cost
effective." Oklahoma v. EPA, 723 F.3d at 1206, 1208-10 (internal quotation
marks omitted). The court held that because the state's cost of compliance
estimate was based upon fundamental methodological flaws, EPA had a
reasonable basis for rejecting the state's BART determination for failure to
comply with the requisite BART guidelines at 1211-12. Moreover, in Alaska
Department of Environmental Conservation v. EPA, 540 U.S. 461, 124 S.Ct. 983,
157 L.Ed.2d 967 (2004), the Supreme Court rejected an argument similar to that
raised here regarding EPA's oversight role in the BACT determination process
under § 167 of the CAA. The Court held that EPA was not limited simply to
verifying that a BACT determination was actually made, concluding instead that
EPA could examine the substance of the BACT determination to ensure that it
was one that was "reasonably moored to the Act's provisions" and was based on
"reasoned analysis." See id. at 485, 490, 124 S.Ct. 983. Although the Court's
analysis was one under § 167, we nonetheless find it persuasive in the context of §
169 A.
Thus, Luminant's contention that "as long as the state performs the required analysis ... EPA
must respect the state's choices and approve its reasonable progress goals" misinterprets our role.
396 Oklahoma v. EPA, 723 F.3dat 1206, 1208-10.
495
-------
Although the analysis ailed on by the Court concerned BART, we believe it concerns the same
basic point regarding our review of the Texas Regional Haze SIP: As we discuss in our
proposal, 397 we disagree with Luminant's view that Texas "performed the required analysis."
As explained more fully in the section of this document that responds to comments dealing with
state and federal roles in the regional haze program, we determined that Texas' reasonable
progress and long-term strategy analysis was flawed and we have authority to review it and if
necessary disapprove it and replace it with a FIP.
With regard to the quote Luminant reproduced from the Eighth Circuit Court's decision in North
Dakota v. EPA cited above, several environmental groups challenged a portion of our final
action on North Dakota's regional haze SIP that ultimately approved North Dakota's reasonable
progress determination for NOx controls for the Coyote Station.398 The environmental groups
objected to North Dakota's decision to reject a control it had evaluated, after having applied the
four reasonable progress factors, and subsequently approving another NOx control as reasonable
progress.
We interpret the Court's statement as meaning broadly that just because a more stringent level of
control could be technically feasible in a particular instance, it does not mean it necessarily must
be required under reasonable progress. We see no conflict with this determination and our
proposed Texas FIP. Extending the Court's reasoning, we too could have proposed additional
controls on other sources within Texas, or higher levels of control on the sources for which we
proposed controls. We proposed, however that neither additional controls nor a higher level of
control on the sources for which we proposed controls was appropriate, after applying the four
reasonable progress factors. In other words, we did not propose the highest level of control
(referred to by the Court as "most reasonable"). Further, in North Dakota's case, we noted
technical flaws in North Dakota's analysis, and we noted that we could have reached a different
conclusion had we conducted the analysis ourselves, but we ultimately determined these issues
did not prevent us from accepting North Dakota's reasonable progress determination. The Court
did not find that our conclusions on the issue were arbitrary, stating in part that, "[e]ven if [the
control in question] were perhaps the most reasonable technology available, the CAA requires
only that a state establish reasonable progress, not the most reasonable progress.399 In contrast,
and as explained in greater detail elsewhere, in reviewing Texas's reasonable progress analysis,
we determined that the technical analysis the state undertook was flawed and thus its reasonable
progress determination was not approvable. We determined that cost effective controls were in
fact available that would have very significant visibility benefits. However, Texas' flawed
analysis masked the benefits of these controls.
Luminant also cites to our Wyoming action in support of its contention that we must respect the
state's choices and approve its reasonable progress goals—even where we would have come to a
different conclusion than the state in considering the factors. Luminant points to the section of
our Wyoming action where we approved Wyoming's decision not to impose additional controls
on oil and gas sources, where we stated in response to a comment, "those costs were not so low
397 See discussion beginning on 79 FR 74838.
398 North Dakota v. EPA, 730 F.3d 750 (8th Cir. 2013); (See EPA's final rule at 77 Fed. Reg. 20.894-945 (April 6,
2012)).
399 Id. at 768.
496
-------
that EPA could find it necessarily unreasonable for the State to not have adopted them."
Luminant seems to be suggesting that because we agreed with the state's decision with regard to
Wyoming's oil and gas sources, we are bound to agree with all state decisions—a notion we
reject—again, our role is more than ministerial. Notwithstanding our comments above, as we
further explain in our response to that comment in our Wyoming final action, oil and gas sources
in Wyoming were already being controlled to a large extent, including BACT controls for new
sources compressor engines, and through Wyoming's oil and gas permitting guidance. For
instance, we discuss in our proposal for that action,400 that Wyoming conducted a thorough
evaluation of potential controls on oil and gas sources. For each source category it estimated a
range for the cost effectiveness, which in many cases varied greatly. We stated, "As shown by
the four-factor analyses, the most reasonable controls are for compressor engines, which the
State already controls through its minor source BACT requirements ...." Thus, although not
disclosed by Luminant in its comment, some of the most cost effective controls were for oil and
gas sources that were already being controlled in Wyoming. Also not disclosed by Luminant is
the fact that we disagreed with Wyoming's control decisions in other instances. For instance,
our Wyoming FIP required controls on the Dave Johnston Unit 3; the Laramie River Units 1, 2,
and 3; and the Wyodak Unit 1. Luminant's contention that we, "must respect the state's choices
and approve its reasonable progress goals—even where [we] would have come to a different
conclusion" is therefore inconsistent with the facts within the very action which it references.
We address Luminant's allegation that we are inconsistent in comparison to our action for
Washington in the response to another comment.
Comment: GCLC states that we have found no instance where EPA has imposed a FIP for the
first planning period requiring additional reasonable progress controls on EGUs that relied on
implementation of CAIR or CSAPR to satisfy BART. 15
Luminant states that in reviewing other states' reasonable progress goals for the first planning
period, EPA has repeatedly and systematically excluded sources that otherwise comply with the
BART requirements, including complying through a regional trading program that is better-than-
BART, from further emission controls in the first planning period without even conducting an
"additional analysis" like it performs here for certain Texas sources.759
Luminant states that EPA proposes, as it did for these other states, to issue a FIP that would
"replace Texas' reliance on CAIR to satisfy the BART requirement for EGUs with reliance on
CSAPR."377 But EPA's proposal otherwise disregards CSAPR's more stringent SO2 and NOx
emission budgets for Texas, as compared to CAIR, as well as the additional trading restrictions
imposed by CSAPR. For all other states that have relied on either CAIR or CSAPR, EPA has
repeatedly found such participation to also satisfy the states' reasonable progress obligation for
the first planning period for those sources.378 But not so for Texas. EPA chooses to completely
disregard CSAPR as an ongoing compliance obligation for Texas sources, even though more
stringent budgets are now in place and will be in place through the 2018 interim goal.379
400 See discussion beginning on 78 FR 34763.
497
-------
Luminant claims this is a critical error by EPA and an inexcusable inconsistency in the proposal.
In determining that CSAPR (like CAIR before it) "is better than BART,"380 EPA conducted
modeling that demonstrated: (1) "that the trading programs of the Transport Rule do not cause
degradation in any affected Class I area"; and (2) that "average visibility improvement of the
'Transport Rule + BART elsewhere' alternative was greater than 'Nationwide BART' on both
the 20 percent best and 20 percent worst days."381
As part of its analysis, EPA modeled the visibility in all Class I areas to determine the visibility
(in deciviews) at each area in 2014 with the implementation of CSAPR and BART in states not
subject to CSAPR.382 This included Big Bend, Guadalupe Mountains, and Wichita Mountains.
EPA used the same model—CAMx—that it now uses to judge the Texas SIP. Table 9 provided
by Luminant shows the visibility at the three Class I areas of interest here that EPA modeled for
2014, assuming CSAPR emission budgets in Texas and other CSAPR states and BART controls
in non-CSAPR states.
Table 9 Provided by Luminant (0061): EPA Modeled Visibility and three Class I Areas
with CSAPR in Place383
Class 1 A"ea
EPA Modeled V^b.My
Cond t'ons (2014) (5C%
Wo'st Days)
EPA Proposed RPG
(2018) (2CR west cays)
llg lend
t «A.
Guadalupe Mountains
14.7 cfv
-
Wichita Mountains
20,2 cfv
: ¦ < ¦
As Table 9 (provided by Luminant) shows, EPA's own CAMx modeling shows that—with
CSAPR in place as it is today—the visibility at these three Class I areas is already well below the
2018 goals that EPA is proposing for these areas—and without the additional SO2 controls that
EPA would impose by the proposed FIP. Yet, despite imposing CSAPR in its FIP for Texas,
EPA dismisses CSAPR's SO2 budgets as "not much different than the CAIR Cap for Texas" and
refuses to include CSAPR in its modeling for this proposal, as it did in its prior modeling. EPA
provides no explanation of its inconsistent modeling approaches. And, as discussed in the next
section, EPA's prior modeling that includes CSAPR more accurately predicts current visibility
conditions in these Class I areas than does the new Texas-only modeling EPA commissioned
from ENVIRON that disregarded CSAPR.
Luminant stated that Texas electric generating unit ("EGUs") are subject to stringent SO2 and
NOx emission limits under EPA's Cross-State Air Pollution Rule ("CSAPR"), and EPA has
determined that CSAPR's limits are better at improving visibility in Class I areas than Best
Available Retrofit Technology ("BART"). We have found no instance of EPA issuing a FIP for
a CSAPR state that would require emission limits beyond what will be required to comply with
CSAPR. EPA's proposed FIP thus treats Texas and Texas EGUs in an inconsistent manner than
EPA's prior actions in other states and arbitrarily ignores EPA's own CSAPR-better-than-BART
findings.
498
-------
Xcel states that EPA's concern about Texas importing unlimited allowances so as to avoid
reducing emissions is not valid here because CSAPR, which replaced CAIR, limits the use of
out-of-state allowances that can be used for compliance, and EPA has expressly allowed other
states to rely on CSAPR to meet RPGs. See Proposed Michigan SIP Approval, 77 Fed. Reg.
46,912, 46,919 (the regional planning organization's "analysis shows emission reductions
equivalent to the scale of CAIR are needed to meet reasonable progress goals.").
Xcel Energy stated that despite EPA's action in Michigan, in this Proposal, EPA arbitrarily
argues that CAIR could be used "in lieu of BART" but not used in the RPG context. While
BART and RPGs are distinctive components of a regional haze strategy, EPA provides no
reasoned basis for allowing consideration of CAIR in the BART context and rejecting it in the
RPG context. 3 That EPA would allow CAIR or CSAPR to substitute for BART, which is a
unit-specific standard with unit-specific performance criteria, but not for demonstrating
reasonable progress, which is a state-wide, multi-source program aimed at reducing the
pollutants of concern for regional haze, is illogical, as well as arbitrary and capricious. In fact,
EPA has done the exact opposite in other RPG determinations and re-affirmed States' reliance on
BART-equivalent analyses. For example, as stated in its proposed approval of the Georgia SIP,
"EPA believes it is reasonable to conclude that any control requirements imposed in the BART
determination also satisfy the RPG related requirements for source review in the first
implementation period." See Proposed Georgia SIP Approval, 77 Fed. Reg. 11,452, 11,469
(Feb. 27, 2012); see also Final Georgia SIP Approval, 77 Fed. Reg. 38,501 (Jun. 28,2012). In
North Dakota, EPA specifically rejected modeling for RPGs that was not conducted in the same
way as BART on the basis that the "ultimate goal is the same." Proposed North Dakota SIP
Approval/Disapproval, 76 Fed. Reg. at 58,629 n. 85; see also Final North Dakota SIP
Approval/Disapproval, 77 Fed. Reg. at 20,906-07.
[UARG (0065) p. 27] UARG stated that the EPA, without providing any explanation, proposes
to treat BART sources in Texas differently from the way all other sources that are subject to
BART in other states have been treated. EPA's Reasonable Progress Guidance states that BART
is likely to satisfy all reasonable progress requirements for the first planning period and leaves
states with wide latitude to make a determination that it does. See Reasonable Progress
Guidance at 4-2 to 4-3. Texas submitted a SIP that adopted the position EPA articulates in its
guidance by relying on compliance with CAIR to satisfy BART obligations for EGUs that are
subject to CAIR in the state. EPA proposes a FIP that would substitute reliance on CAIR with
reliance on CSAPR, CAIR's successor rule, to satisfy those BART obligations. In other states,
sources that have relied on CAIR or CSAPR to satisfy BART have not been subjected to
additional reasonable progress requirements during the first regional haze planning period.
EPA's proposed decision to require more of Texas is inconsistent with the Agency's guidance
and previous regional haze rulemakings.
Luminant stated that EPA steps well outside its authority in attempting to require the installation
of new scrubbers at seven Texas EGUs in 2020. The time period at issue in Texas's submittal,
and thus the scope of EPA's review and authority, is limited to the first regional haze planning
period—2008 to 20 1 8.879 As EPA has explained: "The RHR [regional haze rule] requires control
strategies to cover an initial implementation period extending to the year 2018, with a
499
-------
comprehensive reassessment and revision of those strategies, as appropriate, every 10 years
thereafter."880
The Associations stated that EPA acknowledges that it "cannot assume that the SO2 controls we
are proposing will be installed and operational within this planning period, which ends in 2018."
Id. In this respect EPA's proposal is inconsistent with past regional haze actions, where EPA has
consistently limited the scope of FIPs to control measures that can be implemented during the
interim planning period. See, e.g., 77 Fed. Reg. 20,894, 20,944 (Apr. 6, 2012) (applying a July
31, 2018, compliance deadline in North Dakota FIP); 77 Fed. Reg. 57,864, 57,916 (Sept. 18,
2012) (applying a July 31, 2018, compliance deadline in Montana FIP); 79 Fed. Reg. 52,420,
52,426 (Sept. 3, 2014) (applying a December 31, 2018, compliance deadline in Arizona FIP).
Luminant stated that EPA itself recently explained why it rejected comments that argued for
additional controls in its prior approvals of five other states' regional haze SIPs, all of which
took the same approach and reached the same results as Texas's SIP:
Contrary to the Commenter's assertions, Alabama, Georgia, Kentucky, North
Carolina, and South Carolina did not "exempt [CAIR] sources . . . that would
otherwise be subject to reasonable progress review." Each of these states
considered the four statutory reasonable progress factors in evaluating whether
CAIR would satisfy reasonable progress requirements for the state's EGU sector
and determined that no additional controls beyond CAIR were reasonable for SO2
during the first planning period. As discussed in EPA's Reasonable Progress
Guidance, states may evaluate the need for reasonable progress controls on a
source category basis, rather than through a unit-specific analysis, and have
wide latitude to determine additional control requirements for ensuring
reasonable progress. The guidance also notes that states may consider emissions
reductions from cap-and-trade programs such as CAIR in addition to source-
specific controls.795
Luminant asserted that EPA's different treatment of Texas is not at all justified, nor is it lawful
under EPA's regional consistency regulations. EPA "shall assure that actions taken under the
[CAA]: (1) Are carried out fairly and in a manner that is consistent with the CAA and Agency
policy as set forth in the Agency rules and program directives [and] (2) Are as consistent as
reasonably possible with the activities of other Regional Offices . . . ,"796 EPA has failed at this
obligation, and without any explanation or justification. EPA's proposal does not treat Texas or
Texas operators fairly, nor is it consistent with EPA's many prior actions on other states'
regional haze SIPs, as the example cited above and the many others in these comments illustrate.
Accordingly, EPA's proposal violates EPA's regional consistency regulations, fundamental
notions of fairness, and is contrary to law.797
NRG stated that EPA has repeatedly used interstate emission trading rules such as the Clean Air
Interstate Rule ("CAIR") and the Cross-State Air Pollution Rule ("CSAPR") to satisfy reasonable
progress goals.11 As EPA is proposing to use CSAPR to satisfy Texas' BART obligations,12 it
would only be logical to also use CSAPR to satisfy Texas' reasonable progress obligations as
well.
500
-------
Footnotes:
3 It does not make sense to disregard CAIR or CSAPR in the context of RPGs, because all electric generating units
in the state are subject to their emission limitations, while only some electric generating units are subject to BART
15 See 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77 Fed.
Reg. 34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098 (March
30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26, 2012)
(Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012) (Ohio); 79
Fed. Reg. 24,340 (April 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South Carolina); 77 Fed.
Reg. 24,392 (April 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77 Fed. Reg. 16,937
(March 23, 2012) (West Virginia).
759 Never, in its review of regional haze SIPs for the first planning period, has EPA imposed a FIP requiring
additional "reasonable progress" controls on EGUs that relied on the implementation of CAIR/CSAPR to satisfy
BART. See 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77
Fed. Reg. 34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098
(Mar. 30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26,
2012) (Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012)
(Ohio); 79 Fed. Reg. 24,340 (Apr. 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South
Carolina); 77 Fed. Reg. 24,392 (Apr. 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77
Fed. Reg. 16,937 (Mar. 23, 2012) (West Virginia).
795 79 Fed. Reg. 26,143, 26,145-46 (May 7, 2014) (emphasis added) (alteration in original) (internal citations
omitted).
796 40 C.F.R. § 56.5(a). These requirements also apply to EPA Headquarters officials "who are responsible for
developing the policies governing the implementation and enforcement of the CAA." Nat'l Envtl. Dev. Ass'n's
Clean Air Project v. EPA, 752 F.3d
999, 1009 (D.C. Cir. 2014).
797 Nat'l Envtl. Dev. Ass'n's Clean Air Project, 752 F.3d at 1009-10 (holding that EPA action contrary to its
regional consistency regulations was contrary to law); see also 40 C.F.R. § 56.5(a)(2) (officials in EPA regional
offices "shall assure that actions taken under the act. . . [a]re as consistent as reasonably possible with the activities
of other Regional Offices" (emphasis added)).
759 Never, in its review of regional haze SIPs for the first planning period, has EPA imposed a FIP requiring
additional "reasonable progress" controls on EGUs that relied on the implementation of CAIR/CSAPR to satisfy
BART. See 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77
Fed. Reg. 34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098
(Mar. 30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26,
2012) (Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012)
(Ohio); 79 Fed. Reg. 24,340 (Apr. 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South
Carolina); 77 Fed. Reg. 24,392 (Apr. 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77
Fed. Reg. 16,937 (Mar. 23, 2012) (West Virginia).
378 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77 Fed. Reg.
34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098 (Mar. 30,
2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26, 2012)
(Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012) (Ohio); 79
Fed. Reg. 24,340 (Apr. 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South Carolina); 77 Fed.
Reg. 24,392 (Apr. 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77 Fed. Reg. 16,937
(Mar. 23, 2012) (West Virginia).
15 See 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77 Fed.
Reg. 34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098 (March
30, 2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26, 2012)
(Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012) (Ohio); 79
Fed. Reg. 24,340 (April 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South Carolina); 77 Fed.
Reg. 24,392 (April 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77 Fed. Reg. 16,937
(March 23, 2012) (West Virginia).
501
-------
88° 77 Fecj Reg at 30,252 (EPA proposed approval of Idaho reasonable progress goals and long-
term strategy) 77 Fed. Reg. 30,454, 30,458 (May 23, 2012) (EPA proposed approval of Oregon
reasonable progress goals and long-term strategy).
Footnotes:
367 40 C.F.R. § 51.308(e)(2); see 70 Fed. Reg. 39,104, 39,138-43 (July 6, 2005); Util. Air Regulatory Grp. v.
EPA, 471 F.3d 1333, 1339-41 (D.C. Cir. 2006) (upholding "EPA's substitution of CAIR for BART").
368 See 70 Fed. Reg. 39,104, 39,138-13 (July 6, 2005); Util. Air Regulatory Grp., 471 F.3d at 1339-41 (upholding
"EPA's substitution of CAIR for BART").
369 2009 Texas SIP Narrative at 9-1.
370 North Carolina v. EPA, 550 F.3d 1176, 1178 (D.C. Cir. 2008).
371 76 Fed. Reg. 48,208 (Aug. 8,2011).
372 Order, EME Homer City Generation, L.P. v. EPA, No. 11-1302, at 2 (D.C. Cir. Dec. 30,2011).
373 7 9 Fed. Reg. 71,663 (Dec. 3,2014).
374 77 Fed. Reg. 33,642, 33,643, 33,643 (June 7, 2012). Luminant and some of its affiliated companies filed a
petition for review of EPA's disapproval of this portion of the Texas regional haze SIP in the U.S. Court of Appeals
for the Fifth Circuit. See Luminant Generation Co. LLC v. EPA, 12-60617 (5th Cir.). On motion of an intervenor
in that case, the Fifth Circuit transferred the petition to the U.S. Court of Appeals for the D.C. Circuit (Case No.
13-1178), and it was consolidated with other petitions under the lead case Util. Air Regulatory Grp. v. EPA, No.
12-1342. Luminant further moved to intervene in support of Respondent EPA in Case No. 12-1343, which is a
petition challenging EPA's determination that CSAPR is "better than BART." The D.C. Circuit granted that motion
on September 25, 2012. All of these consolidated petitions for review remain pending and are held in abeyance
pending the D.C. Circuit's disposition of EME Homer City Generation, L.P. v. EPA, No. 11-1302, et al. EPA's
proposal here states that EPA is "not taking comment" on its earlier limited disapproval of Texas's SIP on this basis.
79 Fed. Reg. at 74,821.
375 77 Fed. Reg. at 33,642, 33,643.
376 Id. at 33,654.
377 7 9 Fed. Reg. at 74,823.
378 77 Fed. Reg. 38,515 (June 28, 2012) (Alabama); 77 Fed. Reg. 38,501 (June 28, 2012) (Georgia); 77 Fed. Reg.
34,218 (June 11, 2012) (Indiana); 77 Fed. Reg. 38,006 (June 26, 2012) (Iowa); 77 Fed. Reg. 19,098 (Mar. 30,
2012) (Kentucky); 77 Fed. Reg. 71,533 (Dec. 3, 2012) (Michigan); 77 Fed. Reg. 38,007 (June 26, 2012)
(Missouri); 77 Fed. Reg. 38,185 (June 27, 2012) (North Carolina); 77 Fed. Reg. 39,177 (July 2, 2012) (Ohio); 79
Fed. Reg. 24,340 (Apr. 30, 2014) (Pennsylvania); 77 Fed. Reg. 38,509 (June 28, 2012) (South Carolina); 77 Fed.
Reg. 24,392 (Apr. 24, 2012) (Tennessee); 77 Fed. Reg. 35,287 (June 13, 2012) (Virginia); 77 Fed. Reg. 16,937
(Mar. 23, 2012) (West Virginia).
379 FIP TSD at A-45.
380 77 Fed. Reg. 33,642, 33,648 (June 7, 2012).
381 Id. at 33,652 (emphasis added).
382 EPA, Technical Support Document for Demonstration of the Transport Rule as a BART Alternative, Dock. ID
No. EPAHQ- OAR-2011-0729-0014 (Dec. 2011) ("BART Alternative TSD").
383 Id. at tbl.3-3; 79 Fed. Reg. at 74,887, tbl.43.
Response: Although we proposed to rely on CSAPR to address the BART requirements for
EGUs in Texas, we are not finalizing that proposed action. On July 28, 2015, the D.C. Circuit
Court's issued its decision in EME Homer City Generation v. EPA, 795 F.3d 118 (D.C. Cir
2015), upholding CSAPR but remanding without vacating a number of the Rule's state emissions
budgets. Specifically, the court invalidated a number of the Phase 2401 ozone season NOx
budgets and found that the SO2 budgets as to four states resulted in overcontrol for purposes of
401 CSAPR's effective date was stayed by the D.C. Circuit during a portion of the litigation over the rule. Thus, the
Rule's Phase 1 effective date of January 1, 2012 was delayed until January 1, 2015, and the Rule's Phase 2 effective
date of January 1, 2014 was delayed until January 1, 2017.
502
-------
section 110(a)(2)(D). Texas' ozone season NOx budget and its SO2 budget are both implicated
in this remand. We are in the process of acting on the Court's remand. As a result, at this time
we cannot ensure that CSAPR will continue to be an appropriate alternative to BART for Texas
EGUs. As a result, the comments that we are treating Texas differently than other states where
EPA relied on CSAPR to meet the BART requirements are no longer applicable. Similarly, to
the extent that CAIR or CSPAR could have been relied on to ensure reasonable progress in the
Texas regional haze SIP or FIP, at this point in time CAIR is no longer in place and the CSAPR
budgets for Texas are unclear.
Even assuming, however, that EME Homer City had not invalidated the CSAPR NOx and SO2
budgets for Texas and that we were taking final action to address the BART requirements
through reliance on CSAPR, we do not agree that EPA has been inconsistent in its treatment of
Texas as compared to other states. As explained in our proposed rulemaking, allowing Texas to
rely on CSAPR to meet its reasonable progress obligations is not appropriate, and the fact that
other states did not require additional reasonable progress controls beyond CAIR (or CSAPR)
does not automatically mean all states should not require any additional controls. Such a
simplistic comparison ignores the meaningful differences between Texas and the states cited.
These include the significant impacts that sources in Texas are projected to have on the visibility
at the Wichita Mountains in Oklahoma in 2018 (which includes projected reductions due to
CAIR compliance), the quality of the Texas technical evaluation, and the quality of the
consultations between Texas and Oklahoma.
As explained above, in our review of the regional haze SIPs, we have attempted to take into
account the differences between states in assessing the reasonableness of each state's SIP
submittal. For example, in many of the CSAPR states in the East, CAIR was anticipated to lead
to substantial reductions in SO2 emissions from EGUs.402 In contrast, in Texas, the State
estimated that SO2 emissions from EGUs would fall by only 36% to 350,000 tpy (2018).403 In
addition, for many of the Class I areas at issue, CAIR and other measures were anticipated to
bring about sufficient visibility improvement for the area to exceed the rate of visibility
improvement of the URP glidepath toward natural visibility conditions in 2064. Other factors,
such as the state's percentage impact on downwind Class I areas also differentiates Texas in
404
some cases
Luminant also notes Texas' more stringent NOx and SO2 budgets under CSAPR. To the extent
that these budgets might have changed the anticipated SO2 emissions from sources in Texas, this
is no longer relevant given the remand of these budgets to EPA. Moreover, we note that we did
not propose to require any additional NOx controls in Texas to meet the reasonable progress
requirements. For SO2, at the time of EPA's proposed FIP, it was unclear to what extent Texas
would reduce emissions and to what extent it would rely on credits from reductions in other
402 .402 See, e.g. 77 Fed. Reg. 38515, 38519 (In Alabama CAIR is expected to reduce EGU SO2 emissions by
70%); 77 Fed. Reg. 38501, 38505 (Under Georgia's CAIR rule, EGU SO2 emissions are capped at 149,140 tons, a
70% reduction from 2002 levels); 76 FR 78194, 78,207 (SO2 emission from EGUs in Kentucky estimated to decline
54% by 2018); 77 Fed. Reg. 38185 (North Carolina SIP requires 73% reduction in SO2 emissions); see also 77
Fed. Reg. 11,894, 11902 (South Carolina S02 point source emissions projected to decline 44% to 146,851 tons).
403 77 FR at 74,858.
404 See e.g. 77 FR 11974, 11979 (Feb. 28, 2012) (combined effect of Iowa's emissions on any Class I area less
than 5%).
503
-------
states. As explained in detail in the FIP TSD (see page A-45) and elsewhere in this document,
due to the cost of SO2 credits being lower than originally projected and comments from Texas on
a more recent IPM projection indicating that significant SO2 reductions were not anticipated at
these sources due to MATS and CAIR/CSAPR, we do not have reason to believe significant
additional reductions will result in the near future due to CSAPR requirements (even assuming
Texas CSAPR budgets remained unchanged). Furthermore, TCEQ noted that no large SO2
control projects were planned at most of the sources being evaluated. We also noted that TCEQ
has utilized recent emission data for EGUs when developing projected emissions for 2018 (and
other future years) when developing ozone attainment demonstrations. Overall this information
supports our position that recent actual emissions for specific sources evaluated and CENRAP's
2018 projections including anticipated CAIR reductions are a reasonable representation of future
emission levels in 2018.
Xcel raises similar concerns as Luminant that EPA has allowed other states to rely on CSAPR to
meet RPGs, citing to EPA's action on Michigan. As discussed elsewhere, Michigan had less
than a 5% impact on visibility in downwind states. With respect to its own Class I areas,
Michigan assessed the contribution of three Midwestern states together and concluded that the
47% reduction in SO2 emissions was sufficient for reasonable progress. Michigan also noted
that a number of subject-to-BART non-EGUs also have significant impacts on visibility in
Michigan's Class I areas and BART determinations (by the State and by EPA) require additional
reductions from these sources that will result in significant visibility benefits. In addition,
Michigan noted in its regional haze SIP that some of the largest EGUs, such as DTE's Monroe
power plant and Consumer Energy's Campbell plant, have installed or are in the process of
installing controls.405 More generally, with respect to Xcel's comment that EPA provides no
reasoned basis for rejecting consideration of CAIR in the reasonable progress context, we
disagree that this is what we proposed to do in Texas. In our proposal we took into account the
impact of CSPAR/CAIR on emissions in Texas, but determined that participation in this program
would not ensure reasonable progress at Wichita Mountains and the Texas Class I areas. It is
important to remember that the test for determining whether a BART alternative provides for
greater reasonable progress than BART is based on improvements in visibility on average across
all Class I areas. Therefore, a BART alternative may result in very little visibility improvement
at some Class I areas and large improvements at other Class I areas such that on average across
all Class I areas visibility conditions are better under the alternative than under BART. We
noted in 2005 that the determination that CAIR provided for greater reasonable progress than
BART did not answer the question of whether more than CAIR would be required in a regional
haze SIP:
Our determination that CAIR makes greater reasonable progress than BART for
EGUs is not a determination that CAIR satisfies all reasonable progress
requirements in CAIR affected States. Each State, whether in the CAIR region or
not, is required to set reasonable progress goals for each Class I area within the
State as required in regional haze rule section 308(d)(1), and to develop long term
strategies, considering all anthropogenic sources of visibility impairing pollutants,
as required by section 308(d)(3). In setting the reasonable progress goals, the
405 Michigan regional Haze SIP at 54
504
-------
State is to consider the amount of visibility improvement needed to achieve a
uniform rate of progress towards natural background conditions in the year 2064.
(This uniform rate of progress is sometimes referred to as the default glide-path).
The State is also to consider the statutory reasonable progress factors contained in
CAA section 169A(g)(l).406 In doing so, we anticipate that States will take into
account the degree to which CAIR emissions reductions are projected to bring
visibility conditions at its Class I areas in line with the default glide path. In some
States, the improvements expected from CAIR, combined with the application of
the reasonable progress factors to other source sectors, may result in a
determination that few additional emissions reductions are reasonable for the first
long-term strategy period. Nonetheless, each State is required to set its reasonable
progress goals as provided by the regional haze rule and cannot assume that CAIR
will satisfy all of its visibility-related obligations.407
UARG commented that our Reasonable Progress Guidance states that BART is likely to satisfy
all reasonable progress requirements for the first planning period. This is a misrepresentation of
our Reasonable Progress Guidance. The actual text is:
Note that for some sources determined to be subject to BART, the State will
already have completed a BART analysis. Since the BART analysis is based, in
part, on an assessment of many of the same factors that must be addressed in
establishing the RPG, it is reasonable to conclude that any control requirements
imposed in the BART determination also satisfy the RPG-related requirements for
source review in the first RPG planning period. Hence, you may conclude that no
additional emissions controls are necessary for these sources in the first planning
period.
Note that our Reasonable Progress Guidance refers to instances in which "control requirements
[have been] imposed" on a sources as a result of a BART analysis. It is conceivable that
participation in CAIR would lead to actual control of a particular source. However, as Texas
itself has noted in its SIP, the IPM model analysis used by CENRAP predicts that by 2018 EGUs
in Texas will purchase approximately 125,000 tpy of emissions allowances from out of state.408
406 CAA Section 169A(g)(l) goes on to state that, in determining "reasonable progress," states must consider four
factors: "the costs of compliance, the time necessary for compliance, and the energy and nonair quality
environmental impacts of compliance, and the remaining useful life of any existing source subject to such
requirements." This consideration is commonly referred to as the "four-factor analysis." 406 Correspondingly,
under Section 51.308(d)(1) of the Regional Haze Rule, promulgated in response to this mandate, states must
"establish goals (expressed in deciviews) that provide for reasonable progress towards achieving natural visibility
conditions" for each Class I area within a state. Reasonable progress goals are interim goals that represent
measurable, incremental visibility improvement over time toward the goal of natural visibility conditions. Section
51.308(d)(l)(i)(A) requires states to consider the four statutory factors when establishing their reasonable progress
goals.
407 70 FR 39104, 39143.
408 Texas Regional Haze SIP, page 10-9.
505
-------
This, combined with the precipitous decline in the CAIR allowance market,409 makes it
questionable that any of the BART sources included in our FIP would have installed controls as a
result of CAIR. In addition, as noted elsewhere in this document with regard to CAIR's
successor, on July 28, 2015, the D.C. Circuit Court's issued its decision inEMEHomer City
Generation v. EPA, 795 F.3d 118 (D.C. Cir 2015), upholding CSAPR but remanding without
vacating a number of the Rule's state emissions budgets. Thus, potential control under
CAIR/CSAPR as a substitute for BART is moot. This is consistent with our statement regarding
the Georgia SIP cited by the commenter, where we noted that any control requirements imposed
as BART would also satisfy reasonable progress. EPA's determination that CSPAR would
provide for greater reasonable progress than BART did not result in the imposition of any control
requirements imposed as BART.
Please see our responses to other comments alleging that we erred in proposing controls that
would not be operational until after 2018, and concerning monitored progress of Big Bend, the
Guadalupe Mountains, and the Wichita Mountains.
Comment: Earthjustice stated that, both states and EPA have required controls that provide
benefits of less than 0.5 dv. See, e.g., FIP TSD at A-75 (mentioning controls required in
Wyoming and Arizona that would provide a visibility benefit less than 0.5 deciviews).
Response: We agree with Earthjustice that both states and we have required controls on sources
that provide less than a 0.5 dv visibility improvement. As we note in the above citation to our
FIP TSD, in our FIP for Wyoming, we controlled some sources that had a benefit of 0.3 dv using
CALPUFF410. We explained that given the modeled emissions differences between reasonable
progress with CAMx and BART with CALPUFF and difference in metrics it can be argued that a
0.3 dv benefit with CALPUFF would be on the order of 0.1-0.15 deciview benefit with CAMx
modeling. We noted that this is an estimate just based on emissions and metrics differences and
ignoring the other differences we discussed in our FIP TSD. We also noted that we finalized a
FIP in Arizona that included controls that resulted in a 0.18 -dv and 0.24 dv benefit based on
CALPUFF modeling.411 With regard to that action, we noted that given the modeled emissions
differences between reasonable progress analysis with CAMx and BART analysis with
CALPUFF and difference in metrics it can be argued that a 0.18-0.24 dv benefit with CALPUFF
would be on the order of 0.1 dv or less benefit with CAMx modeling (again, this is an estimate
just based on emissions and metrics differences and ignoring the other differences we discussed
in our FIP TSD).
409 See for instance, Schmalensee, R. and Stavins, R., The SO2 Allowance Trading System: The Ironic History of a
Grande Policy Experiment, Journal of Economic Perspectives, 8/3/2012; Hitaj, C., and Stocking, A., Market
Efficiency and the U.S. Market for Sulfur Dioxide, Working Paper Series Congressional Budget Office,
Washington, D.C., Working Paper 2014-01
410 Wyoming Final FIP FR Vol. 78, No. Ill; pages 34785-,34789.
411 Arizona Final FTP FR Vol 79, No. 170; pages 52464-52477.
506
-------
Comment: Luminant states that the limits EPA is proposing in its FIP are more stringent than
even Best Available Retrofit Technology ("BART") limits for existing EGUs that EPA has
recently approved.826
Footnotes:
826 See, e.g., Alaska: 0.30 lb/MMBtu for Healy Unit #1 (78 Fed. Reg. 10,546, 10,549 (Feb. 14, 2013)); Arizona:
0.23 lb/MMBtu for the Sundt Generating Station (79 Fed. Reg. 9318, 9325 (Feb. 18, 2014)), 0.15 lb/MMBtu for
Cholla and Apache, and 0.08 lb/MMBtu for Coronado (77 Fed. Reg. 72,512, 72,515 (Dec. 5, 2012)); Colorado:
0.11 lb/MMBtu for Tri-State Craig, 0.12 lb/MMBtu for Comanche Station, 0.13 lb/MMBtu for Hayden, and 0.13
lb/MMBtu and 0.26 lb/MMBtu for Martin Drake (77 Fed. Reg. 18,052, 18,073 (March 26, 2012)); Kansas: 0.10
lb/MMBtu for La Cygne, and 0.15 lb/MMBtu for Westar Jeffrey (75 Fed. Reg. 80,754, 80,758 (Dec. 27, 2011));
Montana: 0.08 lb/MMbtu for Colstrip Units 1 and 2, and 0.57 for JE Corette (77 Fed. Reg. 57,864, 57,915 (Sept.
18, 2012)); North Dakota: 0.15 lb/MMBtu for Leland Olds Station, Milton R. Young Station, and Coal Creek
Station, 0.24 lb/MMBtu and 0.16 lb/MMBtu for Stanton Station (76 Fed. Reg. 58,570, 58,595 (Sept. 21,2011));
Nevada: 0.15 lb/MMBtu for Reid Gardner (77 Fed. Reg. 17,334, 17,338 (March 26, 2012)); New York: 0.09
lb/MMBtu for Danskammer Generating Station (77 Fed. Reg. 51,915, 51,928 (Aug. 28, 2012)); Oregon: 0.40
lb/MMBtu for PGE Boardman (76 Fed. Reg. 38997, 39,002 (July 2, 2011)); South Dakota: 0.09 lb/MMBtu for Big
Stone (77 Fed. Reg. 24,845, 24,848 (April 26, 2012)).
827 79 Fed. Reg. at 74,885.
Response: We proposed the following SO2 emission limits, based on a 30 Boiler Operating Day
(BOD) average.
Proposed
SO2 Emission
Limit
Unit
(lbs/MMBtu)
Sandow 4
0.20
C/3
(L)
Martin Lake 1
0.12
"O
Otj
a
P
Martin Lake 2
0.12
Martin Lake 3
0.11
•~
0)
A
Monticello 3
0.06
A
S
•~
Limestone 2
0.08
w
in
Limestone 1
0.08
San Miguel*
0.60
Big Brown 1
0.04
VI
Big Brown 2
0.04
O)
V
Monticello 1
0.04
Monticello 2
0.04
A
A
Coleto Creek 1
0.04
w
m
Tolk 172B
0.06
Tolk 17 IB
0.06
507
-------
We note that some of these limits are above the limits Luminant cites, some are below, and some
are within the range of the limits Luminant cites. For our wet FGD scrubber retrofits, we based
our proposed limits on 98% control with a floor of 0.04 lbs/MMBtu. For our SDA scrubber
retrofit (Tolk only), we based our proposed limits on 95% control with a floor of 0.06
lbs/MMBtu. With the exception of San Miguel412, we based our proposed limits for our wet
FGD scrubber upgrades on 95% control with a floor of 0.04 lbs/MMBtu. This accounts for the
range in values in the above table. In other words, 95% control of a coal with a higher sulfur
content (e.g., Sandow 4), will produce an emission limit that is necessarily higher than 95%
control of a coal with a lower sulfur content (e.g., Limestone). As the percentage of sulfur in
coal varies widely across the U.S., and among the facilities Luminant cites, we do not believe
that the simplistic comparison Luminant proposes (which focuses only on the numerical value of
the limit and ignores the coal sulfur content) is valid.
As we note in our proposal,413 and elsewhere in our response to comments, we believe that
percentage control we have proposed is significantly below the maximum level of control that
can be expected. Furthermore, although we cannot provide citations here due to the
confidentiality protections we are required to extend to those companies that responded to our
information requests, we have seen several instances of engineering firms providing wet
scrubber upgrade and retrofit guarantees of greater than 95% control, and performance testing
following scrubber upgrades resulting in greater than 95% control. We believe that the control
levels we proposed remain appropriate.
Comment: CCP and Xcel stated that the EPA arbitrarily and unreasonably failed to compare the
incremental costs and environmental benefits associated with alternative controls. EPA has done
this comparative cost-incremental analysis for source-specific RPG controls in other regional
haze SIPs/FIPs. See 76 Fed. Reg. 58,631 (Sept. 21, 2011) (analyzing RPGs for a source in the
North Dakota FIP using incremental cost effectiveness analyses for various controls of NOx
emissions).
Xcel stated that EPA has previously accepted a 20-year amortization period for the life of these
types of controls. EPA, Air Pollution Cost Control Manual, at 3-33 (2002) (stating amortization
over 20-30 years is appropriate); Wyoming Regional Haze FIP, 79 Fed. Reg. 5032, 5064-65
(Jan. 30, 2014) (using a 20-year amortization period for the Dave Johnson and Naughton
plants); Arizona FIP, 79 Fed. Reg. 52,240, 52,459 (Sept. 3, 2014) (assuming a 20-year
amortization period); Montana FIP, 77 Fed. Reg. 57,864, 57,882 (Sept. 18, 2012) (using a 20-
year amortization period).
S&L stated that EPA has used a 20-year scrubber useful life in several Regional Haze BART
determinations.27 It is also important to note that the majority of these evaluations were
412 In San Miguel's case, we concluded that the scrubber upgrades it had already performed were reasonably
reflective of the limits of wet scrubber technology, and our proposed S02 limits merely served to lock-in what San
Miguel was already achieving. We discuss San Miguel's emission limits more thoroughly in our response to its own
comment elsewhere.
413 See our Cost TSD, beginning on page 38.
508
-------
completed for retrofit FGD systems (i.e., new systems) and not upgrades to existing scrubbers
and related equipment that has already been operating for more than 30 years.
Stamper (enviros) agreed that, if an EGU owner indicates that a shorter life of the SO2 controls
should be evaluated in EPA's cost effectiveness determination due to a planned shutdown of a
unit, then that shorter lifetime needs to be made into an enforceable requirement. This is
consistent with how EPA has considered shorter equipment lifetimes in cost analyses for BART
determinations, and there is no justification for a different approach for cost effectiveness
analyses done for reasonable progress requirements. For example, Stamper noted that the
owners of Dave Johnston Unit 3 in Wyoming informed EPA that the unit would be shut down in
2027. 79 Fed.Reg. 5045 (January 30, 2014).
Luminant stated that, "[tjhere is no particular threshold for determining significance of visibility
benefit in the regional haze rule."839 Further, "States have latitude to determine these thresholds,"
and "[a]s long as this evaluation is done adequately and the states provide a reasoned basis for
their decisions, EPA will defer to the state "84° EPA has forgotten its own practices and
standards here. Texas fully explained and supported its cost effectiveness analysis, pursuant to
EPA's regulations and guidance, and EPA does not find otherwise. EPA must approve Texas's
submission.
The Associations stated that not all emissions reductions will have the same impacts on visibility
in Class I areas. Key factors such as the type of pollutant at issue, distance from Class I areas,
and prevailing winds can all affect the degree to which certain emissions will contribute to
visibility impairment and, as a result, the visibility benefits that will be produced by reducing
those emissions. In other words, not every ton of emissions reductions is the same. Recognizing
this fact, EPA explains in guidance that "in assessing emission reduction strategies for source
categories or individual, large scale sources, simple cost effectiveness estimates based on a
dollar-per-ton calculation may not be as meaningful as a dollar-per-deciview calculation,
especially if the strategies reduce different groups of pollutants." EPA, Guidance for Setting
Reasonable Progress Goals Under the Regional Haze Program 5-2 (June 1, 2007). EPA has
reaffirmed that view in subsequent SIP reviews, stating its belief "that dollars per deciview is one
of several metrics that can be used to analyze cost of visibility improvement." 77 Fed. Reg.
40,150, 40,156 (July 6, 2012). That is, because of distance, wind patterns and other relevant
meteorological factors, even emissions that might be quite inexpensive to reduce may have no
meaningful impact on downwind visibility.
The Associations stated that the human eye cannot detect changes in visibility of less than one
deciview and, under EPA's own statistical standards, these 2018 "improvements" would be
treated as nonexistent.4 That is why, in another FIP proposal, EPA recently concluded that a
similar incremental visibility improvement was minimal and could not justify the much smaller
difference in cost between wet and dry SO2 scrubbers. EPA, Arkansas FIP Proposal,
Prepublication Version 160-61 (Mar. 6, 2015).
Footnotes:
27 See, e.g., Best Available Retrofit Technology (BART) Determination, American Electric Power, Northeastern
Power Plant, May 30, 2008. BART Five Factor Analysis, Kansas City Power & Light, La Cygne Generation
Station, August 2007.
509
-------
839 77 Fed. Reg. 40,150, 40,156 (July 6, 2012) (approval of Nebraska SIP).
840 Id. (emphasis added).
4 U.S. EPA, Technical Support Document for Demonstration of the Transport Rule as a BART Alternative 24, n.24
(Dec. 2011), Docket ID No. EPA-HQ-OAR-2011-0729-0014 ("All differences that are < 0.05 [deciviews] were
rounded down to 0.0 and are considered to be no degradation").
Response: CCP and Xcel stated that we arbitrarily and unreasonably failed to compare the
incremental costs and environmental benefits associated with alternative controls. They note that
we have done this comparative cost-incremental analysis for source-specific RPG controls in
other regional haze SIPs/FIPs. We disagree that we have ignored environmental costs and
benefits in the reasonable progress analysis in Texas. When performing our control cost analysis
for those facilities that did not have any SO2 control, we also analyzed the next less effective
control option, which is DSI.414 Where NOx controls are at issue, incremental analyses play a
more important role, given the range of options for NOx controls. Thus, we disagree with CCP
and Xcel to include consideration of incremental visibility benefits and control costs.
Xcel stated that we have previously accepted a 20-year amortization period for the life of these
types of controls, citing to our Cost Control Manual, our Wyoming, Arizona, and Montana
Regional Haze FIPs. We believe the citation Xcel makes to our Control Cost Manual refers to
page 3-33 of Chapter 3, "Permanent Total Enclosures" in Section 2, "Generic Equipment and
Devices." The full quote is:
For a PTE, the economic life is the same as the life of the building which might be
20-30 years or of the particular equipment enclosed by the PTE which might be
less. The interest rate value recommended by the Office of Management and
Budget (OMB) is 7 percent. (This replaces the 10 percent rate previously
recommended by OMB.) An economic life of 30 years and an interest rate of 7
percent yields a CRF of 0.080586 [emphasis added].
This is a calculation example for a permanent total enclosure. Our Control Cost Manual has
many calculation examples using different values for the economic life. None of these examples
are meant to specify the actual economic life that should be used for a scrubber. Nevertheless, as
the remainder of the full quote reveals, an economic life of 30 years was in fact used in
calculating the capital recovery factor for that example, which is exactly what we used in all of
our cost analyses.
Regarding S&L's reference to our BART determination for the American Electric Power (AEP)
Northeastern Units 3 and 4, S&L fails to disclose that in that case our Oklahoma FIP415 included
a BART determination for the AEP Northeastern facility that assumed a 30 year life. Subsequent
to that, Oklahoma submitted a SIP that we approved416 that included a revised BART
determination for the AEP Northeastern facility that included requirements that one of the units
shut down by April 16, 2016. The remaining unit is required to comply with a reduced emission
limit of 0.40 lbs/MMBtu (based on DSI control) beginning on April 16, 2016, reducing its
capacity utilization between 2021 and 2026, and ultimately shutting down December 31, 2026.
414 See our Cost TSD for more details.
415 76 FR 81728.
416 79 FR 12944 and 79 FR 12954.
510
-------
Thus, the use of a reduced operating life was appropriate because of the enforceable
commitments described above. Unlike the AEP Northeastern situation, none of the units
affected by our FIP has volunteered to enter into enforceable commitments that would result in
an operating life of less than 30 years.
Regarding S&L's reference to the Kansas City Power and Light La Cygne Units 1 and 2 BART
determinations, these were not included in a FIP—we approved those determinations, submitted
to us by the State of Kansas, except for certain startup and shutdown exemptions. Unit 1 had an
existing SO2 scrubber and was being equipped with an SCR at the time of the BART evaluation.
The State of Kansas did use a 20 year life in assessing BART for these units. The final BART
determination was based on these controls for Unit 1 and the evaluation of whether SCR and a
scrubber was merited for Unit 2. Ultimately, an emission rate of 0.10 lbs/MMBtu corresponding
to the installation of an additional scrubber on Unit 2 was deemed BART. Accordingly, Kansas'
use of a 20 year life did not impact SO2 BART. For NOx BART, Kansas determined that a
weighted average emission rate of 0.13 lbs/MMBtu applied to Units 1 and 2 together was BART.
This was based on a SCR on Unit 1 and combustion controls for Unit 2. The decision not to
require SCR on Unit 2 was based on the determination that the additional visibility benefit of
SCR over the combustion controls was too small.417 Consequently, the use of a 20 year life was
not the determining factor. In any event, it appears that a SCR was ultimately installed on Unit 2
41 8
anyway
Regarding Xcel's reference to our Wyoming FIP, we note that the units referenced, the Naughton
and Dave Johnson facilities were SCR cost analyses, unlike the proposed scrubber in our
Texas/Oklahoma FIPs. In addition, EPA Region 8, which processed the Wyoming FIP, accepted
shorter remaining useful lives for these units based on information supplied by the company,
PacificCorp.419 We agree with Stamper that the owners of Dave Johnston Unit 3 in Wyoming
informed EPA that the unit would be shut down in 2027, and an EGU can use a shorter life by
entering into an enforceable commitment with us and its state.
Regarding Xcel's reference to our Arizona FIP, we used a 20 year life to assess the control costs
for an SNCR installation on a kiln for the Phoenix Cement Company Clarkdale Plant, which is a
much different type of source than the coal fired power plants we evaluated in our proposed
Texas/Oklahoma FIPs. Also, SNCR, a NOx control is a much different type of control than the
SO2 controls we evaluated in our proposed Texas/Oklahoma FIPs. Similarly in our Montana
FIP, we used a 20 year life to assess the SNCR control costs for the Holcim Cement kiln.
Consequently, we do not see any conflict with our use of a 30 year life for assessing the cost
effectiveness of the coal fired power plants we proposed to control in our Texas/Oklahoma FIPs.
We agree with Luminant that there is no particular threshold for determining the significance of
visibility benefit in the regional haze rule. We note the full quote Luminant references is:420
417 76 FR 52616. See also 76 FR 80758.
418 http ://www. sargentlundy. com/proj ects/la-cygne-unit-2/.
419 79 FR 5165.
420 77 FR 40156.
511
-------
There is no particular threshold for determining significance of visibility benefit
in the regional haze rule. Significance is a source- and Class I-specific
evaluation, meaning that it depends on how much visibility improvement is
needed at the Class I area(s), how much a specific source impacts the Class I
area(s), and the cost effectiveness and potential visibility improvement of
available control options. States have latitude to determine these thresholds,421
providing support and a reasonable and adequate basis for why they selected the
thresholds, and to determine BART and reasonable progress controls, in
consultation with other impacted states. As long as this evaluation is done
adequately and the states provide a reasoned basis for their decisions, EPA
will defer to the state [emphasis added].
As we indicate in the above quote, we will defer to the state, "as long as this evaluation is done
adequately and the states provide a reasoned basis for their decisions." As we discuss in our
proposal, we disagree that Texas has in fact supplied a reasoned basis for many aspects of its
decisions on whether to control visibility impacting facilities within its borders. Thus, we
disagree with Luminant that we have "forgotten [our] own practices and standards" and we
disagree that Texas fully explained and supported its cost effectiveness analysis, pursuant to our
regulations and guidance.
The Associations point to our Reasonable Progress Guidance422 and our Nebraska FIP423 as
support for the use of the dollars per deciview metric. We acknowledge that we have in these
documents and elsewhere discussed the potential use of the dollars per deciview metric.
However, as we note in our Oklahoma FIP:424
[T]he BART Guidelines require that cost effectiveness be calculated in terms of
annualized dollars per ton of pollutant removed, or $/ton 425 OG&E provided a
$/deciview analysis for its units and comparable BART determination performed
by us. In our analysis for our BART FIP for OG&E and AEP/PSO, we did not
evaluate $/deciview. We explain that the BART Guidelines list the $/deciview
metric as an optional cost effectiveness measure that can be employed along with
the required $/ton metric for use in a BART evaluation. The metric can be useful
in comparing control strategies or as additional information in the BART
determination process; however, due to the complexity of the technical issues
surrounding regional haze, we have never recommended the use of this metric as
a cutpoint in making BART determinations. We note that to use the $/deciview
421 BART guidelines at 70 FR 39170: However, we believe the States have flexibility in setting absolute thresholds,
target levels of improvement, or de minimis levels since the deciview improvement must be weighed among the five
factors, and States are free to determine the weight and significance to be assigned to each factor. For example, a
0.3, 0.5, or even 1.0 deciview improvement may merit stronger weighting in one case versus another, so one "bright
line" may not be appropriate.
422 Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program, June 1, 2007 rev., page 5-2.
423 77 FR 40156.
424 Response to Technical Comments for Sections E. through H. of the Federal Register Notice for the Oklahoma
Regional Haze and Visibility Transport Federal Implementation Plan, Docket No. EPA-R06-OAR-2010-0190,
12/13/2011, pdf 116.
425 70 FR 39167.
512
-------
metric as the main determining factor would most likely require the development
of thresholds of acceptable costs per deciview of improvement for BART
determinations for both single and multiple Class I analyses. We have not
developed such thresholds for use in BART determination made by us. As
OG&E acknowledges, EPA did not use this metric as part of its proposed BART
determinations for either the Four Corners Power Plant FIP in AZ, or the San Juan
Generating Station FIP in NM. Generally speaking, while the metric can be
useful if thoughtfully applied, we view the use of the $/deciview metric as
suggesting a level of precision in the calculation of visibility impacts that is not
justified in many cases. While we did not use a $/deciview metric, we did,
however, consider the visibility benefits and costs of control together, as noted
above by weighing the costs in light of the predicted visibility improvement.
Our decision was reviewed and upheld in Oklahoma v. EPA, 723 F.3d 1201 by the Tenth Circuit
which ruled:
Oklahoma first suggests EPA should not have rejected the visibility analysis it
conducted in the SIP, which used the dollar-per-deciview method. This argument
is misguided. The EPA rejected the SIP because of the flawed cost estimates.
When promulgating its own implementation plan, it did not need to use the same
metric as Oklahoma. The guidelines merely permit the BART-determining
authority to use dollar per deciview as an optional method of evaluating cost
effectiveness. See 40 C.F.R. pt. 51 app. Y(IV)(E)(1) 426
And in the final rule, the EPA explained why it did not use the dollar-per-
deciview metric used by Oklahoma. "Generally speaking, while the metric can be
useful if thoughtfully applied, we view the use of the $/deciview metric as
suggesting a level of precision in the calculation of visibility impacts that is not
justified in many cases." 76 Fed.Reg. at 81,747. The EPA has never mandated the
use of this metric, and has not developed "thresholds of acceptable costs per
deciview improvement." Id. While the federal land managers have developed
thresholds, these thresholds were apparently developed without input from the
EPA and without notice-and-comment review. EPA Br. at 54 n. 13. In light of
this, we do not find it arbitrary or capricious that the EPA chose not to use the
dollar-per-deciview metric in evaluating BART options in creating the FIP. We
therefore also conclude that any argument by the petitioners that the dollar-per-
deciview measurement proves the scrubbers are not cost effective lacks merit.
See Pet. Reply Br. at 16.
426 We note, however, that in both its final rule and in its brief the EPA asserts that the guidelines require the use of
the dollar-per-ton metric in evaluating cost effectiveness. The guidelines themselves are a bit unclear. In the section
on cost effectiveness, the guidelines mention only the dollar-per-ton metric. 40 C.F.R. pt. 51 app. Y(IV)(D)(4)(c).
However, the guidelines later state that in evaluating alternatives, "we recommend you develop a chart (or charts)
displaying for each of the alternatives" that includes, among other factors, the cost of compliance defined as
"compliance — total annualized costs ($), cost effectiveness ($/ton), and incremental cost effectiveness ($/ton),
and/or any other cost-effectiveness measures (such as $/deciview)." Id. app. Y(IV)(E)(1) (emphasis added).
513
-------
We see no reason, despite that fact that the facilities we evaluated in our proposed
Texas/Oklahoma FIPs were done under the reasonable progress and long-term strategy sections
of the Regional Haze Rule, to deviate from our view of the dollar per deciview metric here. We
also note that the use of the dollar per deciview metric is further complicated in the present case
due to our use of CAMx modeling. As we discuss in our proposal and elsewhere in our response
to comments, there is no way to directly compare the CAMx modeled visibility impacts and
benefits we used in our proposed Texas/Oklahoma FIPs with previous CALPUFF modeling
results.427 Consequently, even if we were to use the dollar per deciview metric in our
Texas/Oklahoma FIPs, we would have no way to compare the results against other modeling and
cost analyses, the vast majority of which employed CALPUFF.
The Associations stated that the human eye cannot detect changes in visibility of less than one
deciview and allege that our "Technical Support Document for Demonstration of the Transport
Rule as a BART Alternative," proves their point that the 2018 visibility improvements in our
proposed Texas/Oklahoma FIP should be treated as nonexistent. They cite to our "Technical
Support Document for Demonstration of the Transport Rule as a BART Alternative" that
references visibility degradations of less than 0.05 deciviews. We address this comment in a
separate response to comment where we discuss consideration of cost versus visibility.
The Associations also point to our proposed Arkansas FIP stating we concluded that a similar
incremental visibility improvement was minimal and could not justify the much smaller
difference in cost between wet and dry SO2 scrubbers. As we have noted elsewhere in our
response to comments, because the visibility modeling in our proposed Arkansas FIP used the
CALPUFF model, while that in our proposed Texas/Oklahoma FIPs used the CAMx model, the
results are not comparable.428 We consequently do not agree with the Associations' contention
that we should conclude that the visibility improvements we modeled in our Texas/Oklahoma
FIPs were not meaningful.
Comment: Luminant stated that the projected deciview improvement from the additional
individual unit controls that EPA proposes are well below the level of improvement that EPA has
previously found do not warrant "reasonable progress" controls. For example, in a final rule
partially approving Wyoming's regional haze SIP, EPA determined that additional EGU
emission controls projected to achieve visibility improvements as high as 0.18 deciview per unit
at the most impacted Class I area were not warranted due to their "relatively modest" visibility
benefit.711 Similarly, for Montana, EPA found that projected visibility benefits from additional
EGU emission controls as high as 0.273 deciview per unit at the most impacted Class I area were
"not sufficient for us to consider it reasonable to impose this option in this planning period."712
Here, by contrast, EPA is claiming that controls on individual Luminant units that are modeled
by ENVIRON (EPA's own contractor) to achieve as little as 0.0136 deciview at the most
impacted Class I area in 2018 (Sandow Unit 4 at Wichita Mountains) would "result in significant
visibility improvements," and it proposes to require the controls on that basis.713 Indeed, for
427 See our FIP TSD, beginning on page A-35.
428 See our FIP TSD, beginning on page A-35, in which we explain why key differences in CALPUFF and CAMx
preclude the comparison of their respective results and why CAMx results for RP are generally much less than
CALPUFF results for BART for the same facility/emissions due to the model inputs and metrics used.
514
-------
Luminant's units, all of the visibility improvements that ENVIRON modeled for EPA are well
below the improvements that EPA found were insignificant in its Wyoming and Montana final
actions.714 EPA's inconsistent action here and its unique treatment of Texas sources are thus
arbitrary and capricious, in violation of EPA's regional consistency regulations, and unlawful.
EPA's attempts to avoid this inconsistency by "adjusting" the modeling that ENVIRON provided
to EPA are likewise arbitrary and capricious and unconvincing. As explained elsewhere in these
comments, these post hoc adjustments are unjustified, based on incorrect data, and not related to
the legal issue at hand (reasonable progress in 2018 judged by conditions in 2018). But even
accepting EPA's fictions and adjustments to the modeling that is commissioned, EPA's inflated
estimated benefits at the majority of the Luminant units that EPA would regulate are still below
the level EPA found in Montana to be "not sufficient" to warrant additional controls.715 In other
words, even at the highest level of benefits that EPA can put forward (which are miniscule), the
benefits are less than EPA found to be not sufficiently large to require additional controls. And
EPA cannot claim that lower costs justify the controls here—a cost-per-deciview analysis (which
EPA fails to conduct) demonstrates conclusively that none of the controls that EPA is proposing
for Texas sources is cost-effective as compared to the controls EPA rejected in Wyoming and
Montana.716
Luminant stated, even using EPA's data, the cost-per-deciview of EPA's proposal is
substantially higher than values that EPA has recently determined are unreasonable to impose in
the reasonable progress context. For example, in its recent review and partial approval of
Wyoming's regional haze SIP, EPA determined that new emission controls for Dave Johnston
Units 1 & 2 with a cost-per-deciview ranging from $10 to $39 million per deciview, depending
on the control and the unit,846 were unreasonable and were not required under the reasonable
progress requirements.847 EPA specifically explained that the cost, "while reasonable if viewed
in isolation, was not necessarily justified this planning period in light of the relatively modest
visibility improvement predicted by the revised modeling (0.11 deciviews—0.12 deciviews at
the most impacted Class I area). As a result, we are approving the state's reasonable progress
determinations of no new controls for Dave Johnston Units 1 and 2[.]"848 Similarly, in reviewing
Montana's SIP, EPA found that controls with a cost-per-deciview ranging from $13 to $23
million, depending on the control and the unit, were not required under the reasonable progress
requirements.849
According to Luminant, were EPA to apply these same standards to Texas sources, the result
would be the same—the small improvement that EPA has estimated would be outweighed by the
substantial cost. The costs-per-deciview of the controls that EPA is proposing for Texas sources
are substantially higher than the costs-per-deciview that EPA found justified Wyoming's and
Montana's decisions not to impose additional controls on their EGUs. Yet again, EPA is
applying a different standard to Texas sources and thus acting in violation of its regional
consistency regulations and in an arbitrary and capricious manner. As it did in Wyoming and
Montana, EPA must "approv[e] the State's reasonable progress determinations of no new
controls"850 for Texas EGUs.
Luminant stated that the EPA judges Texas's SIP using different and more stringent standards
than it has used in other states. For example, in two separate actions approving Idaho's and
Oregon's reasonable progress goals, EPA found that a total deciview impact on a Class I area of
515
-------
0.5 deciview was "relatively small" and that a facility with an impact at or below that level
should not be required to install additional controls to further reduce its impact.10 Here, in
contrast, EPA has determined that the largest impact from any one of Luminant's plants at a
Class 1 area is 0.15 deciviews—less than one-third of what EPA previously found to be
"relatively small" and an "unreasonable" basis for regulation—yet it subjects Luminant's plants
to drastic and costly emission reductions on this basis.11 EPA's proposal thus violates EPA's
regional consistency regulations and is arbitrary and capricious.
Comment: [Luminant (0061) p. 113] Luminant stated that when viewed in the proper context—
the deciview—the visibility impacts that EPA estimates here are clearly below thresholds that
EPA has previously and routinely determined do not warrant additional emission controls under
the reasonable progress requirements. For example, in recently reviewing and approving Idaho's
reasonable progress goals, EPA "independently evaluated whether there are reasonable control
measures available for sources located within Idaho's regulatory jurisdiction" and concluded that
facilities with visibility impacts of 0.5 deciviews or less at the nearest Class I area were
"relatively small,"703 Thus, EPA agreed with the state that additional controls for reasonable
progress "are not reasonable at this time, because even though there are cost effective controls
identified, visibility improvement is anticipated to be relatively small,"704 Further, in finalizing
its approval and responding to comments, EPA again confirmed that, even though "several of the
Idaho stationary sources have visibility impacts between 0.3-1.3 deciviews (dv)," those impacts
were not a "significant contribution to visibility impairment" that warranted reasonable progress
controls.705 EPA has used this same threshold in other states to conclude that "reasonable
progress controls" are not warranted 706
But here, Luminant noted that EPA Region 6 fails to follow the same standard that other EPA
Regions have used. Here, all of the visibility impacts that ENVIRON modeled, at EPA's
request, for Luminant's plants are well below 0.5 dv 707 Indeed, the largest impact that
ENVIRON modeled of any Luminant plant at any Class I area is only 0.151947 deciview
(Monticello's modeled impact at Wichita Mountains),708 less than one-third of the threshold that
EPA employed in its "independent[]" evaluation of Idaho sources to eliminate them from further
regulation.709 EPA fails to explain its departure from the 0.5 deciview threshold it previously
embraced, and its inconsistent action here is arbitrary and capricious and violates EPA's regional
consistency regulations. EPA's use of percentage extinction, as opposed to deciview impact,
appears to be an attempt to mask the fact that the Texas units here have relatively small impacts
on actual visibility at the Class I areas at issue. Under the standards that EPA itself has
previously established, the Luminant units at issue here have minimal impact on visibility and
should not be subject to further controls under the reasonable progress program—"even though
there are cost effective controls identified... ."710
NERA stated that EPA's decision making and the burdens it would impose on states and
individual sources should be commensurate, at least in some measure, across all U.S. states, and
EPA's own regulations require that it act consistently across all regions. If a FIP were to
produce radically different benefits on a dollar per deciview basis in one state, one would at least
expect EPA to provide a defensible basis for the different treatment, which EPA has not done in
its proposed regional haze FIP for Texas.
516
-------
According to NERA, the proposed regional haze FIP for Texas fails to treat Texas in a similar
fashion as other states and fails to provide a rational basis for doing so. For example:
• As EPA has explained in its proposal, the FGD upgrades at four plants are the only
controls that could potentially be implemented and operational during the first planning
period, which ends in 2018.4 Thus, as explained in Luminant's comments, only the
potential benefits of the FGD upgrades can be considered as a basis for EPA's proposed
FIP, which, like Texas's SIP, is limited to the first planning period. The visibility
benefits from the SO2 scrubber retrofits would be considered only in the subsequent
planning period. In this regard:
•
o improvement in visibility from the FGD upgrades would cost approximately $129
million per year per deciview to $651 million per year per deciview (depending
on the unit) on the 20% worst visibility days at WIMO, the Class I area that EPA
claims is the most affected,
o These estimates of dollars per deciview of improvement are mostly more than ten
times higher than costs that EPA has deemed unnecessary and unreasonable in
other regional haze rules. For the Wyoming FIP, additional controls with a cost
effectiveness of only $10 million per year per deciview were deemed not to be
warranted even though they each would have produced about twice as much
absolute visibility impact at the most affected Class I area as EPA's proposed
Texas FGD upgrades. For the Montana FIP, additional controls with a cost
effectiveness of $38 million per year per deciview were deemed not to be
warranted, also having higher absolute visibility impact.
• Even if the proposed new scrubber retrofits on seven Texas units could be considered to
meet the statutory factor that considers time necessary for compliance, those measures
are even less cost effective: between about $400 million per deciview and $1.3 billion per
deciview. These would be 40 to 130 times higher than the upper bound EPA required in
Wyoming, yet provide substantially less deciview benefit to their respective maximally
affected Class I areas.
NERA stated that these estimates of dollars per deciview of improvement are mostly more than
ten times higher than costs that EPA has deemed unnecessary and unreasonable in other regional
haze rules. For the Wyoming FIP, additional controls with a cost effectiveness of only $10
million per year per deciview were deemed not to be warranted even though they each would
have produced about twice as much absolute visibility impact at the most affected Class I area as
EPA's proposed Texas FGD upgrades. For the Montana FIP, additional controls with a cost
effectiveness of $38 million per year per deciview were deemed not to be warranted, also having
higher absolute visibility impact. Even if the proposed new scrubber retrofits on seven Texas
units could be considered to meet the statutory factor that considers time necessary for
compliance, those measures are even less cost effective: between about $400 million per
deciview and $1.3 billion per deciview. These would be 40 to 130 times higher than the upper
bound EPA required in Wyoming, yet provide substantially less deciview benefit to their
respective maximally affected Class I areas.
517
-------
According to NERA, although EPA has rejected using a bright-line dollars per deciview
threshold to justify additional controls for a reasonable progress analysis, cost effectiveness is a
tool that can assure consistency in regulatory decisions across multiple types of sources and
jurisdictions. If regulatory decisions were to result in source-specific controls with high cost
effectiveness in one situation while requiring controls on other sources in another situation with
much lower cost effectiveness, the overall policy would be inefficient. That is, it would produce
less societal benefit than could be achieved at the same societal cost but with a different mix of
control requirements. Cost-effectiveness analysis was developed specifically to avoid this
undesirable societal outcome.
NERA noted that EPA's proposed actions on Texas's and Oklahoma's regional haze SIPs are the
last of EPA's actions on SIPs submitted by states for the first planning period under the Regional
Haze program.22 This makes it possible to determine whether the controls being proposed in this
FIP are consistent with the cost effectiveness that EPA has required in other regional haze
actions for other states. We find that the upgrades proposed in the Texas FIP have extremely
high cost per deciview improvement {i.e., low cost effectiveness) compared to reasonable
progress controls that EPA has evaluated in other states. Inconsistency in cost effectiveness,
however, occurs if a reasonable progress control that EPA has deemed is unwarranted in another
State has an estimated cost per deciview that is substantially less than the cost per deciview
associated with the reasonable progress controls EPA now proposes for Texas.
NERA states that the upgrades proposed in the Texas FIP have extremely high cost per deciview
improvement {i.e., low cost effectiveness) compared to reasonable progress controls that EPA
has evaluated in other states. We find a clear case of such inconsistency in the decisions that
were made by EPA with respect to the regional haze SIPs of the States of Wyoming and
Montana when compared to what EPA now proposes for Texas.23 NERA noted, in its review of
Wyoming's regional haze SIP, EPA initially proposed that Dave Johnston units #1 and #2 should
be required to apply additional NOx controls for reasonable progress. Three levels of NOx
control were considered in EPA's analysis: low NOx burners with overfire air (LNB with OF A),
and LNBs with OFA in combination with either selective non-catalytic reduction (LNB with
OFA and SNCR) or with selective catalytic reduction (LNB with OFA and SCR). EPA found
that these measures each have increasing NOx reduction, increasing cost, and increasing visibility
improvement; they also have decreasing cost effectiveness. Calculations of the cost per deciview
of these options, based on EPA's data and estimates, are in Table 5. As can be seen, the cost
effectiveness of the reasonable progress options considered by EPA in its review of the
Wyoming SIP ranged from $10 million per deciview to $39 million per deciview. It is notable
that EPA could have proposed to require any one of the three levels of control, but decided that
not even the lowest of the three levels of control was warranted. Thus, even a control offering
improvement at the maximally-affected Class I area costing $10 million per deciview was not
seen to be cost effective enough for Wyoming.24 Information was not provided in the Wyoming
FIP that would enable a computation of the cost per deciview for the cumulative visibility
impacts of these controls.
AECOM stated that the EPA Region 6 has previously indicated that a visibility improvement of
0.2 delta-deciviews is too low for applying emission reductions. This action was for the
518
-------
Arkansas RHR SIP, for which EPA stated, "[w]ith regard to the comment that Arkansas sources
contributed 2.0% to visibility impairment at Wichita Mountains during the baseline period and
are projected to contribute 2.3% in 2018, EPA notes that removal of this 2.3% contribution to the
total extinction results in a visibility improvement of only 0.2 dv from the 2018 projected
visibility conditions. Consequently, while we are concerned that the RPG at Wichita Mountains
is not on the glide path, we believe the technical assessment that Arkansas sources do not have a
significant impact at Wichita Mountains is accurate and ADEQ and ODEQ followed consultation
procedures." 106
AECOM stated that the EPA separately determined for the State of Wyoming that a single
source's modeled visibility improvement of 0.19 dv for a Class I area was insignificant to require
additional controls. EPA stated that, "[although the cost-effectiveness for SNCR is reasonable,
we find it reasonable for the State not to select this control technology based on the incremental
visibility improvement for this control technology." 107 Similarly, EPA determined that a
visibility improvement of 0.17 dv was too low for application of emission controls. 108 These
cases indicate that modeled visibility improvements as high as about 0.2 dv109 have been
determined to be too low for requiring emission controls because such controls would result in
minimal improvements.
Xcel Energy noted that the EPA has previously rejected additional controls to achieve reasonable
progress even when visibility improvement was magnitudes greater than EPA expects from the
installation of scrubbers at the Tolk units:
• In Arkansas, EPA concluded that "a visibility improvement of only 0.2 dv" was too low
to apply further emission reductions even when Wichita Mountains was not on the URP
glidepath. Arkansas SIP Approval, 77 Fed. Reg. 14,604, 14,625 (March 12, 2012)
(emphasis added).
• In Arizona, EPA projected benefits of SCR at one project to be 0.41 dv at the most
affected Class I area but still rejected SCR for purposes of reasonable progress. This
level of improvement is more than nine times greater than the visibility improvement that
SDA scrubbers on Tolk would accomplish, at best, at the Guadalupe Mountains. Arizona
FIP, 79 Fed. Reg. 9,318, 9,360 (Feb. 18, 2014).
• In Montana, EPA found a 0.18 dv improvement to be a "low visibility improvement" that
"did not justify proposing additional controls" for SO2 on one source. This level of
improvement is more than four times greater than the visibility improvement that SDA
scrubbers on Tolk would accomplish, at best, at the Guadalupe Mountains. Montana FIP,
77 Fed. Reg. 23,988, 24,012 (Apr. 20,2012).
• In Oregon, even with relatively low costs per ton ($l,816/ton of NOx), minor visibility
improvements were rejected because "adding SNCR only provided an additional 0.18 dv
of visibility improvement over NLNB/MOF A at the Mt. Hood Wilderness Area." This
level of improvement is more than four times greater than the visibility of improvement
that SDA scrubbers on Tolk would accomplish, at best, at the Guadalupe Mountains.
Proposed Oregon SIP, 76 Fed. Reg. 12,651, 12,661 (March 8, 2011).
519
-------
EPA previously rejected similar cumulative visibility improvements of 0.254 dv and 0.273 dv in
the New York SIP as "small." New York SIP Approval, 77 Fed. Reg. at 24,818. Nevertheless,
EPA concluded for Texas that individual sources with de minimis deciview contributions and
similarly small deciview cumulative benefits warrant the substantial costs of additional controls
based on EPA's estimate of "extinction benefits and percentage of total extinction." 79 Fed. Reg.
at 74,882.
[CCP (0075) p. 11] CCP stated that the EPA has rejected controls in the RPG context even when
visibility was improved significantly more than EPA projects from the installation of a WFGD
scrubber at Coleto Creek Unit 1. For example, in Arizona, EPA projected benefits of selective
catalytic reduction technology ("SCR") at one project to be 0.41 dv at the most affected Class I
area but still rejected that control for purposes of reasonable progress. 79 Fed. Reg. 9,318,
9,360 (Feb. 18, 2014) ("Arizona FIP"). In Montana, EPA found a 0.18 dv improvement to be a
"low visibility improvement" that "did not justify proposing additional controls" for SO2 on one
source. 77 Fed. Reg. 23,988, 24,012 (Apr. 20, 2012) ("Montana FIP"). EPA separately
determined for the State of Wyoming that single source's modeled visibility improvement of
0.19 dv for a Class I area was insignificant to require additional controls.(78 FR 34751) And in
Arkansas, EPA concluded that 0.2 dv improvement was too low to apply further emission
reductions for Wichita Mountains. 77 Fed. Reg. 14,604, 14,625 (Mar. 12,2012).
[NRG (0078) p. 2] NRG stated that the EPA has also acknowledged that the actual visibility
impact of its proposal will be much smaller than the 0.284 deciview improvement. For the actual
worst 20 percent of days—the same dataset tracked under the reasonable progress rule EPA is
implementing-the visibility improvement from NRG's facilities would be only 0.057 deciview.2
NRG stated that, notably, EPA has previously found a larger, 0.5 deciview impact to be
insufficient to justify further regulation.3 Thus, it would be inconsistent for EPA to pursue
smaller changes in visibility by regulating Limestone as proposed.
Footnotes:
1 See 79 Fed. Reg. at 74,881, Tbls. 34-36 (Table 34 shows the greatest improvements at 0.135 and
0.149 deciview for Limestone Units 1 and 2, respectively, for the Wichita Mountains as against "average natural
conditions").
2 Id.
3 77 Fed. Reg. 30,248, 30,256 (May 22, 2012); 77 Fed. Reg. 30,454, 30,464 (May 23, 2012).
Footnotes:
11 EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_072913.xlsx.
703 77 Fed. Reg. at 30,255-56 (emphasis added).
704 Id. at 30,256 (emphasis added).
705 77 Fed. Reg. at 66,930-31.
706 See, e.g., 77 Fed. Reg. 30.454, 30,464 (May 23, 2012) (Oregon).
707 See Background, Table 3.
708 Id.
709 77 Fed. Reg. at 30,255.
710 Id. at 30,256; see also 77 Fed. Reg. at 30,464 (same).
520
-------
711 79 Fed. Reg. 5,032, 5,044 5,051 (Jan. 30,2014). EPA relies on its proposed rule for Wyoming that would
have required these controls (FIP TSD at A-75, n.52), but it ignores its final rule that does not.
712 77 Fed. Reg. 23,988, 24,064-67 (April 20, 2012) (proposed); 77 Fed. Reg. 57,864 (Sept. 18, 2012) (final).
713 79 Fed. Reg. at 74,883. Deciview improvement values modeled by ENVIRON are provided in Table 4.
714 See Table 4.
715 See Table 7.
716 See Section XI.B.
846 Based on the costs and visibility improvement values cited by EPA in its Wyoming rule, 79 Fed. Reg 5,032,
5044 (Jan. 30, 2014), NERA calculated these cost-per-deciview values. See NERA Report at 15-16.
847 79 Fed. Reg. at 5,051.
848 Id.
849 NERA Report at 16-17.
850 Id.
23 79 Fed. Reg. 5,032 (Jan. 30,2014).
24 We note that this $10 million/dv is for the 98th percentile deciview improvement, which will be larger than the
average deciview improvement on the 20% worst-visibility days, as is calculated for the proposed Texas FIP
controls. If the average change on the 20% worst days were 10% less than on the 98th percentile day, the cost
effectiveness would be $11 million/dv.
106 77 Fed. Reg. 14,625.
107 78 Fed. Reg. 34,751.
108 78 Fed. Reg. 34,752.
109 EPA's Regional Haze Rule notes that "no degradation" to visibility would be "defined as less than a 0.1 deciview
increase." (64 Fed. Reg. 35,730)
Response: Luminant, NERA and others cite to our Wyoming, Idaho and Montana FIPs and
allege that we are proposing controls in Texas based on visibility improvement values and cost-
per deciview values below what we have previously found do not warrant controls. Luminant
also alleged that we attempted to avoid this inconsistency by "adjusting" the modeling that
ENVIRON provided to us, stating these adjustments (which Luminant explores in more detail in
another comment) are still below the level we found in Montana to be "not sufficient" to warrant
additional controls. We address Luminant's comments concerning "adjustments" to the model
results elsewhere in this document we address comments concerning modeling.
As we have discussed in our proposal,429 our proposed controls were based on our simultaneous
reasonable progress and long-term strategy analyses. These analyses address both (1) the
requirements to consider the four reasonable progress factors for the Texas Class I areas, and (2)
the technical basis required to develop the long-term strategy for the Texas Class I areas and the
Wichita Mountains in Oklahoma. We used the "four factor analysis" method outlined in 40
CFR 51.308(d)(1)(A) that states are directed to use in establishing a RPG. We consider the costs
of compliance, the time necessary for compliance, the energy and non-air quality environmental
impacts of compliance, and the remaining useful life of any potentially affected sources. We
also discussed that, similar to a BART analysis, we were also considering visibility. We noted
that, with the exception of the Tolk facility, the issues relating to the evaluation of three of these
factors: (1) Time necessary for compliance, (2) energy and non-air quality environmental
impacts of compliance, and (3) remaining useful life, are common to all the units we are
analyzing.
That aside, as we have noted elsewhere in our response to comments, because the visibility
429 79 FR 74873.
521
-------
projections cited to by Luminant used the CALPUFF model, while those in our proposed
Texas/Oklahoma FIPs used the CAMx model, the results are not comparable. As we discuss
elsewhere in our response to comments, we also disagree with Luminant that we are required, or
that we should have, conducted a cost-per-deciview (i.e., $/dv) analysis. As we discuss in our
proposal and elsewhere in our response to comments, there is no way to directly compare the
CAMx modeled visibility impacts and benefits we used in our proposed TX/OK FIPs with
previous CALPUFF modeling results of visibility impacts and benefits used in other actions
because of differences in the models, model inputs, and metrics used.430 Many of these
differences result in CAMx modeled visibility impacts and benefits to be much lower than
CALPUFF modeled impacts and benefits relied on in other actions. For example, one difference
between the two model analyses is that CALPUFF modeling is focused on the maximum impact
from the modeled source, whereas the CAMx modeling is focused on the average impact over
the 20% worst days as determined by the monitored data during the baseline. The commenter's
estimates do not consider these differences and attempt to compare CAMx results from our
analysis to CALPUFF modeling results preformed for sources in other states. We consequently
do not agree with the Luminant's contention that we are being inconsistent with our previous
actions.
We also discuss in a separate response to comment that we disagree with the commenters'
assertions that we did not properly compare the cost effectiveness of our proposed controls to
their modeled visibility benefits. We also reject the commenters' assertions that this type of
comparison requires the $/dv metric. We take up the issue of Luminant's reference to
adjustments it states we made to our modeling in responses to comments concerning modeling
elsewhere in this document.
AECOM cited to our proposed Arkansas FIP and our final Wyoming FIP and Luminant cited to
our Idaho and Oregon actions, and NRG cited to our Idaho action as containing instances where
we concluded that visibility improvements greater than what we are proposing in our
Texas/Oklahoma FIPs were deemed too little to warrant controls. As we have noted elsewhere
in our response to comments, because the visibility projections cited to were obtained from the
CALPUFF model, while those in our proposed Texas/Oklahoma FIPs used the CAMx model, the
results are not comparable.
Xcel cites to our Arkansas, Arizona, Montana, Oregon, and New York actions and alleges that
we previously rejected controls that would have resulted in greater visibility than we now require
for Texas. Similarly, CCP cites to our Arizona, Montana, and Wyoming actions. The actions
cited all used CALPUFF modeling. We explain in our FIP TSD431 and elsewhere in our
response to comments, our selection of the photochemical grid model CAMx over CALPUFF to
assess visibility impacts and benefits for Texas sources. We also explained why the results from
our CAMx modeling analysis cannot be compared to CALPUFF results. As a consequence, we
disagree with Xcel and CCP.
We address the allegations that we should not have proposed controls that would be installed
beyond 2018 in our responses to other comments on the issue.
430 FIP TSD at A-3 5.
431 FIP TSD, Appendix A. See discussion beginning on page A-3 5.
522
-------
Comment: [Texas Governor (0066) p. 2-3] The Texas Governor explained that the EPA's
actions are irrationally and arbitrarily discriminatory against the State of Texas. Cf Nw. Austin
Mun. Utit. Dist. No. One v. Holder, 557 U.S. 193, 203 (2009) (emphasizing "our historic
tradition that all the States enjoy equal sovereignty" (internal quotation marks omitted)). It
appears that EPA has devised one set of rules for States it likes and another set for States it
dislikes.
The Texas Governor states that EPA approved California's "regional haze" SIP. See. 76 Fed.
Reg. 34,608 (2011). In doing so, EPA gave the State of California until the year 2307 to
eliminate "regional haze" at Desolation Wilderness and Mokelumne Wilderness, until the year
2106 to eliminate "regional haze" at Joshua Tree National Park, and until the year 2096 to
eliminate "regional haze" at Sequoia National Park. Apparently, for a State like California, EPA
thinks that up to 300 years constitutes "reasonable progress." 42 U.S.C. section 7491. EPA
faulted Texas's plan to eliminate regional haze even faster. In particular, Texas proposed to
eliminate "regional haze" in the Guadalupe Mountains by 2081 and in Big Bend by 2155. While
that rate of haze-elimination clearly would have been "reasonable" in California, EPA
determined that it was "not reasonable" in Texas. 79 Fed. Reg. at 74,843. The Texas Governor
explained that the EPA's capricious discrimination violates the "fundamental norm of
administrative procedure that] requires an agency to treat like cases alike. If the agency makes
an exception in one case, then it must either make an exception in a similar case or point to a
relevant distinction between the two cases." We star Energy, Inc. v. FERC, 473 F.3d 1239,1241
(D.C. Cir. 2007). EPA has done nothing to explain why one set of rules applies to California
while another, stricter set applies to Texas. This is the definition of arbitrary. The Texas
Governor stated that the only self-evident explanation for EPA's discrimination is that California
has fewer coal-fired power plants than does Texas. According to the latest data he has seen, the
Texas Governor stated that Texas has 40 coal-fired electric generating units ("EGUs") while
California only has 10. But Part C of the Clean Air Act does not give EPA the power to conduct
a witch hunt against coal; it only allows EPA to "protect visibility." And Texas's SIP would
reduce the same amount of visible haze as EPA's FIP while costing $2 billion less, and it would
reduce haze faster than California's plan would. EPA cannot premise its FIP authority on its
dislike of coal and/or its desire to play favorites between States.
Response: We do not believe that the type of simplistic comparison between Texas and
California that the comment from the Texas Governor has proposed is valid. There is no
requirement in the Regional Haze Rule, for a "one size fits all" date certain for achieving natural
conditions, as the Governor has apparently implied. Indeed, the Regional Haze Rule is written
expressly anticipating "reasonable progress" and these dates will be different, based on the
unique circumstances present in each state with regard to the number and type of pollution
sources, how much each state's pollution impacts Class I visibility (both in and out of the state),
the types of controls already present on a state's sources, and a number of other relevant factors.
With respect to the comment that EPA is approving a much slower rate of progress for California
Class I areas than Texas Class I areas, we note that the difference is that, as explained elsewhere,
Texas' analysis was not in conformance with the requirements of the statute, which further
analysis revealed resulted in reasonable controls not being required. We note that one reason for
523
-------
this is that Texas failed to reasonably estimate natural visibility conditions at Big Bend and
Guadalupe Mountains. Taking into account a more realistic value for natural visibility
conditions significantly changes the number of years for eliminating anthropogenic visibility
impairment. As shown in Table 44 of our proposal, we estimate that at the rate of visibility
improvement consistent with the controls required in the FIP, the number of years for
eliminating anthropogenic visibility impairment are 173 years for Big Bend and 141 years for the
Guadalupe Mountains. In addition, the difference is that additional reasonable controls of
Texas's sources were identified and that by not requiring those controls, the reasonable progress
goals of the affected Class I areas were not being properly determined. This was not the case in
California, and the California SIP provided sufficient explanation, in accordance with the
regional haze rule requirements, for why the rate of progress to reach natural conditions was
slower than the uniform rate of progress.
The comment also seems to be implying that we examined the number of coal fired power plants
in Texas and California, noted that Texas had more, and incorporated that knowledge in our
proposal. The difference in the number of coal fired power plants in Texas versus any other state
did not enter into our proposed decision making. However, it is reasonable to assume that the
greater number of uncontrolled or under controlled coal fired power plants a state has, the greater
will be its emissions. That is the case with Texas. We demonstrated that the emissions from
these sources impact the visibility at the Wichita Mountains in Oklahoma more than all of
Oklahoma's own point source categories combined. That fact did weigh heavily in our proposed
decision making. Thus, we disagree with the comment that we acted arbitrarily, or that we are
playing favorites with states, as the basis and appropriateness of our analysis and determinations
have been explained extensively throughout our responses and final action, and our proposal did
not result in a more stringent standard for Texas.
We note that the full paragraph of Cf. Nw. Austin Mun. Utit. Dist. No. One that the Texas
Governor cites to is as follows:
The Act also differentiates between the States, despite our historic tradition that
all the States enjoy "equal sovereignty." United States v. Louisiana. 363 U.S. 1.
16. 80 S. Ct. 961. 4 L. Ed. 2d 1025 (1960) (citing Lessee of Pollard v. Hasan.
44 U.S. 212. 3 How. 212. 223. 11 L. Ed. 565 (1845)); see also Texas v. White.
74 U.S. 700. 7 Wall. 700. 725-726. 19 L. Ed. 227 (1869). Distinctions can be
justified in some cases. "The doctrine of the equality of States . . . does not bar . .
. remedies for local evils which have subsequently appeared." Katzenbach. supra.
at 328-329. 86 S. Ct. 803. 15 L. Ed. 2d 769 (emphasis added). But a departure
from the fundamental principle of equal sovereignty requires a showing that a
statute's disparate geographic coverage is sufficiently related to the problem that it
targets.
557 U.S. at 203. We note that the federal law at issue in that case imposed requirements on only
certain States, i.e., disparate treatment of States on the face of the statute. The full context of the
citation reveals that the Court was explaining that, despite the principle of equal sovereignty, a
federal law's treatment of States can be disparate if the disparate treatment is sufficiently related
to the problem being addressed by the statute at issue. We disagree that this holding is at issue
524
-------
here. First, the CAA Regional Haze Rule provisions do not only apply to certain states but rather
apply to all States. Second, as we have discussed in detail elsewhere, our proposal is consistent
with the statute, regulations, guidance, and previous actions in other States. As we explain
elsewhere, any potential differences in our proposal's methodology from other State's is a result
of reasonable, scientifically sound implementation of the statute's requirements in a state as
geographically large and source numerous as Texas, and, most importantly, does not result in
Texas being held to a more stringent standard than other states. Again, and as stated in greater
detail elsewhere, what commenters allege should be national uniformity ignores our authority to
exercise of our judgment, based on the specific facts at hand in reviewing SIP actions. Here, we
are exercising judgment within the parameters laid out in the CAA and our regulations. Because
this is a SIP review action, we believe that we are not only authorized but required to exercise
independent technical judgment in evaluating the adequacy of the State's regional haze SIP, just
as we must exercise such judgment in evaluating other SIPs.
Comment: Luminant stated that in reviewing New Mexico's regional haze SIP, EPA has
already approved a higher (less stringent) reasonable progress goal for the Carlsbad Caverns
National Park (16.92 dv) that uses the same IMPROVE air quality monitor as Guadalupe
Mountains for establishing and determining reasonable progress.505 EPA's proposal to
disapprove Texas's more stringent goal (0.63 deciview lower than New Mexico's RPG)—for the
same monitored location—is the definition of arbitrary and capricious agency action. EPA
provides absolutely no explanation, much less support, for its decision that the more stringent
RPG established by Texas for the exact same monitored location for the same interim deadline is
unreasonable. Nor does EPA provide explanation or support for its completely new and different
proposed RPG for this site. The only mention in EPA's proposal about the common monitor for
the two areas relates to discussions among Texas, New Mexico, and Federal Land Managers
about natural visibility conditions for the two areas.507 There is absolutely no discussion of New
Mexico's RPG for the shared monitor or EPA's prior approval of that RPG as "reasonable."
Indeed, states Luminant, EPA approved New Mexico's RPG for this site based on the same
rationale Texas provided to support its more stringent RPG. As explained by EPA:
[New Mexico] reasonably concluded that the cost of additional controls was not
warranted and concluded that the RPGs are reasonable given projected emissions
reductions from anthropogenic sources and the fact that natural and out-of-state
sources contribute significantly to haze. Because the State has limited ability to
control naturally occurring wildfires and windblown dust, these sources of
visibility impairment will continue to impact visibility at New Mexico's Class I
areas and limit the visibility improvement achievable during the planning
period.508
Luminant states EPA's approval of New Mexico's RPG also illustrates the inconsistency and
arbitrary nature of EPA's treatment of Texas's SIP revision. If EPA were correct that Texas
emissions are impacting visibility at this monitor in a significant way, and that reasonable
progress for this area required emission reductions from sources in East Texas, EPA's position
here would require it to object to New Mexico's RPG for Carlsbad Caverns and "disapprove" the
525
-------
interstate consultation between New Mexico and Texas (and other "upwind" states). Tellingly,
EPA did not do so. Nor did it do so for other states (as discussed in Section III.C and elsewhere
in these comments) that consulted and agreed on apportionment of emission reductions for their
shared impact on Class I areas. EPA's different treatment and proposed disapproval of Texas's
SIP revision is unexplained, arbitrary, and unlawful.
NRG stated that the monitoring data used to determine baseline and natural visibility conditions
for the Guadalupe Mountains are the same dataset that was used to determine baseline and
natural visibility conditions for the nearby Carlsbad Caverns area in New Mexico. However, in
approving New Mexico's regional haze plan, EPA required the State to plan for achieving a less
stringent degree of visibility improvement by 2018. A comparison of the Texas and New
Mexico reasonable progress targets and associated visibility improvement appears in the
following table:
11 """ ' — "¦ - - - - - ^
T«»® and New Mexico Reasonable Progress Goals Based on Data from Guadalupe :
_____ _______ ________
I Guadafeipe Mountains (Texas,
| propQMd for disapproval)m
i
Od-ibce. Ca\ cms (Mew
Mexico, approved by
6PA}
Baseline visibility conditions,
20% worst days
I?,If decivtews (dv)
i/.l'J dv
target (m 2MB, 20?lorat
IS, 3 dv
16.92 dv
'fire, 20% walaw'tl,flCl
0.9 dv
0.27 dv
Baseline visibility conditions,
2®* best days
5,5 dv
5.95 tfv
Lrort fat I'OiS^loKesr
days
5.? dv
6,12 dv
Visibility Improvement squired
by 2018, 20% best days
0.2 civ
(visibility allowed to |
deteriorate on 20* best j
clays) |
NRG noted, as shown in the comparison above, that the 2018 Texas reasonable progress goals
for the Guadalupe Mountains are more stringent than the New Mexico goals that EPA approved
for the adjacent Carlsbad Caverns area. Thus, EPA is proposing to reject a visibility
improvement target that exceeds the improvements that EPA required of New Mexico under the
same baseline conditions at the same monitoring site. This result would be incoherent and
Inconsistent with EPA regulations requiring the agency to assure consistency of decision making
526
-------
through the "fair and uniform application by all Regional Offices of the criteria, procedures, and
policies employed in Implementing and enforcing the act." 40 C.F.R. § 56.3(a).
GCLC noted that EPA proposes to disapprove Texas' RPG of 16.3 dv for the Guadalupe
Mountains. This is despite the fact that EPA has already approved a higher RPG of 16.92 dv set
by New Mexico for the Carlsbad Caverns National Park. These are contiguous parks that share
the exact same, single, air quality monitor as the Guadalupe Mountains. Texas relied on similar
rationale as did New Mexico to establish its RPG for this area, and if any conclusion should have
been reached by EPA, it was that Texas' RPG was too stringent and should be relaxed based on
EPA's prior decision. EPA's proposal to disapprove Texas' RPG for this area is the definition of
an arbitrary and capricious action.
Footnotes:
505 77 Fed. Reg. 36,044, 36,071 (June 15, 2012) (proposed); 77 Fed. Reg. 70,693 (Nov. 27, 2012) (final).
506 IMPROVE 2000 Report at 1-10 to 1-12, tbl.1.2.
507 79 Fed. Reg. at 74,843. Further, while EPA claims that "we discuss the difference between the natural visibility
value calculated by New Mexico for Carlsbad Caverns and that calculated by Texas for the Guadalupe Mountains
elsewhere in our proposal," id. at 74,843 n.218, we have been unable to locate any such discussion in the proposal
or the docket.
508 77 Fed. Reg. at 70,701.
30 79 Fed. Reg. at 74,833.
31 See 77 Fed. Reg. 36,044, 36,071 (June 15, 2012) (finalized at 77 Fed. Reg. 70,693 (Nov. 27,
2012)).
Response: We disagree with Luminant and other commenters that because we approved the
16.92 dv reasonable progress goals for Carlsbad Caverns National Park in New Mexico we
should approve Texas' 16.3 dv reasonable progress goal for Guadalupe Mountains in Texas. The
commenters base this argument on the fact these two Class I areas share a visibility monitor.
This, commenters reason, means that each Class I area should have the same reasonable progress
goal.
The comments indicate a lack of understanding of how reasonable progress goals are established,
as well as the imports of the goals as opposed to the measures adopted to ensure reasonable
progress. First, as we state in the Regional Haze Rule, the reasonable progress goal(s) set by the
state are not enforceable. The reasonable progress goals represent the State's best estimate of the
degree of visibility improvement that will result in 2018 from changes in emissions inventories,
changes driven by the particular set of control measures the state has adopted in its regional haze
SIP to address visibility, as well as all other enforceable measures expected to reduce emissions
over the period of the SIP from 2002 to 20 1 8.432 Given the forward looking nature of reasonable
progress goals and the range of assumptions that must be made as to emissions a decade or more
in the future, we expect there to be some uncertainty in the estimates of visibility in 2018.
There are many Class I areas located in neighboring states that are in relatively close proximity,
as indicated by the following map:433
432 64 FR 35733.
433 http://www.epa.gov/visibility/classimp.gif.
527
-------
Mandatory Class I Areas
There are 156 Class I areas, but only 110 monitors, requiring that many of these Class I areas
share monitors, as shown by the following table:
Monitor
Code
State
Site Name
Class I Area(s)
ACAD
Maine
Acadia National Park
Acadia. ME
MOOS
Maine
Moosehorn NWR
Moosehorn,ME; Roosevelt Campobello, ME
LYBR
Vermont
Lye Brook Wilderness
Lye Brook, VT
GRGU
New
Hampshire
Great Gulf Wilderness
Great Gulf. NH; Presidential Range-Dry River, NH
BRIG
New Jersey
Brigantine National
Wildlife Refuge
Brigantine, NJ
SHEN
Virginia
Shenandoah National Park
Shenandoah. VA
JEFF
Virginia
Jefferson/James River
Face Wilderness
James River Face, VA
DOSO
West
Virginia
Dolly Sods /Otter Creek
Wilderness
Dolly Sods, WV; Otter Creek, WV
MACA
Kentucky
Mammoth Cave National
Park
Mammoth Cave, KY
GRSM
Tennessee
Great Smoky Mountains
National Park
Great Smoky Mountains, TN; Joyce Kilmer-Slickrock
NC
528
-------
SHRO
North
Carolina
Shining Rock Wilderness
Shining Rock, NC
COHU
Georgia
Cohutta
Cohutta, GA
LIGO
North
Carolina
Linville Gorge
Linville Gorge, NC
SWAN
North
Carolina
Swanquarter
Swanquarter, NC
ROMA
South
Carolina
Cape Romain National
Wildlife Ref
Cape Romain, SC
OKEF
Georgia
Okefenokee National
Wildlife Refu
Okefenokee, GA; Wolf Island, GA
SAMA
Florida
St. Marks
St Marks. FL
CHAS
Florida
Chassahowitzka National
Wildlife
Chassahowitzka, FL
EVER
Florida
Everglades National Park
Everglades, FL
BRET
Louisiana
Breton
Breton, LA
SIPS
Alabama
Sipsy Wilderness
Sipsey, AL
SENE
Michigan
Seney
Seney, MI
BOWA
Minnesota
Boundary Waters Canoe
Area
Boundary Waters, MN
VOYA
Minnesota
Voyageurs National Park
Voyageurs, MN
ISRO
Minnesota
Isle Royale National Park
Isle Royale, MN
MING
Missouri
Mingo
Mingo, MO
UPBU
Arkansas
Upper Buffalo Wilderness
Upper Buffalo, AR
HEGL
Missouri
Hercules-Glades
Hercules-Glades, MO
CACR
Arkansas
Caney Creek
Caney Creek, AR
WIMO
Oklahoma
Wichita Mountains
Wichita Mountains, OK
BIBE
Texas
Big Bend National Park
Big Bend, TX
GUMO
Texas
Guadalupe Mountains
National Park
Guadalupe Mountains, TX, Carlsbad Caverns, NM
BAND
New
Mexico
Bandelier National
Monument
Bandelier, NM
SAPE
New
Mexico
San Pedro Parks
San Pedro Parks, NM
WHPE
New
Mexico
Wheeler Peak
Wheeler Peak, NM
SACR
New
Mexico
Salt Creek
Salt Creek, NM
WHIT
New
Mexico
White Mountain
White Mountain, NM
BOAP
New
Mexico
Bosque del Apache
Bosque del Apache, NM
CHIR
Arizona
Chiricahua National
Monument
Chiricahua Natl. Monument, AZ; Chiricahua
Wilderness, AZ; Galiuro, AZ
SAGU
Arizona
Saguaro National
Monument
Saguaro, AZ
PEFO
Arizona
Petrified Forest National
Park
Petrified Forest, AZ
GICL
New
Mexico
Gila Wilderness
Gila, NM
BALD
Arizona
Mount Baldy
Mount Baldy, AZ
529
-------
TONT
Arizona
Tonto National Monument
Tonto, AZ
SIAN
Arizona
Sierra Ancha
Sierra Ancha, AZ
IKBA
Arizona
Ike's Backbone
Ike's Backbone, AZ
SYCA
Arizona
Sycamore Canyon
Sycamore Canyon, AZ
HANC
Arizona
Hance Camp at Grand
Canyon NP
Grand Canyon, AZ
BRCA
Utah
Bryce Canyon National
Park
Bryce Canyon, UT
CANY
Utah
Canyonlands National
Park
Canyonlands, UT; Arches, UT
ZION
Utah
Zion
Zion, UT
CAPI
Utah
Capitol Reef
Capitol Reef, UT
GRSA
Colorado
Great Sand Dunes
National Monument
Great Sand Dunes, CO
MEVE
Colorado
Mesa Verde National Park
Mesa Verde, CO
WEMI
Colorado
Weminuche Wilderness
Weminuche, CO; La Garita, CO; Black Canyon of
Gunnison, CO
WHRI
Colorado
White River National
Forest
Maroon Bells, CO; West Elk, CO; Eagles Nest, CO;
Flat Tops, CO
ROMO
Colorado
Rocky Mountain National
Park
Rocky Mountain, CO
MOZI
Colorado
Mount Zirkel Wilderness
Mount Zirkel,CO; Rawah, CO
BADL
South
Dakota
Badlands National Park
Badlands, SD
WICA
South
Dakota
Wind Cave
Wind Cave, SD
THRO
North
Dakota
Theodore Roosevelt
Theodore Roosevelt, ND
LOST
North
Dakota
Lostwood
Lostwood, ND
MELA
Montana
Medicine Lake
Medicine Lake, MT
ULBE
Montana
UL Bend
UL Bend, MT
BRID
Wyoming
Bridger Wilderness
Bridger, WY; Fitzpatrick, WY;
YELL
Wyoming
Yellowstone National
Park
Yellowstone, WY; Grand Teton, WY; Red Rock
Lakes, WY;
NOAB
Wyoming
North Absoraka
North Absoraka, WY; Washakie, WY
JARB
Nevada
Jarbidge Wilderness
Jarbidge, NV
CRMO
Idaho
Craters of the Moon
NM(US DOE)
Craters of the Moon, ID
SAWT
Idaho
Sawtooth National Forest
Sawtooth, ID
SULA
Montana
Sula (Selway Bitteroot
Wilderness)
Anaconda-Pintler, MT; Selway-Bitterroot, MT
GLAC
Montana
Glacier National Park
Glacier, MT
MONT
Montana
Monture
Bob Marshall, MT; Mission Mountains, MT;
Scapegoat, MT
GAMO
Montana
Gates of the Mountains
Gates of the Mountainsm, MT
CABI
Montana
Cabinet Mountains
Cabinet Mountains, MT
STAR
Oregon
Starkey
Eagle Cap, OR; Strawberry Mountain, OR
HECA
Idaho
Hells Canyon
Hells Canyon, ID
530
-------
MORA
Washington
Mount Rainier National
Park
Mount Rainier, WA
WHPA
Washington
White Pass
Goat Rock, WA; Mount Adams, WA
SNPA
Washington
Snoqualamie Pass,
Snoqualamie N.F
Alpine Lakes, WA
NOCA
Washington
Morth Cascades
North Cascades, WA; Clacier Peak, WA
PASA
Washington
Pasayten
Pasayten, WA
OLYM
Washington
Olympic
Olympic, WA
THSI
Oregon
Three Sisters Wilderness
Three Sisters, OR; Mount Jefferson, OR; Mount
Washington, OR
MOHO
Oregon
Mount Hood
Mount Hood, OR
CRLA
Oregon
Crater Lake National Park
Crater Lake, OR; Diamond Peak, OR; Mnt. Lakes,
OR; Gearhart Mnt, OR
LABE
California
Lava Beds
Lava Beds, CA
REDW
California
Redwood National Park
Redwood, CA
KALM
Oregon
Kalmiopsis
Kalmiopsis, OR
LAVO
California
Lassen Volcanic National
Park
Lassen Volcanic, CA
PORE
California
Point Reyes National
Seashore
Point Reyes, CA
PINN
California
Pinnacles National
Monument
Pinnacles, CA
SAGA
California
San Gabriel
San Gabriel, CA
RAFA
California
San Rafael
San Rafael, CA
BLIS
California
Bliss State Park(TRPA)
Bliss, CA
YOSE
California
Yosemite National Park
Yosemite, CA
HOOV
California
Hoover
Hoover, CA
SEQU
California
Sequoia National Park
Sequoia, CA
SAGO
California
San Gorgonio Wilderness
San Gorgonio, CA
AGTI
California
Agua Tibia
Agua Tibia, CA
JOTR
California
Joshua Tree National
Monument
Joshua Tree, CA
DENA
Alaska
Denali National Park
Denali, AK
TUXE
Alaska
Tuxedni
Tuxedni, AK
TRIN
California
Trinity
Trinity, CA
SIME
Alaska
Simeonof
Simeonof, AK
vns
Virgin
Islands
Virgin Islands National
Park
Virgin Islands
HAVO
Hawaii
Hawaii Volcanoes
National Park
Hawaii Volcanoes, HI
HALE
Hawaii
Haleakala National Park
Haleakala, HI
DOLA
California
Dome Lands Wilderness
Dome Land, CA
KAIS
California
Kaiser
Kaiser, CA; Ansel Adams, CA;
In two instances, Class I areas that share monitors are located in different states. One instance is
Carlsbad Caverns and the Guadalupe Mountains, to which Luminant cites; but the other is the
Great Smoky Mountains National Park monitor, which is also shared with the Joyce Kilmer-
Slickrock Wilderness.
531
-------
In those instances in which the Class I areas share a monitor (interstate or intrastate), baseline
visibility conditions (which are based on monitoring data) would be the same. Indeed, for
Carlsbad and Guadalupe, both New Mexico and Texas came out with the same estimates of
visibility impairment. (17.19 dv of impairment on the 20% worst days and 5.9 dv of impairment
on the 20% best days). 77 Fed. Reg., 36,068; 79 Fed. Reg. 74,832. Given that the same
monitor is used for both of these Class I areas, the two will continue to share the data that is used
to measure progress. However, the Regional Haze Rule does not require or assume that the
reasonable progress goals for two Class I areas that share a monitor must be the same during
each planning period. In the end, the improvement in monitoring visibility conditions at two
Class I areas sharing a monitor must be the same, but there is nothing in the regional haze rule
that requires States to reach the same conclusions as to what is a reasonable goal for improving
visibility over the next ten years:434
In developing its submittal, each State will need to conduct analyses to support its
reasonable progress goals according to information available at the time the plan
is submitted about benefits from the existing CAA programs. Each State should
set its goal(s) taking into consideration input from its stakeholders and based on
the statutory factors described above.
In other words, although New Mexico and Texas share a monitor for the Carlsbad Caverns and
Guadalupe Mountains Class I areas, each state has different sizes and types of sources of
visibility impairing pollution within their regulatory control, and these sources have differing
levels of controls already in place. Thus, each state set its goal(s) "taking into consideration
input from its stakeholders and based on [its evaluation of] the statutory factors... " In addition,
it is not irrelevant in considering this comment to note that New Mexico and Texas are in two
different RPOs. The estimates of visibility conditions, as noted above, depends on assumptions
regarding not only in-state emissions inventories (as noted above) but also inventories in nearby
states, as well international emissions. The Regional Haze Rule includes provisions for state-to-
state consultations (as we describe in our proposal), in which states are expected to share
information, including the control strategies that states intend to include in their regional haze
SIPs. Thus, having consulted and mutually developed and coordinated control strategies, each
states' modeling should have incorporated the other state's controls, which would largely align
their respective reasonable progress goals. However, Texas and New Mexico are in two
different regional planning organizations, and each had different schedules for SIP development,
they ran different visibility modeling, and their regional haze SIPs were constructed on the basis
of different provisions of the Regional haze Rule.435 Also, despite the requirement in the
Regional Haze Rule that all state SIPs be submitted by December 17, 2007, all the states
submitted late SIPs, staggered over the next several years. In fact, much of New Mexico's SIP
was constructed prior to 2003.436 As a consequence of these factors, Texas and New Mexico's
434 64 FR 35733.
435 New Mexico's SIP was submitted under Section 51.309, and Texas' SIP was submitted under Section 51.308.
436 In December, 2003, New Mexico submitted its Regional Haze SIP pursuant to the requirements of sections 169A
and 169B of the CAA and the regional haze rule. However, in American Corn Growers Ass'n v. EPA, 291 F.3d 1
(DC Cir. 2002), the U.S. Court of Appeals for the District of Columbia Circuit issued a ruling vacating and
532
-------
opportunities for consultation were less than ideal, and we had no choice but to conduct a
piecemeal, sequential review of their SIPs, which limited our opportunity to effectively address
interstate issues. We address our obligation for uniformity in response to more detailed
comments on that issue.
It is also worth noting that even though EPA approved New Mexico's reasonable progress goal,
these goals did not accurately reflect the final NOx BART determinations for the two largest
sources of NOx within the state - the San Juan Generating Station and the Four Corners Power
Plant (located on tribal lands) 437 As is clear from our decision to approve New Mexico's goals,
the issue in assessing the reasonable progress goals and long-term strategies is to ensure that the
state has provided a reasonable estimate of the progress to be expected but more importantly has
carefully considered available control strategies and made a reasoned decision as to which set of
measures should be adopted. Unlike the NAAQS, New Mexico's failure to achieve the
reasonable progress goals at Carlsbad does not result in negative repercussions for the state. The
monitored progress in visibility at both Carlsbad and Guadalupe will simply be one more factor
to take into account in the next round of regional haze SIPs.
Comment: CCP noted, based on a cost-effectiveness threshold approved by EPA under the
Clean Air Interstate Rule ("CAIR"), Texas used a $2,700 per ton threshold for screening out
unreasonably costly RPG technologies. See Texas SIP 10-7. The use of this cost threshold as a
screen is consistent with EPA-approved regional haze RPG determinations in Kentucky, see 76
Fed. Reg. 78,194, 78,206 (Dec. 16, 2011); Georgia, 77 Fed. Reg. 38,501, 38,508 (June 28,
2012); North Carolina, 77 Fed. Reg. 11,858, 11,870 (Feb. 28, 2012); and North Dakota, 77
Fed. Reg. 34,801 (June 12, 2012), 76 Fed. Reg. 58,630.
CCP stated that we arbitrarily rejected Texas' $2,700 threshold and unreasonably determined
that "a threshold in the range of $4,000/ton to $5,000/ton would be reasonable for purposes of
identifying potential cost-effective controls for further analysis" in Texas' SIP. See 79 Fed.
74,818, 74,838 (Dec. 16, 2014). EPA's rationale for rejecting Texas' threshold was that "CAIR
was not designed as a reasonable progress strategy." Id. CCP noted EPA previously concluded
that CAIR could be used "in lieu of BART." Id. Thus, concludes CCP, it was arbitrary for EPA
to suggest that the same cost threshold cannot be used in the RPG context.
Further, in promulgating its BART rule, CCP noted that the EPA suggested that $1,500 per ton
was a reasonable cost-effective threshold for eliminating additional controls. 70 Fed. Reg.
39,135-36 ("Based on the data before us, the costs of such controls in most cases are less than $
1,500 per ton."). Thus, a $2,700 cost per ton threshold was also reasonable. Various controls
proposed by EPA would have been rejected using the cost-per-ton threshold once site-specific
factors were properly accounted for.
remanding the BART provisions of the regional haze rule. In 2006, EPA issued BART guidelines to address the
court's ruling in that case. See 70 FR 39104 (July 6, 2005). On January 13, 2009, New Mexico resubmitted
portions of its RH SIP, but not the requirements addressing reasonable progress pursuant to Section 51.309(g). New
Mexico later submitted the remaining portion of its SIP covering Section 51.309(g) on July 5, 2011. Texas
submitted its Regional Haze SIP on March 31, 2009.
437 77 FR at 36073.
533
-------
Xcel Energy stated that Texas used a $2,700 per ton threshold for SO2 reductions that is
consistent with other SIPs where EPA has approved regional haze RPG determinations:
• Kentucky SIP Approval, 76 Fed. Reg. at 78,206 (allowing use of $2,000 per ton SO2 as a
screening threshold based on the Clean Air Interstate Rule ("CAIR") for cost
effectiveness);
• North Carolina SIP, 77 Fed. Reg. 11,858, 11,870 (Feb. 28,2012). EPA approved the
State's decision not to implement reasonable progress controls even though cost
effectiveness values were described as ranging, "from 912 to 1,922 dollars per ton of SO2
removed ($/ton SO2), and the average costs per utility system ranged from $1,231 to
$l,375/ton SO2";
• Final North Dakota SIP Approval/Disapproval, 77 Fed. Reg. 20,894, 20,936 (Apr. 6,
2012) (accepting North Dakota's determination that a level of $2,593 per ton of SO2
removed was not reasonable and too costly in the reasonable progress context even
though it is within the range EPA "ha[ s] considered reasonable in the BART context").
• Proposed North Dakota SIP Approval/Disapproval, 76 Fed. Reg. 58,570, 58,630 (Sept.
21, 2011) (finding that North Dakota reasonably rejected cost effectiveness values for
SO2 control options ranging from about $4,000 to $5,000 per ton).
Xcel Energy stated that despite these prior actions, in this Proposal, EPA unreasonably rejected
Texas' $2,700 threshold and arbitrarily determined that "a threshold in the range of $4,000/ton to
$5,000/ton would be reasonable for purposes of identifying potential cost effective controls for
further analysis" in Texas' SIP. See 79 Fed. Reg. at 74,838. This directly contradicts the
precedent established in North Dakota and approved by EPA. EPA has expressly allowed other
states to rely on CSAPR to meet RPGs. See Proposed Michigan SIP Approval, 77 Fed. Reg.
46,912, 46,919 (Aug. 6, 2012) (the regional planning organization's "analysis shows emission
reductions equivalent to the scale of CAIR are needed to meet reasonable progress goals. . . .
EPA believes that with CSAPR providing the reductions that Michigan expects to obtain from
CAIR, Michigan's long-term strategy can in fact be expected to achieve the state-adopted
reasonable progress goals that Michigan established."). The fact that it is impossible to meet the
RPGs in Texas does not make it less legitimate to rely on the cost thresholds used in CAIR for
identifying reasonable controls.
Xcel Energy stated that despite EPA's action in Michigan, in this Proposal, EPA arbitrarily
argues that CAIR could be used "in lieu of BART" but not used in the RPG context. While
BART and RPGs are distinctive components of a regional haze strategy, EPA provides no
reasoned basis for allowing consideration of CAIR in the BART context and rejecting it in the
RPG context. 3 That EPA would allow CAIR or CSAPR to substitute for BART, which is a
unit-specific standard with unit-specific performance criteria, but not for demonstrating
reasonable progress, which is a state-wide, multi-source program aimed at reducing the
pollutants of concern for regional haze, is illogical, as well as arbitrary and capricious. In fact,
EPA has done the exact opposite in other RPG determinations and re-affirmed States' reliance on
534
-------
BART-equivalent analyses. For example, as stated in its proposed approval of the Georgia SIP,
"EPA believes it is reasonable to conclude that any control requirements imposed in the BART
determination also satisfy the RPG related requirements for source review in the first
implementation period." See Proposed Georgia SIP Approval, 77 Fed. Reg. 11,452, 11,469
(Feb. 27, 2012); see also Final Georgia SIP Approval, 77 Fed. Reg. 38,501 (Jun. 28,2012). In
North Dakota, EPA specifically rejected modeling for RPGs that was not conducted in the same
way as BART on the basis that the "ultimate goal is the same." Proposed North Dakota SIP
Approval/Disapproval, 76 Fed. Reg. at 58,629 n. 85; see also Final North Dakota SIP
Approval/Disapproval, 77 Fed. Reg. at 20,906-07.
Xcel Energy noted that prior EPA reviews comparing cost thresholds used in the BART and
RPG context affirm Texas' use of a cost-per-ton threshold that is actually higher than cost-per-
ton thresholds used in the BART context. See Proposed North Dakota SIP
Approval/Disapproval, 76 Fed. Reg. at 58,630 ("The cost effectiveness value for a new wet
scrubber is $2,593 per ton. While this is within the range of cost effectiveness values that North
Dakota, other states, and we have considered reasonable in the BART context, it is not so low
that we are prepared to disapprove the State's conclusion in the reasonable progress context.");
North Dakota v. EPA, 730 F. 3d 750, 765 (8th Cir. 2013) (noting that RPGs do not necessarily
require BART). Indeed, in promulgating its BART rule, EPA suggested that $1,500 per ton was
a reasonable cost effective threshold for eliminating additional controls. BART Rule, 70 Fed.
Reg. at 39,135-36 ("Based on the data before us, the costs of such controls in most cases are less
than$ 1,500 per ton."). Thus, a $2,700 cost per ton threshold in the RPG context also should be
considered reasonable and should not have been rejected by EPA.
In contrast to prior SIP approvals and EPA statements, Xcel Energy argued that the EPA
arbitrarily concluded that Texas' use of the $2,700 per ton threshold was unreasonable and that a
$4,000 to $5,000 threshold (a range it had previously rejected as "relatively high") was
appropriate. See Proposed North Dakota SIP Approval/Disapproval, 76 Fed. Reg. at 58,630
(rejecting use of wet scrubbers for SO2 controls based on cost effectiveness values of$4,735 and
$5,453 per ton SO2). Even when other states have used a higher cost effective threshold, EPA
has rejected expensive cost controls when visibility benefits are "small." See New York SIP
Approval, 77 Fed. Reg. 24,794, 24,818 (Apr. 25, 2012) (approving State's rejection of selective
catalytic reduction ("SCR") to reduce NOx emissions at cost of $5,358 per ton due to visibility
improvement of only 0.254 at one Class I area). EPA offers no explanation as to why such a
relatively high cost threshold is appropriate for Texas sources, particularly since the projected
visibility benefits at the Class I areas addressed in the Proposal are smaller than those addressed
in the North Dakota SIP and many other states where the cost thresholds were lower. This
renders EPA's screening threshold arbitrary and capricious.
[TCEQ/PUCT (0056) p. 11] The TCEQ disagreed with the EPA's conclusion that $2,700 per ton
was too low of a threshold for cost-effective controls. The EPA stated that CAIR was considered
acceptable in lieu of BART but not necessarily designed as a reasonable progress strategy. The
TCEQ selected the $2,700 per ton threshold because it was used in the CAIR analyses to control
NOx and SO2. CAIR was a contemporary program designed for controlling primary and
precursor pollutants for health-based ozone and particulate matter NAAQS. The cost rate was
not selected because CAIR was considered acceptable for BART, but because it met the high
535
-------
standards for a health-based emissions reduction program. And thus, it was considered more
than adequate for the standards of a visibility-based program.
[UARG (0065) p. 19-20] UARG noted that the EPA takes issue with the cost threshold Texas
used to exclude certain controls from consideration. Texas chose $2,700 per ton, while EPA
prefers $4,000 per ton to $5,000 per ton. Id. at 74,838. This preference is not a reason for
disapproving Texas's SIP. Indeed, the only purported justification that EPA can muster - that
reliance on the $2,700-per-ton cost threshold EPA used in CAIR is inappropriate because "[a]
state should look beyond BART for additional reductions when developing its long-term strategy
to achieve reasonable progress at its Class I areas" - conflicts with EPA's own guidance
indicating that BART may reasonably be concluded to satisfy reasonable progress requirements
in the first planning period. Id:, Reasonable Progress Guidance at 4-2 to 4-3. EPA cites - and
can cite - no basis in the CAA or in the regional haze rule for disapproving Texas's cost
threshold.4
[Xcel Energy (0064) p. 6] Xcel Energy stated that the EPA's rejection of the cost threshold that
Texas applied in deciding whether controls should be required, and EPA's use of significantly
higher cost thresholds, has no legal basis, is inconsistent with EPA's own precedent, and is
arbitrary and capricious.
Earthjustice et al., stated that EPA has additional grounds for disapproving the haze plan Texas
submitted. To begin, Texas considered only reasonable progress controls that would cost
$2,700/ton or less. Texas set this screen based on EPA's CAIR rule, but as EPA explains, CAIR
was as an entirely different Clean Air Act program that was not designed to comply with the
regional haze program's reasonable progress requirements. 79 Fed. Reg. at 74,838. Texas did
not adequately explain why this CAIR threshold would be an appropriate or reasonable screening
tool for a reasonable progress analysis under the haze program. As EPA found, Texas's
$2,700/ton screen arbitrarily screened out promising and cost-effective controls. For example,
updated controls at Tolk would result in 20,000 tons per year of sulfur dioxide reductions at a
cost of just $3,100/ton. Id. The screening also was inappropriate because Texas implemented it
prior to consideration of controls necessary to mitigate visibility impacts from the state's sources
given the large geographic distribution and number of impairing sources. Texas's failure to
analyze such controls based on its arbitrary $2,700/ton screen was unreasonable and undermined
the purposes of the haze program.
In addition, Earthjustice et al., stated that Texas relied on CAIR reductions despite evidence that
Texas EGUs would comply with CAIR by purchasing allowances rather than reducing emissions
and that projected CAIR reductions have not been achieved. Texas relied on 2018 CAIR
projections as the baseline for its cost evaluations. However, Texas projected that its EGUs
would comply with CAIR by purchasing 125,000 tons per year of emissions allowances from
out-of-state sources, rather than reducing their emissions to the CAIR allocation levels. Id. at
74,838, 74,840. Moreover, recent emissions data from Texas power plants shows that in many
cases current emission levels are above those projected for 2018, with no plans for upcoming
controls to reduce emissions. See 79 Fed. Reg. 74842-43. Thus, the use of projected CAIR
emissions as a baseline underestimated the likely actual baseline. Using the likely actual
baseline has the effect of lowering the cost effectiveness ($/ton) values relative to Texas's
536
-------
evaluation. Further, if Texas's visibility assessment also relies on the CAIR projections, then the
state is also underestimating the visibility benefit provided because purchasing allowances will
not result in visibility benefits in the locale where projected. Accordingly, it was arbitrary for
Texas to base its reasonable progress analysis on pollution reductions at Texas EGUs that would
likely never occur.
Finally, Earthjustice et al., stated that Texas failed to consider upgrades to existing pollution
controls. Texas's analysis failed to consider upgrades to existing pollution controls as possible
control measures to improve visibility. A state's reasonable progress analysis and BART
analysis are based on similar statutory factors, and EPA's BART Guidelines explain that when
assessing possible control measures states should consider both new controls and upgrades to a
source's existing controls. 40 C.F.R. Pt. 51, App. Y § IV(D)(1)(6) ("[TJhere will often be
control measures or devices already in place. For such emission units, it is important to include
control options that involve improvements to existing controls . . . ."). As EPA documented,
many heavily-polluting Texas EGUs can cost-effectively upgrade their existing scrubbers to
achieve large pollution reductions, and Texas should have considered these measures. For
example, EPA found that scrubber upgrades at Martin Lake's three units could reduce the coal
plant's sulfur dioxide pollution by 21,000 tpy. 79 Fed. Reg. at 74,841.
Commenter's Reference:
4 Moreover, although the point is not directly relevant to this proceeding, EPA's statement in the preamble to the
proposed rule that "the URP does not establish a 'safe harbor' for the state in setting its progress goals," 79 Fed.
Reg. at 74,834, is wrong and should be corrected by EPA. The language in the preamble to the 1999 regional haze
rule that EPA quotes confirms that there is, in fact, such a safe harbor. EPA there made clear that if a state
determines that the progress the URP would require is reasonable, then "the State should identify this amount of
progress as its reasonable progress goal for the first long-term strategy." 64 Fed. Reg. 35,714, 35,732 (July 1, 1999)
(emphasis added). Only //the state instead "determines that additional progress is reasonable" should the state adopt
a more accelerated or ambitious RPG than the URP. Id. As this preamble language makes clear, the state has no
obligation to undertake additional analysis if it determines the URP is reasonable for its Class I area. The decision
whether to undertake such additional analysis is a matter reserved for the state's exercise of discretion.
Response: UARG and others state that we rejected Texas' $2,700/ton threshold because we
preferred a $4,000 to $5,000 threshold. The quote UARG references actually states: "Sensitivity
analysis performed by CENRAP suggests to us that a threshold in the range of $4,000/ ton to
$5,000/ton would be reasonable for purposes of identifying potential cost-effective controls for
further analysis." Here, we were referring to work done by CENRAP: "Sensitivity analysis
performed by CENRAP suggests to us that a threshold in the range of $4,000/ ton to $5,000/ton
would be reasonable for purposes of identifying potential cost-effective controls for further
analysis. We did not adopt this threshold range, as can be seen by examining the range of cost-
effective ness we proposed, which is significantly below that range. In addition, were we to
actually perform an apples-to-apples comparison of the TCEQ's cost-effectiveness threshold to
the upper limit of the Cost-effectiveness we proposed to find reasonable, we would have to
escalate the $2,700/ton figure forward to 2014 when we performed our cost analysis. Doing so
would yield a cost threshold of $3,448/ton 438 This figure exceeds the cost-effectiveness range
438 The CAIR was proposed on 1/30/2004 (69 FR 4566). Conservatively assume the $2,700/ton figure was adopted
in 2014. Cost-effectiveness was performed in 2014, but we conservatively assume 2013. The CEPCI indices for
2004 and 2013 are 444.2 and 567.3, respectively. Consequently, escalating the $2,700/ton figure to 2013 results in a
value of $3,448/ton (567.3/444.2) X $2,700/ton.
537
-------
we proposed to find reasonable as one of the factors we considered. Thus, the cost-effectiveness
range for the controls we proposed is actually lower (adjusted to account for the increased costs
of pollution control equipment) than the $2,700/ton threshold the TCEQ utilized. Furthermore,
in our proposal, we objected to the manner in which the $2,700 threshold was determined and
applied. As we stated in our proposal: "The TCEQ eliminated controls with an estimated cost-
efficiency greater than $2,700/ton from any further analysis, regardless of their potential
visibility benefits. Given the large number of sources and their large geographic distribution,
some consideration of location and emissions data is needed before controls should have been
eliminated from further analysis. The TCEQ supports its selection of this value with reference to
' 'EPA estimated cost of implementing CAIR was up to $2,700/ton." However, although we
demonstrated that CAIR was acceptable in lieu of BART, CAIR was not designed as a
reasonable progress strategy. A state should look beyond BART for additional reductions when
developing its long-term strategy to achieve reasonable progress at its Class I areas. As a result
of the application of this $2,700/ton threshold, potentially cost-effective controls were not
evaluated at sources that may result in meaningful visibility benefits at Guadalupe Mountains or
Big Bend."
Several commenters point to various previous decisions in which we have approved state
decisions to reject controls at various $/ton figures. These commenters do not mention the other
aspect of BART and reasonable progress determinations in these instances—consideration of
visibility benefit. Ignoring extreme ranges of cost (very high or low $/ton), control decisions are
typically made on the basis of comparing the cost effectiveness in $/ton of pollutant removed to
the visibility improvement that would be obtained from the installation of the control. Texas,
however, rejected all controls with a cost effectiveness greater than $2,700 without consideration
to visibility benefit. Thus, Texas rejected a number of controls with significant visibility benefit
because its estimated cost effectiveness for those controls was above $2,700/ton. Texas also
rejected controls with costs below its $2,700 threshold as a result of its analysis technique: Texas
constructed a large potential control set consisting of a mix of large and small sources, located at
various distances from Class I areas, with a large geographical distribution. Because of the
variation in size, type, and location of these sources, the potential to impact visibility and
potential benefit from controls at a given Class I area varied greatly between the identified
sources. This potential control set identified by the TCEQ included controls on some sources
that would likely result in significant visibility benefits, but also included controls on many
sources with much less anticipated visibility benefits. Because it only estimated the visibility
benefit of all the controls together, the TCEQ was not able to assess the potential benefit of
controlling individual sources with significant, and potentially cost-effective, visibility benefits.
In fact, individual benefits were obscured by the inclusion of those controls with little visibility
benefit that only served to increase the total cost figures. For example, the TCEQ identified SO2
controls at Big Brown to be approximately $l,500/ton, significantly less than its $2,700/ton
threshold. Commenters did not mention this aspect of Texas' control evaluation, focusing only
on Texas' $2,700/ton threshold. This results in a false comparison. Nevertheless we evaluate
each of the commenters' allegations below.
CCP stated that based on a cost-effectiveness threshold approved by us under CAIR, Texas used
a $2,700 per ton threshold for screening out unreasonably costly RPG technologies. CCP stated
that the use of this cost threshold is consistent with EPA-approved regional haze RPG
538
-------
determinations in Kentucky, see 76 Fed. Reg. 78,194, 78,206 (Dec. 16, 2011); Georgia, 77
Fed. Reg. 38,501, 38,508 (June 28, 2012); North Carolina, 77 Fed. Reg. 11,858, 11,870 (Feb.
28, 2012); and North Dakota, 77 Fed. Reg. 34,801 (June 12, 2012), 76 Fed. Reg. 58,630.
As explained elsewhere in our action and responses to comments, although a State may rely on
CAIR (or CSAPR) to meet the BART requirements, the State is still required to evaluate whether
additional controls are necessary for reasonable progress. CCP's statement that EPA approved
other regional haze SIPs that adopted this cost threshold is accurate, but EPA in those cases also
considered whether the use of such a threshold affected the reasonableness of the States'
reasonable progress determinations. Regarding Texas' use of a $2,700/ton cost threshold, we
stated: "The TCEQ supports its selection of this value with reference to "EPA estimated cost of
implementing CAIR was up to $2,700/ton." However, although we demonstrated that CAIR was
acceptable in lieu of BART, CAIR was not designed as a reasonable progress strategy. A state
should look beyond BART for additional reductions when developing its long-term strategy to
achieve reasonable progress at its Class I areas.439 As a result of the application of this
$2,700/ton threshold, potentially cost-effective controls were not evaluated at sources that may
result in meaningful visibility benefits at Guadalupe Mountains or Big Bend. For example,
potential SO2 controls for the Tolk Station were estimated in the Alpine Geophysics analysis to
cost an average of approximately $3,100/ton and result in nearly 20,000 tpy reduced across the
two units. Applying the $2,700/ton threshold, the TCEQ did not consider potential controls on
any EGUs in West Texas to improve visibility at the two Class I areas located in West Texas
despite the potential visibility benefits from controlling these large point sources. Sensitivity
analysis performed by CENRAP suggests to us that a threshold in the range of $4,000/ ton to
$5,000/ton would be reasonable for purposes of identifying potential cost-effective controls for
further analysis.440
CCP's allegation that we are being arbitrary in suggesting the same $2,700/ton threshold should
not automatically be used in the RPG context is misapplied (BART is not reasonable progress).
CCP also states that we suggested that $1,500 per ton was a reasonable cost-effective threshold
for eliminating additional controls and cites to our BART Rule (70 Fed. 39,135-36): "Based on
the data before us, the costs of such controls in most cases are less than $ 1,500 per ton." CCP
concludes that a $2,700 cost per ton threshold was also reasonable. The above quote that CCP
reproduces has nothing to do with the selection of a BART cost threshold as can be seen from an
examination of the text immediately before it:
We are establishing presumptive NOx limits in the guidelines that we have
determined are cost-effective for most units for the different categories of units
below, based on our analysis of the expected costs and performance of controls on
BART-eligible units greater than 200 MW. We assumed that coal-fired EGUs
would have space available to install separated over-fire air. Based on the large
number of units of various boiler designs that have installed separated over-fire
439 Per Section 51.308(e)(5), ' 'After a State has met the requirements for BART or implemented emissions trading
program or other alternative measure that achieves more reasonable progress than the installation and operation of
BART, BART-eligible sources will be subject to the requirements of paragraph (d) of this section in the same
manner as other sources."
440 See " Sensitivity Run Specifications for CENRAP Consultation,'' available in the docket for this action.
539
-------
air, we believe this assumption to be reasonable. It is possible, however, that
some EGUs may not have adequate space available. In such cases, other NOx
combustion control technologies could be considered such as Rotating Opposed
Fire Air ("ROFA"). The limits provided were chosen at levels that
approximately 75 percent of the units could achieve with current combustion
control technology. The costs of such controls in most cases range from just over
$100 to $1000 per ton. Based on our analysis, however, we concluded that
approximately 25 percent of the units could not meet these limits with current
combustion control technology. However, our analysis indicates that all but a
very few of these units could meet the presumptive limits using advanced
combustion controls such as rotating opposed fire air ("ROFA"), which has
already been demonstrated on a variety of coal-fired units. Based on the data
before us, the costs of such controls in most cases are less than $1500 per ton.
As can be seen from an examination of the quote in its proper context, the $l,500/ton reference
was merely our estimate (in 2005) of the cost of a particular type of NOx control and has nothing
to do with the selection of a cost threshold—especially one for the type of SO2 controls we
proposed.
As we note in our proposal, in establishing a reasonable progress goal for any mandatory Class I
Federal area within the state, the state must consider the costs of compliance, the time necessary
for compliance, the energy and non-air quality environmental impacts of compliance, and the
remaining useful life of any potentially affected sources, and include a demonstration showing
how these factors were taken into consideration in selecting the goal, among other requirements
in Section 51.308(d)(1). Thus, cost is only one factor that must be evaluated. Also as we discuss
in our proposal, we considered the visibility that would result from the installation of the controls
as well.441 In most cases, cost and visibility benefit are the determinative factors and are
weighed against one another in determining whether controls are warranted.
Regarding the Kentucky proposal cited by CCP, we did state that Kentucky used a $2,000/ton
threshold. However, in our review, we considered other information. For instance, we stated
regarding Kentucky's evaluation of the Century Aluminum facility:442
The Century Aluminum facility in Hawesville, Kentucky, has four potlines with
2002 base year emissions of 4,985 tons per year of SO2 which were identified as
having a significant contribution at the Mammoth Cave Class I area. VISTAS
evaluated control options and costs for sources within the AOI for the Class I
areas of concern. VISTAS used EPA's AirControlNet software to evaluate
control options and costs for controls. The cost effectiveness of SO2 control
suggested by the VISTAS control cost spreadsheet for potlines 1- 4 at Century
Aluminum is $14,207 per ton of SO2 removed. Since the cost of compliance for
the control option is over seven times greater than the Commonwealth's cost-
effectiveness threshold [$2,000] for reasonable progress, KYDAQ concludes that
there are no cost-effective controls available for these Century Aluminum units at
441 79 FR 74874.
442 76 FR 78206.
540
-------
this time within the cost threshold established for this reasonable progress
assessment for the first implementation period.
Although the use of a specific threshold for assessing costs means that a state may
not fully consider available emissions reduction measures above its threshold that
would result in meaningful visibility improvement, EPA believes that the
Kentucky SIP still ensures reasonable progress. In proposing to approve
Kentucky's reasonable progress analysis, EPA is placing great weight on the fact
that there is no indication in the SIP submittal that Kentucky, as a result of using a
specific cost effectiveness threshold, rejected potential reasonable progress
measures that would have had a meaningful impact on visibility in its Class I area.
EPA notes that given the emissions reductions resulting from CAIR, Kentucky's
BART determinations, and the measures in nearby states, the visibility
improvements projected for the affected Class I area are in excess of that needed
to be on the uniform rate of progress glidepath.
Thus, we do not believe that unlike Texas, the use of a cost threshold in Kentucky's SIP resulted
in the state overlooking potential controls.
Regarding CCP's reference to our final action on Georgia, there is no indication that Georgia
used a cost threshold. Georgia undertook reasonable progress analyses for 15 units, and made
decisions based on its source specific evaluations of costs, visibility benefits, and other factors.
See 77 Fed. Reg. 11,464-11,469. It doing so, it found control costs of over $3,500 to be cost-
effective, at least with respect to the visibility benefits from the controls at issue. Id. at 11,465.
Regarding CCP's reference to our North Carolina proposal, we stated443:
NCDAQ believed that it was not equitable to require non- EGUs to bear a greater
economic burden than EGUs for a given control strategy. The facility-by-facility
costs for EGUs under CSA ranged from 912 to 1,922 dollars per ton of SO2
removed ($/ton SO2), and the average costs per utility system ranged from $1,231
to $1,375/ ton SO2. These costs were estimated using the capital costs from the
CSA 2006 compliance plans and the projected operating costs provided by Duke
Energy and Progress Energy. These costs were used as a guide in determining
cost effectiveness.
However, we also made this statement in our final decision:444
Although the use of a specific threshold for assessing costs means that a state may
not fully consider available emissions reduction measures above its threshold that
would result in meaningful visibility improvement, EPA believes that the North
Carolina SIP ensures reasonable progress. In approving North Carolina's
reasonable progress analysis, EPA is placing great weight on the fact that there is
no indication in the SIP revision that North Carolina, as a result of using a specific
443 77 FR 11870.
444 77 FR 38189.
541
-------
cost effectiveness threshold, rejected potential reasonable progress measures that
would have had a meaningful impact on visibility in its Class I areas.
Thus, we do not believe that unlike Texas, the use of a cost threshold in North Carolina's SIP
resulted in the state overlooking potential controls.
With regard to CCP's reference to our North Dakota federal register notice or notices, we believe
that CCP may have made an error in citation. CCP cited to "North Dakota, 77 Fed. Reg. 34,801
(June 12, 2012), 76 Fed. Reg. 58,630." The citation to 77 FR 34801 is a federal register notice
for a final action on the Minnesota regional haze SIP. We performed word searches for "$" and
"threshold," but did not obtain any results. Unfortunately, we therefore do not have enough
information to respond to this part of the comment. CCP's second cite to "76 Fed. Reg.
58,630" is our proposal regarding the North Dakota SIP/FIP. We see no reference to the use of a
cost threshold on the cited page. We also examined a few pages before and after the citations
and were not able to find anything having to do with a cost threshold, and we performed a word
search on "threshold," without success. Unfortunately, we do not have enough information to
address this comment.
Xcel alleges inconsistency with our proposed North Dakota action, stating we approved North
Dakota's rejection of controls ranging from $4,000 to $5,000/ton. We note, however, that all of
the controls we proposed are significantly below these figures. Xcel also points to our final
action on North Dakota, alleging inconsistency in our acceptance that $2,593/ton was not
reasonable. However, Xcel fails to disclose that the $2,593 value its cites was in fact not used as
a cost threshold that would serve to disqualify a control, regardless of its visibility benefit, as
Texas used its $2,700/ton threshold. In North Dakota's case the $2,593 figure was for the
installation of a new scrubber on the Coyote Station when the unit already controlled its SO2
approximately 66%. Consequently the facts are not similar.
Xcel alleges inconsistency with our New York action. This reference, as with other similar
references made by commenters alleging actions in which we rejected visibility results as being
too small to justify the expense of controls, compare CALPUFF modeling in other actions to
CAMx modeling utilized in our analysis. As we explain in our FIP TSD445 and elsewhere in our
response to comments and final action, the results from our CAMx, analysis cannot be compared
to CALPUFF results due to differences in models, model inputs and metrics. As a consequence,
we disagree with Xcel.
We also disagree with UARG that we erred in stating that the URP does not establish a "safe
harbor' for the state in establishing its progress goals. As we note in our proposal:
If the State determines that the amount of progress identified through the [URP]
analysis is reasonable based upon the statutory factors, the State should identify
this amount of progress as its reasonable progress goal for the first long-term
strategy, unless it determines that additional progress beyond this amount is also
reasonable. If the State determines that additional progress is reasonable based on
445 FIP TSD, Appendix A. See discussion beginning on page A-35.
542
-------
the statutory factors, the State should adopt that amount of progress as its goal for
the first long-term strategy [emphasis added].446
UARG's contention that progress beyond the URP is optional ignores the fact, as is made clear
by the above quote from our proposal, that this determination must be based on the statutory
factors. As we discuss in our proposal,447 Texas's reasonable progress four factor analysis was
highly flawed, and thus could not serve as a legitimate basis for rejecting additional progress
beyond the URP. The CAA directs us to act if a state fails to submit a SIP, submits an
incomplete SIP, or submits a SIP that does not meet the statutory requirements. Thus, the CAA
provides us with a critical oversight role in ensuring that SIPs meet the CAA's requirements. In
this instance, portions of the states' SIPs were not approvable for reasons discussed elsewhere
this responses to comments, the final action, and the proposed rulemaking. Concerning the
comment that BART may be sufficient to satisfy reasonable progress for the first planning
period, BART is a part of reasonable progress but a BART determination is not a shield from
additional review under the reasonable progress and long-term strategy provisions of the
Regional Haze Rule.448 Additional responses to this comment can be found in our responses
related to controls that extend beyond CAIR/CSAPR.
We agree with Earthjustice that Texas' use of a $2,700/ton threshold was not justified and
screened out promising reasonable controls. We also agree that we were conservative in our use
of the projected CAIR emissions and that Texas should have considered scrubber upgrades, as
we did.
Comment: Luminant stated that in other recent regional progress SIP reviews, EPA has
employed CALPUFF, not CAMx, to evaluate individual source impacts for reasonable progress
purposes.721 EPA departs from its prior and established practice here and attempts to justify the
use of CAMx instead. But instead of justifying its use of CAMx as an appropriate model for the
application here, which is its legal obligation, EPA simply offers a list of reasons that it believes
CALPUFF is not appropriate. Whatever reasons there may have been not to use CALPUFF to
undertake the analysis EPA conducted, those reasons do not justify the use of CAMx modeling.
And, as explained below, EPA's own prior analysis demonstrates that CAMx suffers from the
same flaws EPA now claims to find in CALPUFF. In other words, EPA ignores the most logical
conclusion—that given the miniscule impacts at issue here and the significant distances
involved, no model is competent to make accurate predictions.
Footnotes:
721 See, e.g., 79 Fed. Reg. at 5,209 (explaining that EPA "did independently run CALPUFF to model the visibility
improvement from potential controls at Dave Johnston Units 1 and 2 and it is part of the basis of our final decision"
to approve Wyoming's decision not to impose additional controls on those units in its reasonable progress SIP).
Response: We address this comment in the modeling section of this document.
446 64 FR 35732 (July 1, 1999).
447 See discussion beginning on 79 FR 74838.
448 70 FR 39143.
543
-------
Comment: Luminant stated that EPA's attempt to use "natural conditions" here is in no way
validated by its prior action on the North Dakota SIP, as EPA wrongly contends.750 There, North
Dakota chose to use a "hybrid cumulative modeling approach" different from the modeling
developed by its regional planning organization ("RPO")751 No other state in North Dakota's
RPO "opted to develop its own reasonable progress modeling methodology," and EPA
concluded that North Dakota's hybrid approach did not satisfy the regulatory "criteria for the use
of alternative models . . . ,"752 The fact that EPA's disapproval of North Dakota's hybrid
approach was affirmed by the Eighth Circuit753 does not justify EPA's use of "clean background"
conditions by which to judge Texas's SIP. Texas did not use a unique or hybrid approach to
assessing reasonable progress. Texas used the standard approach used by all CENRAP states. In
fact, it is EPA that is using an unorthodox and hybrid approach in this instance, mixing modeling
parameters, without demonstrating that its approach meets the criteria for alternative models.
EPA, like North Dakota, is "not free" "to employ its own visibility model and to consider
visibility improvement in its reasonable progress determinations" "in a manner that [is]
inconsistent with the CAA," as EPA has done here.754 EPA must follow its own regulations and
guidance, just as EPA expects state and regulated entities to do.
Footnotes:
750 FIP TSD at A-39.
751 76 Fed. Reg. at 58,624, 58,627.
752 Id. at 58,624.
753 North Dakota, 730 F.3d at 766.
754 Id. (emphasis added).
Response: We address this comment in the modeling section of this document.
Comment: Luminant stated that EPA has repeatedly approved state reasonable progress goals
that take into account international emissions and do not require domestic sources to further
reduce their emissions to compensate for emissions over which the state has no control—
including New Mexico's reasonable progress analysis for the exact same monitor at Guadalupe
Mountains. 657
Footnotes:
657 See, e.g., 77 Fed. Reg. at 70,701 (New Mexico); 77 Fed. Reg. at 30,256 (Idaho); 77 Fed. Reg. 3681, 3687 (Jan.
22, 2012) (Minnesota); 77 Fed. Reg. 76,174, 76,204 (Dec. 26, 2012) (Washington).
Response: We address the impact of international emissions in our responses to other comments
and in our final action. We did not propose nor do we require that Texas over control its sources
to compensate for emissions from Mexico. None of the actions cited in the footnotes to the
comment demonstrate that we have reasoned otherwise. The fact that a particular state's RPGs
include influences from international emissions has no bearing on its evaluation of its own
sources for emission limits and other measures to make reasonable progress. Our technical
record demonstrates the FIP controls would achieve substantial visibility improvement.
International emissions are not a basis for failing to conduct an appropriate determination of
reasonable progress according to the statutory factors. We reviewed each of the notices that the
commenter cites in alleging an inconsistency, but none of them show any inconsistency with
these points.
544
-------
Comment: Luminant discussed the timing of EPA's proposed disapproval of Texas and
Oklahoma SIPS and proposed FIPS. [Luminant (0061) p. 25]
According to Luminant, Texas submitted its regional haze SIP revision to EPA on March 31,
2009.203 However, EPA delayed its review of Texas's submission—well past the 2-year statutory
deadline for EPA to act and after EPA had taken action on every other state regional haze SIP.
204 Indeed, EPA's proposed disapproval of Texas's SIP revision was published over five years
after Texas submitted its plan to EPA and after Texas had already submitted its five-year
progress report to EPA demonstrating substantial progress at the Class I areas. Yet, EPA
proceeds as if its review is timely and that its delay has no implications. To the contrary, EPA's
tardiness changes the landscape considerably—it means that substantial real-world data is
already available that shows more progress towards natural visibility at the Texas and Oklahoma
Class I areas than even EPA asserts is reasonable and necessary for the first planning period; it
means that EPA has taken action on every other states' regional haze SIP and thus there is a
substantial body of EPA precedent applying the regional haze requirements that constrains
EPA's field of play as to Texas; and it means that EPA's remedial FIP authority is further
constrained by the fact that only a few years will remain in the first planning period by the time
EPA takes final action.
Luminant stated that EPA's proposal would disapprove most of the Texas submittal, including
Texas's RPGs for Big Bend and Guadalupe Mountains and Texas's long-term strategy. EPA
would also "simultaneously" disapprove of Texas's and Oklahoma's consultation through
CENRAP and disapprove Oklahoma's RPG for Wichita Mountains. Based on these
disapprovals, EPA claims the authority to issue a FIP that would impose SO2 emission limits,
and the installation or upgrade of SO2 emission controls, at a handful of Texas EGUs. EPA
estimates the cost of its FIP at close to $2 billion, which would be imposed on four companies.
More than half of this cost would be imposed on Luminant.
Footnotes:
203 79 Fed. Reg. at 74,818.
204 Id. at 74,820.
Response: We have addressed the comments over consideration of recent monitoring data,
controls past the first planning period, EPA's FIP authority, and consistency issues elsewhere.
Comment: Luminant discussed EPA's singular and unprecedented approach for this
rulemaking. [Luminant (0061) p. 25]
Luminant stated, given that EPA delayed acting on Texas's regional haze SIP until it had taken
final action on every other state regional haze SIP for the first planning period,205 there is a
substantial body of precedent from EPA's prior actions regarding what is and is not an
approvable plan for the first planning period. Nevertheless, in EPA's own words, its approach to
Texas is "without. . . prior precedent" and does not use the same methodology or metrics that
EPA employed in reviewing every other regional haze SIP.206 And in a sharp departure from
545
-------
prior practice, EPA chose to consider the remaining portions of the Texas SIP and Oklahoma's
RPG for Wichita Mountains in a simultaneous action, in what is an obvious attempt to expand
the agency's authority beyond the bounds of the statute and regulations as currently written.207
Luminant noted that the EPA concedes, as it must, that its proposed approach to the remaining
portions of the Texas and Oklahoma regional haze SIPs is procedurally and substantively
unprecedented in comparison to EPA's previous actions on other states' SIPs. In a court filing in
2014, EPA's Deputy Regional Administrator of Region 6 explained:
Due to the large distances involved and the large number of sources being
analyzed, EPA is utilizing a different model than the standard models used
previously by EPA and States for Regional Haze SIPs. This model is complicated
and the results it generates are not directly comparable to the modeling platform
used in most other States' Regional Haze submittals.208
EPA explained that "[appropriate review of this model therefore requires thorough technical and
policy analysis and interpretation to ensure compliance with the Regional Haze Rule, without the
benefit ofprior precedent to streamline the process."209 In its proposal, EPA further explains
that, not only is its process and methodology for the Texas SIP unprecedented, but so are the
metrics it employs to determine which individual sources it targets for emissions controls.210 As
discussed below, EPA's new methodology relies on many novel and unsupported assumptions,
including that it linearly assigns visibility impairment to individual sources despite the non-linear
nature of the chemical reactions that produce haze conditions.
Footnotes:
205 Id.
206 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'nv. McCarthy, No. 11-01548, at 5 (D.D.C. 2014)
207 See 79 Fed. Reg. at 74,821.
208 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'nv. McCarthy, No. 11-01548, at 5 (D.D.C. 2014)
(emphasis added).
209 Id. (emphasis added).
210 FIP TSD at A-35 ("We have not established specific metrics for use in evaluating single facility impacts on
visibility impairment (RP) as downwind Class I areas with a photochemical grid model such as CAMx....").
Response: We explain in our proposal and final rule our basis for determining that Texas and
Oklahoma were "intricately linked" and why we addressed them together in one action. As
previously discussed, EPA has the authority to issue a FIP either when EPA has made a finding
that the State has failed to timely submit a SIP or where EPA has found a SIP deficient. Here,
EPA has authority on the latter grounds, and we have approved as much of the Texas and
Oklahoma SIPs as possible under the CAA and adopted a FIP only to fill the remaining gap. Our
action today is consistent with the statute.
We disagree with Luminant's use of Sam Coleman's declaration. While EPA did say that there
was no "prior precedent," this was in regards to the particular type of modeling undertaken;
Luminant takes Coleman's statement out of context. Coleman's statement is related to the
additional modeling we determined was appropriate due to the large distances involved and the
large number of sources being analyzed, which was a unique set of facts not encountered by us in
the Regional Haze context before. Luminant conflates this context and over broadens the scope
546
-------
of Coleman's statement beyond the modeling to our approach in the proposal overall.
Coleman's statement was not stating there was no prior precedent for the basis of our proposal,
nor that our approach was inconsistent with other regional haze SIP actions. We have addressed
EPA's modeling methodology and metrics with specificity elsewhere.
Comment: Luminant (0053-2) stated that EPA's proposal treats Texas differently than other
states and reflects a drastic shift in how EPA reviews state regional haze SIPS. The proposal
creates new standards and requirements that EPA has not imposed on other states in review of
their SIPs for the first planning period. Texas's analysis of the four reasonable progress statutory
factors was consistent with the statute and EPA's regulations and their guidance. In guidance to
states, EPA has consistently said until now that reasonable progress is not required to be
demonstrated on a source-by-source basis. Yet EPA's proposal would require Texas to do just
that.
Response: We address these general comments in our responses to more specific comments
elsewhere.
Comment: The requirements in the proposed FIP are untimely for the first regional haze
planning period due to the EPA's delay in acting on the 2009 RH SIP submittal.
[TCEQ/PUCT (0056) p. 4]
The TCEQ stated that the EPA is evaluating the approvability of the Texas 2009 RH SIP, which
covers the first planning period that runs only through 2018. The EPA has been so untimely in
its review of the 2009 RH SIP that only the proposed scrubber upgrades in the proposed FIP
could possibly be in place by the end of 2018. The projected benefit of the other proposed FIP
controls, the scrubber retrofits, is irrelevant to the approvability of Texas' 2009 RH SIP because
they would not be in place during this first planning period.
Response: We have addressed the installation of controls and their applicability towards
reasonable progress goals and the approvability of Texas's RP elsewhere.
Comment: [Texas Governor (0066) p. 2] The Texas Governor stated that the EPA did not have
the up-to-date facts when it decided that the State was not doing enough to regulate visibility
because EPA took almost six years to act on Texas's proposed SIP. And if EPA had bothered to
look, it would have discovered that the haziness conditions in Big Bend and the Guadalupe
Mountains are much better today than Texas projected way back in 2009. And those conditions
will continue to improve even without EPA's costly-but-ineffectual regulations.
Response: We have addressed comments relating to consideration of recent monitoring data
elsewhere. We have addressed costs and visibility benefit elsewhere.
547
-------
Comment: The proposed rule improperly treats Texas and Oklahoma differently from
other states and therefore violates EPA's regulations. [UARG (0065) p. 24-25]
UARG stated that the EPA proposes to disapprove the Texas and Oklahoma SIPs, in part,
because both states relied on CENRAP modeling that EPA believes was incomplete. EPA does
not explain what legally required elements were missing from the CENRAP modeling, and the
Agency assertion that it is appropriate to disapprove the Texas and Oklahoma SIPs on this basis
is incorrect. To the best of UARG's knowledge, EPA has not rejected reliance on modeling
conducted by any other RPO. Indeed, as far as UARG is aware, EPA has not rejected reliance
on CENRAP modeling for other states that also participated in CENRAP during this regional
haze planning period. This is for a good reason: EPA's regional haze rule expressly
countenances reliance on RPO-developed information. EPA's proposal in this rulemaking to
reject reliance on CENRAP modeling and analysis is therefore inconsistent with EPA's past
actions and its regional haze rule.
Response: Our response to the comments on the CENRAP modeling are detailed elsewhere in
this document, particularly in the consultation and modeling sections.
Comment: EPA cannot treat Texas disparately from other states in promulgating a
regional haze FIP [Xcel Energy (0064) p. 24]
Xcel Energy stated that the disparate treatment of Texas under the Proposal runs afoul of EPA's
mandate for fair and consistent treatment of states under the CAA. EPA has issued "Regional
Consistency" regulations to "[a]ssure fair and uniform application by all Regional Offices of the
criteria, procedures, and policies employed in implementing and enforcing the act." 40 C.F.R. §
56.3(a); see National Environmental Development Association's Clean Air Project v. EPA, 752
F.3d 999 (D.C. Cir. 2014) (finding that EPA violated its Regional Consistency rules by
applying a different interpretation of a single stationary source in jurisdictions outside of the
Sixth Circuit). EPA, specifically regional offices such as the Region 6 office that promulgated
the Texas FIP, have an obligation to ensure that their actions "are consistent as reasonably
possible with the activities of other Regional Offices." 40 C.F.R. § 56.5. EPA's Region 6 has
contravened EPA's regional consistency rules by treating Texas differently than other states in
implementing the RHR, and EPA more broadly has failed to correct any regional inconsistencies
in regional haze FIPs. 40 C.F.R. § 56.3(b) (EPA must "[p]rovide mechanisms for identifying
and correcting inconsistencies by standardizing criteria, procedures, and policies being employed
by Regional Office employees in implementing and enforcing the act."). EPA is "not free to
ignore or violate its regulations while they remain in effect." National Environmental
Development Association's Clean Air Project, 752 F.3d, at 1011 (citing US. Lines, Inc. v.
Federal Maritime Commission, 584 F.2d 519, 526 n.20 (D.C . Cir. 1978).
Xcel Energy stated that the EPA's singling out of Texas for different treatment under the RHR
also is in conflict with the "fundamental principle of equal sovereignty" among the States. See
Shelby County, Ala. v. Holder, 570 U.S. _, 133 S. Ct. 2612, 2616 (2013) (holding the Voting
Rights Act preclearance requirement for particular jurisdictions unconstitutional). The U.S.
Supreme Court has held that states must be treated equally under federal law unless there is "a
548
-------
showing that a statute's disparate geographic coverage is sufficiently related to the problem that
it targets." Id. at 2616-17 (citing Northwest Austin Municipal IItil. Dist. No. One v. Holder,
557 U.S. 193, 203 (2009)). EPA has not identified a provision in the Clean Air Act that specifies
a particular regional haze issue in Texas justifying the disparate treatment of Texas in its FIP.
Furthermore, given the "regional" nature of regional haze, it is arbitrary and capricious to single
out a particular State for more stringent treatment.
Response: We address the applicability of the regional consistency rule and the NEDA CAP
case elsewhere. Regarding Xcel Energy's use of the Shelby County case, the federal law at issue
in that case imposed requirements on only certain States, i.e., disparate treatment of States on the
face of the statute. The language Xcel Energy cites is the Court explaining that treatment can be
disparate if the disparate treatment is sufficiently related to the problem being addressed by the
statute at issue. We disagree that this holding is at issue here. First, the CAA Regional Haze
rule provisions do not only apply to certain States, but rather to all States. Second, as we have
discussed in detail elsewhere, our proposal is consistent with the statute, regulations, guidance,
and previous actions in other States. As we explain elsewhere, any potential differences in our
proposal's methodology from other State's is a result of reasonable, scientifically sound
implementation of the statute's requirements in a state as geographically large and source
numerous as Texas, and, most importantly, does not result in Texas being held to a more
stringent standard than other states. Again, and as stated in greater detail elsewhere, what
commenters allege should be national uniformity ignores our authority to exercise judgment,
based on the specific facts at hand in reviewing SIP actions. Here, we are exercising judgment
within the parameters laid out in the CAA and our regulations. Because this is a SIP review
action, we believe that we are not only authorized but required to exercise independent technical
judgment in evaluating the adequacy of the State's regional haze SIP, just as we must exercise
such judgment in evaluating other SIPs.
Comment: [Earthjustice (0067) p.22] Earthjustice et al., noted that, in comments made at public
hearings on the proposed rule, Texas and industry argued that Texas is being treated differently
than other states because EPA is imposing a novel requirement to assess reasonable progress on
a source-by-source basis. But EPA is using the same basic framework for reviewing Texas's SIP
that is established by the Clean Air Act: whether the plan meets all applicable legal requirements.
42 U.S.C. § 7410(c), (k), (1). EPA's statutory responsibility is to review each haze plan to
ensure that it is "reasonably moored" to the Clean Air Act and that the plan is based on a
reasoned analysis. See Oklahoma v. EPA, 723 F.3d at 1206-12; North Dakota v. EPA, 730 F.3d
at 761 (quoting from Alaska Dep't of Envtl. Conservation v. EPA, 540 U.S. 461, 485 (2004)).
This is precisely the approach that EPA has taken in its review of the Texas plan.
Response: We thank Earthjustice for their comment and acknowledge their support regarding
the legality and consistency of our review of Texas' Regional Haze Regional Haze SIP. We
have further addressed our reasoning for disapproval of the Texas Regional Haze SIP elsewhere.
We have addressed source category and source-by-source assessments elsewhere. We have
addressed EPA's review authority and the Oklahoma and North Dakota cases elsewhere.
549
-------
Comment: EPA Invents New Requirements for Texas Sources. [Luminant (0061) p. 2]
Luminant stated that the EPA has stated that its new "individual source" and "visibility benefit"
approach for Texas is "without. . . prior precedent"7 and treats Texas and Oklahoma different
than all other states. EPA's proposal thus violates principles of administrative law and EPA's
own regulations, which require EPA to treat states and regulated entities in a similar and fair
fashion.8
Footnotes:
7 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'n v. McCarthy, No. 11-01548, at 5 (D.D.C. 2014).
8Nat'lEnvtl. Dev. Ass'n's Clean Air Project v. EPA, 752 F.3d 999, 1010-11 (D.C. Cir. 2014) (holding that EPA
action contrary to its regional consistency regulations was contrary to law); see also 40 C.F.R. § 56.5(a)(2)
(requiring that officials in EPA regional offices "shall assure that actions taken under the act. . . [a]re as consistent
as reasonably possible with the activities of other Regional Offices" (emphasis added)).
Response: We address the comments on individual source and visibility benefit, as well as our
regional consistency rule and the NEDA CAP case elsewhere. We disagree with the
characterization of Luminant's comment on the Declaration and that is explained in more detail
elsewhere. Luminant takes Mr. Coleman's statement out of context. Mr. Coleman's statement
is related to the additional modeling we determined was appropriate due to the large distances
involved and the large number of sources being analyzed, which was a unique set of facts not
encountered by us in the Regional Haze context before. Luminant conflates this context and
misapplies the scope of Coleman's statement beyond the modeling to "individual source" and
"visibility benefit."
Comment: EPA's proposal for Texas is fundamentally inconsistent with its actions on other
states' regional haze sips for the first planning period [Luminant (0061) p. 129]
Luminant stated that EPA concedes that its review and action on Texas's regional haze SIP for
the first planning period is "without. . . prior precedent"794 But it is not simply that EPA's
proposal here is the first of its kind—it also arbitrarily treats Texas's proposal in a fundamentally
different and inconsistent way than EPA has treated other states in reviewing their SIPs for the
first planning period.
In the previous comments, Luminant detailed how EPA created out of whole cloth a new and
unprecedented "visibility analysis" targeting a small group of Texas sources and applied
inconsistent and more stringent thresholds and metrics to these Texas sources than it has for
other states. They also explained how EPA, for the first time, is proposing to unwind a regional
agreement between two states about the apportionment of emission reductions to achieve
reasonable progress goals. And we explain how EPA is offering first-time "interpretations" of
its regulations in an attempt to justify these proposed actions. In Section XIV, they further
explained how EPA is proposing for Texas an unlawful FIP that would reach outside the first
planning period (2008-2018) to require installation of emission controls in 2020—the first time,
according to their research, that EPA has ever attempted to assert such far-reaching authority
under the reasonable progress provisions of the regional haze program.
550
-------
Luminant asserted that EPA was not required, and indeed is not authorized, to make this a
regional haze action of "firsts." Texas's regional haze SIP is not unique, nor was the approach
that Texas used to determine its reasonable progress goals and long-term strategy, and thus there
was no reason for EPA to reinvent its review.
Footnotes:
794 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'nv. McCarthy, No. 11-01548, at 5 (D.D.C. 2014).
795 79 Fed. Reg. 26,143, 26,145-46 (May 7, 2014) (emphasis added) (alteration in original) (internal citations
omitted).
796 40 C.F.R. § 56.5(a). These requirements also apply to EPA Headquarters officials "who are responsible for
developing the policies governing the implementation and enforcement of the CAA." Nat'l Envtl. Dev. Ass'n's
Clean Air Project v. EPA, 752 F.3d
999, 1009 (D.C. Cir. 2014).
797 Nat'l Envtl. Dev. Ass'n's Clean Air Project, 752 F.3d at 1009-10 (holding that EPA action contrary to its
regional consistency regulations was contrary to law); see also 40 C.F.R. § 56.5(a)(2) (officials in EPA regional
offices "shall assure that actions taken under the act. . . [a]re as consistent as reasonably possible with the activities
of other Regional Offices" (emphasis added)).
Response: As stated in our responses elsewhere, we disagree with Luminant's interpretation of
Sam Coleman's declaration. While EPA did say that there was no "prior precedent," this was in
regards to the particular type of modeling undertaken. Thus, Luminant takes Mr. Coleman's
statement out of context. Mr. Coleman's statement is related to the additional modeling we
determined was appropriate due to the large distances involved and the large number of sources
being analyzed, which was a unique set of facts not encountered by us in the Regional Haze context
before. Luminant conflates this context and over broadens the scope of Coleman's statement
beyond the modeling to our proposal overall. Coleman's statement was not stating there was no
prior precedent for our proposal overall, nor was it stating there was no prior precedent for the list
of specific components of the proposal that Luminant states are "firsts."
With regard to comments relating to consultation, controls beyond the first planning period,
visibility/modeling and our clarified interpretation, those comments have all been addressed
elsewhere. We also address our regional consistency rule and the NEDCAP case elsewhere.
Comment: According to the Associations, it is a well-established tenet of administrative law that
"[rjeasoned decision making ... necessarily requires the agency to acknowledge and provide an
adequate explanation for its departure from established precedent." Dillmon v. NTSB, 588 F.3d
1085, 1089-90 (D.C. Cir. 2009) (citingFCC vFox Television Stations, Inc., 129 S. Ct. 1800,
1811 (2009)). Indeed, given that the submitting States relied upon EPA's established guidance
and precedents in crafting their SIPs, EPA is required to provide a "more substantial
justification" for its disapprovals. Perez v. Mortgage Bankers Asssupra, slip op. at 13.
Response: We acknowledge the citations provided by the comment, but the comment begins
with the faulty premise that the disapproval is based on a departure from established precedent.
This is not correct. As explained in the section of this document addressing comments on
consistency, this is not a case where we are departing from established precedent. All EPA
actions to review SIPs presuppose the ability of EPA to disapprove those SIPs that lack a
reasoned basis and do not accord with the requirements of the Clean Air Act. We find that
551
-------
portions of the SIP must be disapproved for not meeting CAA requirements, and the bases for
disapproval are explained at length in our responses and final action.
Comment: EPA improperly rejected Texas' decision not to impose further controls in this
planning period due to low deciview visibility improvement when it has repeatedly not
required further controls in other SIPs. [Xcel Energy (0064) p. 11-13]
Xcel Energy stated that the EPA guidance on implementing RPGs specifically allows states to
consider relative impacts in visibility when setting RPGs to ensure that the measures aimed at
achieving the uniform rate of progress "are reasonable." See RPG Guidance, at 2-3. EPA's RPG
Guidance notes that states should look at "other available measures for the sources and source
categories that contribute significantly to visibility impairment." Id. It was never Congress'
intent under the CAA to regulate every possible source of visibility impairment in a Class I area.
Rather, the intent is only to regulate sources and source categories that significantly impair
visibility and achieve visibility improvement over a long time horizon.
Xcel Energy noted that the EPA concludes that the "visibility benefits of SDA scrubbers on the
Tolk units are projected to occur mainly at the Guadalupe Mountains." 79 Fed. Reg. at 74,882.
EPA estimated that the visibility improvement in 2018 background levels from installing
scrubbers at Tolk 1 and 2 (Units 171 b and 172b) would be 0.022 and 0.024 deciviews (dv),
respectively. 4 Id, at 74,882, Table 36. Even EPA acknowledges that this level of deciview
improvement is "smaller than" deciview improvements from installing scrubbers at emissions
units at the W. A. Parish or Welsh plants. Id. at 74,882; see id. At 74,881, Table 34 (identifying
deciview improvements at the Wichita Mountains ranging between 0.020- 0.025 dv from the
installation of wet scrubbers at Parish Units 5-7 and deciview improvement between 0.022-0
.023 dv from the installation of wet scrubbers at Welsh Units 1-3). Nonetheless, EPA
inconsistently proposes to control Tolk and do nothing at these other units, stating that "the
visibility benefits from installing scrubbers on the W. A. Parish 5, 6, and 7 units; and Welsh 1,
2, and 3 units would not yield large enough visibility benefits to be considered at this time." Id.
at 74,882; see also New York SIP Approval, 77 Fed. Reg. at 24,818 (rejecting controls based on
visibility improvement of 0.254 dv, an amount 10 times greater than the visibility improvement
that would result from EPA's proposed SO2 controls at either Tolk unit).
Xcel Energy noted that the EPA has previously rejected additional controls to achieve reasonable
progress even when visibility improvement was magnitudes greater than EPA expects from the
installation of scrubbers at the Tolk units:
• In Arkansas, EPA concluded that "a visibility improvement of only 0.2 dv" was too low
to apply further emission reductions even when Wichita Mountains was not on the URP
glidepath. Arkansas SIP Approval, 77 Fed. Reg. 14,604, 14,625 (March 12, 2012)
(emphasis added).
• In Arizona, EPA projected benefits of SCR at one project to be 0.41 dv at the most
affected Class I area but still rejected SCR for purposes of reasonable progress. This
level of improvement is more than nine times greater than the visibility improvement that
552
-------
SDA scrubbers on Tolk would accomplish, at best, at the Guadalupe Mountains. Arizona
FIP, 79 Fed. Reg. 9,318, 9,360 (Feb. 18, 2014).
• In Montana, EPA found a 0.18 dv improvement to be a "low visibility improvement" that
"did not justify proposing additional controls" for SO2 on one source. This level of
improvement is more than four times greater than the visibility improvement that SDA
scrubbers on Tolk would accomplish, at best, at the Guadalupe Mountains. Montana FIP,
77 Fed. Reg. 23,988, 24,012 (Apr. 20,2012).
• In Oregon, even with relatively low costs per ton ($1 ,816/ton of NOx), minor visibility
improvements were rejected because "adding SNCR only provided an additional 0.18 dv
of visibility improvement over NLNB/MOF A at the Mt. Hood Wilderness Area." This
level of improvement is more than four times greater than the visibility of improvement
that SDA scrubbers on Tolk would accomplish, at best, at the Guadalupe Mountains.
Proposed Oregon SIP, 76 Fed. Reg. 12,651, 12,661 (March 8, 2011).
Xcel Energy stated that although Texas considered visibility improvements, it concluded there
were insignificant cumulative visibility benefits, measured in deciviews, from requiring
additional controls. See 79 Fed. Reg. at 74,837, Table 10 (identifying estimated deciview
improvements ranging from 0.16 dv in Big Bend to 0.36 dv in Wichita Mountains). EPA
previously rejected similar cumulative visibility improvements of 0.254 dv and 0.273 dv in the
New York SIP as "small." New York SIP Approval, 77 Fed. Reg. at 24,818. Nevertheless, EPA
concluded for Texas that individual sources with de minimis deciview contributions and
similarly small deciview cumulative benefits warrant the substantial costs of additional controls
based on EPA's estimate of "extinction benefits and percentage of total extinction." 79 Fed. Reg.
at 74,882. Based on the foregoing, it was manifestly unreasonable and capricious for EPA to
reject both Texas' calculations of low visibility benefit and EPA's own calculations of low
deciview improvement and, as explained below, turn to a novel and unprecedented method for
justifying scrubbers at Tolk.
Commenter's References:
4 EPA's discussion of why CALPUFF modeling, especially at the large distances involved in Texas, would predict
higher deciview impacts than CAMx is beside the point. The greater complexity and consideration of the interaction
of all relevant sources offered by CAMx, according to EPA, makes it a better model for assessing regional haze for
the long distances between the sources and the Class I areas addressed in the Proposal. At the same time, however,
both models provide results in the same parameter: deciviews. And the CAMx model shows a minuscule deciview
impact by Tolk on the Guadalupe Mountains National Park.
Response: While we disagree with Xcel's characterization of our guidance and attribution of
Congressional intent from it, our proposed FIP followed the statute, Congressional intent, our
regulations, and our guidance cited by Xcel—as we have explained in greater detail elsewhere—
and identified those sources with the largest visibility impacts at Wichita Mountains and the
Texas Class I areas based on consideration of a source's emissions, location, and modeled
visibility impairment. As discussed in more detail in our proposal and in the modeling section of
our responses Texas determined, and we agreed, that it was reasonable to focus the analysis on
point sources of SO2 and NOx.449 This was based on review of emissions and source
449 79 FR 74838
553
-------
apportionment results indicating that these sources were most responsible for anthropogenic
contributions to visibility impairment. We then used a Q/d analysis to identify those sources
with the greatest potential to impact visibility based on emissions and distance. Additional
analysis using photochemical grid modeling was then completed to estimate the visibility impact
from those sources. Based on consideration of facility level and estimated contributions to
visibility from units at the modeled facilities, we identified those sources that had the largest
visibility impacts to analyze for additional controls. Once identified, we performed the four-
factor reasonable progress analysis on these sources to determine if reasonable controls were
available and cost-effective for the first implementation period, which included consideration of
visibility benefit.
We address comments concerning the visibility benefits of controlling the Tolk units, our
determination that those controls are reasonable, and comparing the visibility benefits of controls
on Tolk units at Guadalupe Mountains to visibility benefit of controls on units at Parish and
Welsh at Wichita Mountains in separate responses to comments in the modeling section of this
document. We note that, as discussed in depth elsewhere in this document where we address
comments concerning the visibility metrics we used, we considered visibility benefits in terms of
extinction, percentage of total extinction, and deciview improvement (based on both "dirty" and
"clean" background). The commenter only provides a comparison based on deciview
improvements for controls on Tolk units based on a "dirty" background approach. We evaluated
the visibility benefits of controlling the two units at Tolk for Guadalupe Mountains and
determined that based on evaluation of extinction and percentage of extinction that these controls
would provide for meaningful progress towards the goal of reaching natural visibility conditions
for this progress period at Guadalupe Mountains. From our initial modeling based on the 2018
CENRAP emissions, we estimated that controls on the two units at Tolk would address
approximately 8% of the total visibility impairment from all Texas point sources at Guadalupe
Mountains.
We address comments concerning Texas' approach in considering visibility benefits of controls
in a separate response to comment where we address comments concerning consideration of
costs versus visibility.
We address Xcel's allegations that we have been inconsistent in comparison to our actions, in the
consistency section of this document.
15. Modeling
15.a. Modeling Protocol
General summary: We received comments that EPA should have prepared a modeling protocol
and made it available for public/stakeholder review and comment. The commenters state that a
modeling protocol is required by EPA modeling guidance.
Comment: EPA should have prepared a modeling protocol and made it available for public
review and comment. [NRG (0078) p. 5]
554
-------
NRG stated that the EPA did not provide a modeling protocol for public notice and comment as
part of its proposal. A modeling protocol is the means to detailing and formalizing the
procedures for conducting a modeling study. In this context, a protocol would help to ensure that
sources are identified consistently for potential regulation.
NRG stated that the EPA would certainly have required a state issuing such a plan to provide a
modeling protocol. Notably, EPA's regional haze modeling guidance recommends that states
prepare such a modeling protocol in submitting a regional haze plan to EPA. Alpine Report at
31-33. Thus, to comply with this guidance, EPA would need to make available its own modeling
protocol for public review. Similarly, EPA's BART rules require the use of a modeling protocol:
"There are several steps for making an individual source attribution using a dispersion model: 1.
Develop a modeling protocol." [40 CFR 51, Appx. Y, § III.A.3, Option 1.] And, "[I]f you wish to
use a grid model, you should consult with the appropriate EPA Regional Office to develop an
appropriate modeling protocol." [40 CFR 51, Appx. Y, § III.A.3, Option 3. j
In this instance, NRG noted that development of a modeling protocol for this action would have
facilitated stakeholder participation in the evaluation of potential sources of visibility impairment
and reduced the risk of EPA relying on outdated emissions information.
NRG concluded that it is inappropriate for EPA to fail to provide a modeling protocol where
EPA's own regulations and practice would require an Implementing state to use a modeling
protocol. In this instance, EPA's method to identify sources to regulate appears to lack such a
consistent basis.
Response: EPA is not required to develop a modeling protocol for its reasonable progress/long-
term strategy technical analysis and take public comment on it. Our BART guidelines in the
regional haze rule450, our guidance 451 and 40 CFR Part 51 Appendix W do not require us to
develop a modeling protocol for our own technical work conducted to support review of a SIP
submittal or rulemaking e.g., promulgating a FIP. The recommendation to do modeling
protocols is to help ensure that state analyses will be developed and completed in accordance
with EPA guidance and policy to help states and other stakeholders develop adequate submittals.
We reviewed the Texas SIP submittal that contained its technical analysis to determine whether
it was technically adequate and after our review thereof, we made the determination it was
technically flawed. This required us to perform our own technical analysis so we developed a
workplan and consulted with national experts at EPA HQ (Office of Air Quality Planning and
Standards - OAQPS) that are responsible for developing modeling in support of national
rulemaking, setting policy and modeling/technical guidance for RH analyses. We consulted with
these experts a number of times as we refined our technical approach and developed a workplan.
The docket for the proposal included the modeling files, documentation of how the information
was developed, and included the TSD documents, model output and analysis files, etc. We took
comment on all aspects of the above analyses and techniques.
450 See 77 FR 33642 (June 7, 2012).
451 See EPA, 2007 Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air Quality
Goals for Ozone, PM2.5, and Regional Haze. EPA-454/B-07-002; EPA-R06-OAR-2014-0754-0010;
http://www.epa.gov/ttn/scram/guidance/guide/Draft_O3-PM-RH_Modeling_Guidance-2014.pdf).
555
-------
Comment: EPA's Failure to Follow Its Own Modeling Guidance [Alpine (0078) p. 5, 31]
Alpine stated that it is clear that EPA's action to partially disapprove Texas' SIP is not
supportable based on the data and methods used to back their decision.
Alpine stated that the EPA failed to follow its own regional haze modeling guidance because:
• EPA's approach to evaluating visibility impacts lacked an objective, documented
significance test; and
• EPA failed to publish a supporting modeling protocol.
Alpine stated that the EPA's current draft modeling guidance document,12 as well as previous
final versions of this guidance and current BART modeling protocols are designed to inform air
quality planning by providing documentation that adequately supports and describes the
procedures used in an analysis. Additionally, it is recommended by the Agency that a modeling
protocol be prepared (to facilitate Regional Office review and approval of a modeling analysis)
before the modeling is conducted.
Alpine stated that these protocol documents are designed to detail and formalize the procedures
for conducting any modeling study and effectively communicating to the Agency and impacted
stakeholders, what the blueprint is for the study at hand. An appropriate protocol document
would lead to extensive participation by stakeholders in developing the demonstration and
reduce the possibility, in advance of any analysis, of the modeling entity to have used incorrect,
invalid, or outdated information.
Alpine stated that the EPA did not publish a modeling protocol. The EPA guidance for BART
modeling.13 CENRAP's BART modeling guidance14 and the EPA guidance for the use of
photochemical grid model for regional haze15 all require a modeling protocol be prepared and
submitted for public comment. The modeling protocol specifies how the modeling will be
conducted, how the modeling results will be analyzed and how the modeling results will be used
to inform decisions before any modeling actually occurs.
Alpine noted that the EPA states in the 2007 "Guidance on the Use of Models and Other
Analyses for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5 and Regional
Haze":
Developing and implementing a modeling/analysis protocol is a very important part of an
acceptable modeled attainment demonstration. The protocol should detail and formalize the
procedures for conducting all phases of the modeling study, such as describing the background
and objectives for the study, creating a schedule and organizational structure for the study,
developing the input data, conducting model performance evaluations, interpreting modeling
results, describing procedures for using the model to demonstrate whether proposed strategies
are sufficient to attain the NAAQS and/or regional haze goals, and producing documentation to
be submitted for EPA Regional Office review and approval.
556
-------
Furthermore, Alpine noted that the EPA states in the December 2014 "Draft Modeling Guidance
for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5, and Regional Haze"16:
As with any technical support document designed to inform air quality planning, an attainment
demonstration should be supported by documentation that sufficiently describes the procedures
used in the analysis. In order to facilitate the process of EPA Regional Office review and
approval, we recommend the preparation of two separate supporting documents: one before the
modeling analyses are initiated (modeling protocol) and one after the analyses have been
completed (attainment demonstration package).
Alpine noted that the EPA BART modeling guidance states:
For regional haze applications, regional scale modeling typically involves use of a
photochemical grid model that is capable of simulating aerosol chemistry, transport, and
deposition of airborne pollutants, including particulate matter and ozone Regional scale air
quality models are generally applied for geographic scales ranging from a multistate to the
continental scale. Because of the design and intended applications of grid models, they may not
be appropriate for BART assessments, so States should consult with the appropriate EPA
Regional Office prior to carrying out any such modeling.
According to Alpine, for this modeling the EPA chose to use regional scale photochemical
modeling. Although there is no technical objection to the use of the regional scale photochemical
modeling, the CENRAP BART modeling guidance states:
EPA's BART guidance clearly indicates the need for a detailed modeling protocol to support any
application of alternative models for BART analyses. An example of the content of such a one
atmosphere modeling protocol would be the CMAQ/CAMx modeling protocols developed for
CENRAP (Morris et al., 2004c) and VISTAS (Morris et al, 2004a). In addition, certain
components of the screening and source-specific protocols developed with CALPUFF (Tables 6-
1 and 7-2) would be appropriate. The alternative modeling protocol should be submitted to the
state, regional EPA office and FLM for review and negotiation. Note that EPA's role in the
development of the protocol is only advisory as the "states better understand the BART-eligible"
source configurations" and factors affecting their particular Class I areas (70 FR Part 128 pp
39126).
Alpine stated that while a modeling protocol was prepared in 2004 to support the development of
the 2002 to 2018 modeling platform,17 this project has made significant alterations to the
emissions inventory used in the development of this modeling platform and the state of science
in photochemical model has advanced in the decade since this modeling platform.18 Furthermore,
this existing protocol does not cover how the modeling results will be used to assess if additional
controls are required on specific units.
Alpine stated that since this modeling was conducted by EPA, EPA's modeling protocol should
have been discussed with other outside groups, especially impacted stakeholders in the process
before moving forward with the analysis. The most transparent way to accomplish this would
have been to prepare and share a modeling protocol in order to advise the public and solicit input
557
-------
on the actions that were to be taken with the modeling and analysis. In this case, consistent with
EPA's own guidance on the matter, an updated modeling protocol should have been prepared
and presented for public and other government agency comment.
Footnotes:
12 http://www.epa.gov/ttn/scram/guidance/guide/Draft_03-PM-RH_Modeling_Guidance-
2014.pdf
13 70 CFR Part 128 pp 39125
14 Alpine Geophysics, 2005. CENRAP BART Modeling Guidelines. Prepared by Alpine
Geophysics, LLC, Prepared for Central Region Air Planning Assoc. 15 December 2005.
15 EPA, 2007. Guidance on the Use of Models and Other Analyses for Demonstrating Attainment
of Air Quality Goals for Ozone, PM2.5, and Regional Haze. EPA-454/B-07-002.
16 http://www.epa.gOv/ttn/scram/guidance/guide/Draft_03-PM-RH_Modeling_Guidance-
2014.pdf
17 http://pah.cert.ucr.edu/aqm/cenrap/docs/CENRAP_Draft2.0_Modeling_Protocol_120804.pdf
18 EPA-R06-OAR-2014-0754-0010
Response: As discussed in the previous response, we were not required to develop a modeling
protocol and take public comment on it prior to publishing our proposed Federal Register action
that was subject to public review and comment. BART guidelines, our modeling guidance
documents, and 40 CFR Part 51 Appendix W do not require us to develop a modeling protocol
for our technical work conducted to support our review of SIPs or in rulemaking actions. EPA's
role in this action was to review the Texas and Oklahoma Regional Haze SIPs and either approve
or disapprove and provide FIP measures as needed, and provide all supporting information for
our decisions. EPA has provided the opportunity for comment on all aspects of the modeling that
we conducted as required for notice and comment rulemaking. We also respond to all significant
comments either raised at our public hearings or submitted as written comments. As to the
comment that EPA's BART guidance clearly indicates the need for utilizing an alternative model
for BART analysis, we concur in general that this is a requirement for states and other BART
analysis developers. We agree that Texas did submit a detailed modeling protocol when it
proposed to utilize the CAMx model with CENRAP databases for screening of many sources for
BART applicability. EPA was not conducting a BART modeling exercise but a reasonable
progress technical analysis, but we did use the CAMx model and CENRAP databases with the
same technical options of Plume-in-Grid and Particulate Source Apportionment Tool (PSAT) as
Texas had previously done for their BART screening. We maintain that EPA does not have to
develop a modeling protocol for the reasonable progress technical analysis, and was not required
to develop a modeling protocol prior to our proposal in this action. In sum, we disagree with
commenters as none of the citations provided by the commenter required EPA to develop a
modeling protocol prior to publishing our proposed action.
EPA disagrees that we did not document the significance test and metrics that we used and that
we did not describe our objective approach to identifying the largest impacting sources and then
determining if controls were feasible, cost effective and resulted in improvements to visibility
impacts. All of this information was included in the record with our proposal materials and
detailed in our TSDs that were available in the docket for public review at that time. These
issues are responded to in more detail elsewhere in our responses that address the various
558
-------
aspects of our technical approach and decision making process. We did our own review and
analysis and documented our methodology and decision making process in our proposed action
and our TSD materials. We disagree that we made significant alterations to the emission
inventory, and while we incorporated some advancements in the science, none of these would
create any deviations from the Texas technical analysis that impacts the results. This issue is
addressed in more detail in specific responses to the changes in model and emission inventory in
later responses.
15.b Model Selection/Model Performance/Model Uncertainty/Model Inputs
General summary: We received comments that our selection of the CAMx model rather than
CALPUFF is inappropriate and unjustified. The commenters state that EPA has not justified the
use of CAMx to model visibility impacts from individual sources and at large distances, and
EPA's use of CAMx here is outside of the model's capabilities. Furthermore, the commenters
assert EPA's concerns of using CALPUFF are not clear and concerns of model overprediction
are also present in CAMX and therefore do not justify the use of CAMx. The commenters also
state that EPA failed to consider and discuss bias and uncertainty in model results and instead
relied on the model predictions as definitive results.
We received comments that EPA failed to perform a full model performance evaluation and
instead compared model results to the CENRAP modeling results despite deviations from
CENRAP's modeling protocol. EPA also failed to update the modeled emission inventories or
consider more recent emissions data, such as the 2011 NEI and EPA's recent 2018 emission
inventory showing large reductions from MATs. Recent monitor data indicates that EPA's
modeling is not representative of anticipated future conditions and was not considered during
model performance evaluation.
Comment: [Luminant (0061) p. 115] Luminant stated that EPA's first error is its choice of
CAMx as a modeling platform, despite the fact that there is no precedent for applying CAMx in
this way.719 As EPA itself has explained, "CALPUFF is the best modeling application available
for predicting a single source's contribution to visibility impairment. It is the only EPA-approved
model for use in estimating single source pollutant concentrations resulting from the long range
transport of primary pollutants."720 Moreover, in other recent regional progress SIP reviews,
EPA has employed CALPUFF, not CAMx, to evaluate individual source impacts for reasonable
progress purposes.721 EPA departs from its prior and established practice here and attempts to
justify the use of CAMx instead. But instead of justifying its use of CAMx as an appropriate
model for the application here, which is its legal obligation, EPA simply offers a list of reasons
that it believes CALPUFF is not appropriate. Whatever reasons there may have been not to use
CALPUFF to undertake the analysis EPA conducted, those reasons do not justify the use of
CAMx modeling. And, as explained below, EPA's own prior analysis demonstrates that CAMx
suffers from the same flaws EPA now claims to find in CALPUFF. In other words, EPA ignores
the most logical conclusion—that given the miniscule impacts at issue here and the significant
distances involved, no model is competent to make accurate predictions.
Footnotes:
559
-------
719 79 Fed. Reg. at 74,877-78 (explaining why EPA "chose to use the CAMx photochemical
model instead of CALPUFF").
720 See 70 Fed. Reg. at 39,122.
721 See, e.g., 79 Fed. Reg. at 5,209 (explaining that EPA "did independently run CALPUFF to
model the visibility improvement from potential controls at Dave Johnston Units 1 and 2 and it is
part of the basis of our final decision" to approve Wyoming's decision not to impose additional
controls on those units in its reasonable progress SIP).
Response: See our next response.
Comment: [Luminant (0061) p. 116] Luminant noted that EPA claims one reason it chose
CAMx over CALPUFF was the "large distances between sources and Class I areas."722 But
EPA's modeling guidelines, which specify the use of CALPUFF for visibility modeling of "a
small group of sources," as EPA has done here, does not recommend the use of CAMx for
visibility modeling 723 Nor does EPA provide any support for its apparent view that the distances
involved here are within CAMx's reliable range. And EPA cites no instance in which it has
successfully modeled visibility benefits from individual controls at the distances involved here
with CAMx. Indeed, EPA's deviations from CENRAP standard protocols—such as adjusting the
grid and limiting the facilities analyzed—suggest that EPA acknowledges that its approach here
is unique and outside the model's designed capabilities.724
Footnotes:
722 79 Fed. Reg. at 74,878.
723 40 C.F.R. pt. 51, app. W, § 6.2.1.
724 In fact, the CAMx User's Guide states: "OS AT/PS AT provides ozone/PM attribution to
source regions and categories for a given emissions matrix, but does not provide quantitative
information as to how ozone/PM contributions would change as emissions are altered because
chemical interactions are non-linear." Environ, CAMx User's Guide Version 6.1 at 4 (Apr.
2014).
Response: We further address the comment in reference to App. W recommended models in
response to comments on that specific issue below. While we agree that we did include in our
proposal and Modeling TSD a number of reasons in support of our selection of the
photochemical grid model CAMx over CALPUFF, we also included a number of materials in the
docket to this action that are in general public circulation that document concerns with
CALPUFF.452 One of the primary reasons for our use of CAMx over CALPUFF here is that we
evaluated the Texas SIP for reasonable progress and not BART, and the differences in the
purposes of these analyses supports the use of different models when the resources are available
to utilize a photochemical model. Reasonable progress requires the evaluation of changes in
452 The promulgation of CALPUFF in 2003 included distance limits of a few hundred kilometers (2003 App. W, FR
Vol. 68, No. 72, April 15, 2003), BART guidance indicated 50 km to a few hundred km as reasonable distances for
CALPUFF (2005 BART Guidelines at 70 FR 39104, .39122 ), EPA's 2003 App. W only gave full promulgation for
inert species estimates and did not approve CALPUFF for full chemistry calculations, the BART Guidelines in 2005
allowed for CALPUFF modeling using chemistry to only be used to estimate as a screening tool and part of a 5-
factor BART analysis, and IWAQM Phase 2 report also raised concerns with the effective distance limits of
CALPUFF.
560
-------
emissions from one or more facilities on visibility impairment at downwind Class I areas. In
order to properly account for chemical transformations of those emissions, the model used must
also include the other pollutants in the airshed, for which CALPUFF is not as well suited.
Reasonable progress analyses look at the changes in visibility on the 20% worst days, and this
evaluation was done by most states, including Texas and Oklahoma, by utilizing a
photochemical grid model (PGM) such as CAMx or CMAQ and not CALPUFF. Texas and
Oklahoma relied upon CENRAP's control scenario CAMx modeling that included controls on
some sources in each state that were identified as potentially having cost effective controls
available as possible RP sources. In particular, Texas used the CENRAP modeling as a starting
point for its reasonable progress/long-term strategy technical analysis. As discussed in our TSD,
when we evaluated the technical analysis that Texas submitted, and that was based on
CENRAP's CAMx modeling results, we identified that TCEQ's analysis was flawed, and we
then evaluated how to best perform an evaluation to determine if there were potential reasonable
controls that TCEQ should have evaluated. It would have been inappropriate for us to use
CALPUFF modeling techniques to compare with TCEQ's CAMx based technical analysis;
therefore, we also used CAMx. CENRAP used both CMAQ and CAMx in its Regional Haze
modeling and included emissions from all regional haze emission sources in the modeling
domain to the best that they could be quantified.453 As the BART guidelines indicated, PGMs
(CAMx) is a more intensive tool that also requires additional time and resources to conduct a
proper evaluation compared to CALPUFF, so in many cases CALPUFF may be the only tool
available that EPA and/or states have used in some recent RP analyses. It is also case-by-case
depending on the distance to Class I area and the amount of refinement needed in visibility
modeling and not all situations are equivalent. Considering that Texas had used as its base in its
analysis the CENRAP CAMx modeling and the distances were outside CALPUFF's normal
range, we wanted to make sure that we used the best tools and similar tools to see whether we
were going to reach a different conclusion than Texas, which we knew to be not acceptable
because its technical analysis was highly flawed. As the commenter indicated here and other
commenters have indicated, PGMs are best suited for RH SIPs and identifying the impacts of
sources or groups of sources when the tools allow. After the BART guidelines were published in
the Federal Register on July 6, 2005, ENVIRON, a national consulting firm with expertise in
photochemical modeling, worked under contract with TCEQ to refine the ability of CAMx to
model single facilities and single point sources using advanced techniques of PiG with PS AT and
particulate chemistry in the plume as well as in the CAMx grids. EPA OAQPS and EPA Region
6 consulted with FLM representatives and approved Texas' BART screening modeling protocol
using these new tools in CAMx.454 One of the model apportionments that Texas utilized in their
BART screening was a source group of one source (Luminant's Monticello facility). This is an
453 2005 BART Guidelines at 70 FR 39104; 39123- 39124 (My 6, 2005); "Theoretically, the CALPUFF chemistry
simulations, in total, may lead to model predictions that are generally overestimated at distances downwind of 200
km. Again, States can make judgments concerning the conservativeness or overestimation, if any, of the results.
The use of other models and techniques to estimate if a source causes or contributes to visibility impairment may be
considered by the State, and the BART guidelines preserve a State's ability to use other models. Regional scale
photochemical grid models may have merit, but such models have been designed to assess cumulative impacts, not
impacts from individual sources. Such models are very resource intensive and time consuming relative to
CALPUFF, but States may consider their use for SIP development in the future as they are adapted and
demonstrated to be appropriate for single source applications."
454 See Appendix 9-4: CAMx Modeling Protocol, Screening Analysis of Potentially BART-Eligible Sources in
Texas of the Texas regional haze SIP
561
-------
example, but there were several other source groups that had only a few sources and a number of
facilities did their own CAMx modeling of only their facility to screen out of BART applicability
(See our BART TSD). Therefore, we disagree with the commenter that using CAMx to assess
single source impacts has never been completed previously for RH SIPs.
The commenter briefly mentioned model performance concerns with CAMx that we will
address in more detail in responses to more specific model performance comments below.
However, we do note that EPA's analysis (cited by commenters below and referred to by the
commenter) indicated that CAMx was one of the best performing models (performed
significantly better than CALPUFF in a number of metrics) and the PGM models (such as
CAMx) are the only ones that we can evaluate using Relative Response Factors (RRFs) and
baseline monitoring that allows the overall modeling results to be used in a relative sense to
estimate changes to baseline monitored values based on the relative change in model results
between a basecase and a different modeled scenario. The use of the RRFs and the CAMx
modeling results in a relative and not absolute sense, which removes much of the concerns about
potential model performance issues. CALPUFF has no way to compare in a relative sense and
the results are used directly, which makes CALPUFF results more uncertain. In addition, despite
the point there are refined techniques that may allow CALPUFF to be used at greater distances,
this approach has not been through full model evaluations by EPA, and since CALPUFF was
initially promulgated in 2003 for distance less than 300 km, this raises concerns as to this
approach.
Another reason we were concerned with CALPUFF is that CALPUFF was never fully approved
for chemistry modeling and was only approved to be used for BART assessments as one of the
five factors in the overall RH BART analysis.455 As discussed in our modeling TSD, CAMx and
other PGMs have been approved by EPA in modeling protocols for regional haze analysis,
including full chemistry calculations.
We disagree that there is no precedent for using CAMx in this way. TCEQ also used CAMx with
PiG and PSAT to screen out groups of sources and individual sources for their impacts on Class I
areas for BART screening. TCEQ contracted for the development of PiG with PSAT in order to
do the BART screening with CAMx in 2006-2007. TCEQ screened out a number of Luminant's
EGU sources (for PM impacts only) using the CAMx modeling of small groups of sources and
individual sources. One of the reasons that TCEQ wanted to use CAMx instead of CALPUFF
was due to the distances and potential uncertainties of CALPUFF model results at these
distances. The only difference between our analysis with CAMx and CENRAP and TCEQ's
analysis is we did source apportionment on more individual facilities and sources compared to
the analysis that TCEQ and 6 other facilities conducted and TCEQ included in its RH SIP. We
only refined the CAMx modeling that CENRAP and TCEQ had generated. As discussed in the
TSD, these facilities are very large sources of emissions and each facility is large enough that
looking at individual sources for potential impacts is not dissimilar to what the states or EPA
would also do for a nonattainment SIP for ozone or PM NAAQS. Therefore, our use of CAMx
455 Discussed in 2003 App. W promulgation of use of CALPUFF for LRT (2003 App. W, FR Vol. 68, No. 72, April
15, 2003), 2005 BART guidelines cited by commenter (2005 BART Guidelines, FR Vol. 70, No. 128 p.39122-
39125) and also discussed in response to a comment in EPA's Final FIP in New Mexico FR Vol. 76, No. 162,
August 22, 2011, p. 52431-52434
562
-------
for evaluation of additional potential RP controls is consistent with the state's SIP submission.
We address the comment that App. W does not approve CAMx for long range transport
elsewhere, but we disagree with the commenter and note that the only models approved on a
case-by-case basis that include a 'full PM2.5 chemistry mechanism' are PGMs such as CAMx.
CALPUFF is not approved for full chemistry and therefore PGMs were used for the 'full
chemistry' needs of a regional haze modeling exercise for LTS/RP.
The commenter indicated that EPA cites no instance in which it has successfully modeled
visibility benefits from individual controls at the distances involved here with CAMx and EPA's
deviations from CENRAP standard protocols—such as adjusting the grid and limiting the
facilities analyzed—suggest that EPA acknowledges that its approach here is unique and outside
the model's designed capability. We strongly disagree with the commenter. CAMx has been
used in national EPA rule makings and regulatory impact assessments including the assessments
of additional reductions needed to bring air pollution levels down in specific states and at
specific monitors. There are numerous attainment demonstrations in Texas that have included
modeling the benefits of controls on single facilities or small groups of facilities. Texas has
previously modeled sensitivity of controlling NOx on some of the EGUs in East Texas to
evaluate if the impacts would help reduce ozone levels in the DFW area (2007 DFW Ozone SIP).
Minnesota previously conducted CAMx modeling for a single facility as part of its RH SIP
revisions.
CAMx and CMAQ have been used in a number of regulatory actions and even in Texas's BART
screening of sources, which is very similar to evaluating the benefit of controls on a specific
source. The commenter seems to indicate that we inappropriately adjusted the grid and limited
the facilities analyzed. We did contract for ENVIRON to do a smaller modeling grid (12 km) to
try to better characterize chemistry and transport in the modeling domain of interest (Texas and
Oklahoma and their Class I areas).456 CENRAP, in 2006-2007 originally tried to evaluate and do
a 12 km grid, instead of 36 km grid but it was a large resource burden on computer run time (9-
10 years ago) and the performance improvements were not significant enough to justify the
increased run-time and processing times.457 With the advancements in computing power since
then however, we were able to do a refined 12 km grid over the area of interest to result in a
more accurate characterization of chemistry and transport.
In summary, there are many reasons that support our selection of CAMx over CALPUFF for the
purposes of this action. Again, CAMx is better suited for evaluating the reasonable progress
metric of improvement on the 20% worst days. It is also better suited for evaluating multiple
sources individually in a complex airshed. In addition, many publically available references
point to CALPUFF's potential overprediction and general performance concerns at the distances
at issue here. Any bias issues in CAMx are ameliorated by tethering the model to real
456 TCEQ also included a 12 km nested grid over Texas and the Class I areas surrounding Texas. (TCEQ's
Modeling Guidance- Guidance for the Application of the CAMx Hybrid Photochemical Model to Assess Visibility
Impacts of Texas BART sources at Class I Areas" from this point the document will be referred to as "TX BART
CAMx Guidance."
457 Technical Support Document for CENRAP Emissions and Air Quality Modeling to Support Regional Haze
State Implementation Plans, September 2007 at Section 1.3.4 available in the docket for this action
563
-------
monitoring data, through the use of relative response factors generated by modeling of base and
future cases to predict future monitored values.
Comment: [Luminant (0061) p. 116] Luminant asserted that EPA also claims that it was
"concerned that CALPUFF could overestimate impacts," but that claim is contradicted by prior
EPA findings.725 In prior regional haze actions, EPA has defended its use of CALPUFF against
criticisms that the model "overstates visibility impacts."726 EPA has explained that "it is
uncertain whether the simplified chemistry [in CALPUFF] will always overpredict visibility
impacts."727 And, contrary to its rationale here, EPA has found that "the CALPUFF model
frequently predicted lower nitrate concentrations compared to the Comprehensive Air Quality
Model (CAMx) photochemical grid model. . . ,"728 Moreover, EPA has previously explained,
again contrary to its rationale here, that CALPUFF is used "to estimate the 98th percentile
visibility impairment rather than the highest daily impact value" to compensate for the fact that
CALPUFF "might in some cases predict a maximum 24-hour impact that is an 'outlier.'"729 In
other words, any "overprediction" in CALPUFF is compensated for by eliminating the highest
outlier values. For this reason, there is no justification for EPA's attempt to lower the impact
thresholds it has previously used in other regional haze actions (down from 0.5 deciview to "on
the order of 0.1-0.15 deciview"730) because any overprediction by CALPUFF is already
accounted for in how the results are reported.
Moreover, Luminant noted that even if EPA were correct that CALPUFF overpredicts visibility
impacts, CAMx suffers from the same flaw, particularly at these long distances. Studies
conducted for EPA by ENVIRON—the same contractor that EPA hired to perform its Texas
source modeling—conclude that CAMx overpredicts visibility impacts by three times, as EPA
now claims CALPUFF does.731 EPA provides no contrary analysis or empirical validation here
to demonstrate otherwise. Thus, based on EPA's own prior studies, CAMx is not an accurate tool
for the job that EPA seeks to undertake here. Nor is there any basis or need for adjusting
visibility thresholds previously developed by EPA using CALPUFF modeling in order to make
direct comparisons to EPA's CAMx modeling results here. Even if EPA's claim that CALPUFF
overpredicts is correct, EPA's own prior analysis shows that CAMx does as well. Thus, if CAMx
could be used at all here, to more accurately predict actual visibility conditions, EPA's modeling
would need to be adjusted lower by a factor of one-third, thus providing even less justification
for EPA's proposal.732
Footnotes:
725 79 Fed. Reg. at 74,878.
726 77 Fed. Reg. at 20,908.
727 Id.
728 Id.
729 Id.
730 FIPTSDat A-75.
731 AECOM Report at 6-2 to 6-3.
732 Id. at 6-10.
564
-------
Response: We disagree with the commenter and their characterization of the situation. In the
BART guidelines proposal, EPA originally proposed using the maximum value from CALPUFF
modeling results but in the final EPA weighted the concerns with CALPUFF's simplified
chemistry as the commenter noted and provided guidance that the 98th percentile visibility
impairment should be used to compensate for NOx chemistry concerns, the simplified chemistry,
and a concern of using the maximum distribution point for regulatory decisions (Final BART
Guidelines (70 FR 39104, 39121).The comment that CALPUFF frequently predicts lower nitrate
values than CAMx is not an accurate characterization. The reference is citing to a presentation at
Community Modeling and Analysis System (CMAS) conference in 2010 by Bret Anderson and
others including Erik Snyder of Region 6. The presentation indicated that for one modeling
analysis CAMx results were slightly higher than CALPUFF results, but there were also more
days where CALPUFF nitrates were larger than the CAMx results. The authors of the study
indicated that it could be a number of issues that would need to be investigated further before
drawing conclusions on the results and this was only one limited analysis and does demonstrate
that modeling the same values will always give higher nitrate values from CAMx compared to
CALPUFF results. A more accurate characterization is that initial modeling of a source did
present nitrate values that were sometimes in the CAMx results but not in the CALPUFF results,
but a general conclusion that CAMx will predict higher nitrate values than CALPUFF in all
cases is unfounded and not supported by the limited data.458
The commenter did not seem to understand what was discussed in our TSD (page A-75) on the
point of taking into account the differences in emission inputs and difference in metrics. We
were not lowering the delta dv from 0.5 to 0.1-0.15 due to CALPUFF model performance
concerns. CALPUFF modeling for BART purposes used by states often used the 98th percentile
value (High-eighth-high) to compare to the 0.5 delta-dv threshold for BART applicability. RP
modeling evaluates impacts on 20% worst days, which is an average value over a large number
of days and is statistically much further down on the curve of ranked impacts than the 98th
percentile and is therefore much smaller than 0.5 delta-dv just due to the difference in metrics
utilized for visibility impairment versus BART screening. The other factor that makes
CALPUFF results for BART different and larger than RP modeling with CAMx or CALPUFF is
that BART modeling uses the highest 24-hour emission rate over a 3-5 year period that was often
on the order of twice as much as the values typically used for RP analysis. Therefore, the
difference in emission rate by itself would lead to CAMx RP model values to be on the order of
50% of the value of CALPUFF modeling results for BART screening. Just these two differences
would lead to the CAMx results for RP being much lower than the same source modeled using
the BART screening maximum emissions and 98th percentile value. Thus, the commenter is
wrong and it is reasonable to require RP controls for impacts much lower than 0.5 delta-dv.
We respond to the comments about Environ's model performance analysis below, but we
disagree with the conclusions. We do not conclude, nor does the cited AECOM report conclude
that all CALPUFF and CAMx model results should be calibrated down to l/3rd the model output
458 CMAS 2010 conference Powerpoint "Proof-of-Concept Evaluation of Use of Photochemical Grid Model Source
Apportionment Techniques for Prevention of Significant Deterioration of Air Quality Analysis Requirements"; Bret
Anderson, Kirk Baker, Ralph Morris, Chris Emery, Andy Hawkins, Erik Snyder; 'anderson-proof-of-
concept_20 lO.pptx'
565
-------
values based on the subset of statistics the commenter selected from the two field studies. EPA
does not allow directly calibrating model results,459 especially not based on limited field studies
that are not even in the same climatological situation as our analysis. Furthermore, the alleged
factor of 3 overprediction is based on the direct model results which is not the way we used our
CAMx results. We note that our CAMx modeling results (as noted in our TSD and in other
responses) are used in a relative sense and the absolute model values are not directly compared to
any threshold value. In our CAMx analysis we used Relative Response Factors (as recommended
by EPA's Regional Haze modeling guidance - 2007 Final guidance460 and 2014 Draft
guidance461) that uses the future predicted values at a monitor and divides these values by the
baseline predicted values and then multiplies this ratio times the actual baseline monitored values
to get an estimate of future year values. This minimizes bias issues with PGMs such as our
CAMx modeling. Any bias issues in CAMx are ameliorated by tethering the model to real
monitoring data, through the use of relative response factors generated by modeling of base and
future cases to predict future monitored values.
In conclusion, we disagree with the commenter and will continue to follow EPA's guidance for
using RRFs with our CAMx modeling results. As also discussed in other responses on the size of
the impacts, we continue to maintain that CAMx modeling results in our analysis are naturally
much lower due to the difference in metrics and emission rates that are modeled. If CALPUFF
BART screening modeling would require a source to model maximum 24-hour emission rates of
2000 lb/day of SO2 and the monthly actual emission rate (similar to our CAMx analysis) was
only 1000 lb/day, it would not be reasonable to use a 0.5 delta-dv threshold in both cases. As
discussed elsewhere, we do not support nor do we think any calibrating of our CAMx modeling
results are reasonable, nor required. Further, any calibrating of model results is specifically
restricted by App. W.
Comment: [Luminant (0061) p. 118] Having chosen to use CAMx modeling instead, EPA
should judge the results of the modeling as they were calculated—not attempt to adjust them to
create an artificial justification for its proposal. 739
Footnotes:
737 FIP TSD at A-37.
738 79 Fed. Reg. at 74,881.
739 Further, in directing ENVIRON to use a newer version of CAMx than CENRAP used, EPA
violated its own guidance that "the better course is to rely on modeling based on the same
version of the model that the State employed to ensure we are using a consistent comparison." 77
Fed. Reg. at 20,908 (citing Mont. Sulphur & Chem. Co. v. EPA, 2012 U.S. App. LEXIS 1056
(9th Cir. Jan. 19, 2012)).
459 App. W, Section 7.2.9(a) ".. .Therefore, model calibration is unacceptable."
460 EPA, 2007. Guidance on the Use of Models and Other Analyses for Demonstrating
Attainment of Air Quality Goals for Ozone, PM2.5, and Regional Haze. EPA-454/B-07-002.
461 Draft Modeling Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air
Quality Goals for Ozone, PM2.5, and Regional Haze http://www.epa.gov/ttn/scram/guidance/guide/Draft_03-PM-
RH_Modeling_Guidance-2014 .pdf
566
-------
Response: We address comments concerning "adjustments" to the CAMx model results
elsewhere in this document. As discussed in another response, we reviewed CAMx modeling
results for our 2002 basecase modeling using the updated CAMx version and other changes that
we made and did not see any significant changes in model results. The situation that the
commenter is referring to in Montana is distinguishable in that it dealt with a situation where the
original Montana SIP work had been completed using one dispersion model (ISC) and the
discussion was about to update witha recently promulgated model AERMOD that replaced ISC
as the preferred model. In that situation, AERMOD handled meteorology and dispersion in a
very different way than ISC and changes in modeling results would have most likely been seen,
thus sending the state potentially back to the drawing board to develop a new SIP. EPA knew
this and expected changes in model outputs between these two models and granted a general
grace period of one year for most permitting and other actions that were already in the queue so
that they could be processed with their existing ISC based analyses and future analyses submitted
would use AERMOD. Our situation is distinguishable from the Montana situation in that we did
not change model platforms that have fundamental differences in their formulation. We utilized
modeling generated with an older version of CAMx that has been approved in a number of SIPs
and utilized a new version for our analysis that had some science improvements that are
comparable to scientific tweaks to the existing model formulation and not fundamental changes
in model formulation.462
Comment: The margin of error in different CAMx modeling compilations was not considered.
[Xcel Energy (0064) p. 29]
Xcel Energy stated that model performance varies as a function of many factors including, but
not limited to, the following: pollutant, time of day, time of year, accuracy of input data (e.g.,
emissions inventories, meteorological data, etc.), selection of model options/parameterizations,
resolution, and geographical location. Studies have shown that photochemical models tend to
under-predict sulfate concentrations in general, while for nitrate concentrations they tend to over-
predict in the winter and under-predict in the summer. See Appendix C to comment 0064,
"Compilation and Interpretation of Photochemical Model Performance Statistics," Heather
Simon, et. al., U.S. EPA, Atmospheric Environment (2012). Performance of photochemical
models also appears to be significantly better in the Eastern U.S. than the Western U.S. (west of
the Rocky Mountains). Something as seemingly insignificant as the choice of compiler can lead
to slight differences in modeled results. When utilizing modeling for regulatory purposes, the
bias and error should be reported and taken into consideration when proposing controls. From a
review of the EPA provided documentation, these items were not considered and instead a
definitive result was predicted.
Xcel Energy stated that the EPA also arbitrarily used CAMx modeling to model transport of haze
from sources despite significant known limitations with CAMx involving over-prediction of
emissions at longer distances, and even though EPA has consistently promoted the use of
CALPUFF. See, Documentation of The Evaluation of Calpuff and Other Long Range Transport
Models Using Tracer Field Experiment Data, Environ International Corporation, EPA Contract
462 Environ February 2013 memorandum, "2002 Baseline CAMx Simulation, Texas Regional Haze Evaluation";
'Memo_TXHAZE_2002CAMx_ENV_2 lFeb2013 .docx'
567
-------
No. EP-D-07-102, Work Assignment No. 4-06, Figure 6-13 (2012), available at
http://www.epa.gov/ttn/scram/reports/EPA- 454_R-12-003.pdf.
Response: See our response to the next comment and our responses above to similar comments.
Comment: [Luminant (0061) p. 117] Luminant stated that EPA fails to perform a model
performance evaluation of its additional modeling, as required by EPA's own modeling
guidance.734 Had any state presented a SIP revision without a performance evaluation, EPA
would not have allowed the use of such modeling. Indeed, as a contractor for CENRAP,
ENVIRON conducted an extensive model performance evaluation, which was incorporated into
the Texas SIP.735 EPA provides no explanation of why it is deviating from its own modeling
guidelines in this proposal and certainly no justification for doing so. In fact, even ENVIRON
finds that its own results reported to EPA are "suspect" and warns that "care should be taken in
the interpretation of the visibility projections at these Class I areas."736
Footnotes:
734 ENVIRON Feb. 2013 Memo at 8.
735 2009 Texas SIP Narrative app 8-1, ch. 3 ("Model Performance Evaluation").
736 Id. at 12.
Response: We did not do a detailed model performance of the 2002 basecase because that had
already been done by CENRAP. The only changes we made in the 2002 basecase was to use a
newer version of the CAMx model, include a flexi-nested 12 km computational grid over the
Oklahoma and Texas areas of interest, and incorporate an updated chemical mechanism which
included re-categorizing emissions to the new mechanism to utilize improvements in the science
for our analysis and decisions. As we discussed in our proposal materials, these changes were not
large and did not warrant a full model performance evaluation. We did compare model results
with previous results and determined that model results were very similar and extinctions at
Class I areas were similar and deemed acceptable. We reiterate that all of this information was
included in our record at the time of proposal and was subject to public review. It is not
uncommon as a professional practice in the modeling community to do some small updates such
as we did and not perform a full updated model performance analysis. We did evaluate model
projections on the worst 20% days, best 20% days in comparison to monitored values and we
also compared to previous similar comparisons performed with CENRAP's CMAQ modeling
results.463 Overall there was a significant underestimation of sulfate compounds compared to
monitored values on the 20% worst days. Contrary to the assertions of several commenters that
CAMx results are biased high, this was not true for the 20% worst days at Class I areas in Texas
and near Texas. This is important because it indicates that the CAMx modeling was actually
underestimating the SO2 impacts on most of the 20% worst days, which is opposite of the
comment that CAMx results overestimated by a factor of 3.While all models have some bias
issues and never replicate the atmosphere exactly, the previous modeling results developed by
ENVIRON for CENRAP did include an extensive model performance analysis and were
considered acceptable by CENRAP and TCEQ, and we have previously approved the CENRAP
463 Environ February 2013 memorandum, "2002 Baseline CAMx Simulation, Texas Regional Haze Evaluation";
'Memo_TXHAZE_2002CAMx_ENV_2 lFeb2013 .docx'
568
-------
modeling database in other CENRAP states. Considering the limited scale of the changes, the
analysis of our modeling data compared to 2002 monitoring data, and no distinguishing of any
differences in model projections compared to CENRAP's 2002 modeling, a full performance
analysis was not necessary. Overall, the changes we did in the basecase modeling (12 km grid,
updated CAMx and chemical mechanism and chemical speciation with some sources evaluated
with updated PiG and PSAT) would not be expected to result in much change in model
performance and in fact, they did not.
We respond to the comments about ENVIRON's model performance analysis of CAMx below,
but we disagree with the conclusions made by commenters. We do not conclude, nor does the
report conclude that all CAMx model results should be calibrated down to l/3rd the model output
values based on the subset of statistics the commenter selected from the two field studies. EPA
does not allow directly calibrating model results,464 especially when not based on limited field
studies that are not even in the same climatological situation as our analysis. Furthermore, the
alleged factor of 3 overprediction is based on the direct model results which is not the way we
used our CAMx results. ENVIRON'S thorough model performance evaluation for CENRAP did
not indicate an overprediction bias by a factor of 3 (CENRAP TSD), therefore it does not seem
appropriate to consider that CAMx is performing in the manner as badly as the commenter
suggests. In our evaluation of the basecase monitored to model values for the 20% worst days,
the model was actually underpredicting overall and for sulfate species.465 In the ETEX and
CAPTEX studies performed for EPA, CAMx had some of the best model performance of the
models evaluated and CALPUFF had some of the worst model performance evaluations. We
note that our CAMx modeling results (as noted in our TSD and in other responses) are used in a
relative sense and the absolute model values are not directly compared to any threshold value.
Here, in our CAMx analysis we used Relative Response Factors (as recommended by EPA's
Regional Haze modeling guidance (2007 Final guidance and 2014 Draft guidance) that uses the
future predicted values at a monitor and divides these values by the baseline predicted values and
then multiplies this ratio times the actual baseline monitored values to get an estimate of future
year values. This minimizes bias and error issues with PGMs such as our CAMx modeling. Any
bias issues in CAMx are ameliorated by tethering the model to real monitoring data, through the
use of relative response factors generated by modeling of base and future cases to predict future
monitored values.
As discussed in other responses, the selection of CAMx for this analysis was not arbitrary and
was in-line with both CENRAP and TCEQ selection to use CAMx modeling to assess impacts of
potential RP sources. CAMx with the PiG and PSAT tools was also used by TCEQ for BART
screening, so there were several precedents to point to that utilized CAMx for our work. Based
on recent modeling and guidance development conducted by EPA's OAQPS office, CAMx also
appears to be one of the better tools for this type of analysis.466 As discussed in detail in other
464 App. W, Section 7.2.9(a) ".. .Therefore, model calibration is unacceptable."
465 Environ February 2013 memorandum, "2002 Baseline CAMx Simulation, Texas Regional Haze Evaluation";
'Memo_TXHAZE_2002CAMx_ENV_2 lFeb2013 .docx'
466 Draft Guidance on the use of models for assessing the impacts of emissions from single sources on the
secondarily formed pollutants ozone and PM? s (EPA-454/.P-.1.5-00.1.) and Interagency Workgroup on Air Quality
Modeling Phase 3 Summary Report: Long Range Transport and Air Quality Related Values (EPA-454/P-.1.5-003).
569
-------
responses, CALPUFF has multiple limitations, especially in the context of evaluating impacts for
RP and potential RP controls.
Comment: [AECOM (0061 and 0075) p. 6-2] AECOM stated that it is well-documented that
CAMx has transport limitations and uncertainties. For example, to the best of our knowledge,
there has not been a single rigorous and comprehensive evaluation on the ability of PGMs to
accurately represent the transport of pollutants over specific distance ranges, especially for
distances on the order of at least 500 km. In this regard, EPA's proposal does not address
whether CAMx can accurately represent the impacts of pollutants at the very large distances
from the sources that were simulated. EPA asserts that CALPUFF has some limitations at large
distances, but it does not present data or analysis to show that CAMx does not.
Response: See previous responses for comments above. We address the comment about model
performance evaluations on the order of at least 500 km elsewhere in responses to tracer study
comments in this document. As discussed in responses below to evaluations with tracer studies,
PGM such as CAMx are some of the best performing models in general and have been shown to
perform significantly better than CALPUFF. See our responses to ETEX and CAPTEX tracer
studies and other model performance comments.
Comment: [Luminant (0061), p. i] Luminant stated that the proposal burdens a handful of Texas
generating units, located far away from these Class I areas, with massive costs that would
threaten their continued operation and economic viability and the thousands of jobs they provide.
Luminant stated that, to arrive at this conclusion, EPA invents a methodology found nowhere in
the statute or regulations and applies it in a seemingly random manner. The map below depicts
EPA's skewed and unprecedented approach for Texas, with sources far away from these federal
areas somehow being captured by EPA's novel methodology, while closer sources are excluded.
There is no discernible rationale for this flawed approach.
Map of Sources and Federal Areas Provided by Luminant (0061)
570
-------
WJeMta
Mountain!
VirC^;
I
*
\
Montkdlo
Guadalupe 0
Mountain!
956 lun
Big Band
C\ >
<7
Legend
Te*asSources far^hlcb FPA DrtemnnM New
Cortrolsare h«Quired
TexasSouices forwhlcli EPA DetennlntsNew
Centrali at* NotRetfutred
Okhharn a N Ofi-BART 5 ouft*s th at EFA Does
Mot Evaluate
>
\* .
3anrlnin
M.ltlll Ldke
Hj Blirmi
Luminant asserted that never before has EPA singled out individual sources using the new
approach that EPA employs here to derail the regional haze plans by Texas and Oklahoma. The
record is irrefutable that EPA has routinely approved other states' regional haze plans that look
just like the Texas plan and achieved the same level of progress.
Luminant stated that, with this unprecedented approach, EPA resorts to stretching the science
and the law beyond all recognition to justify its preordained result. EPA fails to follow its own
modeling protocols or to validate its modeling, and it uses results well outside the model's well-
recognized limitations.
Response: These comments are addressed in other responses below where we discuss
identification of sources for additional analysis. The comment that EPA failed to follow its own
modeling protocols and model validation/model performance analysis are addressed in responses
to similar comments above. In response to the comment of utilizing the model outside the
model's limitations we do not agree. Models are tools to be used to assess specific situation as
571
-------
needed, and while the model and analyses we used in this case have not been used many times
for RH purposes that does not prohibit us from doing so if the appropriate modeling tools are
available. As discussed in responses elsewhere, TCEQ used CAMx modeling to assess small
groups and individual point sources to identify their impacts on surrounding Class I areas. TCEQ
sponsored additional scientific developments within CAMx to improve the science of assessing
single source impacts on visibility impairment at downwind Class I areas in 2006/2007 in order
to evaluate a large number of facilities with CAMx for BART screening. Texas's RH SIP
included six CAMx modeling evaluations in 2007 for single facilities to screen out from BART.
Therefore, using CAMx to assess visibility impairment impacts of a single facility or a large
source is not a new or unprecedented approach. CENRAP utilized CAMX modeling to assess
visibility impacts from source types and source regions, such as point sources within a particular
state.
Beyond what has been applied in practice in various SIPs and FIPs, there is no visibility benefit
deciview significance threshold that exists in the RH regulation or guidance. As a result, the EPA
interprets the visibility improvement from the modeling results on a case-by-case basis. As
discussed elsewhere, the EPA believes that it is appropriate to apply different interpretations of
rounding conventions based on the circumstances of the emissions reductions scenario(s) and
geographic scope of the analysis.
Comment: [UARG (0065) p. 26-27] UARG stated that the EPA - apparently for the first time in
a regional haze rule - rejects reliance on the CALPUFF model and instead relies on CAMx. Id.
at 74,877-78. EPA's BART Guidelines, as the proposed rule acknowledges, state that CALPUFF
is "the best regulatory modeling application currently available for predicting a single source's
contribution to visibility impairment." See id. at 74,847. CALPUFF is, moreover, "the only EPA-
approved model for use in estimating single source pollutant concentrations resulting from the
long range transport of primary pollutants." 70 Fed. Reg. 39,104, 39,122 (July 6, 2005). In
previous regional haze rulemakings, EPA has refused to accept anything other than CALPUFF
modeling performed using CALPUFF version 5.8, the version of the model approved in EPA
regulations. EPA has even rejected visibility modeling that was conducted using more up-to-date
versions of CALPUFF that are intended to correct errors in version 5.8. See, e.g., 76 Fed. Reg.
52,388, 52,431 (Aug. 22, 2011) (rejecting modeling results from CALPUFF versions 6.112 and
6.4). The primary rationale that EPA provides to support its use of CAMx here is that "[t]he
large distances between sources and Class I areas are outside the typical range of CALPUFF.
Because of the range, we were concerned that CALPUFF could overestimate impacts." 79 Fed.
Reg. at 74,878. But EPA provides no technical assessment supporting the reliability of CAMx to
model visibility impacts of single sources over long ranges. That the single model on which EPA
heretofore has insisted cannot reliably predict impacts over the distances at issue in this proposed
rule should have signaled to EPA that its proposed rule is arbitrary and irrational and that those
distances are simply too great to allow for imposition of reasonable progress regulations by EPA.
In any event, EPA provides no valid basis for using CAMx uniquely here, instead of CALPUFF.
Response: As discussed in several other responses to comments concerning modeling in this
section, we discuss a number of concerns with using CALPUFF at the distances present in our
analysis. CENRAP and Texas used CAMx modeling for their RP analysis. We disagree with the
points raised by the commenter in the above comment and our positions on these points were
572
-------
discussed in our TSDs. In addition, even Texas raised concerns with CALPUFF in their TX
BART CAMx Guidance.467 Texas argued that with the updates to CAMx that they helped
develop with PiG and PSAT with full chemistry in the PiG, that CAMx was the best tool to use
for modeling impacts of groups of sources and single sources in Texas on nearby Class I areas in
Texas and outside of Texas (multiple locations in the TX BART CAMx Guidance including page
1-1 to 1-10). As the commenter noted, EPA has previously not approved CALPUFF modeling
with more updated chemical mechanisms because the new versions have not gone through
performance testing and approval by EPA and therefore are not acceptable regulatory versions of
the CALPUFF modeling system (see 76 FR 52431, 52434 (August 22, 2011). EPA did provide
all of its technical assessments in the record to this action, including documentation that
CENRAP and TCEQ had used CAMx in similar analyses as part of their work products over 7
years prior to our proposal. As also discussed in similar responses, EPA, TCEQ and CENRAP all
used CAMx modeling for evaluation of potential RP and BART sources, including single facility
analyses. Therefore, including additional similar responses to CAMx vs. CALPUFF elsewhere,
EPA's use of CAMx instead of CALPUFF was an appropriate response given the continued
concerns with using CALPUFF in Texas for distances outside the normal operation of
CALPUFF and was not arbitrary or capricious.
Comment: [Southwestern Public Service [SPS]/Xcel Energy (0053-24 and 0054-4)] SPS (Xcel
Energy) stated that the EPA's proposal is based on out-of-date emissions and meteorological
data. EPA acknowledges that the scrubber retrofits would not be in place by the end of the first
planning period for the regional haze in 2018. As a result, consideration of these expensive
controls should be evaluated fully in the next planning period as part of the Texas state plan due
to EPA in 2018. This would allow consideration of the updated data, current conditions, and the
effect of new programs, and avoid undue reliance on data and modeling that are now over 10
years old.
In addition, by that time SPS would have better sense of its compliance plans for the proposed
Section 111(d) clean power plan. This would allow for more comprehensive planning to assure
the best economic environmental benefit of our customers without impacting long-term progress
towards the goal of attaining natural visibility in Texas and Oklahoma Class I areas.
Response: We address the comments about out-of-date emissions and meteorological data in
this response here. Comments concerning controls beyond the first planning period are addressed
elsewhere in this document.
As discussed elsewhere, EPA's responsibility in this process was to evaluate Texas' RH Plan and
either approve or disapprove the submittal, and if we disapproved it, to promulgate a FIP to
address the deficiencies. EPA evaluated the Texas RH modeling analysis for RP that was based
on 2002 meteorology (as agreed in modeling protocols) and 2002 and 2018 emission inventory
databases provided with the State's SIP submission. Even though we found Texas' technical
analysis flawed, we did use the CENRAP databases in our technical analysis with CAMx, as
Texas did.
467 TCEQ's Modeling Guidance- Guidance for the Application of the CAMx Hybrid Photochemical Model to Assess
Visibility Impacts of Texas BART sources at Class I Areas (Decmber 13, 2007)
573
-------
As discussed below, we did consider recent actual emissions in a number of ways to assess the
visibility impacts and potential benefits from controls on specific sources. We also used the
most recent CAMx model version and updated chemical mechanism that included improvements
to the source apportionment of single point sources and plume in grid algorithms to use the most
recent science for our RP evaluations. These were relatively small changes compared to doing a
full update to all the emission inventories.
The commenter indicated that the meteorology data was outdated. What the commenter is
suggesting is that EPA should have generated an entirely new reasonable progress modeling
demonstration. The development of the RH modeling databases by CENRAP took about 5 years
and much more money and resources than we had available and would have been well beyond
the scope of work in finding Texas' technical analysis to be flawed and developing a
methodology to address the flaws. Updating the meteorology data would have been a major
update, not like the minor updates we did for the technical analysis. Therefore, we did not
believe it was necessary to update the meteorology data.
With regard to comments that we used out-of-date emissions and we should have performed a
more complete update of the inventory during our technical analysis, a full emission inventory
update for all emission categories such as biogenic, mobile, non-road, area, and point sources for
2002 and 2018 was not necessary for our evaluation as to whether additional reasonable progress
controls are appropriate for specific sources or groups of sources. We evaluated the existing
CENRAP 2002 and 2018 emission inventories and whether to update parts of the 2018 emission
inventories. We considered updating the EGU inventory with an emissions inventory estimate
developed for modeling performed for the MATS rulemaking. However, there was considerable
uncertainty in the emission reductions projected in that inventory due to MATS compliance.
Comments from Texas and EGU owners468 on a more recent emission inventory (that included
many of the same assumptions on emission reductions due to MATS that were used to develop
the MATS inventory) indicated that no significant emission reductions or major controls were
planned to be installed on EGUs in Texas for SO2 emissions in response to MATS.
As discussed elsewhere, we performed a Q/d analysis to identify those sources with the greatest
potential to impact visibility based on 2009-2011 emissions. We then performed CAMx
modeling using the same CENRAP 2018 projected emissions as Texas to estimate the individual
visibility impacts due to the projected 2018 emissions of those 38 sources identified by the Q/d
analysis. For this modeling, we did do minor updates to emissions on a few specific facilities to
incorporate recent information that had changed for specific units/facilities to reflect installed
controls or improvements in control efficiencies 469 We thought it was appropriate to conduct our
analysis of these select facilities using the most recent information, since this could impact
potential RP control decisions for those facilities. These changes were not large in comparison
to the overall modeled emission inventory. We also considered recent emissions data in
468 Texas comments on Draft IPM modeling conducted by EPA for potential national rule making platform
provided on June 26, 2014. In this docket materials as "TCEQ comment letter to EPA on draft modeling platform
dated June 24, 2014. '2018 EMP signed.pdf'
469 FIP TSD at A-16
574
-------
estimating and evaluating facility-level and unit-level impacts in order to identify those sources
with the largest visibility impacts for additional control analysis.
After our initial modeling analyses to identify those sources with the largest visibility impacts for
additional control analysis, we developed estimated High and Low control-level emission
estimates based on recent actuals emissions and anticipated control efficiencies of potential
controls to assess the potential visibility benefit of controls. As we explain in detail elsewhere
in the FIP TSD and in response to comments, we determined that recent actual emissions were
more representative of anticipated future emissions than the 2018 CENRAP projected emissions
for these EGU sources in Texas based on available information, including the comments from
Texas and EGU owners on future anticipated reductions. In our analysis of potential controls for
individual units, we estimated the potential visibility benefit of controls based on both additional
reductions to the 2018 CENRAP projected emission levels and reductions to recent actual
emission levels. We also utilized recent actual emissions in our cost estimates.
15.c Modeling Emission Inventory
Comment: [AECT (0074) p. 7-8] AECT stated that the EPA's Proposal lacks a reasonable basis
because EPA's FIP visibility modeling predicts higher visibility impacts from Texas emissions
sources in 2018 than will actually occur because those sources' SO2 emissions that were used in
such modeling are outdated and fail to reflect the substantial SO2 emissions reductions those
sources have achieved or will achieve by 2018.
AECT stated that EPA's FIP visibility modeling predicts higher visibility impacts from Texas
emissions sources in 2018 than will actually occur because those sources' SO2 emissions that
EPA used in such modeling are outdated and fail to reflect the substantial SO2 emissions
reductions those sources have achieved or will achieve by 2018. Specifically, in conducting its
FIP visibility modeling, EPA assumed that Texas emissions sources would emit 749,119 tpy of
SO2 in 2018 based on a projection from the 2002 CENRAP SO2 emissions inventory.23 However,
more recent data show that SO2 emissions from Texas emissions sources have decreased much
faster than EPA projected. Indeed, EPA's most recent national SO2 emissions inventory shows
that, by 2011, actual Texas SO2 emissions were only 558,502 tpy, or only about 75% of the
749,119 tpy of SO2 emissions that EPA assumed in conducting its FIP visibility modeling.
Moreover, when EPA's IPM projections are applied to the 558,502 tpy of SO2, the 2018 Texas
SO2 emissions are projected to be only 259,743 tpy, or only about 35% of the 749,119 tpy of SO2
emissions that EPA assumed in conducting its FIP visibility modeling.
AECT noted that a recent court decision requires that EPA consider the effect of the more recent
and more accurate SO2 emissions data in determining whether any additional SO2 emissions
controls are needed for the identified Texas EGUs to meet the reasonable progress
requirements.24 If EPA uses more recent and more accurate SO2 emissions in its FIP visibility
modeling, the predicted visibility impacts from Texas emissions sources in 2018 will be
significantly lower than the visibility impacts predicted by the FIP visibility modeling that EPA
used to support its Proposal. In fact, AECT is confident that the visibility impacts predicted by
such modeling will be so low that EPA will be required to conclude that there is no support for
the proposed requirement for additional SO2 emissions controls on the Texas EGUs.
575
-------
Footnotes:
23 79 Fed. Reg. at 74858, Table 20
24 Sierra Club v. EPA, 671F. 3d 955 (9th Cir. Jan. 20, 2012) (rejecting an EPA SIP approval that
relied on an emissions projection that had been superseded in the time between the state's SIP
submission and EPA's action on it)
Response: See our response to the next comment.
Comment: EPA's FIP modeling inventories are outdated, fail to integrate substantial
recent reductions, and artificially inflate the modeled Impacts of Texas sources.
[Alpine (0078) p. 5, 26; NRG (0078), p. 2, 3]
NRG stated that the EPA used an outdated base year and future year emissions inventory in its
modeling that had the effect of inflating the modeled impacts of Texas sources on visibility
impairment.
As explained in the attached report by Alpine Geophysics (0078), NRG noted that the EPA
assumed through the current proposal that Texas sources would emit 749,119 tons per year (tpy)
of SO2 in 2018 based on a projection from the 2002 CENRAP emissions inventory. This
projection assumed a decrease of over 200,000 tons per year of SO2 emissions from 2002 to
2018. More recent data show that SO2 emissions have fallen much faster than were anticipated in
2002. EPA's most recent National Emissions Inventory shows that, by 2011, actual Texas SO2
emissions were only 558,502 tpy, or 25% lower than the prior projection for 2018. Applying
EPA's IPM projections indicates that 2018 emissions are now projected to be only 259,743 tpy
502, or 65% lower than the projection relied on by EPA's proposal.4
According to NRG, had EPA used its own more recent emissions data and projections, the
modeled impact of Texas sources on visibility impairment would have been significantly
reduced, making clear that further controls are not needed to assure visibility protection. As a
recent 9th Circuit decision made clear, EPA must use the most up-to-date data available in acting
on state plans.5 Thus, EPA may not ignore the effect of this more recent data that implicates
EPA's action on the Texas and Oklahoma regional haze plans.
NRG concluded that additional controls are not needed to achieve the regulatory targets EPA is
proposing to set, as EPA continues to predict through the IPM model that visibility-impairing
emissions will progressively decline through 2018. Alpine Report at 26-27.
Emission and Modeling Inventories Used to Support Disapproval of Texas's SIP Submittal.
Alpine stated that the modeling utilized by EPA to support the proposed rule was based on
emission estimates that are out of date and demonstrated to be erroneously high.
Alpine noted that, to support EPA's disapproval of portions of Texas' SIP submittal, the Agency
relied on 2002 base year modeling projected to 2018, originally conducted for CENRAP in 2007,
augmented with limited source updates as documented in the SIP review process.9 It is my
576
-------
opinion that EPA's modeling, using this dated modeling platform, generated regional haze
results and modeled source impacts that were erroneously high.
Alpine stated that EPA's modeling inventories and projections are objectively inaccurate and
artificially inflate the modeled impacts of Texas sources on regional haze levels.
Alpine stated that current EPA emission inventory data10 shows reductions in emissions in most
states beyond what was projected in the 2018 CENRAP modeling, including large reductions in
emissions from Texas and the southern States. Emissions from both EGU and non-EGU Texas
point sources are lower than have been projected in the CENRAP modeling and should be
considered in concert with emission reductions recently reported by EPA from all other
contributing sources in the modeling domain.
Comparing emissions in the Texas Regional Haze FIP TSD to EPA's most recently released
modeling platform and emissions projections, Alpine noted that total SO2 emissions, from all
anthropogenic sources, are significantly higher in magnitude in both the FIP base year (2002)
and projection year (2018), relative to base year (2011) and projections (2018) from the new
platform. Alpine provided a figure showing an annual SO2 emissions comparison for Texas,
including the totals reported in the EPA FIP TSD and compared to the recent EPA modeling
platform. (Figure 16 in Alpine comment 0078, not reproduced here)
Alpine stated that much of this difference is attributed to the 2007 vintage of the EPA's FIP
modeling inventories and demonstrates the improved data, methods, and models used to prepare
current, state-of-knowledge emission inventories in the past year. Simply stated, the overall
influence of SO2 emissions from current EPA emissions and projections is significantly lower in
2011 and 2018 than those emissions and projections EPA used to simulate visibility in the Texas
FIP modeling.
As an example, Alpine noted that both EPA and Texas have cited the use of the Integrated
Planning Model (IPM), a multiregional, dynamic, deterministic linear programming model of the
U.S. electric power sector. Version 5.13 of this model reflects state rules, consent decrees, and
announced shutdowns through August, 2013. IPM 5.13 was significantly updated from IPM 2.19
that EPA relied on in its review of TCEQ's SIP and represents electricity demand projections
from the Annual Energy Outlook 2013.
Alpine pointed out that in Table 8 of EPA's FIP TSD, the Agency noted annual emission
estimates of SO2 from the EGU sector. We have added to this table EPA's most current estimate
of Texas' SO2 emissions from EGUs as projected to 2018 with IPM 5.13
Comparison of Texas 2002 Baseline SO2 emissions, 2015 CAIR EGUs Budget and 2018 IPM
Predicted SO2 Emissions. (Table 6 of Alpine comment 0078)
SO2 Emissions
Texas SO2 Emissions (tpy)
CENRAP 2002 base case
550,000
EPA's CAIR budget for Texas EGUs for 2015
225,000
IPM 2.19 projection CENRAP modeled for 2018
350,000
577
-------
IPM 5.13 projection EPA modeled for 2018
144,520
Reviewing these latest IPM projections of EGU point source SO2 emissions, estimates that
represent EPA's latest view of the implementation of the Clean Air Interstate Rule and the
Mercury and Air Toxics Standards, and comparing them to EPA's FIP TSD emission estimates,
Alpine noted that EPA is now projecting annual SO2 emissions from EGUs to be significantly
lower than they did before. In fact, SO2 from the EGU sector is projected to be 205,000 tons
lower, or an approximate 60% lower value than what EPA cites for these sources in the FIP
TSD.
Alpine stated, since the EPA source apportionment modeling uses the dated, higher EGU
emission estimates and projection year inventories, we have a high level of confidence that EPA
is overestimating the SO2 emissions predicted regionally and therefore underestimating the
improvements in visibility at the identified Class I areas.
Alpine concluded, when the combination of current EPA predicted EGU and non-EGU point and
other nonpoint emission reductions are taken into account, both within Texas and in the
surrounding domain, it is reasonable to expect that modeled visibility improvement will be
achieved, consistent with the observational record.
NRG Footnotes:
4 Report of Gregory M. Stella, Alpine Geophysics LLC, Professional Review of Texas Regional
Haze FIP {April 17, 2015) {"Alpine Report"), at 22-23, 26-27.
5 See Sierra Club v. EPA, 671 F.3d 955 (9th Cir. 2012) (rejecting an EPA SIP approval that
relied on an emissions projection that had been superseded in the time between the state's
submission and EPA's action).
Response: As discussed in our FIP TSD (A-15 - A-16), we did start with the CENRAP 2018
Emission Inventory and made some adjustments based on review of information that had
changed for specific units/facilities. These included:
• Updated emissions to 8 facilities and added one new facility:
• One new unit at Sommers/Deely/Spruce power plant site
• Two new units at Sandow 5 Generating Plant (new plant)
• Three new units at WA Parish Station carried over from the 2002 CENRAP inventory
and emission changes to one existing unit
• Emission changes at North Texas Cement (Ash grove) to reflect shutting down two units
and rebuilding the third unit
• Emission changes to reflect recently installed controls or improvements in control
efficiencies on power plants at Sommers/Deely/Spruce, Big Brown, Gibbons Creek,
Sandown Steam Electric Station, Monticello Steam Electric Station, and Fayette Power
Project
As discussed in our FIP TSD (A-15-A-49), the CENRAP modeling was based on an IPM
(Integrated Planning Model) that estimated EGU future emissions in 2018 including reductions
for CAIR across the Eastern half of the United States. This analysis was conducted in 2006 and
projected that Texas would actually be a purchaser of SO2 credits, and not as much high level
578
-------
controls would be placed on Texas EGU sources. Given the length of time between 2006 when
the IPM analysis was conducted, and 2013 when we were conducting this analysis, we had some
concern that projections could be off for the EGUs in Texas. Information available also
indicated that SO2 credits were much cheaper than originally projected, therefore more credits
may have been used in lieu of emission reductions. We also weighed the technique that Texas
has used in estimating emissions from EGUs for future years (including 2018) in ozone
attainment demonstration SIPs in DFW and HGB470. For these photochemical modeling
analyses with CAMx they have relied upon the recent CEM data that is also included in
CAMD's databases in conjunction with information on recently permitted EGUs for estimating
the emissions to model for EGUs in Texas in 2018 as these emission levels are near CAIR Phase
II control levels.
We did consider updating these emissions for the Cross State Air Pollution Rule (CSAPR) or the
Mercury and Air Toxics Standards (MATS), but based on recent information and recent actual
emissions from CEMS we were uncertain that any significant additional reductions would be
expected from Texas EGU sources in the next couple of years. Also, based on recent comments
from the TCEQ, it was also unclear if any further SO2 or NOx reductions would occur due to
these rules even if all litigation was resolved. The TCEQ had provided extensive comments on
recent emission inventory indicating that further significant reductions from current emission
levels for SO2 were not expected due to CSAPR or MATS.471 We thought it was reasonable to
continue to rely upon the initial CENRAP 2018 modeling inventory initially and update the
information that we were more certain about as discussed above. We utilized 2009-2013 CEM
data for EGUs in evaluation and selection of updated High and Low controlled emission levels to
model. Comparison of recent CEM data with CAIR projections indicated that the overall Texas
emissions from EGUs were on track to meet the projected CAIR emissions level (which included
Texas purchasing -125,000 tons of emission credits from other states) without further substantial
reductions. However, comparison of 2018 CENRAP projected emissions to recent actual
emissions for specific facilities identified by us for additional analysis showed that a number of
these facilities have recent actual emissions that are much higher than CENRAP 2018 modeled
emissions.472 For the ENVIRON modeling we did not increase emissions for existing sources
based on recent actuals but we did lower emissions as described above to account for
information on recently installed controls or improvements in control efficiencies.
470 HGB 1997 8-Hour Ozone standard attainment demonstration approved by EPA in 2013, see TSD materials for
2010 "Appendix B Emission Modeling for the HGB Attainment Demonstrtion SIP Revision for the 1997 Eight-
Hour Ozone Standard" on page B-78, "09017SIP_ado_Appendix_B.pdf'.; DFW 1997 8-Hour Ozone standard
attainment demonstration submitted to EPA, see TSD Appendix B: Emission Modeling for the DFW Attainment
Demonstration SIP Revision for the 1997 Eight-Hour Ozone Standard, Page B-39, "AppB_EI_ado.pdf'; DFW 2008
8-Hour Ozone standard attainment demonstration proposed for adoption Dec. 10, 2014 and posted October 2014,
see TSD materials "Appendix B Emissions Modeling for the Dallas-Fort Worth Attainment Demonstration State
Implementation Plan Revision for the 2008 Eight-Hour Ozone Standard" Starting Page 40.,DFWAD_SIP_Appendix
B.pdf
471 TCEQ comment letter to EPA on draft modeling platform dated June 24, 2014. '2018 EMP signed.pdf'
472 See Table A.4-2 of the FIP TSD for a comparison of recent actual emissions to CENRAP 2018 projected
emission levels.
579
-------
At the time we also noted that TCEQ in recent ozone attainment modeling had also used recent
CEM data to represent expected emissions levels from Texas EGUs for future year of 2018 in
recent Houston and DFW area modeling.
We note that Texas had recently submitted comments to us (few months before our proposal) on
a more recent EPA IPM projection that was going to be part of a new modeling platform for
national rule making and in these comments and comments from several EGU owners in Texas,
the assertion was that no significant amount of additional SO2 controls are expected due to
compliance with MATS. 473. The comments also pointed out that as some of our cursory
research had also indicated that no large SO2 control projects were planned at most of the sources
we were evaluating. Therefore, based on Texas' recent comments and other information, we
concluded considerable uncertainty exists as to whether any further reductions of SO2 will occur
beyond current emission levels as a result of compliance with MATS or CSAPR. Overall this
information supported us looking at recent actual emissions to represent future emission levels in
2018 in our evaluations. We did fully contemplate trying to update EGU emissions in Texas, but
based on a full analysis of all the data, we did not use recent national IPM runs discussed by the
commenters because ,as Texas was indicating, EGU emissions in Texas are not anticipated to be
consistent with those emission inventories and are not expected to be much reduced from current
emission levels. In our visibility analysis, we estimated visibility impacts and benefits from
those individual facilities shown to have the largest visibility impacts using both the 2018
CENRAP projected levels and recent actual emission levels that we have determined to be more
representative of anticipated emissions in 2018 for these sources. As demonstrated in our FIP
TSD and proposal, the identified facilities have significant impacts on visibility conditions. Our
technical record makes it equally plain that the required controls reduce impacts from these
sources and result in meaningful visibility benefits towards the goal of natural visibility
conditions. We discuss comments concerning consideration of recent monitored visibility data
elsewhere.
We state in our proposal that observed improvement from the baseline conditions is partially the
result of reductions in the impacts from SO2 emissions and that emissions from non-EGU Texas
point sources are lower than have been projected in the modeling.474 We note that
additional reductions are still needed to meet or exceed the URP goals for 2018 as calculated by
us. As discussed elsewhere, emission reductions at some of the sources that impact visibility the
most are still above the emission level projected in the model and we have demonstrated that
cost-effective controls are available at many of these sources that will result in meaningful
visibility benefits towards the goal of natural visibility conditions. We also note that not many
non-EGU emission reductions are due to enforceable emission limits or installation of controls
and that economic/business fluctuations could result in increases that could negatively impact
visibility. We discuss our consideration of recent visibility modeling elsewhere in this
document.
473 Texas comments on Draft IPM modeling conducted by EPA for potential national rule making platformprovided
on June 26, 2014. In this docket materials as "TCEQ comment letter to EPA on draft modeling platform dated June
24, 2014. '2018 EMP signed.pdf'
474 7 9 FR 74843. We discuss the impact of recent meteorological conditions on monitored visibility conditions
elsewhere in this section of the document.
580
-------
15.d Monitor Data and Model Performance
Comment: EPA's proposal fails to account for real-world data and fails to validate the modeling
it relies on [Luminant (0061) p. 139]
Luminant stated that EPA's proposed disapproval and FIP are also arbitrary and capricious
because they disregard real-world data and are based on modeling results that are not properly
validated. Real-world data collected by EPA's IMPROVE monitors (discussed in Section V)
show that the three Class I areas of interest have already achieved the visibility target EPA sets
for them for 2018 without the additional controls that EPA is proposing. As these data show,
actual conditions have already improved beyond the point that EPA considers reasonable for
2018, and they are projected to meet even the URP for 2018.851 Thus, as explained in Section V,
there is no basis or authority for requiring any additional controls based on estimated impacts to
those areas.
Luminant stated that EPA acts contrary to law and in an arbitrary and capricious manner by
ignoring these real-world data. Despite recognizing that "[mjodel performance at IMPROVE
monitors is of highest importance, because these monitors are sited to be representative of the
visibility conditions impacting each Class 1 Area,"852 EPA refuses to incorporate that data into
its analysis here or to use it to validate its modeling results. This makes no sense and points to
the fundamental flaws throughout EPA's analysis.
Luminant noted that EPA has failed here in its obligation to consider and reconcile the stark
inconsistencies between what its models predicted and reality.853 Inexplicably, and in deviation
from EPA guidance,854 ENVIRON did not perform a model performance evaluation to validate
any of the results it provided to EPA, even though ENVIRON noted significant deviations in its
modeling as compared to real-world conditions.855 Indeed, at Guadalupe Mountains, as to the
most dominant constituents of visibility impairment (soil and coarse mass), ENVIRON found
that the model performance was "suspect" and warned that "care should be taken in interpreting
these modeling results."856 These findings by ENVIRON call into question the foundational
assumption in EPA's proposed FIP—that SO2 impacts from a handful of Texas sources are the
primary contributor to visibility impacts at the three Class I areas—and renders arbitrary and
unsupported EPA's decision to "limit[] our analyses to the consideration of SO2 controls for
these EGU sources."857
According to Luminant, even in the face of these contrary data, EPA did not request ENVIRON
to perform a full model performance evaluation, or do one itself. Instead, EPA simply
"compared" the results of ENVIRON's new modeling to ENVIRON's prior modeling for
CENRAP and concluded that because the results "overall were very similar" (in EPA's
subjective view) the new model results were "validated]."858 But, as discussed in our
Background discussion of EPA's Step 2, EPA's new modeling deviated significantly from
CENRAP's modeling. For example, CENRAP's modeling did not involve Plume in Grid (PiG)
modeling targeted at these selected Texas sources, as did EPA's new modeling that forms the
core of its proposal. There are many other deviations.859 EPA's "comparison" to CENRAP's
581
-------
modeling is thus of no value at all and is certainly not a proper or reliable form of validation.
EPA's novel method of "validation" is not defensible, and EPA's failure to properly validate its
results undercuts EPA's entire proposal. Moreover, EPA's ultimate decision about what units to
regulate was not based on ENVIRON's modeling at all, but instead on EPA's alterations of it,
which EPA does not claim were validated in any manner.
Luminant asserted, if EPA is going to impose emission-reduction obligations based on complex
modeling and "but-for" projections, it must use available means to verify those projections. EPA
may use "predictive models" only where it "provides a complete analytic defense" and
"addresse[s] what appear to be stark disparities between its projections and real world
observations."860 Here, EPA arbitrarily disregards available air-quality data that shows visibility
targets for 2018 have already been achieved or are on track to be achieved without additional
controls and ignores anomalies brought to its attention by ENVIRON. Unexplained
contradictions between EPA's predictions and real-world observations like these undermine the
accuracy of EPA's air-quality projections and render the emission controls based on those
projections arbitrary and unlawful.861
Luminant stated, in this very rulemaking, EPA recognizes the need to reconcile modeled results
with real world data and further recognizes that Texas did this for the modeling it relied on.862
And EPA recognizes that IMPROVE monitoring data is "[t]he starting point" for an assessment
of visibility conditions.863 Yet, despite recognizing this obligation, EPA failed to perform a
model performance evaluation prior to using its new CAMx modeling to evaluate "the 38
facilities in Texas" (identified through EPA's Q/D analysis) for additional controls in the
proposed FIP.864 Under EPA guidance, "[t]he results of a model performance evaluation should
be considered prior to using modeling to support an attainment demonstration or regional haze
assessment."865 EPA claims to have "validated]" its new modeling with reference to CENRAP
modeling,866 but that is not an accepted validation technique per EPA guidance. In fact, it is bad
science. Proper model validation should show, not just that the answer is similar to the answer
from another model, but that the model predicted the right answer for the right reason.867 Clearly,
ENVIRON's CAMx modeling and EPA's alterations of it—which are the core of its proposed
FIP—over-predicted the visibility impacts from these 38 facilities at the three Class I areas (as
research by its own contractor clearly demonstrates), as shown by comparison to current real-
world monitored data, and thus EPA's results cannot be used as a reliable basis for imposing
additional controls. If EPA had performed a thorough review of its modeling analysis and input
data, it would have found several errors in the assumed stack parameters for existing sources,
including the assumption that units would have the exact same stack temperature and velocity
before and after the upgrades and retrofits that EPA is proposing. These types of errors are yet
further reasons EPA should have validated its modeling.
Luminant stated that there is no basis for EPA to ignore this real-world data, which is more
reflective of actual conditions in 2018 than the model EPA has constructed. EPA concedes that
reductions in SO2 emissions are occurring in Texas and have improved visibility in the relevant
Class I areas.868 EPA also concedes that emissions will not worsen between now and 2018.869
And EPA cannot contest that CSAPR's SO2 budgets, which are now in effect and binding, are
lower than CAIR's budgets and impose more stringent caps on out-of-state trading. EPA's
582
-------
decision to ignore CSAPR and other existing regulations for this rulemaking is not only
inexplicable, it is one of the obvious flaws with EPA's predictions.
According to Luminant, indeed, proper validation of EPA's new modeling, against real-world
data, was particularly appropriate here given EPA's own prior studies showing that CAMx
overpredicts visibility impacts at the distances involved here and EPA's own prior regional haze
modeling for these very Class I areas which, as discussed in Section VIII, shows that these areas
meet EPA's own proposed RPGs. It is clear that EPA's new modeling for Texas—in which EPA
refuses to account for CSAPR and other existing programs that will control SO2 emissions
through 2018—significantly overpredicts visibility impacts in 2018. EPA has failed to reconcile
its current modeling with this prior modeling or explain the stark discrepancies. Indeed, EPA's
air quality modeling predicts that air quality will worsen from present conditions after the
implementation of nearly $2 billion of additional SO2 controls. This simply cannot be correct.
There are fundamental flaws in EPA's modeling that proper validation would surely reveal.
Footnotes:
851 See AECOM Report at 2-5 to 2-10.
852 CENRAP Modeling TSD at 34.
853 Cf. NRDC v. Jackson, 650 F.3d 662, 665 (7th Cir. 2011) ("The way to test" predictive models
is to "compare [the] projection against real outcomes." "An agency that clings to predictions
rather than performing readily available tests may run into trouble." (citing Bechtel v. FCC, 10
F.3d 875 (D C. Cir. 1993))).
854 EPA, Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of
Air Quality Goals for Ozone, PM2.5, and Regional Haze 190 (Apr. 2007), available at
http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rhguidance.pdf.
855 ENVIRON Feb. 2013 Memo at 8 ("This project did not include a complete statistical
performance evaluation."); id. at 12 (concluding that its "model results show variation of
dominant light-extinction component"); id. at 19 ("Performance for Soil and especially CM is
suspect and care should be taken in interpreting these modeling results.").
856 Id. at 18-19.
857 FIP TSD at 3.
858 FIP TSD at A-15 to A-16.
859 ENVIRON Memo at 6.
860 Appalachian Power Co. v. EPA, 249 F.3d 1032, 1054 (D.C. Cir. 2001).
861 See Motor Vehicle Mfrs. Ass'n of the U.S. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43
(1983) (agency rule is "arbitrary and capricious" where the agency "failed to consider an
important aspect of the problem" or "offered an explanation for its decision that runs counter to
the evidence before the agency").
862 TX SIP TSD 2014 at 55 ("Model performance evaluation is performed by comparing output
from model simulations with ambient air quality data for the same time period to determine
whether the model's performance is sufficiently accurate to justify using the model for
simulating future conditions.").
863 CENRAP Modeling TSD at 10. See also id. at 34 ("Model performance at IMPROVE
monitors is of highest importance, because these monitors are sited to be representative of the
visibility conditions impacting each Class I Area.").
583
-------
864 FIP TSD at A-15. Because EPA used the ENVIRON modeling at various steps in its FIP
methodology (see, e.g., id. at A-50 ("we then scaled the ENVIRON modeling")), the error
pervades the entire proposal.
865 EPA, Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of
Air Quality Goals for Ozone, PM2.5, and Regional Haze 190 (Apr. 2007), available at
http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rhguidance.pdf.
866 FIP TSD at A-16.
867 See Paul D. Sampson et al., Operational Evaluation of Air Quality Models 3 (1999), available
at: http://www.nrcse.washington.edu/pdf/trsl8_aqmodels.pdf ("The point of model evaluation is
to establish the credibility of a model for use in decision-making. Most model applications
require that the model extrapolate well beyond current precursor and primary emission
conditions that could exist in any model evaluation data set. This is particularly true for issues
that span the urban to global scales, such as oxidants, acidic deposition and visibility. Thus, a
model evaluation needs to test the science in the models. Testing the science means looking for
the "right" kind of answer (right answer for right reason and wrong answer for right reason),
rather than simply looking for good comparisons of final outcome." (quoting Robin L. Dennis et
al., Correcting RADM's Sulfate Underprediction: Discovery and Correction of Model Errors and
Testing the Corrections Through Comparisons Against Field Data, 27 Atmos Environ 975, 997
(1993))).
868 79 Fed. Reg. at 74,843, 74,870.
869 FIP TSD at A-45 ("Overall this information supports looking at recent actual emissions to
represent future emission levels in 2018.").
Response: We did not disregard the recent IMPROVE monitoring data. In fact, the recent
ambient monitoring data at the IMPROVE sites in the three Class I areas (2011-2013) are
influenced by meteorology that has lower than normal transport of pollution from sources in
Texas when compared to the base period on which projections are based (2000-2004) and to the
30-year meteorology analysis of transport to the three Class I areas (1984-2013). Thus,
examining the 2011-2013 time period overstates the progress that can be expected over long
term. In response to comments and information provided by commenters, we conducted further
analysis to appropriately evaluate whether the base period was suited for projections to 2018 and
also an analysis of how the meteorology accompanying the more recent monitoring data for
2011-2013 compared to normal meteorology conditions. We further note that 2014 also was not
quite a normal year475 and likely similarly biased low for visibility impacts at the Class I areas,
but even so monitoring data in 2014 did increase compared to the 2011-13 data. Overall, we
conclude that our evaluation of 2002 and 2018 levels and the controls needed for reasonable
progress are based on representative periods and that recent monitoring trends are not as
representative and not expected to continue if meteorology is more in line with 30-year
climatological and transport norms. See our analysis in response to HYSPLIT analyses and other
information provided about recent and baseline meteorology elsewhere in this section. We note
that recent monitoring data is lower but as discussed in response to previous comments, there are
not large reductions expected between now and 2018. There is always variability in meteorology
that impacts the monitoring trends independent of emission trends. As we have explained in our
475 Some preliminary analyses of meteorology and pollution levels in 2014 indicated a higher frequency of cold
fronts during the summer of 2014 that led to cleaner air from the arctic mixing with the air in the region and resulted
in lower pollution build-up and transport of pollution to Class I areas in Oklahoma and Texas.
584
-------
final action and other responses to comment, being on or near the glide path does not provide
safe harbor. As demonstrated in our FIP TSD and proposal, the identified facilities have
significant impacts on visibility conditions. Our technical record makes it equally plain that the
required controls reduce impacts from the sources that have the largest impact and result in
meaningful visibility benefits towards the goal of natural visibility conditions.
We note that photochemical grid model results are evaluated with Relative Response Factors
(RRFs) to help remove potential bias concerns. While no model is free from bias issues,
previous evaluations of the CENRAP databases we used for our analyses have been evaluated by
us and the CENRAP CAMx model performance was considered adequate because the modeled
outputs compared well to past measured conditions.
Our FIP is based on impacts at all three Class I areas and the model does not perform perfectly
for some species at Guadalupe Mountains, however these are not the pollutants that we are
proposing to reduce to yield improvements in visibility impairment at GUMO, BIBE, and
WIMO. Monitoring data at Guadalupe Mountains does indicate that sulfates also contribute to
higher visibility impairment days, so SO2 reductions would be expected to improve visibility at
GUMO. As discussed in responses elsewhere in this section, there was an extensive model
performance conducted on the CENRAP modeling and the changes we made were relatively
minor and evaluations of speciated light extinction for the 20% best and worst days in
comparison to the CENRAP model values did not show substantial differences for WIMO,
GUMO, and BIBE. Many of the changes we made, such as using PiG and PSAT and a 12 km
fine grid, were all things that TCEQ previously did to the CENRAP databases and protocol
without additional model performance analysis. We approved the protocol that they did not need
to conduct a model performance analysis. From discussions with TCEQ and others at the time,
the changes were not expected to result in substantial differences. We went a step beyond what
TCEQ had previously done by doing some evaluation of the model results to ensure that things
had not substantially changed. Model performance was deemed adequate by CENRAP and
TCEQ and other CENRAP members and as also discussed in other responses, the information
available did not indicate a full performance analysis of 2002 monitor data to modeled data was
necessary. We respond to the comments concerning "adjustments" to CAMx results in separate
responses to comments where we discuss modeling. We do not agree that the inert tracer studies
(CAPTEX and ETEX) indicate that CAMx is overestimating impacts from point sources in our
analysis. We have detailed responses in other parts of this section that address these comments.
The errors in the stack data the commenter raises would not be expected to substantially change
the modeling results and we discuss the impact these stack prameters have on our estimated
visibility benefits of controls elsewhere in this modeling section. See responses above in regard
to recent actual SO2 emissions and how they were factored into our 2018 emissions.
15.e Camx Not Approved Model for LRT
General Summary: We received several comments that CAMx is not the approved model in 40
CFR 51 Appendix W for modeling long-range transport (LRT) for visibility. We received
comments that our selection of the CAMx model rather than CALPUFF is inappropriate and
unjustified and that CALPUFF is the proper LRT model. The commenters state that EPA has
not justified the use of CAMx to model visibility impacts from individual sources and at large
585
-------
distances, and EPA's use of CAMx here is outside of the model's capabilities. Furthermore, the
commenters assert EPAs' concerns of using CALPUFF are not clear and concerns of model
overprediction are also present in CAMX and therefore do not justify the use of CAMx.
Comment: EPA's use of CAMx modeling to review the Texas Regional Haze SIP [AECOM
(0061 and 0075) p. 6-1]
AECOM stated that to evaluate the Texas SIP and develop its proposed FIP, EPA utilized the
Comprehensive Air Quality Model with Extensions (CAMx) to model the visibility impacts to
the three Class I areas from emissions from a small number of Texas point sources, and also to
model the visibility benefits from the SO2 emission controls it would impose on those sources.
(EPA FIP Technical Support Document at A-15-A-16) EPA contracted with ENVIRON to
perform this modeling based on direction from EPA.
AECOM stated that CAMx is a photochemical grid model (PGM), which is a computational
model that is designed to represent the complex physical and chemical processes of multiple
pollutants in the atmosphere. PGM models are capable of estimating the transport, chemistry,
and removal, among other processes, for various gases and particulate matter (PM). Furthermore,
they can provide an estimate of the concentrations of the chemical components that are part of
PM.
For regional air quality modeling, AECOM stated that this capability is important in that these
models can consider the impacts from all possible pollution sources (both natural and
anthropogenic) as long as they are included in the emissions inventory. Once particulate matter
concentrations are known, atmospheric light extinction (bext) can be calculated with the help of
the IMPROVE equation and this can be expressed in deciviews (dv) with the following
definition:
dv = 10 * In (bext/10)
AECOM stated that CAMx is capable of providing the contributions to total PM from individual
or groups of sources through the Particulate Source Apportionment Technology (PSAT). PSAT
'tags' and tracks the species of interest from the emissions' location until it reaches a receptor of
interest (in this case specific Class I areas). It is typical in this type of regional assessment to start
with a one-year 'base case' simulation that both represents current conditions and helps to
identify the biases and shortcomings of the model through a model performance evaluation.
Once the model is deemed suitable for its intended use, one or multiple scenarios representing
different future emission inventories (but the same meteorological conditions as the base case
simulation) are simulated. Once model estimates are known, a subsequent analysis of results is
performed. For regional haze assessment, an analysis including the identification of the 20% best
and 20% worst visibility days in the target Class I areas is typically included.
Although there is significant precedent for the use of PGMs for development of SIPs
(particularly with regard to ozone) and other types of analyses, AECOM stated that CAMx is not
an EPA-approved model for long-range transport in the regional haze context.95 The discussion
586
-------
below identifies the known uncertainties and limitations in the use of these models for a regional
haze analysis.
Footnotes:
95 40 C.F.R. Part 51, Appendix W, § 6.2.3. ("If LRT [long-range transport] is determined to be
important, then refined estimates utilizing the CALPUFF modeling system should be obtained.")
Summary: AECOM stated that CAMx is not an EPA-approved model for long-range transport
in the regional haze context. (More specific comments on model limitations and uncertainties
follow and are addressed in responses to other comments in the modeling responses).
Response: A number of commenters made this same comment that EPA erred in using CAMx,
and that App. W, § 6.2.3 indicates we should have used the CALPUFF modeling system for our
RP analysis. We note that this is a misinterpretation of App. W, § 6.2.3 and CALPUFF's
preferred regulatory status. When EPA promulgated CALPUFF as a preferred model in 2003, we
only approved it for LRT for direct emissions and we did not provide preferred status to
CALPUFF for use of the chemistry module included in CALPUFF that can generate values for
secondary formed pollutants including visibility impairment.476
The 2005 BART Guidelines recommended the use of CALPUFF for assessing visibility
(secondary chemical impacts) but noted that CALPUFF's chemistry was fairly simple and the
visibility results from CALPUFF could be used as one of the five factors in a BART evaluation
and the impacts should be utilized somewhat in a relative sense because CALPUFF was not
explicitly approved for full chemistry calculations.477 The BART guidelines also provided the
option to potentially use PGMs (such as CAMx) in the future if modeling tools available were
appropriate and EPA approved of the technical approaches and how the model would be
utilized.478 The specific regulatory status of CALPUFF in regards to chemistry and secondary
476 FR Vol. 68, No. 72, April 15, 2003, page 18447 and 18442 respectively; "Today's rule recommends a new
modeling system for calculating PSD increment consumption—CALPUFF— that increases efficiency and
accuracy." and "CALPUFF's treatment of chemical transformations, which affect AQRVs. Specific concern was
expressed about the sulfate and aqueous phase chemistry algorithms. As chronicled on the FLAG Web site (above),
these procedures and criteria have been published and received review and comment. However, today's rule
addresses the suitability of CALPUFF for PSD increment consumption and for complex wind situations (with case-
by- case approval), not AQRV analyses.
477 70 FR 39123, 39124. "We understand the concerns of commenters that the chemistry modules of the CALPUFF
model are less advanced than some of the more recent atmospheric chemistry simulations. To date, no other
modeling applications with updated chemistry have been approved by EPA to estimate single source
pollutant concentrations from long range transport." and in discussion of using other models
with more advanced chemistry it continues, "A discussion of the use of alternative models is given
in the Guideline on Air Quality in appendix W, section 3.2."
478 70 FR 39123, 39124. "The use of other models and techniques to estimate if a source causes or contributes to
visibility impairment may be considered by the State, and the BART guidelines preserve a State's ability to use
other models. Regional scale photochemical grid models may have merit, but such models have been designed to
assess cumulative impacts, not impacts from individual sources. Such models are very resource intensive and time
consuming relative to CALPUFF, but States may consider their use for SIP development in the future as they are
adapted and demonstrated to be appropriate for single source applications."
587
-------
formation was also covered in responses to comments previously in a final action on New
Mexico's BART FIP.479
Therefore, there is no specific guideline model approved for full chemistry and the use of
CALPUFF to calculate visibility impairment using CALPUFF's chemistry mechanism is not
considered an application of a model in preferred mode but is allowed to be utilized in a relative
quantitative sense in evaluating for applicability to BART and as part of a BART five-factor
analysis. As we note elsewhere, PGMs such as CAMx have been used in a number of SIP
actions and approved by EPA, and PGMs have advanced state-of-the-science chemistry that has
undergone peer review. CAMx is one of the PGMs that has been approved for many SIP
analyses and was one of the models approved to be used by CENRAP and its members
(including Texas). Further, CAMx with the improvements in PiG with chemistry and PSAT that
TCEQ and ENVIRON developed in 2005-2007 meets the requirements of App. W 3.2.2 and
3.3480 in determining acceptability of a model as an alternative model for assessing visibility
impacts. TCEQ previously requested to use CAMx for BART screening in 2005-2006 and cited
concerns with CALPUFF's chemistry, model performance and the distances of many of the
sources in Texas from Class I areas as being much further than the typical maximum range that
CALPUFF was used (300 km). EPA worked with Texas and conferred with other EPA experts
and FLM representatives about using CAMx instead of CALPUFF in this situation. Texas
worked with the CAMx model developer ENVIRON to modify the model so the PiG tool could
be used with full chemistry in the plume and also PSAT within the plume as well as in the
modeling grid. With these refinements to assess single sources and small groups of sources, EPA
approved the TCEQ's request to use CAMx for BART screening and Texas used it for a large
number of sources, grouped in small groups and in some cases the sources were individually
modeled with CAMx by TCEQ. Six sources in Texas also utilized TCEQ's CAMx BART
screening modeling guidance and conducted single source visibility impact analysis.481 Therefore
there is specific precedent with the CENRAP databases and using CAMx with the appropriate
tools to assess visibility impacts for individual sources and small groups of sources in Texas, and
we note that for some of the BART screening modeling, it was only one or a few sources at a
facility that were specifically modeled and many facilities' emissions were lower than the single
units we evaluated. Arguably, if we applied the rationale that we erred, one could argue that if
we erred in using CAMx in our analysis, then we also erred in approving the BART screening
modeling evaluations with CAMx performed by Texas that allowed many sources to be
determined to not be applicable to a full BART analysis. We do not agree with the comment that
we erred in using CAMx for our analysis.
Neither the regional haze regulations nor Appendix W require the use of a specific preferred
model for photochemical grid modeling for visibility (regional haze), but we have approved the
use of regional scale photochemical grid models such as REMSAD and CMAQ.482 CAMx is
another regional scale photochemical grid model that was utilized by the RPOs and states and
479 EPA's Final FIP in New Mexico FR Vol. 76, No. 162, August 22, 2011, p. 52431-52434
480 FR Vol. 70, No. 216 68232.
481 TCEQ's Modeling Guidance- Guidance for the Application of the CAMx Hybrid Photochemical Model to Assess
Visibility Impacts of Texas BART sources at Class I Areas"
482 40 CFR Part 51 Appendix W, Section 6.2.1 (e&f).
588
-------
approved by EPA. CENRAP conducted its final CAMx source apportionment modeling for the
regional haze analysis to be utilized in consultations of its nine state members in development of
their SIPs. We approved most of these SIPs that included modeling analyses using CAMx and
thus CAMx is clearly acceptable for evaluating long range transport for visibility.483 As we have
stated previously, Texas used CAMx in its reasonable progress analysis to extrapolate the
visibility benefits from all potential controls it identified. Again, as as detailed elsewhere,
although Texas used the CENRAP CAMx source apportionment modeling as the basis for its
analysis, we found such analysis was flawed for RP/LTS.
As discussed above, Texas used CAMx to screen small groups of sources and individual sources
as part of its BART screening and we approved that approach in 2006/7,484 based on modeling
enhancements that Texas contracted to be developed to assist in assessing single point source
visibility impacts on visibility at Class I areas. The visibility impact analysis we performed with
CAMx is commensurate with the work originally done by Texas in 2006/7 for its BART
screening, and many of the sources that Texas modeled with CAMx had emissions of RH
pollutants smaller than individual units that we evaluated in the unit specific analysis we
conducted. Overall, Appendix W gives us discretionary authority in the selection of what
models to use for visibility assessments with modeling systems, and models such as CALPUFF,
CMAQ, REMSAD, and CAMx have all been used for that purpose. As we have stated above,
CALPUFF is not a preferred model when it is uses its chemistry module for estimating visibility
impairment. In this specific situation, we determined that CAMx had the best scientific modeling
approaches and tools and was best suited for the complex analysis that we needed to perform.
15.f National Academy of Science
General Summary: Commenters state that the National Academy of Science advised EPA that
there are considerable uncertainties in ascertaining a precise relationship between individual
sources and regional haze and that such an undertaking would be time-consuming, expensive and
"doomed to failure." Based on those advisements, the commenters draw the conclusion that
EPA's analysis of individual sources is therefore unprecedented and "doomed to failure."
Comment: EPA's modeling choices are arbitrary and capricious and do not support its
conclusions [Luminant (0061) p. 115]
Luminant stated that EPA's unprecedented analysis of individual Texas sources is arbitrary and
capricious and was ill-conceived from its inception. EPA was advised by the National Academy
of Sciences that a "program that focuses solely on determining the contribution of individual
emission sources to visibility impairment is doomed to failure."717 As the NAS explained, "it
would be an extremely time-consuming and expensive undertaking to try to determine, one
source at a time, the percent contribution of each source to haze. . . . [T]here are (and will
probably continue to be) considerable uncertainties in ascertaining a precise relationship between
483 EPA - 454/B-07-002, April 2007; "Guidance on the Use of Models and Other Analyses for Demonstrating
Attainment of Air Quality Goals for Ozone, PM2 5, and Regional Haze.
484 EPA, TCEQ, and FLM representatives verbally approved the approach in 2006 and in email exchange with
TCEQ representatives in February 2007 (see email from Erik Snyder (EPA) to Greg Nudd of TCEQ Feb. 13, 2007
and response email from Greg Nudd to Erik Snyder Feb. 15, 2007).
589
-------
individual sources and the spatial pattern of regional haze."718 Despite this warning, EPA
embarked on its unprecedented effort for Texas. But the results of its efforts only prove the truth
of the NAS's prior findings.
Footnotes:
717 Nat'l Acad, of Scis., Protecting Visibility in National Parks and Wilderness Areas 7 (1993).
718 Id.
Response: The citation to the NAS report is not a fair representation of the current modeling
world. We disagree with the commenters for the following reasons:
• The NAS report was from 1993. There have been considerable advances in model
development and computing power in the 20+ years since that report.
• Computers are also much cheaper and there is a larger population of modelers.
• Modeling is no longer as time consuming or expensive.
• Development of PS AT technology and Plume in Grid with chemistry allows for just this
sort of analysis without requiring the "one at a time" approach envisioned by NAS.
• Model performance has greatly improved and computing power allows for higher grid
resolution and more complicated chemistry and transport than in the past when NAS
issued its report.
Comment: [Luminant (0061) p. 108] Luminant noted that the EPA itself has explained that its
regional haze regulations do not contemplate an individual source / individual control approach
to reasonable progress. EPA, in issuing the rules it claims to be applying here, "avoided inclusion
of any approach in the regional haze rule that required the assessment of the visibility
improvement attributed to an individual source . . . ,"680 EPA's decision in this regard was driven
by an evaluation of The National Academy of Sciences ("NAS"), following the 1990 Clean Air
Act amendments, which found that a "program that focuses solely on determining the
contribution of individual emission sources to visibility impairment is doomed to failure."681 As
the NAS explained:
Because haze is caused by the combined effects of the emissions of many sources, it would be an
extremely time-consuming and expensive undertaking to try to determine, one source at a time,
the percent contribution of each source to haze. For instance, the efforts to trace the contribution
of the Navajo Generating Station to haze in the Grand Canyon National Park took several years
and cost millions of dollars without leading to quantitatively definitive answers. Moreover, there
are (and will probably continue to be) considerable uncertainties in ascertaining a precise
relationship between individual sources and the spatial pattern of regional haze.682
Even though EPA repudiated such an approach in issuing its regulations, EPA embarks on such a
doomed-to-fail effort for Texas, pretending that it can quantify contribution from individual
sources to areas hundreds of miles away, down to the hundredth of a deciview. EPA's approach
for doing so has no support, and there is no precedent for such an undertaking—and certainly no
example of it being done successfully. Indeed, EPA cites no examples of prior SIP reviews in
which it has employed such an "individual control" analysis or any of the thresholds or metrics
that EPA employs in evaluating Texas' SIP and developing its FIP. EPA may not lawfully
590
-------
engage in such an undertaking for Texas, regardless of the merits that EPA may now see in such
an approach.
Footnotes:
680 Am. Corn Growers Ass'n, 291 F.3d at 21 (Garland, J., dissenting) (emphasis added); see also
Response to Petitions at 16 (emphasis added).
681 Nat'l Acad, of Scis., Protecting Visibility in National Parks and Wilderness Areas 7 (1993).
682 Id.
Response: Some of these comments are responded to elsewhere in this document. As noted
earlier, the NAS report is merely a report based upon facts in existence before its release in 1993.
Much of it now is out of date. There is no basis for relying upon it to preclude evaluation of
single facilities with modeling for visibility impacts. The CAA supports the use of updated
science and tools for conducting technical analysis for SIP or FIP development. It is expected
that many of the components of the Regional Haze Rule would be addressed by or rely on
current-day sound science and computing tools, and this will also be the case for furture
implementation of the RHR, including determinations of RP.
EPA promulgated the Regional Haze Rules in 1999 and at that time there were very limited tools
for treating individual sources. In addition, the refined PiG with chemistry, better plume
dispersion, and source apportionment had not all been developed and included in PGMs. As
discussed in other responses, it was the development of these tools in CAMx and the ability to
use PiG with full chemistry and source apportionment that allowed some of the first CAMx
based modeling to be included in a RH SIP. As discussed in more detail in another response,
TCEQ worked with the CAMx model developer to advance the science in CAMx and integrate
these tools. TCEQ then requested EPA to approve the use of CAMx with the tools to evaluate
visibility impacts at distant Class I areas for BART eligible units at either a single facility or a
small group of facilities to be evaluated and ultimately screen facility/ies out of full BART
applicability. This CAMx modeling ultimately screened out a large number of the over 100
BART eligible facilities in Texas. The NAS report was pertinent at the time in 1993 and most of
the 1990s and early 2000s, but in the mid-2000s the modeling tools took a large step forward in
the ability to assess impacts from single facilities using PGMs. Therefore we disagree with the
commenter that we are on a "doomed to failure" approach in using CAMx to evaluate single
sources for RP. We note that recently Minnesota has also utilized PGMs to assess single facility
impacts at Class I areas in recent SIP efforts.
15.g Appendix W requires Modeling a 3-yr Period
General Summary: We received comments that EPA failed to follow its modeling guidance in
Appendix W that recommends modeling data for 5 years, and no less than 3 years of data for
long-range transport applications.
Comment: [Luminant (0061) p. 117] Luminant noted, even if CAMx were proven to be a
reliable model for the purposes EPA is using it for, EPA's application of CAMx deviates from
standard modeling practice in several regards, making the results unreliable. For example, EPA's
own visibility modeling guidelines call for the use of three years of data. EPA's Guideline on Air
591
-------
Quality Models, found in Appendix W to EPA's regulations, allows the use of one year of data
for a single year only for short-range applications. EPA's recommendation for long-range
applications is the use of five years, and no less than three years, of modeling data.733
Footnotes:
733 40 C.F.R. pt. 51, app. W, § 8.3.1.2.
Comment: EPA's limited CAMx modeling (only one year) contradicts EPA's own guidance for
conducting long-range transport modeling. [AECOM (0061 and 0075) p. 1-6, 6-2]
AECOM stated that the EPA's use of CAMx here deviates from standard modeling protocol that
accounts for the meteorological variability that occurs for the time period being modeled. EPA's
CAMx modeling for Texas used only one year of data (2002). (EPA FIP Technical Support
Document at A-15.) The use of a one-year modeling simulation to represent "typical"
meteorological conditions is problematic, especially for the impact assessment on receptors
located at such large distances from the sources. AECOM asserted that EPA's Appendix W
allows the use of one year of data for a single year only for short-range applications and EPA's
recommendation for long-range applications is the use of at least three years of modeling. (40
C.F.R. Part 51, Appendix W, § 8.3.1.2) EPA asserts in its proposal that it is following the BART
Guidelines in several respects. However, the BART Guidelines require "a period of three or five
years of meteorological data" for modeling to assess visibility impacts.98 EPA provides no
explanation of why a photochemical model like CAMx should not be required to follow the same
approach as modeling for a BART analyses.
Footnotes:
9870 Fed. Reg. 39,107
Response to Luminant and AECOM: Both Luminant and AECOM assert that 40 CFR Part 51,
App. W § 8.3.1.2.including § 8.3.1.2.d apply and EPA should have modeled at least 3 years of
meteorology data. The commenters assert that EPA was performing a long-range transport
analysis and should have used 3 years of meteorology data instead of 1 year. App. W § 8.3 and §
8.3.2.1 more directly address the meteorological requirements for dispersion models for PSD
purposes, including the LRT model CALPUFF. PGMs are very complicated modeling analyses
and very resource intensive and EPA has issued specific modeling guidance documents for PGM
analyses to clarify and provide additional guidance and clarifications on when and how to use the
models. As discussed in response to another comment about the appropriateness of using
CAMx, the BART guidelines allow a PGM to be used (with EPA approval), but the BART
guidelines do not prescribe the meteorological period that is necessary if a PGM is used. EPA's
modeling guidance for PGMs stresses the need to pick representative periods to model that may
be only one year or less of meteorology, but are representative of the baseline period (in this case
2000-2004).485 EPA's general modeling guidance is to utilize multiple years of data for
dispersion models, but EPA's guidance for analyses with PGMs is more complicated and there
has always been a balance between modeling a long period and the available resources and
485 EPA-454/B-07-002, April 2007; 'Guidance on the Use of Models and Other Analyses for Demonstrating
Attainment of Air Quality Goals for Ozone, PM2 5 and Regional Haze'
592
-------
reasonableness in the amount of computation time and other resources needed for a PGM based
analysis. Updated Draft modeling guidance released for comment in December 2014 continue to
reflect that one full year should be modeled for Regional Haze.486
EPA, FLM representatives, CENRAP and TCEQ all agreed to conduct an analysis for LTS and
RP at Class I areas using the CAMx and CMAQ photochemical grid models. In the protocols
that were established and approved for CENRAP states and most other RPOs, EPA and the
RPOs (and their members) agreed to model 2002 and this was deemed acceptable given the
resource intensity nature of a one year annual simulation with photochemical grid modeling.487
The agreement to model only one meteorological year for photochemical grid modeling was
approved by EPA regional offices and EPA's Office of Air Quality Planning and Standards,
which promulgated the RH rules, BART rules, App. W, and issued modeling guidance for
photochemical modeling for RH SIPs. One thing that is important to note is while the
meteorology is only done for one year (2002) and then also modeled with 2018 emissions, the
model RRF values (2018 result/2002 results) are multiplied times the monitoring data
representative of 2000-2004. The use of the 5 years of monitoring data and the RRF approach,
which estimates future modeled values by multiplying the ratio of future model divided by base
model to yield a 2018 value that should be representative of the 5-year average of meteorology in
the future case.488 This analysis technique is not perfect, but it is a way that EPA modeling
guidance attempts to incorporate 5-years of monitored data to make the modeling more
representative of a 5-year period instead of just using 2002 monitoring data that would only
represent the influences of 1-year.
If we applied the commenters' logic and their interpretation of App. W § 8.3.2.1, and required
three years of meteorology, at a minimum there are several other things that would have to be
completed: 1. If 3 years of PGM modeling were conducted, that would require the generation of
two additional years of meteorology data, meteorology performance evaluations and refinements
until an acceptable met model performance was achieved; 2. two more years of year-specific
emission inventories;3. and two more years of RH PGM modeling with model performance
evaluations and adjustments until adequate model performance was achieved. Considering that
national rulemaking that EPA conducts has typically only utilized one year of meteorology for
the baseline period and attainment SIPs and RH SIPs require only one year of meteorology (or
less for ozone), the conclusion that would be reached by following the comment would be totally
unreasonable in comparison to all other PGM modeling conducted for regulatory purposes.
Given that following the commenter's logic and interpretation would result in a workload 2 times
larger than the original RH SIP development just to do RP or BART screening if following the
BART guidelines as the commenter indicates is not a reasonable interpretation.
486 DRAFT December 2014 'Modeling Guidance for Demonstrating Attainment of Air Quality Goals for Ozone,
PM2.5 and Regional Haze', p. 16.
487 Id. p. 149. ; "Thus, the preferred approach for regional haze-related model applications is to simulate an entire,
representative year (i.e., one whose mean derived deciview values for "20% worst" and "20% best" days
approximates mean values for deciviews averaged over 5 years for the best and worst days). States can then base the
RRF values on the best and worst days in each Class I area for the modeled year (the ~24 best and worst days from
the modeling year).
488 Id. several sections of the guidance document.
593
-------
We note that if we applied the commenters' interpretation to the existing Texas RH SIP and
BART screening modeling, we would have to disapprove all the BART screening modeling
conducted with CAMx and also disapprove the RH SIP because they did not have 3 to 5 years of
meteorology data.
The commenters failed to take into account a number of issues here. First, there is a general
question of whether App. W § 8.3.1.2.d applies directly in this case as they assert. The analysis
was for RP and to identify if some sources had large impacts on RP and if they did, whether
there were reasonable available controls for these sources which might result in lowering
visibility impairment for the 20% worst days (an improvement in RP). Second, CENRAP, using
2002 meteorology and CAMx, conducted an analysis on groups of sources, source types and split
by regions and potential controls on sources. Third, Texas conducted an RP technical analysis
starting with the results of the CENRAP analysis. Texas included its RP technical analysis in its
SIP submittal. We conclude that in this case, EPA's modeling guidance and requirements are
consistent with using the 1 year of meteorology. EPA will continue to revisit our PGM guidance
periodically and update as needed to address any future concerns and further advancements and
changes in modeling resources, etc.
The commenter also asserted that EPA indicated we were following the BART guidelines in
several respects, but did not provide any context to evaluate how this was indicated in our record
and related to their specific comment. We did rely upon the BART Guidelines for assistance in
assessing the reasonable progress factors, as applicable. As discussed above, the part of the
BART guidelines that discusses that a three to five year meteorological dataset should be used to
assess visibility benefits for BART, is not applicable to our analysis for RP using CAMx.
15.h ETEX and CAPTEX Tracer Studies and CAMx Model Uncertainty
General summary: We received comments that EPA's CAMx modeling significantly
overstates visibility impact and improvements on which EPA based its proposal. Commenters
describe the ETEX and CAPTEX tracer studies and conclude that the results of these studies
prove that CAMx overestimates visibility impacts by a factor of 3. The results also show an
overestimate in CALPUFF results by a factor of 6 (ETEX) or a factor of 3 to 4 (CAPTEX).
When this factor of 3 overprediction is taken into consideration, modeled visibility improvement
from controls are small and should not be required.
Comment: [AECOM (0061/0075) p. 1-6] AECOM stated that the EPA's CAMx modeling
significantly overstates the visibility improvements on which EPA is basing its proposal.
Analysis performed for EPA by ENVIRON in 2012, which EPA fails to address in this proposal,
shows that CAMx overstates visibility impacts by a factor of three at the distances EPA attempts
to model here.
Response: See our response to the next comment.
594
-------
Comment: [AECOM (0061 and 0075) p. 6-2, AECOM/CCP (0075) App A] Luminant and CCP
attached an AECOM report titled, "Analysis of the U.S. EPA 's Proposed Rule on the Texas and
Oklahoma Regional Haze State Implementation /7
-------
• the locations of the CAMx predictions are subject to larger uncertainty the farther the
simulated transport distance, since an accumulation of errors in the wind speed and
direction can result in large erroneous displacements of modeled emissions; and
• the magnitude of the predictions is likely to be overstated because of limitations in the
application of vertical wind direction shear effects that tend to disperse pollutants more
effectively than the models indicate.
CAPTEX. CAPTEX was a series of 3-hour perflourocarbon tracer releases from Dayton, Ohio,
and Sudbury, Ontario, conducted during the fall of 1983. The releases were timed to take
advantage of forecast, steady winds from the west and northwest, since the sampling sites were
located to the east of the release points. The sampling sites were arrayed in arcs, at
approximately 100-km intervals, at distances of between roughly 300 km to 1100 km from the
release point. The 62 sampling sites in the United States and 24 sampling sites in Canada took
either 3-hour or 6-hour samples, depending on the distance from the release point.
Based on the results of the CAMx evaluation as reported in Appendix C of the 2012
EPA/ENVIRON report, AECOM indicated that for the two CAPTEX experiments evaluated:
The use of CAMx with plume-in-grid employed (which EPA employs for its review of the Texas
SIP), the fractional bias results ranged from about 0.6 to 1.2, corresponding to over-predictions
ranging from a factor of 2 to 4. The over-prediction tendency for the large distances associated
with CAPTEX is consistent with the ETEX results for CAMx.
CALPUFF has fractional bias evaluation results slightly higher than 1.0, corresponding to
over-predictions between 3 and 4. The over-prediction tendency for the large distances
associated with CAPTEX is consistent with the ETEX results.
According to AECOM, the implication for the tendency for CAMx and CALPUFF to more
significantly over-predict impacts of emission sources hundreds of kilometers from a target is
that the modeled impact of controls for these distant sources is overstated.
Response: We disagree with the commenters' conclusion about the ETEX and CAPTEX tracer
studies and the relevance of these tracer study analyses. The analysis provided allegedly
indicating that CAMx overestimates visibility impacts by a factor of 3 is an incorrect
interpretation and has flaws in the evaluation and conclusions. Our regulations do not allow for
the calibration of model results to try to adjust for potential biases.489 Furthermore, the bias
amount indicated by the commenter is flawed and is based on limited sampling of model
performance evaluations that exist and the tracer tests are limited to only inert pollution
dispersion and not chemistry evaluations.
In order to develop a coherent understanding of performance of LRT models used for regulatory
purposes, for this action, we examined all of EPA's prior LRT model evaluation efforts. These
evaluation studies included the 1986 8-Model Study (Carhart, et al., 1989), Rocky Mountain
Acid Deposition Model Assessment Project (Godowitch, 1989a, 1989b), IWAQM Phase I and II
489 App. W, Section 7.2.9(a) "... Therefore, model calibration is unacceptable."
596
-------
(EPA, 1998a, 1998b), and the 2012 EPA LRT model evaluation study (ENVIRON, 2012).490
From this review, we drew the following conclusions that are relevant to commenters concerns:
1 The results of both the 1986 and 2012 EPA evaluation efforts clearly identified that LRT
model performance is highly sensitive to the quality of the meteorological inputs, meaning
that LRT model performance will vary with the performance of the meteorological
simulation used to supply meteorological information to the LRT model. Thus, in context,
one cannot arbitrarily conclude that any modeling platform will consistently overpredict
with every application as LRT model performance is inextricably linked to prognostic
meteorological model performance.
2 Carhart et al. (1989) concluded that, in general, LRT models tend to underestimate the
horizontal spreading of the plume at ground level resulting in too high peak (centerline)
concentrations when compared to the observations. For Lagrangian models compared in the
1986 study, this was believed to be due to using sigma-y dispersion (Turner) curves that are
representative of near-source and are applied for longer (> 50 km) downwind distances. The
underestimation of the lateral spread of plumes will consistently result in overprediction of
ground level concentrations, resulting in higher bias and error scores. This observation is
consistent with the findings of the EPA 2012 model study which showed that the model
CALPUFF underestimated the observed lateral distance of the plume by as much as 51% for
the SRL75 database and 34% for the GP80 on the 100-km monitor arcs (ENVIRON, 2012).
This phenomenon was also observed in the original report published in 1998 as part of the
IWAQM Phase II process. CALPUFF underestimated lateral plume spread by as much as
24%) for the GP80 study and as much as 49% for SRL75 (EPA, 1998a). At greater distances,
both the 1998 and 2012 studies showed that this phenomena reversed itself, with much
greater lateral distances compared to observed (GP80, 600-km arc). This would result in a
lower model bias at greater distances.
The graphical analysis of the ETEX results provided in the 2012 report also shows an
underestimation of the lateral spread of almost all models compared to observations. This
appears to be an artifact that affects most models used for LRT purposes, irrespective of whether
the dispersion is based upon Turner stability classes or not. However, of all the models evaluated
in the EPA 2012 study, CAMx scored the highest (best) across all spatially related evaluation
metrics for ETEX and second best overall for CAPTEX 3 and 5, indicating that the degree in
spatial error is consistently lower than other models in the EPA 2012 study, and thus we feel
490 Carhart, R.A., A.J. Policastro, M. Wastag, and L. Coke, 1989: Evaluation of Eight Short-Term Long-Range
Transport Models Using Field Data. Atmos. Environ., 23, 85 - 105.
Godowitch, J.M. 1989a. Evaluation and Sensitivity Analyses Results of the MESOPUFF II Model with CAPTEX
Measurements. EPA-600/3-89-056. U.S. Environmental Protection Agency, Atmospheric Research and Exposure
Assessment Laboratory, Research Triangle Park, NC.
Godowitch, J.M., 1989b: Evaluation and Testing of the MESOPUFF II Model System with CAPTEX
Measurements. 6th Joint Conference On Applications of Air Pollution Meteorology, American Meteorological
Society, Anaheim, CA, 56-59.
USEPA, 1998a: A Comparison of CALPUFF Modeling Results to Two Tracer Field Experiments. Tech. Rep., EPA-
454/R-98-009, Research Triangle Park, NC, 48 pp. http://www.epa.gov/scram001/7thconf/calpuff
USEPA, 1998b: Interagency Workgroup on Air Quality Modeling (IWAQM) Phase 2 Summary Report and
Recommendations for Modeling Long Range Transport Impacts. Tech Rep., EPA-454/R-98-009, Research Triangle
Park, NC, 160 pp. http://www.epa.gov/scram001/7thconf/caipuff/phase2.pdf
597
-------
confident in placing greater overall confidence in CAMx model predictions compared to other
model platforms used for single source purposes, such as CALPUFF.
As stated in a response above, our CAMx modeling analysis utilized a technique called RRF that
limits the potential impacts of modeling performance issues since the modeling results are used
in a relative sense to adjust montored data and absolute modeling values are not directly used.
Due to this and other reasons, we do not think that the CAMx modeling overstates the impacts.
In fact, several pieces of information related to CENRAP and our modeling indicates the impacts
are underestimated. CENRAP and TCEQ did a thorough model performance analysis evaluation
in development of the RH modeling and did not see gross overprediction biases as the
commenter asserts but actually underestimation issues on many of the higher pollutant days.
We did evaluate our model projections on the worst 20% days, the best 20% days in comparison
to speciated monitored values, and we also compared to previous similar comparisons performed
with CENRAP's CMAQ modeling results.491 Overall there was a significant underestimation of
total visibility impairment and sulfate compounds compared to monitored values on the 20%
worst days with slightly higher underestimation of sulfates in our modeling compared to
CENRAP's modeling. Contrary to the assertions of several commenters that CAMx results are
biased high, this was not true for the 20% worst days at Class I areas in Texas and near Texas.
This is important because it indicates that the CAMx modeling was actually underestimating the
SO2 impacts on most of the 20% worst days, which is opposite of the comment that CAMx
results overestimated by a factor of 3.While all models have some bias issues and never replicate
the atmosphere exactly, the previous modeling results developed by ENVIRON for CENRAP
did include an extensive model performance analysis and were considered acceptable by
CENRAP and TCEQ, and we have previously approved the CENRAP modeling database in
other CENRAP states. Considering that our modeling and CENRAP's modeling is
underestimating sulfate impacts and overall light extinction, this information is much more
compelling and directly related to our analysis, where the two tracer study results are not as
directly comparable. Based on our assessment and CENRAP/TCEQ's assessment of the
CENRAP based modeling databases we strongly disagree with the comment that the CAMx
results are overestimated, especially considering the tracer data analysis is based on an inert
tracer and our analysis is a cumulative analysis with chemistry and for the same dataset that is
pertinent to this specific action.
As discussed in a separate response to comment in this section, the CALPUFF modeling
submitted by the commenter had flaws and is not appropriate even before they did their
inappropriate calibrating of results, as discussed below.
We note that the commenters proposed calibrating of delta-dv results is not accurate, even if it
were permissible. As noted in other comments and in our TSD, a change in concentrations of a
certain % does not result in the same percentage of change in the del-dv value since the visibility
impairment calculations are not based on a linear relationship and are dependent on the
logarithmic light extinction formula and background levels. Therefore the difference in model
versus monitor performance for concentrations based on an inert tracer do not change
491 Environ February 2013 memorandum, "2002 Baseline CAMx Simulation, Texas Regional Haze Evaluation";
'Memo_TXHAZE_2002CAMx_ENV_2 lFeb2013 .docx'
598
-------
proportionally the same as the calculated del- dv would change due to the non-linearity of the
calculation of the inverse Mm value and the del-dv. The Method 8 "new" or "revised"
IMPROVE and del-dv equations are documented in a comment elsewhere in this document and
also in our TSD. It should also be pointed out that the studies used inert tracers, therefore the
evaluations of the different models and set-ups only analyze performance of dispersion and not
the chemistry part of the models.
Comment: [AECOM (0061) p. 6-10, AECOM (0075) p. 6-11] AECOM stated that the
magnitude of most modeled visibility benefits relative to average natural conditions for BIBE
and GUMO for the emission controls specified by EPA are at or below 0.2 delta-deciview.
However, these modeled benefits do not consider the CAMx over-prediction tendency of a factor
of about 3 for transport distances on the order of 500 km, which are comparable to the distances
involved in this analysis. To account for this factor of 3 over-prediction, the modeled visibility
benefits should be adjusted in two ways:
Divided by 3 to account for the CAMx over-prediction over long distances, and
Reduced further to account for the adjustments in natural conditions (default assumptions
need to add natural impairment, as discussed above), which would reduce the deciview
change.
AECOM provided Tables 6-1 through 6-4 (not reproduced here) to illustrate adjustment of EPA's
estimated deciview improvement from natural and 2018 background conditions by a factor of 3
to account for CAMx over-prediction for 14 EGUs. Table 6-2 showing the amended deciview
improvement from average natural conditions accounting for a CAMx factor of 3 over-prediction
indicates that the resulting visibility benefits are below the 0.2 delta-deciview threshold for all
affected units. The 2018 background modeled benefits in Table 6-3 are below 0.03 delta-
deciviews when considering the factor of 3 overprediction in Table 6-4.
Response: We disagree with the use of a a 0.2 delta dv threshold by the commenter as discussed
in responses below. As discussed in the response to the previous comment, we do not agree
with the conclusion that our CAMx results are overpredicting by a factor of 3 and note that
EPA's Appendix W prohibits calibrating model results. As discussed in the previous response,
the overall model performance on the 20% worst days is actually underpredicting sulfates and
total extinction at the Class I areas in Texas and at the Wichita Mountains. Therefore the
information using 'calibrated' CAMx model values in the commenter's tables is invalid and is
not an accurate assessment nor acceptable under EPA's regulations in Appendix W. The
commenter has not supported its contention that our modeling or the CENRAP/TCEQ modeling
databases are overpredicting at all and definitely not by a factor of 3. EPA disagrees that our
visibility impacts and benefits from our CAMx model results are overestimated and they are
potentially underestimated.
Comment: [CCP (0075) p. 2] CCP stated that the EPA's model significantly overstates the
visibility improvements on which EPA is basing its proposal. When properly adjusted to account
for this recognized over-prediction bias, the impacts from Coleto Creek Unit 1 are insufficient to
599
-------
justify installation of controls. CCP's own modeling further demonstrates the lack of impact of
emissions from Coleto Creek Unit 1 on visibility conditions and the lack of an improvement in
visibility that can be attributed to the proposed controls.
[CCP (0075) p. 9] CCP stated that the EPA arbitrarily used CAMx modeling to model transport
of haze from Coleto Creek Unit 1 and other sources, even though EPA has consistently promoted
use of CALPUFF and despite significant known limitations with CAMx involving over-
prediction of emissions at longer distances. Estimates suggest that CAMx model predictions are
consistently over-predicting observations by a factor of 3. Coleto Creek Unit 1 is more than
500km from the nearest Class I area and thus was thus unfairly subject to these model biases and
estimates for visibility should be adjusted by a factor of 3 (suggesting a total impact of less than
0.1 dv, and EPA has acknowledged impacts of 0.2 dv as too low to justify further controls, see
77 Fed. Reg. 14,604, 14,625 (March 12, 2012) ). EPA's modeled benefits in haze reduction from
Coleto Creek Unit 1 do not reliably indicate actual impacts and when adjusted for this over-
prediction bias do not support the additional controls proposed for Coleto Creek Unit 1.
Response: See our responses to comments above and elsewhere in this section for comments on
the calibration of the model, CAMx overprediction by a factor of 3, and model selection of
CAMx over CALPUFF. We disagree with the commenter on all of these specific issues and the
comment that the Coleto Creek impacts should be 'calibrated' to 1/3 the modeled is not
substantiated and not allowed by EPA's regulations.
The commenter is incorrect in their comparison of the estimated impacts of Coleto Creek to a
previous action that found that an impact of 0.2 dv was "too low to justify further controls."
Regarding the comment that EPA found that a level of 0.2 dv improvement was too low to
justify additional controls in Arkansas, the commenter is misrepresenting EPA's decision and not
providing the proper context about the decision as it relates to appropriate interstate consultation.
We note that the estimated visibility impairment contribution from all sources in Arkansas on
Wichita Mountains was 2.3% of the total extinction. For proper context, compare this with the
estimated 1.5% total contribution from just a single source (i.e. Big Brown) at Wichita
Mountains. Modeling results showed that complete removal of Arkansas' contribution would
result in the 0.2 dv improvement in 2018 at Wichita Mountains. Both Arkansas and Oklahoma
agreed, through consultation, that additional reductions from sources in Arkansas for reasonable
progress at Wichita Mountains was unnecessary, and EPA agreed that the consultation was
satisfactory. The same conclusion could not be made regarding the consultation between
Oklahoma and Texas as the magnitude of emission reductions and associated visibility benefits
to Wichita Mountains due to controls on Texas' sources was much more significant relative to
Arkansas' impact on Wichita Mountains.
Here, the commenter is erroneously attempting to compare the determination of the significance
of impacts from all the emissions of an entire state (Arkansas) to the impacts from a single
facility. EPA source apportionment modeling for 2018 estimated the contribution to visibility
impairment at Wichita Mountains from the Coleto Creek Unit 1 to be 0.5% and contribution to
impairment at Big Bend from Coleto Creek Unit 1 was modeled to be 0.49%, which are very
600
-------
sizeable when you consider that these impacts are approximately 25% of the impact from all the
sources in Arkansas.
Comment: [CCP (0075) p. 10] CCP stated that the AECOM technical reports attached to their
comment [AECOM (0061/0075) and AECOM/CCP (0075)] document the well-recognized bias
of models used by EPA to over-predict actual impacts in the large modeling domains that were
necessary to evaluate the emissions from Coleto Creek Unit 1 on Class 1 areas that are each over
500 km away. EPA failed to consider the over-prediction bias in presenting and evaluating
visibility impacts that may be attributed to Coleto Creek Unit 1 or the proposed controls. The
modeling results with the over-prediction bias correction indicate that actual impacts on Class 1
areas are below levels EPA has used as thresholds for requiring further controls.
Both modeling approaches, which represent somewhat independent analyses, result in consistent
conclusions showing the lack of any significant impact on haze conditions. The CALPUFF
modeling performed by CCP and included in these comments with worst-case emissions for all
pollutants indicate that for the three Class I areas, the visibility impact predictions (corrected for
model bias) indicate haze levels below those where EPA rejected additional controls for
visibility under the Regional Haze Program. The CAMx modeling performed by EPA that
assesses the effects of the WFGD controls indicate that such controls would result in an
equivalent visibility result, such that the controls are ineffective in changing the haze levels. In
fact, the likely actual impacts are consistent with levels that reflect "no degradation" at all.(64
FR 35730)
Response: See responses to comments above and elsewhere in this document for the calibration
of the model, CAMx overprediction by a factor of 3, and model selection of CAMx over
CALPUFF. We disagree with the commenter on all of these specific issues and the comment that
the Coleto Creek impacts should be 'calibrated' to 1/3 the modeled is not substantiated and not
allowed by EPA's regulations.
For responses to comment for the CALPUFF modeling performed by CPP, CALPUFF model
calibration for potential bias, consistency with other actions, overall impacts and no degradation
see responses to comments elsewhere in this document.
In summary, the commentor is incorrect on these issues and presented critically flawed
CALPUFF modeling and calibrations of modeling results which is not allowed by App. W. Our
analysis that controls for Coleto Creek are beneficial is still valid.
15.i Back Trajectory analysis
General summary: Commenters provided back trajectory data (72 hrs, 500m) and monitored
data for 2002 and 2011-2013 for the 20% worst days for Big Bend, Guadalupe Mountains, and
Wichita Mountains. The commenters conclude that this data shows that only a small number of
back trajectories come from regions with sources being analyzed and considered for controls.
For Big Bend, the back-trajectories submitted by the commenters show the majority of back-
trajectories coming from Mexico. For Guadalupe Mountains, back-trajectories also primarily
601
-------
come from Mexico and visibility is mostly due to natural sources. Back-trajectories for Wichita
Mountains rarely come from sources that EPA is proposing to control.
Comment: [Luminant (0061) p. 105] Luminant noted that back-trajectory data (discussed in
section 6i) show that EPA has clearly targeted the wrong source of emissions in its analysis. The
subset of Texas sources that EPA has chosen to analyze (to the exclusion of other sources) very
rarely are linked to these Class I areas on the days of concern. EPA's small group analysis
(discussed below) is thus flawed from the outset. And the emission limitations that EPA
proposes in its FIP for these sources will not address visibility conditions at these areas and are
completely disconnected from the data before the agency showing that the impacts to these areas
come mainly from other sources.
[Luminant (0061) p. 102] Luminant stated that, in point of fact, the visibility impact from
international emissions dwarfs the impact from all Texas sources (not to mention the handful of
sources EPA is targeting), including on the key worst 20% days. To illustrate, we asked AECOM
to develop back trajectories for the 20% worst days at the Texas Class I areas to determine the
source of the haze-causing emissions. AECOM identified the 20% worst days from publicly-
available IMPROVE data for the 2011-2013 period, and conducted back trajectories to identify
the source of the haze-causing emissions. Figure 10 shows the results of this analysis for Big
Bend. As seen in Figure 10, the overwhelming majority of haze-causing emissions on the key
20% worst days originates from sources in Mexico, including very large sources of SO2
emissions close to the U.S.-Mexico border.660 In fact, for Big Bend, approximately 96% of the
back trajectories passed through Mexico,661 Figure 11 shows the same information for
Guadalupe Mountains—where approximately 77% of the 20% worst day trajectories passed
through Mexico.662
602
-------
FIGURE 10: BACK TRAJECTORIES ON THE 20% WORST DAYS FOR BIG BEND, 2011-2013
Max S02 in 2011-13 (tpyP
«
0.0 -100
•
too 1 -1 000
*
1.000 1 - 10,000
•
10.000.1- 50 000
•
50.000.1- 100.000
•
100.000 1 -200 000
Uealc* M»p
-m—1 rw
1
.. A
>, \ -ft :v. -
/» I * * ; ;v» on| t v. **\
# A \ t * *' ' " J"
# \ \ M' • V-v
* 5 • ¦J .'a
* • • %,> v_* ** * » 1 »w -
¦t *'
-5;
t#
- *' v-
\ *~
, - » **• , X
X
raxfirt
Big
tsii HtSE. 0»lw»«,
UaptrjUdia, 6 Wer to
C*V
V
L«5»«a
*• 72-fir8aet Trajectory
A Texas F IP 8c wees
~ Class I Areas
M P«Wi0«0 <»**«**?
-mflPTSOTj&^A t-2
V-
\ %
t m
Big Bend National Park
Back Trajectories & Annual
Point Source Emissions
AICOM
Sail.
I
603
-------
FIGURE 11: BACK TRAJECTORIES ON THE 20% WORST DAYS FOR GUADALUPE MOUNTAINS,
2011-2013
Max S02 in 2011-13 (tpy)
• 0.0 -100
• 100.1 -1.000
• 1000 1 ¦ 10 000
9 10,053 1-50.300
A 60,003 1 -100.000
100,000 1 -200.000
Ckiida lup-t II aunrtlni
'Jrtorji Part
V
£m, HEJE, D*L»n»
Ug*r»a
-*- 72*f C«dt Tr»,eOcnr
A Texis FIP Eoj'ccs
riciiMi
J'» : •< .-oc; -.rt-!
Guadalupe Mountains
National Park
Back Trajectories & Annual
Point Source Emissions
Mexico
604
-------
Moreover, Luminant noted that this analysis also shows that very few of the 20% worst days are
impacted by Texas sources, much less the targeted Luminant sources. Indeed, for Big Bend, only
14 out of 69 trajectories (about 20%) can be traced to one of the Texas sources for which EPA is
proposing further emission reductions.663 And even more telling, only 3 out of 69 trajectories
(about 4%) can be traced to a Luminant facility. For Guadalupe Mountains, only 5 out of 66
trajectories (about 8%) can be traced to a Luminant facility.664 Yet, those Luminant units would
bear well over half of the over $2 billion in costs that EPA claims is reasonable. There is a wide
gulf between the data and EPA's findings.
According to Luminant, nevertheless, EPA's proposal ignores this highly relevant information,
rendering EPA's conclusions and proposal incomplete and unsupported. EPA purports to
determine what amount of progress towards natural visibility at these Class I areas is
"reasonable," without taking into account the source of the majority of the problem—
international emissions. EPA has thus failed to consider an important aspect of the problem and
failed to "examine the relevant data and articulate a satisfactory explanation for its action,
including a rational connection between the facts found and the choice made."665 Based on the
data presented, EPA must approve Texas's reasonable progress goals, as it has done for other
states facing similar international contributions.
Luminant Footnotes:
660 As Figure 10 shows, and as further explained in AECOM's report, the "boundary condition"
emissions also originate from Mexico point sources south of the modeling domain. EPA's failure
to account for these boundary condition emissions in its review of Texas's SIP is arbitrary and
capricious.
661 AECOM Report at 5-1.
662 Id.
663 AECOM Report at 4-3.
664 Id. at 4-7.
665 Motor Vehicle Mfrs. Ass'n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (internal
quotations omitted).
Analysis of Back Trajectories for 3 Recent Years [AECOM (0061 and 0075) p. 4-1]
To further assess EPA's review of Texas' 2018 RPGs for BIBE and GUMO, AECOM conducted
backward trajectory analyses of air parcels arriving at those parks on the 20% worst haze days
for the 2011-2013 period. The 20% worst haze days were identified from the IMPROVE data
available at the Federal Land Manager Environmental Database.72 Because EPA does not
account for international contributions in its proposed 2018 RPGs, BIBE and GUMO are the
focus of this analysis because of their proximity to Mexico. However, the more distant, but still
very significant, Mexican haze contributions to WIMO are also separately discussed in the next
section.
AECOM stated that the backward trajectories show the areas from which emissions originated
on the 20% worst days, and thus provide a basis for determining what sources should further
reduce emissions to achieve visibility improvements. As discussed further below, the analysis for
605
-------
these two areas shows that the overwhelming amount of haze-forming emissions on the 20%
worst days originate from Mexico, and not the Texas EGUs that EPA is proposing to regulate.
This is consistent with Texas' conclusion in its SIP that, "[t]he Texas Class I areas are close to
Mexico, and international transport of emissions from Mexico and Central America significantly
influence regional haze at these areas."73 In particular, Texas found that "52 percent of the
impairment at Big Bend and 25 percent of the impairment at Guadalupe Mountains is from
Mexico and further south."74 Our analysis confirms Mexico's significance to Texas. EPA's
proposal does not consider a backward trajectory analysis. The influence of international
emissions is discussed in detail in a subsequent section.
AECOM stated that back trajectories were performed for the 20% worst visibility days measured
at each national park's IMPROVE monitor to characterize the nature of these regional haze
episodes. Each trajectory was run using the NOAA Air Resources Laboratory (ARL) HYSPLIT
Trajectory Model75 to track the air parcel backward 72 hours, or three days, prior to its arrival at
the national park's IMPROVE monitor location from 3 p.m. CST time (21 UTC). A midday time
was selected to characterization the air when the boundary layer would likely be well mixed. An
initial height of 500 m above ground level (AGL) was used with the ETA Data Assimilation
System (EDAS) archive meteorological data which has a 40-km horizontal resolution and a
three-hour temporal resolution. This 500 m height represents the air mass within the boundary
layer and is commonly used, particularly by users of the HYSPLIT web interface as it is the
default setting 76
AECOM Footnotes:
72 Colorado State University Cooperative Institute for Research in the Atmosphere (CIRA), cited
2014: Federal Land Manager Environmental Database. Web site:
http://views.cira.colostate.edu/fed/DataWizard/.
73 https://www.tceq.texas.gov/assets/public/implementation/air/sip/haze/2SIP_ado_rev.pdf at 10-
9.
74 https://www.tceq.texas.gov/assets/public/implementation/air/sip/haze/2SIP_ado_rev.pdf at 10-
10.
75 Draxler, R.R. and Rolph, G.D., 2014: HYSPLIT (HYbrid Single-Particle Lagrangian
Integrated Trajectory) Model Version 4. NOAA Air Resources Laboratory, College Park, MD.
Big Bend National Park Back Trajectory [AECOM (0061 and 0075) p. 4-2]
AECOM stated that a total of 69 days represented the 20% worst days at BIBE where data were
available. The associated IMPROVE data composition plot is presented in Figure 4-1 which is
identical to the previous composition plot, except it provides a more detailed look at 2011-2013.
Also, the days where trajectories that passed near the Coleto Creek facility are noted by "C".
Many of the worst days occurred in 2011, when Texas experienced one of its worst single-year
droughts on record as discussed in a previous section.
AECOM Figure 4-2 shows the back trajectories at BIBE from 2011-2013 on the 20% worst days.
The 38 facilities in Texas that were modeled by EPA are labeled in the map. The back
trajectories illustrate that BIBE most often experienced air from the west, south, and southeast
transport regimes. The most recent facility-aggregated point source SO2 emissions for Texas and
606
-------
its neighboring states (2011-2013)77 and Mexico (2008)78 are shown. It is apparent from the
trajectories for the 20% worst haze days that BIBE very rarely experienced northeasterly
transport from the Texas facilities indicated for additional emission controls in EPA's proposed
FIP. Only 14 out of the 69 trajectories (approximately 20%) passed within 193 km of the Texas
sources for which EPA is proposing new SO2 emission limits. This analysis was performed using
ArcGIS. The farthest distance a trajectory must travel to reach a Texas facility with a proposed
new emission limit is approximately 860 km away from BIBE. A trajectory uncertainty of
22.5%, the middle of the uncertainty range discussed by NOAA ARL (15 - 30%79), was used.
Thus, 22.5% of 860 km is 193 km. For Coleto Creek, there were only 5 trajectories (about 7%)
that passed within 126 km trajectory uncertainty range (22.5% of 560 km). Using the same
methodology, only 3 out of 69 trajectories (approximately 4%) originated from or within 193 km
of a Luminant facility.
In contrast, trajectories indicate that emissions from Mexico, particularly the facilities such as
Carbon I and Carbon II on the Texas/Mexico border, heavily influenced BIBE haze conditions.
These point sources stand out as large emitters just southeast of BIBE and only 230 km away.
Trajectories from the west and south further indicate the transport of natural dust events from the
Chihuahuan Desert and/or fire activity. For example, as shown in AECOM Figure 4-3 in
comment 0061/0075, one of the top five worst haze days (April 21, 2011) indicated southerly to
southwesterly transport which brought in smoke from wildfires. As noted previously, TCEQ has
identified numerous dust and fire events, particularly in 2011, that adversely influenced
particulate matter measurements. Furthermore, a recent long-term dust event climatology
identified 16 local dust events during 2000-2007 that specifically affected BIBE.80 According to
the 2011 NEI (version 1), wildfires in 2011 in Texas alone account for approximately 188,970
tons of PM2.5 and 222,985 tons of PM10.81 These statistics further support the TX ALT NC
natural conditions presented above.
AECOM Footnotes:
77 EPA, cited 2014: Air Markets Program Data 2013. Web site: http://ampd.epa.gov/ampd/.
78 EPA, cited 2014: Emissions Modeling Clearinghouse: 2011-based Modeling Platform,
201 1NEIv2-based Platform (2011v6.2). Web site: http://www.epa.gov/ttn/chief/emch/.
79 http://www.arl.noaa.gov/documents/workshop/NAQC2007/HTML_Docs/trajerro.html.
80 Tong, D. Q., M. Dan, T. Wang, and P. Lee, 2012. Long-Term Dust Climatology in the
Western United States Reconstructed from Routine Aerosol Ground Monitoring. Atmospheric
Chemistry and Physics, 12, 5189-5205.
81 http://www.epa.gov/ttnchiel/net/201 linventory.html.
Guadalupe Mountains National Park Back Trajectory [AECOM (0061 and 0075) p. 4-6]
AECOM stated that a total of 66 days represented the 20% worst days at GUMO for 2011-2013
where data were available. The associated IMPROVE data composition plot is presented in
AECOM Figure 4-4 where "C" denotes days trajectories that passed Coleto Creek. The chart
indicates that GUMO often measured more naturally-occurring haze-forming species (CM,
OMC, and soil species) than sulfate and nitrate species from manmade activities, particularly in
2011 and 2013.
607
-------
Similar to BIBE, as shown in AECOM Figure 4-5, back trajectories to GUMO on the 20% worst
days show that GUMO's haze conditions were influenced by west, south, and southeast transport
regimes. GUMO infrequently experienced northeasterly transport from the direction of the
facilities targeted in EPA's proposed FIP during the 20% worst haze days where 21 out of 66
(approximately 32%) trajectories originated from the Texas FIP sources. The farthest distance a
trajectory must travel to reach a Texas facility with a proposed new emission limit is
approximately 960 km away from GUMO. Like the approach taken with BIBE, the trajectory
uncertainty of 22.5% was used. Thus, 22.5% of 960 km is about 216 km. With this procedure,
we find that only 8 trajectories (about 12%) originated from within 182 km of Coleto Creek
(22.5% of 800 km), and we find that only 5 trajectories (about 8%) originated from within 216
km of a Luminant facility.
As mentioned previously for BIBE, AECOM stated that GUMO is also significantly influenced
by emissions from Mexico and by natural events such as windblown dust and fires. In relation to
natural events, it is instructive to note that the aerosol composition chart showed GUMO is
influenced significantly by uncontrollable naturally-caused haze. Soil and CM represent larger
amounts of the total extinction than at BIBE, which reinforces the reasoning behind GUMO's
higher TX ALT NC natural conditions. Research has shown that GUMO experiences more dust
storms than most other Class I areas in the southwestern U.S. where 27 events where identified
from 2000-2007.83 Both TCEQ and the New Mexico Environment Department have submitted
demonstrations to EPA of several exceptional event days for particulate matter measurements. 84
Also, according to the 2011 NEI (version 1), wildfires in 2011 in Texas and New Mexico
account for approximately 270,070 tons of PM2.5 and 318,683 tons of PM10.85
Response: The commenters' back trajectory analysis for the base period and 2011-2013 is
flawed and did not follow the NOAA draft guidance they cited and appropriate HYSPLIT
modeling techniques.492 We note that the commenter tried to use a draft NOAA document that
indicated potential error in HYSPLIT centerlines may be 15-30%) and the commenter used the
middle value of 22.5%. EPA notes that the 15-30% value is not supported by any technical
analysis by the model developer NOAA and is basically just rough estimate. There are a number
of uncertainties in HYPSLIT analyses and on the scale of 72-hours. Luminant's HYSPLIT
analysis also used a more restrictive value of 40 km for uncertainty.
The HYSPLIT model yields an estimate of the path an air mass has traveled before reaching a
monitor at a specific location and time. Specifically, the model provides the centerline of the
probable path. By knowing where an air mass has traveled before reaching a monitor, one can
consider what potential areas and emission sources could have contributed to the pollution
loadings in the atmosphere that has been trnaported to the monitor. HYSPLIT back trajectories
alone do not determine inclusion or exclusion of an area but in fact are supporting information
that compliments other information.
The commenter's analysis assigned a trajectory uncertainty of 22.5% (halfway between 15-30%)
mentioned in the HYSPLIT user guide) in determining an error-bounded minimum distance for a
trajectory centerline to pass the point source to claim contribution. In other words, the
492 NOAA is National Oceanic and Atmospheric Administration. NOAA is the developer of HYSPLIT and has
previously provided draft guidance on the use of the HYSPLIT model.
608
-------
commenter used this uncertainty bound to determine if a back trajectory centerline was near
enough to a source to indicate that emissions from the source may have been included in the air
mass transported to the receptor. The commenter's analysis used a straight-line distance between
source and receptor to calculate this distance. However, the on-line HYSPLIT user information
explicitly states an error may be estimated to be 15-30% of the travel distance of the trajectory.
Since none of the trajectories are straight-line but take a longer path, all the analyses provided by
the commenters based on the 22.5% uncertainty value are invalid and biased to underestimate the
number of trajectories that may be 'close' to the source in question since the actual travel path is
significantly longer and would result in wider uncertainty values when the back trajectory is
nearest to the source. Furthermore, limiting the uncertainty bound to 40 km in assessing the
HYSPLIT data for Wichita Mountains is invalid and significantly biased to underestimate the
number of trajectories that may be 'close' to the sources in question. EPA's OAQPS office was
consulted on this matter and the determination from the office expert was that there are a number
of variables that impact the uncertainty and there is not enough information to support selecting
any specific value, but it is likely that the actual uncertainty level is higher than the unsupported
value of 15-30%). Some of the variables that increase the uncertainty are: resolution of the
meteorology and amount of relevant upper air data to help with meteorological model
performance, grid resolution, time and length of transport and amount of mixing variability in the
atmosphere, accuracy of convective mixing in the modeling, etc. There are also some
uncertainties in the back trajectory model itself as is documented in EPA's model performance
analysis.493 Therefore we do not agree with the use of the unsupported trajectory uncertainty to
try and limit the potential uncertainty.
HYPLIT back trajectories have some value but they do not do any pollutant dispersion or
chemistry, so they are not a full evaluation of transport of emissions from sources to Class I
areas. We put more weight on CENRAP's CAMx source apportionment modeling and our
CAMx modeling in determining whether a source impacts a Class I area and what the potential
visibility improvements from reductions at a source would yield. The commenter is putting too
much weight on the HYPSPLIT back trajectories and not enough weight on the CAMx modeling
results. Since the overall uncertainty of HYSPLIT back trajectory analysis has a number of
variables it is hard to impossible to try and create uncertainty ranges as clear decision points of
'near' a source or 'not near' a source. If HYSPLIT back trajectories show no winds or very
limited amounts of back trajectories in the general direction it might indicate the source has a
relative small chance of impact if no other analysis exists. In this case we have CENRAP's and
TCEQ's source apportionment modeling and our source apportionment modeling that use the
same meteorology and both the CENRAP/TCEQ source apportionment modeling and our CAMx
modeling indicate the sources that we have identified for control analysis have large
contributions to visibility impairment at the Class I areas. Given the limits of a HYSPLIT
analysis and larger uncertainties, and the results of our technical anlaysis, we do not agree with
the commenter that we are evaluating the wrong sources. As the facts in our FIP TSD and
proposal demonstrate, the identified facilities have significant impacts on visibility conditions.
Our technical record makes it equally plain that the required controls reduce impacts from these
sources and result in meaningful visibility benefits towards the goal of natural visibility
493 Documentation of The Evaluation of Calpuff and Other Long Range Transport Models Using Tracer Field
Experiment Data, Environ International Corporation, EPA Contract No. EP-D-07-102, Work Assignment No. 4-06,
Figure 6-13 (2012), available at littp://www.epa. gov/ttn/scratn/reports/EPA- 454 R-12-003.pdf
609
-------
conditions.While EPA's analysis includes many elements and steps we disagree that we have not
laid out our logical steps in performing our analysis and supporting the proposed control
decisions and our final decisions.
The commenters assessments of what they consider to be near is inappropriate in the context of
the HYSPLIT capability. The results that the commenter provides indicating that for many of the
sources there are few trajectories that go near the source is biased extremely low due to the
inappropriate and flawed uncertainty analysis.
EPA conducted our own HYSPLIT back trajectories to do a more thorough analysis and to
respond to the comments provided with HYPSLIT results. EPA had two different sets of
HYSPLIT runs that were completed. We did one run for all days in a year for a 30-year period to
develop a climatological baseline for comparing subsets of years and the subset of 20% worst
days to compare with the data provided by the commenters. The other set of HYSPLIT runs we
conducted were to evaluate for only the 20% worst days at each of the three Class I areas for the
baseline period of 2000-2004 and 2011-2013. We did not use the trajectory uncertainty approach
as we have a number of concerns with the unsupported values and think the range given does not
capture the full uncertainty range of HYPLIT results. For our analysis, we did the following:
a. Run HYSPLIT trajectories for the dates and locations using
i. ED AS 40 km meteorology when available and the highest resolution sets
available for the older years
ii. Receptor heights 100m AGU
iii. 72 hour backward
We used 100 m since the visibility impairment is monitored either by a monitor or by a visitor
looking at vistas and not towards 500 m and above. In addition, our initial evaluation, discussed
elsewhere in this section, shows that the 2011-13 time period is not representative of
climatological norms regarding the transport wind flows to the three Class I areas.
We received a number of comments that argued that the 2011-13 time period was normal or
worse than normal and monitoring data for 2011-13 was on track for meeting 2018 RPGs at the
three Class I areas. To respond to these comments we conducted HYSPLIT runs for a 30 year
period (1985 to 2014) develop a climatological transport pattern record for each of the Class I
areas). We also evaluated the all days and 20% worst days for the period of 2000-2004 and for
the period of 2011-2013 to compare these two groups of years to the 30 year transport patterns.
We have included some of the plots that we generated in the figures below. We also find that the
base time period 2000-2004 was more representative of climatological norms.
We performed HYSPLIT modeling of a 30-year period (1985-2014) and concluded that in years
with wind flow patterns consistent with the climatological norms over that period a significant
number of days have back trajectories that did include areas where the sources proposed for
additional controls are located. Furthermore our analysis of the 2011-13 period indicates that it
was less representative of normal pollution transport patterns also showed a number of back
trajectories went through or near the areas with the sources being considered for controls.
610
-------
Therefore these back trajectories do indicate the controls being considered for these sources
would be expected to reduce visibility impacts at the three Class I areas.
Our analysis of 30-years of back trajectories to assess whether the 2011-13 and 2000-2004
periods were within the climatological norm also indicated that the base period (2000-2004) was
more similar to the climatological norm than the 2011-2013 period, so we conclude that using
the base period is more representative for projecting 2018 levels.
In sum, the number of trajectories that go near the sources in Texas is large enough to not rule
these sources out from consideration for potential control. In general, we have treated back
trajectories as a tool to potentially screen an area out if no trajectories go through an area but if
some trajectories go through an area then the area may be evaluated further or, as in this case, the
full analysis may rely on more sophisticated tools such as CAMx. Moreover, our back trajectory
analyses show that the sources for which we are imposing controls through the FIP are the
approporate ones.
We have additional HYSPLIT plots and a summary of the results in the docket materials but we
have provided some additional plots here in support of our conclusions.
Figures EPA1-6 are back trajectory plots for different periods.
Figures EPA 7-12 are incremental probability plots. These plots compare a specific subset of
plots to the 30 years of back trajectories that were completed for the climatological transport
characterization for each of the Class I areas. The yellow dot on the map indicates the start point
of the back trajectory and the blue color shading indicates areas that had less back trajectories
through the plot than the 30 year average and red areas indicate areas that had more back
trajectories through the plot area than the 30 year average. The gradual shading represent slightly
less/more to the darker shading which is a larger amount of difference compared to the 30 year
average. For example a deep red indicated that the back trajectories passed through a particular
area significantly more often than the 30 year average.
Big Bend:
Figures EPA 1 and EPA 2 illustrate that the baseline period for BIBE had more back
trajectories/transport from areas around the sources we considered for controls during the 2000-
2004 baseline than during the 2011-2013 period. Figures EPA 7 and EPA 8 indicate that the
baseline period had slightly above average transport for all days from central and south Texas
and the 2011-13 period was characterized with less transport from south Texas and higher
transport from the border area in Mexico to the Southeast of BIBE. Overall, it can be seen that
the 2011-13 period 20% worst days had only a few back trajectories to northeast Texas and this
period is atypical compared to the climatological norm and the baseline. Looking at the baseline
period and climatological analysis support that controls will be beneficial in the areas with the
sources we have proposedto control.
Guadalupe Mtns.:
Figures EPA 3 and EPA 4 illustrate that the baseline period for GUMO had more back
trajectories/transport from areas around the sources we considered for controls during the 2000-
2004 baseline than during the 2011-2013 period. Figures EPA 9 and EPA 10 indicate that the
baseline period had slightly above average transport for all days from north central through south
611
-------
Texas and the 2011-13 period was characterized with less transport from north central Texas
through higher transport from the border areas in Mexico to the Southeast of GUMO. Overall, it
can be seen that the 2011-13 period 20% worst days had only a few back trajectories to northeast
Texas and this period is atypical compared to the climatological norm and the baseline. Looking
at the baseline period and climatological analysis support that controls will be beneficial in the
areas with the sources we have proposed to control
Wichita Mtns.:
Figures EPA 5 and EPA 6 illustrate that the baseline period for WIMO had significantly more
back trajectories/transport from areas around the sources we considered for controls during the
2000-2004 baseline than during the 2011-2013 period. Figures EPA 11 and EPA 12 indicate that
the baseline period had nearly normal levels of transport for all days from east Texas, but
slightly less days from central Texas and the 2011-13 period was characterized with much less
transport from north central Texas through southeast and south central Texas. Overall, it can be
seen that the 2011-13 period 20% worst days had only a few back trajectories to northeast Texas
and this period is atypical compared to the climatological norm and the baseline. Looking at the
baseline period and climatological analysis support that controls will be beneficial in the areas
with the sources we have proposed to control
As discussed above the narrow and inappropriate definition of what constituted 'near' for a back
trajectory made all the comments and conclusions suspect. EPA performed this analysis to
provide an unbiased analysis. The results indicate that the 2011-13 was somewhat atypical for
transport to BIBE and GUMO and very atypical for transport to WIMO
The commenter indicated that a number of back trajectories went through Mexico but failed to
mention that many of these also went through or near Texas. Therefore, sources in Mexico and
Texas could both contribute emissions to the visibility impairment at the Class I Areas. We have
concluded that the back trajectory data provided by the commenter do not support their
assertions that transport from the regions with those sources we are controlling is rare. The data
they have provided are inconsistent with the HYSPLIT guidance and general practices and are
for years that are not representative of normal climatological patterns with respect to transport
wind flow to the Class I areas. Furthermore, the back trajectories submitted by the commenter
do in fact show transport from regions in Texas for some days. Our additional analysis identified
the normal wind patterns over a 30-year period and determined that based on normal conditions,
transport does occur from the regions in Texas with those sources we are controlling.
We respond to comments on international and boundary emissions elsewhere in this document.
Comment: EPA's "additional analysis" starts with the wrong group of sources
[Luminant (0061) p. 109]
To illustrate the fault in EPA's approach, Luminant performed a back trajectory analysis for the
Wichita Mountains—the Class I area that EPA claims will see the most improvement from its
FIP. They performed this analysis for 2002 (the single year that EPA uses for its modeling base
case and future projections) and also for 2011-13 to show more recent meteorological conditions.
A back trajectory analysis, unlike a Q/D analysis, shows whether the wind blowing toward the
Class I area actually originated near a source of interest or from some other region. As shown in
612
-------
Table 12 below, the Texas sources for which EPA proposes controls are among the sources with
the fewest number of days with a back trajectory from Wichita Mountains passing nearby the
facility.
TABLE 12: SUMMARY OF THE TOTAL BACK TRAJECTORIES FROM WICHITA MOUNTAINS NEAR
CERTAIN FACILITIES IN 2002 AND 2011-2013
(HIGHLIGHTED ROWS ARE FACILITIES FOR WHICH EPA PROPOSES ADDITIONAL CONTROLS)685
FIP
Controls
Proposed?
Total Back Trajectories Near Facilities on
State
Facility
20% Worst Days / Total # of Days in
Period
2002
2011-2013
TX
Works No. 4
No
6/24
11/70
TX
Oklaunion Power Station
No
2/24
14/70
TX
San Miguel - Lignite Fired PR
Yes
0/24
12/70
TX
Coleto Creek
Yes
5/24
7/70
TX
TXI - Midloathian Plant
No
5/24
5/70
TX
Calaveras Plant
No
3/24
7/70
TX
Twin Oaks Power
No
4/24
6/70
TX
Holcim - Midloathian Plant
No
5/24
4/70
TX
Fayette Power Project
No
4/24
5/70
TX
Limestone Electric Generating Station
Yes
3/24
6/70
TX
WA Parish EGS
No
3/24
5/70
TX
Big Brown SES
Yes
3/24
5/70
OK
Sooner Generating Station
No
1/24
7/70
TX
Sandow SES
Yes
1/24
6/70
TX
Gibbons Creek
No
1/24
6/70
MEX
CARBON II (International)
No
0/24
7/70
OK
PSO Northeastern Power Station
No
1/24
5/70
OK
Grand River Dam Auth - Chouteau
No
0/24
5/70
OK
Muskogee Generating Station
No
0/24
4/70
TX
Big Spring Carbon Black
No
1/24
4/70
TX
Monticello SES
Yes
3/24
0/70
TX
Welsh Power Plant
No
3/24
0/70
TX
AEP Pirkey Power Plant
No
1/24
1/70
TX
Harrington Station Power Plant
No
0/24
2/70
TX
Sid Richardson - Borger Carbon Black Pit
No
0/24
1/70
TX
Orion - Borger Carbon Black Plant
No
0/24
1/70
TX
Tolk Station
Yes
0/24
1/70
TX
Martin Lake SES
Yes
0/24
1/70
Luminant noted, as Table 12 shows, the units for which EPA proposes controls very rarely have
the potential to impact Wichita Mountains. Indeed, Luminant facilities are among the facilities
with the fewest days with a trajectory originating from nearby and should have easily been
screened out of EPA's analysis. And EPA's choice of which sources to evaluate was critical
because, as discussed in the Background section and elsewhere in these comments, the
subsequent steps were constructed to all but assure that some impact would be shown. Much
greater occurrences are seen among other Texas sources, international sources, and Oklahoma
sources, for which EPA did not even consider whether to require additional controls. It is
apparent, then, that EPA's pre-ordained analysis was flawed from the outset, and had EPA's
613
-------
analysis started in a different and proper manner, it would have reached a different result. This is
yet another way that EPA's methodology was arbitrary and capricious and cannot support its
proposal.
Footnotes:
683 See FIP TSD at A-4.
684 Id
685 Source back trajectory data is from the NOAA HySplit model, using EDAS 40km resolution
meteorological data, 120 hour duration, ending at 3 p.m. at the WIMO IMPROVE monitor
location 500m above ground. The analysis measures the distance from each facility to the hourly
back trajectory endpoints, provided in the data files for each day from the Hy Split output, using
the Haversine formula. A trajectory is near a facility when the facility is within a linearly
increasing radius up to 40 km during the initial 6-hour period, or within 40 km for the remainder
of the 5-day period. The 20% Worst days are days based on the total extinction monitored at the
WIMO IMPROVE visibility monitor, available at:
ftp://vista.cira.colostate.edu/Public/AirQuality/Data/Aerosol/IMPROVE/DataSetsBySiteAVIMO
l.csv.
Response: As discussed above, we do not agree with Luminant's narrow definition of 'near' by
limiting the definition to within 40 km of the back trajectory center line. It is unreasonable and
not accurate to use 40 km as a maximum distance from the HYSPLIT centerline for transport
distances that would occur for 5 days. For example, applying this threshold a HYSPLIT
centerline might travel on the order of 600 km from GUMO or BIBE back towards Texarkana,
TX, but if a facility was not within 40 km of the centerline it would not be considered as a source
that could have contributed to pollution levels at the West Texas Class I areas. This does not
match with the unsupported NOAA uncertainty range of up to 30 %. While we do not think 30%
is an upward limit on HYSPLIT uncertainty, a back trajectory centerline distance of 600 km the
bound would be +/- 180 km, not 40 km. This is a fundamental flaw in their analysis and overly
biases the results to very low numbers. See our response to multiple comments above regarding
sources and HYSPLIT analyses. We address the other comments from Luminant regarding the
idneitifcation of sources for additional control analysis elsewhere in this document.
Figure EPA 1 - Big Bend 20% worst days 2000-2004 100 m Back Trajectory
614
-------
615
-------
Figure EPA 2 - Big Bend 20% worst days 2011-2013 100 m Back Trajectory
Arkansas
OK 13 h 0 ffi a^uachita
TOCK STATION
Mountains
WORKSfNO 4
I ^ v _ ' ' WELSH POWER-PLANT [J| .)
j ^__JwONTICELLO STEAM ELECT
MARTIN LAKE ELECTRIC ALTSTATION Q L _-Mjs.si
r \ -v—r
BIG BROWN STEAM-ELECTRIC STATION p \ (
Texas j YIMESTONE ELECTRIC GENERATI
SANDOW STEAM ELECTRIC STATION Louisiana I
tervds* Desert!"
GUMOI
-WAJPARISH ELECTRIC GENE
^COLETO CREEK POWER STATION
Sommers-.Deelv-Sp ruce.
;^San Miguel ft
iSierra»Madre Oriental Mountains
wmih
cidental Mot
lmagfceJN&SA:
616
-------
Figure EPA 3 - Guadalupe Mountains 20% worst days 2000-2004 100 m Back Trajectory
¦Arkansas,
Oklahoma
Ouachita Mountains
TOLK STATION
WORKS(NO 4
Mew Mexico
WELSHPOWER Pi:ANT
* ? ' B -'d i r K : " tBlpfevA'" fy
^_^-WIONirilCELLO STEAM E
ie^t MARTIN LAKE ELEGTRI^gTAJION Q_J,-
jpy BIG BROWN STEAM-EL-ECT-RIC'STATION n \
GUM01 Texas I ^LIMESTONE] ELECTRICjGENj
SANDOW STEAM 'ELECTRIC STATION Louisiana
(ChlSfa^uoniDosert t
BIBE1
'Sierra Mad re Oriental Mountains
i . \
tad re Occidental M
Image 'NASA
lma2Q 1i5 Digitjl^Qbe
617
-------
Figure EPA 4 - Guadalupe Mountains 20% worst days 2011-2013 100 m Back Trajectory
WORKSfNO 4Oklahoma
TOLK STATION
iNew Mexico
T^WEbSH-PO^PLANT $
k " i ' | ^^^ONiTICEL-L-O-STf
iMARTIN LAKE E L E C T R1 CA^ST AJI ON Wjs-J
BIG BROWN STEAM.EL-EGTRIG'STATION Q 1\ {
rilMm / " Tevas ^ VLIMESTONE ElECTRI
GUM01 / lexas J \ '"V .• \
SANDOW STEAM ELECTRIC STATIO^0u;lsian
<;GOdEmCREEK POWER
* V 3 JBL*1| Big B end .Desert
'ffjChibuahuanJpesei'l.
S* " -U^El' San Miguel'©
\A -r. ¦ r t \
>iega>Madre Oriental Mountains^
618
-------
Figure EPA 5 - Wichita Mountains 20% worst days 2000-2004 100 m Back Trajectory
Missouri
-Arkansas
- /I
/Oklahoma
WIM01- "
•WORKS NO 4 \
Ouachita Mountains,.
WELSH POWER PLANT J
- MONTICELLO STEAM ELECTRIC-STATION
^ i (
V'MARTIN LAKE ELEC'T-RICAL-SIATION
OWN STEAM ELECTRIC STATION n
(LIMESTONE ELECTRIC GENERATION STATION
>1 ' - •? I J
Louisiana
SANDOW STEAM ELECTRIC STATION
¦ ;.-c . \
SommersvDeely-Sprupg+40ustsCC^^T" A—^
VQ' ^ nousio,wavparisH ELECTRIC GENER^TJNG'STATION
R- J--' PS
San Migu€l»p;WL <;COLETO CREEK POWER STATION
619
-------
Figure EPA 6 - Wichita Mountains 20% worst days 2011-2013 100 m Back Trajectory
620
-------
Figure EPA 7 - Big Bend Incremental Probability all days 2000-2004 100 m Back Trajectory
621
-------
Figure EPA 8 - Big Bend Incremental Probability all days 2011-2013 100 m Back Trajectory
622
-------
Figure EPA 9 - Guadalupe Mountains Incremental Probability all days 2000-2004 100 m Back
Trajectory
623
-------
Figure EPA 10 - Guadalupe Mountains Incremental Probability all days 2011-2013 100 m Back
Trajectory
624
-------
Figure EPA 11 - Wichita Mountains Incremental Probability all days 2000-2004 100 m Back
Trajectory
625
-------
Figure EPA 12 - Wichita Mountains Incremental Probability all days 2011-2013 100 m Back
Trajectory
626
-------
15.j CALPUFF Modeling for Coleto Creek
General Summary: Commenters submitted CALPUFF modeling for Coleto Creek Unit 1 for
2004-2006. Results indicate that visibility impacts from the facility are below the 0.5 dv subject
to BART threshold. The commenter states that tracer studies suggest CALPUFF overestimates
visibility impacts by a factor of 4.5 (on average) and adjusts the CALPUFF model results down
by this factor. The commenter concludes that Coleto Creek's impacts are very small and any
visibility benefit from controls would be even smaller.
Comment: [AECOM/CCP (0075), whole report] CCP attached a report by AECOM titled,
"Analysis of Visibility Impacts from Coleto Creek Unit 1 - Using CALPUFF(AECOM/CCP,
0075).
Introduction. AECOM noted that, in July 2005, EPA issued BART Guidelines that provide
guidance to the States in making Best Available Retrofit Technology (BART) determinations for
large power plants and other BART-eligible sources. The BART Guideline states that modeling
may be necessary to support a decision by the States about which BART-eligible sources "cause
or contribute" to visibility impairment and are subject to BART. EPA recommended that the
modeled value be compared to the "contribution" threshold of 0.5 delta dv and the "cause"
threshold of 1.0 delta dv to determine if a source does not contribute to visibility impairment and
therefore is not subject to BART. In states where source-specific subject-to-BART tests are
conducted, sources are exempted from BART review if their predicted impact is below 0.5 delta-
dv for a peak predicted day. Impacts of facilities below this level for the 98th percentile
predicted day are presumed not to cause or contribute to visibility impairment. AECOM noted
that this contribution threshold was also used in Texas in Texas BART screening with
CALPUFF494, but Texas electrical generating units participating in CAIR (or CSAPR) are
exempt from facility-specific BART for SO2 and NOx emissions.
AECOM stated that the purpose of their analysis was to perform a facility-specific subject-to-
BART analysis even though Coleto Creek Unit 1 is exempted from BART for SO2 and NOx
emissions by rule to assess the potential for the unit to impact visibility conditions in Class I
areas. The analysis provides information as to whether that unit would be determined not to
contribute to regional haze based upon CALPUFF modeling for impacts at the Class I areas that
are the subject of EPA's proposed FIP (Big Bend National Park, Guadalupe Mountains National
Park and Wichita Mountains Wilderness). The location of Coleto Creek relative to these Class I
areas is shown in Figure 1-1. As shown in Figure 1, all three Class I areas are well beyond 300
km from Coleto Creek (in fact, at least twice that distance). At these distances, EPA guidance,495
is to use CALPUFF "with caution" because of a demonstrated overprediction tendency and the
application of the model beyond a distance for which it has not been validated for use.
494 This is stated in the Texas BART Rule, available in Appendix 9-2 of the Texas regional haze SIP
495 Interagency Workgroup on Air Quality Modeling, 1998, available at
http://www.epa.gov/scram001/7thconf/calpuff/phase2.pdf
627
-------
For the CALPUFF modeling, the commenter indicated that there currently exists a 6-kilometer
resolution 3-year (2001-2003) CALMET meteorological database produced by CENRAP that
covers the necessary area needed to perform the CALPUFF modeling for Coleto Creek Unit 1
and the three Class I areas. This CALMET database relied solely on the 36-km MM5 data as
input to CALMET combined with 6-km terrain and land use data. Running CALMET in this
manner is known as "No Obs" mode; it does not include supplemental meteorological data from
surface and upper air stations. Use of this limited CALMET data would require the conservative
use of the highest modeled visibility impairment impact for determining if the source causes or
contributes to visibility impairment. The 98th percentile (or 8th highest) ranked days impact can
be used for determining if a source causes or contributes to visibility impairment impacts if
"refined" CALMET meteorological data is used in the analysis. Refined CALMET data includes
the MM5 data (or equivalent) along with actual concurrent surface, upper air and precipitation
observations. Therefore, AECOM prepared three years (2004-2006) of "refined" CALMET
meteorological data for use in the CALPUFF visibility modeling for Coleto Creek Unit 1.
The remainder of AECOM's CALPUFF report describes the emissions selected for modeling,
the preparation of the refined CALMET meteorological data, the CALPUFF modeling (including
post-processing), and analysis of the modeling results.
AECOM indicated that in summary, the results of the CALPUFF analysis detailed in their report
indicated visibility impacts (corrected for model over-prediction bias) due to peak Coleto Creek
emissions for the three Class I areas that are considered insignificant and consistent with "no
degradation".
AECOM obtained annual and daily emissions for Coleto Creek Unit 1 over the past 15 years
(2000-2015) from the EPA Clean Air Markets Program Database. The annual SO2 and NOx
emissions trends from 2000-2014. They noted that with the exception of a spike in 2009, the
annual emissions profile has remained relatively constant since 2002. Therefore, AECOM used a
six-year look-back period in order to determine what they thought was the conservatively high
peak daily emissions to be used for the CALPUFF modeling.
In addition to SO2 and NOx emissions, they also included estimates of emissions of H2S04 and
PM10. AECOM reviewed the daily actual emissions of SO2 and NOx emissions from Coleto
Creek Unit 1 for the last six calendar years (2009-2014). They excluded days with emissions due
to abnormal (e.g., malfunction) conditions were excluded from consideration, consistent with the
BART Rule guidance. The highest normal daily SO2 emission rate (6,806.7 lb/hr) day was
January 27, 2010. The highest daily NOx emission rate (1.3891 lb/hr) occurred on December 11,
2009. AECOM provided a plot of the daily average SO2 and NOx emission rates for the last six
years (Figure 2-1 of AECOM report). A separate figure (Figure 2-2 of the report) shows that
while the highest SO2 and NOx emission days are outliers, they are not substantially higher than
some of the other higher days in the period. AECOM indicated that it is conservative to use the
highest daily emissions to characterize a full 3-year period of modeling, and also to consider that
peak emissions for SO2 and NOx (which were not on the same day) were coincident and occurred
on every day of the three-year modeling simulation.
628
-------
AECOM used the model inputs provided in Table 2-1 of the AECOM report to run CALPUFF.
AECOM indicated that the PMio and H2SO4 emissions are consistent with values proposed in the
Coleto Creek Unit 2 permitting effort. The PMIO emissions were further speciated into three
components, all of which have different light scattering properties: (1) PMF (fine particulate
matter: soils); (2) EC (elemental carbon); and (3) SOA (secondary organic aerosols based on
guidance provided by the National Park Service (NPS) for visibility modeling.
CALMET Meteorological Data Processing. AECOM developed a refined three-year (2004-2006)
CALMET database using 4-km grid spacing. The modeling domain was based on a 50-km buffer
around the source and a 50-km buffer around each of the three Class I areas to account for puff
recirculation. The modeling domain is shown in Figure 3-1 of the AECOM report. This design
allows for an 856 km (east-west) x 780 km (north-south) domain extent and, at a 4-km
resolution, there are 214 x 195 horizontal grid cells.
AECOM used a Lambert Conformal Conic (LCC) coordinate system to account for the curvature
of the Earth's surface. The LCC projection for this analysis was based on the WGS-84 datum
and standard parallels of 30 and 60 degrees North, with an origin of 31.7 degrees North and
100.6 degrees West.
AECOM used the latest EPA-approved version of CALMET (Version 5.8.4, Level 130731) to
produce three-dimensional wind fields for three years (2004-2006). Advanced meteorological
data in the form of prognostic mesoscale meteorological data, the Fifth Generation Mesoscale
Model (MM5), were used to provide a superior estimate of the initial wind fields. This
application considered 3 years (2004-2006) of prognostic MM5 meteorological data all at a 36-
km resolution.
These prognostic meteorological data sets were combined with the 4-km grid resolution terrain
and land use data to more accurately characterize the wind flow throughout the modeling
domain. The gridded terrain data was derived using the U.S. Geological Survey (USGS) 90-
meter grid spacing Digital Elevation Model (DEM) files. These files were processed in the
TERREL pre-processor program. The gridded land use data was derived from USGS 1:250,000
Composite Theme Grid land use files. Figures 3-2 and 3-3 of the AECOM report show a
depiction of the terrain and land use data throughout the modeling domain, respectively.
The Step 2 wind fields were produced using the input of available National Weather Service
(NWS) hourly surface and twice-daily upper air balloon sounding data within and just outside
the modeling domain. Hourly surface data from both first-order and second-order stations also
were considered in this analysis. Hourly precipitation data from stations within and just outside
of the modeling domain were also included. Figure 3-4 of the AECOM report shows the
meteorological stations that were used in the CALMET modeling.
The CALMET modeling was conducted using the MREG=1 switch which requires specific
settings to conform to EPA-recommendations. In addition, the CALMET settings followed the
August 31, 2009 EPA guidance document, "Clarification on EPA-FLMRecommended Settings
for CALMET"9.
629
-------
CALPUFF Modeling and Post-Processing. AECOM used the current EPA-approved version of
CALPUFF (Version 5.8.4, Level 130731) to model emissions of SO2, NOx, H2SO4, and PM10
from Coleto Creek Unit 1 in order to estimate visibility impacts at the three closest Class I areas.
The CALPUFF modeling was conducted using three years (2004-2006) of CALMET
meteorological data as discussed above. The CALPUFF modeling results were post-processed
with the CALPOST (Version 6.221, Level 080724) and POSTUTIL (Version 1.56, Level
070627) utility programs. For visibility post-processing. Additional details on the modeling
inputs and settings are provided in the AECOM report.
CALPUFF Modeling Results. Modeling was conducted by AECOM using CALPUFF for three
years (2004-2006) to estimate visibility impacts at three Class I areas, Big Bend National Park,
Guadalupe Mountains National Park and Wichita Mountains Wilderness, due to emissions from
Coleto Creek Unit 1. It is noteworthy that all of these Class I areas are well beyond 300
kilometers from the Coleto Creek Facility. Therefore, the CALPUFF modeling results of
visibility impacts are likely to be overstated.
The AECOM modeling exercise was treated primarily as a typical subject-to-BART modeling
exercise even though Coleto Creek Unit 1 is exempt from BART for SO2 and NOx emissions by
rule. The modeling used peak daily average emission rates for SO2 and NOx from recent
operating years, along with allowable emissions of particulate matter, as detailed in sections
above.
All the modeling was performed in a manner that would meet the expectations of Federal Land
Managers for a refined CALMET and CALPUFF modeling analysis. The refined modeling
analysis allows for the use of the 98th percentile (8th highest day) results from each year to be
used for determining if the source "causes or contribute" to visibility impairment.
The overall objective for the AECOM modeling exercise was to evaluate modeled visibility
impacts relative to a natural conditions background. Table 6-1 shows the results of the
CALPUFF modeling analysis. The results shown in Table 6-1 indicate that for the three Class I
areas being considered for the proposed FIP, the predicted impacts are less than 0.5 delta-dv in
all cases. Based on these results, Coleto Creek Unit 1 does not contribute to visibility impairment
at any of the modeled Class I areas. Furthermore, Coleto Creek Unit 1 would NOT be subject to
BART even if it were not exempt by rule. In addition, we also believe that, based on long-range
transport evaluation studies conducted by EPA, the over-prediction tendency of CALPUFF
results are likely between a factor of between 3 and 4 to 6 for the distances being evaluated
(equating to an average over-prediction bias of at least 4.5). Attachment A provides additional
details on the basis for this estimate and a summary of the EPA-conducted evaluations.
Therefore, the amended peak CALPUFF impact is likely on the order of 0.1 delta-dv or less, as
shown in Table 6-2 of the AECOM report.
AECOM stated that it is noteworthy that EPA Region 6 has previously indicated that a visibility
improvement of 0.2 delta-deciviews is too low for applying emission reductions. This action was
for the Arkansas RHR SIP, for which EPA stated, "[w]ith regard to the comment that Arkansas
sources contributed 2.0% to visibility impairment at Wichita Mountains during the baseline
period and are projected to contribute 2.3% in 2018, EPA notes that removal of this 2.3%
630
-------
contribution to the total extinction results in a visibility improvement of only 0.2 dv from the
2018 projected visibility conditions. ...Consequently, while we are concerned that the RPG at
Wichita Mountains is not on the glide path, we believe the technical assessment that Arkansas
sources do not have a significant impact at Wichita Mountains is accurate and ADEQ and ODEQ
followed consultation procedures."14
AECOM noted that the EPA separately determined for the State of Wyoming that a single
source's modeled visibility improvement of 0.19 dv for a Class I area was insignificant to require
additional controls. EPA stated that, "[a]lthough the cost-effectiveness for SNCR is reasonable,
we find it reasonable for the State not to select this control technology based on the incremental
visibility improvement for this control technology."15 Similarly, EPA determined that a visibility
improvement of 0.17 dv was too low for application of emission controls.16 These cases indicate
that modeled visibility improvements as high as about 0.2 dv17 have been determined to be too
low for requiring emission controls because such controls would result in minimal
improvements.
AECOM concluded that Coleto Creek should be excluded from the EPA FIP proposal because a
conservative screening analysis with CALPUFF indicates that its peak daily emissions, even if
assumed to be continuous for a 3-year modeling period, results in a modeled impact that below
the haze contribution threshold. In addition, the CALPUFF modeling results imply that the
current Coleto Creek emissions produce a visibility result in the three Class I areas that is both
imperceptible and without any significant visibility impact. The CALPUFF results with worst-
case emissions for all pollutants indicate that for the three Class I areas, the visibility impact
predictions (corrected for model bias) indicate a haze level consistent with "no degradation".
Due to the insignificance of the unit's emissions on the Class I area, the controls proposed by
EPA in the FIP on Coleto Creek Unit 1 would be ineffective in changing the haze levels.
Footnotes:
1 US EPA, 2005. BART guidelines, Federal Register update: July 6, 2005.
2 This is stated in the Texas BART Rule, available in Appendix 9-2 of the Texas regional haze
SIP.
3 "CAIR" is the Clean Air Interstate Rule and "CSAPR" is the Cross-State Air Pollution Rule.
4 79 FR 74818.
5 Interagency Workgroup on Air Quality Modeling, 1998, available at
http://www.epa.gov/scram001/7thconf/calpuff/phase2.pdf (see Appendix C).
6 70 FR 39129.
8 Documentation is available at http://www2.mmm.ucar.edu/mm5/.
9 Available at:
http://www.epa.gOv/scram001/guidance/clarification/CALMET%20CLARIFICATION.pdf
14 77 Fed. Reg. 14,625.
15 78 Fed. Reg. 34,751.
16 78 Fed. Reg. 34,752.
17 EPA's Regional Haze Rule notes that "no degradation" to visibility would be "defined as less
than a 0.1 deciview increase." (64 Fed. Reg. 35,730)
631
-------
Response: We have reviewed the CALPUFF modeling provided for Coleto Creek Unit 1 and do
not agree with the conclusions that Coleto Creek's impacts are small. We have a number of
concerns with the CALPUFF modeling provided and the technical and policy conclusions that
were made in the comments. These comments are linked to an overall conclusion about the
impacts and whether the impacts are small as the commenter suggests. Overall, our CAMx based
analysis of sources is more robust and even reports provided or cited by the commenter indicate
that CAMx is more accurate than CALPUFF and more robust. Furthermore (as addressed in
similar comments) CAMx is the better tool to use for analyzing potential RP controls and
potential benefits as well as identifying the largest contributors of visibility impairment to Class I
areas. CAMx and CMAQ were used by CENRAP and the CENRAP states to evaluate RP on
20% worst and best days and also in specific RP control grouping analyses conducted by
CENRAP for its members (Nine States including Texas and Oklahoma). CAMx was the best tool
to assess if installing RP controls at Coleto Creek would yield visibility benefit and the relative
benefit compared to other Texas sources with the larger impacts on visibility at Class I areas.
Even if the commenter had provided acceptable CALPUFF modeling, we would still weigh the
CAMx model results as more scientifically robust and more accurate in this situation than
CALPUFF results. CALPUFF results are not geared to evaluate benefits and impacts on the 20%
worst and best monitored days as they evaluate only a single facility's impacts and not the entire
airshed.
To be thorough, we have reviewed the cited CALPUFF report and modeling and we will address
our concerns with each subtopic and then with the overall conclusions made by the commenter.
Introduction: The commenter indicates that Texas and EPA recommended the 0.5 delta dv as the
contribution threshold. This is correct in the context the commenter raised for BART screening
modeling with a regulatory analysis utilizing CALPUFF modeling results of visibility impacts.
This is not correct in the context of CALPUFF modeling results for potential visibility benefits
for BART controls on a source nor is it appropriate to use as a threshold for potential RP
controls. 496. The commenter raised that EPA guidance indicates to use caution (and consult with
EPA which was not done in AECOM's situation) in utilizing CALPUFF at the distances that
exist between Coleto Creek and the three closest Class I areas. As the commenter indicated
CALPUFF is known to potentially overpredict nitrates in general regardless of distance. Further,
CALPUFF has not been validated for these distances in general, and CALPUFF's model
performance at these distances is not totally known. As discussed later, performance analyses
with CALPUFF indicates it does not perform as well as some other models (including CAMx).
This is a concern to EPA and was a general concern to Texas in 2006. Because of concerns with
model performance of CALPUFF at great distances (example: 500-1000 km), Texas actually
requested to use CAMx with Plume-in-Grid (PiG) and Particulate Source Apportionment
(PSAT) to screen potential BART sources. EPA approved the approach of screening of small
groups of sources for BART applicability with CAMx modeling using a 0.5 delta-dv maximum
impact (not 98th percentile) and using a natural "clean" background approach in 2007. As
discussed in other responses, EPA used the same technical tools that Texas did in 2006 for
BART screening for our RP analysis using updated CAMx with the PiG and PSAT tools to
assess visibility impacts and benefits of controls on individual sources. While the 0.5 delta-dv
496 If a BARTsource had impacts above 0.5 dv, but the controls being evaluated were less than 0.5 dv in benefit over
baseline, this would not preclude the source from being controlled
632
-------
value has some context in the BART situation to assess visibility impacts, there is nothing in the
Regional Haze Rules that indicate that 0.5 delta-dv visibility benefit is a threshold of significance
for a single facility in analyzing for potential RP controls.497
AECOM indicated that the only CALMET data that existed was the CENRAP dataset (2001-
2003) that was known as a "No Obs" data set that required using the maximum CALPUFF value
instead of the 98th percentile. This is not correct, a number of facilities in Oklahoma, Texas and
Louisiana, reprocessed the 2001-2003 CENRAP met data with CALMET to include observations
which allowed use of the 98th percentile. We do not see the need for CALPUFF analysis, but
AECOM should have used these datasets instead of creating 2004-2006 datasets if they wanted
to conduct CALPUFF modeling. AECOM's CALPUFF modeling strayed from these agreed
upon protocols in TCEQ's SIP materials. These CALPUFF modeling protocols were agreed to
between EPA, Federal Land Manager Representatives, TCEQ and CENRAP in 2006-2007 and
there are a number of areas where AECOM's modeling did not follow the previous protocols that
makes AECOM's CALPUFF modeling not acceptable nor relevant.
AECOM used emissions from 2009-2014 CEM data and filtered the data to come up with what
they thought was representative emission rates to model. AECOM did not use the 2000-2004
data period that all other applicants used for BART screening modeling. They did not use CEM
data in selecting emission rates for the period that they modeled (2004-2006).498 The BART rule
indicates you should model 24-hour maximum emission rates for the period modeled. For BART
screening with CALPUFF all sources used either 2001-2003 (met period modeled for CALPUFF
pursuant to approved protocols) or 2000-2004 which included the modeled years. AECOM did
not compare the rates they modeled to emission rates that would have been modeled if they had
used 2001-2003, or 2000-2004 as other BART applicants used.
CALMET Meteorological Data Processing: AECOM did not justify why the 2004-2006 period
was used instead of just re-running CALMET for 2001-2003 with observations or using
available datasets for the 2001-2003 period. As discussed above many sources did refined
modeling protocols with EPA, FLM and Texas to recreate the CALMET data with observations
and these files were available through TCEQ and/or EPA. The CENRAP modeling protocol
included meteorological performance analysis of the MM5 met data (comparing MM5 model
values to actual observations) to ensure that the MM5 data was adequately replicating
meteorological conditions prior to CALMET processing. MM5 model performance analysis was
one of the requirements from the EPA and CENRAP approved modeling protocols. Once the
MM5 data was deemed to meet acceptable model performance it was then used by CENRAP's
contractors to generate the CENRAP CALMET data. AECOM did not provide any
497 BART Guidelines, FR Vol. 70 No. 128 July 6, 2005 p.39129 "Even though the visibility improvement from an
individual source may not be perceptible, it should still be considered in setting BART because the contribution to
haze may be significant relative to other source contributions in the Class I area. Thus, we disagree that the degree
of improvement should be contingent upon perceptibility. Failing to consider less-than-perceptible contributions to
visibility impairment would ignore the CAA's intent to have BART requirements apply to sources that contribute to,
as well as cause, such impairment."
498 BART Guidelines, FR Vol. 70 No. 128 July 6, 2005 p. 39129. "We recommend that the State use the highest 24-
hour average actual emission rate, for the most recent three or five year period of meteorological data, to
characterize the maximum potential benefit."
633
-------
meteorological performance analysis to substantiate that its model was performing adequately
before it generated CALMET meteorological files. AECOM also used a smaller grid than EPA
would have approved by only creating a 50 km buffer beyond the source and the Class I areas of
interest. Given the location of Coleto Creek is relatively close to the gulf coastline, Figure 3-1 of
AECOM's report demonstrates how plumes could go outside of either the eastern or southern
boundaries and they would no longer be tracked, even if the winds shifted and brought the
plumes back into the computational grid. EPA has required a much larger buffer to make sure
that plumes do not transport out of the grid because the impacts are then lost and do not re-enter
even if wind fields would re-entrain the pollutant plumes. In addition, AECOM also used a
smaller grid then we would normally accept by using a 4 km grid instead of a 6 km grid as was
outlined in CENRAP and TCEQ's modeling protocol and CALMET files. The origin used for
the projections also deviates from the CENRAP and TCEQ's modeling protocol.
CALPUFF Modeling and Post-Processing: BART guidelines require the modeling should
utilize emissions representative for the baseline period and the meteorology should be from the
baseline period499. Both Texas and CENRAP's BART modeling protocols agreed to use 2001-
2003 CALMET meteorology using MM5 databases that had been approved by EPA, FLM
representatives, States and Regional Planning Organizations, including CENRAP.500 The
modeling submitted for Coleto Creek did not follow these agreed protocols.and the computation
grid size and buffer are large enough to result in acceptable modeling results regardless of what
regulatory purposes the modeling is conducted.
We noted a number of issues when we reviewed the modeling files themselves: 1) AECOM used
a much smaller computational grid than we would recommend and what is in CENRAP's
modeling protocol; 2) AECOM has less vertical grids and the spacing of the grids in the lower
atmosphere is larger (individual layers are vertically thicker) than we would recommend, which
likely leads to artificially high dispersion and lower impacts; 3) No primary Carbon emissions
(PMC) were included and a very low geometric standard deviation of particle size was used; 4)
AECOM limited maximum plume travel to 1 grid cell, which is not appropriate for a 4 km grid
(especially considering mid and higher level winds) and EPA has typically used a maximum
plume travel of 10 grids with a 6 km grid; 5) AECOM used very low background ozone values to
fill missing data (20-38 ppb) when EPA recommends using 80 ppb currently and TCEQ and
CENRAP modeling protocols used 40 ppb or higher (which was acceptable at the time); and 6)
AECOM used a constant ammonia level of 1 ppb in CALPUFF modeling and a varying
ammonia level that was either 1 ppb or just above 1 ppb in the postutil501 processing when
CENRAP and TCEQ modeling protocols indicate 3 ppb should be used in both CALPUFF and
postutil. Overall most of these deviations/inconsistencies in AECOM's modeling would result in
more dispersion and/or lower visibility impairment at the Class I areas, thus artificially
underestimating the impacts of Coleto Creek.
499 BART Guidelines, FR Vol. 70 No. 128 July 6, 2005; p. 39162 "2. With the accepted protocol and compare the
predicted visibility impacts with your threshold for ' 'contribution."... "We recommend that States use the 24 hour
average actual emission rate from the highest emitting day of the meteorological period modeled, unless this rate
reflects periods start-up, shutdown, or malfunction."
500 Texas BART Modeling Protocol to Determine Sources Subject to BART in the State of Texas and CENRAP
BART Modeling Guidelines
501 Postutil is the name of the computer code used to post process CALPUFF output files.
634
-------
CALPUFFModeling Results: EPA does have concerns with using CALPUFF for these distances
(distance between Coleto Creek and each Class I area) that are in the range of 500 km or more.
Due to the number of issues identified in the grid used, CALMET, CALPUFF and postutil
modeling files above, and the impacts these would have on model results, we do not agree that it
is likely the impacts are overstated. Furthermore, the emission rates modeled were not compared
to the baseline period used by TCEQ and CENRAP (2000-2004 or 2001-2003) and it is unclear
if the emission rates were representative of baseline emissions or included the benefit of controls
on RH pollutants. Even if the emissions represent recent emissions visibility impacts from Coleto
Creek, the modeling issues identified would most likely lead to higher values if corrected.
It is clear from the issues identified that AECOM did not provide modeling the EPA and FLM
representatives would find acceptable. AECOM provided modeling results that indicated the
highest 98th percentile value was 0.427 delta-dv at WIMO and, given the modeling issues, this
value could be above 0.5 delta-dv if an appropriate modeling analysis was conducted. AECOM
used limited model performance data from inert tracer experiments performed for areas of the
country and in Europe that are not representative of meteorology, terrain, and dispersion in Texas
and Oklahoma. As further discussed in another specific response, use of this data is not allowed
and there is not enough information to support that the model performance issues seen in the
CAPTEX and ETEX tracer studies also exist at exactly the same level in this analysis.502
AECOM did not provide any information to substantiate that there were model performance
issues with their modeling analyses for Coleto Creek specifically. Regardless, as we previously
discussed, the 0.5 delta-dv is not necessarily a safe harbor from being considered for RP. As
discussed in our TSD, Coleto Creek was identified as one of the largest visibility impairing
sources at Class I areas in Texas and Oklahoma using CAMx, which is the model that CENRAP,
Texas, and Oklahoma utilized in evaluating potential facilities and controls for RP.
AECOM's comment that EPA Region 6 had previously indicated that a visibility improvement
(from removing impacts of all Arkansas sources) at WIMO of 0.2 delta-dv is too low for
applying emission reductions for RP is comparing 'apples to oranges'. The CAMx source
apportionment modeling results (0.2 delta-dv) is for all the sources in Arkansas and is for the
20% worst days, not the maximum 98th percentile day that CALPUFF results are given from
AECOM. We note that the estimated visibility impairment contribution from all sources in
Arkansas on Wichita Mountains was 2.3% of the total extinction. For proper context, compare
this with the estimated 1.5% total contribution from just a sinsle source (i.e. Big Brown) at
Wichita Mountains. Modeling results showed that complete removal of Arkansas' contribution
would result in the 0.2 dv improvement in 2018 at Wichita Mountains. Both Arkansas and
Oklahoma agreed, through consultation, that additional reductions from sources in Arkansas for
reasonable progress at Wichita Mountains was unnecessary, and EPA agreed that the
consultation was satisfactory. The metric difference alone is not a reasonable comparison.
Comparing one source's impacts with CALPUFF modeling to the impacts from a number of
facilities in Arkansas, which would have a much lower individual impact if evaluated
individually with CAMx, is not an accurate comparison. It can be concluded that the 0.2 delta-
dv impacts when using CAMx and 20% worst metrics was small enough to rule out a state with
many sources and the individual source impacts would be much smaller, but it is not reasonable
to use this value in comparing with CALPUFF modeling results for a single facility.
502 App. W, Section 7.2.9(a) ".. .Therefore, model calibration is unacceptable."
635
-------
AECOM also points to an EPA notice of proposed approval for sources in Wyoming that EPA
evaluated Wyoming's BART control selection. The first example given was where EPA
indicated that visibility improvement from installation of SNCR of 0.19 delta-dv was not
selected based on the incremental visibility improvement.503 In this case EPA had evaluated two
controls, LNB with OF A that resulted in a 0.13 delta-dv and SNCR controls that would result in
0.19 delta-dv visibility improvement. EPA was evaluating two controls, LNB with OFA which is
cheaper and SNCR that is significantly more expensive and in this case EPA proposed
concurrence with Wyoming's decision that LNB with OFA gave approximately 2/3 of the
visibility benefit of SNCR for a much lower cost and the additional 0.06 delta-dv that SNCR
would net was not worth the additional cost. The other EPA proposed approval of state BART
was for General Chemical Boiler D NOx BART analysis. Wyoming proposed approval of BART
controls as existing LNBs with SOFA added on and also evaluated SNCR and SCR as
technically feasible. The cost effectiveness for the three options evaluated were $1280/ton for
LNB with SOFA installed, $3176/ton for SNCR, and $3510/ton for SCR. LNB with SOFA
netted a 0.07 delta-dv benefit. The incremental cost and benefits of SNCR over LNB with SOFA
were $2913/ton and 0.05 delta-dv. The incremental cost and benefits of SCR over SNCR were
$4342 and 0.05 delta-dv. EPA was proposing approval of Wyoming's BART decision because
the incremental benefits over the 0.07 delta-dv for the BART controls of LNB with SOFA were
only 0.05 del-dv for SNCR and an additional 0.05 for SCR but SCR annualized costs were
$3,399,266 compared to the BART control selected that was less than $1 million a year
($943,549). In this case Wyoming determined and EPA proposed approval that the visibility
benefits were not worth the additional costs. It should also be pointed out that these visibility and
incremental visibility benefits were all based on CALPUFF modeling and not comparable to our
CAMx modeling. In either case these visibility benefits are much lower than the commenters
CALPUFF non calibrated model results. We also note that these evaluations were for NOx
controls and not SO2 and there was some concern that the visibility benefits from NOx controls
could be overestimated. While CALPUFF model calibration is inappropriate, if you took the
commenters approach the values above should also be scaled by the same factors, so they are not
intellectually consistent in their argument that Coleto Creek's impact is below levels that EPA
has determined to not control sources in other actions.
Lastly, the commenter referred in a footnote to a cite from EPA's Regional Haze Rule indicating
that EPA proposed a target of no degradation to visibility would be defined as less than a 0.1
deciview increase. Again the commenter is mixing metrics and model results inappropriately.
The 0.1 deciview was in the context of no degradation of overall visibility conditions on the
specific subset of 20% best days (based on monitoring) using photochemical modeling results
(such as CAMx).504 The 20% best days are defined as having the lowest monitored light
extinction at a Class I area, and not directly tied to the days when a CALPUFF model may
indicate a high impact from a source. Given that these are the lowest monitored days it is
unlikely that any day (or very few days) that the source (Coleto Creek) was impacting the Class I
area as the transport winds that would transport Coleto Creek's emissions would also transport
emssions from many other sources in Texas and/or Mexico to the respective Class I area.
Therefore it is highly unlikely (see our responses to the Back trajectory comments) that any of
503 78 Fed. Reg. 34751-34752.
504 Regional Haze Rules, FR Vol. 64 July 1, 1999 p. 35730-35734
636
-------
the days (or few days) that CAMx would transport the impacts of Coleto Creek emissions to a
Class I area would be one of the 20% best days thus the context of 0.1 delta-dv is not directly
relatable. The 20% best days are the cleanest (lowest light extinctions) of the days monitored and
the light extinction at a specific Class I area commensurate with 0.1 delta-dv on the cleanest days
would result in a delta-dv much lower than 0.1 delta-dv on less clean days when emissions from
Coleto Creek and other sources would transport to the Class I area of interest. Reference to the
1999 RHR where an increase of less than 0.1 dv on the 20% best days was "no degradation;"
this applies to overall conditions, not impacts from a single source.
Even setting aside these issues, AECOM's flawed CALPUFF modeling results indicated impacts
from Coleto Creek of 0.427 delta-dv, well above 0.1 delta-dv. (Table 6-1 from AECOM's
CALPUFF modeling report is included at the end of this response). As discussed in our TSD and
other responses we were specifically evaluating sources to identify the largest impacting sources
on the 20% worst days using photochemical modeling which for a number of reasons was more
appropriate than using CALPUFF on individual sources. It was this analysis that supports that
Coleto Creek does have one of the largest impacts on visibility impairment on the 20% worst
days and photochemical grid models are what CENRAP and Texas used to identify potential
sources and potential benefits of controls for RP.
In summary, there are a number of issues with the commenter's CALPUFF modeling: (1) it
utilizes the wrong years for modeling; (2) the modeling does not comply with the original BART
CALPUFF modeling protocol that Texas and EPA approved; (3) modeling grid is too small and
inappropriately sized horizontally and vertically; (4) a number of inappropriate issues with the
CALPUFF inputs (emissions, background ozone data fill, ammonia levels, etc.); and (5) it uses
some inappropriate assumptions, including the calibrating of modeling results based on limited
analyses using other databases and locations that are not directly comparable to assessing
impacts from Coleto Creek's units. The 0.5 dv threshold was utilized as a BART threshold, but
our action is for reasonable progress and the 0.5 dv threshold was not set as an applicable
threshold in the Regional Haze Rule for reasonable progress (see responses concerning costs and
visibility benefit in this document). We used a photochemical grid model, which is more
scientifically robust than the CALPUFF modeling system and is more appropriate for assessing
visibility impacts and benefits on the 20% worst days and longer transport distances, such as the
distances between Coleto Creek and the Class I areas in Texas and Oklahoma. We performed a
multi-tiered analysis in order to identify the Texas facilities with the largest impacts on visibility
at Class I areas (in Texas and Oklahoma) and Coleto Creek's facility did rank as one of the
largest impacting sources of the more than 1,600 sources considered in Texas. As discussed in
another response in this section, calibrating CALPUFF model output values is not contrary to
EPA regulations. We discuss the commenters' use of the tracer studies in this section but their
analysis and conclusions are flawed and not representative of the larger collection of information
available that also is discussed in more detail in our reponse on the issue. In conclusion, based
on our analysis with CAMx, we think both the visibility impacts of the sources and the benefits
from the proposed emission reductions are large enough to be beneficial for reasonable
progress.505
505 Federal Register Vol. 64, No. 126 Thursday, July 1, 1999 p. 35721; ".. .specific control measures a State must
implement in its initial SIP for regional haze. That determination can only be made by a State once it has conducted
637
-------
Table 6-1 From AECOM's Report: CALPUFF Modeling Results (without Amendment for Over-
prediction Tendency)
Class I Area
8th Highest Days Impact (delta-dv)
2004
2005
2006
3-Year Average
3-Year Peak
Big Bend National Park
0.293
0.240
0.140
0.224
0.293
Guadalupe Mountains National Park
0.149
0.199
0.170
0.173
0.199
Wichita Mountains Wilderness
0.427
0.389
0.314
0.377
0.427
15.k Consideration of Maximum Impacted Day or Other Days
General summary: Commenters stated that focusing on visibility impacts on the 20% worst
days ignores larger impacts from these sources and other sources on other days. This approach is
also inconsistent with CALPUFF modeling for BART of the maximum impact from a source for
comparison with a 0.5dv threshold. Consideration of impacts on other days will identify sources
for control analysis that will result in visibility improvement on other days and make progress
towards the goal of natural visibility conditions.
Comment: Initial Source Apportionment Modeling for 38 Q/D Sources [Gray (0070) p.7]
Dr. Gray stated that the second method EPA used to narrow the scope of its review is source
apportionment modeling. The metric used by EPA (the worst 20% of days, or W20 days) masks
visibility impacts from sources that should have been considered in its control review. (FIP TSD,
Section A.2)
Dr. Gray stated that the facilities that were identified as potentially large contributors to
downwind Class I areas via the Q/D analysis were then evaluated by EPA "for
meteorology/transport to determine which of the 38 facilities had large impacts during the
average 20% Worst Days and also their impacts on specific days within the 20% worst days."
Modeling was conducted for the 2002 and projected 2018 baseline emissions scenarios using the
CAMx dispersion model (v5.41) with plume-in-grid (PiG) treatment of close-to-source
emissions and a 12-km nested grid over Texas (within the 36-km grid previously used for
the necessary technical analyses of emissions, air quality, and the other factors that go into determining reasonable
progress. As discussed in section 11(F), because of the regional, multistate nature of visibility impairment in Class I
areas,44 EPA recommends that these analyses and the determination of the extent of emissions reductions needed
from individual States be developed and refined through multistate planning efforts using the best available
technical tools, such as regional-scale modeling."
638
-------
CENRAP modeling).
Dr. Gray stated that the EPA relied largely on the emissions inventory used by CENRAP12 and
made minimal adjustments based on definitive changes at the facilities in question, i.e. new units
and new controls. EPA lowered emissions based on new controls and actual emissions; it did not
adjust emissions upwards to reflect increases relative to the CENRAP modeled emissions. The
model results indicated that sulfate is the main constituent that contributes to visibility
impairment at the modeled Class I areas for both the best 20% (B20) and worst 20% (W20) days.
Dr. Gray commented that by restricting their analysis to the W20 days, for which a small subset
of the 38 sources may dominate the impacts, visibility impacts from other sources that occur on
days that are not part of the W20 will be "lost" in this analysis. A BART analysis for some of
these "other sources" (which would focus on the maximum or 98th percentile13 24-hour average
impact from each individual source) could potentially show significant visibility impacts (above
the 0.5 dv threshold, for example). A moderate concentration impact on a moderately impaired
day (not one of the W20 days) will have a larger dv impact than the same concentration
increment on a W20 day.
According to Dr. Gray, the EPA's RPG analysis only identifies the source (or sources) that
contribute significantly to the W20 days and therefore may only control the W20 days without
sufficiently controlling emissions from other sources that contribute significantly to non-W20
days. The analysis also needs to consider non-W20 days and must confirm that the proposed
controls are also effective during non-W20 days. After significant progress is made on reducing
visibility impacts on the W20 days, including the proposed Texas controls (and presumably other
reductions to meet the desired glide path), there may be other "uncontrolled" Texas sources that
still cause significant impacts on other non-W20 days (which may then become the new W20
days - one cannot simply focus on the W20 days when evaluating a control plan). In the long-
term, after the proposed controls are placed on the sources affecting the W20 days, many of the
days slightly below W20 may still have large visibility impairments (if a different mix of sources
contribute to their impacts than for the W20 days which were "controlled"). Furthermore,
reasonable progress also requires a demonstration of no degradation on the best 20% days (B20
days).
Dr. Gray stated that the EPA should consider the merits of adding additional sources to the set of
proposed sources for control based on a broader definition of "impacts" (i.e. on more than just
the W20 days). Adding more sources to the more detailed apportionment modeling may result in
greater visibility benefits on the W20 days and also on other non-W20 days, which ultimately
will need to be controlled as well (lest they become W20 days). This approach is ultimately more
consistent with the Clean Air Act mandate to ameliorate all anthropogenic visibility impairment.
Footnotes:
12 EPA noted that the 2018 emission projections that were used in the CENRAP modeling were prepared in 2006
and were based on a number of assumptions at that time regarding future controls (including the CAIR program).
The effect that current rules or proposed rules (including CSAPR and MATS) will have on 2018 emission rates is
uncertain but appears to include few additional major SO2 reductions. FIP TSD at A-16, A-45.
13 The 98th percentile metric is typically used with CALPUFF. Results from CAMx have been interpreted using the
1st high value. 79 Fed. Reg. 74,848.
639
-------
Comment: Commenter states that our analysis focused only on the visibility impacts from
sources on the 20% worst days. Focusing only on average visibility impacts on the 20% worst
days ignores larger impacts from the sources identified by EPA for additional analysis and
impacts from other sources that may be significant on other days. An analysis for some sources
not analyzed for controls using CALPUFF and focusing on the 98th percentile or maximum
impacted day could potentially show visibility impacts above the 0.5 dv threshold applied to
BART sources based on CALPUFF modeling to determine if the contribute to visibility
impairment. The commenter states that EPA should also consider analyzing additional facilities
for controls based on consideration of visibility impacts on days not included in the 20% worst
days. While the controls EPA identified based on impacts on the 20% worst days will improve
visibility on the most impacted days, there may be days with significant but slightly less
visibility impairment outside of the 20% worst days that are impacted by different sources not
addressed by this action. Those days could possibly become the new 20% worst days in the
future. To improve visibility in a manner consistent with the Clean Air Act mandate to
ameliorate all anthropogenic visibility impairment, impacts on days outside of the 20% worst
should be considered.
Response: Under the reasonable progress and long-term strategy requirements of the Regional
Haze Rule, the state or EPA in promulgating a FIP must establish reasonable progress goals that
provide for improvement on the most impaired days, demonstrate that the established goals are
reasonable and develop coordinated emission management strategies to achieve those goals.506
The most impaired days are defined as the average visibility impairment for the 20% of
monitored days in a calendar year with the highest amount of visibility impairment.507 Because
the rule focuses on improving visibility on the most impacted days, we believe it is reasonable
and appropriate to focus our analysis on sources that significantly impact visibility on those 20%
worst days. While we generally agree with the commenter that this may ignore visibility impacts
from sources that impact visibility on days other than the most impaired days, visibility
impairment on the current 20% worst days will be reduced as a result of controls implemented to
address visibility impairment for this first planning period, and we believe that in the future the
most impaired days may shift and be impacted by different sources. Analysis and development
of future regional haze SIPs for future planning periods can aim to address those sources that
impact any new set of most impaired days. Furthermore, targeted reductions at those sources
that significantly impact the most impaired days will also result in improved visibility on days
outside of the most impaired days.
CALPUFF modeling is used to provide estimates of the maximum visibility impacts from a
source based on maximum emissions and simplified chemistry, irrespective of the relationship to
the 20% worst days. It is possible that CALPUFF modeling of some of the subset of the 38
sources identified based on Q/d that were not analyzed for additional controls could show
significant impacts on the maximum or 98th percentile day, but our CAMx photochemical
modeling (which includes all emissions sources and has a realistic representation of formation,
506 40 CFR 51.308(d)(1)
507 40 CFR 51.301, This is the definition in the Regional Haze Rule, but it contains an obvious typographical error.
It should be interpreted to mean that visibility on the most impaired days is defined as stated.
640
-------
transport, and removal processes of particulate matter that causes visibility degradation) provides
additional information that allows for the identification of the sources with the greatest impacts
on the 20% worst days.
15.1 Context for Evaluation of Modeling Results (CALPUFF vs. CAMx)
Comment: Context for Evaluation of Modeling Results [Gray (0070) p.8]
Dr. Gray noted that the modeling performed for EPA's FIP TSD is inherently different from that
performed in the context of most BART analyses, and thus the two cannot be straightforwardly
compared. EPA's discussion underestimates the differences between the two, later leading it to
undervalue the results of its modeling - both the impacts from Texas sources and the benefits of
controlling them. (FIP TSD, Section A.3)
Dr. Gray described EPA's approach, noting that on pages A-35 through A-40 of the FIP TSD,
EPA discusses five issues related to the comparison of the modeled visibility impact levels from
Reasonable Progress (RP) analyses using CAMx (such as conducted by EPA for this FIP) to
modeled visibility impacts from BART analyses that typically rely on the CALPUFF model.
First, according to EPA, an RP analysis is designed to control visibility impacts from
contributing sources on peak (W20) days, whereas a BART analysis is designed to control peak
visibility impacts from individual sources (whenever they occur). As such, the two types of
analysis are addressing fundamentally different questions, which make the model results not
directly comparable.
Second, different metrics are extracted from each type of analysis. The estimated visibility
impact for an individual source when using the RP methodology will represent the visibility
impact (extinction) from the source averaged over the W20 days (typically 72 days), whereas the
BART analysis measures visibility impact in terms of the maximum or 98th percentile impact
(typically the 8th highest daily averaged extinction for each year). As EPA noted, "RP metric
results (average impact over the 20% worst days) could easily be several times less than the
CALPUFF based BART metrics (1st or 8th high single day impact)."
Third, there is also a significant difference in how emission rates are input to the models in the
two different analyses. BART guidelines require that maximum 24-hour (or hourly) average
emission rates be modeled by CALPUFF over a 3 to 5 year period. The RP analysis uses annual
average emission rates as input to the CAMx model, which is typically run for only one year.
After examining this issue, EPA concluded that, "RP results using the same metrics would be
50% or less than the BART based results just due to the differences in emissions modeled for a
facility."
Fourth, although CAMx uses a significantly more complex chemical mechanism than
CALPUFF, EPA recognized that it is impossible to determine how the different chemical
mechanisms would affect the modeled estimates of visibility impairment.
641
-------
Fifth, the use of either "clean" or "dirty" backgrounds influences two parts of the analysis:
modeling and determining deciview (dv) impacts or benefits. Generally, a "clean" analysis
measures a source's impact against natural background conditions, while a "dirty" analysis
includes the impacts from other sources. EPA states that, "A facility's visibility impairment
impacts are substantially lower with a dirty background analysis compared to a clean background
analysis." With regard to modeling, EPA states that CALPUFF is a clean analysis; the natural
background pre-cursors react only with emissions from that source. CAMx, on the other hand,
typically models all sources together. As EPA describes, this "limits the amount of ammonia
(and other pre-cursors) that are available to react with the specific facility emissions," and can
therefore lower the impact of any given source. EPA notes that "Due to the inclusion of all these
other sources at 2018 estimated emission levels, the estimated impacts from a [given]
source...are less than the results" would be with only the given source.
Converting extinction to deciviews is also affected by the chosen background. In the specific
example presented in the TSD FIP, the exact same extinction impact yielded a deciview impact
three times greater with a "clean" background analysis than with a "dirty" background analysis.
EPA explains that "Results based solely on a degraded background, will rarely if ever
demonstrate an appreciable effect on incremental visibility improvement in a given area. Rather
than providing for incremental improvements towards the goal of natural visibility, degraded
background results will serve to instead maintain those current degraded conditions. Therefore,
the visibility benefit estimated based on natural or "clean" conditions is needed to assess the full
benefit from potential controls." EPA addresses the issue by calculating the deciview impacts
based on annual average natural conditions.
Dr. Gray commented that, as EPA noted, it is difficult to compare the CAMx visibility results
from its RP analysis to CALPUFF visibility results from a typical BART analysis. There are
significant differences between the two that ultimately cannot be addressed in a generalized
comparison. Nevertheless, since both analyses provide visibility improvements in terms of
deciviews, it is worth discussing how comparable visibility impacts or benefits would be
represented in the output metrics using each methodology.
According to Dr. Gray, EPA's distinction between the policy purposes of BART and RP
analyses (to reduce peak impacts from a source versus reducing impairment on the W20 days,
respectively) obscures their shared underlying goal - the Clean Air Act goal of eliminating any
anthropogenic visibility impairment. As discussed above, EPA's emphasis on the W20 metric
restricts its analysis and causes it to lose sight of sources that do contribute to visibility
impairment. Furthermore, the difference that EPA asserts between the respective purposes of
BART and RP is not always, or even typically, the case. In some previous source-specific RP
analyses, EPA and states have used CALPUFF to model visibility benefits, and have used largely
the same methodology as in BART modeling (i.e. use of 24 hour or hourly maximum emissions,
a "clean" background condition, and a maximum or 98th percentile metric).14
Dr. Gray stated that the EPA addressed the need for using natural or "clean" conditions as
background levels when estimating the benefits of controls by re-calculating the deciview
improvements for the various control measures under consideration using average natural
conditions (estimated from IMPROVE monitoring data). Tables A-6.2 through A-6.5 of the FIP
642
-------
TSD show that the estimated visibility benefits (deciview improvements) based on annual
average natural conditions are much greater than the estimated visibility benefits based on the
"dirty" 2018 modeled background for each of the evaluated control measures. For example, from
Table A-6.2 it can be seen that the cumulative dv improvements in all 19 modeled Class I areas
using natural conditions are between 3.4 and 4.7 times higher than the dv improvements
estimated using the modeled 2018 background for the same controls, and that the ratio is greater
than 4.0 for all facilities (and control measures) other than Tolk (ratio = 3.4).
Dr. Gray stated, as EPA discusses, it is necessary to compare results against natural conditions
because comparisons to the existing or near-term degraded background will obscure the benefit
from controls. However, EPA's approach underestimates impacts in two ways. First, as EPA
notes, its modeling in CAMx uses emissions from other sources, which can lower the extinction
impact of each source. Second, EPA compares the resulting extinction values to annual average
natural background conditions. Using the 20% best days of natural conditions produces impacts
that are approximately 30% higher.15
Dr. Gray stated that even after the visibility improvements are re-calculated to consider "clean"
background conditions, the estimated visibility benefits using EPA's RP analysis (CAMx) will
still be lower than the benefits that would be estimated using a BART analysis (CALPUFF) for
the same control measure due to a difference in metrics (BART relies on the maximum or 98th
percentile impact from CALPUFF, whereas EPA's interpretation of its CAMx results consider
the average impact during the W20 days), and also due to a difference in how the modeled
emission rates are specified (annual averages versus maximum emission rates).16 The model
results shown in Tables A.4-la-c can be used to compare the average and maximum modeled
extinction on W20 days for the top ten impacting sources at WIMO, GUMO, and BIBE. As
expected, the maximum impacts were much higher (by a factor of 2 to 3 or more) than the
average for the W20 days. Because this is only the maximum impact during the W20 days (not
necessarily the source's maximum impact overall) this still may underestimate the difference
between EPA's presented results and those that would be used in a BART analysis.
Thus, according to Dr. Gray, there are at least two effects to combine: the difference in metrics
(the use of maximum or 98th percentile dv improvements overall will be at least a factor of 2
greater than average dv improvements on the W20 days), and the modeled emission rate
difference (using maximum emission rates, rather than average emission rates, will yield
visibility impacts that are at least twice as large). Each of these provides a factor of two or more;
thus, it can be concluded that the visibility benefits estimated from a BART analysis will be at
least 4 times the visibility benefits obtained from an RP analysis (both using natural background
conditions) for a comparable modeled source.
A spreadsheet containing a summary of the deciview improvement on average W20 days was
provided with comment 0070.
Footnotes:
14 See e.g. EPA's rulemakings regarding the regional haze plans for Montana, North Dakota, Arizona, and
Colorado. 77 Fed. Reg. 23988, (Apr. 20, 2012); 76 Fed. Reg. 58570 (Sept. 21, 2011); 79 Fed. Reg. 9318 (Feb. 18,
2014); 77 Fed. Reg. 18052 (Mar. 26, 2012).
15 See results in EPA's spreadsheet: "TX116-007-_33_Vis_modeling_summary.xlsx"
643
-------
16 EPA recognized this difference in suggesting that Texas use double the annual average emission rates in its BART
CAMx modeling, as an estimate of maximum emission rates. Texas SIP, Appendix 9-5, p. 2-6.
Further, as described in the attached Report from Dr. Gray (0070), EPA undervalues the impacts
from Texas sources and the benefits of controlling them in its comparison between the CAMx
and CALPUFF visibility models. Visibility results from CAMx and CALPUFF are difficult to
compare because they model visibility impairment in fundamentally different ways. Here, EPA
relied in part on CAMx modeling results to screen out 29 sources from consideration for
reasonable progress controls; EPA concluded that the CAMx modeling showed impacts that
were too low to justify further analysis.
A rough comparison between the models shows that, in general, results obtained in CALPUFF
would be at least 4 times as large as results from using the CAMx model. See Gray Report at 9-
11. If EPA had considered the magnitude of the impact from these 29 sources using CALPUFF,
or at least attempted to translate the CAMx results to CALPUFF results, the values would show
significant impacts that should have led EPA to consider controls for these 29 sources.
Response: We agree that we and some states have used CALPUFF to model visibility benefits
as part of the reasonable progress analysis, and have used largely the same methodology as in
BART modeling (i.e. use of 24 hour or hourly maximum emissions, a "clean" background
condition, and a maximum or 98th percentile metric). This approach can provide information on
the relative visibility benefits of controls to inform the evaluation of cost-effectiveness as part of
the four factor analysis and has the benefit that it is immediately comparable to modeling used
for BART determinations. However, this approach models the impacts from the single facility
with limited chemistry and focuses on the maximum impacts from a source rather than the
visibility impairment on the 20% worst days. Due to the large time and resource demands of
photochemical modeling, CALPUFF modeling has been used to provide information on visibility
benefits of controls on individual sources for reasonable progress in some cases, but a full
photochemical modeling approach using CAMx, as was utilized here, provides information to
assess visibility impacts and visibility improvement for the 20% worst days. We discuss why we
believe it is reasonable and appropriate to focus our analysis on sources that significantly impact
visibility on those 20% worst days in more detail in the response above. We address the selection
of the CAMx photochemical model in our analysis over the CALPUFF model is a separate
response to comment in the modeling section of this document.
We agree with the comment that "modeling performed for EPA's FIP TSD is inherently different
from that performed in the context of most BART analyses, and thus the two cannot be
straightforwardly compared." In the FIP TSD, we discuss the difficulties in attempting to
compare CAMx results for reasonable progress to CALPUFF model results for BART due to
differences in models, model inputs, and metrics used.508 While we estimate that due only to
emission differences between CALPUFF modeling for BART and CAMx modeling for
reasonable progress, CAMx model results for reasonable progress would be 50% or less, we also
note that "due to chemistry mechanisms, pre-cursor concentrations, and other differences that
would introduce variation in comparisons it is impossible to come up with an answer on how this
508 FIp TSD at A_35
644
-------
issue should be factored into a comparison of model results from CAMx and CALPUFF except
to conclude that they would likely give differing values."509 We agree with the commenter that
many of these differences result in CAMx modeled visibility impacts and benefits that are much
lower than the CALPUFF modeled visibility impacts and benefits relied on in other actions.510
However, as discussed above and in the FIP TSD, due to all of the differences in the CALPUFF
and CAMx model results, it is not possible to directly compare these model results. Therefore, it
is not appropriate to use these comparisons and thresholds established for CALPUFF modeling
for BART to determine the significance of the modeled visibility impairment and benefits using
CAMx.
To evaluate the projected visibility benefits of controls from our CAMx modeling in our cost
evaluation, we considered a number of metrics, such as change in deciviews under 2018
projected levels of air pollution at the three Class I areas and under estimated natural visibility
conditions, change in light extinction, and change in the percentage of total light extinction. We
also considered the visibility benefit of emission reductions from recent actual emission levels
versus CENRAP 2018 projected emission levels at these sources. As we discuss further in our
FIP TSD and in responses in our RTC document, to provide context regarding the significance of
individual source impacts, we compared the individual source impacts with CENRAP source
apportionment modeling results for impacts from all emission sources within a state and impacts
from all emission sources within a state within a specific source type. We also compared these
individual source impacts to the impact levels used by the states for triggering consultation with
another state about its overall impacts, and the estimated range of anticipated visibility benefits
resulting from required controls in other actions.511 We evaluated recent FIPs that included
controls for reasonable progress using CALPUFF modeling512, and conservatively estimated that
just based on emissions and metric differences, the visibility benefits in those actions would be
well in the range of CAMx modeled visibility benefits for the required controls in this action.
As we discuss in detail elsewhere in this document where we respond to comments on our
identification of sources to evaluate for additional controls, we utilized a 0.3% contribution to
total visibility impairment on a unit-level basis to identify those sources with the largest visibility
impacts. We then considered the estimated visibility benefit of controls alongside the four
statutory factors to identify cost-effective controls that will achieve reasonable visibility benefits
required during this planning period towards the national goal. We disagree with the commenter
that we undervalued the results of our CAMx modeling. We discuss our determination
concerning the impacts and potential visibility benefits from controls on the "other 29" sources
that were screened out of additional control analysis in separate responses to comments in this
section of this document.
Regarding the comment concerning the use of annual average natural background conditions
versus the 20% best days of natural conditions, we agree that use of the 20% best days would
result in higher calculated deciview impacts. We calculated the "natural background" using the
509 FIP TSD at A-37
510 For example the use of maximum or 98th percentile dv improvements compared to average dv improvements on
the W20 days and maximum emissions compared to average emissions.
511 FIP TSD at A-75
512 Wyoming 78 FR 34785-34789, Arizona 79 FR 52464-52477
645
-------
20% worst days natural conditions, annual average natural conditions, and the 20% best days.
Initially, we used the three available values for natural conditions to perform this calculation.
We note that EPA provided additional guidelines with narrowly defined flexibility regarding the
averaging period to be used for calculating natural background. These guidelines indicate that
the states may use either annual average natural conditions or the average of the best 20% days
for natural conditions.513 Ultimately, we determined that the annual average value was the more
appropriate, conservative approach for this specific analysis at this time. We note this is
consistent with the selection of natural visibility background used by Texas in their BART
screening modeling using CAMx.514
15.m Consideration Of Additional Controls And Impacts At Other Class I Areas
General summary: We received comments that we should have required additional controls on
sources above what we proposed in our FIP to assure even greater reasonable progress. The
commenter states that certain controls are reasonable and consistent with the proposed controls
when impacts at Class I areas other than the Texas Class I areas and Wichita Mountains are
considered. In addition, the commenters state that some specific facilities fall above the 0.3%
impact threshold for impacts at other Class I areas and should have been evaluated for controls.
Visibility impairment from the 29 sources not analyzed for controls are still significant and
should be required. Commenters stated that EPA evaluated controls for Parish and Welsh but
did not require controls despite significant visibility benefit and reasonable costs. In addition,
EPA should have analyzed oil and gas sources and NOx controls for certain point sources in
Texas.
15.n Impacts At Other Class I Areas and Consideration of Additional Controls
Comment: Impacts at Carlsbad Caverns [Earthjustice (0067) p.46]
Earthjustice et al. stated that degrading visibility conditions on the best days at Carlsbad Caverns
provides further support for controls at those Texas sources that have the highest impacts at that
Class I area. Tolk, Big Brown, Big Spring Carbon Black, Sommers-Deely-Spruce, and
Harrington are the top 5 contributors to visibility impairment at both Guadalupe Mountains and
Carlsbad Caverns, and each is over EPA's 0.3% contribution threshold. Under the Regional Haze
Rule, EPA must adopt emission limits necessary to ensure no further visibility degradation on
the least impaired days at all Class I areas. Evidence in the record demonstrates that emission
controls at Tolk, Big Spring Carbon Black, Sommers-Deely-Spruce, and Harrington are cost
effective.
Earthjustice et al. concluded that given the impacts of these sources at Carlsbad Caverns, there is
no reason not to require emission reductions at those sources.
513 Paisie, JW. . Regional Haze Regulations and Guidelines for Best Available Retrofit Technology (BART)
Determinations. Memorandum to Kay Prince, Branch Chief EPA Region 4. July 19, 2006.
514 Texas regional haze SIP, appendix 9-5
646
-------
Response: We approved the RPG for the 20% best days at Carlsbad Caverns in a separate
action on November 27, 2012 (77 FR 70693). The commenter presents the highest modeled
visibility impacts on the 20% worst days but the comment is focused on impacts on the 20% best
days. The commenter also incorrectly compares facility-wide estimated impacts to the unit level
0.3%) threshold that we selected to identify sources for additional control analysis. We note that
this action does require controls at Tolk and Big Brown that will result in some reduction in
visibility impairment at Carlsbad Caverns on the 20% worst and 20% best days. Furthermore,
the coal-fired JT Deely units at Sommers-Deely-Spruce complex are scheduled to shut down by
2018 further reducing visibility impairment at Carlsbad Caverns and other impacted Class I
areas. Overall, visibility impairment on the 20% best days at Carlsbad Caverns from Texas
sources is projected to decrease and therefore Texas sources are not anticipated to contribute to
any degradation in visibility at Carlsbad Caverns on the 20% best days. Comments concerning
additional controls at specific facilities mentioned above are discussed in more detail in separate
response to comments. Comments concerning consideration of impacts at Class I areas other
than Big Bend, Guadalupe Mountains, and Wichita Mountains are addressed below in this
section.
Comment: Commenter 0053-26 (Sierra Club) questioned why the Caney Creek area in Arkansas
is being left out of the rule.
[Gray (0070) p. 15] Dr. Gray stated that the EPA's source selection failed to consider impacts at
Class I areas outside of Texas and Oklahoma, both individually and cumulatively. Throughout
the analyses in Appendix A, EPA has focused on the modeled visibility impacts at WIMO, BIBE
and GUMO, despite the fact that the model estimated the visibility impacts at 16 additional Class
I areas in the states surrounding Texas. Examination of Figures A.l-3a-h, as well as the CAMx
modeling results (impacts on W20 days) at all modeled Class I areas, reveals that there were
significant visibility impacts from Texas facilities at Caney Creek Wilderness Area (CACR),
Carlsbad Caverns National Park (CAVE), Breton Wilderness Area (BRET), Salt Creek
Wilderness Area (SACR), Hercules-Glades Wilderness Area (HEGL), Upper Buffalo Wilderness
Area (UPBU), and White Mountain Wilderness Area (WHIT), in addition to the WIMO,
GUMO, and BIBE. Utilizing more Class I areas in the analysis would assist in the identification
of sources that contribute significantly to Class I areas other than WIMO, GUMO, and BIBE.
The benefits of applying controls on these sources will not be apparent if the focus is only on
those Class I areas.
In fact, in terms of percent contribution to extinction, Dr. Gray stated that 14 of the 38 modeled
sources had the highest impacts at Salt Creek (SACR; 9 sources) or Caney Creek (CACR; 5
sources). These sources are listed below.
647
-------
Table 5: Sources nidi Highest % Ceiitriliitwa al CACR or SACK
KEYSTONE PLANT
_ .V.
X A
N/A
SACR
GOLDSMITH GASOLINE PLANT
c :t%
X A
X A
SACR
1 ' !¦¦¦¦¦!¦ : 1 . ' ; • ...
' 18%
X A
•>;?%
SACR
R11EP70X OAS PLANT
0 20%
X A
N/A
SACR
SHERHAN PLANT
a
N A
X A
SaC ?.
BOP.GEP: C AP30X SLAl K PIT
S 25%
X A
0.11*4
SAC?,
BIG SPRING CARBON BLACK
0.19%
0.13%
SACR
TOLK STATION
1.01H
1,51*
SACR
HARRINGTON STATION
i.:§%
ij#w
0J4H
SACR
w'mh «f motels mat not
2, Esiuatri
Dr. Gray stated that the facility and unit impacts above EPA's 0.3% threshold are listed in bold.
Four of the facilities with units having the highest impacts at CACR or SACR were evaluated for
additional controls (Martin Lake, Monticello, Tolk, and Welsh). The benefits from controls on
these units will be underestimated if only WIMO, BIBE, and GUMO are evaluated.
Additionally, when impacts to SACR are reviewed, Harrington has impacts above EPA's 0.3%
threshold. Even by EPA's criteria, there is no justification for not reviewing the Harrington units
for controls.
Dr. Gray stated that the summary tables in the TSD FIP that display the estimated visibility
benefits at WIMO, GUMO, and BIBE for implementing source controls should also include the
cumulative impacts for all 19 modeled Class I areas. For example, the cumulative impacts at all
19 Class I areas should be added to Tables 20 through 22 (Section 13) and Table A.6-5 (Section
A.6; the suggested additional data for Table A.6-5 are shown below).
Footnote:
20Based on spreadsheets "TX116-007-_29_Source_selection_analysis_TX_RH-l-31-14.xlsx" and "TX116-007-
_30_Source_selection_analysis_TX_RH-es_l-31-14.xlsx", tab "All Class I areas data".
Earthjustice et al. commented that we should have considered cumulative visibility impacts and
impacts at Class I areas outside of Texas and Oklahoma. Our CAMx model results also show
significant impacts at Class I areas in Arkansas, New Mexico, Louisiana and Missouri. The
commenter states that consideration of impacts at these other Class I areas would result in
identification of additional sources with reasonable controls and that focusing the analysis on
Texas and Oklahoma will result in a failure to identify these additional sources, and
underestimate of the benefit of controls. The commenter provides a table showing that in terms
of percent contribution to extinction, 14 of the 38 modeled sources had the highest impacts at
648
-------
Salt Creek (SACR; 9 sources) or Caney Creek (CACR; 5 sources). Additionally, when impacts
to SACR are reviewed, Harrington has impacts above the 0.3% threshold and therefore there is
no justification for not reviewing the Harrington units for controls. The commenter also states
that we should provide cumulative visibility benefits to the visibility benefit tables in the FIP
TSD.
Tabi« 3; Sources with Hiffctsi % Costrftmtion at CACR or SACK*
Faaiir-.
i'fUt1
Cm?
Acijti
i i.v . I Area
N"
C'&CV'
W* * V 'WW MV A M'ln "t *
jAA j
¦— ¦_
"LrivXw
HW PIRKET POWER ?L7
1.05H
1,15*
0,20%
CXCP
rot**
0.70H
CACR
MAITW LAKE ELECTRICAL STATION
Uf*
1,11%
C-l'P.
J2EY5T0KE PLANT
c c«%
—
MA
SACR
golds:.!ithgasoiine plant
X A
N/A
SACR
BOP.&EEl C AF30N BLACK
? -.5%
X A
0.18%
SAC?.
FTLLEPTON C-A> PLANT
e
N A
N'A
SACR
SHEErLAX PLANT
G 25%
X A
XA
SAC1
BOP.GER C ARSON BLAC X PIT
: :s4.
N'A
SAC P.
BIG SPRING C AE30N BUCK
0.19%
0.13%
SACR
TOLK STATION
1.11*
0.49H
SACR
HARRINGTON STATION
L20H
0.4OH
®J4%
SACR
1. P* :ox:ec nkchiaut unpad: wee not
estomaiel generally law multiple emission points.
2. Estimated ait impacts based on actual emustcos per unit
Response: We focused our control analysis on the Texas Class I areas and the Wichita
Mountains. As discussed in more detail elsewhere, we are disapproving portions of the Texas
and Oklahoma regional haze SIPs, including the Texas long-term strategy consultation, the
Oklahoma reasonable progress consultation, the Oklahoma established reasonable progress goal
for Wichita Mountains and the Texas reasonable progress/long-term strategy analysis and
consideration of reasonable controls at Texas sources necessary to establish the Texas and
Oklahoma reasonable progress goals. In developing a FIP to address the deficiencies in the
Oklahoma and Texas SIPs, we had to analyze the visibility impacts and the availability of
reasonable progress controls at Texas sources that impact visibility at the two Texas Class I areas
and the Wichita Mountains and establish reasonable progress goals including consideration of an
appropriate reasonable progress control analysis for these areas.
As discussed in the Texas TSD that accompanies the proposed FIP, during consultations for
Arkansas and Missouri Class I areas, the participating states determined that the projected 2018
CENRAP modeling and other findings based on existing and proposed controls arising from
local, state, and federal requirements indicated that these Class I areas were on the glidepath and
projected to more than meet the URP goal for the first implementation period ending in 2018.
Arkansas Department of Environmental Quality (ADEQ) and Missouri Department of Natural
Resources (MDNR) determined that additional emissions reductions from other states were not
649
-------
necessary to address visibility impairment at their Class I areas for the first planning period, and
we approved this portion of the AR RH SIP.515 In this situation, based on consideration of the
overall estimated visibility improvements anticipated by 2018, the relative impacts from
Arkansas sources compared to the total impact from sources in other states, and the fact that
additional reductions were anticipated from Arkansas sources due to BART, we found that this
was reasonable. We note that we disapproved portions of the BART analysis and the reasonable
progress analysis in the Arkansas RH SIP and have proposed a FIP to address these deficiencies,
including BART and reasonable progress controls that will significantly reduce the visibility
impairment to levels well below the uniform rate of progress at Caney Creek.516 Similarly,
Louisiana determined that impacts from Texas sources on their Class I areas were not significant
enough to warrant additional controls for this planning period and we approved that
determination. We also approved the consultation in New Mexico, as well as the RPGs for New
Mexico's class I areas in a previous, separate action.517 And at that time, we did not receive any
comments on that final approval for consultation or comments concerning the impact from
sources in Texas. As discussed in that action, based on our review of the New Mexico regional
haze SIP, including the Western Regional Air Partnership (WRAP) modeling and emission
inventories and additional reductions anticipated due to BART, we determined that the
established RPGs for 2018 for the New Mexico Class I areas were reasonable and that "we
anticipate additional visibility improvement in 2018 beyond the modeled visibility conditions."
In light of the determinations promulgated in these previous actions and the identified
deficiencies in the Texas and Oklahoma Regional haze SIPs, we determined that it was
appropriate to focus our analysis on the Texas Class I areas and Wichita Mountains and consider
the impacts and potential visibility benefit from controls at these three Class I areas to address
these deficiencies. We note that many of the sources identified by the commenter as having their
largest impacts at Caney Creek and Salt Creek also have similar impacts at the Class I areas of
interest and were included in our control analysis. We also note that visibility improvements will
occur at Class I areas in Arkansas, Missouri, Louisiana and New Mexico as a result of the
controls required in this action For example, the benefit at Caney Creek from the required
controls were estimated at Big Brown (0.179 dv for each unit), Martin Lake (0.35 — 0.44 dv),
and Monticello (0.189 - 0.264 dv). In our analysis, we provided estimates of visibility benefits
from the all the FIP required controls at a number of Class I areas in other states.518 Specific
comments concerning Pirkey and Harrington Station are addressed in separate responses to
comments in this section. In focusing on the impact of Texas sources on Wichita Mountains and
the Texas Class I areas, we addressed the identified deficiencies and establish RPGs for these
three Class I areas considering our technical analysis and the emission reductions required in this
action. We expect New Mexico, Arkansas, Louisiana, and Missouri to consider remaining
impacts from Texas sources on their Class I areas including the information on visibility impacts
from specific sources provided by our analysis, as well as incorporate corrections and updates to
emission reductions in consultations and development of their regional haze SIPs for the next
planning period.
515 77 FR 14604 (March 12, 2012)
516 80 FR 18944
517 77 FR 70693 (November 27, 2012)
518 See TX-116-007-_33_Vis_modeling_summary.xlsx
650
-------
Comment: EPA Inappropriately Eliminated Sources from Detailed Four-Factor Analysis
of Reasonable Progress Controls. [Earthjustice (0067) p. 45]
Furthermore, even if the CAMx modeling results are considered alone, there would be
significant visibility benefits at Big Bend, Guadalupe Mountains, and Wichita Falls associated
with controlling some of the other 29 sources and some of the remaining 1600 sources. The total
visibility impacts from the other 29 sources collectively account for similar visibility impacts as
the total of the nine larger selected sources. While the other 29 sources each have smaller
individual visibility impacts than sources like Martin Lake, Big Brown, or Monticello, as Dr.
Gray explains, that is not surprising since the facilities EPA is proposing to control are amount
the top 15 SO2 emitters in the entire United States.
Earthjustice et al. stated that there are additional control measures or emission limitations that are
very likely cost effective and could be readily implemented for Texas sources that met EPA's
0.3% contribution criteria to be evaluated. As discussed in more detail in the Stamper Report,
these include: new SO2 controls at Harrington station, upgraded SO2 controls at Oklaunion
Power Plant and H.W. Pirkey Power Plant, evaluation of control measures for PPG Industries
Works No. 4, enforceable requirements for shutdown of two units at the Sommers-Deely-Spruce
Complex, reasonable controls for Big Spring Carbon Black Plant, and pollution reduction
measures for the TNP One Steam Electric Station. Given that most of the Class I areas impacted
by Texas sources are not on track to achieve natural background visibility by 2064, EPA should
amend its proposed FIP to include cost-effective pollution control requirements for the seven
additional sources evaluated in the Stamper Report.
Response: Comments concerning additional controls at specific facilities mentioned above are
discussed in more detail in separate response to comments in this section. Comments concerning
the "other 29 sources" and "remaining 1600" are addressed in a separate response to comment
below.
Comment: [Gray (0070) p. 3] Dr. Gray noted that EPA Region 6 has developed a technical
support document (TSD) for the Oklahoma and Texas Regional Haze Federal Implementation
Plans (FIP TSD). In the FIP TSD, EPA presents the technical support for its proposed control
plan for Texas point sources of nitrogen oxides (NOx) and sulfur dioxide (SO2). Dr. Gray
provided an overview of EPA's visibility modeling, EPA's narrowing process for source
selection, and EPA's evaluation of the visibility benefits from controls. Dr. Gray concluded:
A. The controls EPA is proposing to require will provide significant visibility benefits
and should be required in the final rule. EPA provides ample technical support for
controlling the proposed 14 units at 7 facilities. In fact, EPA's approach ultimately
undervalues the benefits from these controls.
B. More sources should have been reviewed for emissions controls. EPA's analysis
narrows its scope several times, limiting the sources that it reviews for controls from over
1600 to just 9. In so doing, it fails to evaluate reasonable progress controls to limit
emissions responsible for more than half of the impairment from Texas point sources.
651
-------
C. The controls EPA is proposing not to require will also provide significant visibility
benefits and should be required in the final rule.
The overview of EPA's approach provided by Dr. Gray was not excerpted.
Response: We appreciate the commenter's support for the required controls. Our responses to
specific comments from Dr. Gray on EPA's approach and additional controls are included
elsewhere in this document.
Comment: Impacts at WIMO, GUMO, and BIBE [Gray (0070) p. 13]
Dr. Gray stated that, in reviewing the impacts at WIMO, GUMO, and BIBE, EPA should have
used a lower threshold and included more sources for control review. As discussed above, there
are a large number of sources collectively contributing to cumulative impairment.
Dr. Gray stated that the CAMx modeling results show that these "other 29" sources all have
much smaller individual visibility impacts than sources like Martin Lake, Big Brown, or
Monticello. This is hardly surprising, since those facilities are among the top 15 SO2 emitters in
the US. 18 However the combined impacts from the collection of 29 sources add up to a
significant visibility impact at WIMO, GUMO, and BIBE (in addition to other Class I areas).
Dr. Gray stated that Figures A.3-la-c in the FIP TSD illustrate that Texas sources are responsible
for a substantial fraction of the extinction at WIMO, GUMO, and BIBE. Figures A.3-2 through
A. 3-4 show the estimated contributions to extinction at these three Class I areas from each of the
38 modeled Texas point sources. It can be seen from these figures, and by examination of the
source contributions for the other 16 modeled Class I areas, that a large number of different
sources have significant impacts in each of the Class I areas, including many facilities that were
not included in the "small group of sources" (as selected in Section A-4) that were considered for
control measure implementation.
According to Dr. Gray, collectively, the 29 sources that EPA eliminated contribute roughly 33%,
31%, and 24% of Texas's point source contribution at GUMO, WIMO, and BIBE, respectively
(roughly 2-4% of the total extinction). These 29 sources alone contribute more than all of the
point sources from any other state at each of these Class I areas. 19
Dr. Gray stated that these figures clearly demonstrate that there would be additional visibility
benefits at WIMO, GUMO and BIBE associated with evaluating controls for the "other 29"
sources. The total visibility impacts from the "other 29" sources collectively account for similar
visibility impacts as the total of the nine larger selected sources. At GUMO, the "other 29"
sources are collectively responsible for almost 3% of the total visibility impact, which is more
than the total impact from the nine selected sources combined.
By the same measure, Dr. Gray stated that the EPA's source apportionment work makes entirely
clear it was justified in further reviewing the 9 sources it chose. These sources, individually and
collectively, have significant impacts at WIMO, GUMO, and BIBE. Their impacts alone are
652
-------
greater than the point source impacts from any other state, and collectively their impacts have
increased versus the emissions that were assumed by the original CENRAP modeling.
Dr. Gray stated that overall, Texas point sources contribute 14% to the total modeled extinction
at WIMO. Examination of the modeled extinction (average of W20 days) at WIMO (Figure A.6-
le) reveals that the nine selected facilities account for about 6.6% of the total visibility
impairment at WIMO. EPA's proposed control scenario (consisting of controls on seven of the
nine selected facilities) would address 5.8% of the total visibility impairment at WIMO. The
"other 29" Texas point sources account for 4.4%, and the "remaining 1600" smaller sources
account for an additional 3.1% of the total visibility impairment at WIMO. By comparison,
Oklahoma point sources contribute roughly 3.9%.
Dr. Gray stated that the model results for BIBE are shown in Figure A.6-If. Texas point sources
contribute 8% to the total visibility impairment at BIBE. The nine selected facilities account for
about 2.1% of the average extinction at BIBE. The recommended controls would address 1.9%
of total visibility impairment at BIBE. The "other 29" (not considered for controls) account for
another 2.0% and "the other 1600" Texas sources account for the remaining 4.0% of the total
visibility impairment at BIBE. Coleto Creek alone accounts for over 6% of the total Texas point
source visibility impact.
Dr. Gray stated that Figure A.6-lg shows the model results for GUMO. Texas point sources
contribute 9% to the total extinction (average of W20 days) at GUMO, of which the nine
selected sources account for 2.4%. EPA's recommended controls would address 2.2% of the total
extinction at GUMO. The "other 29" Texas point sources are responsible for 2.9% and the
"remaining 1600" sources account for 3.4% of the total visibility impairment at GUMO. Tolk
accounts for nearly 8% of the total Texas point source impact.
Dr Gray summarized these impacts at WIMO, BIBE, and GUMO in the exhibit below (Table 2
of comment 0070). Dr. Gray concluded that clearly, review of controls is warranted at all 9
sources EPA reviewed, along with the "other 29" sources it eliminated, and at least some of the
1600+ additional point sources.
Relative Contribution to Extinction Impacts
(Table 2 provided by commenter 0070)
I 2 ;v ,E .. : f ?. ];ir * ¦
Footnotes:
18 According to EPA's 2011 National Emission Inventory (NEI), Martin Lake was the 7th largest emitter in US
-------
(68,933 tpy), the Big Brown facility was the 8th largest SO2 emitter in the US (64,198 tpy), Monticello was the 13th
largest (54,439 tpy), and Parish was the 15th largest SO2 emitter in the US (49,547 tpy).
19 See FIP TSD at A-64 through A-66 and underlying data in spreadsheet "TX116-007-_23_extinction_charts.xlsx".
[Gray (0070) p. 17] Dr. Gray stated that combining the results from the original 2018 CAMx
modeling for the 38 facilities (Environ, 2013) with the results from the later 2018 modeling with
high controls on the 9 selected facilities (Environ, 2014) provides clear evidence that more
widespread emission controls would provide greater improvements in visibility at the Class I
areas within the region. The average percent change to the extinction coefficient on the W20
days at four key Class I areas (BIBE, CACR, GUMO and WIMO) from the 9 selected facilities
under the assumed high control level (see Section A. 5) was applied to the modeled extinction
coefficients from each of the remaining 29 of the original 38 facilities. Adjusted total extinction
coefficients and deciview values were calculated and the change in deciview was examined.
Note that this analysis assumes that controls similar to those considered in the high-control
facility could be applied to all remaining 29 facilities, and that those controls would be equally
effective.21 The objective of this exercise was to examine whether more widespread controls are
likely to result in further improvements in visibility. Table 2 summarizes the results of this
scaling analysis.
T^,r y,-,-|iV)At 2# Percent W®rstDays « nil* im-rtth mt>-T,un in Yr-rtrtbtv .
:01s Hish Hidi uh .dVrUurto
n „ ~7% . lilgfc „ . taYisMSiv
fa/F-dj-tif \ outi !>b on 1 yafiv!" t'U . , ,, Hiffite «-
S1 i .-if ikOf' .-.>!¦ ;tc»liw, r , ¦ ,, V >1J • '>ti T-
t .IfUJRc' ! ,1 ,r r ., ,
un
654
-------
evaluated for additional controls and the percentage addressed by controls proposed in the FIP.
The commenter states that collectively these sources are responsible for a significant percentage
of the total visibility impairment at WIMO, BIBE and GUMO and controls should have been
considered for some of these sources. The commenter also provides a scaling analysis to assess
potential visibility benefits from controlling additional sources out of the 38 initially identified
by the Q/d analysis.
Response: We appreciate the commenter's support for the required controls and agree that the
visibility impacts from the nine facilities identified for additional modeling for control analysis
alone are greater than the combined point source impacts from any other individual state. Our
Reasonable Progress Guidance discusses the steps to follow in identifying reasonable controls
and establishing reasonable progress goals. The key pollutants contributing to visibility
impairment at each Class I area should be determined. "Once the key pollutants contributing to
visibility impairment at each Class I area have been identified, the sources or source categories
responsible for emitting these pollutants or pollutant precursors can also be determined. There
are several tools and techniques being employed by the RPOs to do so, including analysis of
emission inventories, source apportionment, trajectory analysis, and atmospheric modeling"
(page 3-1). As discussed in more detail in our proposal and in a separate response to comment in
the modeling section of the RTC document, we determined that it was reasonable to focus our
analysis on point sources of SO2 and NOx.519 This was based on review of emissions and source
apportionment results indicating that these sources were most responsible for anthropogenic
contributions to visibility impairment. We then used a Q/d analysis to identify those sources
with the greatest potential to impact visibility based on emissions and distance. Additional
analysis using photochemical grid modeling was then completed to estimate the visibility impact
from those sources. Based on consideration of facility level and estimated contributions to
visibility from units at the modeled facilities, we identified those sources that had the greatest
visibility impacts to analyze for additional controls. We agree with the commenter that
collectively the "Other 29" sources and "1,600+" sources contribute a sizeable percentage of the
total visibility impairment. However, on an individual basis, these point sources have lower
contributions and smaller potential for visibility improvements relative to the nine facilities
evaluated for additional controls. For example, the proposed controls on only 7 facilities address
5.8% of the total visibility impairment at the Wichita Mountains, while controls on all of the
"Other 29" sources would address 4.4% of the total visibility impairment. Consistent with our
guidance, we identified those key pollutants and sources with the greatest impact on visibility
impairment for this first planning period. We also note that the "Other 29" includes impacts
from San Miguel and the PPG Glass Works facility that were considered for additional controls,
and the JT Deely units that are scheduled to shutdown in 2018.
The Regional Haze Rule requires the identification of reasonable progress controls and the
development of coordinated emission control strategies in order to make reasonable progress
towards the goal of natural visibility conditions. Faced with a very large and unwieldy universe
of sources, we followed our guidance and chose an approach that focused on the portion of the
universe of Texas sources that contributed the greatest impact to visibility impairment, by
establishing a threshold of 0.3% contribution to total visibility impairment on a unit basis for this
519 79 FR 74838
655
-------
planning period, thereby identifying a reasonable set of units at nine facilities to analyze for
additional controls.520 Our four-factor analysis concluded that controls on units at seven of the
nine facilities analyzed for additional controls were required. As these controls are implemented,
the percentage impact from those facilities not controlled will become larger (on a percentage
basis) and will be analyzed in future planning periods. In other words, some of the "Other 29"
will be identified as the greatest impacting sources and should in turn be analyzed for additional
reasonable progress controls in a future planning period. This methodology can be used as a
consistent procedure to identify facilities for additional control analysis in this and future
planning periods and would ensure continuing progress towards the goal of natural visibility
conditions. The USDA Forest Service commented that "the methodology and metrics that EPA
used are the most comprehensive seen to date for any SIP/FIP in the country that we have
reviewed, and should serve as a model for future efforts to consider the contribution and/or
potential benefits of individual sources to visibility."
With regards to the scaling analysis presented by the commenter, we agree with the commenter
that additional emission reductions would result in additional visibility benefit at the Class I
areas. However, as the commenter states, the estimates presented assume that controls are
technically feasible and cost-effective for all of these sources and that those controls will result
in significant emission reductions for each facility. The additional estimated improvement
requires significant emission reductions at multiple units across 29 facilities and at all but one
Class I area (GUMO) results in less visibility improvement than controlling units at just 9
facilities. As discussed above, on an individual basis, the units at these 29 point sources have
lower contributions and smaller potential for visibility improvements relative to the nine
facilities evaluated for additional controls. Some of those remaining 29 sources may be
identified for additional control analysis in future planning periods as the sources with the largest
visibility impacts and potential for visibility benefit are controlled.
Comment: Other Considerations [Gray (0070) p. 16]
Dr. Gray stated that some of the "other 29" sources were removed from consideration after the
updated emissions inventory indicated that recent actual emission rates had been much lower
than modeled emission rates. As with the Q/D analysis above, this source selection relies on
actual emissions that are not enforceable. In particular, EPA relied on lower than modeled
emissions to justify not reviewing controls for Oklaunion, Sommers Deely Spruce, Works No. 4,
and HW Pirkey. These sources should also be included in the more detailed evaluations of
control measure effectiveness; at a minimum, the emission levels that EPA relies on should be
enforceable limits.
Dr. Gray stated that even though the RP analysis that is used to evaluate potential controls is
done on a unit by unit basis, the threshold for significance impacts should still be the total facility
impact, which is the approach EPA recommends that states take when determining whether a
facility is subject to BART. Emissions from separate units at a single facility are typically
520 As discussed elsewhere, San Miguel has already upgraded its scrubber and therefore it was not included in our
modeling analysis of additional controls and not included among the nine facilities discussed here. In our FIP, we
are finalizing our determination that San Miguel maintains an emission rate consistent with recent monitoring data.
656
-------
transported together, which means that the impacts from each unit are effectively combined at all
downwind receptors. Sources for which this is true (e.g. Big Spring Carbon Black) should be
evaluated for controls.
Response: We respond to the comments concerning Oklaunion, Sommers-Deely-Spruce, Works
No. 4, Pirkey, and Big Spring Carbon Black in separate responses elsewhere in this section. For
a BART analysis, the combined impact from all BART-eligible units considering maximum
actual emissions is compared to the selected threshold to determine if the facility is subject-to-
BART. The five-factor BART analysis is then applied to the subject-to-BART sources, typically
on a unit-by-unit basis, as potential controls would be installed on each unit. As discussed in
more detail where we address comments concerning our identification of sources for additional
control analysis, we initially evaluated facility-wide impacts in narrowing the list of sources for
additional control analysis. We then examined estimated unit level impacts. We concluded that
this was a reasonable way to identify sources to analyze for additional controls and established a
0.3% visibility impact threshold to identify those sources with the largest visibility impacts for
additional evaluation. We also considered additional information to determine whether or not
sources near this threshold should be included in our analysis. We agree with the commenter
that the analysis of additional controls is typically done on a unit by unit basis. Our initial
consideration of both facility-wide impacts and then ultimately, unit level impacts allowed us to
identify a reasonable set of sources with the greatest visibility impacts to evaluate further for unit
level controls. We believe that it is appropriate to consider actual emissions because these
current emission levels are anticipated to continue in the near future. This approach is consistent
with the methodology utilized in photochemical modeling to develop realistic projections of
basecase and future visibility conditions and the approach for a BART five-factor analysis
utilizing actual emissions rather than permitted emission limits. We discuss the use of recent
actual emissions in more depth elsewhere in this section of the RTC document. Should actual
emissions at a specific unit increase in the future, that will be considered during future planning
periods.
Comment: EPA should impose pollution reduction requirements and/or emission
limitations on additional sources to enable the class I areas affected by Texas sources to
achieve greater reasonable progress towards the national visibility goal [Stamper (0068) p.
44, 56]
[Stamper (0068) p. 56] Stamper stated that there are additional control measures and/or emission
limitations that are very likely cost effective and/or readily implemented for several Texas
sources that met EPA's 0.3% contribution criteria to be evaluated. EPA should consider adoption
of these controls and emission limitations as part of its reasonable progress control measures in
its Texas FIP. It is likely that additional cost-effective and reasonable controls are also available
for sources that do not meet EPA's 0.3% threshold, but that nonetheless cumulatively contribute
to visibility impairment at the evaluated Class I areas. Given that most of the Class I areas
impacted by Texas sources are not on track to achieve natural background visibility by 2064,
EPA is justified in requiring implementation of cost effective pollution control requirements at
the 7 sources discussed above.
657
-------
[Stamper (0068) p. 44] Stamper stated that the EPA determined Texas sources to evaluate for
controls based initially on an analysis of Q/D (i.e., annual emissions of a source/unit in tons per
year, divided by the distance to the closest Class I area in kilometers). EPA used a Q/D value of
10 as a threshold for initial identification of sources to be evaluated for additional pollution
controls and summed SO2 and NOx emissions together for the Q values for each source/unit. FIP
TSD at A-l. EPA states that the Q/D value of 10 is supported in the BART Guidelines and also
has been suggested by the Federal Land Managers as an initial screening tool.158 Id. at A-l to A-
2. EPA initially calculated the Q/D values based on the 2009 emission inventory, but then
updated it with 2010 to 2011 emissions information. Id. at A-7. EPA identified 37 facilities as a
result of these analyses which it then further evaluated for impacts during the average 20% worst
days at the impacted Class I areas. Id. at A-l5.
Stamper stated that the EPA used a threshold of 0.3% for a unit's impact on a Class I area to
determine if further evaluation was necessary for reasonable progress controls. FIP TSD at A-49
to A-50. It is not clear how this 0.3% threshold was derived. Further, it must be noted that in
other reasonable progress analyses performed by EPA, a value of Q/D over a specific threshold
was sufficient to justify a four-factor analysis of reasonable progress controls (i.e., no modeling
was done in addition to a Q/D analysis to select sources to evaluate for reasonable progress
controls).159 EPA also seemed to primarily focus on impacts at three Class I areas (Wichita
Mountains, Guadalupe Mountains, and Big Bend) and did not fully consider impacts at other
Class I areas in selecting sources to evaluate for reasonable progress controls.
Stamper stated that there were several facilities with Class I area impacts at or higher than EPA's
0.3%) threshold which EPA ultimately did not evaluate for additional reasonable progress
controls. Those sources include the Works No. 4 Glass Plant, Sommers-Deely-Spruce Power
Station, Oklaunion Power Station, H.W. Pirkey Power Plant, Big Spring Carbon Black Plant, and
Harrington Station. Id. at A-51. EPA did not evaluate these facilities for further controls for
various reasons, most of which are not supportable. Below, we review these sources and the
additional controls that should have been evaluated by EPA as part of its Texas regional haze
FIP.
Footnotes:
158 As discussed in the Visibility and Health Modeling Technical Support Document accompanying the Conservation
Group's comments on EPA's proposed action, it is questionable whether using a Q/D value of 10 is sufficient for
identifying all of the Texas sources that should be considered for control to address regional haze. EPA's BART
guidelines state that "a larger number of sources causing impacts in a Class I areas may warrant a lower contribution
threshold" for BART evaluations, and EPA should the same guidance to reasonable progress control evaluations.
See 40 C.F.R. Part 51, Appendix Y, Section III. A.
159 See EPA's rulemakings regarding the regional haze plans for Montana, Wyoming, and North Dakota. 77
Fed.Reg. 23988, (April 20, 2012); 24058-9; 78 Fed.Reg. 34738, (June 10, 2013); 3476376 Fed.Reg. 58570, 58624-5
(September 21, 2011).
Response: We appreciate the commenter's support for the required controls. We respond to
specific comments concerning additional controls on units at the facilities identified by the
commenter in separate responses to comments below. We respond to comments concerning our
selection and use of a 0.3%> contribution to total visibility impairment threshold and our use of a
Q/d threshold of 10 in separate responses to comments elsewhere in this section of this document
where we address comments on the identification of sources for additional control analysis. We
658
-------
discuss comments concerning impacts at other Class I areas in a separate response to comment
above.
Comment: EPA Should Include Emission Limits on Additional Sources In Order To
Make Greater Reasonable Progress. [Earthjustice (0067) p.2, 41]
[Earthjustice (0067) p.2] Earthjustice et al. stated that while EPA should require the controls it
proposed in its federal plan, EPA should consider additional control measures as well. EPA's
analysis demonstrates that significant emissions reductions at the Welsh and W. A. Parish power
plants are cost-effective, would significantly improve visibility, and meet all four of the statutory
factors for reasonable progress controls. EPA should require controls at the Welsh and W.A.
Parish facilities.
In addition, Earthjustice et al. stated that the EPA failed to conduct a detailed analysis of
potential controls at several facilities where controls are available that would result from a four
factor reasonable progress analysis. In particular, EPA should conduct such analyses of the
potential reasonable progress controls at the following facilities: Harrington Station, the
Oklaunion Power Plant, H.W. Pirkey Power Plant, PPG Industries Works No. 4, the Sommers-
Deely-Spruce Complex, the Big Spring Carbon Black Plant, and the TNP One Steam Electric
Station. We likewise urge the agency to take a closer look at the cost-effective control measures
available for reducing emissions from oil and gas sources in Texas.
[Earthjustice (0067) p.41] Earthjustice et al. stated that as demonstrated in the Gray (0070) and
Stamper (0068) Reports, EPA's proposed FIP should be revised to include controls on additional
sources. First, EPA should require SO2 controls for the W.A. Parish and Welsh units. Second,
EPA should conduct detailed four-factor analyses for the Harrington Station, the Oklaunion
Power Plant, H.W. Pirkey Power Plant, PPG Industries Works No. 4, the Sommers-Deely-Spruce
Complex, the Big Spring Carbon Black Plant, and the TNP One Steam Electric Station.
Response: We respond to comments concerning controls at specific facilities mentioned in the
comment in a separate responses to comments below in this section.
Commenter 0053-11 stated that every single one of the coal-fired power plants in Texas should
be considered.
While appreciative of the 230 tons of SO2 emissions proposing to be cut, commenter 0053-25
suggested that the coal-fired industry should be completely dismantled considering global
climate change.
Commenter 0053-41 stated that the best available SO2 and NOx control technology should be
required for all coal-burning facilities in Texas. Given that the TCEQ failed to develop a plan to
mitigate pollution from Texas facilities that clearly contribute to haze in Texas and our
neighboring states, it is appropriate for EPA to implement a plan to do so.
659
-------
Commenter 0053-63 stated that coal plants in Texas have made more pollution than in other
states. The commenter urged the EPA to move forward as quickly as possible and to include all
Texas coal plants in the proposal.
Commenter 0054-39 (Sierra Club) applauded EPA's rule, and noted these power plants (e.g., Big
Brown, Martin Lake, Monticello and others) are the largest polluters among 2,000 industrial
plants in the state of Texas. The commenter stated that millions of people will benefit from the
230,000 tons of SO2 reduction and any associated reductions in fine PM, sulfates, and nitrates.
These units were built in the 1970's before NSR permitting, so regulation is overdue. The
commenter stated that this rule could be strengthened by included more than 15 of the nearly 40
power plants in Texas.
Commenter 0053-26 urged EPA to expand in terms of coal plants, noting that 19 coal plants in
Texas are some of those are the oldest and dirtiest coal plants in the country.
Comment: Multiple commenters made statements about Texas coal-fired power plant
emissions. Commenters suggested that more coal-fired plants should be affected by the
proposed action. There also were comments that these coal-fired power plants contribute to
climate change and impact our health and environment, including smog. There was a comment
that National Park Service data for Big Bend shows that at least a third of the haze pollution
comes from sulfates. Other commenters noted that Texas sources produce more haze-causing
pollution than Oklahoma and Arkansas combined.
Response: Elsewhere in this document, we address specific comments concerning our
identification of sources to evaluate for additional controls and determination of required
controls. We identified those sources with the greatest visibility impacts and considered the
estimated visibility benefit of controls alongside the four statutory factors to identify cost-
effective controls that will achieve reasonable visibility benefits required during this planning
period towards the national goal. We agree that emissions from coal-fired power plants can
affect climate change, impact our health and environment. We also agree with the National Park
Service Data concerning sulfates. We also agree that Texas sources produce more haze-causing
pollution than Oklahoma and Arkansas combined.
Harrington
Comment: EPA Must Consider Requiring New SO2 Controls At Harrington Station.
[Stamper (0068) p. 44]
The attached Stamper report to Earthjustice, et al comments stated that the EPA did not evaluate
or propose any pollution controls for Harrington Station. Although EPA did not explain why
Harrington Station was not evaluated further in its FIP TSD, it appears it was because none of
the three Harrington units had impacts over 0.3%. See FIP TSD at A-52. EPA listed Harrington
Station as ranking number 5 in terms of percent extinction causes in Guadalupe Mountains
National Park, with a maximum contribution to the 20% worst days at the Park of almost 1%.
Given EPA's 0.3% threshold, this is a significant impact.
660
-------
Further, Stamper argued that the EPA ignores its modeling that shows the Harrington plant's
contribution to extinction at Salt Creek Wilderness Area in New Mexico is 1.2%.160 On average,
that equates to 0.4% contribution per each Harrington unit. Thus, the Harrington plant's impacts
on the Salt Creek Wilderness Area alone should have been sufficient justification for analyses of
pollution control measures for the Harrington Station using EPA's 0.3% threshold (including
EPA's 0.3%) impact per unit threshold).
In addition, Stamper noted that the Harrington units had a combined 0.828%) contribution to
extinction at the White Mountains Wilderness Class I area in New Mexico.161 Overall at the 18
Class I areas modeled by EPA, the Harrington plant has the 5th highest cumulative contributions
to extinction of the 38 facilities evaluated by EPA.162 It is arbitrary for EPA to identify an
extinction contribution threshold of 0.3% per unit, but then only consider that contribution based
on impacts at three Class I areas (i.e., Wichita Mountains, Guadalupe Mountains, and Big Bend)
when sources like the Harrington plant have the highest impact at other Class I areas. Since there
is ample justification to conduct a pollution control evaluation for the Harrington Station, we
have conducted an analysis for SO2 controls at this currently unscrubbed plant, as discussed
below.
Stamper stated that the Harrington Station consists of three EGUs, each with a generating
capacity of 360 MW, that burn subbituminous Powder River Basin coal. None of the units have
SO2 controls, and all have low NOx burners with overfire air. Unit 06IB has an ESP for PM
control, while Units 062B and 063B each have baghouses for PM control.
Stamper stated that the EPA's cost spreadsheets were used to evaluate the cost effectiveness for
new wet FGDs at 98%> control, a NID™ circulating dry scrubber (CDS) at 98%> control, an SDA
at 95% control, and DSI at 50% control. The SO2 control costs for the Harrington units were
calculated using the same IPM cost modules used by EPA, but for the reasons discussed above,
cost effectiveness was based on a 5-year annual average emissions baseline, 5-year annual
average SO2 rate in lb/MMBtu, and 5-year average gross heat rate and MW-hrs generated, based
on actual operating data from 2009 to 2013. DSI was only evaluated at 50% SO2 removal
efficiency because, as stated above, EPA needs to provide additional justification to consider
DSI as able to achieve 80-90% control, especially without significantly increasing particulate
matter. The results of these analyses are provided in Table 16 of comment 0068.
SO2 Control Cost Effectiveness Evaluation for Harrington Station Units 061B, 062B, and
063B163 submitted by Stamper (0068, Table 16)
661
-------
Harrington
Unit
so:
C OUH ol
Total
Annualized
Coses
Tons so:
Removed
Cost
Effectiveness
06 IB
Wei FGD
SIS ioO 560
5 05 7 tpv
S3,591 ton
t
NID™ CDS
$ P.340 <>45
5.057 tpv
$3,429- ton
06 IB
SDA
$1 ~ 06° 445
4.S4S tpv
S3.52I ton
06 IB
DSI at 50%
SS.S46.370
1738 rpy
S3.:3: ton
0o:b
Wet FGD
$;$,*66 725
5.411 tpv
S3.505 ton
06-B
NID™ CDS
SIS.144.000
5.411 tpv
S3.353 ton
062B
SDA
$i? $50 232
5.170 tpv
S3,448 ton
06:8
DSI at 50%
$7,35 5.809
2.93 7 tpv
! u
£
0
06 5B
Wef FGD
SIS S4S.S70
5.640 tpv
S3,342 rou
06 5 B
NID™ CDS
S1S.052.0P5
5.640 tpv
S3.201 ton
06 3B
SDA
<17.77; s:s
5.40S tpv
$3,286 ton
0655
DSI 3t50%
S 7.4391) 16
3,052 tpv
$2.43? ton
According to Stamper, all of these costs are within the range of cost effectiveness values for controls
required by EPA and states to meet regional haze requirements. Thus, these costs are reasonable. A
NID™ CDS would provide the most cost effective SO2 reductions. While DSI is most cost effective,
that is based on only achieving 50% SO2 removal, and thus it will not achieve as significant a level
of SO2 reduction as a scrubber at 95-98% SO2 removal.
Stamper stated, as shown in EPA's FIP TSD, these SO2 controls (particularly SO2 scrubbers) have
been used at numerous facilities. The energy and non-air quality impacts of these controls are widely
known and have not generally been cited as justification for not installing SO2 scrubbers. FIP TSD at
7-8. Further, the cost estimates provided in the above table address the energy costs, water costs, and
waste disposal costs associated with these controls.
With respect to the remaining useful life of the Harrington units, Stamper was not aware of any
planned retirements of the units that would justify assuming a shorter life of the pollution controls
than 30 years. As discussed in EPA's FIP TSD, there is ample justification that a scrubber or DSI
system will have a life of at least 30 years.
Stamper concluded that the Harrington units should be able to install SO2 controls by the end of
2018.
Stamper stated that the EPA should have thus considered these cost effective controls for the three
EGUs at Harrington Station. Based on a review of the four factors, all of SO2 these controls are
justified as reasonable progress controls. Such controls will reduce the Harrington station's
contribution to regional haze at the Salt Creek and White Mountains Class I areas, as well as at
Guadalupe Mountains National Park and at nearby Carlsbad Caverns National Park.
Footnotes:
160 See TX116-07-_29_Source_selection_analysis_TX_RH-l-3 l-14.xlsx at "All Class I areas data" tab.
161 Id.
162 Id.
662
-------
163 See spreadsheets with filenames "Wet_FGD_Cost_IPM_TX_Sources_Revised_VS_Mar 27 2015.xlsx," "NID
CDS_Cost_IPM_TX_Sources_VS_Mar 27 2015.xlsx," "SDA_Cost_IPM_TX_Sources_Revised_VS_Mar
27_2015.xlsx," and "DSI_Cost_IPM_TX_Sources_Revised_VS_Mar 27_2015.xlsx," at Exhibits 34, 35, 36, and 37
to this report.
Stamper/Earthjustice et al. state that we should have performed an additional control analysis on
the Harrington facility. The modeled visibility impacts from these units exceeds the 0.3%
visibility impact threshold at Salt Creek, cumulative impacts are significant, and it is arbitrary to
establish a 0.3% threshold but then only consider impacts at Wichita Mountains, Guadalupe
Mountains and Big Bend. The commenter also states that the maximum contribution to visibility
impairment at Guadalupe Mountains from the facility approaches 1%. Stamper provides
additional analysis of the cost and other factors associated with potential SO2 controls.
Estimated costs of controls range from approximately $2500/ton to $3200/ton for DSI at 50%
control and $3200/ton to $3400/ton for NID™ CDS at 98 % control. Wet FGD and SDA were
also analyzed. The commenter concludes that SO2 controls are cost-effective and should have
been considered for the Harrington units.
Response: As discussed in a separate response to comment above, to address the deficiencies
we identified in the Oklahoma and Texas regional haze SIPs, we focused on identifying sources
for additional control analysis that could potentially address those significant visibility impacts at
Wichita Mountains and the two Texas Class I areas. While we acknowledge commenters
concerns for other Class I area impacts, we had previously approved the consultation in New
Mexico, as well as the RPGs for New Mexico's Class I areas in a previous, separate action.521
The largest estimated percent of total visibility impairment on the 20% worst days for one of the
Harrington units based on modeled emission rates were 0.137%, 0.056%, and 0.024% at
Guadalupe Mountains, Wichita Mountains, and Big Bend, respectively. Percentage impacts
based on recent actual emissions were even lower. Because the visibility impacts were well
below a 0.3% threshold on a unit basis, we did not perform additional analysis for controls on
these units for this planning period and focused on those sources with the largest impacts for this
first round of SIPs at the Class I areas of interest. For future regional haze planning periods, we
expect New Mexico to consider impacts from Texas sources including the information on
visibility impacts from specific sources provided by our analysis, as well as incorporate
corrections and updates to emission reductions in consultations and development of their RH
SIPs for the next planning period.
Oklaunion
Comment: EPA Must Consider Upgraded SO2 Controls at the Oklaunion Power Plant.
[Stamper (0068) p.47]
Earthjustice/Stamper stated that the EPA did not evaluate the Oklaunion Power Station for
reasonable progress controls, even though the 2018 modeling showed the Oklaunion Power
Station contributed 0.567%> to extinction at the Wichita Mountains Class I area. FIP TSD at A-
51. It appears that EPA did not evaluate the Oklaunion Power Station for reasonable progress
521 77 FR 70693 (November 27, 2012)
663
-------
controls because the "Estimated Facility Impact Adjusted to reflect 2008 to 2012 Avg.
Emissions" was reduced to 0.286%.164 Id. According to the emissions data provided by EPA, the
Oklaunion Power Station had, on average, lower recent annual SO2 emissions based on the
average of 2008 to 2012 emissions (3,611 tpy averaged over 2008-2012 compared to 7,157.8 tpy
that was modeled by CENRAP in the 2018 modeling165). However, it does not appear that EPA
evaluated why SO2 emissions at the Oklaunion Power Station have been lower than what was
modeled in CENRAP's 2018 modeling and if such reductions were permanent. Further, EPA did
not evaluate whether there were any limits on emissions that would ensure SO2 emissions from
the Oklaunion Power Station remain at the level of the 2008 to 2012 average rate of 3,611 tpy.
Stamper noted that a 2012 TCEQ Permit Amendment Technical Review document identifies the
current allowable SO2 emission rates of the Oklaunion Power Station as 14,936.0 tpy or 3,410
lb/hr (3-hour average).166 This annual limit is more than twice as high as the 7,157.8 tpy SO2 rate
modeled by CENRAP in the 2018 modeling.
Stamper stated that the Oklaunion Power Station is a 720 MW coal-fired EGU that is equipped
with a wet limestone scrubber for SO2 control that was installed in 1986.167 According to data
submitted to EPA's Clean Air Markets Database, the unit is also equipped with low NOx burners
(dry bottom only) and an electrostatic precipitator. Based on coal data submitted to the Energy
Information Administration, the unit burns Powder River Basin subbituminous coal with
uncontrolled SO2 rate (based on sulfur in the coal) ranging from 0.73 lb/MMBtu to 1.07
lb/MMBtu over 2011 to 2013.168 According to emissions data submitted to EPA's Clean Air
Markets Database, the unit emitted SO2 at a rate of 0.18 to 0.21 lb/MMBtu on an annual average
basis over 2011 to 2013169. Thus, the unit's scrubber appears to be achieving about 75% - 80%
SO2 removal. The wet scrubber was designed for an 86.8% removal efficiency.170 However, it
also has provisions for bypass.171
Stamper stated that the maximum allowable emission rate for Oklaunion of 3,410 lb/hour (3-hour
average) equates to an emission rate of 0.50 lb/MMBtu, given the 6,800 MMBtu/hr size of the
EGU.172 Based on the uncontrolled SO2 emissions in the coal burned in recent years at the
Oklaunion facility, the SO2 limits require, at best, 53% SO2 removal. Given all of this, the public
has no assurances that the Oklaunion Power Station will emit SO2 at rates lower than the 7,157.8
tpy SO2 rate modeled by CENRAP in the 2018 modeling - in fact, SO2 emissions could be much
higher. Thus, EPA's decision to exclude the Oklaunion Station from a review of reasonable
progress controls is not justified. The Oklaunion Station is ranked 6th of the Texas EGUs
modeled by EPA in terms of extinction impacts at the Wichita Mountains Class I area, which is
based on the modeling of an annual SO2 emission rate that is less than half of the EGU's
allowable annual SO2 rate. EPA must evaluate scrubber upgrades for this EGU to ensure that its
impacts on visibility at Wichita Mountains and other Class I areas are minimized.
Stamper noted, as discussed in EPA's FIP Cost TSD at pages 25-28, there are numerous options
that are available for upgrades to wet scrubbers. Although EPA did not provide any of the cost
information for the scrubber upgrades it evaluated, other scrubber upgrade cost data was publicly
available in BART analyses. The costs per kW for those example scrubber upgrades was
calculated, and it was found that the costs ranges from $l/kW up to a high of $182/kW for
Cholla Unit 4 which replaced an existing scrubber tower with a new wet scrubber.173 The
664
-------
average costs for scrubber upgrades in the publicly available data was $68/kW.174 Excluding the
maximum and minimum scrubber upgrade costs (which reflect a new scrubber retrofit at Cholla
Unit 4 and the $l/kW scrubber upgrade costs at Hay den Units 1 and 2), the range of costs for
scrubber upgrades based on the cost data we were able to collect was $12/kW to $144/kW.
Given that these were all wet scrubber upgrades, it is reasonable to assume a similar range for a
scrubber upgrade at the Oklaunion Power Station, which at 720 MW size equates to a range of
total capital costs of $8,640,000 to $103,680,000 for scrubber upgrades. Using a cost recovery
factor reflective of a lifetime of the upgraded scrubber of 30 years and a 7% interest rate, these
costs equate to range of annualized costs of $696,384/year to $8,356,608/year.
Stamper stated, for the same reasons provided by EPA, it is reasonable to assume that at least
95% control can be achieved with upgrades to the Oklaunion wet scrubber. See, e.g., 79 Fed.Reg.
74877 (December 16, 2014), see also FIP Cost TSD at 27. Based on the worst case uncontrolled
SO2 in the coal in the last few years at the Oklaunion Power Station, 95% control would equate
to an emission rate of 0.05 lb/MMBtu and 2,696 tons of SO2 reduced per year, on average.175
Thus, using the range of annualized cost of scrubber upgrades of $696,3 84/year to
$8,356,608/year, the cost effectiveness of scrubber upgrades at the Oklaunion Power Station
would likely be in the range of $258/ton to $3,099/ton. This range of costs is reasonable as it is
within the costs per ton that EPA has found to be cost effective to meet BART and reasonable
progress requirements.
Stamper stated that this is an estimate of the possible range of cost effectiveness for scrubber
upgrades at the Oklaunion Power Station. Given that scrubber upgrades are most likely available,
feasible and cost effective, EPA should undertake further analysis of the upgrades that could be
done at the Oklaunion Power Station to reduce this plant's visibility impacts on the Wichita
Mountains and other Class I areas.
Footnotes:
164 EPA also stated that "if just the impacts from SO2 were examined, the [Oklaunion] facility's impacts would be
below the 0.3% value." FIP TSD at A-53. However, it must be noted that 87% of Oklaunion's contribution to
extinction at Wichita Mountains Class I area are due to SO2. See TX166-010-
03_EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_072913.xlsx at "fac data" tab.
165 See TX116-07-_29_Source_selection_analysis_TX_RH-l-3 l-14.xlsx at "All ClassI" tab.
166 See Permit Amendment, Source Analysis & Technical Review, Public Service Company of Oklahoma,
Oklaunion Power Station, Permit Number 9015/PSDTX325M2, downloaded from TCEQ's Air/NSRPermits
(NSRP) library available at https://webmail.tceq.state.tx.us/gw/webpub, Ex. 56. See also Emission Sources -
Maximum Allowable Emission Rates, Permit Numbers 9015 and PSDTX325M2, dated February 3, 2012,
downloaded from TCEQ's Air/NSRPermits (NSRP) library available at https://webmail.tceq.state.tx.us/gw/webpub,
Ex. 57.
167 See Burns & McDonnell, Utility FGD Design Trends, at 21 (Ex. 67).
168 See Worksheet entitled "Pirkey and Oklaunion Coal Info" at Oklaunion tab, data from EIA-923 for 2011 to 2013
(Ex. 58).
169 See Worksheet entitled Oklaunion and HW Pirkey CAMD 2000 to 2014, Ex. 59.
170 See Burns & McDonnell, Utility FGD Design Trends, at 21 (Ex. 67).
171 Id. at 22.
172 See Emission Sources - Maximum Allowable Emission Rates, Permit Numbers 9015 and PSDTX325M2, dated
February 3, 2012 (Ex. 57).
173 See SO2 Scrubber Upgrade Cost Data worksheet (Ex. 48).
174 Id.
175 See Worksheet entitled Oklaunion and HW Pirkey CAMD 2000 to 2014, at Row 38, Ex. 59.
665
-------
Response: We respond to this comment below where we address comments concerning controls
on Pirkey, Oklaunion, and Twin Oaks.
H.W. Pirkey
Comment: EPA Must Consider Upgraded SO2 Controls at the H.W. Pirkey Power Plant.
[Stamper (0068) p.49]
Earthjustice/Stamper stated that the EPA also did not evaluate the H.W. Pirkey Power Plant for
reasonable progress controls, even though the 2018 modeling showed the H.W. Pirkey Power
Plant contributed 0.50% to extinction at the Wichita Mountains Class I area. FIP TSD at A-51.
As with the Oklaunion Power Station, EPA did not evaluate the H.W. Pirkey Power Plant for
reasonable progress controls because the "Estimated Facility Impact Adjusted to reflect 2008 to
2012 Avg. Emissions" was reduced to 0.097%. Id. According to the emissions data provided by
EPA, the H.W. Pirkey Power Plant had, on average, lower annual SO2 emissions in 2008 to 2012
average than what was modeled in the CENRAP modeling (3,809 tpy compared to 19,635 tpy
that was modeled by CENRAP in the 2018 modeling176). However, it does not appear that EPA
evaluated why SO2 emissions at the H.W. Pirkey Power Plant have been lower than what was
modeled in CENRAP's 2018 modeling. Further, EPA did not evaluate whether there were any
limits on emissions that would ensure SO2 emissions from the H.W. Pirkey Power Plant remain
at the level of the 2008 to 2012 average rate of 3,809 tpy. Moreover, the H.W. Pirkey Power
Plant was modeled to have a much higher contribution to the Caney Creek Class I area than
Wichita Mountains, with a 1.047 % contribution to extinction at Caney Creek Class I area.177 It
appears that EPA did not even consider these much higher impacts from the H.W. Pirkey Power
Plant at the Caney Creek Class I area.
Stamper stated that the H.W. Pirkey Power Plant is approximately a 650 MW EGU that burns
lignite coal and uses natural gas for ignition and flame stabilization. The EGU is equipped with a
wet limestone scrubber, low NOx burners and overfire air, and an ESP, according to EPA's
Clean Air Markets Database. The wet limestone scrubber had a design SO2 removal efficiency of
85%) and has provisions for bypass.178
Stamper stated that a review of annual SO2 emissions and calculated annual SO2 rates (based on
CAMD SO2 emissions and heat input data) shows that SO2 emissions have varied widely at this
EGU:
Between 2000 to 2005, annual SO2 emissions varied from 11,699 tpy to 19,693
tpy.179 SO2 emissions decreased significantly beginning in 2006, but not
consistently: Annual SO2 emission rates varied from 0.07 lb/MMBtu to 0.29
lb/MMBtu, and annual SO2 emissions varied from 1,953 tpy to 7,339 tpy.180
Based on coal data submitted to the Energy Information Administration, the
average uncontrolled SO2 emission rate over 2009 to 2013 was 2.39 lb/MMBtu,
assuming all sulfur in the coal is emitted.181 Based on the annual average SO2
rates being achieved in recent years, this means the H.W. Pirkey Power Plant has
been achieving between 85%>-95%> SO2 removal.
666
-------
Stamper noted that while this is clearly a reduction in SO2 emissions from years past, a review of
the Title V permit for the H.W. Pirkey Power Plant shows that the most stringent SO2 limit
applicable to the EGU is 1.2 lb/MMBtu, based on the NSPS Subpart D.182 With a heat input
capacity of approximately 7,000 MMBtu/hr183, this means that the annual allowable SO2
emissions of the Pirkey plant are 36,792 tpy.
Given that it appears SO2 removal efficiencies approaching 95% control have been achieved at
the H.W. Pirkey Power Plant, Stamper stated that it seems very likely that upgrades to the
scrubber have been made in recent years. Indeed, a Sargent & Lundy paper which is in the
docket for the Texas Regional Haze FIP discusses recent upgrades of the wet scrubber at H.W.
Pirkey Unit l.184
Specifically, Stamper stated that this paper indicates that the natural oxidation wet limestone
FGD at H.W Pirkey Power Plant has 4 absorber modules, each with 33% capacity, thus leaving
one spare, with a full flow bypass duct and a reheat system to maintain a dry stack.185
Modifications had previously been made to the FGD system to add an absorption tray to each
absorber, add a dibasic acid (DBA) feed system, and reduced limestone grind size.186 Additional
modifications were made in 2006 to achieve 100% flue gas scrubbing, including installation of a
new bypass damper and modifications to the existing outlet ductwork, modifications to the stack
to provide for wet operation, modifications to the reagent preparation train, and alterations to the
mist eliminator washing system.187 According to the Sargent & Lundy paper, since 2006, the
H.W. Pirkey Power plant has "achieved zero bypass."188 Further, the system "operates regularly
at 97%) SO2 removal efficiency...."189 However, based on the CAMD SO2 data discussed above,
it does not appear that the Pirkey Plant regularly operates the scrubber to achieve 97% SO2
removal efficiency. In fact, SO2 removal efficiency based on an annual average basis is often
around 85-86%) removal efficiency. In the absence of a stringent SO2 emission limit, it appears
that the incentive to consistently reduced SO2 emissions is not always present.
Thus, Stamper concluded that the EPA's decision to exclude the H.W. Pirkey Power Plant from a
review of reasonable progress controls is not justified. As EPA has proposed for the San Miguel
plant, EPA should propose a lower SO2 limit than the currently applicable 1.2 lb/MMBtu limit
for the H.W. Pirkey Power Plant to ensure that the recent scrubber upgrades are operated in a
manner to consistently achieve low SO2 emission rates. Based on the uncontrolled SO2 emission
rates (assuming all sulfur in the coal is emitted as SO2), the uncontrolled SO2 emission rates have
averaged 2.39 lb/MMBtu over the last five years. At the minimum, EPA must impose an SO2
limit reflective of 95% control, or 0.12 lb/MMBtu, at the H.W. Pirkey Power Plant to ensure that
the upgraded wet scrubber is consistently maintained and operated to achieve the lowest SO2
emission rates. Given that the upgraded scrubber can achieve 97% control, an emission limit of
0.12 lb/MMBtu provides more than an adequate compliance margin. Incorporating this limit into
the Texas reasonable progress FIP will ensure that the H.W. Pirkey Power Plant's has a lower
contribution to haze at the Caney Creek and the Wichita Mountains Class I areas.
Footnotes:
176 See TX116-07-_29_Source_selection_analysis_TX_RH-l-3 l-14.xlsx at "All ClassI" tab.
177 Id.
178 See Burns & McDonnell, Utility FGD Design Trends, at 13-14 (Ex. 67).
667
-------
179 See Worksheet entitled Oklaunion and HW Pirkey CAMD 2000 to 2014, Ex. 59.
180 Id.
181 See Worksheet entitled "Pirkey and Oklaunion Coal Info" at Pirkey tab, data from EIA-923 for 2011 to 2013 (Ex.
58).
182 See Federal Operation Permit, H.W. Pirkey Power Plant, November 22, 2010, at 30 (Ex. 60).
183 See TCEQ's Permit Renewal Source Analysis & Technical Review, Southwestern Electric Power Company, HW
Pirkey Power Plant, Permit Number 6269 and PSDTX64, at 3 (Ex. ~
184 See Caverly, Don et al., Results of FGD Upgrade Projects on Low-Rank Coals, in Docket ID EPA-R06-OAR-
2014-0754-0008, filename TX166-008-
063_Results_of_FGD_Upgrade_Projects_on_Low_Rank_Coals_EP2007.pdf.
185Id. at 3.
186 Id.
187 Id. at 4-7.
188 Id. at 5.
189 Id. at 7.
Response: We respond to this comment below where we address comments concerning controls
on Pirkey, Oklaunion, and Twin Oaks.
Twin Oaks
Comment: EPA Should Have Considered Pollution Reduction Measures for the TNP One
Steam Electric Station (Twin Oaks Power Plant). [Stamper (0068) p.54]
Stamper stated that although the TNP One Steam Electric Station (also known as Twin Oaks) did
not make EPA's 0.3% extinction contribution threshold, EPA should have evaluated pollution
control measures at this facility. Sulfur dioxide emissions in recent years have been significantly
higher than what was modeled for this facility. The table below compares what was modeled for
the TNP One Steam Electric Station compared to SO2 emissions data in EPA's Clean Air
Markets Database.
668
-------
Table 18: T?IP Out Steam Electric Station (Twin, Oaks) Actual E uu ieiis Compared to
2§18 Modeled Ebpsiohs,*'
TNP One Steam
Electric Station Unit
•¦¦¦<>: py
1
2008 "
2.325
1
2009
2.471
1
2010
2.975
1
_
2.7*8
1
2012
2,206
1
2013
2,272
1
2014
I
Emissions Modeled
for 2018
#41
2
MA
2.131
^ ooo
2,235
2010
2.429
¦1
on
2012
1,832
2
2015
3.061
2
2014
:
Emissions Modeled
for 201S
6-i 7
Based on the table, Stamper stated that SO2 emissions have been around three times as high as
what was modeled for each unit. Had more accurate emissions been modeled for the TNP One
Steam Electric Station, this facility would likely have shown contributions to haze closer to
EPA's 0.3% threshold.
Stamper stated that the TNP One Steam Electric Station units are both fluidized bed units with
limestone injection, according to data submitted to EPA's Clean Air Markets Database. Annual
SO2 emission rates have been quite variable over the past 7 years, ranging from 0.33 lb/MMBtu
to 0.54 lb/MMBtu.208 According to the Statement of Basis for the Federal Operating Permit for
the TNP One Steam Electric Station, the units burn primarily lignite, subbituminous coal, and
natural gas, but the units are also able to burn the following: oil filter fluff, tire-derived fuel,
petroleum coke, 3-M pelletized tape, waste oil sorbet, waste crankcase oil, waste oil-based floor
sweep, waste wax-based floor sweep, waste hydraulic oil, and waste oily rags.209 A review of the
Title V permit shows that the most restrictive limit on SO2 emissions appears to be a 70% SO2
control requirement when emissions are less than 0.60 lb/MMBtu, pursuant to NSPS Subpart
Da.210
Stamper stated that the EPA should evaluate additional SO2 control technology for the Twin
Oaks Power Plant. For example, the units could add dry scrubbers similar to the Sandow 5
generating plant, which could improve SO2 removal on these fluidized bed boilers to close to
99% control. EPA could also simply impose restrictions on burning of higher sulfur fuels, such
as a prohibition or limits on burning petroleum coke and waste oil and/or EPA could require a
blend of subbituminous and lignite coal to reduce inlet SO2 emissions. Further, given that the
669
-------
Twin Oaks units each have baghouses, DSI could be quite cost effective at providing further
reductions in SO2 from the current fluidized bed with limestone injection.
Stamper stated that given that the Twin Oaks facility has much higher impacts than evaluated by
EPA, due to its SO2 emissions being much higher than modeled and considering the units' high
SO2 emission rate for units with SO2 controls, the Twin Oaks units should have been evaluated
for reasonable progress controls as part of the Texas FIP.
Footnotes:
207 EPA's Clean Air Markets Database Emission Data for Twin Oaks (TNP One Steam Electric Station) is attached
as Ex. 64. The emissions modeled for 2018 are from TX116-07-_29_Source_selection_analysis_TX_RH-l-31-
14.xlsx at "All Class I areas" tab.
208 See EPA's Clean Air Markets Database Emission Data for Twin Oaks (TNP One Steam Electric Station), Ex. 64.
209 See Statement of Basis of the Federal Operating Permit, Optim Energy Twin Oaks, LP, at page 2 (Ex. 65).
210 See Federal Operating Permit for Optim Energy Twin Oaks, LP, April 4, 2011, at page 22 (Ex. 66).
Response: We respond to this comment below where we address comments concerning controls
on Pirkey, Oklaunion, and Twin Oaks.
Comment: Stamper/Earth Justice et al. commented that we should have evaluated additional
controls or reduced emission limits on Oklaunion Unit 1, H.W. Pirkey Unit 1 and Twin Oaks
Units 1 and 2. Modeling results showed the percent of total extinction from the Oklaunion unit
at Wichita Mountains was 0.567% and an adjusted impact based on 2008-2012 emissions of
0.286%. For Pirkey, modeling results show a contribution to extinction of 0.50% and an
adjusted contribution of 0.097% at Wichita Mountains. The commenter also states that modeled
impacts from Pirkey at Caney Creek are even larger than those modeled at Wichita Mountains.
While the commenter agrees that actual emissions from the Oklaunion and Pirkey units are lower
than the emission rate modeled, the commenter states that permitted emissions are much higher
and that there is no assurance that actual emissions will continue at the current, lower rate. The
commenter estimates that for Oklaunion, a scrubber upgrade to 95% efficiency would result in a
reduction of approximately 2700 tons per year and estimates that the cost would likely be in the
range of $258/ton to $3,099/ton. This range of costs is reasonable as it is within the costs per ton
that EPA has found to be cost effective to meet BART and reasonable progress requirements.
The commenter states that the scrubber at Pirkey has been upgraded and is capable of achieving
higher emission reductions than under current operations. Therefore, similar to our proposed
action on San Miguel, we should require an emission limit to ensure the scrubber is operated to
obtain emission reductions consistent with a 95% control efficiency.
For Twin Oaks, the commenter states that although modeled visibility impacts fell below the
0.3% threshold, modeled emissions are a factor of 3 lower than actual emissions. The
commenter suggests that because visibility impacts at actual emissions would be larger, we
should have evaluated additional controls at this facility.
Response: Here we respond to comments concerning controls on Pirkey, Oklaunion, and Twin
Oaks. As explained in our proposed action and FIP TSD522, and discussed in a separate response
to comment, we selected a threshold of 0.3% of the total visibility impairment on the 20% worst
522 See FIP TSD at A-39
670
-------
days to identify units for additional control analysis. This threshold was established to identify a
reasonable set of units that had the largest visibility impacts for additional control analysis for
this planning period. While impacts at the modeled emission rate of 7,157.8 tpy SO2 from the
2018 CENRAP projected emissions for Oklaunion unit 1 exceeded that threshold, we noted that
actual emissions (3,611 tpy annual average 2008-2012) are much lower and result in an
estimated impact of 0.286%, which was less than the threshold. We also noted that a portion of
that impact was due to nitrate emissions from the source and therefore, the percentage impact
that was due to sulfur emissions would be even smaller than 0.286% and below the 0.3%
threshold. This is in contrast to the other EGUs evaluated where sulfur impacts accounted for
almost the entire visibility impact on the 20% worst days. Because this unit fell below the
threshold based on our consideration of past actual emissions, we did not evaluate additional
controls or revised permit limits for this unit for this planning period. Similarly, actual emissions
for Pirkey Unit 1 (4263 tpy annual average 2008-2012) are much lower than the modeled
emissions (19483 tpy) from the 2018 CENRAP projected emissions and result in an estimated
impact of 0.097%, much lower than the 0.3% threshold and the unit at this facility was not
selected for additional control analysis. Emissions data from CAMD show that since the
scrubber upgrade in 2006, annual emissions at Pirkey have been much less than half the modeled
emissions. As discussed in a separate response to comment in this section of the RTC document,
we focused our analysis on impacts at the Wichita Mountains Class I area in Oklahoma and the
two Class I areas located in Texas. We note, however, that the estimated percent contribution to
visibility impairment at Caney Creek from Pirkey also falls below the 0.3% threshold (0.203%)
when we consider actual recent emissions. As explained in detail in the FIP TSD523 we
determined it was necessary to consider recent actual emissions from EGUs due to uncertainty in
2018 projected emissions completed in 2006, the cost of SO2 credits being lower than originally
projected, and comments from Texas on a more recent IPM projection indicating that significant
SO2 reductions were not anticipated at these sources and no large SO2 control projects were
planned at most of the sources being evaluated.524 We also noted that TCEQ has utilized recent
emission data for EGUs when developing projected emissions for 2018 (and other future years)
when developing ozone attainment demonstrations. While some facilities were considered for
additional control because estimated percent contribution to visibility impairment from actual
emissions were higher than the CENRAP 2018 projected levels and above the 0.3% threshold
(e.g. Parish Plant, Welsh), Pirkey and Oklaunion were not evaluated further for controls or
revised permit limits for this planning period as they fell below the threshold when recent actual
emissions were considered. Any increases in actual emissions at these facilities in the future
should be considered during development of the regional haze SIP for future planning periods.
In future planning periods, as the facilities with the largest impacts are controlled, the percent of
total visibility impairment due to these lower impact facilities will increase and they will need to
be considered for additional control.
With regards to Twin Oaks, when we consider recent actual emission levels, the estimated
contribution to visibility impairment still falls below the 0.3% threshold for these units (0.041%
523 See FIP TSD at A-45
524 Texas comments on Draft IPM modeling conducted by EPA for potential national rule making platform provided
on June 26, 2014. In this docket materials as "TCEQ comment letter to EPA on draft modeling platform dated June
24, 2014. '2018 EMP signed.pdf.
671
-------
at modeled emissions and 0.12% at actual average emissions for each unit). Because these units
fell below our threshold we did not consider additional controls or revised emission limits at this
facility for this planning period.
Works No. 4
Comment: EPA Should Have Evaluated Control Measures for Works No. 4
[Stamper (0068) p.51]
Stamper stated that the PPG Industries Works No. 4 glass plant was shown to contribute 0.863%
to extinction at the Wichita Mountains Class I area. FIP TSD at A-51. According to EPA, the
NOx emissions from Works No. 4 are the primary source of the plant's impacts on Wichita
Mountains Class I area. FIP TSD at A-27. Further, each of the two units contribute more than
EPA's 0.3%) threshold to extinction at Wichita Mountains, with Unit 1 contributing 0.448% and
Unit 2 contributing 0.415%. FIP TSD at A-52.
Despite those significant impacts, Stamper noted that the EPA decided to "drop this source from
consideration this time." Id. EPA stated that the modeled NOx emissions were over five times
higher than recent actual emissions and permit limits. Id. Based on the permit documents in
EPA's docket, it appears that may be true for Unit 1, for which a NOx limit of 894.25 was
proposed in a permit application for an oxy-fuel conversion at the Melting Furnace No. I.190 Unit
l's modeled NOx emission rate was 4,526.8.191 However, the Melting Furnace Unit 2 is
currently subject to a 2,947.49 tpy allowable NOx emission rate,192 which is only 30%> less than
the 4,191.9 tpy NOx rate at which the Unit 2 melting furnace was modeled. Since the modeling
of Unit 2 showed a 0.415%> contribution to visibility impairment at Wichita Mountains, it is
likely that modeling of Unit 2's lower allowable NOx rate would continue to show impacts at
EPA's 0.3% contribution threshold.
As justification for not analyzing further controls, Stamper noted that the EPA also states that
NOx controls had been installed at one of the two units. FIP TSD at A-52. That is true for Unit 1.
According to the 2007 permit application for the Unit 1 oxy-fuel conversion at Works No.4, the
NOx emissions were anticipated to be reduced from 2,889.84 tons per year to 894.25 tpy193, a
69%) reduction in emissions from baseline emissions. EPA should have evaluated a similar oxy-
fuel conversion for Unit 2 at the Works No. 4 glass plant. Given that Unit 2 had over a 0.3%>
impact at Wichita Mountains and that PPG Industries apparently determined it was cost effective
to convert Unit 1 to oxy-fuel firing, it seems likely that these same controls would be cost
effective for Unit 2 as well. Such controls would likely reduced the Unit 2 melting furnace's
impacts at the Wichita Mountains Class I area to well less than EPA's 0.3%> contribution
threshold.
Footnotes:
190 See October 2007 Standard Permit Registration PPG Industries, Inc., Wichita Falls Plant, Line 1 Melting Furnace
Oxy-Fuel Conversion, at 13 (in Docket ID EPA-R06-OAR-2014-0754-0008, filename TX166-008-096 PPG -
Std_Pmt_App_83132_partl.pdf).
191 See TX116-07-_29_Source_selection_analysis_TX_RH-l-3 l-14.xlsx at "All ClassI" tab.
192 See Permit Alteration Source Analysis and Technical Review, PPG Industries, Inc., at 1 (in Docket ID EPA-R06-
OAR-2014-0754-0008, filename TX166-008-099_TRV-PPG_Industries_Permit898_Projectl78473.pdf).
672
-------
193 See October 2007 Standard Permit Registration PPG Industries, Inc., Wichita Falls Plant, Line 1 Melting Furnace
Oxy-Fuel Conversion, at 11 (in Docket ID EPA-R06-OAR-2014-0754-0008, filename TX166-008-096 PPG -
Std_Pmt_App_83132_partl.pdf.
Stamper/Earthjustice et al. comment that additional analysis of controls should have been
performed for the PPG Industries Works No. 4 glass plant. Modeled visibility impacts were
above the 0.3% threshold for these two units. Actual emissions for unit 1 are much lower than
modeled, but impacts from unit two would likely be above the 0.3% threshold even after
accounting for actual emission levels. Furthermore, the recent implementation of NOx controls
at unit 1 shows that controls are available, cost-effective, and can significantly reduce emissions
and should be analyzed for unit 2.
Response: We agree with the commenter that modeled percent visibility impairment from the
PPG Glass Works was above the 0.3% threshold for each of the two units. As discussed in
section 4.3 of the FIP TSD and section VII.D.4 of the proposed FIP, we performed additional
analysis and gathered additional information from the facility in order to consider recent
emission reductions, revised permit limits and the potential for additional controls at this facility.
Across the 2 units, average actual annual emissions are only 44% of the projected 2018
emissions for NOx and 81% of the projected SO2 emissions.525 Furthermore, permitted emission
levels are now at levels below the 2018 projected emissions.526 These reductions are the result of
installation of an oxy-fuel conversion on furnace 1 and a more recent fuel conservation project
that involved relocating the burners on furnace 2 at the time furnace 2 was shutdown for
rebricking. We also noted that a furnace typically lasts ten to twelve years until rebricking is
required. In consideration of the emission reductions already occurring at the facility, the
anticipated lifetime of the furnaces, and the fact that furnace 2 had undergone rebricking within
the past few years, we determined it was reasonable to not require additional controls at this time
and encouraged the State of Texas to consider additional controls when Furnace No. 2 is
scheduled for its next rebricking.
Sommers-Deely-Spruce
Comment: EPA Should Adopt Enforceable Requirements for the Shutdown of Two Units
at the Sommers-Deely-Spruce Complex. [Stamper (0068) p.52]
Stamper stated that the EPA noted in its FIP TSD that the Sommers-Deely-Spruce complex
contributes to extinction at the Wichita Mountains, Guadalupe Mountains, and Big Bend Class I
areas in the range of 0.2-0.23%. FIP TSD at A-53. However, these percent contributions reflect
only 40%) of the modeled 0.569%) contribution at Big Bend National Park and 40%> of the
0.558%) contribution at Wichita Mountains.194 It appears that EPA may have discounted the
modeled impacts due to recently installed pollution controls, but the FIP TSD does not provide
525 See Table 31 of the FIP TSD for a summary of modeled, permitted and actual emissions.
526 Standard Permit Registration, PPG Industries, Inc., Wichita Falls Plant, Account No. WH-0040-R. Submitted by
ENVIRON, dated October 11, 2007; Permit Alteration, Permit Number: 898, Flat Glass Manufacturing Facility,
Wichita Falls, Wichita County, Regulated Entity Number: RN102522950, Customer Reference Number:
CN600124614, Account Number: WH-0040-R
673
-------
justification for only evaluating 40% of this facility's impacts when considering whether to
evaluate the facility for additional reasonable progress controls.
Stamper stated, based on the original modeling, the Sommers-Deely-Spruce Complex
contributed 0.3% or more at the Salt Creek Wilderness Class I area and the White Mountains
Wilderness Class I area in addition to the Wichita Mountains, Guadalupe Mountains, and Big
Bend Class I areas.195 Thus, these units have widespread impacts on regional haze. In fact, before
EPA's adjustment to the modeled impacts, the facility was the 6th highest in cumulative Class I
area impacts of the 38 sources modeled by EPA.196
However, Stamper stated that the EPA did not evaluate or propose additional controls because
the owners of the Sommers-Deely-Spruce complex have indicated that two "of their dirtiest
sources" will be shut down by 2018. FIP TSD at A-53. Given the widespread impacts from the
Sommers-Deely-Spruce facility and that EPA is relying on these unit shutdowns to exclude this
facility from further review for reasonable progress requirements, EPA should adopt
requirements for these units to cease operation by 2018 as part of its FIP.
Footnotes:
194 See TX116-07-_29_Source_selection_analysis_TX_RH-l-31-14.xlsx at "All Class I areas" tab, at cells U77 and
U78.
195 Id. at "All Class I areas" tab.
196 Id.
Stamper/Earthjustice commented that we noted the FIP TSD that the Sommers-Deely-Spruce
complex contributes to extinction at the Wichita Mountains, Guadalupe Mountains, and Big
Bend Class I areas in the range of 0.2-0.23%. However, these percent contributions reflect only
40% of the modeled 0.569% contribution at Big Bend National Park and 40% of the 0.558%
contribution at Wichita Mountains. The commenter states that it appears we discounted the
modeled impacts due to recently installed pollution controls, but the FIP TSD does not provide
justification for only evaluating 40% of this facility's impacts when considering whether to
evaluate the facility for additional reasonable progress controls. The commenter states that the
modeled impacts from the Sommers-Deely-Spruce Complex contributed more than 0.3% at Salt
Creek and White Mountains in addition to Wichita Mountains, Big Bend and Guadalupe
Mountains. Due to these visibility impacts we should have included a requirement for the
shutdown of the two J.T. Deely units in the FIP.
Response: The commenter confuses unit level impacts with facility level impacts. The
Sommers-Deely-Spruce complex is comprised of 6 units: two gas-fired units (O.W. Sommers),
two older coal-fired units (J.T. Deely), and two newer coal-fired units (J.K. Spruce). The
modeled percentage of total visibility impairment from the facility are 0.558%) at Wichita
Mountains and 0.569%) at Big Bend. Each of the J.T. Deely units are responsible for
approximately 40% of the total SO2 emissions from the facility so in estimating the maximum
unit level impacts from the individual units at this facility, we applied a 40% factor. On a unit
basis, the estimated percentage of total visibility impairment is 0.223% at Big Bend and 0.221%
at Wichita Mountains. Unit level impacts at all other modeled Class I areas were below these
levels. As explained in our proposed action and FIP TSD, and discussed in detail in a separate
response to comment, we selected a threshold of 0.3% of the total visibility impairment (total
674
-------
extinction) on the 20% worst days to identify units for additional control analysis. This threshold
was established to identify a reasonable set of units that had the greatest visibility impacts for
additional control analysis for this planning period. The units at the Sommers-Deely-Spruce
complex all fell below this 0.3% threshold and we determined no additional control analysis was
necessary for this planning period. However, we noted in the FIP TSD that because the two J.T.
Deely units impacted Wichita Mountains, Big Bend and Guadalupe Mountains at levels around
0.2%, we considered whether further evaluation of controls on these units was appropriate. As
we noted in the FIP TSD, Sommers-Deely-Spruce has indicated that the two J.T. Deely units are
scheduled to be shutdown by 2018, further supporting our decision that no further analysis was
needed if in fact these units would be retired in the first planning period.527
Big Spring Carbon
Comment: EPA Should Have Evaluated Reasonable Controls for the Big Spring Carbon
Black Plant. [Stamper (0068) p.52]
Stamper stated that, according to EPA's FIP TSD, the Big Spring Carbon Black Plant was
modeled to have a contribution to extinction of 0.482%> a Guadalupe Mountains Class I area,
0.304%o at Wichita Mountains Class I area, and 0.173% at Big Bend Class I area. FIP TSD at A-
51. According to the modeling results presented in EPA's "Source Selection Analysis"
spreadsheet, the Big Spring Carbon Black Plant was also modeled to contribute 0.579%> at Salt
Creek Class I area, 0.541% at White Mountains Class I area, 0.482%> at Guadalupe Mountains
Class I area, and 0.304%> at Wichita Mountains Class I area.197 Other Class I areas at which the
Big Spring Carbon Black plant was modeled to contribute between 0.2% - 0.3% include the
Pecos Wilderness Class I area, Wheeler Peak Wilderness Class I area, and the San Pedro Parks
Class I area.198 In fact, the Big Spring Carbon Black Plant's cumulative impacts on visibility at
the 18 Class I areas modeled shows that it has the 7th highest cumulative impacts of the 38
facilities evaluated by EPA.199
Stamper stated that the EPA acknowledged that the Big Spring facility impacts were above 0.3%
and stated that there are 9 units with sizeable emissions. However, EPA did not evaluate the Big
Spring plant for reasonable progress controls, because "it was unclear whether [the 9 units] could
be controlled through one scrubber or would be treated as 9 units with individual impacts much
smaller." FIP TSD at A-54. It appears that EPA did not investigate SO2 scrubber controls or
other reasonable progress controls for the Big Spring Carbon Black plant any further. Given that
the Big Spring Carbon Black plant contributes at least 0.3% to extinction at four Class I areas
and overall is ranked 7th in terms of cumulative visibility impacts at 18 Class I areas, EPA's
decision not to evaluate this plant for controls is not justified.
Stamper stated that the total SO2 and NOx emissions modeled for the Big Spring Carbon Black
Plant were 1,135 tpy for NOx and 17,823 tpy for SO2.200 The Big Spring Black Carbon Plant's
actual emissions in recent years have been reported to be lower than these levels, as shown in the
table below, although the plant also appears to have significant startup and shutdown emissions
and it is not clear whether those emissions are included in the reported annual emissions.
527 See FIP TSD at A-53
675
-------
Table I" Bis Spring
~artwn Black Annual Emissions*1
Year
S02» tpy
NOx, tpy
PM10, i.
2013
- S.-'S
SIS
70
2012
uos
565
70
: 1
6.973
601
74
2010
¦>070
609
75
2000
its ?
567
70
2001
EJ2
551
6S
Stamper noted, according to permit documents available via TCEQ's website, carbon black is
produced at the Big Spring plant in three units using the oil furnace process. TCEQ describes the
carbon black production process at the Big Spring plant as follows:
The oil furnace process produces carbon black by the incomplete combustion of
feedstock oil. The reactor is heated by the combustion of natural gas with low
excess air in the combustion section of the reactor. The feedstock oil is injected
into the reactor downstream of the combustion section, where it undergoes
incomplete combustion, producing carbon black, reduced sulfur compounds
(COS, CS2, H2S), acetylene, and carbon monoxide. The reactor effluent consists
of carbon black suspended in the reactor tail gas. The carbon black is filtered from
the tail gas, pelletized, dried, and stored. A portion of the tail gas is burned in the
dryers. During normal operating conductions, excess tail gas in combusted in an
incinerator."202
According to the Title V permit, Stamper noted that there is one incinerator stack (Unit 13 A) and
two dryer stacks (Units 12A and 7A), as well as three flares and 6 feedstock tanks.203 Thus, there
are three units, not nine as EPA asserted, that EPA could have evaluated for SO2 controls.
According to TCEQ, approximately 35% of the tail gas is used for the dryers204, thus 65% of the
tail gas should go to the incinerator. At the minimum, EPA should have evaluated the addition of
an SO2 scrubber for the incinerator and whether the flue gas from all three emission points (the
two dryer stacks and the incinerator) could be routed through one scrubber. Further, EPA should
consider imposing strict limitations on the operations of the flares, which currently are allowed
to be used for up to 792 hours (almost 10% of the year) each.205 If a scrubber is installed on the
incinerator and if venting tail gas to the flares is limited, that will ensure the scrubber is utilized
to reduce SO2 as much as possible. Last, the feedstock oil sulfur content is allowed to be as high
as 4.0 (annual average) - 4.5% (instantaneous).206 That is a very high allowable sulfur content.
EPA should have evaluated whether lower sulfur content oil could be used as feedstock to reduce
overall SO2 emissions.
Stamper concluded that the Big Spring Carbon Black Plant clearly meets EPA's criteria for
pollution control evaluation based on its contribution to extinction at 7 different Class I areas.
Given the significant and widespread regional haze impacts of the Big Spring Carbon Black
plant and given that there are really only three units to consider for SO2 controls instead of 9
units as EPA indicated in its FIP TSD, EPA does not have adequate justification for failing to
676
-------
evaluate reasonable progress controls for this facility, especially in light of the fact that there are
really only three units to consider for SO2 controls instead of 9 units as EPA indicated in its FIP
TSD. Controlling SO2 emissions at the Big Spring Carbon Black plant would have wide reaching
benefits for the Class I areas impacted by this plant.
Footnotes:
197 See TX116-07-_29_Source_selection_analysis_TX_RH-l-31-14.xlsx at "All Class I areas" tab.
198 Id.
199 Id.
200 Calculated from emissions listed in "AllClassI" tab in TX116-07-_29_Source_selection_analysis_TX_RH-l-31-
14.xlsx.
201 This data was collected from TCEQ's Annual Contaminant Summary Reports, available through TCEQ's Central
Registry Query at http://wwwl5.tceq.texas.gov/crpub/index.cfm?fuseaction=home.welcome.
202 See TCEQ, Construction Permit Amendment, Review Analysis & Technical Review, Big Spring Carbon Black
Manufacturing Plant, Permit No. 6580 at 7th page (Ex. 61).
203 See September 25, 2012 Federal Operating Permit for Sid Richardson Carbon Big Spring Facility, at 16th page
(Ex. 62).
204 See Statement of Basis of the Federal Operating Permit for Sid Richardson Carbon Company Big Spring Facility
at page 3 (Ex. 63).
205 See TCEQ, Construction Permit Amendment, Review Analysis & Technical Review, Big Spring Carbon Black
Manufacturing Plant, Permit No. 6580 at 2nd page (Ex. 61).
206 Id. at 7th page.
[NPS (0077) p. 2-3, 4-5] And, according to the NPS, while Big Spring Carbon's facility impacts
were above EPA's 0.3% visibility impact threshold, EPA did not include Big Spring in its
additional visibility modeling because" ... it was unclear whether [emissions] could be controlled
through one scrubber or would be treated as 9 units with individual impacts much smaller."
Instead, EPA should have conducted a four-factor analysis to resolve these "unclear" issues.
We received comments from NPS and Stamper/Earthjustice et al. that we should have performed
a four factor analysis to evaluate controls at Big Spring Carbon. The commenters state that the
modeled visibility impact for Big Spring was above the 0.3% threshold used to identify sources
for additional control analysis at Wichita Mountains and Guadalupe Mountains. Furthermore,
impacts at some other Class I areas were greater than 0.3% or in the 0.2-0.3% range, and
cumulatively, impacts from Big Spring were seventh highest compared to impacts from other
modeled facilities. The commenters also state that we did not evaluate the Big Spring plant for
reasonable progress controls, because "it was unclear whether [the 9 units] could be controlled
through one scrubber or would be treated as 9 units with individual impacts much smaller." FIP
TSD at A-54. Earthjustice provided a description of the facility from the permit and identified
that according to TCEQ, approximately 35% of the tail gas is used for the dryers, thus 65% of
the tail gas should go to the incinerator, and there are only three units to control. The commenter
continues that we should have evaluated a scrubber for the incinerator, investigated if the
emissions from the dryers and incinerator could be routed to one stack for control, and
considered limiting flaring and the sulfur content of the feedstock. The commenter also notes
that actual emissions from the facility are lower than the emissions modeled by us. The
commenter concludes that the facility meets the criteria for additional control analysis and that
controlling the facility would result in visibility benefits at a number of Class I areas.
677
-------
Response: The commenters confuse facility level and unit level impacts. As explained in our
proposed action and FIP TSD, and discussed in detail in a separate response to comment, we
selected a threshold of 0.3% of the total visibility impairment (total extinction) on the 20% worst
days to identify units for additional control analysis. This threshold was established to identify a
reasonable set of units that had the greatest visibility impacts for additional control analysis for
this planning period. Big Spring Carbon Black Plant was modeled to have a contribution to
extinction of 0.482%> at Guadalupe Mountains Class I area, 0.304%> at Wichita Mountains Class I
area, and 0.173% at Big Bend Class I area on a facility basis. In our proposed action, we stated
that it was unclear from the emission inventory what the individual unit impacts would be and
that they could be divided into up to 9 individual units with much smaller impacts that would fall
below the threshold. Based on additional information from the commenter, actual emissions
from the facility are much less (approximately 50%) than modeled values. Considering actual
emission levels, the estimated impact from the facility falls below 0.3% at any Class I area, and
therefore the impact from any individual unit would fall below the 0.3% threshold used to
identify units for additional control analysis for this planning period. Any increases in actual
emissions at this facility in the future should be considered during development of the regional
haze SIP for future planning periods
Parish and Welsh
Comment: EPA Should Include Emission Limits on Additional Sources In Order To
Make Greater Reasonable Progress. [Earthjustice (0067) p.2, 41]
[Earthjustice (0067) p.2] EPA's analysis demonstrates that significant emissions reductions at the
Welsh and W. A. Parish power plants are cost-effective, would significantly improve visibility,
and meet all four of the statutory factors for reasonable progress controls. EPA should require
controls at the Welsh and W.A. Parish facilities.
[Sierra Club mass mail (0072 and 0073)] Approximately 4,550 commenters submitted email
communications expressing concern that several large coal plants in Texas avoided requirements
to reduce their emissions, including the Welsh and Pirkey coal plants in NE Texas, and the state's
largest coal plant, the NRG Parish plant near Houston.
Two public hearing commenters (0053-11 and 0053-28) specifically suggested that the EPA
consider including W.A. Parish.
[Earthjustice (0067) p. 41] Stamper noted that the EPA found that the visibility benefits from
installing scrubbers on W.A. Parish Units 5, 6, and 7 would not provide enough visibility
improvement to be considered for reasonable progress controls at this time. EPA also found
scrubber upgrades at Welsh Units 1, 2 and 3 were not justified because the visibility
improvements were not large enough. 79 Fed.Reg. 74882 (December 16, 2014). It must be noted
that EPA found that the cost effectiveness of scrubber upgrades at these units was reasonable,
and that visibility is not one of the factors in a four factor reasonable progress analysis. FIP TSD
at 30.
678
-------
Stamper noted that the EPA's rationale for these contrasting decisions is that the Wichita
Mountains (which the W.A. Parish and the Welsh Plants are closest to) has a higher total
extinction for the baseline conditions and 2018 projections, so that the relative improvement in
visibility impairment at Wichita Mountains from scrubber retrofits at the W.A. Parish and Welsh
units would be less noticeable than the visibility improvement at Guadalupe Mountains (which
has lower extinction for baseline and 2018 conditions) from the SDA additions at the Tolk units.
FIP TSD at 28-29, 31.
Stamper stated, as EPA has acknowledged, the use of CAMx modeling to evaluate benefits from
the pollution controls evaluated means that EPA evaluated visibility improvements compared to
a "dirty background" (because CAMX "takes into account the entire pollution load in the
atmosphere in 2018"). FIP TSD, Appendix A at A-37. EPA stated that "[a] facility's visibility
impairment impacts are substantially lower with a dirty background analyses compared to a
clean background analysis." Id. at A-38. EPA further states:
...as a Class I area becomes more polluted, any individual source's contribution to
changes in impairment becomes geometrically less. Therefore the more polluted
the Class I area would become, the less control would seem to be needed from an
individual source.
EPA TSD, Appendix A at A-39, quoting final North Dakota Regional Haze SIP and FIP (77
Fed.Reg. 20912). See also 79 Fed.Reg. 74880-1 (December 16, 2014). EPA states that this is
why visibility benefits of a particular control should be evaluated against natural visibility
conditions, not dirty visibility conditions, and EPA states it was upheld on this point by the Eight
Circuit in North Dakota v. EPA, 730 F.3d 750, 766 (8th Cir. 2013). FIP TSD at A-39.
Earthjustice et al. stated that the EPA's plan must include "all measures necessary to obtain
[Texas's] share of the emission reductions needed to meet the progress goal for" each Class I
area impacted by Texas sources. 40 C.F.R. § 51.308(d)(3)(ii) (emphasis added). W.A. Parish and
Welsh have significant visibility impacts at several Class I areas, including Wichita Mountains,
Big Bend, Guadalupe Mountains, and Caney Creek in Arkansas. EPA found that emissions
reductions at each of the seven units at Welsh and W.A. Parish would be cost effective, and
would yield appreciable visibility improvements at several Class I areas. EPA should revise the
FIP to include SO2 controls at the seven Welsh48 and Parish units. EPA acknowledges that
emission controls at each of the Welsh and W.A. Parish units are within the range of cost that
EPA has previously found reasonable and cost effective. 79 Fed. Reg. 74884. Indeed, there is no
serious dispute that additional controls for each of these units would be cost effective and
reasonable under EPA's four-factor analysis. Nevertheless, EPA concluded that installing or
upgrading the scrubbers at either facility would not provide enough visibility improvement to be
considered for reasonable progress controls at this time. As visibility is not a factor in
determining reasonable progress controls, 42 U.S.C. 7491(g)(1), visibility improvement should
not be used to justify a no control outcome.
However, Earthjustice et al. stated that even if it were appropriate to consider visibility
improvement, controls at Welsh and Parish would provide significant visibility benefits.49
Earthjustice et al. and Stamper noted that EPA's modeling showed that the deciview visibility
679
-------
benefits with scrubber retrofits at the WA Parish units and the Welsh units were greater than the
visibility benefits from the retrofits EPA proposed for the Tolk units. Controls at the Welsh and
Parish units would have benefits ranging between 0.137 - 0.146 dv at Caney Creek alone. As the
exhibits below demonstrate, SO2 controls at the W.A. Parish units would improve visibility at
Wichita Mountains by 0.102 dv to 0.127 dv, providing almost the same visibility improvement
as SO2 controls at Monticello Units 1 and 2 at Guadalupe Mountains National Park and
providing greater visibility benefits than SO2 controls at the Tolk units at Guadalupe Mountains.
In the exhibits below, Stamper pulled out the visibility improvement for each Class I area
compared to natural background just for wet FGD systems at each EGU evaluated, because the
assumed level of control of 98% can be met with either a wet FGD system or a circulating dry
scrubber as discussed above. Stamper ranked units from greatest deciview improvement to least
deciview improvement for these three Class I areas.
Visibility Improvement at Wichita
Mountains from Potential SO2 Controls50
provided by Earthjustice et al. (0067, Table 5
and Stamper (0068, Table 12)
Emission I iur
Dttfrinr Improvement
Big Brown 2
0.438
Big Brown 1
0 456
Monticello 1
0,254
Monticello 2
0 21
Coleto Creek
OZlvj
WA Parish 6
o.ir
WA Pa risk 5
o.i r
VeMi 3
0.1 lo
Welsh 2
0 111
Wri-h 1
<. ii.-i
WA P.i lis li "
o.io;
Tolk 172B
(J 03 7
MI'IB
0 054
Visibility Improvement at Big Bend National
Park from Potential SO2 Controls51 provided
by Earthjustice et al. (0067, Table 6 and
Stamper (0068, Table 13)
Visibility Improvement at Guadalupe
Mountains National Park from Potential
SO2 Controls52
provided by Earthjustice et al. (0067,
Table 7 and Stamper (0068, Table 14)
680
-------
Unit
DffMtw Improvement
CoJeto Creek
0.136
Big Brown 1
1 0.089
Big Brown 2
o m
WA Parish 6
O.Of ¦
WA Parish 5
ii,054
WA Parish 7
0.047
Monticello 1
0.022
Monticello 2
0.020
Tolk 17: B
oiii
Tolk 171B
0,013
Webb 1
0.009
Welsh 2
0,009
Welsh 3
1 n.iw
F inj -toll
Unit
Dedviev Improvement
Big Brown i
0.105
Big Brown 2
0105
Tolk 172B
0,09®
Totkl7JB
0 090
Colefo Creek
0.044
Monticello I
0.02?
WA Parish 6
0.027
Moniieeio 2
0.025
WA Parish 5
0.024
WA Parish '
0,021
Welsh 3
0.013
i
0012
Welsh 2
0.H12
Stamper stated that, as demonstrated in EPA's Cost TSD and in Stamper's analysis (see Table 10
to comment 0068), the cost effectiveness of scrubber retrofits at W. A. Parish Units 5, 6, and 7
are very reasonable. Given EPA's four factor analysis did not identify any issues with scrubber
retrofits at the W.A. Parish units, EPA's decision not to require scrubber retrofits at W.A. Parish
Units 5, 6, and 7 has not been justified.
With respect to the Welsh units, Stamper stated that the EPA also found that SO2 scrubber
retrofits were cost effective, but ultimately EPA did not propose SO2 controls for the Welsh units
because the visibility improvements were not large enough. 79 Fed.Reg. 74882 (December 16,
2014). The modeling presented in EPA's proposed rulemaking shows the most significant
benefits from SO2 scrubber retrofits at the Welsh units at Wichita Mountains Class I area based
on average natural visibility conditions, ranging from 0.109 dv to 0.116 dv as shown in Table 12
above. However, the Welsh power plant is closer to the Caney Creek Class I area in Arkansas, a
Class I area for which EPA has not presented modeling results in its proposed rulemaking. In
fact, the Welsh EGUs are the closest Texas EGU to Caney Creek at 161 kilometers distance,
slightly closer than the Monticello units which are 165 kilometers away. EPA's modeling files
show that SO2 scrubber retrofits at Welsh Units 1, 2 and 3 would result in visibility
improvements of 0.137 dv, 0.140 dv, and 0.146 dv, respectively, at the Caney Creek Class I
154
area.
Earthjustice et al. stated that the EPA's rationale for excluding controls at W.A. Parish and
Welsh is that baseline conditions at the closest Class I area (i.e., Wichita Mountains) are more
impaired than conditions at Guadalupe Mountains, so that the relative improvement in visibility
impairment at Wichita Mountains from scrubber retrofits at the W.A. Parish and Welsh units
would be less noticeable than the visibility improvement at Guadalupe Mountains from the SDA
additions at the Tolk units. FIP TSD at 28-29, 31. This rationale is flawed, for at least two
reasons. First, EPA's analysis seems to imply that Wichita and Guadalupe are in competition to
determine which Class I area will receive benefits from controls; but the fact that controls at Tolk
will benefit Guadalupe Mountains should not be used to forgo improvement at Wichita
Mountains. Improving visibility at the two areas is not mutually exclusive.
681
-------
Second, this rationale undermines the core purposes of the Regional Haze Rule and the Clean Air
Act's visibility provisions. Under this view, Class I areas that are already impaired are less
deserving of protection. Third, while EPA considers the total extinction at Wichita and
Guadalupe, EPA does not consider the significance of visibility improvement in light of the
number of sources impacting the two areas and magnitude of the sources' impacts. If Wichita
suffers from greater light extinction because many sources contribute small amounts to
impairment, then it may be appropriate to address impairment at Wichita by controlling sources
like Welsh and Parish.
Moreover, Earthjustice et al. stated that when Arkansas developed its reasonable progress goal
for Caney Creek, it relied upon TCEQ's projected 2018 SO2 emission rates. As shown in the
table below, however, Welsh's actual SO2 emissions from 2009 through 2013 were significantly
higher than Texas projected. Stamper explained that when the TCEQ consulted with the
Arkansas Department of Environmental Quality on the projected 2018 impacts at Caney Creek
Class I area from Texas sources, TCEQ identified the 2018 SO2 emission rates in the CENRAP
2018 modeling of the Welsh plants shown in the table below. Welsh Units 2 and 3 are emitting
SO2 at much higher rates than was evaluated in the 2018 CENRAP modeling.
Assumed SO2 Emission Rates for Welsh Units in 2018 CENRAP Modeling Compared to
2009 to 2013 SO2 Emissions
provided by Earthjustice et al. (0067, Table 8) and Stamper (0068, Table 15)
Welsh Unit
2018 S02 Emissions
in CENRAP .
Modeling, tpy**5
2009 to 2013 average
SOI Emissions, '
tpy*
¦ :'cent Reduction
Necessary 1© Achieve
Emissions upon
wMcli Caner Creek
RPG Was Based
1
11.721
9.061
*y
1.223
§,453
87%
3
1:22?
9.543
87%
T < ! 1
14»1 "11
28.067
50%
Thus, the RPG for Caney Creek is predicated on emissions from Welsh that are 50 percent less
than actual emissions.55 To meet the RPG for Caney Creek, EPA must require additional
emissions reductions from Welsh. Stamper noted, as the above exhibit demonstrates, the 2018
RPG for Caney Creek is dependent on at least a 50% reduction in SO2 emissions at the Welsh
plant as compared to 2009 to 2013 annual average emissions.157 Stamper stated that if EPA is not
going to require SO2 controls at the Texas EGU that is the closest to the Caney Creek Class I
area, then EPA must revise the Caney Creek RPG to account for this change.
Moreover, even with EPA's proposed FIP on Texas sources, Stamper stated that the EPA has
proposed RPGs for the Wichita Mountains, Big Bend and Guadalupe Mountains Class I areas
that would not achieve natural visibility conditions within the next 50 years (or by 2064).
Specifically, EPA projects that, with its proposed FIP controls, it will take 82 years, 173 years,
and 141 years, respectively, for the achievement of natural background visibility conditions at
Wichita Mountains, Big Bend, and Guadalupe Mountains. 79 Fed.Reg. 74887 (December 16,
2014). Given that EPA has found that it is cost effective to retrofit scrubbers that will reduce SO2
682
-------
by 95-98% from the W.A. Parish Units 5, 6, and 7 and at the Welsh Units 1, 2 and 3, EPA is not
justified in exempting these units from SO2 controls. This is particularly important for the Welsh
units, at which controls will enable the Caney Creek Class I area to reach its RPG.
For all these reasons, and as described in the attached Stamper Report, Earthjustice et al. stated
that the EPA should require SO2 emission controls at Welsh and Parish in order to make
reasonable progress at Wichita Mountains, Big Bend, Guadalupe Mountains, and other Class I
areas. Stamper concluded that EPA has no justification for not requiring the SO2 scrubber
retrofits at the W.A. Parish Units 5, 6, and 7 or at Welsh Units 1, 2, and 3. EPA has found that
SO2 scrubber retrofits are cost effective and reasonable at all of these units. FIP TSD at 30.
Installation of these controls will aid in the achievement of reasonable progress towards the
national visibility goal at several Class I areas.
Earthjustice et al. Footnotes:
48 We note that pursuant to a federal court consent decree with Sierra Club, Welsh Unit 2 is required to retire by
December 31, 2016, which could affect the reasonableness of controls on that unit. See Sierra Club v. U.S. Army
Corps of Engineers, et al.. No. 4:10-cv-04017-RGK (W.D. Ark. Consent Decree entered Dec. 22, 2011).
49 As explained in the attached Report of H. Andrew Gray (0070), on a unit basis, operating controls at the Welsh
and Parish units would improve visibility at Wichita Mountains by more than 0.1 dv— the "threshold" EPA applied
in it's the regional haze context in Wyoming and Arizona. We are not suggesting that the amount of improvement
deemed significant should be the same across the country. However, EPA needs to explain why the improvement it
deemed significant enough to merit controls in two other states' plans is not enough to merit controls in the Texas
plan.
50 Figures are taken from the FIP TSD, Table A.6-4. The table lists the deciview improvement from 98% SO2
control with WFGD or CDS evaluated against natural background conditions.
51 Figures are taken from the FIP TSD, Table A.6-4. The table lists the deciview improvement from 98% SO2
control with WFGD or CDS evaluated against natural background conditions.
52 Figures are taken from the FIP TSD, Table A.6-4. The table lists the deciview improvement from 98% SO2
control with WFGD or CDS evaluated against natural background conditions.
53 See March 25, 2008 letter from TCEQ to ADEQ, at 9 (Table entitled Units Inside the Caney Creek Area of
Influence") in Appendix 4_3b of 2009 Texas Regional Haze plan (Ex. —).
54 2009 to 2013 SO2 emissions data in EPA's cost spreadsheets in EPA-R06-OAR-2014-0754-0008.
55 It must also be noted that even with the retirement of Welsh Unit 2, which is currently slated for December 31,
2016 (FIP TSD at 21), the Welsh plant's SO2 emissions will exceed the level that was modeled for the plant in the
2018 CENRAP modeling based on 2008-2013 average emissions.
Stamper Footnotes:
153 This is discussed in further detail in the Visibility and Health Modeling Technical Support Document submitted
with the Conservation Organizations' comments in this rulemaking.
154 See TX-116-007-_33_Vis_modeling_summary.xlsx.".
155 See March 25, 2008 letter from TCEQ to ADEQ, at 9 (Table entitled Units Inside the Caney Creek Area of
Influence") in Appendix 4_3b of 2009 Texas Regional Haze plan.
156 2009 to 2013 SO2 emissions data in EPA's cost spreadsheets in EPA-R06-OAR-2014-0754-0008.
157 It must also be noted that even with the retirement of Welsh Unit 2, which is currently slated for December 31,
2016 (FIP TSD at 21), the Welsh plant's SO2 emissions will exceed the level that was modeled for the plant in the
2018 CENRAP modeling based on 2008-2013 average emissions.
[Gray (0070) p.20] Dr. Gray noted that the EPA is not proposing to require the retrofit with SO2
scrubbers of the WA Parish (Units 5, 6, and 7) and Welsh (Units 1, 2, and 3). These controls do
provide significant visibility benefits and should be required. On a unit basis, the benefits from
683
-------
each of these controls at WIMO are above 0.1 dv - EPA's estimated "threshold" used in
Wyoming and Arizona- and are well above our revised estimated "threshold" of 0.045 dv. Thus,
using 0.1 -0.15 dv as a threshold (as compared to average natural conditions) would necessitate
including each of the WA Parish Units 5, 6, and 7 and all three Welsh units on the list of sources
recommended for WFGD retrofits.
Dr. Gray noted that wet scrubbers at the Parish and Welsh units provide a benefit of greater than
0.045 dv at 3 Class I areas. Furthermore, WIMO is not the Class I area that benefits the most
from controls on the Welsh units. These controls have benefits ranging between 0.137 - 0.146 dv
at Caney Creek (versus 0.109 - 0.116 dv at WIMO).
Dr. Gray stated that examination of Table A.6-4 in the FIP TSD and Table 5 shows that the
cumulative visibility benefits (delta dv) at all 19 modeled Class I areas, when compared to
average natural conditions, were between 0.33 and 0.43 dv for each individual unit at WA Parish
and Welsh with the exception of WA Parish Unit 8. On an individual unit basis, the cumulative
benefit from these controls are similar to (or above) the cumulative benefits from the controls on
Limestone Units 1 and 2 that EPA is proposing to require (0.401 and 0.433 dv, respectively). The
total of the modeled cumulative visibility impacts for all three WA Parish units was 1.13 dv. The
total modeled cumulative visibility impact for all three Welsh units was 1.23 dv. It is clear that
the visibility benefits offered by these controls is significant enough to warrant their requirement.
Likewise, Dr. Gray stated that the EPA proposed not to require a WFGD upgrade at W.A. Parish
Unit 8. As we have seen, reductions in SO2 from W.A. Parish have visibility benefits at Class I
areas in the region. The SO2 emitted from Unit 8 is no different. And in fact, reducing emissions
from Unit 8 is more cost effective, on a per ton basis, than reductions from WFGD retrofits on
the other three units.27 By the same token, the WFGD upgrade at Unit 8 provides greater dv
benefit per cost than the WFGD retrofits. On a cost per ton and cost per dv basis, this control is
on par with the other recommended controls. It is reasonable and should not be ignored merely
because the sum total of dv improvement is low.
[Gray (0070) p.20] Dr. Gray noted that the EPA's interpretation of the visibility benefits from
the potential controls that it evaluated underestimated the significance of those benefits. EPA
calculated, but failed to consider, impacts to Class I areas outside of Texas and Oklahoma, both
individually and cumulatively. EPA also underestimated how these benefits compare to visibility
improvement provided by previous determinations. EPA's analysis supports both the controls
that it proposed to require as well as those it proposed not to require. (FIP TSD, Section A.6)
Dr. Gray noted that Tables A.6-3 and A.6-4 show the emissions reductions and the
corresponding visibility improvements (dv) that would be expected for SO2 scrubber upgrades or
retrofits on each of the 21 selected units. The modeled visibility benefits are tabulated for
WIMO, GUMO, BIBE, and also for the cumulative benefit across all 19 modeled Class I areas.
Dr. Gray noted that the EPA attempted to establish a reasonable threshold for CAMx-derived
visibility (dv) impacts as compared to those that have been established for previous BART
evaluations using CALPUFF. For example, in their recent FIP for Wyoming, EPA proposed
controlling sources that had a modeled benefit, using CALPUFF, of 0.3 delta-dv. A recently
684
-------
finalized Arizona FIP included controls on sources with modeled visibility benefits of 0.18 and
0.24 delta-dv. Considering the differences between RP analyses using CAMx and BART
evaluations using CALPUFF, and the difference in metrics, EPA concluded that the 0.18 to 0.3
delta-dv benefits obtained with CALPUFF would be on the order of a 0.1 to 0.15 dv benefit with
CAMx modeling. This estimate was based on just the metrics and emissions difference and
ignored the other differences discussed above.
Dr. Gray noted that the EPA evaluated the modeled visibility benefits (dv and extinction) and
other information for each source unit and concluded that "all of the scrubber upgrades in Table
A.6.3 would yield visibility benefits, with the exception of WA Parish Unit 8 which has a very
small benefit." FIP TSD at A-75. Regarding the sources that currently do not currently have SO2
controls, they concluded that "many of the scrubber retrofits in Table A.6.4 would yield visibility
benefits." Id. WFGD retrofits were recommended for all uncontrolled units other than Tolk,
Welsh and WA Parish. SDA scrubbers were recommended for the two Tolk units. EPA's
proposed control scenario was summarized in Table 1, above.
Dr. Gray commented that after reviewing the effectiveness of the various control measures,
EPA's proposed control plan includes control measures on 14 units at seven Texas facilities. The
controls that EPA proposes are reasonable and should be required because they are cost-effective
controls that can be applied to sources that impact visibility at Class I areas.23 Visibility is not
one of the four factors by which reasonable progress controls are evaluated. However, because
EPA has evaluated and used visibility benefits in its review, we discuss EPA's results in our
comments. With that perspective, the controls EPA has proposed to require are clearly
reasonable in that they provide significant visibility benefits at multiple Class I areas.
Dr. Gray stated, when considering the difference in the RP and BART analyses, and the
differences in metrics involved, EPA concluded that recent decisions in Wyoming and Arizona
to require controls that provided benefits of 0.18 to 0.3 delta dv, using CALPUFF, would be
comparable to a delta dv of 0.1 to 0.15 using CAMx.24 As discussed above, the differences due
to metrics is more likely to actually result in a factor of 4, meaning that a visibility improvement
result from CAMx would be roughly 4 times greater when modeled with CALPUFF. The
comparable estimated "threshold" for the modeled impact would then be even lower than 0.1 dv
- reducing 0.18 - 0.3 by a factor of 4 gives a range of 0.045 - 0.075 dv (using an annual average
natural background).25 As demonstrated in Table 5 of comment 0070 (not reproduced here), the
benefits from the proposed controls are all well above this threshold at one or more of the three
Class I areas EPA evaluated in its analysis (WIMO, BIBE, and GUMO).
Furthermore, Dr. Gray contended that these controls provide benefits at the Class I areas that
EPA did not utilize in its analysis. In total, each of the proposed controls has a benefit greater
than 0.045 dv at 2 or more Class I areas (up to 8 Class I areas). The results in Table 5 illustrate
the benefits at other Class I areas as well as the cumulative benefits in each case, which range
from 0.401 to 1.236 dv.
Dr. Gray noted that the EPA is not proposing to require the retrofit with SO2 scrubbers of the
WA Parish (Units 5, 6, and 7) and Welsh (Units 1, 2, and 3). These controls do provide
significant visibility benefits and should be required. On a unit basis, the benefits from each of
685
-------
these controls at WIMO are above 0.1 dv - EPA's estimated "threshold" used in Wyoming and
Arizona- and are well above our revised estimated "threshold" of 0.045 dv. Thus, using 0.1-0.15
dv as a threshold (as compared to average natural conditions) would necessitate including each
of the WA Parish Units 5, 6, and 7 and all three Welsh units on the list of sources recommended
for WFGD retrofits.
Dr. Gray noted that wet scrubbers at the Parish and Welsh units provide a benefit of greater than
0.045 dv at 3 Class I areas. Furthermore, WIMO is not the Class I area that benefits the most
from controls on the Welsh units. These controls have benefits ranging between 0.137 - 0.146 dv
at Caney Creek (versus 0.109 - 0.116 dv at WIMO).
Dr. Gray stated that examination of Table A.6-4 in the FIP TSD and Table 5 shows that the
cumulative visibility benefits (delta dv) at all 19 modeled Class I areas, when compared to
average natural conditions, were between 0.33 and 0.43 dv for each individual unit at WA Parish
and Welsh with the exception of WA Parish Unit 8. On an individual unit basis, the cumulative
benefit from these controls are similar to (or above) the cumulative benefits from the controls on
Limestone Units 1 and 2 that EPA is proposing to require (0.401 and 0.433 dv, respectively). The
total of the modeled cumulative visibility impacts for all three WA Parish units was 1.13 dv. The
total modeled cumulative visibility impact for all three Welsh units was 1.23 dv. It is clear that
the visibility benefits offered by these controls is significant enough to warrant their requirement.
Likewise, Dr. Gray stated that the EPA proposed not to require a WFGD upgrade at W.A. Parish
Unit 8. As we have seen, reductions in SO2 from W.A. Parish have visibility benefits at Class I
areas in the region. The SO2 emitted from Unit 8 is no different. And in fact, reducing emissions
from Unit 8 is more cost effective, on a per ton basis, than reductions from WFGD retrofits on
the other three units.27 By the same token, the WFGD upgrade at Unit 8 provides greater dv
benefit per cost than the WFGD retrofits. On a cost per ton and cost per dv basis, this control is
on par with the other recommended controls. It is reasonable and should not be ignored merely
because the sum total of dv improvement is low.
Dr. Gray stated that Table A.6-5 (page A-76) should include a pair of columns for cumulative
impacts (for all 19 Class I areas), as shown below:
686
-------
t_ utmu.iiivi-.i-j!. i.-s. i ,-y r.v '
Packetound "CI
-------
meet the progress goal for" each Class I area impacted by Texas sources. 40 C.F.R. §
51.308(d)(3)(ii) (emphasis added). Furthermore, we are not justified in exempting these units
from SO2 controls because the Class I areas are not projected to be on track to meet the natural
visibility goal by 2064. In support of these comments, the commenter provides a comparison of
the visibility benefits of all evaluated controls at Wichita Mountains, Guadalupe Mountains, and
Big Bend in terms of deciview improvement ("clean background"). The commenters also
compare the cumulative benefit at all 19 modeled Class I areas for the Parish and Welsh units
(0.33 to .43 dv for each unit) to the Limestone units that we proposed to control (0.401 - 0.433
dv for each unit)528, and provide the facility-wide cumulative visibility benefit from controls at
Parish and Welsh (2.408 dv) compared to the cumulative benefit of the required controls (10.924
dv).529 Furthermore, the commenter states that we did not consider benefits from controlling
Welsh at Caney Creek ranging between 0.137- 0.146 dv which are larger than the benefits at
Wichita Mountains.
Earthjustice et al. states that our rationale for not controlling these sources is flawed for at least
two reasons. First, the analysis seems to imply that Wichita and Guadalupe are in competition to
determine which Class I area will receive benefits from controls; but the fact that controls at Tolk
will benefit Guadalupe Mountains should not be used to forgo improvement at Wichita
Mountains. Improving visibility at the two areas is not mutually exclusive. Second, this rationale
undermines the core purposes of the Regional Haze Rule and the Clean Air Act's visibility
provisions. Under this view, Class I areas that are already impaired are less deserving of
protection. Third, while we consider the total extinction at Wichita and Guadalupe, we do not
consider the significance of visibility improvement in light of the number of sources impacting
the two areas and magnitude of the sources' impacts. If Wichita suffers from greater light
extinction because many sources contribute small amounts to impairment, then it may be
appropriate to address impairment at Wichita by controlling sources like Welsh and Parish.
Earthjustice et al. comments that the current emissions at Welsh are higher than what was
projected in the 2018 emissions used in the CENRAP 2018 modeling that was the basis for
consultations between Arkansas and Texas and establishing the RPG at Caney Creek.
Earthjustice et al. provides a comparison between 2018 modeled emissions and recent average
annual emissions and estimates that a 50% reduction is necessary to meet the modeled emission
level. They note that Welsh unit 2 is scheduled to shutdown in 2016, but this will still leave
actual emissions above modeled levels. Therefore, if we are not going to require SO2 controls at
the Texas EGU that is the closest to the Caney Creek Class I area (Welsh), then EPA must revise
the Caney Creek RPG to account for this change.
Earthjustice et al. also states that the estimated unit level benefits exceed the "threshold" of 0.1
dv used in our actions in Arizona and Wyoming and the revised 0.045 dv threshold that they
develop based on their analysis of the estimated differences between CALPUFF and CAMx
modeling results. Using 0.1-0.15 dv as a threshold (as compared to average natural conditions)
would necessitate including each of the WA Parish Units 5, 6, and 7 and all three Welsh units on
528 See Table 5 of Visibility and Health Modeling TSD to comments from the conservation organizations prepared
by Dr. Gray
529 See Table 6 of Visibility and Health Modeling TSD to comments from the conservation organizations prepared
by Dr. Gray
688
-------
the list of sources recommended for WFGD retrofits. The commenter notes that they are not
suggesting that the amount of improvement deemed significant should be the same across the
country. However, we need to explain why the improvement it deemed significant enough to
merit controls in two other states' plans is not enough to merit controls in the Texas plan. The
commenter also states that the evaluated controls provide benefits at the Class I areas that EPA
did not utilize in its analysis. In total, each of these controls has a benefit greater than 0.045 dv at
2 or more Class I areas (up to 8 Class I areas).
The commenter also discusses upgrading the scrubber at Parish unit 8. Emission reductions are
cost-effective on a $/ton basis and should not be ignored because the sum total of dv
improvement is low.
We also received comments during the public hearing stating that we should have required
controls on Welsh and Parish.
Response: We disagree with this comment. We considered the estimated visibility benefit of
controls on units at Parish and Welsh alongside the four statutory factors in considering whether
controls are cost-effective and will achieve reasonable visibility benefits required during this
planning period towards the national goal. We discuss comments concerning our consideration
of visibility benefits under reasonable progress elsewhere in this document. We determined that
additional controls were not required under the reasonable progress analysis at this time for the
Parish and Welsh units. As discussed in our proposal, we also determined that it was not
reasonable to meet the uniform rate of progress goal for this planning period. We established
RPGs for Wichita Mountains and the Texas Class I areas consistent with our determination of
reasonable progress controls. Therefore, the long term strategy we established for Texas in the
FIP does include all measures determined necessary to obtain Texas' share of emission
reductions for this planning period as required under 40 C.F.R. § 51.308(d)(3)(ii).
Considering the visibility benefits and costs, we disagree that we should have required controls
on units at Parish and Welsh. In evaluating the cost of controls, we also weighed how effective
the reductions were in achieving visibility benefits. We considered the anticipated visibility
benefit in deciviews (for both a "dirty background" and a "clean background") as well as the
reduction in extinction and the percentage of visibility impairment addressed by the controls.
Based on our evaluation of these visibility metrics within the cost factor of the four-factor
reasonable progress analysis, we determined that additional controls on Parish and Welsh were
not required for reasonable progress for the first planning period. In the FIP TSD and the
proposed FIP, we note lesser visibility improvement benefits at the three Class I areas for the W.
A. Parish and Welsh units compared to the benefits at other facilities that mainly impact the
Wichita Mountains. The visibility benefits at Wichita Mountains of scrubber retrofits at the
Parish and Welsh units (0.102—0.126 dv) are approximately half of the visibility benefit of
scrubber retrofits at the Coleto Creek and Monticello units (0.2 -0.254 dv), and much less than
the visibility benefit from controlling units at Big Brown (0.436-0.438 dv). We also note that
when considering the costs of controls and the relative visibility benefit, the Parish scrubber
retrofits would be slightly more expensive with respect to $/ton but would be much less effective
in improving visibility at the Wichita Mountains, when compared to the required controls at units
Monticello or Coleto Creek. For the Welsh scrubber retrofits, the costs ($/ton) are approximately
689
-------
50% greater than the cost of scrubber retrofits at Monticello or Coleto Creek and would result in
approximately 50% less visibility improvement at Wichita Mountains. Furthermore, as we noted
in the FIP TSD and also noted by the commenter, Welsh unit 2 is under a consent decree530 to
shut down by the end of 2016, eliminating the visibility impact from this unit.
As we discuss in a separate response to comment, we focused our analysis on visibility impacts
and benefits at Wichita Mountains, Big Bend and Guadalupe Mountains. We agree that visibility
benefits from controls on Welsh units at Caney Creek are larger than the benefits at Wichita
Mountains. However, we note that the visibility benefit at Caney Creek from controls on the
Welsh units (0.137 - 0.146 dv) are smaller than the benefit from the required controls at Big
Brown (0.179 dv for each unit), Martin Lake (0.35—0.44 dv), and Monticello (0.189 - 0.264 dv)
and much more expensive on a $/ton basis.531
With regards to the comparison of cumulative visibility benefits of controlling units at Welsh and
Parish to those at Limestone, the commenter does not consider the considerable difference in
costs (both $/ton and total capital costs) between a scrubber upgrade and a scrubber retrofit.
Upgrading the scrubbers at Limestone is much less expensive than installing new scrubbers at
the Welsh or Parish units. In addition, we determined that the cumulative visibility benefits of
each new scrubber at the Parish and Welsh units would be less than those at each of the units
where we proposed scrubber retrofits and less than that at each of the units with proposed
scrubber upgrades with the exception of Limestone, at a cost significantly higher than the
estimated cost of scrubber upgrades. Similarly, the total cumulative visibility benefit of
controlling the three units at Welsh and the four units at Parish is less than half the benefit from
all the required scrubber retrofits or all the required scrubber upgrades, and at a greater average
$/ton cost. While controlling the Welsh and Parish units would result in some additional
cumulative visibility improvement, based on our evaluation and weighing of the cost and
consideration of the visibility benefits of these controls at Wichita Mountains, we determined
their individual projected visibility improvements do not merit the installation of scrubbers at
this time. We encourage the State of Texas to re-evaluate this determination as part of its next
regional haze SIP submittal and we note that as the required controls are implemented the
significance of impacts and potential benefits from the Parish and Welsh units will increase in
terms of percentage of extinction. As discussed below, we disagree with comments that this
determination is inconsistent with the determination to require controls at Tolk or with the
determination of required controls in other states for the purpose or reasonable progress.
We disagree with the commenter's statement that the fact that controls at Tolk will benefit
Guadalupe Mountains was used to forgo improvement at Wichita Mountains from controlling
Welsh and Parish. The commenters mistake our explanation for determining that it is reasonable
to control Tolk during this planning period as a rationale for not controlling Welsh or Parish.
530 See Sierra Club et al v. U.S. Army Corps of Engineers, civil 4:10-cv-04017-RGK, also letter from JohnM.
McManus to Mike Wilson, dated May 2, 2013. Under the terms of a consent decree, after the Turk Plant commences
commercial operation, Unit 2 will be restricted to a 60% annual capacity factor during any rolling 12-month period.
Thereafter, Unit 2 must be retired no later than December 31, 2016.
531 See TX-116-007-_33_Vis_modeling_summary.xlsx in the docket to this action for visibility benefits of controls.
Cost-effectiveness of controls on Welsh units are more than $1000/ton more than controls on units at Big Brown,
Monticello, and Martin Lake.
690
-------
As discussed in the FIP TSD, we evaluated visibility benefits at Wichita Mountains and
determined that the visibility improvements for controls at Big Brown, Monticello and Coleto
Creek were significant. We also concluded that scrubber installations on Big Brown 1 and 2
would also yield significant benefits at the Guadalupe Mountains, and that a scrubber installation
on the Coleto Creek unit would also yield significant visibility benefits at Big Bend.532 We
evaluated the visibility benefits of controlling the two units at Tolk at Guadalupe Mountains and
determined that based on evaluation of extinction and percentage of extinction that these controls
would provide for meaningful progress towards the goal of reaching natural visibility conditions
for this progress period at Guadalupe Mountains. From our initial modeling based on the 2018
CENRAP emissions, we estimated that controls on the two units at Tolk would address
approximately 8% of the total visibility impairment from all Texas point sources at Guadalupe
Mountains. Similarly, controlling the one unit at Coleto Creek would address over 6% of the
total visibility impairment from all Texas point sources at Big Bend. In contrast controlling all
three units at Welsh or the three unscrubbed units at Parish would result in addressing a smaller
percent of the total visibility impairment from Texas point sources at Wichita Mountains, and the
required controls at other facilities (e.g. Big Brown, Monticello) result in addressing a larger
percent of the visibility impairment at Wichita Mountains. In explaining our determination in
the FIP TSD, we simply noted that in terms of one of the metrics considered, deciview visibility
benefit, the visibility benefit at Guadalupe Mountains from controlling the units at Tolk was less
than the visibility benefit at Wichita Mountains from controlling the units at Welsh or Parish and
pointed out that our decision also considered other metrics, such as percentage extinction and
extinction relative to the benefit from controls at units at other facilities.
We disagree with the commenter and believe our consideration of these metrics is consistent
with the purpose of the reasonable progress analysis in ensuring that based on consideration of
the four factors, reasonable controls are identified that will allow for reasonable progress during
this planning period Our methodology and metrics identified reasonable controls for sources in
Texas that are cost-effective and result in meaningful visibility improvement towards the goal of
natural visibility conditions.. We agree with the commenter that Wichita Mountains suffers from
greater light extinction because many sources contribute to impairment and we have identified
controls at units across a number of facilities to address a large percentage (-41%) of the
visibility impairment from Texas point sources.533 While controlling Welsh and Parish units
would result in some additional visibility improvement at Wichita Mountains, based on our
evaluation and weighing of the cost and consideration of the visibility benefits of these controls
at Wichita Mountains, we determined their individual projected visibility improvements do not
merit the installation of scrubbers at this time. We encourage the State of Texas to re-evaluate
this determination as part of its next regional haze SIP submittal and note that as the required
controls are implemented the significance of impacts and potential benefits from the Parish and
Welsh units will increase in terms of percentage of extinction.
As to the comment concerning the emissions modeled for Welsh and the RPG for Caney Creek,
we disagree with the commenter that we must require emission reductions at Welsh for Caney
532 FIP TSD at page 28
533 Based on initial modeling using 2018 CENRAP emissions, we estimate that controls on Big Brown, Martin Lake,
Monticello, Coleto Creek, Sandow, and Limestone addresses over 41% of the total impairment due to Texas point
sources
691
-------
Creek to meet the RPG or that we must reset the RPG at Caney Creek if Welsh's emissions are
not reduced. The Caney Creek RPG was established based on CENRAP 2018 modeling results
and projected emissions that included an estimate of emission reductions that would result in
Texas due to compliance with CAIR, not a source-specific limit at any EGU in Texas. As
discussed in the Texas TSD, during consultations for Caney Creek, the participating states
determined that the projected 2018 CENRAP modeling and other findings based on existing and
proposed controls arising from local, state, and federal requirements indicated that Caney Creek
was on the glidepath and projected to exceed (do better than) the URP goal for the first
implementation period ending in 2018. Arkansas Department of Environmental Quality (ADEQ)
determined that additional emissions reductions from other states were not necessary to address
visibility impairment at Caney Creek for the first planning period, and we approved this portion
of the Arkansas regional haze SIP534. Earthjustice et al. provides a comparison between 2018
modeled emissions and recent average annual emissions and estimates that a 50% reduction is
necessary to meet the modeled emission level. The commenter mistakenly uses a theoretical
emission rate that we calculated for other purposes rather than the actual measured emissions at
the facility as reported to the US EPA Clean Air Markets Division. The 2009-2013 average
emissions of SO2 for the facility were reported to be 24,523 tpy, not the 28,067 tpy used by
Earthjustice. We note that the modeled emissions of 14,171 tpy for the three Welsh units is
relatively close to the estimated annual emissions from the facility based on recent actuals after
the shutdown of unit 2 in 2016 of approximately 16,500 tpy and the shutdown will eliminate
visibility impairment from this unit. Furthermore, despite Welsh being the closest facility in our
analysis to Caney Creek, other facilities (e.g. Big Brown, Monticello, and Martin Lake) had
larger impacts due to emissions, stack parameters, and meteorology/transport. Significant
emission reductions at these facilities are required in this action, resulting in significantly more
visibility benefit at Caney Creek than would be achieved by limiting Welsh's emissions to the
levels modeled by CENRAP, and overall less visibility impairment than the ADEQ's established
RPG based on CENRAP's modeling. We also note that we have proposed additional controls in
Arkansas that will result in additional visibility improvements at Caney Creek.535
We disagree that there is an inconsistency in the use of "thresholds" for determination of
reasonable progress controls in Arizona, Wyoming and in determining controls on Welsh and
Parish were not required for reasonable progress this planning period. Earthjustice et al. states
that the estimated unit level benefits exceed the "threshold" of 0.1 dv used in our actions in
Arizona and Wyoming and the revised 0.045 dv threshold that they develop based on their
analysis of the estimated differences between CALPUFF and CAMx modeling results and that
we must explain why these visibility benefits were considered significant enough to merit
controls in Arizona and Wyoming but not in Texas. We agree with the commenter that the
amount of improvement deemed significant should be determined on a case by case basis and
might not be the same across Class I areas. There is no explicit threshold for determining
significance of visibility benefit in the regional haze rule. Significance is a source- and Class I
area-specific evaluation, meaning that it depends on how much visibility improvement is needed
at the Class I area(s), how much a specific source impacts the Class I area(s), and the cost
effectiveness and potential visibility improvement of available control options. States have
534 77 FR 14604
535 80 FR 18944
692
-------
latitude to determine these thresholds536, providing support and a reasonable and adequate basis
for why they selected the thresholds, and to determine BART and reasonable progress controls,
in consultation with other impacted states. In the case of Wyoming, our proposed action found
that Low NOx burners with Overfire Air (LNB w/ OF A) were reasonable controls on the Dave
Johnston units 1 and 2, based on consideration of the low costs of approximately $1000/ton and
visibility improvement from controls at each unit of about 0.3 dv (CALPUFF modeling).537 In
our final action we determined based on revised modeling that controls on Dave Johnston units 1
and 2 ranging from 0.11 to 0.12 dv based on CALPUFF modeling were not necessarily justified
for this planning period.538 In Arizona, controls were determined to be required for the Rillito
Plant Kiln 4 based on estimated visibility improvements of 0.18 dv and 0.16 dv with CALPUFF
modeling at the two most impacted Class I areas in Arizona, the higher anticipated improvement
in the western unit of Saguaro National Park, and a cost-effectiveness of approximately
$l,850/ton.539 As we discuss in depth elsewhere in this document, CAMx modeled visibility
impacts and benefits are generally much lower than the CALPUFF modeled visibility impacts
and benefits relied on in other actions due to the differences between these two modeling
platforms, the model inputs, and the metrics used. However, due to all of the differences in the
CALPUFF and CAMx model results, it is not possible to directly compare these model results or
develop thresholds based on these comparisons. We evaluated these recent FIPs that included
controls for reasonable progress using CALPUFF modeling, and conservatively estimated that
just based on emissions and metric differences, the visibility benefits in those actions would be
well in the range of CAMx modeled visibility benefits for the required controls in this action.
To evaluate the projected visibility benefits of controls from our CAMx modeling in our cost
evaluation, we considered a number of metrics, such as change in deciviews under 2018
projected levels of air pollution at the three Class I areas and under estimated natural visibility
conditions, change in light extinction, and change in the percentage of total light extinction. We
also considered the visibility benefit of emission reductions from recent actual emission levels
versus CENRAP 2018 projected emission levels at these sources. As we discuss further in our
FIP TSD and in responses in our RTC document, to provide context regarding the significance of
individual source impacts, we compared the individual source impacts with CENRAP source
apportionment modeling results for impacts from all emission sources within a state and impacts
from all emission sources within a state within a specific source type. We also compared these
individual source impacts to the impact levels used by the states for triggering consultation with
another state about its overall impacts, and the estimated range of anticipated visibility benefits
resulting from required controls in other actions.540 The determination of reasonable progress
controls must be based on consideration of a number of factors, and to fully consider the
visibility benefits, several metrics should be evaluated. As we discussed with relation to Tolk's
impacts on Guadalupe Mountains, extinction and percentage extinction are also important
536 BART guidelines at 70 FR 39170: However, we believe the States have flexibility in setting absolute thresholds,
target levels of improvement, or de minimis levels since the deciview improvement must be weighed among the five
factors, and States are free to determine the weight and significance to be assigned to each factor. For example, a
0.3, 0.5, or even 1.0 deciview improvement may merit stronger weighting in one case versus another, so one "bright
line" may not be appropriate.
537 78 FR 34785
538 79 FR 5051
539 79 FR 52420
540 FIP TSD at A-75
693
-------
factors. In both Wyoming and Arizona, the impacted Class I areas are projected to be much
cleaner in 2018 than Wichita Mountains. The appropriate "threshold" in terms of dv may need
to be lower in order for progress to be made towards natural conditions at cleaner Class I areas.
Failure to take this into account would result in eliminating reasonable controls on potential
sources as a Class I area approaches its reasonable progress goal (e.g. cleaner Class I airsheds),
having the undesired effect of increasing the difficulty of meeting the RPG.
Lastly, we agree with the commenter that on a $/ton basis, scrubber upgrades on Parish unit 8 are
very cost-effective. However, the visibility benefit and reduction in emissions from this control
is very low when compared to all the other evaluated scrubber upgrades. The estimated visibility
benefit from upgrading the scrubber would be an order of magnitude less than all the other
evaluated scrubber upgrades and not large enough to require as reasonable progress for this
planning period.
15.o Consideration of NOx Controls
Comment: Selection of pollutants for reasonable progress analysis [NPS (0077) p. 1-2]
The NPS agreed with the Texas Council on Environmental Quality and EPA that"... the
predominant anthropogenic pollutants that affects the state's ability to meet the URP goals in
2018 on the worst 20% days at the Texas Class I areas are largely due to sulfate and nitrate,
primarily from point sources." However, EPA states that,"... we are limiting our analyses to the
consideration of SO2 controls for these EGU sources, as our modeling indicates that the impacts
from these sources on the 20% worst days are primarily due to sulfate emissions." EPA has
focused its analysis on controlling SO2 on the premise that NOx (and nitrate) is not the
predominant cause of visibility impairment. While this may be true in many of the Class I areas
impacted by emissions from Texas sources, IMPROVE data provided in EPA's Table 25 show
that the nitrate contribution from Texas sources is almost as great as their sulfate contribution at
Salt Creek Wilderness Area in New Mexico, which may be indicative of nitrate impacts at
Carlsbad Caverns NP. Although EPA does not specify the contribution from Texas at Carlsbad
Caverns NP, its Table 25 does show 27% of the nitrate impact at Salt Creek WA attributed to
Texas. Additionally, while nitrate contributions at Guadalupe Mountains NP are lower than at
Salt Creek WA, they are still significant.
According to the NPS, regional Haze regulations (40 CFR 308(d)(1)) require that the reasonable
progress goals must provide for an improvement in visibility for the 20% most-impaired days
and ensure no degradation of visibility on the 20% least-impaired days. Nevertheless, EPA
approved reasonable progress goals for New Mexico that show degradation on the 20% least-
impaired days for Carlsbad Caverns NP. EPA attributed 44% of the nitrate impairment and 26%
of the sulfate impairment at Carlsbad Caverns NP to emissions sources in CENRAP (including
Texas). EPA noted that, for both Salt Creek WA and Carlsbad Caverns NP, "[contributions of
nitrate from CENRAP states and New Mexico from mobile sources are projected to decrease
significantly, while contributions from area source emissions, including emissions from oil and
gas production in New Mexico and the CENRAP states are projected to increase."
694
-------
The NPS stated that nitrate is a significant contributor to visibility impairment in some Class I
areas impacted by emissions from Texas sources, and Texas sources contribute significantly to
the projected degradation of visibility of the least-impaired days at Carlsbad Caverns NP.
Because we do not have an IMPROVE monitor at Carlsbad Caverns NP, we are concerned that
impacts similar to those at Salt Creek WA may be occurring at Carlsbad Caverns NP, which is
near several sources of NOx (oil and natural gas) and ammonia (cattle feed lots). As discussed
later, we request that EPA begin consideration of the impact of oil and natural gas development
in western Texas and southeastern New Mexico on visibility at nearby Class I areas.
[NPS (0077) p. 4] The NPS suggested that nitrate is a significant contributor to visibility
impairment in some Class I areas impacted by emissions from Texas and sources of NOx that
impact those Class I areas should be evaluated.
[Stamper (0068) p.5] Stamper noted that EPA has proposed scrubber retrofits on 7 EGUs that
currently have no SO2 controls. EPA only evaluated controls on SO2 because its modeling
indicated that the impacts from these sources are due primarily to sulfate emissions. FIP TSD at
3. While the majority of the Class I impacts from Texas sources do seem to be dominated by
sulfur dioxide emissions, there are some facilities that EPA should have evaluated for NOx
emissions.
[Sierra Club mass mail (0072 and 0073)] The commenters stated that it also appears that the EPA
is not asking the coal plants to do anything to reduce their emissions of nitrogen oxides (or
NOx), as was done in the neighboring state of New Mexico. The commenters asked that the EPA
please consider the impacts of NOx pollution and the additional coal plants as the rule is
finalized in 2015.
Multiple public hearing commenters also suggested that EPA should have included NOx
emissions in the proposal (0053-11, 0053-26, 0053-28, and 0053-46).
The NPS suggested that nitrate is a significant contributor to visibility impairment in some Class
I areas impacted by emissions from Texas and sources of NOx that impact those Class I areas
should be evaluated. NPS asserts that we focused our analysis on controlling SO2 on the
premise that NOx (and nitrate) is not the predominant cause of visibility impairment. At Salt
Creek, the nitrate impact from Texas sources is 27% of the total nitrate impact and is nearly as
big as the sulfate impact from Texas sources. NPS states that nitrate contributions at Salt Creek
and Guadalupe Mountains are significant, and it is likely this is also the case at Carlsbad
Caverns. We also received comments from Earthjustice et. al and others that other facilities
should have been evaluated for NOx controls and that NOx controls should have been evaluated
for the facilities that we are requiring install SO2 controls.
Response: With regards to comments on additional controls for NOx, as discussed in the
proposed FIP, we agree with Texas that the predominant anthropogenic emissions impacting
visibility are nitrate and sulfate emissions, primarily from point sources.541 Projected 2018
statewide emissions of SO2 are largely due to point sources, and approximately 50% of statewide
NOx emissions are from point sources. The remaining portion of NOx emissions are roughly
541 79 FR 74838
695
-------
evenly distributed among area and mobile sources. NPS is incorrect in their assertion that we
focused our analysis on SO2 controls because nitrate is not the predominant cause of visibility
impairment at the impacted Class I areas. As described in more detail in the FIP TSD, in our
initial analysis we focused on point sources and we identified facilities with the greatest potential
to impact visibility based on a Q/d analysis considering both SO2 and NOx emissions. We then
used photochemical modeling to estimate the visibility impacts due to the emissions from these
facilities, considering SO2, NOx and all other emitted pollutants. Based on the results of that
visibility modeling, we identified a subset of facilities for additional control analysis . At this
point, we examined the source apportionment results for each identified facility and determined
that with the exception of the PPG GlassWorks facility, the visibility impacts due to these
facilities was almost entirely due to their sulfate emissions. Therefore, we determined that to
address the visibility impacts on the 20% worst days from these sources, it was only necessary to
evaluate sulfate controls for this planning period. Because impacts from the PPG Glassworks
were significant and NOx emissions were responsible for a large portion of that impact, we
investigated NOx controls for this facility as discussed in a separate response to comment in this
section of the RTC document. We note that the Q/d analysis did identify a number of facilities
based on emissions and their distance to Guadalupe Mountains or Carlsbad Caverns.542
Photochemical modeling showed that visibility impacts from these sources (considering NOx,
SO2 and other pollutant emissions combined) were small and no additional control analysis for
these sources was needed for this planning period.543
While overall visibility conditions at some Class I areas may show a significant impact from
nitrate, sulfate impacts were significant and these sulfate impacts are primarily due to point
sources. The contributing NOx emissions are spread out over numerous sources and source
categories, including mobile sources. For Salt Creek, 2018 CENRAP modeling shows a
contribution of 2.43 Mm-1 from nitrate from Texas sources (point, area, and mobile sources
combined) compared to 3.5 Mm-1 from sulfate from Texas sources (primarily point sources).
Our analysis identified those sources that had the greatest visibility impacts, which we then
further analyzed for controls. This analysis did not identify any individual point sources (with
the exception of PPG Glassworks) with significant visibility impacts due to NOx emissions
among the group of sources with the greatest visibility impacts. We address additional
comments concerning impacts at Carlsbad Caverns in a separate response to comment above.
We address comments concerning oil and gas sources in separate response to comments below.
15.p Oil and Gas Sources
Comment: EPA Should Analyze Reasonable Progress Controls for Oil and Gas Sources.
[Earthjustice (0067) p. 47]
Earthjustice et al. stated that the EPA proposed to "agree with the TCEQ's decision to focus the
analysis of the four statutory factors on point sources." 79 Fed. Reg. at 74,838. In so doing, it
542 For example, Fullerton Gas Plant, Goldsmith Gas Plant, Guadalupe Compressor Station, Keystone Compressor
Station, Keystone Gas Plant and Pegasus Gas Plant
543 See the FIP TSD and "EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_072913.xlsx" in the docket for
visibility modeling results for these facilities.
696
-------
failed to review at least one category of non-point sources that impacts visibility and for which
reasonable controls are available, namely area sources of NOx and specifically NOx emissions
from oil and gas sources. EPA should disapprove Texas's failure to review area sources of NOx,
and should evaluate and require reasonable controls on these sources.
Earthjustice et al. stated that the EPA and Texas both acknowledge that anthropogenic nitrates
are a significant source of visibility impairment at Class I areas in the region, and that sources of
nitrates were relatively evenly distributed among point, mobile, and area sources as of the base
year 2002. 79 Fed. Reg. at 74,834, 74,838. In that modeling, area sources comprised between 12
and 17% of the nitrate impacts at the Class I areas most impacted by Texas sources. Id. This is a
significant proportion unto itself, and as discussed below, it was inappropriately dismissed by
both Texas and EPA. Given that both point and mobile source NOx emissions have decreased
while area source emissions have not, the proportion of nitrate impacts from area sources is
likely to have increased significantly since the 2002 baseline.56
Furthermore, area source emissions as a whole are not enforceably limited - unlike a major
stationary source which is limited to existing emission rates and must be re-evaluated for
significant increases, area sources are not evaluated comprehensively. The potential for
unevaluated, uncontrolled growth makes establishment of reasonable progress controls on
existing and new sources all the more important to ameliorate existing and prevent future
impairment.
Earthjustice et al. stated that Texas noted in its SIP that the largest contributor to area source
NOx was upstream oil and gas emissions. However, it failed to present a detailed breakdown of
emissions, so it is impossible to determine what percent of area source NOx was included in this
category, and whether there were other categories that should also have been evaluated for
reasonable progress controls.
With regard to upstream oil and gas emissions, Earthjustice et al. stated that Texas advanced
three rationales for dismissing evaluation of controls: first, that it was taking all measures it had
determined were reasonable in its Dallas-Fort Worth (DFW) ozone SIP; second, that it was
implementing a grant program for retrofitting gas-fired, rich burn compressor engines; and third,
that it would reexamine these sources in its five year progress report. Texas SIP Appendix 10-1
("By that time, we expect to have much improved information on the inventory and the
economic and technical feasibility of additional controls."). Each of these rationales is flawed
and none considers the statutory four factors involved in a reasonable progress analysis. Texas's
rationale should be rejected by EPA in favor of an actual control evaluation.
Earthjustice et al. stated that Texas's idea of what was reasonable in the context of its DFW
ozone SIP is not necessarily the same as what is required for reasonable progress under the
Regional Haze Rule, and EPA has the responsibility to review this analysis in the context of the
Regional Haze Rule. Further, the DFW SIP covers only a small portion of the state. As seen in
the image below, oil and gas wells are present throughout the state, including significant
development in close proximity to Wichita Mountains, Guadalupe Mountains, Carlsbad Caverns,
and Salt Creek. There are major NOx emissions outside of the DFW area - the Permian Basin in
697
-------
western Texas, for instance, is estimated to be responsible for roughly 166,000 tons of NOx per
year.57 At a minimum, the controls evaluated for the DFW area could be implemented elsewhere.
Earthjustice et al. provided a map of active oil and gas wells in Texas (Figure 1 in comment
0067, not reproduced here). 58
Earthjustice et al. noted that Texas's second rationale for not considering controls on oil and gas
sources—its reference to its grant program—only serves to highlight the fact that there are
available controls for at least some of these sources that could be evaluated and required.
Earthjustice et al. argued that Texas's third reason fares no better. Texas promised to re-examine
oil and gas sources in its five year progress report. 79 Fed. Reg. at 74,834 ("The TCEQ also
noted uncertainty in upstream oil and gas emission estimates."). However, incomplete
information is not an excuse for failing to review the information that exists when the SIP was
developed. As EPA noted in its evaluation of Wyoming's reasonable progress review, "[i]f the
State determined that additional information was needed to potentially control oil and gas
sources, the State should have developed the information." 78 Fed. Reg. at 34,765.
Furthermore, Earthjustice et al. explained that Texas's claim that it would re-examine sources of
oil and gas emissions in its five year progress report has been disproven by the progress report
Texas recently submitted to EPA. Although its final submission contains a discussion of oil and
gas emissions that was not present in the version made available for public review,59 it is limited
entirely to a summary of Texas's attempts to improve its emissions inventory. The promised
discussion of "economic and technical feasibility of additional controls" - or even a meaningful
summary of the inventory itself relative to the 2002 baseline or 2018 projections - does not exist.
For these reasons, Earthjustice et al. stated that the EPA's proposed approval of Texas's
dismissal of area sources of NOx is inappropriate. Texas did not perform a four factor analysis of
this significant source of visibility impairing pollution. EPA should disapprove the oil and gas
portion of Texas's haze SIP and substitute a thorough control analysis based on updated
information and consideration of the four reasonable progress factors. It has been widely
demonstrated that the emissions from these sources are significant and that reasonable controls
are available, cost-effective and are otherwise likely to result in controls following a four factor
analysis.60
Footnotes:
56 See, e.g., Texas 2014 Five-Year Regional Haze State Implementation Plan Revision, February 26, 2014, p. 4-5
through 4-7. [This document is attached to comment 0067 - Item 14]
57 2011 Oil and Gas Emission Inventory Enhancement Project for CenSARA States, Table E-2.
58TCEQ, Texas Oil and Gas Wells, available at
http://www.tceq.state.tx.us/assets/public/implementation/barnett_shale/bs_images/txOilGasWells
.png [This document is attached to comment 0067 - Item 15]
59 Texas 2014 Five-Year Regional Haze State Implementation Plan Revision, Proposal, June 18, 2013. [This
document is attached to comment 0067 - Item 13]
60 See, e.g., 2011 Oil and Gas Emission Inventory Enhancement Project for CenSARA States; Summary of Oil and
Gas Sector TSD: Significant Stationary Source of NOx Emissions, October 2012; Conservation Organization's
August 26, 2013 comments to EPA Re: Docket ID No. EPA- R08-OAR-2012-0026, Comments on EPA's Re-
Proposed Approval, Disapproval, and Promulgation of Implementation Plans; State of Wyoming; Regional Haze
State Implementation Plan; Federal Implementation Plan for Regional Haze, 78 Fed. Reg. 34,738 (June 10, 2013);
698
-------
Conservation Organizations' comments to EPA Re: Docket ID No. EPA-R08- OAR-2012- 0026, comments on
EPA Proposed Approval, Disapproval, and Promulgation of Implementation Plans; State of Wyoming; Regional
Haze State Implementation Plan; Federal Implementation Plan for Regional Haze; Environmental Commenters'
August 20, 2014 comments to EPA Re: Comments of Clean Air Task Force, National Parks Conservation
Association, Earthjustice, WildEarth Guardians, and Southern Utah Wilderness Alliance on Managing Emissions
From Oil and Natural Gas Production in Indian Country: Advanced Notice of Proposed Rulemaking, 79 Fed. Reg.
32,502 (June 5, 2014). [Document is attached to comment 0067 - Items 11, 12, 16 and 17]
Earthjustice et al. commented that we failed to review area sources of NOx and specifically NOx
emissions from oil and gas sources. We should have disapproved Texas RH SIP for failing to
perform a four factor analysis on NOx area sources and should have evaluated and required
reasonable controls on these sources. Anthropogenic nitrates are a significant source of visibility
impairment, and sources of nitrates were relatively evenly distributed among point, mobile, and
area sources as of the base year 2002. The 2002 CENRAP modeling showed area sources
comprised between 12% and 17% of the nitrate impacts at the Class I areas most impacted by
Texas sources. This is a significant portion and while point and mobile source NOx emissions
have decreased, impacts from area sources have likely increased. The TX RH SIP states that oil
and gas sources are the largest contributor of NOx to area source NOx emissions but provides no
additional information quantifying oil and gas versus other area sources.
Earthjustice summarizes the TX RH SIP with regards to dismissing evaluation of controls for oil
and gas emissions: first, that it was taking all measures it had determined were reasonable in its
Dallas-Fort Worth (DFW) ozone SIP; second, that it was implementing a grant program for
retrofitting gas-fired, rich burn compressor engines; and third, that it would reexamine these
sources in its five year progress report. Earthjustice states that there are flaws in this reasoning
because the DFW ozone SIP is limited to sources in that region and that controls considered
under regional haze may differ from those considered for ozone planning, the grant program
serves to demonstrate the availability of controls, and finally that the five-year progress report
fails to examine additional controls for these sources. Lastly, the commenter provides additional
information that they explain demonstrates that these emissions are significant and controls are
available and cost-effective.
Response: As discussed in the proposed FIP, we agreed with Texas that visibility impairment
due to anthropogenic emissions is largely due to sulfate and nitrate, primarily from point sources.
Projected 2018 statewide emissions of SO2 are largely due to point sources, and approximately
50% of statewide NOx emissions are from point sources. The remaining portion of NOx
emissions are roughly evenly distributed among area and mobile sources. As discussed in detail
elsewhere, we evaluated Texas point sources and identified a reasonable set of sources with the
greatest estimated visibility impacts for additional control analysis. In identifying these sources
we considered impacts from emissions of NOx, SO2 and other visibility impairment pollutants.
Visibility impacts from NOx emissions from area sources are relatively small compared to
impacts from point sources of SO2 and NOx at the Class I areas impacted by Texas emissions.
Table 4 of the TX TSD (copied below) summarizes the percentage of sulfate and nitrate impacts
from Texas sources for point, mobile and area sources at five Class I areas impacted by Texas
emissions for 2002. Impacts from NOx emissions from area sources range between 12% and
17%) and are less than half the impact from NOx point source emissions at these Class I areas
(ranging from 26.6%> to 35.8%>). The 2018 modeling projections estimate impacts from NOx
699
-------
emissions from all Texas area sources combined will comprise 1.08% of the total projected
visibility impairment in 2018 at Salt Creek, compared to 9.03% from Texas point sources (NOx
and SO2 combined). The 2018 modeling projections estimate impacts from NOx emissions from
all Texas area sources combined will comprise 1.35% of the total projected visibility impairment
in 2018 at Wichita Mountains, compared to 13.22% from Texas point sources (NOx and SO2
combined).544 Area source NOx impacts at other Class I areas are smaller in terms of percent
impact than at these two Class I areas. Oil and gas emissions are the largest component of area
source emissions but are only part of the total NOx area source emissions. Furthermore, impacts
from sulfate are larger than nitrate impacts on the 20% worst days at all Class I areas.545 Based
on our review of the CENRAP modeling results and the TCEQ's analysis in Chapter 11 and
appendix 10-1 of the Texas regional haze SIP we found that the predominant anthropogenic
pollutants that affects Texas' ability to meet the URP goals in 2018 on the worst 20% days at the
Texas Class I areas are largely due to sulfate and nitrate, primarily from point sources. We
focused on point source emissions of NOx and SO2 and evaluated controls on those sources with
the largest impacts on visibility for this planning period.
We agree with the commenter that cost-effective controls have been identified for some area
sources associated with oil and gas production. However, we agree with Texas that it was
reasonable to not examine these sources for additional control at this time as these sources are
not the primary contributors to visibility impairment for this planning period. We evaluated
Texas point sources and identified a reasonable set of sources with the largest estimated visibility
impacts for additional control analysis. Similar to that process of eliminating those point sources
that have less potential to contribute to visibility impairment from additional control analysis, we
eliminated area sources from additional control analysis for this planning period. As the
visibility impact from the identified point sources are reduced through reasonable controls, the
impact from area sources will become a larger percentage of the total visibility impairment and
should be addressed in future planning periods. As discussed in the Texas regional haze SIP546,
improved emission inventory information, along with experience gained in implementation of
controls for the Dallas/Fort Worth area and through the grant program established to assist with
the retrofitting of gas-fired, rich burn compressor engines, will be useful in consideration of
controls for these sources in future planning periods. Texas committed to additional examination
of these sources and has included information on current efforts to improve the emission
inventory in their regional haze Progress Report SIP. Any specific comments concerning the
adequacy of the TX Progress Report SIP will be addressed when we take action on that separate
SIP submittal. We note that any new sources such as oil and gas sources are covered by new
source performance standards.
Percentage Source Category Contributions to SO4 and NO3 at the Five Class I Areas Texas Most
Impacts
Big Bend Guadalupe Mountains
544 CENRAP PSAT data available in the docket for this action as
CENRAP_PSAT_Tool_ENVIRON_Aug27_2007.mdb
545 See Tables 25 and 26 of the TX TSD for a summary of nitrate and sulfate impacts from Texas sources on Class I
areas in nearby states.
546 See section 10-1.2 of the Texas regional haze SIP
700
-------
Point
Mobile
Area
Point
Mobile
Area
so4
67.1
2.8
6.9
75.6
3.5
8.5
N03
26.6
28.6
14.3
29.2
36.5
13.9
Wichita Mountains
Salt Creek
White Mountain
Point
Mobile
Area
Point
Mobile
Area
Point
Mobile
Area
S04
78.2
3.7
9.2
73.8
3.9
8.1
75.2
4.1
8.1
N03
28.1
44.7
13.4
35.8
29.9
17.1
27.9
40.3
12.0
Comment: Oil and gas emissions near Carlsbad Caverns NP and Guadalupe Mountains NP
[NPS (0077) p. 4 and 0077-A3]
The NPS suggested that the EPA should consider emission reductions from some oil and natural
gas facilities in western Texas by addressing combined emissions from this small number of
sources.
The NPS expressed concern about the increasing emissions from oil and natural gas development
near Carlsbad Caverns NP and Guadalupe Mountains NP. The NPS provided a spreadsheet
(0077-A3) with Texas and New Mexico Q/D values for several oil and gas plants. For example,
the 2011 National Emissions Inventory (NEI) includes eight oil and gas facilities in Texas with
combined SO2, NOx, and PM10 emissions of over 10,000 ton/yr and cumulative Q/d greater than
10 at both Carlsbad Caverns NP and Guadalupe Mountains NP. In New Mexico, the 2011 NEI
includes 14 oil and gas facilities with combined SO2, NOx, and PM10 emissions of over 19,000
ton/yr and cumulative Q/D greater than 10 at Carlsbad Caverns NP and Guadalupe Mountains
NP. The NPS noted that EPA recommended future consideration of controls on natural gas
facilities in Arizona as part of its FIP actions there, and that EPA determined that Non-Selective
Catalytic Reduction represented reasonable progress at a natural gas compressor station in its
Montana FIP. The NPS requested that EPA consider similar actions in Texas by addressing
combined emissions from this small number of oil and gas sources.
Response: NPS provided a spreadsheet containing 2011 emission inventory and Q/d data for 8
011 and gas facilities in Texas. Of those facilities, only Fullerton Gas Plant and Goldsmith Gas
Plant had a Q/d value greater than 10 for either Carlsbad Caverns or Guadalupe Mountains. As
discussed in a separate response to comment we used a Q/d threshold of 10 to initially identify
those individual facilities with the potential to significantly impact visibility at a nearby Class I
area. The Fullerton and Goldsmith facilities were included in our initial source apportionment
modeling to estimate the visibility impact of these facilities. The 2018 modeled visibility impact
was 0.106% of the total visibility impairment from the Fullerton Plant at Carlsbad Caverns and
Guadalupe mountains. The 2018 modeled visibility impact from Goldsmith was 0.072% of the
total visibility impairment at Carlsbad Caverns and Guadalupe Mountains. This is much smaller
than the modeled impact from Tolk (0.646%) or Big Brown (0.502%) and well below the 0.3%
unit impact threshold we applied in identifying sources for additional control analysis. We
anticipate that individual impacts from the other 6 facilities would also be small considering their
emissions and distance from these Class I areas. The NPS also provided information on
emissions and Q/d for 14 oil and gas facilities in New Mexico. We reviewed and approved New
701
-------
Mexico's RH SIP addressing reasonable progress and long-term strategy requirements in a
separate action.547 Furthermore, SO2 emissions from these facilities with SO2 emissions greater
than 100 tpy are limited under the SO2 Milestone and Backstop Trading Program.548 As the
commenter states, in the Montana FIP we determined that Non-Selective Catalytic Reduction
represented reasonable progress at a natural gas compressor station. We note that this facility
was identified for additional analysis because it had a Q/d value greater than 10. We also note
that this is the only source with required controls in Montana under the reasonable progress
requirements. As discussed elsewhere, we initially relied on a Q/d analysis to identify 38 sources
with the greatest potential to impact visibility. We then conducted a more refined analysis using
photochemical modeling to estimate the visibility impacts from the 38 sources and identify the
subset of sources with the largest visibility impacts. Some of these sources may be identified for
additional control analysis in future planning periods as the sources with the largest visibility
impacts and potential for visibility benefit are controlled. We address comments concerning oil
and gas sources that fall under the area source category in a separate response to comment above.
15.q Identification of Sources using Q/d and Photochemical Modeling
General Summary: We received comments on the methodology used to identify sources for
analysis. Commenters stated that our analysis, beginning with a Q/d analysis and the use of a
0.3% of total impairment threshold for identifying sources for additional analysis was arbitrary,
capricious, or improper. In addition, commenters contend that the Q/d analysis selects the wrong
sources because it does not consider stack parameters or meteorology. Other commenters
suggested that all 38 facilities identified as having the greatest potential to impact visibility by
the Q/d analysis should have undergone a four-factor analysis. We also received comments that
a lower threshold should have been used, that the threshold was applied inconsistently, and that
the 0.3% threshold screened out sources that have a significant visibility impact and should have
been evaluated for controls.
Q/d Analysis
Q/d General summary: We received comments on the methodology used to initially identify
sources for analysis. Commenters stated that EPA included both NOx and SO2 emissions for the
Q/d screening even though the subsequent control analysis only focused on SO2 emissions. In
addition, the Q/d analysis selects the wrong sources because it does not consider stack
parameters or meteorology. Back-trajectory data also shows that EPA selected the wrong group
of sources.
We also received comments that our application of Q/d is inconsistent with other states in that it
does not capture the majority of emissions in the State, comments that we should have used a
547 77 FR 70693
548 Under Section 309 of the Federal Regional Haze Rule, nine western states and tribes within those states have the
option of submitting plans to reduce regional haze emissions that impair visibility at 16 Class I areas on the
Colorado Plateau. Three states ~ New Mexico, Utah, and Wyoming ~ and Albuquerque-Bernalillo County
exercised this option by submitting plans to EPA by December 31, 2003.
702
-------
lower Q/d threshold, and considered permitted emissions rather than actual emissions.
Commenters also suggested that all 38 facilities identified from the Q/d analysis should have
undergone a four-factor analysis.
Comment: Luminant provided a summary of EPA's "additional" visibility benefit analysis
of a "small group' of Texas sources. [Luminant (0061) p. 27]
Step 1: EPA singles out 38 facilities in Texas out of 1,600 by re-doing the Q/D analysis
that Texas used in its SIP to identify sources that EPA would evaluate for additional
controls.217
At Step 1. EPA identified a list of 38 Texas facilities, out of 1,600 Texas point sources, for
visibility modeling.226 EPA used an emission-over-distance ratio or "Q/D" to perform this
screening. However, EPA used both SO2 and NOx emissions for its screen, even though its
subsequent analysis was focused exclusively on SO2 emissions.227 EPA did not select, or
consider for selection, any sources from Oklahoma or any state other than Texas. EPA claimed
that it had to target only a limited number of facilities for its analysis because of "computation
resource limitations" of the model that it had selected (i.e., CAMx).228 However, EPA never
explains why it could not have easily created source regions in its analysis to encompass impacts
associated with sources in other states.
217 FIP TSD at A-2 to A-4.
224 79 Fed. Reg. at 74,880-81.
225 Id. at 74,882-84.
226 Id. at 74,877.
227 FIP TSD at A-4 to A-5.
228 Id. at A-4.
Luminant stated that the Q/d analysis used both SO2 and NOx emissions even though subsequent
analysis focused only on SO2 emissions. Luminant also comments that while EPA states that
computation resource limitations required limiting the number of facilities for analysis, EPA
does not explain why source regions were not created to encompass impacts associated with
other states.
Response: The TCEQ focused its control strategy analysis on point source emissions of SO2
and NOx, as the sources of these pollutants are the main anthropogenic pollutants that affect
visibility at Class I areas in Texas. In our review of the TX SIP, we agreed with Texas that it
was appropriate to focus the analysis on point sources of NOx and SO2. In order to identify
those point sources with the potential to have the greatest visibility impacts, we performed a Q/d
screening analysis considering combined emissions of NOx and SO2 emissions. This screening
led to the identification of 38 facilities for additional analysis. We then performed source
apportionment modeling for these 38 facilities to estimate the visibility impact due to the
emissions from each of these sources. The source apportionment modeling provided information
on the total visibility impairment due to all of the source's emissions, as well as the visibility
impairment specifically due to emissions of NOx, SO2, elemental carbon, or other species
emissions. Our complete analysis identified those sources with the greatest visibility impacts at
Wichita Mountains and the Texas Class I areas based on consideration of a source's emissions
703
-------
(modeled and recent emissions data), location, and modeled visibility impairment. Once
identified, we performed additional control analysis on these sources to determine through the
four-factor analysis if controls were available and cost-effective. Analysis of source
apportionment data for the sources identified for additional control analysis revealed that for all
these sources with the exception of one, their contribution to visibility impairment on the 20%
worst days was almost completely due to sulfate emissions. Therefore the control analysis for
these sources was focused on sulfate controls since controlling other emitted species, such as
NOx, would not result in significant visibility improvement on the 20% worst days for this
planning period. Source apportionment modeling for the PPG Flat Glass plant showed visibility
impacts from both NOx and SO2 emissions. As discussed in the proposal and FIP TSD, NOx
controls were considered along with recent emissions and permit data for the PPG Flat Glass
facility. Contrary to Luminant's comments that our analysis focused only on SO2 emissions, we
considered visibility impacts from both NOx and SO2 emissions, and focused additional control
analysis on those pollutants that were primarily responsible for the modeled visibility impact on
the 20% worst days from those sources.
As Luminant states, we did not create source regions to analyze impacts from other states. This
was not necessary because CENRAP's source apportionment modeling provided estimates of
visibility impacts from other states and source categories. The CENRAP states relied upon this
data in assessing visibility impacts from source categories and specific pollutants from source
within other states. We also relied upon this data in our analysis of the Texas and Oklahoma
regional haze SIP and reasonable progress at the relevant Class I areas. The commenter takes
our statement on "computational resources" out of context. As we discuss in the FIP TSD, due
to computation resource limitations, it is not possible to include a large number of facilities in the
photochemical modeling episode utilizing source apportionment. In other words, we could not
use source apportionment on all 1600+ point sources to evaluate their visibility impacts. The
Q/d analysis and use of the threshold value of ten identified those facilities with the greatest
potential to impact visibility due to their location and size. As discussed in a separate response
to comment below, this approach is a widely used method as an initial step to evaluate a
facility's potential to impact air quality and identify those sources with large enough emissions
close enough to a receptor to need additional analysis. We then performed photochemical
modeling with source apportionment as a more refined approach for these identified facilities to
estimate their visibility impacts considering emissions, stack parameters, chemistry, and
meteorology/transport. As explained in more detail elsewhere, our analysis focused on sources
within Texas. We reviewed Oklahoma's analysis of its sources and agreed with Oklahoma's
determination that no additional controls were necessary for this planning period.
Comment: [Luminant (0061), p. i] Luminant stated that the proposal burdens a handful of Texas
generating units, located far away from these Class I areas, with massive costs that would
threaten their continued operation and economic viability and the thousands of jobs they provide.
Luminant stated that, to arrive at this conclusion, EPA invents a methodology found nowhere in
the statute or regulations and applies it in a seemingly random manner. The map below depicts
EPA's skewed and unprecedented approach for Texas, with sources far away from these federal
704
-------
areas somehow being captured by EPA's novel methodology, while closer sources are excluded.
There is no discernible rationale for this flawed approach.
Map of Sources and Federal Areas Provided by Luminant (0061)
Wichita
Mourrtaini
B4ttrn/ '
9 66 ton
%
$
T " "%j M«l till Like
\ ftt Bttrni i 1
Bis Bind
i
• , •
Santlnw
Legend
Texas Sources for which EPA C pterin inn New
controlsane BeQuired
Tt*as Sources forwhlcfi EPA Determines New
Contmli are Mot ReqimH
0 kijJwn a N on-BART 5 0 wte i th a t E P& C> oe t
Not Evaluate
Luminant asserted that never before has EPA singled out individual sources using the new
approach that EPA employs here to derail the regional haze plans by Texas and Oklahoma. The
record is irrefutable that EPA has routinely approved other states' regional haze plans that look
just like the Texas plan and achieved the same level of progress.
Luminant stated that, with this unprecedented approach, EPA resorts to stretching the science
and the law beyond all recognition to justify its preordained result. EPA fails to follow its own
modeling protocols or to validate its modeling, and it uses results well outside the model's well-
recognized limitations.
705
-------
Response: Comments concerning costs and economic viability, consistency with actions in other
states, and modeling limitations are addressed elsewhere. Comments concerning our selection of
sources for additional analysis are addressed in the response to the comment below.
Comment: EPA's "additional analysis" starts with the wrong group of sources
[Luminant (0061) p. 109]
Luminant stated that, moreover, even if the type of individual source visibility analysis that EPA
performed were permissible under the statute and regulations, EPA started the analysis with the
wrong set of sources, and thus it reached the wrong result. EPA chooses which sources to include
in its analysis based on a quantity over distance or "Q/D" analysis to narrow approximately
1,600 Texas sources to 38 (Step 1, as discussed in our Background section). But this rough
metric does nothing to determine which sources may be reasonably controlled for the benefit of
these Class I areas. This Q/D approach ignores unit specific characteristics of the sources that
effect their potential for visibility impacts,683 ignores all sources outside of Texas which have
demonstrated impacts on these areas (including international and Oklahoma sources), and also
ignores one of the most important factors—meteorological conditions like wind patterns. EPA
claims that it had to use this Q/D threshold to narrow its target sources "[d]ue to computation
resource limitations," [footnote 684] but the result is that EPA's analysis too is of limited value
and, in fact, arbitrary. By selecting only a few sources, and then purporting to determine which
of them contribute significantly to visibility impairment by comparing them relative to each
other, EPA's analysis assumes what it seeks to prove. The truth is that all of the sources EPA
evaluated have minuscule, if any, impact on these areas, and trying to distinguish among them is
a futile effort—and comparing them is a useless exercise.
It is apparent, then, that EPA's pre-ordained analysis was flawed from the outset, and had EPA's
analysis started in a different and proper manner, it would have reached a different result. This is
yet another way that EPA's methodology was arbitrary and capricious and cannot support its
proposal.
Footnotes:
683 See FIP TSD at A-4.
684 Id
685 Source back trajectory data is from the NO A A Hy Split model, using ED AS 40km resolution meteorological data,
120 hour duration, ending at 3 p.m. at the WIMO IMPROVE monitor location 500m above ground. The analysis
measures the distance from each facility to the hourly back trajectory endpoints, provided in the data files for each
day from the HySplit output, using the Haversine formula. A trajectory is near a facility when the facility is within a
linearly increasing radius up to 40 km during the initial 6-hour period, or within 40 km for the remainder of the 5-
day period. The 20% Worst days are days based on the total extinction monitored at the WIMO IMPROVE visibility
monitor, available at:
ftp://vista.cira.colostate.edu/Public/AirQuality/Data/Aerosol/IMPROVE/DataSetsBySite/WIM01.csv.
Luminant presents a map of sources and Class I areas and comments that it demonstrates our
"skewed and unprecedented" approach, with sources far away from these federal areas
"somehow being captured by EPA's novel methodology, while closer sources are excluded."
Luminant also made general comments that the approach is random, "found nowhere in the
statute," unprecedented, and inconsistent with our actions on other states.
706
-------
Luminant commented that we started our reasonable progress control analysis with the wrong set
of sources. Luminant states that we used a Q/d analysis to identify which sources to include in
our analysis and that this approach is flawed because it ignores unit specific characteristics and
ignores one of the most important factors - metrological conditions such as wind patterns.
Luminant comments that this analysis is of limited value and arbitrary and attempts to determine
which of a few sources contribute significantly to visibility impairment by comparing them
relative to each other. Luminant claims that all of the sources EPA evaluated have minuscule
impacts on these areas, and that trying to distinguish among them or compare them is futile and
useless.
Luminant submits a back trajectory analysis for 2002 and 2011-2013 for Wichita Mountains that
it claims supports the comment that we started the analysis with the wrong set of sources.
Luminant states that the back trajectories show that transport rarely occurs from the Luminant
facilities to the Wichita Mountains Class I area and that much greater occurrences are seen
among other Texas sources, international sources, and Oklahoma sources.
Luminant states that EPA's choice of which sources to evaluate was critical because the
subsequent steps were constructed to all but assure that some impact would be shown. EPA's
"pre-ordained" analysis was flawed from the outset, and had the analysis started in a different
and "proper" manner, it would have reached a different result. Luminant concludes that this is
another way that EPA's methodology was arbitrary and capricious.
Response: We disagree with Luminant's assertion that our analysis, beginning with a Q/d
analysis was arbitrary, capricious, or improper. We also disagree with the assertion that our
analysis identified the "wrong" sources or that the analysis was "pre-ordained." The map
presented by the commenter does not account for the size of the emission source or other
important factors that contribute to the level of the visibility impairment due to the source. Just
taking into account the level of emissions from the sources presents a very different picture of
which sources may be causing visibility impairment at a class I area.549 As explained below and
elsewhere in this document, our complete analysis identified those sources with the greatest
visibility impacts at Wichita Mountains and the Texas Class I areas based on consideration of a
source's emissions, location, and modeled visibility impairment. Once identified, we performed
additional control analysis on these sources to determine through the four-factor analysis if
controls were available and cost-effective. Specific comments from Luminant concerning our
approach as being new, unprecedented and inconsistent with previous actions are addressed in
this and other response to comments in this document. Specific comments concerning sources in
Oklahoma are addressed in a separate response to comment in this document.
As we discuss at length in the FIP TSD, EPA, States (including Texas) and RPOs (including
CENRAP) have used a Q/d analysis to identify those facilities that have the potential to impact
visibility at a Class I area based on their emissions and distance to the Class I area. These
identified facilities could then be considered for further evaluation to estimate visibility impacts,
549 For example, see Figure A. l-3a of the FIP TSD for a map of Q/d values for the Texas sources at Wichita
Mountains.
707
-------
and then undergo the reasonable progress analysis for determination of reasonable controls. The
BART guidelines550 discuss identifying sources with the potential to impact visibility based on a
Q/d approach consistent with the method followed in this action. Furthermore, this approach has
also been recommended by the Federal Land Managers' Air Quality Related Values Work Group
(FLAG)551 as an initial screening test to determine if an analysis is required to evaluate the
potential impact of a new or modified source on air quality related values (AQRV) at a Class I
area. In the Texas regional haze SIP, TCEQ relied on a Q/d approach as one of the initial steps
to identify sources for additional analysis.552 We used a similar Q/d approach to identify 38
sources, from the more than 1600 point sources in Texas, that had the most potential to impact
visibility due to their location and size. In other words, we started by looking at every point
source in Texas553 and narrowed the field to a much smaller subset of sources with the most
potential to impact visibility based on their emissions and location. This approach is a widely
used method as an initial step to evaluate a facility's potential to impact air quality and identify
those sources with large enough emissions close enough to a receptor to need additional analysis.
Using this methodology, we considered every point source in Texas and narrowed the list to a
much smaller list of facilities with the greatest potential visibility impacts based on just
emissions and distance.
Following the Q/d analysis we took the additional step of using photochemical modeling,
utilizing CAMx with Plume-in-Grid (PiG) and Particulate Source Apportionment Tagging
(PSAT). As the commenter states, the Q/d analysis does not take into account stack parameters,
meteorological conditions, or chemistry. Given the large geographic distribution of sources and
distances to the Class I areas, we recognized that it was highly likely that only a subset of these
38 facilities would have the greatest visibility impacts on downwind Class I areas once
meteorological and transport conditions, atmospheric dispersion, chemistry and stack parameters
were taken into consideration, as CAMx with Pig and PSAT can do. EPA initially tagged
emissions from the 38 facilities identified though the Q/d analysis in order to estimate the
potential of emissions from a facility to impact visibility. This is a more refined approach than
the initial Q/d analysis performed by both Texas and EPA because unlike a Q/d analysis that
only considers emissions and distance, this accounts for emissions, location, stack parameters,
meteorological conditions, and models both chemistry and transport to the Class I areas. The
results of this modeling were used to verify our initial identification of sources and further
eliminate sources from a full four-factor reasonable progress analysis based on facility-level
impacts and consideration of estimated unit level impacts, as described in detail in the FIP TSD.
550 See 40 CFR part 51, app. Y, § III (How to Identify Sources "Subject to BART")
551 Federal Land Managers' Air Quality Related Values Work Group (FLAG), Phase I Report—Revised (2010)
Natural Resource Report NPS/NRPC/NRR—2010/232, October 2010. Available at
http://www.nature.nps.gov/air/Pubs/pdf/flag/FLAG_2010.pdf
552 TX RH SIP Appendix 10-1. "The group of sources was further reduced to eliminate sources that are so distant
from any of the ten Class I areas that any reduction in emissions would be unlikely to have a perceptible impact on
visibility. The list was restricted to those sources with a ratio of estimated projected 2018 base annual emissions
(tons) to distance (kilometers) greater than five to any Class I area."
553 The Texas point sources are defined as industrial, commercial, or institutional sites that meet the reporting
requirements of 30 Texas Administrative Code (TAC) §101.10. Permitted point sources in Texas are required to
submit annual emissions inventories. The data are drawn from TCEQ's computer-based State of Texas Air Retrieval
System (STARS). Annual emission data from 2009 were utilized to calculate the Q/D value for all point sources
with reported emissions in Texas. 2009 emissions data available in the docket as "2009statesum.xlsx"
708
-------
As discussed in the FIP TSD, our modeling results showed that the 38 identified sources from
the Q/d analysis were responsible for approximately 80% of the total impact from Texas point
sources on the 20% worst days at Wichita Mountains. Similarly, the 38 identified facilities were
responsible for approximately 50% of the impact from Texas point sources on the 20% worst
days at both Guadalupe Mountains and Big Bend. Contrary to Luminant's claim, the Q/d
analysis successfully narrowed the list of sources from over 1600 to those sources with the
greatest contributions to visibility impairment at Class I areas in Texas and Oklahoma. Our
additional analysis utilizing photochemical modeling and source apportionment showed that
some of these sources have large impacts when compared not only to other sources in Texas but
when compared to the CENRAP modeled impacts from all point source emissions combined in
other states. For example, the Monticello and Big Brown facilities are projected to contribute
approximately 1.3 Mm"1 and 1.2 Mm"1, respectively to visibility impairment on the 20% worst
days at Wichita Mountains in 2018 based on the CENRAP 2018 projected emissions for these
facilities. This is 1.7% and 1.5% of the total visibility impairment at Wichita Mountains. In our
FIP TSD we noted that Texas used an impact extinction level threshold of 0.5 Mm"1 (a level less
than half of the estimated impact from the Monticello or Big Brown facilities) from all sources in
a state as a threshold for inviting a state to consult.554 Oklahoma selected a threshold of 1.0 Mm"
1 to determine which states should consult in analyzing visibility impairment at Wichita
Mountains. We also noted that the largest projected contribution from all point sources within a
state at Wichita Mountains after Texas (14%) is Oklahoma at 3.9%. Elimination of all point
sources in Oklahoma would result in less visibility benefit (3.9%) than the visibility benefit from
required controls (greater than 5%).
The CENRAP RPO model results also support our selection of facilities in that these results
indicated that over 50% of the visibility impairment at Wichita Mountains due to point sources in
Texas is due to EGU sources (7.7% of the total visibility impairment), the majority due to EGU
sources located in East Texas modeled region (6.5% of the total visibility impairment). Our final
analysis identified those emission units with the largest visibility impacts, including the top three
impacting sources at Wichita Mountains (Big Brown, Monticello and Martin Lake), that are all
EGUs located in the East Texas modeled region. As these facts demonstrate, the identified
facilities have significant impacts on visibility conditions. Our technical record makes it equally
plain that the required controls reduce impacts from these sources and result in meaningful
visibility benefits towards the goal of natural visibility conditions.
In summary, our analysis properly identified the sources in Texas with the greatest individual
visibility impacts for additional control analysis. Luminant is incorrect in their assertion that the
visibility impacts from the identified sources are miniscule, or that we started with the wrong
sources. Starting from the entire universe of Texas point sources, we systematically eliminated
those facilities that had less potential to impact visibility based on careful consideration of
emissions, location, and finally modeled visibility impacts. After identifying those facilities with
the greatest visibility impacts, we performed the four factor analysis to evaluate whether
reasonable progress controls were available and cost-effective.
554 See Texas Regional Haze SIP Appendix 4-1: Summary of Consultation Calls and Section X.A. of the Oklahoma
Regional Haze SIP
709
-------
With regard to the back-trajectory analysis for Wichita Mountains submitted by the commenter,
we respond to this comment in a separate response to comment.
Comment: EPA Inappropriately Eliminated Sources from Detailed Four-Factor Analysis
of Reasonable Progress Controls. [Earthjustice (0067) p. 45]
Earthjustice et al. stated that after accepting Texas's narrowing of the scope of review to only
point source emissions of NOx and SO2, EPA further narrowed its review by first employing a
Q/D (or "Q over D") analysis, which compares emissions to distance, and secondly by using
source apportionment modeling. This process pared a list of more than 1600 sources down to just
9 facilities. Two aspects of EPA's method for screening reasonable progress sources led EPA to
inappropriately screen out sources that have significant visibility impacts.
First, considering the high number of sources contributing to cumulative impacts at multiple
Class I areas in the region, and its use of actual (rather than potential or allowable) emissions,
Earthjustice et al. stated that EPA should have used a lower Q/D threshold than 10 for
determining which reasonable progress sources to analyze in detail. EPA's evaluation of impacts
on just the worst 20% of days rather than peak impacts from each facility eliminates sources that
should have been reviewed. Likewise, EPA's emphasis on just 3 of the Class I areas impacted
obscures sources that have higher impacts at other Class I areas or have larger cumulative
impacts.
[Gray (0070) p.6] Dr. Gray stated that the first method EPA used to narrow the scope of its
review is a screening tool known as a Q/D (or "Q over D") analysis, which compares emissions
to distance. EPA used a threshold of 10 but should have used a lower threshold that would have
resulted in the inclusion of more sources in its control review. (FIP TSD, Section A.l)
According to Dr. Gray, by using a Q/D threshold of 10 and basing its calculations on actual
emissions, EPA removed from further analysis a number of sources, the control of which may
provide opportunities for reasonable, cost-effective visibility improvement. The end result of this
process was to ignore roughly 49%, 39%, and 22% of the contribution from Texas point sources
at BIBE, GUMO, and WIMO, respectively (roughly 3-4% of the total extinction in each case).6
At GUMO, these ignored sources contribute nearly 3 times more than the next highest7 state
point source contribution (from New Mexico). At BIBE, they contribute 50% more than next
highest state point source contribution (from Louisiana). And at WIMO, they contribute roughly
the same amount as the next highest state point source contribution (from Oklahoma).8 Similar to
EPA's guidelines around setting a threshold to determine sources subject to BART, any of the
thresholds used in EPA's reasonable progress analysis should take into consideration "the
number of emissions sources affecting the Class I areas at issue and the magnitude of the
individual sources' impacts. In general, a larger number of sources causing impacts in a Class I
area may warrant a lower contribution threshold."9 The same principle applies here, where there
are 1600+ Texas point sources collectively contributing to impairment. Indeed, Texas considered
a Q/D threshold of 5 in evaluating sources to consider. 79 Fed. Reg. 74,836. EPA correctly notes
the Federal Land Managers' Air Quality Related Values Work Group (FLAG) recommendation
of a Q/D threshold of 10. This is in the context of new or modified sources, and as such typically
uses the source's potential or allowable emissions rather than actual emissions. Because actual
710
-------
emissions are lower than potential emissions, sometimes by a significant margin, and because
actual emissions are not enforceable,10 a lower Q/D threshold should be used in this context (or
potential emissions should be used with the Q/D threshold of 10).
Dr. Gray stated that the EPA also selected the threshold of 10 to limit the number of facilities for
analysis because of resource constraints (the inability to perform source apportionment on large
numbers of facilities using photochemical modeling). While there are obviously logistical and
resource-related limits to the number of facilities that can realistically be evaluated, this approach
somewhat artificially limits the scope of analysis and excludes sources that do contribute to
visibility impairment. EPA could have performed CALPUFF analyses, which are far less
resource intensive, for at least some of the facilities. EPA also could have set up something like
its model plants for BART - i.e. assuming that any source meeting specific criteria was having a
sufficient impact to justify a four factor analysis for reasonable progress control.
Dr. Gray stated that the several sources that screened out of EPA's Q/D analysis were modeled
using CAMx as a part of Texas's BART analysis and appear to have significant impacts on
visibility (up to 0.42 dv maximum impact based on permit allowable emissions).11
Dr. Gray stated that these may be smaller sources relative to the impacts from some of the largest
Texas sources, but collectively they are significant contributors, and cost-effective, reasonable
controls may be available for many of them.
Footnotes:
6 See FIP TSD at A-64 through A-66 and underlying data in spreadsheet "TX116-007-_23_extinction_charts.xlsx".
7 Next highest after Texas.
8 Id.
9 40 C.F.R. Part 51, App'x Y III.A.
10 The challenges of relying on non-enforceable actual emissions are apparent in the many adjustments EPA was
required to perform due to updates to actual emissions, e.g. in its Q/D analysis and source apportionment work.
11 Texas SIP at 9-15.
Earthjustice et al. comments that EPA should have used a lower Q/d threshold considering the
number of sources contributing to visibility impairment and the use of actual rather than potential
or allowable emissions. The commenter also states that peak impacts and impacts at other Class
I areas should be considered. Using a Q/d value of ten based on actual emissions resulted in
EPA ignoring roughly 49%, 39%, and 22% of the contribution from Texas point sources at
BIBE, GUMO, and WIMO, respectively. The commenter also states that EPA limit the number
of facilities for analysis because of resource constraints but could have used CALPUFF to model
some of the sources and as a model plant to identify sources for additional analysis.
The commenter states that EPA screened some sources that had large impacts based on Texas'
CAMx modeling as part of its BART screening.
Response: Comments concerning impacts on the 20% worst days versus the peak impacts and
consideration of impacts at other Class I areas are discussed in separate response to comments.
We disagree with the commenter that a different threshold for the Q/d analysis was appropriate.
The use of a threshold value of ten based on actual emissions is consistent with previous EPA
actions in Arizona, Wyoming and Montana. We also note that Texas used a Q/d threshold of 5,
711
-------
but applied it separately to total NOx and SO2 emissions. The purpose of the Q/d analysis is to
identify those sources that have the greatest potential to impact visibility at the Class I areas. As
seen in table A. 1-2 of the FIP TSD, many of the sources have Q/d values well above 10. Q/d
values ranged from 10 to 425, and more than half of the facilities have Q/d values over 30.
Using a threshold value less than 10 would only serve to identify additional sources that based on
emissions and distance will have much less potential visibility impacts relative to those with the
largest Q/d. We agree with the commenter that a lower threshold may be appropriate in some
situations. It may be necessary to use a lower threshold in future planning periods as controls
are implemented to address those sources with the largest Q/d values, but as explained below,
that is not the case at this time.
With regard to the comment that the use of a Q/d threshold of ten ignores a large portion of the
visibility impairment attributable to Texas point sources, we note that of the 1600+ point sources
in Texas that contribute to visibility impairment, the 38 identified facilities for the Q/d analysis
contribute more than half of the total visibility impairment due to Texas point sources. In the
case of Wichita Mountains, the 38 facilities contribute almost 80% of the total impact from
Texas point sources. Clearly, this relatively small subset of Texas point sources are responsible
for a large portion of the total visibility impairment. We agree that while some sources may have
very small visibility impacts, aggregate impacts can be significant. However, while there are
undoubtedly thousands of sources within Texas that individually have small contributions to
regional haze; there are also many sources that, even in isolation, have relatively large visibility
impacts. In this first planning period, we identified the most significant sources that impact
visibility, determined whether cost-effective controls were available for these sources, and
balanced the costs of those controls against their visibility benefits. Given the iterative nature of
the regional haze program, we think that it was a reasonable approach to require only those cost-
effective controls with the largest benefits this planning period. We expect that Texas will
control additional sources, which by then will be the largest contributors to impairment, during
future planning periods.
Furthermore, we believe that it is appropriate to consider actual emissions levels as these are
representative of the emission levels that would be contributing to measured visibility
impairment. This is consistent with the BART guidelines recommendations for using actual
emissions rates to assess visibility impacts and the use of actual emissions from the baseline
period in calculating cost-effectiveness of controls as part of the BART five-factor analysis. We
also considered actual emissions in estimating the visibility benefit and emission reductions from
controls in our reasonable progress analysis. We considered 2009 emission levels for the Q/d
analysis and then compared the analysis to 2010 and 2011 Q/d values and determined there were
no significant differences in the sources above the threshold. Should actual emissions from any
source increase in the future, that should be considered in assessing the sources visibility impacts
and potential benefits from controls in future planning periods.
We discuss our selection of the CAMx photochemical model over the CALPUFF mode in a
separate response to comment. We note that the CAMx photochemical model allowed for a
consistent basis to assess all identified sources in Texas, including those sources beyond the
typical accepted range (<300km) of the CALPUFF model. This modeling approach also
provides for an analysis of the visibility impairment on the 20% worst days.
712
-------
We agree that a few sources had impacts ranging from 0.1 to 0.42 dv based on Texas' CAMx
modeling as part of its BART screening were screened out based on our Q/d analysis. Texas'
model results for BART screening using CAMx used a conservative approach based on a
combination of either annual emission rate multiplied by a factor of 2 to estimate maximum
short-term emissions and short-term permit allowable emissions from the 2006-2007 time frame
and focused on the maximum impact from the source. Our Q/d analysis examined actual annual
emissions from 2009, 2010, and 2011, and the photochemical modeling is based on projected
actual emissions and examining the impact on the average of the 20% worst days. We note that
some of these facilities are refineries subject to EPA refinery consent decrees that result in
emission reductions from 2006 allowable levels. As the sources currently with the largest
visibility impacts are finalized for controls in this action, some of the sources not being
controlled today by the FIP in this first planning period may subsequently rise to become the
sources with the largest visibility impacts for analysis in the next or future SIP planning periods.
Comment: Selection of sources for reasonable progress analysis [NPS (0077) p. 2-3, 4-5]
The NPS agreed with EPA that"... based on their visibility impacts, a smaller subset of the
facilities that we have initially analyzed should be further evaluated to determine ... if cost-
effective controls are available ..." However, EPA's approach to define sources in Texas for
further analysis is not consistent with the approach used in EPA's FIPs for Arizona, Wyoming,
and Montana.
The NPS stated that EPA used the ratio of facility emissions (Q) to distance of the facility from a
Class I area (d) to identify 38 facilities in Texas for further evaluation. These facilities emitted
617,000 tons of SO2, NOx, and PM10 in 20111, which represents 68% of the statewide total.
However, in its Texas FIP, EPA evaluated only nine of these 38 facilities under the reasonable
progress provisions with total SO2 emissions of 346,000 ton/yr representing 38% of the statewide
total SO2, NOx, and PM10 emissions and 67% of statewide total SO2 emissions. This is both a
smaller number of facilities evaluated and a smaller percentage of statewide emissions than EPA
evaluated in its FIPs for Arizona2 and Montana3. Had EPA relied upon its Q/d analysis in Texas
as it did in Arizona and Montana, it would have achieved approximately the same relative degree
of evaluation of the statewide emissions. Instead, EPA used visibility impact as a fifth factor to
eliminate 29 of the 38 facilities from consideration.
The NPS expressed concern about this statement in the FIP TSD:
When we examined the impacts, we noted that some source impacts are quite low
and some impacts were spread among several sources at the facility, making
individual unit impacts even smaller. We therefore concluded that some of these
impacts did not warrant further evaluation for this planning period and dropped
them from Table A.4-3.
The NPS asserted that the EPA's approach in Texas in dealing with facilities with multiple
emission units appears to be inconsistent with its approach in Arizona where it modeled several
emission units (at Apache, Cholla, Coronado, Nelson Lime) together to predict the resulting
713
-------
visibility improvement. We recommend that any visibility modeling evaluate the combined
visibility benefits from controlling the affected emission units at a given facility, as EPA did in
Arizona.
[NPS (0077) p. 4-5] The NPS concluded that the EPA should have relied upon its Q/D analysis
in Texas, as it did in Arizona and Montana, to select sources for the four-factor reasonable
progress analysis. This would have broadened the scope of these analyses to encompass a degree
of evaluation similar to that carried out by EPA in Arizona and Montana.
Footnotes:
1 National Emissions Inventory
2 In its Arizona FIP, EPA evaluated 15 facilities under the BART and reasonable progress provisions with total NOx
emissions of 38,000 ton/yr representing 78% of the statewide NOx emissions. (EPA had previously approved
Arizona's SO2 and PMIO SIP provisions.)
3 In its Montana FIP, EPA evaluated 13 facilities under the BART and reasonable progress provisions with total
SO2, NOx, and PM10 emissions of 44,000 ton/yr representing 71% of the statewide total emissions.
[NPS (0077) p. 4] The NPS expressed appreciation for the 230,000 tons of electric generating
unit (EGU) annual SO2 reductions EPA is proposing and anticipate significant benefits from
these reductions. The remaining 145,000 tons of annual SO2 emissions from its EGUs would still
rank Texas EGUs as the seventh-largest SO2 emitter among US EGU rankings - assuming no
other SO2 reductions occur. (And Texas would still rank #2 with its 135,000 tons of annual EGU
NOx emissions.) Review of the information presented in Figures 1-3 indicates that the
recommended controls on the nine "finalist" facilities address less than half of the visibility
impairment attributed to the 38 facilities selected by the initial screening process (in Table A.l-
2). Considering that controls were determined to be cost-effective at several of the 29 facilities
that were ultimately exempted from additional controls, we believe that implementation of all
controls determined to be cost-effective at these 38 facilities should be required.
[NPS (0077) p. 3-4 As NPS noted above, if visibility impacts are modeled, the combined
visibility benefits from controlling the affected emission units at a given facility should be
evaluated, as EPA did in Arizona. By breaking a facility down into ever smaller pieces, each
partial impact may be diminished to the point where it becomes relatively insignificant. As EPA
concluded in its FIP TSD:
The cost-effectiveness of the scrubber retrofits for the Welsh and Parish units are
within a $/ton range that we have previously found to be cost-effective in BART
determinations. However, we do not believe that their individual projected
visibility improvements merit the installation of scrubbers at this time. We
encourage the State of Texas to re-evaluate this determination as part of its next
regional haze SIP submittal.
According to NPS, the result of this "divide and exempt" approach is that very large emission
sources (like Welsh and Parish) with impacts almost three times EPA's 0.3% threshold of
concern can escape addition of controls that are determined to be "... within a $/ton range that we
have previously found to be cost-effective in BART determinations."
714
-------
[Earthjustice (0067) p. 45] Earthjustice et al. stated that, in addition to using a Q/D threshold,
the second method EPA used to narrow the scope of its review is source apportionment
modeling. EPA eventually settled on a threshold that screened out sources that contribute less
than 0.3% to the total light extinction at Class I areas. In other reasonable progress analyses that
EPA has performed or approved, a value of Q/D over a certain threshold alone qualified the
source for a four factor analysis. Instead of following that practice, EPA first eliminated sources
with a Q/D less than 10, which left EPA with 38 sources; EPA then added a second step, source
apportionment modeling, to screen sources for a detailed, four-factor reasonable progress
analysis, which left EPA with 9 sources.
Earthjustice et al. submitted a report by Dr. Andrew Gray as comment 0070. As explained by Dr.
Gray, the metric used by EPA (the worst 20% of days, or W20 days) masks visibility impacts
from sources that should have been considered in its control review. EPA should consider adding
additional sources to the set of sources reviewed for control. By eliminating 29 of the 38 sources
remaining after the first screening step, EPA missed an opportunity to identify additional
reasonable controls on sources with visibility impacts that are still large relative to many other
sources during the worst 20% of days at Big Bend, Guadalupe Mountains, and Wichita Falls.
Those sources may also contribute significantly to visibility impairment during other non-W20
days and at other Class I areas.
The NPS agreed with EPA that"... based on their visibility impacts, a smaller subset of the
facilities that we have initially analyzed should be further evaluated to determine ... if cost-
effective controls are available ..."
Earthjustice et al. stated that "It is reasonable to identify sources that will help achieve
reasonable progress this period and compel such reductions in the proposed FIP." They also state
that our modeling "confirmed that individual sources had significantly different impacts, and it is
proper to prioritize sources with significant visibility impacts for a four factor reasonable
progress analysis."
NPS and Earthjustice et al. commented that consistent with our approach in Arizona, Wyoming
and Montana, we should have relied on only the Q/d analysis to identify facilities for additional
control rather than eliminating some of the identified sources based on source apportionment
modeling.
NPS states that this would result in a larger portion of the total emissions from Texas being
analyzed for additional controls and be more consistent with our actions in Arizona and
Montana. Even after the 230,000 tons of SO2 reductions required by this rule, Texas would still
rank as the seventh highest SO2 emitter among US EGUs. NPS expressed concern that some
facilities were eliminated based on our review of the facility's estimated visibility impacts and
consideration of individual unit impacts. NPS suggests that if visibility impacts are modeled the
combined visibility impacts from all emissions units should be considered. NPS states that
consideration of unit-level impacts rather than facility-level impacts diminishes the impacts to
the point where they become relatively insignificant. NPS states that the recommended controls
on the nine "finalist" facilities address less than half of the visibility impairment attributed to the
38 facilities selected by the initial screening process and that considering that controls were
715
-------
determined to be cost-effective at several of the 29 facilities that were ultimately exempted from
additional controls, all controls determined to be cost-effective at these 38 facilities should be
required. Earthjustice et al. also comments that the impacts from the other 29 sources are still
large relative to many other sources that impact visibility on the 20% worst days.
Response: In developing FIPs for Montana, Arizona, and Wyoming, we relied on a Q/d analysis
to identify sources for additional control analysis. The Q/d analysis only considers emissions
and distance and does not consider how meteorology, chemistry or stack parameters influence
the potential to impact visibility. In Arizona, after eliminating sources based on more recent
emissions and those sources analyzed under BART, units at 9 facilities were identified for
additional analysis and reasonable controls were required at 2 units at 2 facilities. In Wyoming,
after eliminating sources analyzed under BART, 3 units at 2 facilities were identified for
additional control analysis and no reasonable controls were required on these facilities. In
Montana, ultimately 9 facilities were identified for additional control analysis and reasonable
controls were required at 2 units at 1 facility. A Q/d analysis in Texas resulted in identification
of 38 facilities, located at a considerable range of distances from Class I areas. We note that this
is a much larger number of sources identified than in the other states mentioned by the
commenter.
As discussed in a separate response to comment and in the FIP TSD, we determined it was
appropriate to use photochemical modeling to assess the visibility impact from those sources
identified by our Q/d analysis. In the same way that Q/d is used as an estimate of the potential
visibility impact due to emissions and distance, the photochemical modeling aims to estimate the
visibility impact albeit in a much more refined manner that accounts for chemistry and
meteorological conditions. We also note that some RPOs and states used a combination of back
trajectory analysis, source apportionment modeling results, and Q/d as a more refined approach
to identify sources for additional control analysis for RP.555 Our modeling results indicated that
a subset of the 38 facilities were the primary contributors to visibility impairment at each Class I
area. There are a number of different approaches used by states in identifying sources for
reasonable progress evaluation but it usually centered around the general premise of evaluating
the biggest sources and/or the biggest impactors on visibility. Figures A.3-2, A.3-3 and A.3-4 of
the FIP TSD show the percent contribution to visibility impairment as a percentage of the total
impairment due to Texas sources and the total impairment due to Texas point sources. Some of
the facilities identified via Q/d were shown to have relatively small visibility impacts and
controls on these facilities would be much less effective in improving visibility. For example,
the top 10 facility level modeled visibility impacts at Wichita Mountains accounts for almost
75% of the total visibility impact from all 38 facilities combined. Therefore, we determined it
was reasonable to eliminate some of these small impacting sources from the list of sources
needing additional control analysis for reasonable progress for this first planning period. As
discussed elsewhere in the FIP TSD and response to comments, we eliminated those facilities
555 To select the specific point sources that would be considered for each Class I area, VISTAS first identified the
geographic area that was most likely to influence visibility in each Class I area and then identified the major SO2
point sources in that geographic area. The distance-weighted point source SO2 emissions (Q/d) were combined with
the gridded extinction-weighted back-trajectory residence times. The distance weighted (Q/d) gridded point source
S02 emissions are multiplied by the total extinction-weighted back-trajectory residence times (Q/d * Bext-weighted
RT) on a grid cell by grid cell basis and then normalized VISTAS Area of Influence Analyses, 2007 available in
the docket for this action.
716
-------
that had relatively low facility-wide impact levels. For EGUs, because of the large differences
noted for some of the facilities between the projected 2018 emissions modeled and actual
emissions, we scaled the modeled visibility impacts to consider visibility impacts at higher or
lower recent emission levels. After evaluation of facility-level and estimated unit-level impacts,
we established a 0.3% visibility impact threshold on a unit basis to identify those sources with
the greatest visibility impacts for additional control analysis.
Ultimately, we identified 21 units at 9 facilities based on a threshold of percent contribution of
visibility impairment of 0.3% that underwent the four-factor analysis, including consideration of
additional visibility modeling results, for additional controls. We used the 0.3% threshold only
as a way to identify a reasonable set of sources with the largest visibility impacts to evaluate
further. Based on our evaluation of the four-factors, including consideration of visibility
benefit, we determined that additional controls should be required at 14 units across 7 facilities.
In addition to these 9 facilities, as discussed elsewhere, we also examined controls for the PPG
Glassworks and San Miguel, and eliminated Sommers-Deely-Spruce from additional analysis
due to a planned shutdown of the JT Deely units. Those sources eliminated from additional
control analysis for this planning period should be reevaluated in the next planning period.
We disagree with the NPS that all cost-effective controls at the 38 facilities should be required
during this planning period. As discussed more fully elsewhere, we believe states (or EPA when
promulgating a FIP) can consider the visibility impacts of sources when determining what
sources to analyze under the four-factor framework. We also believe that States (or EPA when
promulgating a FIP) are permitted, but not required, to consider visibility improvement in
addition to the four statutory factors when making their reasonable progress determinations.
In selecting what sources or source categories undergo the four-factor analysis, we focused on
the sources that have the greatest visibility impacts, as determined by consideration of both Q/d
and our more refined analysis relying on photochemical modeling. We identified units at nine
facilities for reasonable progress control analysis. We note that NPS states that they agree with
us on this matter in their comments. The NPS comments state "The NPS agreed with EPA that
'... based on their visibility impacts, a smaller subset of the facilities that we have initially
analyzed should be further evaluated to determine ... if cost-effective controls are available
We also note that Earthjustice et al. also supports this approach. Earthjustice et al. stated that "It
is reasonable to identify sources that will help achieve reasonable progress this period and
compel such reductions in the proposed FIP." They also state that our modeling "confirmed that
individual sources had significantly different impacts, and it is proper to prioritize sources with
significant visibility impacts for a four factor reasonable progress analysis."
We note that in every action that required controls for reasonable progress, visibility was
considered at some point in the reasonable progress analysis. As discussed in more detail in a
separate response to comment, while collectively the visibility impacts from the 29 sources not
analyzed for controls may be significant, the individual point source visibility impacts are much
smaller relative to the subset of sources we identified that impact visibility the most and the
relative visibility benefit from controlling these 29 other sources will be much smaller than the
facilities we did identify to control. The USDA Forest Service supports this overall approach
and comments that "In summary, while the USDA Forest Service has expressed concern to EPA
that the use of visibility as a factor to be considered within the reasonable progress context may
717
-------
be outside the statutory framework established for RP (see Clean Air Act, Section 169A (g)(1)),
the methodology and metrics that EPA used are the most comprehensive seen to date for any
SIP/FIP in the country that we have reviewed, and should serve as a model for future efforts to
consider the contribution and/or potential benefits of individual sources to visibility." As the
sources currently with the largest visibility impacts are controlled, the "other" sources will
become the sources with the largest visibility impacts for analysis in the next planning period.
Our methodology identified those sources with the largest contributions to visibility impairment
and addresses a significant portion of the total SO2 emissions in the state of Texas. As the
commenter states, we addressed 67% of Texas' total SO2 emissions by evaluating the 9
identified facilities for controls and we believe this a very significant portion of the state
emissions. This does not include the significant emissions addressed in considering potential
controls at San Miguel and the planned shutdown of the JT Deely units. Furthermore, given the
large geographic distribution of sources across Texas and the large distances to Class I areas,
there is a large portion of the state's total emissions that based on distance,
meteorology/transport, and stack parameters are not anticipated to contribute significantly to
visibility impairment at the Class I areas examined for this planning period. For example,
facilities with significant emissions of SO2 located in the Houston area will contribute
significantly to the total state emissions level, but many of these facilities are too far away and
spread around multiple emission sources and lower stacks to significantly impact visibility at
Wichita Mountains or the Texas Class I areas. We disagree with the commenter that only
considering the percentage of emissions addressed or the number of facilities identified is an
appropriate comparison because not all emissions or facilities have the same visibility impact,
especially in a state with as many sources and large geographic extent as Texas. What is of the
most concern is that the sources with significant visibility impacts are identified and analyzed for
controls, and we believe our approach accomplishes just that. We note that our analysis resulted
in the identification of a much greater number of reasonable controls than the analyses performed
in Arizona, Montana, or Wyoming. The required controls result in a reduction in 230,000 tpy of
SO2, a much larger reduction than the total emissions in Arizona or Montana. We also note that
Texas has a significant number of coal-fired EGUs and therefore, total EGU emissions are large.
We disagree with the NPS that our approach is inconsistent with the modeling approach in
Arizona for BART. In Arizona, emission units were modeled together to estimate facility level
visibility improvement from controls. We used CAMx modeling with source apportionment in
our initial modeling to assess facility-level visibility impacts. We also estimated unit-level
impacts from these facility impacts based on unit level emissions. As we state in the FIP TSD
"For our analysis we tried utilizing an extinction percentage of 1% for a facility's impacts with a
consideration that some facilities have two or three units and this metric would equate to 0.5% or
0.33%) extinction per unit."556 We concluded that this was a reasonable way to identify sources
to analyze for additional controls and established a 0.3% visibility impact threshold to identify
sources for additional evaluation. Our initial consideration of both facility-wide impacts and
then ultimately, unit level impacts allowed us to identify a reasonable set of sources with the
greatest visibility impacts to evaluate further for unit level controls. We also note that we used
source apportionment modeling to evaluate controls at these units and provided the reduction in
556 FIP TSD at A-49
718
-------
extinction levels anticipated from controls on each unit. Just as we have estimated the visibility
benefit of all controls together (all scrubber retrofits and all scrubber upgrades),557 we present all
the necessary data to assess the visibility improvement from controlling all units at a facility. As
we state in the FIP TSD, because our four-factor RP analysis evaluating potential controls would
be completed on a unit specific basis, we evaluated visibility benefit on a unit basis as well. We
disagree that our consideration of unit-level impacts rather than facility-level impacts diminished
the impacts to the point where they became relatively insignificant. We considered both facility-
level and unit-level impacts in identifying those sources with the greatest visibility impacts. By
conducting our reasonable progress analysis on a unit-level, we have identified those units with
the largest visibility impacts and evaluated cost-effective controls on those units that will achieve
reasonable visibility benefits required during this planning period towards the national goal.
While we agree that visibility impairment on a facility-level for Welsh and Parish is greater than
that from Coleto Creek, controlling one unit at Coleto Creek would yield almost twice the
visibility benefit of controlling a single unit at Welsh or Parish and controls on Coleto Creek are
also more cost-effective on a $/ton basis. As the required controls are implemented, the
percentage impact from those facilities and units not controlled will become larger (on a
percentage basis) and should be analyzed in future planning periods.
We address specific comments concerning consideration of controls on Parish and Welsh
elsewhere in this section of the RTC document, as well as in the Cost-benefit and Cost
Comments sections. We also address comments concerning visibility impairment on days other
than the 20% worst days in a separate response to comment.
0.3% Threshold
0.3% Threshold General Comments: We received comments on the selection and use of a
0.3% of total impairment threshold for identifying sources for additional analysis. Commenters
stated that the 0.3% threshold is without precedent, is arbitrary and not supported statistically or
otherwise. Furthermore, the commenters state that the threshold is inconsistent with thresholds
approved by EPA in other states for identifying sources for evaluation for controls.
We also received comments that a lower threshold should have been used, that the threshold was
applied inconsistently, and that the 0.3% threshold screened out sources that have a significant
visibility impact and should have been evaluated for controls.
Comment: EPA's Source Apportionment Modeling Is Reasonable for Determining
Priority Sources That Require Controls in the First Planning Period. [Earthjustice (0067)
p.35]
Earthjustice et al. stated that the EPA's modeling results showed that controlling a small number
of sources will result in significant visibility benefits at both Texas Class I areas. EPA's source
apportionment modeling confirmed that individual sources had significantly different impacts,
and it is proper to prioritize sources with significant visibility impacts for a four factor
reasonable progress analysis. However, as we detail below, additional sources should also make
557 See tables 43 and 44 of the proposed FIP for an example and Vis modeling summary.xls in the docket for all data
and calculations
719
-------
this first cut in order to make reasonable progress by reducing significant visibility impacts that
Texas sources have on Class I areas in the surrounding states.
Earthjustice et al. stated that the EPA's modeling results showed that controlling a small number
of sources will result in significant visibility benefits at both Texas Class I areas. EPA's source
apportionment modeling confirmed that individual sources had significantly different impacts,
and it is proper to prioritize sources with significant visibility impacts for a four factor
reasonable progress analysis. Additional sources should also make this first cut in order to make
reasonable progress by reducing significant visibility impacts that Texas sources have on Class I
areas in the surrounding states
Response: We agree with Earthjustice et al. that source apportionment modeling confirms that
individual sources have significantly different visibility impacts, significant visibility benefits
can be achieved by controlling a small number of sources and it is proper to consider and
prioritize sources with significant visibility impacts for a four factor reasonable progress
analysis. We address comments concerning additional controls elsewhere in this document.
Comment: [Luminant (0061) p. 112] Luminant noted, even if the regulations permitted EPA to
deviate from using the deciview metric, EPA cites no precedent for using a 0.3% extinction
threshold or any rational basis for it. EPA concludes that "any unit with an estimated impact
greater than 0.3% [extinction] would be further evaluated" for further reductions, but units with a
smaller impact would not.700 But, this new threshold that EPA would require of Texas is well
below the 1% threshold used by other states (and approved by EPA) for excluding sources from
additional reductions under the reasonable progress analysis.701 Had EPA used the same 1%
threshold that it has previously approved for other states, EPA's analysis would correctly
confirm that it is not reasonable to require further reductions from any of the Texas units that
EPA examined, given that none of the units at issue were modeled to have a contribution of 1%
or more to visibility impairment at any Class I area.702
Footnotes:
700 FIP TSD at A-50.
701 See, e.g., 77 Fed. Reg. 3,691, 3,704 (Jan. 25, 2012) (proposing approval of Virginia's "one percent" "threshold to
determine which emissions units would be evaluated for reasonable progress control"). In its proposal here, EPA
acknowledges its prior approval of a 1% impact threshold "used by states in development of sources for RP
evaluation..." FIP TSD at A-49. But it attempts to avoid applying that threshold here by asserting that "utilizing
an extinction percentage of 1% for a facility's impacts . .. would equate to 0.5% or 0.33% extinction per unit." Id. at
A-49. But in these other instances, EPA has approved a 1% per-unit, not per-facility, threshold, and thus EPA's
attempt at distinction fails. See, e.g., 76 Fed. Reg. 78,194, 78,206 (Dec. 16, 2011) ("[T]he Commonwealth [of
Kentucky] established a threshold to determine which emissions units would be evaluated for reasonable progress
control. . .. The Commonwealth then identified those emissions units with a contribution of one percent or more to
the visibility impairment at that particular Class I area, and evaluated each of these units for control measures for
reasonable progress, using the . .. four 'reasonable progress factors' .... KYDAQ identified 10 emissions units at
five facilities in Kentucky . . . with SO2 emissions that were above the Commonwealth's minimum threshold for
reasonable progress .. . ." (emphasis added)).
702 FIP TSD at A-52.
Luminant commented that there is no precedent or rational basis for EPA's use of a 0.3%
extinction threshold for determining which units would be further evaluated for controls.
720
-------
Luminant states that this threshold is less than and inconsistent with thresholds used by other
states and approved by EPA. Luminant cites to EPA's approval of Virginia's and Kentucky's
regional haze SIPs as examples. Luminant further states that had EPA used a 1% extinction
contribution for emission units in Texas, all units would have fallen below the thresholds and
eliminated from analysis for further control.
Response: We address comments concerning the precedent and rational basis for our threshold
for determining which units would be further evaluated for reasonable progress controls in a
separate response to comment below.
The commenter is incorrect in their comparison of EPA's application of a 0.3% threshold in our
analysis of Texas sources and the use of a 1% threshold used by Kentucky, Virginia and other
VISTAS states in identification of sources for reasonable progress analysis for controls. We
utilized emissions data and source apportionment modeling to estimate the visibility impairment
from 38 facilities (identified by a Q/d analysis) and applied a threshold of 0.3% of the total
modeled visibility extinction at a Class I area to the estimated impact from an emission unit. In
other words, the estimated extinction (in inverse megameters) due to emissions from the
emission unit at a Class I area was divided by the total extinction modeled at the Class I area due
to emissions from all emission sources. The 1% threshold used by Kentucky, Virginia, and other
VISTAS state was applied to a different metric developed by VISTAS that relied on a
normalized assessment of residence time and Q/d (emissions divided by distance) 558, and did not
rely on modeled visibility impairment. We note that the methodology utilized by the VISTAS
states identified those sources with the greatest potential to impact visibility, based on
consideration of emissions, location, and typical wind patterns.
The commenter's assertion that we should have used a threshold such that "all units would have
fallen below the threshold and eliminated from further controls" is inconsistent with the
requirements and purpose of the regional haze rule. The Regional Haze Rule requires the
identification of reasonable progress controls based on consideration of the four factors and the
development of coordinated emission control strategies in order to make reasonable progress
towards the goal of natural visibility conditions. Despite the fact that we and Texas agree that
impacts from point sources are significant, the commenter suggests that it would be appropriate
to select a threshold that results in no evaluation of any point sources. While based on the four
factor analysis, it may be appropriate to determine that controls are not required on some or even
all evaluated sources, failing to demonstrate how the four factors were considered for the sources
of group of sources with the most significant visibility impacts is clearly at odds with the rule.
We note that Texas did identify a group of sources for additional control analysis based on Q/d,
and that analysis, while flawed, included some of the same sources we identified in our analysis.
We discuss Texas' four-factor analysis in depth in our proposal, TX TSD and in response to
comments elsewhere in this document.
558 To select the specific point sources that would be considered for each Class I area, VISTAS first identified the
geographic area that was most likely to influence visibility in each Class I area and then identified the major SO2
point sources in that geographic area. The distance-weighted point source SO2 emissions (Q/d) were combined with
the gridded extinction-weighted back-trajectory residence times. The distance weighted (Q/d) gridded point source
S02 emissions are multiplied by the total extinction-weighted back-trajectory residence times (Q/d * Bext-weighted
RT) on a grid cell by grid cell basis and then normalized VISTAS Area of Influence Analyses, 2007 available in
the docket for this action.
721
-------
Comment: Source Apportionment Results - Selection of Sources for Further Evaluation
[Gray (0070) p. 11]
Dr. Gray stated that the EPA used the source apportionment modeling to narrow the scope of its
review from 38 sources to 9 sources. Because the threshold EPA used was too high, and the
agency failed to consider cumulative impacts and impacts outside of WIMO, GUMO, and BIBE,
EPA eliminated sources it should have reviewed in its control analysis. Controls on an additional
set of sources would provide meaningful visibility benefits.(FIP TSD, Sections A.3 and A.4)
Dr. Gray commented that as a preliminary matter, EPA's source apportionment step was not
inherently necessary: in other reasonable progress analyses that EPA has performed or approved,
a value of Q/D over a specific threshold alone qualified the source for a four factor analysis. 17
EPA should have taken that approach here and evaluated controls on all 38 sources.
Dr. Gray stated that the EPA's source selection process focuses on controlling only the largest
sources that contribute to visibility impairment at only three of the modeled Class I areas. By
eliminating 29 of the 38 initially identified sources, EPA missed an opportunity to include more
sources in its subsequent analysis, and thus to identify additional reasonable controls. These 29
sources alone contribute more than all of the point sources from any other state; this is a
significant contribution, and these sources should have at least been evaluated for reasonable
controls.
As with the Q/D analysis above, Dr. Gray concluded that the EPA's threshold should have been
lower to take into account the multitude of sources collectively impacting these Class I areas.
Footnotes:
17 See EPA's rulemakings regarding the regional haze plans for Montana, Wyoming, and North Dakota. 77 Fed.
Reg. 23988, (Apr. 20, 2012); 24058-9; 78 Fed. Reg. 34738, (June 10, 2013); 34763; 76 Fed. Reg. 58570, 58624-5
(Sept. 21, 2011).
Response: We respond to comments concerning evaluation of controls on all 38 sources in a
separate response to comment above. We address comments concerning the visibility impacts
and consideration of controls for the other 29 sources that were not selected for additional
control analysis based on our 0.3% threshold in a separate response to comment above.
Comment: Proposed reasonable progress and long-term strategy determination for San
Miguel [San Miguel (0060) p. 2-3]
San Miguel stated that they do not believe that it should be included as one of the 15 electric
generating units ("EGUs") with a proposed SO2 limit.1 EPA, as part of its additional visibility
analysis in its FIP TSD document, narrowed the list of sources for additional analysis. San
Miguel was identified as having a modeled impact of 0.207% on the most-impacted area (based
on estimated unit average extinction percentage impacts on the worst 20% of days for class I
722
-------
areas), as represented on page A-51 of the FIP TSD document. This is below EPA's own 0.3%
threshold that EPA used to identify sources for further evaluation. Rather, EPA appears to
include San Miguel for further analysis based on the "estimated facility impact adjusted to reflect
2008-2012 average emissions," that demonstrate a 0.333% impact. However, as the model
reflects, emissions going forward for San Miguel should be far below those levels and below
EPA's 0.3%) threshold. Given that EPA's modeled projected impact for San Miguel is below
EPA's level for additional visibility analysis, and the continued reduction in emissions from the
San Miguel plant with its upgraded scrubber, EPA should completely remove San Miguel from
the list of units with source-specific limits.
Footnotes:
1 As discussed in Section II of this letter, San Miguel believes that the Proposed FIP should be withdrawn and that
EPA should approve of Texas' SIP submission. Nothing in this comment letter should be construed as an
endorsement of this Proposed FIP, though San Miguel does include the following substantive comments if EPA
intends to proceed despite this opposition.
Response: We established a 0.3% contribution to total visibility impairment threshold to
identify sources for reasonable progress control analysis. As the commenter states, the modeled
visibility contribution at Big Bend was estimated to be 0.207%. However, the CENRAP 2018
estimated emissions for the one unit at this facility developed for the 2007 CENRAP modeling
were much lower (6,600 tpy SO2) than recent average actual emissions (10,601 tpy SO2 average
annual emissions 2009-2013). The estimated visibility impact at the higher, actual emission
level was estimated to be 0.333%. Because this estimated visibility impact exceeded the 0.3%
threshold, we evaluated additional controls for this unit. We found that San Miguel has already
upgraded its scrubber and we proposed that it maintain an emission rate consistent with current
controls and recent monitoring data from Dec 2013 - June 2014. As the commenter states,
"emissions going forward for San Miguel should be far below those levels and below EPA's
0.3%) threshold." The FIP's required emission limit will ensure that the existing scrubber is
operated to maintain those lower emission levels and limit the visibility impact from the source.
Similarly, had the facility not completed these recent upgrades, we would have evaluated the
sources for upgrades and likely would have established the same emission limit after examining
the four regulatory factors.
Comment: Luminant provided a summary of EPA's "additional" visibility benefit analysis
of a "small group' of Texas sources. [Luminant (0061) p. 27]
Step 3: EPA concludes, based on this new modeling, that it is "worth investigating" a
"smaller set of sources."218
218 Id. at A-28; see 79 Fed. Reg. at 74,878
At Step 3. EPA reviewed ENVIRON's modeling of entire facilities (many of which included
multiple point sources of emissions) to "narrow[] the list to [a] smaller group of sources" for "a
second round of CAMx modeling."247 EPA concluded that while "there are slight differences in
the projected values" between ENVIRON's modeling for CENRAP and its modeling for EPA,
"the conclusions are consistent with the original CENRAP work."248 In evaluating these
results—again, results that were "consistent" with the CENRAP modeling results already
723
-------
considered by Texas and Oklahoma—EPA simply "concluded it was worth investigating
whether the installation of cost effective controls on a small group of sources, out of the universe
of sources in Texas, would result in a significant reduction in Texas' contribution to the visibility
impairment at Class I areas."249 EPA, however, did not examine "the universe of sources in
Texas," but only these few, nor did it explain the metric it was using to determine which sources
were "worth investigating."
Footnotes:
247 79 Fed. Reg. at 74,878.
248 FIP TSD at A-28.
249 Id.
Response: We disagree with Luminant that we did not examine the universe of sources or that
we did not explain the metrics and process used to determine which sources we examined for
additional controls. In the FIP TSD, we state that in evaluating the impacts from individual
sources it can be seen that even a smaller set of sources make up the majority of the total
impairment impacts from the 38 facilities at these three Class I Areas. Luminant does not
provide the proper context in their comment in describing how we evaluated the available
information and identified sources for additional analysis. As discussed in a separate response to
comment, we stated that our modeling was consistent with the CENRAP modeling in discussing
the overall model performance and visibility projections for the Class I areas. However, our
modeling evaluated the visibility impacts from 38 individual sources in Texas, whereas the
CENRAP modeling only tagged emission source categories within a state or region (e.g. all area
sources in Oklahoma). Our modeling provided additional information to investigate the
visibility impacts from individual or a small group of sources. Our evaluation began with
examining all Texas point sources in the 2009 TCEQ point source emission inventory, a list that
included over 1,600 facilities. We calculated the Q/d value for all of these facilities and
eliminated those sources with a Q/d value less than 10 for all nearby Class I areas. Modeling
was then performed for the 38 remaining sources to estimate the visibility impact from these
individual facilities. As we stated in the FIP TSD, "In evaluating the impacts from individual
sources it can be seen that even a smaller set of sources make up the majority of the total
impairment impacts from the 38 facilities at these three Class I Areas. Therefore, we concluded
it was worth investigating whether the installation of cost effective controls on a small group of
sources, out of the universe of sources in Texas, would result in a significant reduction in Texas'
contribution to the visibility impairment at Class I areas." As described in more detail in a
separate response to comment, we established a 0.3% of total visibility impairment threshold to
identify a reasonable set of units with the largest contributions to visibility impairment at the
examined Class I areas for additional control analysis. We then considered the estimated
visibility benefit of controls alongside the four statutory factors to identify cost-effective controls
that will achieve reasonable visibility benefits required during this planning period towards the
national goal.
Comment: Step 4: EPA "decide[s] to examine the [ENVIRON] results in several different ways
to help in identifying a subset of sources for further visibility modeling and control analysis."219
EPA evaluates all 38 facilities, but apparently draws no conclusions to distinguish among the
facilities that ENVIRON modeled.
219 FIP TSD at A-41
724
-------
At Step 4. EPA "decided to examine the [ENVIRON] results in several different ways to help in
identifying a subset of sources for further visibility modeling and control analysis."250 The first
way EPA "examine[d] the results" was to "rank[] the top 10 impacting facilities [not individual
units] for each of the three Class I areas."251 EPA did not explain why a "top 10" list was
relevant. In doing so, EPA compiled its "top 10" list—not in terms of deciviews as reported to
EPA by ENVIRON—but as a percentage of total extinction in terms of inverse megameters
(l/Mm).252EPA says it made this decision "to somewhat normalize the total extinction
differences between the differing Class I areas."253 But EPA never explains why it needed to
compare impacts between the three Class I areas, nor did EPA provide any other rationale for
using percentage of total extinction as its deciding metric, instead of deciviews as required in the
regulations.
From this exercise, EPA concluded that "a number of facilities made the top 10 for more than
one Class I area."254 EPA also reviewed recent SO2 and NOx emissions data (2008-2012) for 19
of the 38 facilities and compared it to the emission rates ENVIRON used in its modeling for
EPA. EPA noted that some facilities had higher recent emissions than were modeled by
ENVIRON, and others had lower emissions.255 However, despite looking at recent emissions
data from these selected units,, EPA did not consider in its analysis recent visibility monitoring,
which, as discussed below, shows that these Class I areas already meet EPA's 2018 interim
goal. Moreover, as discussed further below, in its analysis, EPA also decided not to consider or
account for SO2 and NOx limitations imposed on Texas sources under the CSAPR, which
became effective on January 1, 20 1 5.256 Instead, EPA concluded that "[o]verall this information
supports looking at recent actual emissions to represent future emission levels in 2018."257 EPA
did not eliminate any of the 38 facilities at this step based on this assessment.
250 Id. at A-41.
251 Id.
252 To the best of our knowledge, EPA has never used this metric in acting on other regional haze SIPs.
253 Id.
254 Id.
255 Id. at A-49.
256 Id. at A-45.
257 Id.
Response: As Luminant states, we considered the modeling results in a number of different
ways in order to identify the subset of sources with large visibility impacts at the nearby Class I
areas. Our initial Q/d analysis identified 38 facilities with the greatest potential to impact
visibility based on emissions and location. Modeling was then performed for the 38 sources to
estimate the visibility impact from these individual facilities. As we stated in the FIP TSD, "In
evaluating the impacts from individual sources it can be seen that even a smaller set of sources
make up the majority of the total impairment impacts from the 38 facilities at these three Class I
Areas. Therefore, we concluded it was worth investigating whether the installation of cost
effective controls on a small group of sources, out of the universe of sources in Texas, would
result in a significant reduction in Texas' contribution to the visibility impairment at Class I
areas." As a first step, we identified the top ten impacting facilities at each Class I area. We
focused this analysis on Wichita Mountains, Big Bend and Guadalupe Mountains. This provided
information on the size of the visibility impacts from each facility relative to each other at each
725
-------
Class I area. This showed that a handful of the top impacting facilities were responsible for a
significant portion of the visibility impairment from the modeled sources at each Class I area.
We also noted that a number of facilities were included in the top impacting facilities for more
than one Class I area. All of this information provided additional support to the conclusion that a
subset of the sources were responsible for the greatest visibility impairment at the Class I areas
and that controlling a small number of facilities in Texas would result in significant visibility
improvements at the impacted Class I areas. We presented the data in terms of extinction and
percent of total visibility impairment, as calculated by ENVIRON.559 We note that identification
of the top ten facilities is not impacted by the choice of using deciview, or extinction, or percent
impact at each Class I area.
In the FIP TSD, we also discuss why we utilized extinction and percent extinction metrics and
how the fact that Guadalupe Mountains has cleaner background visibility conditions than
Wichita Mountains should be considered.560 For example, a source that has a smaller absolute
impact [in terms of extinction] on a relatively cleaner area but a higher percentage impact might
be considered for control so that the cleaner area can potentially make progress. We explain that
using the percentage of total visibility impairment metric allows us to somewhat normalize the
extinction differences between Class I areas so that we can utilize the same approach at each
Class I area and identify a reasonable set of sources to analyze that if controlled would result in
meaningful visibility benefits towards meeting the goal of natural visibility at every Class I area.
For every Class I area to have the opportunity to reach the natural visibility goals, it is necessary
to identify the sources or source categories that significantly impact visibility, identify available
controls and analyze whether those controls are reasonable. Had we established a strict threshold
based on extinction or deciview, we would have had to establish a different threshold for each
Class I area. As stated by the USDA Forest Service in its supportive comments, the use of this
methodology and metrics, including the use of a small percentage threshold on the 20% worst
days is linked to the concept of reasonable progress. We believe it could serve as the model for
future efforts to consider the contribution and potential benefits of individual sources to
visibility. After identifying which sources to analyze for additional controls based on the
percentage impact on a unit basis, we determined which controls were reasonable based on
consideration of the four factors, including comparison of cost to the anticipated visibility benefit
(deciview improvement, extinction, percentage of total extinction, and the percentage of the total
impact from Texas point sources addressed by the control).
As discussed in detail in a separate response to comment below where we address comments
concerning the legality of the visibility metrics we used, we disagree with Luminant's assertion
that use of the deciview metric is required by the regulations. In this separate response to
comment we also further discuss the utility of the percentage of visibility impairment and
extinction metrics and the need to consider a number of different metrics to fully consider the
potential visibility benefits of controls to address visibility impairment on the 20% worst days.
We also discuss that states and RPOs, and particularly Texas, routinely relied on light extinction
and percent of total visibility impairment metrics when assessing the various contributions to
559 Electronic file in the docket "EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_072913.xlsx" see spreadsheet
under tab "ToplOSRC" prepared by ENVIRON and FIP TSD Tables A.4-la, b and c
560 FIP TSD at A-41
726
-------
visibility impairment due to either source categories or pollutant species from other states and
international sources.561
We address comments concerning recent monitoring data in a separate response to comment.
As the commenter states, we considered recent emissions data and compared it to the CENRAP
2018 projected emissions that were modeled. We noted that the data indicates that a number of
facilities have actual emissions that are much higher than modeled.562 For instance, Big Brown,
Sandow, and Martin Lake were all significantly higher than modeled rates, with Martin Lake
having over 90% more SO2 emissions than modeled based on the 2018 CENRAP projections.
As explained in detail in the FIP TSD563 we determined it was necessary to consider recent actual
emissions from EGUs due to uncertainty in 2018 projected emissions completed in 2006, the
cost of SO2 credits being lower than originally projected, and comments from Texas on a more
recent IPM projection indicating that significant SO2 reductions were not anticipated at these
sources and no large SO2 control projects were planned at most of the sources being evaluated.564
We also noted that TCEQ has utilized recent emission data for EGUs when developing projected
emissions for 2018 (and other future years) when developing ozone attainment demonstrations.
Therefore, based on Texas' recent comments and other information, we concluded considerable
uncertainty exists as to whether any further reductions of SO2 will occur beyond current emission
levels as a result of compliance with MATS or CSAPR. Overall this information supports
looking at recent actual emissions to represent future emission levels in 2018.
Comment: Step 5: Instead of asking ENVIRON to update its modeling based on recent data,
EPA in-house staff "scale[s]" the visibility impacts modeled by ENVIRON to "adjust[]" the
results based on 2008-2012 annual SO2 emissions at each of the 38 facilities.220
Step 6: EPA "drop[s]" 20 of the 38 facilities from further consideration because "some
source impacts are quite low and some impacts were spread among several sources at the
facility, making individual unit impacts even smaller."221 EPA never explains what it
considers to be "quite low" or "even smaller."
Step 7: EPA in-house staff again "scales" the ENVIRON modeling results to a unit level
for the remaining 18 facilities and establishes a 0.3% light extinction "threshold" and
concludes that "any unit with an estimated impact greater than 0.3% would be further
evaluated."222 EPA never explains the basis or origin of its 0.3% extinction test.
220 Id. at A-49 to A-50.
221 Id. at A-50 to A-52.
222 Id. at A-50.
561 See Chapter 11 and Appendix 4-1: Summary of Consultation Calls of the Texas regional haze SIP for examples
of the use of extinction (inverse megameters) and percentage of total extinction metrics
562 FIP TSD at A-47, Table A.4-2
563 See FIP TSD at A-45
564 Texas comments on Draft IPM modeling conducted by EPA for potential national rule making platform provided
on June 26, 2014. In this docket materials as "TCEQ comment letter to EPA on draft modeling platform dated June
24, 2014. '2018 EMP signed.pdf.
727
-------
At Step 5. EPA "scaled" the visibility impacts modeled by ENVIRON to "adjust[]" the results
based on 2008-2012 annual SO2 emissions at each of the 38 facilities.258 EPA does not scale NOx
emissions, even though ENVIRON's modeling showed that many of the 38 facilities contribute
to visibility based primarily on their NOx emissions.259 Nor does EPA provide any support for or
detail on its "scaling" methodology, nor is this approach justified or appropriate.260 The "scaling"
methodology is a direct contradiction to the appropriate application of the PSAT modeling
results, as specified in the CAMx User's Guide. The guidance is clear that the non-linear
chemical reactions in the photochemical model cannot be used for estimating improvement
associated with different control scenarios.261 EPA then compares average extinction at each
facility (both modeled and adjusted SO2 impacts) for each of the three Class I areas262 and
highlights those facilities that had impacts (either modeled or adjusted, in terms of percentage
extinction) of 1% or more on a facility basis or "0.5% or 0.33% extinction per unit."263 EPA says
that "[t]his reduced the sources we had to examine to about a dozen facilities," although it does
not specifically identify those facilities.264 EPA does not eliminate from further review any of the
38 facilities at this step based on this assessment.
At Step 6. without reference to its 1% facility threshold, EPA arbitrarily "drop[s]" 20 of the 38
facilities from further consideration because "some source impacts are quite low and some
impacts were spread among several sources at the facility, making individual unit impacts even
smaller."265 EPA never explains what it means by "quite low" and "even smaller," but
nevertheless concludes that "some of these impacts did not warrant further evaluation for this
planning period."266 This step leaves 18 facilities for further review.
At Step 7. EPA again "scales" ENVIRON's facility modeling to unit level for the remaining 18
facilities (which results in a list of 40 units).267 EPA then establishes a 0.3% light extinction
"threshold" for further evaluation.268 EPA concludes that "anj unit with an estimated impact
greater than 0.3% [at any one of the three Class I areas] would be further evaluated."269 EPA
fails to cite any precedent or guidance establishing a 0.3% threshold, nor does it provide any
rationale for such a threshold.
Footnotes:
258 Id. at A-50.
259 Id. at A-47 to A-48; EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_072913.xlsx ("Avglmpacts" and
"emissions data").
260 Section 7 of the attached report from AECOM explains how EPA's linear scaling method further exaggerated the
visibility benefits that EPA asserts would result from the controls in its proposed FIP.
261 CAMx User's Guide v6.1, Section 1.2.
262 FIP TSD at A-51.
263 Id. at A-49.
264 Id.
265 Id. at A-50 to A-52.
266 Id. at A-50.
267 Id.; TX116-007-_30_Source_selection_analysis_TX_RH-es_l-31-14 ("WIMO & TX Facility (short)" and
"WIMO & TX (Short) EI Ranked").
268 FIP TSD at A-50.
269 Id. (emphasis added).
728
-------
Response: As stated by the commenter, we adjusted facility level impacts for the 38 facilities by
actual emissions because we determined recent actual emissions were more representative of
future emission levels in 2018. We discuss this in more detail in the response to comment above
and elsewhere in this document. As we state in the FIP TSD: "We used a linear scalar approach
to scale the impacts based on the modeled emission rate and associated impact level to result in
an estimated impact based on actual emissions. Since as we have noted earlier, the extinction
from most sources was almost all due to SO2 emissions, we scaled the impacts based on SO2
emissions."565 The commenter is incorrect in their assertion that ENVIRON's modeling showed
that many of the 38 facilities contribute to visibility based primarily on their NOx emissions. In
support of this statement, the commenter cites to table A.4-2 in the FIP TSD, but this table
merely shows the modeled emissions versus actual emission levels for the 38 sources, and not
the visibility impacts due to NOx or SO2 emissions from these sources. The modeling results
also cited to by the commenter, and displayed in the FIP TSD in figures A-2.2, A-2.8, and A-
2.14 clearly show that visibility impairment from these sources is predominantly due to sulfate.
For example, at Wichita Mountains, the only facilities with any significant level of impact from
nitrates are Oklaunion and PPG Glassworks and all sources (with the exception of PPG
glassworks) have sulfate impacts that are several times greater than nitrate impacts. For those
facilities with the largest modeled visibility impacts, the contribution to visibility impairment
from nitrate is very small compared to the sulfate contribution.
We discuss the use of scaling visibility impacts by emissions in more detail in a separate
response to comment below. We determined that the linear relationship we developed to
extrapolate extinction due to emission rates was a reasonable approach in our technical analysis.
At this point in our analysis, we scaled the visibility impact up (or down) based on the ratio of
actual SO2 emissions to modeled emissions, as an estimate of the visibility impacts anticipated
based on actual emissions from the facility. We considered the modeling results and our scaled
modeling results in a number of different ways in order to identify a reasonable subset of sources
with the largest visibility impacts at the nearby Class I areas. As we stated in the FIP TSD, "In
evaluating the impacts from individual sources it can be seen that even a smaller set of sources
make up the majority of the total impairment impacts from the 38 facilities at these three Class I
Areas. Therefore, we concluded it was worth investigating whether the installation of cost
effective controls on a small group of sources, out of the universe of sources in Texas, would
result in a significant reduction in Texas' contribution to the visibility impairment at Class I
areas."
The commenter's description in Step 5 does not accurately describe the methodology we used to
identify sources for control. The section of the FIP TSD the commenter cites to here is a
summary of our approach, describing the different ways we considered the modeling and scaled
results.566 In this summary, we explain that we looked at break points of 1%, 0.5% and 0.3%.
We compared these potential thresholds to thresholds used by states to determine which upwind
states should consult based on their impact from all sources within the state. We estimated that,
based on the thresholds used by states for consultation, an impact from an individual source that
may be large enough to be considered for potential control may conservatively be in the range of
0.2% to 1.33%). We then explain that we tried utilizing a 1%> facility impact threshold with a
565 FIP TSD at A-49.
see FIp TSD at A_49
729
-------
consideration that some facilities have two or three units and this metric would equate to 0.5% or
0.33% extinction per unit. We conclude this summary with a general statement that "this was a
reasonable way to arrive at some common breakpoints/drop-offs in potential visibility
improvements. This reduced the sources we had to examine to about a dozen facilities and
helped set a context for what level of impacts may warrant further modeling to assist in a full
four factor analysis for our RP/LTS analysis."567
For EGUs, because of the large differences noted for some of the facilities between the projected
2018 emissions modeled and actual emissions, we scaled the modeled visibility impacts to
consider visibility impacts at higher or lower recent emission levels. We then ranked the
facilities based on their maximum modeled impact at a Class I area as shown in Table A-4.3 of
the FIP TSD. When we examined the facility-level impacts, we noted "that some source impacts
are quite low and some impacts were spread among several sources at the facility, making
individual unit impacts even smaller. We therefore concluded that some of these impacts did not
warrant further evaluation for this planning period and dropped."568 As discussed in the
summary above, we considered thresholds of 1%, 0.5% and 0.3%. We also considered a unit
level threshold of 0.5 or 0.33%. We eliminated those facilities that had relatively low facility-
wide impact levels (well less than 0.3%) when compared to the impact threshold levels above.569
We also eliminated facilities that when considering the large number of units at the facility,
would have low unit-level impacts. For example, Waha Plant had a facility-level impact of
0.242% which was spread across 5 emission units.570
For the remaining sources, we used unit level emissions to divide the facility level impacts into
estimated unit-level visibility impacts. As we explain in the FIP TSD and elsewhere in this
response to comments document, we examined unit-level visibility impacts because our four
factor RP analysis evaluating potential controls would be completed on a unit specific basis. As
we describe below in more detail, we established a 0.3% of total visibility impairment threshold
to identify sources for additional control analysis.
Examination of the unit-level scaled impacts (which were typically higher than the modeled
impacts) reveals a number of units with impacts of approximately 0.3%.571 This can be clearly
seen in the figure below showing the unit level impacts of the units listed in Table A.4-4 of the
FIP TSD. We concluded this was a reasonable common breakpoints/drop-off in visibility
impacts and potential visibility improvements. We established a 0.3% threshold only as a way to
identify a reasonable set of units with the largest contributions to visibility impairment at the
examined Class I areas for additional control analysis. As we discuss in more detail in a separate
response to comment below, we continued to evaluate whether to include or exclude sources that
were close to the cutpoint, or for which we had additional information that would indicate they
should be excluded in the second round of visibility modeling.572
567 FIP TSD at A-49
568 FIP TSD at A-50
569 The impacts of the facilities eliminated at this step ranged from 0.005% to 0.196%.
570 See TX116-007-_30_Source_selection_analysis_TX_RH-es_l-31-14 for modeled and scaled visibility impacts
for facilities and units.
571 See Table A.4-4. Nine units have impacts ranging from 0.274% to 0.333%
572 FIP TSD at A-53.
730
-------
1.100%
1.000%
0.900%
0.800%
0.700%
0.600%
0.500%
0.400%
0.300%
0.200%
0.100%
0.000%
We disagree with the commenter that there is no rational basis for our threshold for determining
which units would be further evaluated for reasonable progress controls. As an initial step, most
states considered source type emission levels (e.g. point, mobile, area, etc,) and photochemical
modeling source apportionment to identify the pollutant(s) and source type(s) that contribute the
most to visibility impairment and eliminate other pollutants and source groups from additional
evaluation based on consideration of extinction and/or percent of total extinction. For example,
Texas determined, and we agreed, that based on the CENRAP model results the predominant
anthropogenic emissions impacting visibility are nitrate and sulfate emissions, primarily from
point sources. Therefore Texas focused its control strategy analysis on point source emissions of
SO2 and NOx.573
There are a number of different approaches used by states in identifying sources for reasonable
progress evaluation but it usually centered around the general premise of evaluating the biggest
sources and/or the biggest impactors on visibility. To identify point sources for additional
control analysis many states relied on a Q/d analysis. Some states relied on a simple analysis of
emissions to determine which sources should be analyzed. As discussed elsewhere, the VISTAS
573 Texas Regional Haze SIP at 10-2 and Table 3 of Appendix 10-1 showing percent contribution from different
source types to S04 and N03 at the five Class I areas
Estimated Maximum Unit Impact Adjusted to reflect 2008-2012
Avg. Emissions
• •
•
9
•
L_«
• •
•
•
•
1 • •
•
•
•
••
• * *.
»
w w 1
5 10 15 20 25 30 35 40
731
-------
RPO relied on a metric derived from Q/d and residence-time.574 As discussed in depth
elsewhere, in selecting which sources or source categories would undergo the four-factor
analysis, we focused on the sources that have the greatest visibility impacts as determined by
consideration of both Q/d and our more refined analysis relying on photochemical modeling.
Our modeling results indicated that a subset of the 38 facilities identified by the Q/d analysis
were the primary contributors to visibility impairment at each Class I area. We then used the
modeling results to narrow the group of sources further because it was reasonable to conduct a
full four-factor analysis only for this subset of sources with the largest visibility impacts, based
on facility-level and consideration of estimated unit level impacts. We used the 0.3% threshold
(as described above) only as a way to identify a reasonable subset of sources with the largest
visibility impacts to evaluate further for reasonable controls. As discussed in the FIP TSD, we
also considered including sources that fell under the threshold based on additional
information.575 At this point, the resulting set of sources served as a starting place from which to
further analyze individual source impacts and potential benefits from controls in subsequent
modeling, and identify reasonable controls using the four-factors. We discuss the relative level
of visibility impacts and visibility benefits from controlling those sources eliminated from
additional analysis in a separate response to comment.
We note that there is no explicit threshold for determining significance of visibility benefit in the
regional haze rule. As we state in the preamble to the final Regional Haze Rule and Guidelines
for BART Determinations: "Depending on the facts regarding the number of sources affecting a
class I area and their modeled impacts, the State could set a threshold that captures those sources
responsible for most of the total visibility impacts, while still excluding other sources with very
small impacts."576 Significance is a source- and Class I area-specific evaluation, meaning that it
depends on how much visibility improvement is needed at the Class I area(s), how much a
specific source impacts the Class I area(s), and the cost effectiveness and potential visibility
improvement of available control options. States have latitude to determine these thresholds577,
providing support and a reasonable and adequate basis for why they selected the thresholds, and
to determine BART and reasonable progress controls, in consultation with other impacted states.
Comment: [Xcel Energy (0064) p. 6] Xcel Energy stated that the EPA's method of categorizing
sources based on percent of light extinction is non-statistical and unsupportable. Using an
574 To select the specific point sources that would be considered for each Class I area, VISTAS first identified the
geographic area that was most likely to influence visibility in each Class I area and then identified the major S02
point sources in that geographic area. The distance-weighted point source S02 emissions (Q/d) were combined with
the gridded extinction-weighted back-trajectory residence times. The distance weighted (Q/d) gridded point source
SO2 emissions are multiplied by the total extinction-weighted back-trajectory residence times (Q/d * Bext-weighted
RT) on a grid cell by grid cell basis and then normalized VISTAS Area of Influence Analyses, 2007 available in
the docket for this action.
575 FIp TSD at A_53
576 79 FR 39121
577 BART guidelines at 70 FR 39170: "However, we believe the States have flexibility in setting absolute thresholds,
target levels of improvement, or de minimis levels since the deciview improvement must be weighed among the five
factors, and States are free to determine the weight and significance to be assigned to each factor. For example, a
0.3, 0.5, or even 1.0 deciview improvement may merit stronger weighting in one case versus another, so one "bright
line" may not be appropriate." (We note that the deciview values here are based on CALPUFF modeling results
calculated against a natural "clean" background).
732
-------
acceptable statistical categorization method, Tolk should be excluded from the group of
controlled sources. EPA also imposes on Tolk the highest cost-per-ton controls of all sources that
EPA proposes to control for the smallest expected visibility benefit of any source (as measured
in deciviews, EPA's preferred visibility metric).
Response: We address Xcel's comments regarding the method we used to identify sources for
additional reasonable control analysis is the response to comment below. We address Xcel's
comments regarding the cost-effectiveness and visibility benefits of the required controls in
elsewhere in the response to comments document.
Comment: Xcel Energy stated that the EPA's 0.3% contribution threshold for the four-
factor reasonable progress evaluation is arbitrary.
According to Xcel Energy, one of the most important errors that EPA made in the Proposal is its
selection of a "natural break" in the visibility impact data that EPA used to require controls on
some sources and not on others. FIP TSD, at A-49. EPA's analysis and its selection of a "natural
break" threshold are seriously flawed. As discussed above, EPA departed from its own guidance
and approach to regional haze rules across the country. Instead of relying on deciview impacts to
identify sources of visibility impairment and to estimate visibility benefits from controls, EPA
resorted to the use of light extinction. This is, on its face, problematic. However, EPA then
utilized its analysis of light extinction caused by various sources as a way to decide which
sources should be controlled and which could wait until the next planning period.
Specifically, Xcel Energy noted that the EPA plotted the light extinction data for 19 facilities (38
emission units) obtained from CAMx modeling to evaluate visibility impacts.9 EPA calculated
the percent of extinction for the average impacts on the worst 20% (W20%) days at the most
impacted Class I area for each emission unit using the PM source apportionment tool ("PSAT").
According to the FIP TSD, EPA then reviewed these adjusted values and identified "natural
break points that indicated a significant drop-off in impacts" that would allow them to "select a
natural subset of the largest impacting sources" to be included in their reasonable progress
analysis. FIP TSD, at A-49. EPA identified natural break points around 1%, 0.5%, and 0.3%. FIP
TSD, at A-49. EPA provided no mathematical justification in the FIP TSD to support its
assertion that these values represent the natural breaks for the sample population. The selection
of the 0.3%) light extinction threshold seems to have been made by "eyeballing" the data instead
of performing a statistical analysis.
The following scatter plot in Xcel Energy Figure 4 shows the percent extinction for the average
impacts on the W20%> days at the most impacted area for the 38 emission units.
Scatter Plot of % Extinction on W20% Days at the Most Impacted Area (Figure 4 provided
by Xcel Energy)
733
-------
12
Based on the above plot, there is no obvious break in the data at 0.3% light extinction.
To identify natural breaks in the data using more scientific and mathematical methods, Xcel
Energy had the data imported into the ESRI ArcGIS software, excluding zeroes. In ArcGIS,
there are multiple classification options that can be selected. The current sample population data,
"percent of extinction for the average impacts on the worst 20% (W20%) days at the most
impacted Class I area for each emission unit" is a temporally and spatially non-linear data set.
For this type of data, Jenks' optimization is the most appropriate statistical methodology to
calculate classification. 10
Xcel Energy noted that Jenks' optimization is a method of statistical data classification that
partitions data into classes using an algorithm that calculates groupings of data values based on
the data distribution. Jenks' optimization seeks to reduce variance within groups and maximize
variance between groups. Jenks' optimization uses an iterative method to calculate multiple
breaks in the datasets to determine the natural break with the smallest variance. Class breaks are
identified that best group similar values and that maximize the differences between classes. For
the purpose of this analysis, two classification options were selected to define the natural breaks
to determine which sources should be included in the four-factor reasonable progress analysis
and which sources should be excluded.
Natural Breaks for% Extinction on W20% Days at the Most Impacted Area Based on
Jenks' Optimization (Figure 5 provided by Xcel Energy)
734
-------
3.0-
2.5-
2.0-
1.5-
1.0-
0 5-
0.0
0.000000
8
1
0.257500
1
0.515000
0.772500
1.030000
As shown in Exhibit (Xcel Energy Figure 5) above, Jenks' optimization, a statistically sound
method that is appropriate to the dataset, would set the natural break for including sources in the
group subject to the reasonable progress four-factor analysis at 0.41 %. EPA's utilization of a
threshold of 0.3% extinction to determine which units should be included in a reasonable
progress four-factor analysis is unjustified and its selection appears to be subjective. It results in
an arbitrary threshold specific to this data set that cannot be applied or replicated across other
reasonable progress analyses.
Xcel Energy concluded that the EPA should set the threshold at 0.41% for purposes of
determining which sources should be evaluated for controls under the four-factor reasonable
progress analysis. This would put Tolk in the group of sources excluded from the consideration
of controls.
Footnotes:
9 The percent extinction values obtained from the CAMx modeling were adjusted to reflect 2008-2012 average
emissions. FIP TSD, at A-52 (Table A.4-4).
10 The other classification schemes provided in ArcGIS are not appropriate for the data. The Equal Interval
methodology calculates the interval based on the equal number of values in each class irrespective of the variance in
the data. The calculated value of the equal interval is 0.51 %. The Quantile method classification applies to linear
data sets and is not applicable here because the current dataset is a composite of random values derived as a percent
extinction at the most impacted Class I area and, therefore, is a spatially and temporally non-linear dataset. The
Standard Deviation classification determines numbers of classes to keep the variance between the mean values of
the classes to less than one standard deviation. The variance between the current set of values is so large that the
Standard Deviation method cannot be applied without classifying the data into a minimum of four classes of
sources. Since the purpose of this analysis is to classify the data into two classes, this methodology is not
appropriate. Nonetheless, even if one of these classifications were used for the determination of a reasonable
progress analysis threshold, none of the methodologies calculate a natural break at 0.3%.
Xcel commented that the 0.3% threshold applied by EPA to identify sources for additional
control analysis is non-statistical and unsupported. Xcel states that EPA identified natural
735
-------
break points around 1%, 0.5%, and 0.3%. FIP TSD, at A-49 but provided no mathematical
justification in the FIP TSD to support its assertion that these values represent the natural breaks
for the sample population. Xcel states that based on a statistical analysis using Jenks
optimization, of the data a threshold of 0.41% should have been adopted and the estimated
impact for Tolk Station would fall below this threshold. Therefore, Xcel concludes that Tolk
Station should not have been included in the sources identified for additional analysis by EPA.
Response: We address comments concerning our use of metrics other than deciview impact in a
separate response to comment. We note that in some places the commenter incorrectly
characterizes the 0.3% threshold as being determinative of which sources are controlled. The
commenter states that we erred in our "selection of a "natural break" in the visibility impact data
that EPA used to require controls on some sources and not on others." However, this threshold
was only used to identify those sources to evaluate further for additional controls. The
commenter's description does not accurately describe the methodology we used to identify
sources for control. The section of the FIP TSD the commenter cites to here is a summary of our
approach, describing the different ways we considered the modeling and scaled results.578 In this
summary, we explain that we looked at break points of 1%, 0.5% and 0.3%. We compared these
potential thresholds to thresholds used by states to determine which upwind states should consult
based on their impact from all sources within the state. We estimated that, based on the
thresholds used by states for consultation, an impact from an individual source that may be large
enough to be considered for potential control may conservatively be in the range of 0.2% to
1.33%). We then explain that we tried utilizing a 1% facility impact threshold with a
consideration that some facilities have two or three units and this metric would equate to 0.5% or
0.33%) extinction per unit. We eliminated some facilities based on consideration of lower
facility-level impacts and/or the number of units at the facility. For the remaining sources, we
estimated unit -level impacts. As we explain in the response to comment above, examination of
the unit-level scaled impacts revealed a number of units with impacts of approximately 0.3%.579
We concluded this was a reasonable common breakpoint/drop-off in visibility impacts and
potential visibility improvements. We used the 0.3% threshold only as a way to identify a
reasonable set of sources with the largest visibility impacts to evaluate further for reasonable
controls. As discussed in the FIP TSD and above, we also considered including sources that fell
under the threshold based on additional information.580 Ultimately, we identified 21 units at 9
facilities with the largest visibility impacts to evaluate for additional control analysis including
visibility modeling including both units at Tolk Station.581
The scatter plot presented by the commenter does not provide the data ranked by impact and
therefore obscures the trends in the data. As can be seen in the plot (using the same data set)
provided in our response to the comment above, there are a number of units with impacts of
approximately 0.3%, our selected threshold. As discussed elsewhere, we also considered
inclusion of additional units near this threshold based on additional information.
578 FIP TSD at A-49
579 See Table A.4-4. Nine units have impacts ranging from 0.274% to 0.333%
58° FIp TSD at A_53
58i pip tsd a( A-56, Table A.5-1 shows the 21 units
736
-------
We disagree with the commenter that our methodology in identifying a threshold was flawed and
we disagree with the methodology developed by the commenter and their suggested threshold of
0.41% is appropriate. The Jenks optimization method seeks to divide the data into a specified
number of groups by identifying a group of sources most like each other and most different for
the other groups. By selecting only one break point, the method applied by Xcel seeks to divide
the data set into only two groups rather than looking for multiple groupings within the data. We
disagree with the commenter's assertion that the purpose of this analysis is to classify the data
into two groups, those sources to be evaluated and those not included in our evaluation. The
purpose of this analysis is to identify a reasonably broad set of sources with the largest visibility
impacts that will then be analyzed through the four factor analysis, including consideration of
modeled visibility benefits of controls. This set of sources would include those sources with the
very largest impacts and potentially a subset of sources with smaller impacts. Examination of
the data clearly reveals that some facilities have relatively very large (-1%) impacts and should
possibly be considered a separate grouping. For example, using the same data set and
methodology but seeking to divide the data into three groups reveals break points at 0.2% and
0.6%. Applying the same methodology to the adjusted facility level impacts yields break points
at approximately 1% (0.974%) and 0.333%).
Furthermore, despite the fact that Texas and we agree that based on CENRAP 2018 source
apportionment data NOx and SO2 impacts from point sources are the most significant
anthropogenic sources impacting visibility at the Class I areas, the commenter supports use of a
threshold of 0.41% that would fail to capture even the facility with the very largest modeled
visibility impact at Guadalupe Mountains (Tolk Station). While based on the four factor
analysis, it may be appropriate to determine that controls are not required on some or even all
evaluated sources, failing to demonstrate how the four factors were considered for the sources of
group of sources with the most significant visibility impacts is clearly at odds with the rule. As
we state in the preamble to the final Regional Haze Rule and Guidelines for BART
Determinations: "Depending on the facts regarding the number of sources affecting a class I area
and their modeled impacts, the State could set a threshold that captures those sources responsible
for most of the total visibility impacts, while still excluding other sources with very small
impacts."582
Comment: EPA claims that it used a "percent impacts approach" (as opposed to deciview
impacts approach required by the regulations) "because of its linkage to the reasonable
progress concept."270 But EPA fails to acknowledge that its actual reasonable progress
regulations use deciviews as the required unit of measurement, not percent extinction. EPA
claims that "a source that has a smaller absolute impact on a relatively cleaner area but a higher
percentage impact might be considered for control so that the cleaner area can potentially make
progress,"271 but EPA fails to explain that the relevant progress is measured in 2018 against
actual visibility conditions, not artificially pristine conditions.
270 Id. Compare 40 C.F.R. § 51.308(d)(1) ("For each mandatory Class I Federal area located within the State, the
State must establish goals (expressed in deciviews) that provide for reasonable progress towards achieving natural
visibility conditions." (emphasis added)).
582 79 FR 39121
737
-------
271 FIP TSD at A-50.
Response: We address this comment in a separate response to comment where we discuss the
legality and utility of the visibility metrics we used. As discussed in detail in that separate
response to comment, we disagree with Luminant's assertion that use of the deciview metric is
required by the regulations. In this separate response to comment we also discuss the utility of
the percentage of visibility impairment and extinction metrics.
Comment: Step 8: EPA examines 40 units at the remaining 18 facilities and concludes
that 20 units have a modeled or estimated impact greater than 0.3% at one or more of the
three Class I areas.
At Step 8. based on its new 0.3% threshold, EPA then examines the 40 units at the remaining 18
facilities and concludes that 20 units have a modeled or estimated impact greater than 0.3%
extinction at one or more of the three Class I areas.272 EPA, inexplicably, does not consider, or
provide anywhere in the record, the deciview impact from each unit at this step, even though it
later attempts to isolate the visibility benefit, by unit, of each control it would impose. EPA's use
of percentage extinction masks the extremely miniscule impact of these facilities on actual
visibility. For reference, the 0.291%) total extinction that EPA / ENVIRON's modeling attributes
to the entire W. A. Parish Station (all four units) at Wichita Mountains (which EPA notes in
Table A.4-3 of its FIP Technical Support Document ("TSD")) equates to a deciview impact of
0.02538 dv (average of 20%> worst days); the 0.44%> total extinction attributed to the entire Big
Brown Plant at Big Bend equates to a deciview impact of 0.036157 dv; and the 0.444%> total
extinction attributed to the entire Sommers Deely Plant at Guadalupe Mountains equates to a
deciview impact of 0.037297 dv.273 In other words, in terms of deciviews (the unit of
measurement required by the regulations), the impact from these facilities is beyond miniscule
and from each individual unit would be even smaller, though EPA does not report those unit-
level values. Luminant provided a table (see Table 3 to comment 0061 that shows EPA's
estimated total visibility impact (in deciviews, average 20%> worst days) for Luminant-operated
plants (all of which include multiple units)274, at each of the three Class I areas at issue.275
TABLE 3: ENVIRON / EPA MODELED TOTAL VISIBILITY IMPACTS OF LUMINANT PLANTS AT THREE
CLASSIAREAS(DECIVIEWS,2f!%WORSIDAYS)276
Unit
WIMO
Modeled Total Visibility
Impact (deciviews,
average 20% worst days)
(21.34722 dvtotal)
BIBE
Modeled Total Visibility
Impact (deciviews,
average 20% worst days)
(17.68373 dvtotal)
GUMO
Modeled Total Visibility
Impact (deciviews,
average 20% worst days)
(17.18758dvtotal)
Big Brown
(Units 1 &
0.139274
0.036157
0.042224
Martin Lake
(Units 1,2, &
0.081423
0.013585
0.01869
Monticello
Units (1,2, & 3)
0.151947
0.016356
0.020473
738
-------
Sandow Unit 4
0.032759
0.013648
0.009957
Sandow Unit 5
0.008672
0.003584
0.002271
272 Id. at A-52, tbl.A.4-4.
273 EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_072913.xlsx ("Avglmpacts").
274 As reflected in the table, ENVIRON modeled Sandow Unit 4 and Sandow Unit 5 as separate facilities.
275 EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_072913.xlsx ("AvgJmpacts'Vcol. K); TX116-007-
_29_Source_selection_analysis_TX_RH-l-31-14.xlsl.
Response: We explain in the response to comment above how we used the 0.3% threshold as a
way to identify a reasonable subset of sources with the largest visibility impacts to evaluate
further for reasonable controls. As discussed in the FIP TSD, we also considered including
sources that fell under the threshold based on additional information.583 Ultimately, we
identified 21 units at 9 facilities to evaluate for additional control analysis.584
We disagree with the commenter that the use of percentage extinction masks the impact of these
facilities on actual visibility. To the contrary, as we explain in response to comments below
where we discuss the use of natural background conditions and on the legality of visibility
metrics we utilized, deciview impacts calculated based on dirty background (as the commenter
has presented here) significantly understates the potential benefit from controls and the visibility
impact from a source. As explained in the preamble to the final Regional Haze Rule and
Guidelines for BART Determinations using a metric that is dependent on current degraded
background visibility conditions results in a paradox that the dirtier the existing air, the smaller
an individual source's visibility impact will become.585 In other words, as visibility conditions
worsen at a Class I area, the visibility impact from a source would appear to be smaller and
controls would be less likely. Therefore, to fully assess the potential visibility impacts and
benefits of controls at a source, deciview impacts based on natural "clean" background
conditions, as well as extinction and percent extinction should be considered. We explain in the
FIP TSD, "Results based solely on a degraded background, will rarely if ever demonstrate an
appreciable effect on incremental visibility improvement in a given area. Rather than providing
for incremental improvements towards the goal of natural visibility, degraded background results
will serve to instead maintain those current degraded conditions. Therefore, the visibility benefit
estimated based on natural or "clean" conditions is needed to assess the full benefit from
potential controls." 586
583 FIp TSD at A_53
584 FIP TSD at A-56, Table A.5-1 shows the 21 units
585 Using existing conditions as the baseline for single source visibility impact determinations would create the
following paradox: the dirtier the existing air, the less likely it would be that any control is required. This is true
because of the nonlinear nature of visibility impairment. In other words, as a Class I area becomes more polluted,
any individual source's contribution to changes in impairment becomes geometrically less. Therefore the more
polluted the Class I area would become, the less control would seem to be needed from an individual source. We
agree that this kind of calculation would essentially raise the "cause or contribute" applicability threshold to a level
that would never allow enough emission control to significantly improve visibility. Such a reading would render the
visibility provisions meaningless, as EPA and the States would be prevented from assuring "reasonable progress"
and fulfilling the statutorily-defined goals of the visibility program. Conversely, measuring improvement against
clean conditions would ensure reasonable progress toward those clean conditions. 70 FR 39124
586 FIP TSD at A-39
739
-------
We note that, consistent with the deciview values presented by the commenter above, the
Monticello and Big Brown facilities are projected to contribute approximately 1.3 Mm"1 and 1.2
Mm"1, respectively to visibility impairment on the 20% worst days at Wichita Mountains in 2018
based on the CENRAP 2018 projected emissions for these facilities. While the commenter
asserts that these impacts are miniscule, these individual facility impacts are 1.7% and 1.5% of
the total visibility impairment at Wichita Mountains. In our FIP TSD we noted that Texas used
an impact extinction level threshold of 0.5 Mm"1 (a level less than half of the estimated impact
from the Monticello or Big Brown facilities) from all sources in a state as a threshold for inviting
a state to consult.587 Oklahoma selected a threshold of 1.0 Mm"1 to determine which states
should consult in analyzing visibility impairment at Wichita Mountains. We also noted that the
largest projected contribution from all point sources within a state at Wichita Mountains after
Texas is Oklahoma at 3.9%. Elimination of all point sources in Oklahoma would result in less
visibility benefit (3.9%) than the visibility benefit from required controls (greater than 5%). We
estimated that the required controls provide for over 3 dv improvement at the Wichita Mountains
when estimated using a "clean" background. The required controls result in a greater than 5%
improvement in overall visibility conditions at the Wichita Mountains on the 20% worst days.
We also estimate that the required controls significantly reduce the projected delay in meeting
natural visibility, helping to achieve that goal 25 to 30-years earlier at Big Bend and the
Guadalupe Mountain by our projections. Our final analysis identified those emission units with
the largest visibility impacts. As these facts demonstrate, the identified facilities have significant
impacts on visibility conditions. Our technical record makes it equally plain that the required
controls reduce impacts from these sources and result in meaningful visibility benefits towards
the goal of natural visibility conditions.
The commenter is incorrect in stating that we do not report unit-level deciview impacts. Based
on our photochemical modeling results, we estimated uncontrolled unit level deciview impacts
based on recent actual emission levels and 2018 CENRAP projected emissions levels and
provide that data in the record.588 We also present visibility benefit of the analyzed control
levels for each unit in terms of extinction and deciviews.589
Comment: Step 9: EPA deviates from its own 0.3% threshold and "exclude[s]" some of the 20
units with impacts above the 0.3% threshold from "the second round of visibility modeling," but
"include[s]" other units with impacts below the 0.3% threshold based on "additional
information."223 EPA's ad hoc review results in a list of 21 units.
At Step 9. even though EPA previously established a 0.3% threshold, it disregards that cut-off by
"exclud[ing]" some of the 20 units with impacts above EPA's 0.3% threshold from "the second
round of visibility modeling," but "includ[ing]" other units with impacts below EPA's 0.3%
threshold based on so-called "additional information."277 For example, a single unit at the
Oklaunion Power Station modeled a 0.567% impact at Wichita Mountains, well above EPA's
587 See Texas Regional Haze SIP Appendix 4-1: Summary of Consultation Calls and Section X.A. of the Oklahoma
Regional Haze SIP
588 TX116-007- 33 Vis modeling summary available in the docket for this action. See worksheet "no control" for
extinction, and deciview impacts (clean and dirty background) of uncontrolled emissions
589 FIP TSD at Tables A.6-la,b,c and A.6-2a,b,c
740
-------
0.3% threshold.278 Nevertheless, EPA excluded the plant from further consideration because "its
impacts are a combination of NOx and SO2 impacts," and EPA found that "if just the impacts
from SO2 were examined, the facility's impacts would be below the 0.3% value."279 As
discussed in more detail below, back trajectories from Wichita Mountains travel near this facility
on the 20% worst days more frequently than other units that EPA chose to regulate. EPA also
fails to explain why it included NOx as part of its Q/D analysis, but then excluded NOx impacts
to haze conditions at Wichita Mountains in later steps of its analysis. Nor did EPA afford other
sources the same treatment as Oklaunion and examine "just the impacts from SO2" for other
units, even though the data suggests that other units (including Luminant's Sandow Unit 4) with
a combination of NOx and SO2 impacts above 0.3% would fall below that threshold when "just"
SO2 was considered.280
Further, even though none of the modeled results for Martin Lake Unit 1 were above EPA's
0.3% threshold, EPA arbitrarily included this unit "because it was above [EPA's threshold]
based on actuals and very close to the cutpoint with modeled values."281 Pirkey Power Plant, like
Martin Lake Unit 1, was above EPA's threshold for one scenario but not the other—despite
being well above the threshold based on modeled results (0.501%), it was below based on EPA's
adjusted results.282 Without further explanation, EPA excluded Pirkey from additional
consideration.283 EPA's selective treatment of sources at this step resulted in a "final" list of 21
units at only 9 facilities.284
277 FIP TSD at A-53.
278 Id. at A-52.
279 Id. at A-53.
28°TX116-007-_22_EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_MSF_v5 ("Avg_Impacts") (showing
Sandow Unit 4 with combined impacts of 0.376% at Wichita Mountains, but with SO2 impacts below those of
Oklaunion).
281 FIP TSD at A-53.
282 Id. at A-52.
283 Id. at A-54.
284 Id.
Response: Luminant does not present the full context of our explanation of how we determined
which sources would be included in the reasonable progress analysis and additional modeling.
In the FIP TSD, we state "we continued to evaluate whether to include or exclude sources that
were close to the cutpoint, or for which we had additional information that would indicate they
should be excluded in the second round of visibility modeling."590 We continue by providing
specific information that we relied on in determining whether or not to include that source in the
reasonable progress analysis after considering the modeled visibility impacts. Luminant also
does not provide an accurate description of our rationale for excluding Oklaunion from
additional reasonable progress analysis for this planning period. As we explain in a separate
response to comment, while impacts at the modeled emission rate of 7,157.8 tpy SO2 from the
2018 CENRAP projected emissions for Oklaunion unit 1 exceeded the 0.3% threshold (0.567%),
we noted that actual emissions (3,611 tpy annual average 2008-2012) are much lower and result
in an estimated impact of 0.286%, which was less than the threshold. Because this unit fell
below the threshold based on our consideration of past actual emissions, we did not evaluate
590 FIP TSD at A-53.
741
-------
additional controls or revised permit limits for this unit for this planning period. We also noted
that a portion of that impact was due to nitrate emissions from the source and therefore, the
percentage impact that was due to sulfur emissions would be even smaller than 0.286% and
below the 0.3% threshold. This is in contrast to the other EGUs evaluated where sulfur impacts
accounted for almost the entire visibility impact on the 20% worst days. We explain our
consideration of both NOx and SO2 impacts in our Q/d analysis and in evaluating the modeled
visibility impacts in a separate response to comment. We note that visibility model results
indicated that visibility impairment from those sources with the greatest visibility impacts was
almost entirely due to SO2 emissions, with the exception of Oklaunion as discussed above, and
PPG Glass Works, as discussed elsewhere in this document. Therefore, the control analysis for
these sources was focused on sulfate controls since controlling other emitted species, such as
NOx, would not result in significant visibility improvement on the 20% worst days for this
planning period.
We disagree with the commenter that our identification of sources for reasonable progress
analysis was "arbitrary" or that our application of the 0.3% threshold was "selective." We note
that we considered recent actual emissions at EGUs as more representative of future 2018
emissions from these sources.591 We explain in the FIP TSD, that similar to Oklaunion, based on
consideration of recent actual emissions, estimated percent contribution to visibility impairment
from Pirkey fell well below the 0.3% threshold.592 Therefore, Pirkey was excluded from
additional analysis. In this same manner, unit 1 at Martin Lake was just below the threshold at
0.296%) based on modeled emissions, however we noted that actual emissions are much higher
for this unit and estimated impacts based on actual emissions were well over the 0.3% threshold.
Furthermore, we noted that the two other units at Martin Lake were above the threshold
considering both recent actual and modeled emissions. Therefore all Martin Lake units were
included in our reasonable progress analysis.
With regard to the comment that we did not 'afford' other sources the same treatment as
Oklaunion and examine "just the impacts from SO2" for other units, even though the data
suggests that other units (including Luminant's Sandow Unit 4) with a combination of NOx and
SO2 impacts above 0.3% would fall below that threshold when "just" SO2 was considered, the
commenter is incorrect. With the exception of Oklaunion and PPG glassworks, modeled
visibility impacts from the individual sources are almost entirely due to sulfate impacts. For
example, contrary to the commenter's assertion, only 1% of the total modeled visibility impact at
Wichita Mountains from Sandow 4 is due to NOx emissions compared to over 11% in the case
of Oklaunion.593 The commenter is correct that the modeled extinction due to sulfate is less for
591 As explained in detail in the FIP TSD (see page A-45) we determined it was necessary to consider recent actual
emissions from EGUs due to uncertainty in 2018 projected emissions completed in 2006, the cost of SO2 credits
being lower than originally projected, and comments from Texas on a more recent IPM projection indicating that
significant SO2 reductions were not anticipated at these sources and no large SO2 control projects were planned at
most of the sources being evaluated. We also noted that TCEQ has utilized recent emission data for EGUs when
developing projected emissions for 2018 (and other future years) when developing ozone attainment demonstrations.
592 pip tsd a( A-54, "Pirkey had high modeled emissions and was above 0.3%, but the value was less than 0.1% for
the value based on actuals, so we did not include in our additional visibility modeling."
593 See TX116-007-_22_EPA_txbart3612k_Vis_2002_2018_PSAT_Projected_MSF_v5 ("Avg lmpacts") (showing
total extinction from Sandow 4 as 0.277 Mm-1 and nitrate extinction of 0.00295Mm-l compared to total extinction
from Oklaunion as 0.417 Mm-1 and nitrate extinction of 0.0463 Mm-1)
742
-------
Sandow 4 than for Oklaunion, but based on these modeled emissions, both Oklaunion and
Sandow 4 impacts are above the 0.3% threshold, even when only considering SO2 emissions.
Therefore, the commenter is incorrect in stating that had we only considered SO2 some of the
other facilities would fall below the threshold. Furthermore, when considering recent actual
emissions, the estimated visibility impact from Sandow 4 is much larger than the modeled value
due to recent actual emission being much larger than the modeled values. As discussed above,
actual emissions at Oklaunion are less than modeled.
We address comments concerning back trajectories provided by the commenter in a separate
response to comment.
Comment: EPA's evaluation lacks clear and objective metrics. [NRG (0078) p. 5]
NRG stated that the EPA's proposal does not identify a clear or objective basis on which EPA
identified sources to target for new controls as part of this action.
NRG stated that the EPA should identify such a basis, as EPA's own regulations and practice
would require an implementing state to have such a threshold. For example, BART rules require
use of such a threshold:
One of the first steps in determining whether sources cause or contribute to
visibility impairment for purposes of BART is to establish a threshold (measured
in deciviews) against which to measure the visibility impact of one or more
sources ... 40 CFR 51, Appx. Y, § III.A.l.
Once such a metric is set, it is a straightforward exercise to "compare the
predicted visibility impacts with your threshold for 'contribution. " 40 CFR 51,
Appx. Y, § III.A.3, Option 1.
Instead, NRG noted that the EPA's proposal relies on a series of subjective and ad hoc decisions
about which sources should make emission reductions and the level of reductions that should be
made. Among the numerous steps that lacked clear explanation, EPA began by considering a
small number of the numerous emissions sources in the state, investigated a smaller subset
further without clear explanation, excluded approximately half of those sources based on a non-
numerical assessment that their visibility impacts were low, then imposed an unprecedented
"light extinction" threshold to exclude some (but not all) facilities below the threshold from
controls. This approach does not reflect a reasonable application of the legal criteria applied to
determinations of reasonable progress goals.
Response: We disagree with the comment. We established a threshold of 0.3% contribution to
total visibility impairment to identify sources to evaluate for reasonable progress controls. We
discuss in the FIP TSD and in this RTC document, the steps we followed to identify those
sources with the largest visibility impacts for additional analysis. We began using a Q/d analysis
with a threshold of ten to identify sources with the greatest potential to impact visibility based on
emissions and location. We then used photochemical modeling to estimate the visibility
743
-------
impairment due to each of these facilities. Our modeling results indicated that a subset of the 38
facilities identified by the Q/d analysis were the primary contributors to visibility impairment at
each Class I area. We eliminated a number of sources based on consideration of facility-level
impacts that fell below 0.3%. We then used the 0.3% threshold on a unit basis only as a way to
identify a reasonable set of sources with the largest visibility impacts to evaluate further for
reasonable controls. We disagree with the comment and provided a clear explanation of what
thresholds we applied and our methodology to identify sources for reasonable control analysis.
As we discuss elsewhere, we believe this methodology and the metrics we utilize are consistent
with the Regional Haze Rule, the CAA and our guidance.
15.r Legality of Visibility Metrics Used
General comment: We received comments that that EPA established the deciview as the
required metric for establishing and tracking progress towards the reasonable progress goals.
EPA's use of extinction or % extinction and establishment of thresholds is arbitrary, capricious,
illegal and without precedent.
Comment: [USDA Forest Service (0083) p. 2] The USDA Forest Service noted, in the
document entitled "Technical Support Document for the Oklahoma and Texas Regional Haze
Federal Implementation Plans" (FIP TSD), EPA stated the following:
We concluded that any unit with an estimated impact greater than 0.3% would be
further evaluated. We believe that using a percent impacts approach is appropriate
because of its linkage to the reasonable progress concept. For example, a source
that has a smaller absolute impact on a relatively cleaner area but a higher
percentage impact might be considered for control so that the cleaner area can
potentially make progress. Since we had recent actual emissions, and any
feasibility of controls would likely be based on reductions from actuals, we
weighed the estimated impacts based on actuals in addition to the modeled impact
levels.
In the TSD, the EPA identified that difficulties arise when these modeled visibility impact levels
from RP analysis using photochemical modeling are compared to BART modeling metrics for
individual sources developed in support of other regional haze actions. The USDA Forest service
concurs with the EPA policy analysis that the analytical methods and model metrics used in the
RP context should be linked to the reasonable progress concept, including use of small
percentage threshold on the 20% worst and best days (20W or 20B) rather than using the FLAG
2010 or Appendix Y fixed threshold of 0.5 dv, 98th percentile to define significance. A potential
further ramification identified by your analysis is that use of the BART/FLAG approach could
result in eliminating a larger number of potential sources as a Class I area approaches its
reasonable progress goal (e.g. cleaner Class I airsheds), having the reverse effect of increasing
the difficulty of meeting the RPG. We applaud EPA's effort to ensure that methods and metrics
used in the RP context are meaningful and provide tangible information to support making
reasonable progress towards the national visibility goal.
744
-------
In summary, while the USD A Forest Service has expressed concern to EPA that the use of
visibility as a factor to be considered within the reasonable progress context may be outside the
statutory framework established for RP (see Clean Air Act, Section 169A (g)(1)), the
methodology and metrics that EPA used are the most comprehensive seen to date for any
SIP/FIP in the country that we have reviewed, and should serve as a model for future efforts to
consider the contribution and/or potential benefits of individual sources to visibility.
[Luminant (0061) p. Ill] Luminant stated that the various metrics and thresholds that EPA
employs in subsequent steps in its analysis—which it uses to judge which sources to further
regulate and which not—are random, unsupported, arbitrary and capricious, and inconsistent
with thresholds that EPA has previously used to judge reasonable progress in other states. EPA
concedes that, despite having already acted on every other state's reasonable progress goals,
there is no "prior precedent" for the approach it has taken for Texas.686 And EPA further
concedes that it has never before applied the various "thresholds" that it creates from whole cloth
for the current proposal.687 None of these thresholds withstands even the barest scrutiny.
686 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'n v. McCarthy, No. 11-01548, at 5 (D.D.C. 2014)
687 FIP TSD at A-35 ("We have not established specific metrics for use in evaluating single facility impacts on
visibility impairment (RP) as downwind Class I areas with a photochemical grid model such as CAMx . . . .").
[Luminant (0061) p. Ill] Luminant noted, at the outset, EPA's reliance on percentage
extinction as a visibility metric at critical steps is contrary to EPA's regulations and arbitrary and
capricious.688 As EPA has previously explained: "The RHR establishes the deciview (dv) as the
principal metric for measuring visibility."689 EPA has explained the reason the regulations use
deciviews as the required metric for establishing and tracking reasonable progress goals:
This visibility metric expresses uniform changes in haziness in terms of common
increments across the entire range of visibility conditions, from pristine to
extremely hazy conditions. . . . The deciview is a useful measure for tracking
progress in improving visibility, because each deciview change is an equal
incremental change in visibility perceived by the human eye. Most people can
detect a change in visibility at one deciview.690
Despite the fact that the deciview is "the required visibility metric identified in the [regional
haze regulations],"691 EPA abandons the deciview in its proposal here at critical junctures in its
analysis and instead uses percentage extinction to decide which sources are regulated and which
are not.692 EPA claims that it used "the percent [extinction] approach to somewhat normalize the
total extinction differences between the differing Class I areas."693
But EPA's explanation makes no sense and, in fact, contradicts EPA's rationale for adopting the
deciview as the required metric in its regulations. As EPA previously explained in adopting the
Regional Haze Rule, there is good reason it adopted the deciview as the required metric—it
reflects actual visibility conditions and "provid[es] a scale that relates visibility to perception."694
In other words, to measure progress toward the national goal, it is critical that changes in
visibility be considered against actual ambient conditions at the relevant Class I area (i.e., EPA's
so-called "dirty background"), not against a theoretical background that does not exist in the real
world. As EPA explained, "EPA supports the use of the deciview metric as calculatedfrom
745
-------
ambient monitoring data for tracking changes in regional visibility."695 And EPA further
explained that the "fundamental advantage of using the deciview" in its regulations is that it
"expresses uniform changes in haziness in terms of common increments across the entire range
of visibility conditions . . . ,"696 In other words, EPA's proposal here has it exactly backwards. It
is not necessary or appropriate to "normalize the total extinction differences between the
differing Class I areas."697 Instead, it is appropriate to consider each Class I area, as it exists and
as it is projected to exist at each interim step in the process. Under the binding regulations, the
relevant question here is what is a reasonable amount of visibility improvement at these Class I
areas in 2018 in terms of deciviews—not whether these three areas achieve some "normalized"
amount of improvement.698 In the end, EPA's explanation for using percentage extinction, as
opposed to deciviews, is unsupported by the record, inconsistent with the statute and regulations,
and not rationally connected to EPA's final decision, and it is thus unlawful.699
Footnotes:
686 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'n v. McCarthy, No. 11-01548, at 5 (D.D.C. 2014)
687 FIP TSD at A-35 ("We have not established specific metrics for use in evaluating single facility impacts on
visibility impairment (RP) as downwind Class I areas with a photochemical grid model such as CAMx . . . .").
688 Id. at A-50.
689 77 Fed. Reg. at 30,250.
690 Id.
691 77 Fed. Reg. 30,454, 30,459 (May 23, 2012) (emphasis added).
692 FIP TSD at A-50 ("We concluded that any unit with an estimated impact greater than 0.3% extinction would be
further evaluated.").
693 Id. at A-41.
694 64 Fed. Reg. at 35,727.
695 Id. (emphasis added).
696 Id. (emphasis added).
697 FIP TSD at A-41.
698 40 C.F.R. § 51.308(d)(1).
699 See Motor Vehicle Mfrs. Ass'n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (holding that an agency
decision is unlawful if the agency has failed to demonstrate that it has "examine [d] the relevant data and
articulate [d] a satisfactory explanation for its action including a rational connection between the facts found and the
choice made" (internal quotations omitted)).
[UARG (0065) p. 25-26] UARG explained that in this rulemaking, EPA has adopted new
analytical approaches that single out Texas and Oklahoma for unique and unfair treatment. EPA,
for instance, appears largely to have abandoned use of the deciview in assessing the Texas and
Oklahoma regional haze SIPs, opting instead to use an inverse megameter metric. 79 Fed. Reg.
at 74,839 n.208. Such an approach is inconsistent with every other EPA regional haze
rulemaking action of which UARG is aware. Further, this approach conflicts with EPA's
regional haze rule, which speaks in terms of deciviews and requires that the deciview metric be
used. 40 C.F.R. § 51.308(d)(1) ("the State must establish goals (expressed in deciviews)"); id. §
51.308(d)(l)(i)(B) ("the State must... determine the uniform rate of visibility improvement
(measured in deciviews)"); id. § 51.308(d)(2) ("the State must determine the following visibility
conditions (expressed in deciviews)"). Indeed, EPA's proposed rule appears to acknowledge that
use of the deciview metric is required, yet EPA ignores that requirement. See 79 Fed. Reg. at
74,826. EPA's approach obscures the differences between its own evaluation of visibility
impacts and the evaluation conducted by Texas. EPA, for instance, provides no direct
comparison of the visibility impacts it calculated with those estimated by the state. Id. at 74,839.
EPA's only explanation for relying on the inverse megameter, moreover, contradicts EPA's
746
-------
long-held position that the deciview is the most reliable metric. Compare id. at 74,839 n.208
("Extinction is an appropriate measure for the visibility impairment contribution from individual
sources because it avoids the sensitivity of the logarithmic transformation for calculating
deciviews to the overall level of visibility impairment including the impacts of other sources.")
with 64 Fed. Reg. at 35,727 ("the fundamental advantage of using the deciview remains: the
deciview metric expresses uniform changes in haziness in terms of common increments across
the entire range of visibility conditions, from pristine to extremely hazy conditions"). Indeed,
EPA itself reverts to reliance on deciviews, at times, to support its proposed FIPs. 79 Fed. Reg. at
74,885-87. In short, EPA provides no rational basis for substituting inverse megameters for
deciviews in its proposed rule.
[CCP (0075) p. 12] CCP stated that the EPA's extensive reliance on "source apportionment
modeling" based on "extinction" to determine that controls at Coleto Creek Unit 1 are necessary
is misplaced. Id. At 74,839. EPA attempts to justify its reliance on extinction in the Proposed
Rule, but "extinction" is never mentioned in EPA's Guidance for Setting Regional Progress
Goals Under the Regional Haze Program or in its own RPG rules. Rather, EPA's regional haze
rules confirm that visibility conditions for establishing RPGs should be "measured in deciviews"
or "expressed in deciviews." See 40 C.F.R. § 51.308(d)(1) ("the State must establish goals
(expressed in deciviews) that provide for reasonable progress"); id. at § 51.308(d)(l)(i)(B)
("determine the uniform rate of visibility improvement (measured in deciviews); id. at §
51.308(d)(2) (requiring calculations of baseline and natural visibility conditions to be "expressed
in deciviews") (emphasis added). EPA also has repeatedly confirmed the use of deciviews versus
extinction when analyzing visibility associated with regional haze SIPs. See, e.g., 77 Fed. Reg.
11,455 (Feb. 27, 2012) ("Georgia SIP") ("[t]he deciview is a more useful measure for tracking
progress in improving visibility than light extinction itself because each deciview change is an
equal incremental change in visibility perceived by the human eye."); 64 Fed. Reg. 35,725
(discussing benefits of using deciviews versus extinction).
[Xcel Energy (0064) p. 6, 13-14] [Xcel Energy (0064) p. 6] Xcel Energy stated that the EPA
ignored its own guidance and precedent in using "light extinction" rather than deciview impacts,
to justify inclusion of Tolk in the group of stationary sources controlled under the Proposal.
[Xcel Energy (0064) p. 13-14] Xcel Energy stated that, disregarding the low deciview
improvement projected from installing SO2 controls on the Tolk units and EPA's own guidance
focusing on deciview improvement, EPA chose to "evaluate^ other metrics, such as extinction
benefit or percent of extinction benefits," apparently for the sole purpose of justifying inclusion
of Tolk in this Proposal. 79 Fed. Reg. at 74,882 (emphasis added). EPA looks at light extinction
to argue "that the overall visibility benefit for installing scrubbers on the Tolk units was superior
to either the W. A. Parish or the Welsh units." /d. In a footnote, EPA says that "[extinction is an
appropriate measure for the visibility impairment contribution from individual sources because it
avoids the sensitivity of the logarithmic transformation for calculating deciviews to the overall
level of visibility impairment including the impacts of other sources." 79 Fed. Reg. at 74,839 n.
208. EPA ignores the other problems with using light extinction for such purposes. But, even if
there were a technical justification for use of light extinction rather than deciview impacts, EPA
never explains why it uses this approach for the first time in reviewing Texas' SIP or as
determinative solely as to the Tolk Generating Station.
747
-------
Xcel Energy stated that the EPA's resort to "light extinction" is directly contrary to its own
guidance. Using light "extinction" to assess visibility impairment is never mentioned in EPA's
RPG Guidance or in its own regional haze rules. Rather, EPA's regional haze rules repeatedly
state that visibility conditions for establishing RPGs should be "measured in deciviews" or
"expressed in deciviews." See 40 C.F.R. § 51.308(d)(1) ("the State must establish goals
(expressedin deciviews) that provide for reasonable progress"); id. at§ 51.308(d)(l)(i)(B)
("determine the uniform rate of visibility improvement (measured in deciviews"); id. at § 51.308(
d)(2) (requiring calculations of baseline and natural visibility conditions to be "expressed in
deciviews") (emphasis added). EPA also has repeatedly confirmed the use of deciviews versus
light extinction when analyzing visibility associated with previous regional haze SIPs. See, e.g.,
Proposed Georgia SIP Approval, 77 Fed. Reg. at 11,455 ("[t]he deciview is a more useful
measure for tracking progress in improving visibility than light extinction itself because each
deciview change is an equal incremental change in visibility perceived by the human eye.");
RHR, 64 Fed. Reg. at 35,725 (discussing benefits of using deciviews versus extinction). EPA has
no reasoned basis to, for the first time, ignore deciview impacts and use "light extinction" to
justify imposing controls on Tolk.
Xcel Energy stated that it is indicative of the arbitrariness of EPA's use of light extinction solely
for justifying inclusion of Tolk in the group of sources subject to the four-factor reasonable
progress analysis that EPA then discusses the visibility benefits of controls only in terms of
deciview improvement. EPA claims that all of the information it considered regarding the
benefits of proposed controls is included in the FIP Technical Support Document ("TSD"). 79
Fed. Reg. at 74,882. However, EPA fails to present the benefits of proposed controls in terms of
light extinction in either in the Proposal or the FIP TSD. Failing to consistently represent impacts
in terms of light extinction arbitrarily obfuscates the purported benefits of the proposed controls.
Luminant, Xcel, CCP, and UARG commented that use of the inverse megameter or percentage
extinction metric is contrary to EPA regulations and is arbitrary and capricious. The commenters
state that the deciview is the required visibility metric. The commenters cite to language in our
actions in Idaho (77 FR 30250), Oregon (77 FR 30454), and Georgia (77 FR 11455) as well as
the Regional Haze Rule (64 FR 35727) that discuss the benefits of using the deciview metric and
the establishment of the deciview as the principal metric for measuring visibility. They also state
that using light extinction is never mentioned in EPA's RPG guidance or in the regional haze
rules, and is inconsistent with other EPA actions on regional haze.
CCP and Xcel comment we relied on extinction and percent of total extinction to justify
controlling Coleto Creek and Tolk Station. Xcel continues by stating that we used light
extinction rather than deciview as determinative solely for Tolk Station. Xcel also comments
that we failed to present the benefits of proposed controls in terms of light extinction in either the
Proposal or the FIP TSD.
Luminant also comments that it is not necessary or appropriate to "normalize the total extinction
differences between the differing Class I areas."594 Instead, it is appropriate to consider each
Class I area, as it exists and as it is projected to exist at each interim step in the process. Under
594 FIP TSD at A-41
748
-------
the binding regulations, the relevant question here is what is a reasonable amount of visibility
improvement at these Class I areas in 2018 in terms of deciviews—not whether these three areas
achieve some "normalized" amount of improvement.
UARG also comments that our approach obscures the differences between our evaluation of
visibility impacts and the evaluation conducted by Texas, and that we fail to provide a direct
comparison of the visibility impacts we calculated to those calculated by the State.
The USD A Forest Service expressed their support in their comments on the methods and metrics
used in our analysis. The USDA Forest service "concurs with the EPA policy analysis that the
analytical methods and model metrics used in the RP context should be linked to the reasonable
progress concept, including use of small percentage threshold on the 20% worst and best
days.. .We applaud EPA's effort to ensure that methods and metrics used in the RP context are
meaningful and provide tangible information to support making reasonable progress towards the
national visibility goal." They conclude by stating that "the methodology and metrics that EPA
used are the most comprehensive seen to date for any SIP/FIP in the country that we have
reviewed, and should serve as a model for future efforts to consider the contribution and/or
potential benefits of individual sources to visibility."
Response: We disagree with the commenters that our use of metrics other than deciviews for
certain purposes is contrary to the regional haze regulations. The commenters fail to distinguish
between the metrics used to describe overall visibility conditions at a Class I area and the metrics
that can be used to describe the visibility impairment due to an individual source, group of
sources, a state's sources, or some other portion of the visibility impairment at a Class I area. In
describing the overall visibility conditions at a Class I area, we established the deciview as the
principle metric. This applies to the calculation of current, baseline, and natural visibility
conditions at a Class I area, as well as the reasonable progress goal established as the visibility
condition goal for the Class I area at the end of the current planning period. We agree with the
commenters that the use of the deciview metric is required in a number of places within the rule
that discusses overall visibility conditions and assessing progress towards meeting the desired
visibility conditions. Specifically, the state must 1) establish reasonable progress goals
expressed in deciviews (40 CFR 51. 308(d)(1)); 2) determine the uniform rate of progress in
deciviews (40 CFR 51.308(d)(l)(i)(B)); and 3) determine the baseline and natural visibility
conditions expressed in deciviews and the number of deciviews by which baseline conditions
exceed the natural conditions (40 CFR 51.308(d)(2)). Consistent with these requirements, we
calculated the baseline and natural visibility conditions, the uniform rate of progress, and the
number of deciviews by which baseline conditions exceed the natural conditions in deciviews for
Big Bend and the Guadalupe Mountains, as well as established reasonable progress goals for the
Wichita Mountains and the Texas Class I areas in deciviews.
The deciview metric provides a scale that relates to visibility perception and therefore is useful in
assessing the overall visibility conditions that are being or will be perceived at the Class I area.
The commenters cite to several actions and the Regional Haze Rule where the benefits of using
the deciview metric are discussed, however this is only discussed in the context of overall
visibility conditions, such as determining current or natural visibility conditions. This is very
different from the fraction of visibility impairment attributable to a source or group of sources.
749
-------
We note that in the final Regional Haze Rule we do in fact mention the use of light extinction as
another metric that states may choose to use.595
There is no requirement to use the deciview metric in describing the visibility impairment due to
a source or group of sources as part of the analysis required for identifying reasonable controls
under reasonable progress. In describing how to identify sources or source categories
responsible for visibility impairment, our guidance596 provides States with considerable
flexibility to utilize various tools and techniques that would necessarily involve the use of
various metrics other than deciviews. Many states and RPOs, including Texas and CENRAP,
relied on a Q/d analysis, described and discussed in depth in separate responses to comments and
in our proposed FIP, to identify sources for additional control analysis. The Q/d analysis relies
on an annual emissions divided by distance metric, not deciviews. The VISTAS RPO relied on a
metric derived from Q/d and residence-time597, not deciviews. Some states relied on a simple
analysis of emissions to determine which sources should be analyzed.
When assessing the various contributions to visibility impairment due to either source categories
or pollutant species from other states and international sources, Texas routinely relied on light
extinction and percent of total visibility impairment metrics. For example, Chapter 11 of the
Texas regional haze SIP describes the contributions due to sulfate, nitrate, and other pollutants
on the 20% worst and 20% best days at Guadalupe Mountains and Big Bend in terms of light
extinction (inverse megameters, Mm"1). Similarly, the extinction metric is used by Texas (see
section 11.2.3 of the TX RH SIP) to assess the level of impact on other Class I areas from Texas
sources. Texas relies on the percent of total visibility metric in identifying the types of sources
(e.g. point, area, mobile) that contribute the most to visibility impairment from sulfate and
nitrate598 and in identifying the relative contributions to visibility impairment due to emissions
from various source areas (e.g. Texas, Mexico, neighboring states).599 Texas also used the
extinction metric to determine which states significantly impact the Texas Class I areas, applying
an impact extinction level threshold of 0.5 Mm"1 from all sources in a state as a threshold for
inviting a state to consult.600 Source apportionment modeling performed by the RPOs was
utilized by every state to assess the various contributions to visibility impairment at their Class I
areas in terms of light extinction and percent contribution to total light extinction. The CENRAP
PM source apportionment tool (CENRAP PSAT tool)601 utilized by all CENRAP states,
595 "The final rule maintains the deciview as the principle visibility metric used in establishing reasonable progress
goals, in defining baseline, current, and natural conditions, and in tracking changes in visibility conditions over time.
States may choose to express visibility changes in terms of other metrics, such as visual range or light extinction, as
well as in terms of deciview." 64 FR 35,727
596 "Once the key pollutants contributing to visibility impairment at each Class I area have been identified, the
sources or source categories responsible for emitting these pollutants or pollutant precursors can also be determined.
There are several tools and techniques being employed by the RPOs to do so, including analysis of emission
inventories, source apportionment, trajectory analysis, and atmospheric modeling." Guidance for Setting
Reasonable Progress Goals Under the Regional Haze Program, U.S. EPA, OAQPS, June 1, 2007, page 3-1
597 VISTAS Area of Influence Analyses, 2007, available in the docket for this action.
598 Texas Regional Haze SIP at Table 3 of Appendix 10-1 showing percent contribution from different source types
to S04 and N03 at the five Class I areas
599 Texas Regional Haze SIP Table 10.8 at p. 10-10
600 See Texas Regional Haze SIP Appendix 4-1: Summary of Consultation Calls
601 Available in the docket to this action as "CENRAP_PSAT_Tool_ENVIRON_Aug27_2007.mdb"
750
-------
including Texas and Oklahoma, to review the results of the source apportionment modeling
provides results in two ways: light extinction (inverse megameters) and percentage of total
extinction. In our action, we also utilized the methodology and metrics used by the RPOs to
evaluate the source apportionment results, the only difference being that our source
apportionment modeling provided information on visibility impacts from individual sources
instead of source categories, or regions/states. In the FIP TSD, we provide information on
visibility impacts from the individual sources in terms of extinction, percentage of total
extinction, and in deciviews.602
We evaluated the information in terms of light extinction and percentage of total impact to
identify a reasonable subset of sources with the largest visibility impacts to analyze for additional
controls. Because the overall visibility conditions at different Class I areas can vary greatly,
particularly Class I areas in the Eastern U.S. compared to Class I areas in the Western U.S., we
determined that it is not enough to consider just the magnitude of extinction from a facility, we
must also consider the percentage of total impairment metric at each Class I area. As we state in
the FIP TSD, "We believe that using a percent impacts approach is appropriate because of its
linkage to the reasonable progress concept. For example, a source that has a smaller absolute
impact [in terms of extinction] on a relatively cleaner area but a higher percentage impact might
be considered for control so that the cleaner area can potentially make progress." Using the
percentage of total visibility impairment metric allows us to somewhat normalize the extinction
differences between Class I areas so that we can utilize the same approach at each Class I area
and identify a reasonable set of sources to analyze that if controlled would result in meaningful
visibility benefits towards meeting the goal of natural visibility at every Class I area. Contrary to
Luminant's comment, these source apportionment model results are tied to projected conditions
in 2018 at each Class I area. The estimated light extinction is based on projected 2018 emission
levels and the percentage impact is based on the percentage of the total modeled visibility
impairment in 2018. We note that we also considered recent actual emissions at EGUs as more
representative of future 2018 emissions from these sources.603 For every Class I area to have the
opportunity to reach the natural visibility goals, it is necessary to identify the sources or source
categories that significantly impact visibility, identify available controls and analyze whether
those controls are reasonable. Had we established a strict threshold based on extinction, we
would have had to establish a different threshold for each Class I area. For example, had we
selected a threshold of 0.35 Mm-1 on a facility basis we would have identified ten facilities for
additional analysis based on modeled visibility impacts at Wichita Mountains. However, use of
this same threshold applied to either Big Bend or Guadalupe Mountains fails to identify the
largest individual sources impacting visibility at these Class I areas for additional analysis to
determine if reasonable controls are available. Using a percentage approach, such as the 0.3% of
total visibility impairment on a unit basis we used in this action, results in identification of a
subset of sources that include those sources with the greatest visibility impacts at each class I
602 As discussed in more detail elsewhere, we estimated deciview impacts using "clean" background and "dirty"
background conditions in our assessment of visibility impacts and benefits for individual sources.
603 As explained in detail in the FIP TSD (see page A-45) we determined it was necessary to consider recent actual
emissions from EGUs due to uncertainty in 2018 projected emissions completed in 2006, the cost of SO2 credits
being lower than originally projected, and comments from Texas on a more recent IPM projection indicating that
significant S02 reductions were not anticipated at these sources and no large S02 control projects were planned at
most of the sources being evaluated. We also noted that TCEQ has utilized recent emission data for EGUs when
developing projected emissions for 2018 (and other future years) when developing ozone attainment demonstrations.
751
-------
area. As stated by the USDA Forest Service in its supportive comments, the use of this
methodology and metrics, including the use of a small percentage threshold on the 20% worst
days is linked to the concept of reasonable progress. We believe it could serve as the model for
future efforts to consider the contribution and potential benefits of individual sources to
visibility. After identifying which sources to analyze for additional controls based on the
percentage impact on a unit basis, we determined which controls were reasonable based on
consideration of the four factors, including comparison of cost to the anticipated visibility benefit
(deciview improvement, extinction, percentage of total extinction, and the percentage of the total
impact from Texas point sources addressed by the control).
We disagree with Luminant's comment that "under the binding regulations, the relevant question
here is what is a reasonable amount of visibility improvement at these Class I areas in 2018 in
terms of deciviews—not whether these three areas achieve some 'normalized' amount of
improvement." The RHR requires that we identify reasonable controls based on consideration of
the four statutory factors and then establish a reasonable progress goal that reflects the
anticipated amount of visibility improvement from implementation of those controls in
additional to all other "on the books" controls. One cannot determine what a reasonable amount
of visibility improvement at each Class I area is without evaluating reasonable controls at the
sources or source categories responsible for the visibility impairment. As an initial step we
identify those sources or source categories with the largest visibility impacts at each Class I area.
As discussed above, by using the percentage impact approach we were able to somewhat
normalize the extinction impacts and identify those sources with the most significant impacts at
each class I area. Secondly, within the four-factor analysis we took into consideration the
projected visibility benefit of the controls to identify cost-effective controls that will achieve
reasonable visibility benefits required during this planning period towards the national goal. At
this step, we considered the anticipated visibility benefit in deciviews604 (for both a "dirty
background" and a "clean background") as well as the reduction in extinction and the percentage
of total visibility impairment addressed by the controls.
With regards to CCP and Xcel's comments concerning our reliance on extinction and percent of
total extinction to determine if controls should be required at Coleto Creek and Tolk Station, we
disagree with the comment. As we state in our proposed action and discussed in the FIP TSD,
we weighed deciview benefits, as well as extinction benefits and percentage of total extinction
information in making our proposed findings about the benefits of potential controls at all the
evaluated units.605 In considering controls on Tolk, we noted that the visibility benefits occur
mainly at Guadalupe Mountains. We also noted that in terms of deciview improvement, the
visibility benefit from controlling Tolk at Guadalupe Mountains was smaller than the benefit of
controls at Welsh or Parish on visibility at Wichita Mountains. However, in comparing the level
of visibility benefit at one Class I area to another, the percent extinction benefits are more useful
because they avoid the deciview metric's sensitivity to the logarithmic transformation that
depends on the total visibility impairment that can vary from Class I area to Class I area. As we
604 In the FIP TSD we explain in depth the difficulties that arise in trying to use the deciview metric in the context of
CAMx modeling for reasonable progress compared to the use of deciviews in the context of single-source BART
modeling using CALPUFF (see FIP TSD at A-35.), as well as the need to use clean background conditions for the
deciview metric (see FIP TSD at A-39)
605 79 FR 74882
752
-------
discuss above, using the percent extinction metric somewhat normalizes extinction between
different Class I areas. In particular, the Wichita Mountains has a much higher total extinction
for the baseline and the 2018 projection than the Guadalupe Mountains, so the relative
improvement in extinction levels is higher when the Tolk units are controlled for the Guadalupe
Mountains, than if the W. A. Parish or the Welsh units were controlled for the Wichita
Mountains. We also considered the level of impact from all Texas point sources at each Class I
area and the amount of that impact addressed by individual controls. We determined that
controls were reasonable for Coleto Creek and Tolk based on consideration of the four factors,
including comparison of cost to the anticipated visibility benefit (deciview improvement,
extinction, percentage of total extinction, and the percentage of the total impact from Texas point
sources). We note that Coleto Creek alone accounts for over 6% of the total Texas point source
visibility impact at Big Bend and Tolk accounts for nearly 8% of the total Texas point source
impact at Guadalupe Mountains.
Xcel is incorrect in commenting that we failed to present the benefits of the proposed controls in
terms of light extinction. Tables A.6-la, A.6-l.b, and A.6-1.C in the FIP TSD present the
average change in extinction levels for different controls at Wichita Mountains, Guadalupe
Mountains, and Big Bend. We also note that tables A.6-2.a, A.6-2.b, and A.6-2.C present the
change in deciview for different controls at the Class I areas. The FIP TSD also presents
modeled visibility impacts from the facilities in terms of extinction and percent of total
extinction.
As to UARG's comment that our approach obscures the differences between our evaluation and
Texas' and that we failed to provide a direct comparison of visibility impacts, Texas did not
assess the visibility impact from individual sources and only estimated the total visibility benefit
from controlling those sources that Texas identified based on Q/d and consideration of cost-
effectiveness. As we discuss further in our FIP TSD and in other responses elsewhere in our
RTC document, to provide context regarding the significance of our estimated individual source
impacts and benefits from controls, we compared the individual source impacts with CENRAP
source apportionment modeling results for impacts from all emission sources within a state and
impacts from all emission sources within a state within a specific source type, examining
extinction and percent extinction. We also compared these individual source impacts to the
impact levels used by the states for triggering consultation with another state about its overall
impacts (extinction levels), and the estimated range of anticipated visibility benefits resulting
from required controls in other actions.606
As we discuss in more detail elsewhere where we discuss comments concerning legal deference,
we disagree with Luminant's cited use of Sam Coleman's declaration and it is taken out of
context. While EPA did say that there was no "prior precedent," this was in regards to the
particular type of modeling undertaken; Luminant takes Coleman's statement out of context.
Coleman's statement is related to the additional modeling we determined was appropriate due to
the large distances involved and the large number of sources being analyzed, which was a unique
set of facts not encountered by us in the Regional Haze context before. Luminant conflates this
context and over broadens the scope of Coleman's statement beyond the modeling to our
approach in the proposal overall. Coleman's statement was not stating there was no prior
606 See our FIP TSD at A-75.
753
-------
precedent for the basis of our proposal, nor that our approach was inconsistent with other
regional haze SIP actions.
15.s Adjustment of CAMx Results: Linear Extrapolation and Nox Vs. SO2, Natural
Background, Recent Actual Emissions
General Summary: We received comments on the method EPA used to adjust CAMx results.
EPA developed a linear relationship between emissions and extinction and then adjusted CAMx
modeled extinction linearly with emissions to match proposed controlled emission levels. The
commenters state that the relationship between emissions and light extinction is not linear and
that interactions between nitrate and sulfate create a complicated relationship. The CAMX user
guide supports that the relationship is non-linear.
We also received comments on the calculation of a dv impact or improvement based on natural
"clean" background conditions and the estimated visibility impacts/improvement based on recent
actual emissions rather than projected 2018 emissions.
15.t Linear Extrapolation
Comment: [AECOM (0061 and 0075) p. 7-1] AECOM stated that there are anomalies in how
EPA considered and manipulated the results of the CAMx modeling that was performed. For
example, with these modeling results, EPA formulated the visibility benefits by using a linear
approach that provides an estimate of the extinction for each facility as a function of controlled
emissions. However, the relationship between a change in emissions to the corresponding change
in deciviews is not linear since: 1) the relationship between an emissions rate and the
corresponding effect on light extinction is a complex non-linear system (see section 1.2 of the
CAMx Users Guide Version 6.1 (April 2014)), depending on multiple parameters described in
more detail below; and 2) the relationship between light extinction and deciviews is an
exponential relationship as shown above.
AECOM stated that the assumption of a linear response ultimately assumes that the modeled
amount of particulate concentrations will be reduced linearly by decreasing emissions, but there
is no certainty this will occur. In the atmosphere, there is competition between sulfate and nitrate
for the available ammonia, the interaction that results from this competition creates a rather
complicated system. As stated in Seinfeld's widely used textbook on Atmospheric Chemistry,
"reductions in aerosol sulfate will result in partial replacement of the reduced aerosol mass by
available nitric acid. The sulfate decreases frees up ammonia to react with nitric acid and to
transfer it to the aerosol phase."113 Therefore, sulfate reductions can be accompanied by an
increase in the aerosol nitrate, a compound which also contributes to visibility impairment. Thus,
the assumption of a linear relationship between SO2 emissions reductions and visibility
improvement does not accurately represent the complexity of the sulfate-nitrate-ammonia system
in the atmosphere and could overstate the claimed visibility improvements expected from the
EPA FIP requirements.
AECOM stated that, in particular, a limitation of the linear model used by EPA is that although
small perturbations relative to model inputs might be approximated by a linear relationship, EPA
754
-------
extended the linear treatment to larger variations. In fact, the proposed level of controls is well
outside the range actually used in the photochemical modeling. This was particularly true for the
wet flue gas desulfurization controls in which the proposed emissions range from 521 to 700 tons
per year (tpy) of SO2 (depending on the facility), but the high control modeling scenario
considered emissions in the range of 1,110 to 4,595 tpy. Therefore, EPA assumes, without
justification, that the linear model used to estimate the response to the visibility impairment from
emissions will still be valid when the proposed emissions are a factor of 2 to 6 smaller than the
emissions used in the CAMx modeling scenario with the highest controls. There is no basis for
this assumption.
Footnotes:
113 Seinfeld J. H. and Pandis S. N. (1998) Atmospheric Chemistry and Physics: From Air Pollution to Climate
Change, 1st edition, J. Wiley, New York.
Step 11: In clear contradiction to the CAMx User's Guide, EPA linearly scales the results
of the CAMx modeling from ENVIRON to estimate the visibility benefit associated with
larger changes in emission rates at the 21 remaining units.
(EJ Gray report pg 18-19): EPA's approach here results in reasonable estimates of the visibility
benefits associated with the evaluated controls. The physical (atmospheric) system being
modeled by CAMx consists of a number of processes (plume rise, transport, chemistry,
deposition, etc.). Many of these are "linear" in nature (meaning a scaling of emissions would
result in a scaling of ambient concentration), but some are not (such as the chemical mechanism).
However, for relatively small changes in total SO2 emissions (from all sources), the chemistry
(and the overall model results) can be approximated as a linear process, which is the approach
EPA has taken here.
The high and low emission rates, developed from actual emissions data and realistic control
percentages, are representative of emission rates under typical high and low control scenarios for
each unit rather than being associated with control levels from each of the possible controls.
Importantly, the collection of emission rates for the high and low control levels for the various
units at each facility, together with the 2018 baseline facility emission levels, spanned the
possible values of emissions at each facility so that the slope (for each facility) could be
determined using linear regression.
The linear regression procedure that EPA used to relate each unit's SO2 emission rate to its
extinction contribution at each modeled Class I area relies on all other variables (other than the
SO2 emission rate from the specific source unit) being constant. Unfortunately, the background
conditions (which are made up of contributions from all pollutant sources in the model) are not
exactly the same for the low and high (and 2018 baseline) emission scenarios, because the other
20 controlled units are at different levels between the low and high (and baseline) model runs.
Examination of the model results for the low versus high control scenarios (with all sources) at
numerous modeled Class I areas show that the modeled extinction levels are not very far apart
for the purposes of linear regression, so the uncertainty in the estimation of the regression slopes
(change in extinction per ton of emissions) for the individual source units should not be large due
to this slight mismatch. Thus, EPA's approach was ultimately reasonable for these purposes.
755
-------
A separate linear regression was performed for each of the nine facilities and at each of the 19
modeled Class I areas. For many of these regressions, few distinct data points were used; for
example, at facilities with only one unit (Coleto Creek and Sandow 4) there were only three data
points used in the regressions. Many of the high or low control levels at different units within the
same facility were only slightly different, resulting in similar emission levels and modeled
extinctions, which essentially created duplicate regression points. For many of the facilities, the
2018 baseline facility-wide data point essentially represented a "doubled" or "tripled" unit
emission and extinction impact.
Finally, it should be noted that some of the regressions consisted of a single point from the 2018
baseline facility-wide model results combined with one or more additional points that are located
"approximately" at the origin (zero modeled impact at a very low relative emissions level). The
resulting slopes for these regressions are defined almost entirely by the 2018 baseline model
results. However, the source unit/Class I area impacts in the regressions where this issue was
observed were very small relative to the impacts that are the primary focus of this analysis (1-2
orders of magnitude lower); their uncertainty does not impact the overall conclusions. For all
impacts that play a significant part in EPA's analysis, the regressions appear to have been
developed with a reasonable, although small, set of practical data points.
In sum, due to the observed linearity between modeled incremental extinction and SO2 emission
rates, and the fact that the overall modeled extinction levels (from all sources) weren't very
different between the low, high, and 2018 baseline scenarios, the resulting regression slopes
provide a reasonable estimation of the change in extinction at each Class I area that would result
from a change in SO2 emission rate from units at each facility.
[Gray (0070) p. 18] Dr. Gray stated that the EPA used CAMx to establish a mathematical
relationship between the tons reduced by a given control at a specific unit, and the resulting
visibility benefit to each Class I area. This method produced a reasonable estimate of the
visibility benefits from the potential controls. (FIP TSD, Sections A.5 and A.6)
Dr. Gray noted that the EPA's approach to estimate the visibility impacts of the 21 selected
source units and the benefits of emission reductions from each unit, EPA developed a linear
relationship between tons of emissions reduced and visibility benefit in extinction, allowing it to
estimate benefits from any level of control. Using actual emissions data and generalized
emission reduction percentages, EPA created a pair of high and low control estimates for each
unit. It then modeled the benefits from each scenario using CAMx.
Dr. Gray noted that the high and low control level model results for each unit were combined
with the earlier facility-wide model results to create a set of modeled visibility levels (average
extinction during W20 days) at each Class I area. EPA found that, "For each facility and Class I
area, the modeled data was linear with high correlation. Therefore we used the linear fit to
extrapolate the anticipated visibility impact/benefit from a given level of emission/control." The
resulting linear regression slope gives the visibility benefit (change in extinction) per ton of SO2
emissions reduced.
Dr. Gray noted that the EPA then calculated the tons reduced by each control to estimate the
visibility benefit from that control. EPA updated the baseline uncontrolled emission rates for
756
-------
each unit based on CEM data for 2009-2013. These baseline emission rates (the maximum and
minimum emissions years were eliminated and the three remaining years were averaged) were
used to estimate the amount of tons reduced for each control measure. Then the linear
relationship described above was used to estimate the visibility improvement corresponding to
each potential control measure evaluated at the 21 selected source units.
Dr. Gray commented that the EPA's approach here results in reasonable estimates of the
visibility benefits associated with the evaluated controls. The physical (atmospheric) system
being modeled by CAMx consists of a number of processes (plume rise, transport, chemistry,
deposition, etc.). Many of these are "linear" in nature (meaning a scaling of emissions would
result in a scaling of ambient concentration), but some are not (such as the chemical mechanism).
However, for relatively small changes in total SO2 emissions (from all sources), the chemistry
(and the overall model results) can be approximated as a linear process, which is the approach
EPA has taken here.
Dr. Gray stated that the high and low emission rates, developed from actual emissions data and
realistic control percentages, are representative of emission rates under typical high and low
control scenarios for each unit rather than being associated with control levels from each of the
possible controls. Importantly, the collection of emission rates for the high and low control levels
for the various units at each facility, together with the 2018 baseline facility emission levels,
spanned the possible values of emissions at each facility so that the slope (for each facility) could
be determined using linear regression.
Dr. Gray stated that the linear regression procedure that EPA used to relate each unit's SO2
emission rate to its extinction contribution at each modeled Class I area relies on all other
variables (other than the SO2 emission rate from the specific source unit) being constant.
Unfortunately, the background conditions (which are made up of contributions from all pollutant
sources in the model) are not exactly the same for the low and high (and 2018 baseline) emission
scenarios, because the other 20 controlled units are at different levels between the low and high
(and baseline) model runs. Examination of the model results for the low versus high control
scenarios (with all sources) at numerous modeled Class I areas show that the modeled extinction
levels are not very far apart for the purposes of linear regression, so the uncertainty in the
estimation of the regression slopes (change in extinction per ton of emissions) for the individual
source units should not be large due to this slight mismatch. Thus, EPA's approach was
ultimately reasonable for these purposes.
Dr. Gray stated that a separate linear regression was performed for each of the nine facilities and
at each of the 19 modeled Class I areas. For many of these regressions,22 few distinct data points
were used; for example, at facilities with only one unit (Coleto Creek and Sandow 4) there were
only three data points used in the regressions. Many of the high or low control levels at different
units within the same facility were only slightly different, resulting in similar emission levels and
modeled extinctions, which essentially created duplicate regression points. For many of the
facilities, the 2018 baseline facility-wide data point essentially represented a "doubled" or
"tripled" unit emission and extinction impact.
Finally, Dr. Gray stated that it should be noted that some of the regressions consisted of a single
757
-------
point from the 2018 baseline facility-wide model results combined with one or more additional
points that are located "approximately" at the origin (zero modeled impact at a very low relative
emissions level). The resulting slopes for these regressions are defined almost entirely by the
2018 baseline model results. However, the source unit/Class I area impacts in the regressions
where this issue was observed were very small relative to the impacts that are the primary focus
of this analysis (1-2 orders of magnitude lower); their uncertainty does not impact the overall
conclusions. For all impacts that play a significant part in EPA's analysis, the regressions appear
to have been developed with a reasonable, although small, set of practical data points.
In sum, Dr. Gray concluded that due to the observed linearity between modeled incremental
extinction and SO2 emission rates, and the fact that the overall modeled extinction levels (from
all sources) weren't very different between the low, high, and 2018 baseline scenarios, the
resulting regression slopes provide a reasonable estimation of the change in extinction at each
Class I area that would result from a change in SO2 emission rate from units at each facility.
Footnote:
22 Regression results can be found in tab "sorted PSAT data" in EPA's spreadsheet: "TX116-007-
_3 3_Vis_modeling_summary. xlsx"
CCP and Luminant commented on the method we used to adjust CAMx results based on
emission difference between modeled emissions and controlled emission levels. The method
relies on adjusting CAMx modeled extinction linearly with emissions to match proposed
controlled emission levels. The commenters state that the relationship between a change in
emissions to the corresponding change in deciviews is not linear since: 1) the relationship
between an emissions rate and the corresponding effect on light extinction is a complex non-
linear system and 2) the relationship between light extinction and deciviews is an exponential
relationship. The CAMx user guide supports that the relationship is non-linear. As stated in a
widely used textbook on Atmospheric Chemistry, "reductions in aerosol sulfate will result in
partial replacement of the reduced aerosol mass by available nitric acid. The sulfate decreases
frees up ammonia to react with nitric acid and to transfer it to the aerosol phase."607 Therefore,
sulfate reductions can be accompanied by an increase in the aerosol nitrate, a compound which
also contributes to visibility impairment. Thus, the assumption of a linear relationship between
SO2 emissions reductions and visibility improvement could overstate the claimed visibility
improvements expected from the FIP requirements. The commenters state that small
perturbations relative to the model inputs can be approximated as linear. However, the
commenters assert that we extended the linear treatment to larger variations. The commenters
continue that the proposed level of controls is well outside the range actually used in the
photochemical modeling. For example, the proposed emissions ranged from 521 to 700 tons per
year (tpy) of SO2 (depending on the facility), but the control modeling scenario considered
emissions in the range of 1,110 to 4,595 tpy. The use of a linear model is not justified when the
proposed emissions are a factor of 2 to 6 smaller than the emissions used in the CAMx modeling
scenario with the highest controls.
607 Seinfeld J. H. and Pandis S. N. (1998) Atmospheric Chemistry and Physics: From Air Pollution to Climate
Change, 1st edition, J. Wiley, New York.
758
-------
We also received comments from Earthjustice et al. that summarized and supported the
methodology used to adjust the modeling results:
"EPA's approach here results in reasonable estimates of the visibility benefits associated with the
evaluated controls. The physical (atmospheric) system being modeled by CAMx consists of a
number of processes (plume rise, transport, chemistry, deposition, etc.). Many of these are
"linear" in nature (meaning a scaling of emissions would result in a scaling of ambient
concentration), but some are not (such as the chemical mechanism). However, for relatively
small changes in total SO2 emissions (from all sources), the chemistry (and the overall model
results) can be approximated as a linear process, which is the approach EPA has taken here...
For all impacts that play a significant part in EPA's analysis, the regressions appear to have been
developed with a reasonable, although small, set of practical data points.
In sum, due to the observed linearity between modeled incremental extinction and SO2
emission rates, and the fact that the overall modeled extinction levels (from all sources) weren't
very different between the low, high, and 2018 baseline scenarios, the resulting regression slopes
provide a reasonable estimation of the change in extinction at each Class I area that would result
from a change in SO2 emission rate from units at each facility."
Response: We disagree with the comments from Luminant and CCP that the methodology used
to estimate visibility benefits from control level emissions was unjustified or unreasonable, and
agree with Earthjustice that our approach was reasonable. The linear relationship we developed
to extrapolate extinction due to controlled emission rates was a reasonable approach in our
technical analysis.
The commenter is incorrect in suggesting that we developed a linear relationship between
emissions and deciviews and then commenting that this is flawed because the relationship
between light extinction and deciviews is exponential. We developed a linear relationship
between emissions and light extinction (inverse Megameters), not deciviews.
We agree with the commenters, that in general, the relationship between downwind
concentrations and emissions can be complicated and non-linear due to complex chemistry,
including the fact that reductions in sulfur emissions can result in an increase in ammonium
nitrate. Each modeled emission scenario took this complex chemistry into account in estimating
the visibility impacts for that scenario. We estimated control efficiencies for a high and low
control case scenario that would span the range and give a reasonable approximation of emission
reductions of potential controls and maximize the number of data points available to estimate the
visibility benefit due to a reduction in emissions.608. Using the unit level High and Low modeled
visibility impacts and the 2018 facility level modeling described in the FIP TSD, we examined
the relationship between the various levels of emissions from a modeled site and the modeled
visibility impact at each Class I area. For each facility and Class I area, the available modeled
data was linear with high correlation and the modeled emission levels were relatively close to the
estimated control levels examined. Therefore we used the linear fit to extrapolate the anticipated
visibility impact/benefit from a given level of emission/control.609 We agree with Luminant
608 See FIP TSD at A-54 for a more detailed description
609 See Vis modeling summary.xlsx in the docket for this action for our calculations and estimates of visibility
benefits from the examined levels of controls.
759
-------
and CCP that small perturbations relative to the model inputs can be approximated as linear.
However, we disagree with the commenters that we extended the linear treatment to large
variations, and we note errors as discussed below in the commenter's assessment of the
differences between modeled and required control levels. The variations between the modeled
High control levels and the control levels required in the FIP are relatively small. We agree with
Earthjustice et al. that the small level of uncertainty in the visibility benefit from these controls
introduced by the linear extrapolation does not impact the overall conclusions. Luminant and
CCP are incorrect in stating that the proposed emissions are a factor of 2 to 6 smaller than the
emissions used in the CAMx modeling scenario with the highest controls. The table below
summarizes the modeled emission levels and compares them to the baseline actual emissions and
the controlled emission levels that were used to extrapolated visibility impacts and benefits. The
modeled high control scenario emissions for units with proposed scrubber retrofits range from
1,110 to 1,675 tpy compared to a range of 521 to 700 tpy for the proposed control emissions.
The absolute difference between the modeled and control emission rates at individual facilities
with required scrubber retrofits ranges from 84 tpy to 1,062 tpy, or a factor of 1.08 to 2.73
smaller. The absolute difference between the modeled and control emission rates at individual
facilities with proposed scrubber upgrades ranges from -30 tpy to 535 tpy, or a factor of 1.6 or
less. The largest difference of 1,062 tpy (a factor of 2.73 difference) for units with required
scrubber retrofits is for Big Brown unit 1 and amounts to small percentage difference (-3%) in
reduced emissions going from 30,667 tpy uncontrolled emissions down to 1676 tpy (High
control case) or 614 tpy (required control level). Using the linear relationship we developed,
this also equates to a 3% difference in the reduction of visibility impact (extinction) or visibility
benefit going from the baseline level to the control level. This is a small perturbation from the
modeled levels, a small difference in estimated extinction benefit from the modeled and required
control level, and does not impact our overall decisions on the significance of visibility benefits
from the required controls. In every case, the required control level emissions are the same or
less than the high control level modeled, and the visibility benefits from controls at the required
control level will be the same or more than those modeled at the High control level. Therefore,
the High level modeled visibility benefit can be seen as a lower bound and even these support
our decision.
The table below summarizes the modeled emission levels and compares them to the baseline
actual emissions and the controlled emission levels that we extrapolated visibility impacts for.
Difference
between
high
control
modeled
3yr
EPA/
emissions
average
CENRAP
high control
and
2009-2013
2018
FIP
modeled to
control
excluding
Phase 1
low
High
control
controlled
level
Unit
max and
modeled
controlled
control
emission
emissions
emissions
#
Facility
min
emissions
emissions
emissions
level
ratio
(tpy)
1
Big
Brown
30667
23328
20108
1676
614
2.73
1062
760
-------
2
Big
Brown
30814
23831
13343
1668
646
2.58
1022
1
Coleto
Creek
16059
16225
9839
1492
699
2.14
794
liml
Limestone
10913
12817
7423
2474
2467
1.00
8
lim2
Limestone
11946
5023
5230
2615
2615
1.00
0
1
Martin
Lake
24495
11442
19278
3856
3706
1.04
150
2
Martin
Lake
21580
12080
11652
3884
3663
1.06
221
3
Martin
Lake
19940
12495
7444
3722
3551
1.05
171
1
Monticello
17865
19298
13686
1355
537
2.52
818
2
Monticello
16429
19853
9201
1346
522
2.58
824
3
Monticello
13857
11978
3531
1851
1571
1.18
281
4
Sandow 4
22289
8477
22978
4596
4625
0.99
-30
171b
Tolk
10031
11584
7450
1209
836
1.45
373
172b
Tolk
11034
10549
4520
1103
1018
1.08
85
5
WA
Parish
14157
3763
10944
1397
708
1.97
689
6
WA
Parish
15307
3840
7423
1419
704
2.02
716
7
WA
Parish
12335
3324
8108
1244
602
2.07
642
8
WA
Parish
2586
4548
1790
1371
836
1.64
535
1
Welsh
8084
1236
5893
1110
610
1.82
500
2
Welsh
8256
1233
3974
1117
647
1.73
470
3
Welsh
8609
11815
5014
1124
650
1.73
474
15.u Natural Background
Comment: EPA's use of "natural background" is inconsistent with its regulations and
arbitrary and capricious [Luminant (0061) p. 119]
Luminant stated that EPA's use of an artificial "natural conditions" background by which to
judge the Texas and Oklahoma RPGs for 2018, instead of the projected actual conditions in 2018
that CENRAP and ENVIRON modeled, is contrary to the regulations, inconsistent with agency
precedent, and arbitrary and capricious. The relevant issue here for the reasonable progress
analysis is what will be the visibility in deciviews at the three Class 1 areas of interest in 2018.
That issue is not addressed by looking at visibility changes against an artificial "natural
conditions" background which will not exist in 2018, without reference to or attribution of other
actual emissions of precursors to visibility impairment and the chemical interaction of those
emissions within the atmosphere. EPA concedes as much.740 EPA admits that "[t]he deciview
improvement based on the 2018 background conditions provides an estimate of the amount of
benefit that can be anticipated in 2018 and the impact a control/emission reduciton [sic] may
have on the established RPG for 2018."741 Yet, in its individual unit assessment, EPA does not
761
-------
appear to consider the actual deciview improvement in 2018 but instead only a theoretical
improvement against artificial conditions that do not exist and can never exist given the
uncontrollable impact of international and other emissions. EPA's analysis thus does not address
the relevant legal issue and is not rationally connected to EPA's final decision (i.e., what is a
reasonable progress goal for 2018). And, as shown in Tables 6 and 7, EPA's use of a "natural"
background artificially overstates the estimated visibility benefit from the individual controls that
EPA proposes.
Luminant asserted that EPA's use of a "natural background" also fails to take into account the
substantial amount of SO2 emissions from international sources that Texas has no authority to
address. EPA has found that "the projected emissions from international sources will in some
cases affect the ability of States to meet reasonable progress goals."742 Thus, EPA specifically
instructed that "EPA does not expect States to restrict emissions from domestic sources to offset
the impacts of international transport of pollution."743 Yet, here, EPA would do just that. EPA
decides what domestic sources to regulate based on an analysis that assumes the elimination of
all man-made sources of pollution from Mexico in 2018—a feat that Texas cannot legally
achieve and that EPA does not believe can be attained even by 2064. And this would be no
simple feat, given that "52 percent of the impairment at Big Bend and 25 percent of the
impairment at Guadalupe Mountains is from Mexico and further south."744 Because EPA's
natural background assessment ignores these substantial uncontrollable emissions, it is arbitrary
and capricious and cannot support EPA's proposal.
Moreover, Luminant noted that EPA's reason for adopting the deciview as the required metric
for tracking reasonable progress is the exact opposite of the reason it advances here to justify the
use of a "natural background" by which to judge Texas sources. The very reason EPA adopted
the deciview in its regulations is that it "provid[es] a scale that relates visibility to perception"
and "expresses uniform changes in haziness in terms of common increments across the entire
range of visibility conditions, from pristine to extremely hazy conditions "745 Thus, in adopting
the deciview in its regulations, EPA has already rejected the notion it puts forward here—that
regional haze decision-making should be based on an artificial pristine background in order to
ultimately achieve the national goal. The national goal, EPA has previously found, is furthered
by "the use of the deciview metric as calculated from ambient monitoring data for tracking
changes in regional visibility."746
According to Luminant, whatever theoretical shortcoming that EPA now perceives in the
deciview as the metric forjudging visibility impairment and reasonable progress (based on the
fact that it reflects the "nonlinear nature of visibility impairment"747), EPA's own regulations
dictate that the deciview is the required metric for determining reasonable progress.748 EPA may
not effectively amend those regulations by this action. EPA's use of "natural conditions"
background is disconnected from both reality and the regulatory issue at hand—i.e., what will be
the visibility at these areas in 2018 measured in deciviews—and thus it is unlawful.749
Luminant stated that EPA's attempt to use "natural conditions" here is in no way validated by its
prior action on the North Dakota SIP, as EPA wrongly contends.750 There, North Dakota chose to
use a "hybrid cumulative modeling approach" different from the modeling developed by its
regional planning organization ("RPO")751 No other state in North Dakota's RPO "opted to
762
-------
develop its own reasonable progress modeling methodology," and EPA concluded that North
Dakota's hybrid approach did not satisfy the regulatory "criteria for the use of alternative models
"752 xhe fact that EPA's disapproval of North Dakota's hybrid approach was affirmed by the
Eighth Circuit753 does not justify EPA's use of "clean background" conditions by which to judge
Texas's SIP. Texas did not use a unique or hybrid approach to assessing reasonable progress.
Texas used the standard approach used by all CENRAP states. In fact, it is EPA that is using an
unorthodox and hybrid approach in this instance, mixing modeling parameters, without
demonstrating that its approach meets the criteria for alternative models. EPA, like North
Dakota, is "not free" "to employ its own visibility model and to consider visibility improvement
in its reasonable progress determinations" "in a manner that [is] inconsistent with the CAA," as
EPA has done here.754 EPA must follow its own regulations and guidance, just as EPA expects
state and regulated entities to do.
Footnotes:
740 FIP TSD at A-37.
741 Id. at A-39.
742 64 Fed. Reg. at 35,736.
743 Id.
744 2009 Texas SIP Narrative at 10-10.
745 64 Fed. Reg. at 35,727 (emphasis added).
746 Id. (emphasis added).
747FIP TSD at A-38 to A-39.
748 40 C.F.R. § 51.308(d)(1).
749 See Motor Vehicle Mfrs. Ass'n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (holding that an
agency decision is unlawful if the agency has failed to demonstrate that it has "examine [d] the relevant data and
articulate [d] a satisfactory explanation for its action including a rational connection between the facts found and the
choice made" (internal quotations omitted)).
750 FIP TSD at A-39.
751 76 Fed. Reg. at 58,624, 58,627.
752 Id. at 58,624.
753 North Dakota, 730 F.3d at 766.
754 Id. (emphasis added).
Luminant: Step 11: Instead of basing its decision on the modeling provided to EPA by
ENVIRON, EPA adjusts the results of ENVIRON's modeling to "estimate[]" "the
visibility benefit. . . based on natural [] conditions," instead of ENVIRON's modeled
conditions in 2018.224
[Luminant (0061) p. 118] Luminant noted, while EPA extolls the virtues of CAMx, it arbitrarily
mixes modeling approaches in its analysis, making the results unreliable and nonsensical. EPA
goes to great lengths to note the differences between CAMx modeling (which it performed) and
CALPUFF modeling (which it did not), yet it arbitrarily mixes modeling standards to reach a
contrived result. For example, as EPA explains, "CAMx is a full photochemical model with all
the other sources quantified and added to the modeling," whereas CALPUFF modeling "is
conducted to determine a facility's impact on a Class I area with no consideration of other
pollutants in the air (other than EPA's estimate for natural background conditions) . . . ,"737 But
instead of using the results of the CAMx modeling that EPA directed ENVIRON to conduct,
EPA attempts to translate those results to a "natural background" (as CALPUFF would have
used). That is, EPA adjusted the results of the CAMx modeling from ENVIRON to "estimate[]"
"the visibility benefit. . . based on natural . . . conditions," which it said "is needed to assess the
763
-------
full benefit from potential controls."738 But there is absolutely no precedent for mixing the
modeling approaches, and EPA cites none. Having selected CAMx for its modeling—which
necessarily includes all sources in the modeling domain and which would more closely reflect
actual benefit—EPA cannot then complain that a different "background" is appropriate and
adjust the CAMx results on an ad hoc basis. EPA provides zero support for this novel approach.
If EPA had wanted to review projected benefits against a "clean" background, it should have
employed CALPUFF modeling for its impacts determination, as it has done in all other
reasonable progress assessments to date. Having chosen to use CAMx modeling instead, EPA
should judge the results of the modeling as they were calculated—not attempt to adjust them to
create an artificial justification for its proposal. 739
Footnotes:
737 FIP TSD at A-37.
738 7 9 Fed. Reg. at 74,881.
739 Further, in directing ENVIRON to use a newer version of CAMx than CENRAP used, EPA violated its own
guidance that "the better course is to rely on modeling based on the same version of the model that the State
employed to ensure we are using a consistent comparison." 77 Fed. Reg. at 20,908 (citing Mont. Sulphur & Chem.
Co. v. EPA, 2012 U.S. App. LEXIS 1056 (9th Cir. Jan. 19, 2012)).
[UARG (0065) p. 21] As part of their argument that EPA's proposed FIPs for Texas and
Oklahoma are unlawful, UARG stated that the EPA's assessment is irrationally based on
visibility conditions modeled against "natural conditions." EPA acknowledges in its FIP TSD
that visibility improvement "based on the 2018 background conditions provides an estimate of
the amount of benefit that can be anticipated in 2018 and the impact a control/emission reduciton
[sic] may have on the established RPG for 2018." EPA, Technical Support Document for the
Oklahoma and Texas Regional Haze Federal Implementation Plans (FIP TSD) at A-39 (Nov.
2014), Doc. ID No. EPA-R06-OAR-2014-0754 0007 ("FIP TSD"). Instead of considering real-
world visibility impacts, however, EPA relies on a hypothetical visibility benefit measured
against pristine conditions that will not exist. Indeed, EPA's approach entirely ignores visibility
impairment due to emissions from non-U.S. sources, over which Texas has no control. The
effects of those emissions are significant. Texas concluded that they account for "52 percent of
the impairment at Big Bend and 25 percent of the impairment at Guadalupe Mountains." 2009
Texas SIP at 10-10. Because EPA's assessment is based on unsupportable assumptions that
distort its analysis of reasonable progress, its proposed FIP is arbitrary and capricious.
Luminant and UARG comment that the use of an artificial "natural conditions" background by
which to judge the Texas and Oklahoma RPGs for 2018, instead of the projected actual
conditions in 2018 that CENRAP and ENVIRON modeled, is contrary to the regulations,
inconsistent with agency precedent, and arbitrary and capricious. The commenter states that
relevant issue for reasonable progress is visibility in deciviews at the three Class I areas in 2018
and is not addressed by looking at visibility changes against artificial "natural conditions" which
will not exist in 2018 and do not consider other emissions and chemical interaction of those
emissions, as we discuss in the FIP TSD. Luminant states that the analysis does not address the
relevant legal issue and is not rationally connected to the final decision (i.e., what is a reasonable
progress goal for 2018). The commenter also states that the use of natural background overstates
the estimated visibility benefit from the proposed controls.
The commenters continues that we provide an estimate of the deciview improvement from
controls in 2018 and the impact controls have on the RPG for 2018, but consider only the
764
-------
deciview improvement on a natural background that do not and will not exist given
uncontrollable impacts of international and other emissions. Luminant asserts that this is
inconsistent with the Regional Haze Rule that states that "the projected emissions from
international sources will in some cases affect the ability of States to meet reasonable progress
goals." And "EPA does not expect States to restrict emissions from domestic sources to offset
the impacts of international transport of pollution."610 Luminant claims we decided which
sources to control based on analysis that assumes elimination of all anthropogenic emissions in
Mexico, a significant portion of the visibility impairment, and is therefore arbitrary and
capricious and cannot support our proposal.
Luminant also states that the use of "natural background" is inconsistent with our adoption of the
deciview as the required metric for tracking reasonable progress. They state " In adopting the
deciview in its regulations, EPA has already rejected the notion it puts forward here—that
regional haze decision-making should be based on an artificial pristine background in order to
ultimately achieve the national goal. The national goal, EPA has previously found, is furthered
by 'the use of the deciview metric as calculated from ambient monitoring data for tracking
changes in regional visibility.'611" The deciview is the required metric for determining
reasonable progress and we may not effectively amend those regulations by this action. The use
of "natural conditions" background is disconnected from both reality and the regulatory issue at
hand—i.e., what will be the visibility at these areas in 2018 measured in deciviews—and thus it
is unlawful. Luminant cites to Motor Vehicle Mfrs. Ass'n v. State Farm Mut. Auto. Ins. Co., 463
U.S. 29, 43 (1983) (holding that an agency decision is unlawful if the agency has failed to
demonstrate that it has "examine[d] the relevant data and articulate[d] a satisfactory explanation
for its action including a rational connection between the facts found and the choice made"
(internal quotations omitted)).
Luminant also comments that the use of natural background is not validated by the prior action in
the North Dakota SIP, because North Dakota used a "hybrid cumulative modeling approach" and
we found that this approach did not satisfy the regulatory "criteria for the use of alternative
models.. ,"612 Luminant claims that the disapproval of this hybrid approach was affirmed by the
Eight Circuit does not justify the use of "clean background" conditions to judge Texas SIP as
Texas did not use a unique or hybrid approach. Luminant asserts that we are using an
unorthodox and hybrid approach without demonstrating that it meets the criteria for alternative
models and we, like North Dakota, are "not free" "to employ its own visibility model and to
consider visibility improvement in its reasonable progress determinations" "in a manner that [is]
inconsistent with the CAA," as we have done here.613 We must follow our own regulations and
guidance, just as we expects state and regulated entities to do.
Response: We disagree with the commenter that the use of "natural conditions" is contrary to
the regulations, inconsistent with agency precedent, and arbitrary and capricious. We disagree
with the commenter that the analysis does not address the relevant legal issue, is unlawful and is
not rationally connected to the final decision (i.e., as defined by the commenter as what is a
610 64 Fed. Reg. at 35,736
611 64 Fed. Reg. at 35,727
612 76 Fed. Reg. at 58,624, 58,627
613 North Dakota v. EPA, 730 F.3d 750, 766 (8th Cir. 2013)
765
-------
reasonable progress goal for 2018). We have provided in the proposal, the TSDs, the final
action notice, and in the RTC document an examination of all the relevant data, thorough
explanations for our final actions and choices, and included rationales describing the connections
between the facts and our choices.
We disagree with the commenter that our use of the "natural background" metric is contrary to
regulations. As we discuss in a separate response to comment concerning the legality of the
deciview, extinction and percent extinction metrics, the commenter fails to distinguish between
the required metric used to describe overall visibility conditions at a Class I area at a given point
in time and the range of metrics, such as deciview impairment calculated based on natural
background conditions and percent of total extinction, that can be used to describe the visibility
impairment due to an individual source, group of sources, a state's sources, or some other
contribution to the visibility impairment at a Class I area. As explained below, it is necessary to
consider the visibility benefit of controls on a "clean" background basis to assess the full benefit
from potential controls.
The Regional Haze Rule requires that we identify reasonable controls based on consideration of
the four statutory factors and establish a reasonable progress goal that reflects the anticipated
amount of visibility improvement from implementation of those controls in additional to all other
"on the books" controls. Specifically, Section 51.308(d)(l)(i)(A) requires consideration of the
four factors and a demonstration of how these factors were taken into consideration in selecting
the goal. We analyzed the time necessary for compliance, energy and non-air environmental
impacts, the remaining useful life, and the costs of compliance including consideration of the
anticipated visibility benefits of specific controls on individual units. As discussed in depth
below, in considering the anticipated visibility benefits from individual controls, it was
appropriate to consider estimated benefits on a "clean" or "natural" background.
In the FIP TSD, we discuss the need to estimate visibility benefits using both a "clean" and
"dirty" background.614
The deciview improvement based on the 2018 background conditions provides an
estimate of the amount of benefit that can be anticipated in 2018 and the impact a
control/emission reduction may have on the established RPG [reasonable progress
goal] for 2018. However, this estimate based on degraded or "dirty" background
conditions underestimates the visibility improvement that would be realized for
the control options under consideration. Because of the non-linear nature of the
deciview metric, as a Class I area becomes more polluted the visibility
impairment from an individual source in terms of deciviews becomes
geometrically less. Results based solely on a degraded background, will rarely if
ever demonstrate an appreciable effect on incremental visibility improvement in a
given area. Rather than providing for incremental improvements towards the goal
of natural visibility, degraded background results will serve to instead maintain
those current degraded conditions. Therefore, the visibility benefit estimated
based on natural or "clean" conditions is needed to assess the full benefit from
potential controls.
614 FIP TSD at A-39
766
-------
In considering the visibility benefits, we considered deciview improvements based on "clean"
and "dirty" background as well as the reduction in extinction and percent extinction. We
disagree with the comment that this approach is arbitrary and capricious because the "clean"
background analysis assumes elimination of all anthropogenic emissions in Mexico. By
definition, the "clean" background analysis using natural conditions eliminates the impact from
all other anthropogenic sources, domestic and international. . It is not reasonable to assess the
visibility benefit of controls, the value of installing a control in the immediate future that will
permanently reduce visibility impacts from a source, only in a manner that is dependent on the
current level of emissions or impact from other sources or other countries. For example, in
considering only the estimated visibility benefit from controlling Big Brown using a "dirty"
background, an increase in visibility impacts from Mexico emissions or emissions from another
Texas point source would result in a decrease in the calculated visibility benefit in deciviews
from installing controls on Big Brown, making controls appear less beneficial. By using a metric
that is independent of all other emission sources ("clean"), we avoid this paradox that the dirtier
the existing air, the less likely it would be that any control is required. This was also explained
in the preamble to the final Regional Haze Rule and Guidelines for BART Determinations.615
The use of "clean" background is consistent with our regulations, and necessary to assess the full
potential benefit from controls and does not overstate the visibility benefit. We also discuss the
consideration of deciview benefits and impacts based on "clean" and "dirty" background
conditions in our response to comments concerning our analysis of cost versus visibility
elsewhere in this document. We address the footnote about Montana Sulfur elsewhere in this
document.
Our use of "clean" background is also consistent with the methodology used by Texas for its
BART visibility analysis, which also relied on CAMx photochemical modeling with source
apportionment. The TCEQ utilized this approach in assessing the visibility impacts from
individual sources and groups of sources to determine their significance for BART screening.
As detailed in the screening analysis protocol developed by TCEQ and reviewed by us, "The
source's HI [haze index] is compared to natural conditions to assess the significance of the
source's visibility impact. EPA guidance lists natural conditions (bnatural) by Class I area in
terms of Mm-1 (EPA, 2003b) and assumes clean conditions with no anthropogenic or weather
interference. The visibility significance metric for evaluating BART sources is the change in
deciview (del-dv) from the source's and natural conditions haze indices'"616
615 Using existing conditions as the baseline for single source visibility impact determinations would create the
following paradox: the dirtier the existing air, the less likely it would be that any control is required. This is true
because of the nonlinear nature of visibility impairment. In other words, as a Class I area becomes more polluted,
any individual source's contribution to changes in impairment becomes geometrically less. Therefore the more
polluted the Class I area would become, the less control would seem to be needed from an individual source. We
agree that this kind of calculation would essentially raise the "cause or contribute" applicability threshold to a level
that would never allow enough emission control to significantly improve visibility. Such a reading would render the
visibility provisions meaningless, as EPA and the States would be prevented from assuring "reasonable progress"
and fulfilling the statutorily-defined goals of the visibility program. Conversely, measuring improvement against
clean conditions would ensure reasonable progress toward those clean conditions. 70 FR 39124
616 TX RH SIP, Appendix 9-5, "Screening Analysis of Potential BART-Eligible Sources in Texas" at 2-11, emphasis
added.
767
-------
Luminant asserts that use of natural visibility conditions is inconsistent with the Regional Haze
Rule that states, "the projected emissions from international sources will in some cases affect the
ability of States to meet reasonable progress goals." And "EPA does not expect States to restrict
emissions from domestic sources to offset the impacts of international transport of pollution."617
We disagree with the commenter and believe this approach using a "clean" background is in fact
consistent with the Regional Haze Rule. As we discuss in a separate response to comment
concerning the legality of the extinction and percent extinction metrics, the commenter confuses
the analysis of visibility benefits from specific controls discussed above and the overall
assessment of progress towards meeting the RPG or URP and does not provide the proper
context for this comment. The Regional Haze Rule States:
The EPA agrees that the projected emissions from international sources will in
some cases affect the ability of States to meet reasonable progress goals. The EPA
does not expect States to restrict emissions from domestic sources to offset the
impacts of international transport of pollution. We believe that States should
evaluate the impacts of current and projected emissions from international sources
in their regional haze programs, particularly in cases where it has already been
well documented that such sources are important. At the same time, EPA will
work with the governments of Canada and Mexico to seek cooperative solutions
on transboundary pollution problems. 64 Fed. Reg. 35714, 35736 (July 1, 1999).
Consistent with this, we agreed with Texas that it was not reasonable to meet the uniform rate of
progress at Big Bend and Guadalupe Mountains, in part due to the impact from sources outside
of the United States. Based on the CENRAP modeling, Texas determined and we agree that the
level of impact from Mexican sources was significant. Using natural visibility conditions to
assess the visibility benefit of a specific control in Texas is not inconsistent with the recognition
that impacts from Mexico are a significant portion of the total visibility impairment, and in no
way requires reductions to offset impacts from Mexico. The established reasonable progress
goals include consideration of the impact from Mexico and other source regions, and also
includes consideration of the visibility benefits of identified reasonable controls.
As discussed above, in order to fully assess the potential visibility benefit of controls, an analysis
utilizing natural background conditions is necessary. We discuss consideration of international
emissions in more depth in a separate section of this document.
We also disagree with the commenter and find that the use of natural background is supported by
our previous action on North Dakota's regional haze SIP and the associated Eighth Circuit Court
decision. The commenter does not present the full context of our rationale in disapproving North
Dakota's assessment of visibility benefit of RP controls. The full text of that determination is:
In addition to evaluating the four statutory factors, North Dakota also considered
the visibility impacts associated with the control options for each RP source.
However, in modeling visibility impacts, North Dakota used a hybrid cumulative
modeling approach that is inappropriate for determining the visibility impact for
individual sources. As with the modeling North Dakota conducted for its NOX
617 64 Fed. Reg. at 35,736
768
-------
BART analysis for MRYS [Milton R. Young Station] Units 1 and 2 and LOS
[Leland Olds Station] Unit 2, the approach fails to compare single- source
impacts to natural background. While there is no requirement that States, when
performing RP analyses, follow the modeling procedures set out in the BART
guidelines, or that they consider visibility impacts at all, we find that North
Dakota's visibility modeling significantly understates the visibility improvement
that would be realizedfor the control options under consideration. Accordingly,
we are disregarding the modeling analysis that North Dakota has used to support
its RP determinations for individual sources.618
We disapproved the reasonable progress visibility modeling performed by North Dakota
specifically because it failed to use a natural background approach and therefore understated the
visibility benefit from potential controls. While we note that it does not appear that the modeling
approach satisfied the Appendix W criteria for the use of alternate models, that is in the context
of the overall visibility projection modeling that North Dakota developed using a
CALPUFF/CMAQ hybrid model which used obsolete settings. Failure to satisfy Appendix W
was not raised in the context of the individual source visibility assessment using CALPUFF. The
Eighth Circuit Court's decision affirmed our position that the use of degraded, or dirty
background for individual source visibility assessment, was not consistent with the Clean Air
Act.619 The relevant section of the Eighth Circuit Court's decision on this point reads:
Although the State was free to employ its own visibility model and to consider
visibility improvement in its reasonable progress determinations, it was not free to
do so in a manner that was inconsistent with the CAA. Because the goal of §
169A is to attain natural visibility conditions in mandatory Class I Federal areas,
see 42 U.S.C. § 7491(a)(1), and EPA has demonstrated that the visibility model
used by the State would serve instead to maintain current degraded conditions, we
cannot say that EPA acted in a manner that was arbitrary, capricious, or an abuse
of discretion by disapproving the State's reasonable progress determination based
upon its cumulative source visibility modeling.620
The use of natural background conditions to assess visibility benefits of individual controls, as
we have done here in this action, is consistent with the goals of the CAA. Furthermore, use of
natural visibility background is not a unique or unorthodox approach and the appropriate choice
of metric is not relevant to a demonstration meeting the criteria for alternative models. Using
CAMx modeling consistent with EPA, Texas and CENRAP modeling protocols does not
constitute an "alternative model." Furthermore, in evaluating impacts from individual sources or
groups of sources, the use of a natural background and PSAT is consistent with the protocol
developed by TCEQ and reviewed by us and utilized in their BART screening modeling.621
Additional comments concerning deviations from Appendix W are addressed elsewhere.
618 76 FR 58627 (September 21, 2011) emphasis added
619 North Dakota v. EPA, 730 F.3d 750, 766 (8th Cir. 2013)
620 North Dakota v. EPA, 730 F.3d 750, 766 (8th Cir. 2013)
621 TX RH SIP, Appendix 9-5, "Screening Analysis of Potential BART-Eligible Sources in Texas" at 2-11: "The
source's HI is compared to natural conditions to assess the significance of the source's visibility impact. EPA
769
-------
Finally, we disagree with the commenter that in directing ENVIRON to use a newer version of
CAMx than CENRAP used, EPA violated its own guidance that "the better course is to rely on
modeling based on the same version of the model that the State employed to ensure we are using
a consistent comparison." 77 Fed. Reg. at 20,908 (citing Mont. Sulphur & Chem. Co. v. EPA,
2012 U.S. App. LEXIS 1056 (9th Cir. Jan. 19, 2012)). This is covered in more detail in a
response elsewhere in this document.
Comment: EPA's analysis of visibility benefit of individual SO2 controls is not a lawful
basis for its disapprovals or FIPs [Luminant (0061) p. 106]
Luminant asserted that EPA's use of visibility modeling of specific control measures at
individual sources as the lynchpin of its methodology is unlawful and contrary to the statute.
Texas was not required to conduct such an analysis to support its reasonable progress goals and
long-term strategy, and EPA may not use such an analysis to disapprove Texas's SIP. EPA
explains that "[p]rior to doing the control cost evaluations discussed in the sections above,"
EPA conducted CAMx modeling runs "to evaluate the benefits of the proposed controls and their
associated emission decreases on visibility impairment values."669 EPA uses projected or
estimated visibility benefit at several steps in its process to either include or exclude sources
from regulation.670 EPA's approach is random and novel. At some steps, EPA looks to visibility
impact from the facility as a whole while at other steps looks to visibility benefit estimated from
certain emission reductions at individual units. EPA contracts ENVIRON to perform visibility
modeling, but ultimately does not rely on the modeling resulting for its conclusions—choosing
instead to rely on its own calculated and inflated predictions of "benefit."
Footnotes:
669 79 Fed. Reg. at 74,877-78 (emphasis added).
670 See supra pages 30-43.
Response: We disagree with the comment. As discussed in depth elsewhere, the disapproval of
the TX RH SIP was based on our evaluation of the analysis in the SIP. Additional analysis
including modeling of specific controls at individual sources was necessary to determine if the
noted flaws in the SIP affected the ultimate determination in the SIP and to inform the FIP. The
control scenario modeling performed by ENVIRON for EPA was conducted at the same time as
the analysis of potential controls and controlled emission rates. Because the exact controlled
emission rates were not available at the time, we modeled a High and Low control level. As
discussed in depth in a response to comment above, the modeling results provided estimates of
visibility benefits at these estimated controlled rates, representative of the level of control
anticipated due to use of scrubbers and DSI. These modeled visibility benefits then had to be
adjusted to reflect the final determination of the proposed controlled emissions based on results
of the analysis of the potential controls and control performance. After adjusting the modeled
results to reflect the actual estimated controlled emission rates, the estimated visibility results
guidance lists natural conditions (bnatural) by Class I area in terms of Mm-1 (EPA, 2003b) and assumes clean
conditions with no anthropogenic or weather interference. The visibility significance metric for evaluating BART
sources is the change in deciview (del-dv) from the source's and natural conditions haze indices"
770
-------
were evaluated using a number of metrics, including visibility benefit at both 2018 degraded
background conditions and annual average natural conditions. As discussed in the comment
above, we disagree with the comment that this results in "inflated" predictions of benefit.
We disagree with the commenter and believe that our approach is reasonable and consistent with
the Regional Haze Rule. The RHR requires that we identify reasonable controls based on the
four statutory factors and establish reasonable progress goals that reflect the anticipated amount
of visibility improvement from implementation of those controls in additional to all other "on the
books" controls. After an initial Q/d analysis, we modeled emissions from 38 facilities with the
greatest potential visibility impacts in order to further evaluate the potential of emissions from
these sources to impact visibility. This is a more refined approach than the initial Q/d analysis
performed by both Texas and EPA because unlike a Q/d analysis that only considers emissions
and distance, this accounts for emissions, location, stack parameters, meteorological conditions,
and models both chemistry and transport to the Class I areas. The results of this modeling were
used to further eliminate sources from undergoing the four-factor analysis based on consideration
of facility-level impacts and estimated unit level impacts, as described in detail in the FIP TSD
and a separate response to comment. We then considered the four factors, including visibility
benefit of controls on those individual units identified through this process as having largest
visibility impacts. In evaluating the visibility benefits from controls, we considered deciview
visibility improvements, as well as reductions in extinction and percent extinction.
Comment: Luminant provided a summary of EPA's "additional" visibility benefit analysis
of a "small group' of Texas sources. TLuminant (0061) p. 271
Step 10: After further altering the modeling results that ENVIRON previously provided,
EPA hires ENVIRON a second time to conduct "further visibility modeling" using
CAMx for the 21 remaining units. ENVIRON models the improvement in visibility in
2018 at the three Class I areas from various SO2 emission control scenarios that EPA
directs ENVIRON to include in the model.
Step 11: Instead of basing its decision on the modeling provided to EPA by ENVIRON,
EPA adjusts the results of ENVIRON's modeling to "estimate[]" "the visibility benefit. .
. based on natural [] conditions," instead of ENVIRON's modeled conditions in 2018.224
In clear contradiction to the CAMx User's Guide, EPA linearly scales the results of the
CAMx modeling from ENVIRON to estimate the visibility benefit associated with larger
changes in emission rates at the 21 remaining units.
Step 12: Based on its estimates of visibility benefits, EPA excludes 7 of 21 units from
regulation and establishes emission limitations for the remaining 14 units.225
At Step 10. after altering ENVIRON's first modeling results to reach its "final" list of target
sources, EPA again hired ENVIRON (which it excluded from Steps 3-9) to conduct "further
visibility modeling" using CAMx for the 21 remaining units on EPA's list. EPA explained that
its "final modeling analysis" was "used to evaluate the benefits of the proposed controls and their
associated emission decreases on visibility impairment values."285 However, ENVIRON did not
771
-------
model the units with specific controls identified through consideration of the reasonable progress
four factors; instead, the visibility modeling was performed "[p]rior to" EPA's analysis of the
costs of controls and other statutory factors.286 Indeed, the controls EPA modeled did not even
match the controls it would later propose—EPA "formulated model runs that would span the
range of potential controls/emissions we planned to examine."287 Using this approach,
"ENVIRON assisted EPA in conducting two additional [model] runs," which "used a general
estimate for what low controls and high control might achieve."288 In other words, EPA simply
"provided [] ENVIRON" two estimates of SO2 emissions for each of the remaining 21 units—
one so-called "low control" value and one so-called "high control" value.289
These two additional model runs misrepresented the Luminant facilities, and potentially all
facilities, where additional scrubber controls were evaluated in ways that overstated the visibility
benefit from the proposed controls. For example, after adding new scrubbers or upgrading the
existing scrubbers at each unit, the stack temperatures and velocities will be considerably lower
than they would be prior to the addition or upgrade. The typical stack temperatures for a coal-
fueled boiler with and without a wet scrubber are 180°F and 350°F, respectively. Lower stack
temperatures result in less dispersion of air emissions and thus less visibility impact at large
distances. In air dispersion modeling for other purposes (e.g., a Prevention of Significant
Deterioration permit amendment), EPA would surely not allow the use of the same stack
parameters for the source before and after a scrubber installation project, as EPA did here. EPA
provides no basis for its inconsistent approach here.
ENVIRON's additional CAMx modeling "focused on calculating the extinction and visibility
impacts and benefits" at the three Class I areas from EPA's specified SO2 reductions.290 "In
evaluating the impacts and benefits of potential controls," EPA "utilized a number of metrics."291
EPA explained that by using CAMx, which "include[d] modeling all emissions in the modeling
domain, the model results are inherently a degraded background analysis and the results are
impacted by emissions from other sources."292 Thus, EPA explained that the CAMx results
"provide[] an estimate of the amount of benefit that can be anticipated in 2018 and the impact a
control may have on the established RPG for 2018."293 Table 4 shows the results of ENVIRON's
CAMx modeling for Luminant's units under the "high" SO2 controls scenario established by
EPA, in terms of deciview improvement in 2018 at each Class I area. As shown in Table 4 to
comment 0061 provided by Luminant (not reproduced here), the most improvement that EPA's
controls would produce for any Luminant unit at any Class I area is miniscule— 0.0678 deciview
for Big Brown Unit 2 at Wichita Mountains.
Instead of relying on these modeling results provided by ENVIRON, which show miniscule
projected benefits, EPA adjusted the ENVIRON results by "updating] the baseline uncontrolled
emission for each unit based on [SO2 emissions] data for 2009-2013."295 EPA's adjustment
substitutes higher "scaled" emission numbers that artificially increase projected visibility
improvement from the proposed controls. Luminant provided a table (Table 5 to comment 0061
[not reproduced here]) that shows the results of EPA's adjustments to ENVIRON's CAMx
modeling for Luminant's units under the "high" SO2 control scenario established by EPA, in
terms of deciview improvement in 2018 at each Class I area. But as Table 5 shows, even the
"scaled" improvements are all less than a tenth of a deciview.
772
-------
Response: As the commenter states, the control scenario modeling performed by ENVIRON for
EPA was conducted at the same time as the analysis of potential controls and controlled emission
rates. Because the exact controlled emission rates were not available at the time, we modeled a
High and Low control level.622 As discussed in depth elsewhere, the modeling results provided
estimates of visibility benefits at these estimated controlled rates, representative of the level of
control anticipated due to use of scrubbers and DSI. These modeled visibility benefits then had
to be adjusted to reflect the final determination of the proposed controlled emissions based on
results of the analysis of the potential controls and control performance. After adjusting the
modeled results slightly to reflect the actual proposed emission rates, the estimated visibility
results were evaluated using a number of metrics, including visibility benefit at both 2018
degraded background conditions and annual average natural conditions.
We disagree with the commenter that the High and Low control modeled scenarios
misrepresented the Luminant facilities, and potentially all facilities, where additional scrubber
controls were evaluated in ways that overstated the visibility benefit from the proposed controls.
The only example the commenter raises is the stack temperature and flow velocity. We agree
with the commenter that we did not account for the difference in stack temperature and velocity,
however these differences would only serve to make the visibility benefit of controls larger.
The addition of a scrubber would result in a decrease in stack temperature and as the commenter
states, lower stack temperatures would result in less dispersion of air emissions and thus less
visibility impact at large distances. Therefore, the visibility impact from a unit with a scrubber
would be even less than what was modeled and the visibility benefit larger. These changes
would not impact our overall determination of reasonable controls.
As to the comment that we adjusted the modeled results by updating the baseline uncontrolled
emission for each unit based on SO2 emissions data for 2009-2013, this was a necessary step to
assess the visibility benefit of controls relative to the visibility impairment due to future
anticipated emission levels at these units without the required controls. Comparison of 2018
CENRAP projected emissions to recent actual emissions showed that a number of facilities have
actual emissions that are much higher than CENRAP 2018 modeled emissions.623 For instance,
Big Brown, Sandow, and Martin Lake actual emissions were all significantly higher than 2018
CENRAP modeled rates, with Martin Lake having over 90% more SO2 emissions than projected
by CENRAP for 2018. Both Pirkey and Oklaunion had much smaller actual SO2 emissions than
projected. As we discuss in the FIP TSD, we believe that recent actual emissions are more
representative of anticipated future emissions at the sources evaluated than the CAIR projections
developed in 2006 and adopted by CENRAP.624
The CENRAP modeling was based on an IPM (Integrated Planning Model) that estimated EGU
future emissions in 2018 including reductions for CAIR across the eastern half of the United
States. This analysis was conducted in 2006 and projected that Texas would actually be a
purchaser of SO2 credits, and not as much high level controls would be placed on Texas EGU
sources. Given the length of time between 2006 when the IPM analysis was conducted, and 2013
when we were conducting this analysis, we had some concern that projections could be off for
622 See FIP TSD at A-54
623 See Table A.4-2 in the FIP TSD
624 FIP TSD at A-45
773
-------
the EGUs in Texas. Information available also indicates that SO2 credits are much cheaper than
originally projected, therefore more credits may have been used in lieu of emission reductions.
We also weighed the technique that Texas has used in estimating emissions from EGUs for
future years (including 2018) in ozone attainment demonstration SIPs in Dallas-Fort Worth and
Houston-Galveston-Beaumont625. For these photochemical modeling analyses with CAMx Texas
has relied upon the recent CEM data that is also included in CAMD's databases in conjunction
with information on recently permitted EGUs for estimating the emissions to model for EGUs in
Texas in 2018 as these emission levels are already near levels projected under CAIR Phase II
control such that further emission reductions are doubtful in the absence of some new
requirements.
The actual SO2 allowances for Texas under CSAPR are not much different than the CAIR Cap
for Texas, so large additional reductions over current emission levels were not expected.
However, because we had earlier projected with IPM that controls for the Mercury and Air
Toxics Standards (MATS) may generate the installation of additional scrubbers in Texas that
could potentially result in further SO2 reductions, we again investigated this possibility. Texas
recently submitted comments to us on a more recent IPM projection that was at the time intended
by EPA to be part of a new modeling platform for national rule making.626 In these comments
and comments from several EGU owners in Texas, the assertion was that no significant amount
of additional SO2 controls are expected due to compliance with MATS. The comments also
pointed out that, as some of our cursory research had also indicated, no large SO2 control
projects were planned at most of the sources we were evaluating. Therefore, based on Texas'
recent comments and other information, we concluded considerable uncertainty exists as to
whether any further reductions of SO2 will occur beyond current emission levels as a result of
compliance with MATS or CSAPR. Overall this information supports looking at recent actual
emissions to represent future emission levels in 2018.
In summary, this adjustment from CENRAP 2018 to the a baseline calculated from recent actual
emissions was not an "artificial adjustment" and was necessary to account for the large
difference between specific unit-level emissions in the 2018 CENRAP emissions and a baseline
more representative of anticipated future emission levels in 2018. The results considering the
2018 CENRAP emissions baseline were also needed to provide a comparison with the Texas
regional haze SIP and an estimate of the change from the 2018 CENRAP modeled reasonable
progress goal to a new reasonable progress goal including the controls required in the FIP. The
visibility benefit of individual controls calculated based on the CENRAP 2018 emissions
baseline represents the additional level of visibility benefit from controlling individual units,
consistent with the assumptions/emission projections in the Texas regional haze SIP.
625 HGB 1997 8-Hour Ozone standard attainment demonstration approved by EPA in 2013, see TSD materials for
2010 "Appendix B Emission Modeling for the HGB Attainment Demonstration SIP Revision for the 1997 Eight-
Hour Ozone Standard" on page B-78, "09017SIP_ado_Appendix_B.pdf'.; DFW 1997 8-Hour Ozone standard
attainment demonstration submitted to EPA, see TSD Appendix B: Emission Modeling for the DFW Attainment
Demonstration SIP Revision for the 1997 Eight-Hour Ozone Standard, Page B-39, "AppB_EI_ado.pdf'; DFW 2008
8-Hour Ozone standard attainment demonstration proposed for adoption Dec. 10, 2014 and posted October 2014,
see TSD materials "Appendix B Emissions Modeling for the Dallas-Fort Worth Attainment Demonstration State
Implementation Plan Revision for the 2008 Eight-Hour Ozone Standard" Starting Page 40.,DFWAD_SIP_Appendix
B.pdf
626 TCEQ comment letter to EPA on draft modeling platform dated June 24, 2014. '2018 EMP signed.pdf'
774
-------
We disagree with the commenter that these estimated visibility benefits based on 2018 projected
conditions are small, considering that they represent a "dirty" background approach and the
visibility benefits averaged over the 20% worst days. As we discuss in detail elsewhere, a
"dirty" background approach underestimates the visibility benefits of controls and a "clean"
background approach is necessary to fully assess the potential benefit from controls.
Comment: At Step 11. still not satisfied with the miniscule benefits being predicted, EPA
further adjusted the results of the ENVIRON's CAMx modeling to "estimate^" "the visibility
benefit. . . based on natural ... conditions," which it said "is needed to assess the full benefit
from potential controls."297 EPA noted that CALPUFF, which it did not use to evaluate these
Texas sources, "simulates 'clean' background conditions with no other sources included than the
source(s) being evaluated."298 As EPA explained:
CALPUFF modeling (for BART and other analyses) is conducted to determine a
facility's impact on a Class I area with no consideration of other pollutants in the
air (other than natural background conditions) to challenge and consume the pre-
cursors that are modeled to react with the facility's emissions . . . CAMx is a full
photochemical model with all the other sources quantified and added to the
modeling, such that emissions from other facilities, non-point sources, mobile
sources, etc., all react with available pre-cursors such as ammonia . . . CAMx
takes into account the entire pollution load in the atmosphere in 2018... ,299
EPA estimated that the same emission reductions compared against a "natural" background
would yield a deciview "improvement 3 times greater than" against actual conditions.300 Based
on this explanation, EPA "estimated" what the visibility change would be from the SO2 controls
at each unit against "natural conditions," in isolation, as if there were no other sources of
emissions. EPA used two "natural conditions"—the natural conditions of the 20% worst days
and the average annual natural conditions. EPA does not explain why it used two different
"natural conditions" or what the difference is between the two. Both are presented below in
Tables 6 and 7. Note that EPA's "natural background" artificially eliminates all international
sources of emissions, which in reality contribute significantly to visibility impairment in these
areas and which EPA concedes Texas cannot control or reduce.
Luminant provided a table (Table 6 to comment 0061 [not reproduced here]) that shows the
results of EPA's estimates under EPA's "high" control scenario, in terms of deciview
improvement against a "natural" background at each Class 1 area using ENVIRON's 2018
baseline emissions.
EPA also adjusted the projections by combining both its "scaled" emissions and its "natural"
conditions after-the-fact alterations to ENVIRONS's modeling. These adjustments, together,
result in further inflated visibility improvement numbers that EPA relies on for its proposal.302
Luminant provided a table (Table 7 to comment 0061 [not reproduced here]) that shows the
results of EPA's dual adjustments.
775
-------
Response: As discussed in depth in a separate response to comment in this section of this
document, we disagree with the commenter and believe it is necessary to use a "natural
background" approach to fully assess the visibility benefits of controls. Use of a "dirty"
background when evaluating an individual source's visibility impact and potential controls,
underestimates the visibility benefit of controls. By definition, the "clean" background analysis
using natural conditions eliminates the impact from all other anthropogenic sources, domestic
and international. This approach is aimed at assessing the full potential visibility benefit of
controls. It is not reasonable to only assess the visibility benefit of controls, the value of
installing a control in the immediate future that will permanently reduce visibility impacts from a
source, in such a manner that is dependent on the current level of emissions or impact from other
sources or other countries. For example, in considering only the estimated visibility benefit from
controlling Big Brown using a "dirty" background, an increase in visibility impacts from Mexico
emissions or emissions from another Texas point source would result in a decrease in the
calculated visibility benefit in deciviews from installing controls on Big Brown, making controls
appear less beneficial. By using a metric that is independent of all other emission sources
("clean"), we avoid this paradox that the dirtier the existing air, the less likely it would be that
any control is required. This was also explained in the preamble to the final Regional Haze Rule
and Guidelines for BART Determinations.627 The use of "clean" background is consistent with
our regulations, and necessary to assess the full potential benefit from controls and does not
overstate the visibility benefit. We also discuss the consideration of deciview benefits and
impacts based on "clean" and "dirty" background conditions in our response to comments in the
Cost versus visibility section of this document. TCEQ utilized this same "adjustment" to
natural background in assessing visibility impacts from individual sources or groups of sources
for BART screening purposes using CAMx modeling and source apportionment.628
As the commenter states, we calculated the "natural background" using the 20% worst days
natural conditions and the annual average natural conditions. We note that we also made this
calculation using the 20% best days. Initially we utilized the three available values for natural
conditions to perform this calculation. Furthermore, EPA provided additional guidelines with
narrowly defined flexibility regarding the averaging period to be used for calculating natural
background. These guidelines indicate that the states may use either annual average natural
conditions or the average of the best 20% days for natural conditions.629 Ultimately, we
determined that the annual average value was the more appropriate, conservative approach for
this specific analysis at this time. WE note this is consistent with the selection of natural
visibility background used by Texas in their BART screening modeling. In summary, it was
627 Using existing conditions as the baseline for single source visibility impact determinations would create the
following paradox: the dirtier the existing air, the less likely it would be that any control is required. This is true
because of the nonlinear nature of visibility impairment. In other words, as a Class I area becomes more polluted,
any individual source's contribution to changes in impairment becomes geometrically less. Therefore the more
polluted the Class I area would become, the less control would seem to be needed from an individual source. We
agree that this kind of calculation would essentially raise the "cause or contribute" applicability threshold to a level
that would never allow enough emission control to significantly improve visibility. Such a reading would render the
visibility provisions meaningless, as EPA and the States would be prevented from assuring "reasonable progress"
and fulfilling the statutorily-defined goals of the visibility program. Conversely, measuring improvement against
clean conditions would ensure reasonable progress toward those clean conditions. 70 FR 39124
628 TX RH SIP, Appendix 9-5, "Screening Analysis of Potential BART-Eligible Sources in Texas" at 2-11
629 Paisie, J.W.. Regional Haze Regulations and Guidelines for Best Available Retrofit Technology (BART)
Determinations. Memorandum to Kay Prince, Branch Chief EPA Region 4. July 19, 2006.
776
-------
appropriate to use a "clean" background approach and to adjust the baseline to reflect recent
actual emissions.
Comment: At Step 12, EPA used the so-called visibility benefits it had estimated to decide
which units would be subject to regulation. For example, EPA said that there was a "decrease in
visibility improvement benefits at the three Class I areas for the W.A. Parish and Welsh units
compared to the benefits at other facilities that mainly impact WIMO."304 EPA did not, however,
state what threshold cut-off it was applying, nor did it compare its estimated visibility
improvements to the projected costs for each unit. That is, EPA did not rely on or even calculate
the amount of visibility improvement per dollar spent at each of the units it examined. Instead,
based exclusively on its estimates of visibility benefits, EPA excluded 7 of 21 units from
regulation and established emission limitations for the remaining 14 units.305 EPA also
established emission limits for San Miguel, which it did not model for visibility improvement.306
Response: We disagree with the commenter. Our determination of reasonable controls was
based on an evaluation of the four factors, including cost and consideration of estimated
visibility benefits of controls in terms of deciview, and reduction in extinction and percent
extinction. As we discuss in more detail elsewhere, we reject Luminant's contention that we
should have used the $/dv metric. While we do not believe it is appropriate to use a $/dv metric,
we did consider the cost-effectiveness in $/ton and the relative visibility benefit anticipated from
the controls. We did not establish a threshold cut-off for visibility benefit for controls but
considered cost and visibility benefit in a relative sense in making our control decisions.
Significance of visibility benefits is a source- and Class I area-specific evaluation, meaning that
it depends on how much visibility improvement is needed at the Class I area(s), how much a
specific source impacts the Class I area(s), and the cost effectiveness and potential visibility
improvement of available control options.
For example, in the FIP TSD and the proposed FIP, we note lesser visibility improvement
benefits at the three Class I areas for the W. A. Parish and Welsh units compared to the benefits
at other facilities that mainly impact the Wichita Mountains. The visibility benefits at Wichita
Mountains of scrubber retrofits at the Parish and Welsh units (0.102—0.126 dv) are
approximately half of the visibility benefit of scrubber retrofits at the Coleto Creek and
Monticello units (0.2 -0.254 dv), and much less than the visibility benefit from controlling units
at Big Brown (0.436-0.438 dv). We also note that when considering the costs of controls and the
relative visibility benefit, the Parish scrubber retrofits would be slightly more expensive with
respect to $/ton but would be much less effective in improving visibility at the Wichita
Mountains, when compared to the required controls at units Monticello or Coleto Creek. For the
Welsh scrubber retrofits, the costs ($/ton) are approximately 50% greater than the cost of
scrubber retrofits at Monticello or Coleto Creek and result in approximately 50% less visibility
improvement at Wichita Mountains. Furthermore, as we noted in the FIP TSD and also noted by
the commenter, Welsh unit 2 is under a consent decree630 to shut down by the end of 2016,
eliminating the visibility impact from this unit.
630 See Sierra Club et al v. U.S. Army Corps of Engineers, civil 4:10-cv-04017-RGK, also letter from JohnM.
McManus to Mike Wilson, dated May 2, 2013. Under the terms of a consent decree, after the Turk Plant commences
777
-------
Comment: [TCEQ/PUCT (0056) p. 18] The TCEQ urged the EPA to remove all text about
benefits of emission reductions from "actual emission levels" from its final action and technical
support documents. These discussions exaggerate the potential benefits of the EPA's proposed
FIP and are irrelevant to the approvability of the 2009 RH SIP.
In comments on the EPA's proposed FIP, the TCEQ noted that both Table 44: CalculatedRPGs
for 20% Worst Days ... and Table 45: Anticipated Visibility Benefit... should be removed from
the final action because they tabulate calculated benefits that will not occur by 2018, the only
year that is appropriate for evaluating the visibility impacts of proposed controls. The 2018
visibility conditions that the 2009 RH SIP will produce are the appropriate starting points for
evaluating the effects of the EPA's proposed FIP.
The TCEQ stated that the EPA inappropriately suggests in its proposal and technical support
documents that emission rates in 2011, 2012, or 2013 are relevant to what the Texas 2009 RH
SIP will achieve by 2018. The RHR sets 2018, the last year in the first planning period, as the
time by which a state's SIP must provide for reaching the state's RPG. The RHR does not imply
the need for particular emission levels during any intermediate year between the baseline period
and 2018.
The TCEQ stated that there is no technical basis for the EPA's selection of actual emissions from
2009 through 2013 as the base from which to calculate the benefit of applying the FIP controls.
During the 2009 through 2013 period, the emissions were not affected by the full range of
additional emission reduction requirements contained in the 2009 RH SIP.
The TCEQ stated that choosing 2011 ignores seven more years of emissions reductions required
under Texas' long-term strategy. As Texas' 2014 Five-Year RH SIP submittal shows in Figure 4-
1: Texas Modeled Emissions Inventory Summary for 2002 and Figure 4-2: Updated Texas
Emissions Inventory Summary for 2005, the SO2 and NOx emissions in Texas are already lower
than the straight line between the 2000 through 2004 baseline condition period and the 2018 SO2
and NOx emissions estimates used to develop the 2009 RH SIP.15
The TCEQ stated that Table 45 misleads a reader to believe that the EPA's proposed FIP action
would produce a 0.62 deciview improvement in visibility at Wichita Mountains. However, as
discussed in comment A.2., the potential 0.14 deciview improvement at Wichita Mountains is
almost certainly an overstatement of the incremental benefit from the proposed FIP in 2018
because SO2 emission reductions are occurring due to other requirements and the actual SO2
emissions will likely be lower than those in the CENRAP 2018 emissions projections.
Footnotes:
15 See https:/ jwww.tceq.texas.govjassets/publicjimplementation/air/sip/haze/13012SIP ado. pdf.
commercial operation, Unit 2 will be restricted to a 60% annual capacity factor during any rolling 12-month period.
Thereafter, Unit 2 must be retired no later than December 31, 2016.
778
-------
Response: We disagree with the commenter that using recent actual emissions exaggerates the
benefit of controls. As explained in detail in the FIP TSD (see page A-45) and elsewhere in the
response to comments, we determined it was necessary to consider recent actual emissions from
EGUs due to uncertainty in 2018 projected emissions completed in 2006, the cost of SO2 credits
being lower than originally projected, and comments from Texas on a more recent IPM
projection indicating that significant SO2 reductions were not anticipated at these sources and no
large SO2 control projects were planned at most of the sources being evaluated. We also noted
that TCEQ has utilized recent emission data for EGUs when developing projected emissions for
2018 (and other future years) when developing ozone attainment demonstrations. Overall this
information supports looking at recent actual emissions to represent future emission levels in
2018.
We agree with TCEQ that overall state-level emissions have decreased from the baseline.
However, TCEQ provides no specific information to support the assertion that additional
emission reductions are anticipated at the specific sources we are requiring controls on between
now and 2018. We provide a comparison of the current emissions to the 2018 CENRAP
projected emissions in Table A.4-2 of the FIP TSD. We note that the data indicates that a
number of facilities have actual emissions that are much higher than modeled. For instance, Big
Brown, Sandow, and Martin Lake were all significantly higher than modeled rates, with Martin
Lake having over 90% more SO2 emissions than modeled. As we mention above, we do not
anticipate any additional significant reductions at any of these facilities due to MATs or
CSAPR/CAIR, or any other requirements by 2018. This is consistent with comments from
Texas on a more recent IPM projection indicating that significant SO2 reductions were not
anticipated at these sources and no large SO2 control projects were planned at most of the
sources being evaluated.631 It is unclear if the implementation of the sulfur dioxide (S02)
National Ambient Air Quality Standard (NAAQS) will result in any emission reductions at these
specific sources as areas in Texas have yet to be designated and it is unclear the amount of
emission reductions that would be necessary to attain the standard for those areas that might be
designated non-attainment. We believe that recent actual emissions (2009-2013 annual average,
excluding the maximum and minimum years) is a reasonable estimate of the anticipated future
emissions in 2018 at these sources. We note that none of the effected sources provided
information in their comments suggesting emission reductions were anticipated in the near
future. Furthermore, there are no enforceable emission limits on these specific sources to secure
any specific emission reductions beyond current emissions at this time. Based on the above
information, we disagree with the commenter's assertion that actual emissions will be lower than
projected for the sources with required controls or that the estimated 0.14 dv improvement at
Wichita Mountains due to scrubber upgrades is an overstatement.
We utilize the 2009-2013 baseline to estimate the potential visibility benefits from controls
required by the FIP, noting that emission reductions will occur from this emission level and not a
2018 level estimated by CENRAP in 2006, which in many cases is much lower than current
actual emissions. We disagree with the commenter that it is not appropriate to include this
631 Texas comments on Draft IPM modeling conducted by EPA for potential national rule making platform provided
on June 26, 2014. In this docket materials as "TCEQ comment letter to EPA on draft modeling platform dated June
24, 2014. '2018 EMP signed.pdf.
779
-------
information in our proposal. Table 45 is necessary to present the full visibility benefit of the
required controls that would result from reducing recent actual emission levels at the affected
units down to the controlled emission levels. The FIP controls are estimated to provide for a 0.62
dv improvement in projected 2018 visibility conditions (as projected by CENRAP) at Wichita
Mountains when all required controls are implemented when considering reductions from recent
actual emission levels. This provides for an apples-to apples comparison between the level of
additional estimated visibility improvement from the CENRAP 2018 projected visibility
conditions based on reducing CENRAP 2018 emissions in Tables 43 and 44 and reducing actual
emissions in Table 45.
We include an estimate of the visibility benefit from the required controls compared to the
CENRAP modeled 2018 baseline as shown in Tables 43 and 44. This was necessary to assess
how the RPG estimated by the CENRAP modeling would change based on the required controls.
As this was the 2018 baseline used for the CENRAP modeled RPGs, we needed to use the same
baseline to calculate the adjustment to the RPG. Table 43, which TCEQ does not object to,
contains this information for the scrubber upgrades that are required to be in place by 2018.
Table 44 includes this same information for the scrubber retrofits. It is appropriate to include
this information to assess the visibility benefit of these required controls with respect to the
CENRAP modeled RPG.
We disagree with the commenter that the visibility benefits from the scrubber upgrades are
irrelevant to the approvability of the 2009 TX RH SIP. As discussed in more detail elsewhere,
we disapproved the TX RH SIP in part because we determined that the analysis of reasonable
controls in the Texas FIP was flawed. In order to determine if the ultimate decision that no
additional controls were necessary was reasonable, and if we determined it was not reasonable
then to inform our development of a FIP, we had to conduct our own analysis. Our analysis,
including the analysis of the visibility benefits of scrubber retrofits, showed that cost-effective
controls were in fact available that resulted in significant visibility benefits and provided for
meaningful improvements in visibility conditions at the Class I areas of interest. Because these
retrofits are not anticipated to occur before 2018, we evaluated the visibility benefit from them
separately, however this does not impact the fact that they are reasonable controls. We discuss
the comment on requiring controls beyond 2018 in a separate response to comment.
15.v EPA Miscalculated The Rpgs, Rpgs Do Not Account for Reductions Outside of Texas
(E.G. OK FIP)
General Summary: EPA's methodology to estimate revised RPGs for Big Bend, Guadalupe
Mountains, and Wichita Mountains is without precedent and is not supported by the record. The
commenters also state that the revised RPGs are incorrect because they do not account for
reductions in OK emissions.
Comment: EPA's proposed RPGS, and thus its FIPS, are unlawful and arbitrary and
capricious [Luminant (0061) p. 123]
780
-------
Luminant stated that the last step in EPA's reasonable progress analysis—its attempt to
"quantify[]" new RPGs for the three Class I areas as part of its proposed FIPs—is one of the
most unsupportable aspects of EPA's proposal. EPA's new proposed RPGs suffer from many
flaws that render them unlawful and arbitrary and capricious, even putting aside the numerous
legal and technical errors and deviations in EPA's analysis previously discussed. When EPA
undertakes to issue a FIP, "EPA steps into the State's shoes" and "must meet the same
requirements" as the state.767 Here, however, EPA's recalculated RPGs do not meet even the
most basic requirements for establishing an RPG that a state would be required to meet. EPA's
RPGs, if developed and submitted to EPA by a state, would surely be disapproved by EPA.
767 77 Fed. Reg. at 40,164.
First, Luminant noted that EPA cites no precedent or support for its method of "adjusting" or
"reset[ing]" an RPG in the manner that is has.768 EPA did not develop its proposed RPGs as
states would—that is, by modeling all existing and reasonable controls and comparing the results
to the URP.769 Instead, EPA simply took CENRAP's prior modeling (using one modeling
platform) and apparently subtracted the results of its new unit-level modeling of specific Texas
sources (using a different modeling platform) from the CENRAP results. EPA provides no
explanation of how it accomplished this calculation or any basis for it. For example, EPA says
that 0.14 deciview is the "[predicted additional benefit [at Wichita Mountains] due only to FIP
scrubber upgrades," Id. at 36, and it subtracts that number from the 2018 CENRAP projection,
but nowhere does EPA explain how it calculates the 0.14 deciview benefit, nor can we locate that
number in any of the various spreadsheets in the record770 And, and although EPA has now
reviewed every other state's regional haze SIP, we have found no examples of where EPA has
re-calculated an RPG in this manner. To allow for meaningful comment on its current proposal
and new methodology, EPA must issue an amended notice or notice of data availability that
explains how EPA calculated the final RPGs in its proposed FIP and the justification for doing so
in this unorthodox manner.
768 79 Fed. Reg. at 74,886, 74,889.
769 EPA Reasonable Progress Guidance at 2-3.
770 FIP TSD at 36-37. For example, EPA says that 0.14 deciview is the "[predicted additional benefit [at Wichita
Mountains] due only to FIP scrubber upgrades," Id. at 36, and it subtracts that number from the 2018 CENRAP
projection, but nowhere does EPA explain how it calculates the 0.14 deciview benefit, nor can we locate that
number in any of the various spreadsheets in the record.
EPA then attempts to "quantif[y]" new RPGs for the three Class I areas.312 EPA "quantifies]"
the new RPGs using "the results of our reasonable progress analysis of point sources as
described in detail in our FIP TSD."313 However, in doing so, EPA does not use the visibility
modeling from ENVIRON or its own unit-level quantification of benefits that were used to
justify the controls EPA determined to impose in its FIP. Nor does EPA model the visibility
conditions in light of the recently finalized BART requirements on Oklahoma sources. Instead,
EPA reverts back to the original CENRAP modeling used by Texas and Oklahoma to quantify
the states' original RPGs. EPA apparently subtracts from the 2018 CENRAP projection the so-
called visibility improvement that EPA separately estimated (under different modeling) from the
controls it would impose on Texas sources by 2018
781
-------
Response: We disagree with the comment and believe we took a reasonable approach to
estimate the change in overall visibility impairment anticipated due to the required controls and
provided all calculations for review. We also disagree with the commenter's description of how
the states estimated the reasonable progress goals. While our guidance suggests that the
reasonable progress goals should be established by modeling all existing and reasonable controls,
in practice all RPOs including CENRAP completed the modeling early in the process. The 2018
CENRAP modeling was completed before any states had completed their BART and RP
determinations. In many cases, the 2018 projection included an assumption of BART level
controls and "on the book" controls. Once final BART determinations and reasonable progress
determinations were completed, the RPO did not go back and remodel to reassess the reasonable
progress goals. In our proposed action in Arkansas632, as well as our actions in Arizona633 and
Hawaii634, the modeled reasonable progress goals were adjusted based on a methodology of
scaling of visibility extinction components in proportion to emission changes. We noted that
although we recognize that this method is not refined, it allows us to translate the emission
reductions achieved through the FIP into quantitative RPGs, based on modeling previously
performed by the RPOs. However, in this case, our analysis using CAMx and source
apportionment, provided a somewhat more refined means to estimate the visibility benefit from
specific individual controls on the 20% worst days in 2018. While there is limited precedent for
adjusting the RPO calculated RPGs to account for emission reductions achieved in a FIP or
revised SIP, we took a reasonable approach based on the information available. We adjusted
each RPG established by Texas or Oklahoma for 2018 by the amount of visibility benefit
anticipated from all scrubber upgrades estimated by our modeling analysis based on CAMx
source apportionment modeling. In estimating the deciview visibility benefit in 2018 compared
to the CENRAP modeled 2018 RPGs, we considered reductions from 2018 CENRAP emissions
levels and 2018 "dirty" background conditions. We believe that this is a reliable estimate of the
amount of visibility benefit anticipated from controls (e.g. 0.14 dv for the Wichita Mountains)
beyond the projected 2018 CENRAP RPGs. We then simply adjusted the RPGs established by
the state by the amount of visibility benefit anticipated from the additional controls.
The commenter is incorrect in stating that we did not provide this calculation in the record. We
discuss the proposed RPGs in Section 13 of the FIP TSD and direct the reader to the appropriate
document in the record for our calculations and estimates of visibility benefits.635 For example,
the 0.14 deciview benefit is seen in the referred to document available in the docket636 ("Vis
modeling summary.xls," spreadsheet "Summary Vis 2018"), as well as in the FIP TSD at Table
20, 21 and A.6-6. In describing this value, we state that it is the total visibility benefit from all
scrubber upgrades beyond the 2018 CENRAP modeled emissions calculated based on the 2018
"dirty" background conditions. The spreadsheet includes all the calculations behind these
values. The included spreadsheet sums the estimated reduction in extinction from all the
required scrubber upgrades based on 2018 CENRAP emission levels and then calculates the
deciview improvement based on "dirty" 2018 background conditions. Because we provided all
632 80 FR 18944, 18997
633 79 FR 52420, 52468
634 77 FR 31692, 31708
635 See FIP TSD at p36 and footnote 24 on the same page.
636 See Vis modeling summary.xls in the docket to this action. The extinction levels for individual sources are
calculated in the "ext. summary" spreadsheet column "O" and the total extinction benefit and deciview benefit
calculations are found in the "summary vis 2018" spreadsheet.
782
-------
the relevant information and calculations in the docket at proposal, we disagree that we "must
issue an amended notice or notice of data availability that explains this calculation and the
justification for doing so in this unorthodox manner".
As discussed above, we adjusted the CENRAP modeled reasonable progress goals to translate
the emission reductions required in this FIP for Texas sources into quantitative reasonable
progress goals. Luminant's comment concerning consideration of required emission reductions
in Oklahoma is addressed in a separate response to comment below.
We note that unlike the emission limits that apply to specific reasonable progress sources, the
reasonable progress goals are not directly enforceable. Rather, the reasonable progress goals are
an analytical tool used by EPA and the states to estimate future visibility conditions and track
progress towards the goal of natural visibility conditions.
With regard to the comment that we subtracted from the 2018 CENRAP projection the visibility
improvement that we separately estimated (under different modeling) from the controls it would
impose on Texas sources by 2018 the commenter is correct we did use the CENRAP CMAQ
projections and our CAMx model projection in our calculations.
Comment: Second, Luminant noted that EPA's proposed RPGs have no relationship to EPA's
underlying analysis of Texas sources and its justification for the new emission limits. EPA
justifies its proposal to require emission reductions at a few Texas sources by altering
ENVIRON's modeling in at least two ways: 1) using "adjusted actual emissions" (which do not
reflect the actual emissions at these units, as shown in Table 11 and Figure 9) and 2) based on
projected visibility improvement against "natural conditions." As shown in Tables 4, 5, 6, and 7,
these two alterations artificially inflate the projected visibility improvements that EPA relies on
to justify its decision to require additional controls. As discussed elsewhere in these comments,
EPA provides no justification for altering ENVIRON's modeling in this manner. But even
putting that error aside, when EPA subsequently calculates the RPGs for the Class I areas, it does
so on an entirely different basis. EPA explains that its "proposed RPGs are established based on
an adjustment of the 2018 RPGs established by Texas and Oklahoma that were based on the
2018 CENRAP modeling."771 CENRAP modeling was performed—correctly—against realistic
2018 conditions, not the artificial background EPA claims to use for its decision-making, and
without the alterations that EPA makes for its unit-level assessment. In other words, EPA's
"adjustment" to the RPGs is apparently quantified against the actual "dirty" background that
EPA previously rejected in its analysis as inconsistent with the Clean Air Act.772 And, EPA's
justification for its proposal (visibility improvement using "adjusted" emissions against "natural
conditions") is entirely disconnected from and does not support its final decision to adjust the
RPGs.
Response: Luminant's comments concerning the use of natural background and actual
emissions are addressed in detail elsewhere. We disagree with the comment that we provide no
justification for the "adjustments" made to the model results to account for recent actual
emissions and natural background. We provide a thorough explanation of why it was
appropriate to consider recent actual emissions and use natural background conditions to assess
783
-------
the full visibility benefit of individual controls in separate response to comments. We disagree
with the comment that the proposed RPGs have no relationship to EPA's underlying analysis of
Texas sources. The commenter fails to recognize that the methodology necessary to assess
visibility impacts from individual sources is different from the methodology necessary to
estimate overall visibility conditions in 2018. Our determination on what controls were
reasonable was made considering the four factors and the estimated full potential for visibility
benefit from individual controls based on consideration of natural background and recent actual
emissions. Based on our determination of the reasonable controls, we then estimated the
reasonable progress goal for 2018. In order to estimate the RPG for 2018, we started with the
CENRAP modeled RPG and adjusted it to account for the additional emission reductions due to
the required controls. This calculated adjustment to the 2018 CENRAP RPG had to be based on
reductions from CENRAP 2018 emission levels that were modeled for the 2018 CENRAP RPG
and considering the 2018 "dirty" background since the adjustment is made to 2018 visibility
conditions. As we describe in the FIP TSD, CAMx is inherently a "dirty" background analysis
because it models all emissions from all sources. However, as discussed elsewhere, it is
appropriate to adjust the CAMx estimated extinction to a natural background analysis to estimate
the full benefit of a potential control.
We disagree with the commenter that our adjusted actual emissions do not reflect the actual
emissions at these units, as shown in Table 11 and Figure 9 of the Luminant comments. We used
actual annual emissions taking a three year average of emissions from 2009-2013, excluding the
maximum and minimum years. Table 11 and Figure 9 show emissions from this same period
and more recent data but break up annual emissions into quarters and combine emissions across
all Luminant units with proposed controls. We note that 3rd quarter SO2 emissions are roughly
constant from 2009-2014 and this period is typically when high sulfate impact days are observed.
We used the most recent complete data that was available at the time we developed our analysis,
2009-2013, and excluded the max and minimum years to get a reasonable estimate of typical
actual emissions. We provide additional information on recent emissions data from these
sources in a separate response to comment.
Comment: Third, Luminant noted that EPA's RPGs are contrary to its own regulations. EPA's
regulations define an RPG in terms of "an improvement in visibility for the most impaired days .
. . and no degradation in visibility for the least impaired days . . . ,"773 Texas's and Oklahoma's
SIPs thus provide visibility goals for both the "worst" and "best" days.774 EPA's proposal,
however, only addresses half of the states' RPGs—i.e., the "worst" days—but provides
absolutely no analysis of the part of the RPGs addressing the "best" days. Thus, EPA's
"additional analysis" is an inadequate basis for disapproving Texas's and Oklahoma's RPGs and
an unlawful basis for its FIP. EPA's analysis provides absolutely no basis for disapproving
Texas's and Oklahoma's RPGs for the 20% best days, even though EPA claims to be
disapproving them.775
Footnotes:
767 77 Fed. Reg. at 40,164.
768 79 Fed. Reg. at 74,886, 74,889.
769 EPA Reasonable Progress Guidance at 2-3.
784
-------
770 FIP TSD at 36-37. For example, EPA says that 0.14 deciview is the "[predicted additional benefit [at Wichita
Mountains] due only to FIP scrubber upgrades," Id. at 36, and it subtracts that number from the 2018 CENRAP
projection, but nowhere does EPA explain how it calculates the 0.14 deciview benefit, nor can we locate that
number in any of the various spreadsheets in the record.
771 Id. at 36.
772 Id. at A-38 to A-39.
773 40 C.F.R. § 51.308(d)(1).
774 2009 Texas SIP Narrative at 10-3 to 10-4.
775 7 9 Fed. Reg. at 74,822; Oklahoma TSD at 11.
[UARG (0065) p. 21] UARG stated that the EPA's assessment also fails to address reasonable
progress for the 20 percent best days, as required by the regional haze rule. EPA's assessment is
focused exclusively on visibility impacts on the 20 percent worst days and, accordingly, fails to
satisfy the minimum requirements of the regional haze rule
Response: We disagree with the comment. Our basis for disapproving the relevant reasonable
progress goals for the 20% best days arises, as was noted in our proposal, from our determination
that the analysis developed by Texas to evaluate reasonable progress controls was flawed and
additional controls are necessary for the first planning period. Finalizing requirements for
additional controls, as we now accomplish with our final rule, makes "visibility on these days
better than Texas projects," as we noted in our proposal.637'638 The submitted reasonable
progress goals for the 20% best days did not consider reductions from the reasonable controls, so
they cannot be approved. We understand the comment to request a quantitative assessment of
the projected visibility conditions for the 20% best days. These calculations have been
completed and add to our position that visibility will be better than Texas projects. These
numbers, following the same methodology that we employed with the 20% worst days, are
summarized in the table below.
Estimated RPG for the 20% Best Days Based on Predicted Benefit of Scrubber Upgrades
Beyond 2018 CENRAP Projected Visibility Conditions
Predicted
Additional
additional
benefit
Total
benefit
from
proposed
controls
RPG
Baseline
(dv)
2018
CENRAP
Projection
(dv)
benefit
due only to
FIP
scrubber
upgrades
(dv)
Proposed
RPG (dv)
predicted
due to
FIP
scrubber
retrofits
(dv)
Assuming
all
controls in
place by
2018
Wichita
Mountains
9.78
9.23
0.007
9.22
0.032
0.039
9.19
Big Bend
5.78
5.6
0.008
5.59
0.042
0.050
5.55
637 79 FR 74843.
638 "No degradation," as distinctly needed for the 20% best days, is ensured because added controls do not
significantly impact the 20% best days and would serve only to improve visibility on these days. Even so, what we
provide as the 20% best day reasonable progress goals for 2018 (i.e., the "least impaired days") for Big Bend,
Guadalupe Mountains and Wichita Mountains numerically differ from the numbers that Texas had submitted by
very small amounts. By the design of 40 CFR 51.308(d)(1), improvements forthe most impaired days provide a
more vital benchmark for progress that may be made.
785
-------
Guadalupe
Mountains
5.95
5.7
0.001
5.70
0.030
0.031
5.67
Comment: [Luminant (0061) p. 125] Luminant stated that EPA's proposed RPG for the Wichita
Mountains is unlawful because it fails to take into account emission reductions for all Oklahoma
sources. EPA's regulations require a state, and thus EPA when it steps into the state's shoes, to
account for all emission reductions that are expected to result from compliance with existing
rules and other reasonable measures and compare the results to the URP in order to determine
reasonable progress.776 But EPA fails to do that here in determining what is reasonable progress
for the Wichita Mountains. For example, EPA's RPG does not take into account the additional
SO2 emission reductions from several Oklahoma sources that are required by EPA's recent
BART FIP and Oklahoma's revised SIP finalized in 2014.777 As EPA explains:
[I]n our earlier action on the Oklahoma Regional Haze SIP, we disapproved the
SO2 BART determinations for six EGUs at three power plants in Oklahoma and
promulgated a FIP setting more stringent SO2 emission limits for these EGUs.
Although we subsequently approved a SIP revision from Oklahoma addressing
the BART requirements at one power plant, and removed the FIP requirements
for this facility, our FIP and the revised Oklahoma SIP require greater
reductions overall in emissions of SO2 than was assumed [by Oklahoma] in
setting the RPGs for the Wichita Mountains 778
EPA faults Oklahoma's RPG for Wichita Mountains because it does not take these reductions
into account.779 Yet, in assessing Oklahoma's RPG and calculating its proposed RPG for Wichita
Mountains, EPA itself fails to calculate or quantify the visibility benefit at Wichita Mountains
from these now required controls.780 Indeed, although EPA claims the failure to account for these
reductions is one basis for its disapproval of Oklahoma's RPG,781 EPA's remedial FIP also fails
to take these on-the-books emission reductions into account.782 Thus, for the exact same reason
that EPA offers for disapproving Oklahoma's RPG—that it does not account for all existing
required controls—EPA's FIP is equally deficient and cannot stand. As discussed in Section V,
EPA's RPG also fails to include SO2 and NOx reductions from EPA's Cross-State Air Pollution
Rule ("CSAPR"), which became effective January 1, 2015, and serves as the BART-equivalent
for Texas sources. EPA's RPG for Wichita Mountains thus fails to comply with EPA's
regulations.
Footnotes:
776 EPA Reasonable Progress Guidance at 2-3.
777 76 Fed. Reg. 81,728, 81,753 (Dec. 28, 2011); 79 Fed. Reg. 12,944, 12,954 (Mar. 7, 2014).
778 Oklahoma TSD at 11 (emphasis added) (internal citations omitted).
779 Id.
780 79 Fed. Reg. at 74,886.
781 Oklahoma TSD at 11.
782 79 Fed. Reg. at 74,886.
Response: The comment is taken out of context and does not fully capture the rationale for our
disapproval. The full text is:
786
-------
However, for the reasons discussed below, we are proposing to disapprove
Oklahoma's RPG for the Wichita Mountains. First, in our earlier action on the
Oklahoma Regional Haze SIP, we disapproved the SO2 BART determinations for
six EGUs at three power plants in Oklahoma and promulgated a FIP setting more
stringent SO2 emission limits for these EGUs.8 Although we subsequently
approved a SIP revision from Oklahoma addressing the BART requirements at
one power plant,9 and removed the FIP requirements for this facility, 10 our FIP
and the revised Oklahoma SIP require greater reductions overall in emissions of
SO2 than was assumed in setting the RPGs for the Wichita Mountains. Second,
we are proposing to disapprove Oklahoma's RPGs for the Wichita Mountains
because they were based on an incomplete consultation with Texas under
51.308(d) (l)(iv) that resulted in inadequate reasonable progress towards the
national visibility goal. For these reasons, the RPGs for Wichita Mountains do
not reflect the degree of visibility improvement anticipated from these
639
measures.
We considered the comment concerning consideration of the reductions required by the BART
FIP in Oklahoma in setting the 2018 reasonable progress goals and we note that the CENRAP
2018 modeling did include an assumption for anticipated BART reductions for the Oklahoma
sources reflecting an emission limit of 0.15 lb/MMBtu on all six coal-fired EGU subject-to-
BART units. We believe these assumptions are a reasonable approximation of the anticipated
BART reductions in Oklahoma at this time, considering the uncertainty of the timing of the
reductions for some of the sources and the uncertainty in the final control scenario chosen by the
owners to meet the requirements. The BART requirements for the Sooner and Muskogee
facilities must be met by January of 2019; therefore, it is unclear if any reductions at these
facilities will occur during the first planning period to be considered in the 2018 reasonable
progress goal. Furthermore, while the BART FIP requires emission limits consistent with the
installation of scrubbers at four units across the two facilities, there is uncertainty at this time as
to the facilities owner's plans for compliance through installation of controls or retirement of the
some of the BART units and the schedule for those plans. In the case of the AEP Northeastern
facility, an Oklahoma SIP revision has been approved that requires, in addition to emission limits
on the Northeastern facility, an enforceable commitment that obligates ODEQ to "obtain and/or
identify additional SO2 reductions within the State of Oklahoma to the extent necessary to
achieve the anticipated visibility benefits estimated" by the CENRAP in the 2018 CENRAP
projections.640 The required enforceable emission limits in the Oklahoma and Texas FIPs
remedy the deficiencies in the SIPs and our finalized reasonable progress goals for 2018 properly
consider the visibility benefits anticipated by those required emission reductions anticipated by
2018. We address comments concerning emission reductions due to CSAPR elsewhere in this
document.
We note that unlike the emission limits that apply to specific reasonable progress sources, the
reasonable progress goals are not directly enforceable. Rather, the reasonable progress goals are
an analytical tool used by EPA and the states to estimate future visibility conditions and track
progress towards the goal of natural visibility conditions.
639 OK TSD at 11.
640 79 FR 12954
787
-------
15.w General Four-Factor Analysis Comments
Comment: Every factor of EPA's proposed reasonable progress analysis fails.
[GCLC (0063) p. 14]
GCLC stated that the four-factor analysis for the purpose of Texas' SIP has no relation or bearing
on Texas' LTS analysis. EPA has failed to demonstrate that its proposed FIP will comply with
the four-factor analysis for determining reasonable progress.65 The Proposed FIP will result in
miniscule and non-perceivable changes in visibility compared to the Texas SIP; at the same time,
it will cause serious economic harm and jeopardize electric reliability in the state, all while
imposing requirements that would extend beyond the planning period.
According to GCLC, one of the primary reasons EPA's analysis is so flawed is due to its reliance
on the BART guidelines and EPA's insistence on applying source-specific limitations for a rule
that is meant to look beyond specific sources and look at source-categories or state-wide
contributions more broadly. This is not to say that EPA's source-specific four-factor analysis is
correct, either, as it seriously underestimates the harms of the rule to those individual units
regarding all of the factors of the reasonable progress analysis.
Footnotes:
65 See 42 USC§ 7491(g)(1).
Response: We address specific comments concerning these issues are addressed elsewhere in
this document.
Comment: EPA's Proposal to Require Installation of Scrubbers at Tolk Would Produce
Miniscule Visibility Benefits. [Xcel Energy (0064) p. 6, 25-28]
[Xcel Energy (0064) p. 6] Xcel Energy stated that the EPA's proposal to impose costly controls
on Xcel Energy's Tolk Generating Station ("Tolk") is technically and legally unjustifiable. It is
based on an incomplete and inadequate analysis of the four statutory factors. It also effectively
ignores the minute-to-nonexistent visibility benefit that would result from the proposed controls
Response: We address specific comments concerning our four-factor analysis and the
justification for the required controls elsewhere in this document. We address comments
concerning the magnitude of visibility benefit of controls where we address comments on our
analysis of cost versus visibility in this document.
16. Interstate Visibility Transport
Comment: Multiple commenters argued that EPA should not disapprove NAAQS infrastructure
SIPs as part of this action.
788
-------
[NRG (0078) p. 14] NRG argued that the proposed infrastructure SIP must not be finalized
because they rest on the same reasoning as EPA's disapproval of Texas' regional haze plan. 79
Fed. Reg. at 74,892 (proposed 40 C.F.R. § 52.2304(d)-(e)).
[UARG (0065) p. 32-33] UARG stated that EPA has no authority to disapprove Texas's Section
110(a)(2)(D)(i)(II) interstate transport SIP revisions.
[GCLC (0063) p. 18-19] GCLC stated that EPA's proposed disapproval of Texas' infrastructure
SIP submittals for interstate transport and visibility protection is contrary to the CAA. EPA has
no basis to disapprove portions of Texas' NAAQS SIPs. Texas' SIP submittals comply with all
requirements of the CAA and must be approved.
[Luminant (0061) p. 49, 142] Luminant argued that EPA's proposed disapproval of Texas's
NAAQS infrastructure SIPs is contrary to law. EPA claims that those prior submissions do not
satisfy the so-called "good neighbor" provision of the Clean Air Act, § 110(a)(2)(D)(i)(II). 79
FR 74823. According to EPA, that provision required Texas's SIP revisions that addressed the
NAAQS to "have adequate provisions to prohibit in-state emissions from interfering with
measures required to protect visibility in any other state." Id. EPA's proposal is flawed both
legally and factually.
[Luminant (0061) p. 49] Luminant noted that the EPA proposes to disapprove Texas's SIP
submittals for multiple NAAQS (1997 PM2.5; 1997 ozone; 2006 PM2.5; 2008 ozone; 2010 NO2;
and 2010 SO2) because Texas, like many states, "relied on the improvement in visibility expected
to result from the implementation of CAIR in developing their long-term strategy." 79 FR 74888.
EPA proposes that its CSAPR FIP, which relies on emission reductions from the implementation
of CSAPR in lieu of BART, would address "this deficiency in the Texas SIP." Id.
[Luminant (0061) p. 49] Luminant noted that EPA proposes to disapprove Texas's SIP
submittals for three NAAQS (1997 PM2.5; 2006 PM2.5; and 2010 SO2) because of EPA's
"proposed conclusion that additional control of SO2 emissions in Texas is needed to prevent
interference with measures required to be included in the Oklahoma SIP to protect visibility." Id.
EPA proposes that its reasonable progress FIP "addresses this deficiency in the Texas SIP." Id.
Structural requirements. [Luminant (0061) p. 142; GCLC (0063) p. 18-19], Luminant and
GCLC argued that EPA misinterpreted the scope and intent of CAA § 110(a)(2)(D)(i)(II). EPA
is attempting to impose substantive requirements and emissions limitations via the §
110(a)(2)(D)(i)(II), rather than what is clearly contemplated in the CAA - a requirement to have
structural requirements in place. This present-day interpretation conflicts with EPA's own
historical interpretation of this CAA provision, including that the purpose of this provision is to
"assure that the air agency's SIP contains the necessary structural requirements for the new or
revised NAAQS." EPA, Guidance on Infrastructure State Implementation Plan (SIP) Elements
under Clean Air Act Sections 110(a)(1) and 110(a)(2), at 2 (Sept. 2013). (emphasis added).
EPA has gone on to add that" the infrastructure SIP submission process provides an opportunity
for the responsible air agency, the public, and the EPA to review the basic structural
requirements of the air agency's air quality management program in light of each new or
789
-------
revisedNAAQS." Id. (emphasis added). EPA's proposal here does not find that Texas's SIP does
not contain the basic structural requirement needed to protect visibility—such as adequate
funding, monitoring and reporting requirements, and an enforcement program. Instead, EPA's
disapproval is premised on EPA's assertion that the Texas SIP does not meet the specific
requirements of a separate Clean Air Act provisions—the visibility provision at CAA §169A (42
U.S.C. § 7491). The specific requirements of CAA §169A are not "basic structural
requirements" that are required to be addressed in a § 110(a)(2)(D)(i)(II) infrastructure SIP.
Therefore, EPA is in error.
Luminant noted that the specific substantive requirements for a state's regional haze SIP are
found in CAA § 169A, not in § 110(a)(2)(D)(i)(II). It is axiomatic that the more specific
statutory provision controls over the more general one.874 Congress plainly intended § 169A to
provide the substantive requirements for a state's regional haze SIP, not § 110(a)(2)(D)(i)(II).
Certainly, § 110(a)(2)(D)(i)(II) does not create any additional requirements for Texas's regional
haze SIP beyond those in § 169A, and thus for all the reasons that EPA must approve Texas's
submission under § 169A as discussed in these comments, it must approve it under
CAA § 110(a)(2)(D)(i)(II).
SIP Timing. [Luminant (0061) p. 142; GCLC (0063) p. 18-19], Luminant and GCLC asserted
that EPA's proposed disapproval of Texas' infrastructure SIPs also conflicts with the timing of
NAAQS and regional haze SIP submittal deadlines. GCLC explained that there is no statutory or
regulatory syncing of time lines between the NAAQS and regional haze programs. Luminant
noted that regional haze SIPs must be submitted at specific intervals (every 10 years) and in
accordance with specific deadlines, which have nothing to do with the deadlines for NAAQS SIP
revisions.875 Luminant and GCLC explained that if EPA continues to proceed with this proposed
plan of effectively reading the substantive requirements of § 169A into § 110(a)(2)(D)(i)(II), it
would force states to disregard the regional haze SIP deadlines in favor of the NAAQS deadlines.
This result contradicts the clear requirements of the CAA and highlights the errors of EPA's
proposed action. Luminant noted that if a state fails to submit a regional haze SIP EPA would
then likely claim the authority and obligation to issue a FIP within two years addressing all of the
substantive requirements of a regional haze SIP, including establishing reasonable progress goals
and a long-term strategy. This outcome and EPA's position make no sense and contravene the
statute and EPA's own regulations.
Footnotes:.
874 Morales v. Trans World Airlines, Inc., 504 U.S. 374, 384 (1992) ("[I]t is a commonplace of statutory
construction that the specific governs the general. . .."); Gozlon-Peretz v. United States, 498 U.S. 395, 396 (1991)
("[A] specific [statutory] provision controls one of a more general application ....").
875 40 C.F.R. § 51.308(f).
Response: The EPA acknowledges the background information provided by the commenter.
We take no position with respect to the commenter's synopsis. We are not approving Texas's
Regional Haze submission under § 169A. EPA's rationale for disapproval of Texas' regional
haze plan is discussed elsewhere in this document. EPA has the authority to disapprove Texas's
110(a)(2)(D)(i)(II) interstate transport SIP revisions. "Section 110(a)(2) specifies the substantive
elements that infrastructure SIP submissions need to address, as appropriate, for EPA approval."
EPA, Guidance on Infrastructure State Implementation Plan (SIP) Elements under Clean Air Act
790
-------
Sections 110(a)(1) and 110(a)(2), at 1-2 (Sept. 2013) (EPA i-SIP Guidance) (emphasis added).
"Under section 110(a)(2)(D)(i)(II), an infrastructure SIP submission cannot be approved with
respect to prong 4 (visibility transport) until the EPA has issued final approval of SIP provisions
that the EPA has found to adequately address any contribution of that state's sources to impacts
on visibility program requirements in other states." EPA i-SIP Guidance at 32-33. This
requirement is substantive.
One way in which prong 4 may be satisfied for any relevant NAAQS is through
an air agency's confirmation in its infrastructure SIP submission that it has an
approved regional haze SIP that fully meets the requirements of 40 CFR 51.308 or
51.309. 40 CFR 51.308 and 51.309 specifically require that a state participating
in a regional planning process include all measures needed to achieve its
apportionment of emission reduction obligations agreed upon through that
process. See, for example, 40 CFR 51.308(d)(3)(ii). A fully approved regional
haze SIP will ensure that emissions from sources under an air agency's
jurisdiction are not interfering with measures required to be included in other air
agencies' plans to protect visibility. Id, at 33.
EPA has disapproved such SIPs for failure to comply with this provision for various other states.
See 78 FR 46142, July 30, 2013 (Arizona); 77 FR 14604, March 12, 2012 (Arkansas); 76 FR
52388, August 22, 2011 (New Mexico); 76 FR 81728, December 28, 2011 (Oklahoma). We
disagree that our disapproval is contrary to the CAA and that Texas's SIP submittals comply
with all requirements of the CAA.
With respect to the PM2.5, NO2 and ozone NAAQS submittals, as we discussed in our proposal,
we gave limited disapproval to the Texas Regional Haze SIP based on its reliance on CAIR.
CAIR provided limits on emissions of SO2 and NOx. SO2 is a precursor for PM2.5. NOx is a
precursor for ozone. NO2 is a component of NOx. With CAIR no longer in effect and
controlling Texas may not rely on its regional haze SIP to ensure that emissions from Texas do
not interfere with measures to protect visibility in nearby states.
We did not misconstrue the scope and intent of CAA section 110(a)(2)(D)(i)(II) as it explicitly
provides that states must have SIPs with adequate provisions to prevent interference with the
efforts of other states to protect visibility. See 76 FR 81728, 81738, December 28, 2011. As
noted in our proposal, when Congress enacted section 169A as part of the 1977 CAA
Amendments it also amended section 110 to require that all SIPs "contain adequate provisions
prohibiting . . . any source or other type of emissions activity within the State from emitting any
air pollutant in amounts which will ... interfere with measures required to be included in the
applicable implementation plan for any other State ... to protect visibility" (79 FR 74818, 74824,
December 16, 2014). The two sections work together. In many other i-SIP actions across the
country, we've allowed states to rely on their approved regional haze plan to meet the
substantive requirements of the visibility component of 110(a)(2)(D)(i)(II) because the regional
haze plan achieved at least as much emissions reductions as projected by the RPO modeling. See
76 FR 34608, June 14, 2011 (California); 79 FR 60985, October 9, 2014 (New Mexico); 76 FR
36329, June 22, 2011 (Idaho); and 76 FR 38997, July 5, 2011 (Oregon).
791
-------
Our position concerning the requirements of the visibility protection portion of CAA
110(a)(2)(D)(i)(II) does not contradict and render superfluous the Regional Haze SIP deadlines
and planning periods. In the SIP submittal for a new or revised NAAQS a state may refer to its
Regional Haze SIP, as Texas has done. So long as a state has an approved Regional Haze plan
upon which it relies to comply with 110(a)(2)(D)(i)(II), that requirement of the relevant NAAQS
SIP is met, regardless of the intervals in which these submittals are due. Likewise, if the state
relies on an unapprovable regional haze plan to meet 110(a)(2)(D)(i)(II), then the state's relevant
NAAQS SIP is not approvable. Because we found that the Texas Regional Haze SIP is deficient,
the interstate transport and visibility protection requirement of CAA 110(a)(2)(D)(i)(II) is not
met.
We proposed to rely on CSAPR and the emission reductions required in our Texas FIP to address
these deficiencies in Texas' SIP submittals, but we have determined that it is not appropriate to
finalize this determination at this time. Given the uncertainty following the D.C. Circuit Court's
partial remand of the CSAPR budgets, we do not consider it appropriate to rely on CSAPR at this
time to address the interstate visibility transport obligation with respect to visibility. Therefore,
today's action does not finalize the portion of our proposed FIP addressing Texas' visibility
transport obligations, as that portion of the FIP would have partially relied on CSAPR. We will
address the visibility transport requirements for Texas in a future rulemaking, once the issues
surrounding the partial remand are resolved.
Comment: [TCEQ/PUCT (0056) p. 18-19] The TCEQ disagreed with the EPA's proposed
disapproval of §110(a)(2)(D)(i) requirement for visibility protection for the Texas infrastructure
SIP submittals for ozone, particulate matter (PM2.5) nitrogen dioxide (NO2), and SO2 NAAQS.
The EPA fails to go into any detail on the reasons for disapproving these multiple, separate SIPs.
The TCEQ noted that for the 1997 eight-hour ozone standard, the EPA only states that Texas
originally failed to make a timely submission, and notes that CAIR was then promulgated and
implemented by the EPA. Texas was not in CAIR for ozone, and subsequently submitted a
separate transport SIP for the 1997 eight-hour ozone NAAQS. The EPA neglects to offer any
reason or explanation for why this submission was inadequate or deserving of disapproval, other
than the promulgation and implementation of the CSAPR. Although Texas was included in
CSAPR for the 1997 eight-hour ozone standard, Texas has from the beginning challenged that
inclusion, and litigation over the matter is on-going. Additionally, the EPA failed to act on, or
even mention the Texas ozone transport SIP submission before including Texas in CSAPR for
the 1997 ozone standard.
The TCEQ noted that for the 1997 PM2.5 NAAQS, Texas was included in CAIR, and
subsequently complied with CAIR requirements. The EPA included Texas in CSAPR for the
1997 PM2.5 NAAQS at final promulgation of the rule, without having given Texas proper notice
of this inclusion by including Texas in the proposed rule. Texas has challenged its inclusion in
CSAPR for the 1997 PM2.5 NAAQS, and litigation over this matter is also on-going. The linkage
of Texas to a single monitor in an area already attaining the relevant NAAQS is a clear case of
over-control, something explicitly prohibited by the CAA, as acknowledged by the Supreme
Court. See EPA v. EMEHomer City Generation, L.P., 134 S.Ct. 1584, at 1608 (April 29,
792
-------
2014). Texas also submitted a transport SIP for the 2006 PM2.5 NAAQS. Although this SIP did
rely on CAIR, the EPA has failed to offer any substantive reason why this is inappropriate, given
that CSAPR replaced CAIR, and the sole Texas linkage in the final CSAPR for 2006 PM2.5 are to
the same inappropriate monitor in an area already attaining the NAAQS.
As for the 2008 ozone, 2010 SO2, and 2010 NO2 standards, the TCEQ stated that Texas has
submitted transport SIPs for each of these standards demonstrating that Texas does not have
transported emissions out of state that interfere with attainment or maintenance in any downwind
state.
The TCEQ argued that the EPA fails to offer any rational or reasoned explanation for why these
SIP submissions are inadequate. In fact, the EPA fails to offer any analysis of these SIP
submissions at all; therefore, this proposed disapproval is arbitrary, capricious, and not
supportable.
Response: In the infrastructure SIP submittals for the ozone, PM2.5, NO2 and SO2 NAAQS
Texas indicated that the Regional Haze SIP, which relied on CAIR, fulfilled its obligation for
interference with measures required to be included in the SIP for any other state to protect
visibility. With respect to the ozone, PM2.5, NO2 and SO2 NAAQS, because CAIR is no longer
controlling Texas emissions of NOx and SO2 from EGUs the portions of the SIPs for these
NAAQS do not demonstrate that Texas emissions would not interfere with measures to be
included in the SIP for any other state to protect visibility. Additionally, with respect to the
PM2.5 and SO2 NAAQS, our finding that additional SO2 controls are needed to prevent
interference with measures required to be included in the Oklahoma SIP to protect visibility
means that the portions of the SIPs for these NAAQS do not demonstrate that that Texas
emissions would not interfere with measures to be included in the SIP for any other state to
protect visibility. We discuss EPA v. EME Homer City elsewhere in this document. In this
rulemaking we are taking no action on the portions of the SIP submittals that pertain to
prohibiting Texas emissions that interfere with attainment or maintenance of the NAAQS in any
downwind state (CAA 110(a)(2)(D)(i)(I)). We plan to address these portions of the SIPs in
separate rulemaking.
Comment: [Luminant (0061) p. 142] Luminant noted that § 7410(a)(2)(D)(i)(II) specifically
limits EPA's authority to require one state to adopt binding emission limits for the benefit of
another state. Section 7410(a)(2)(D)(i)(II) requires only that a SIP contain "adequate provisions
. . . prohibiting. . . any source or other type of emissions activity within the State from emitting
any air pollutant in amounts which will—. . . interfere with measures required to be included in
the applicable implementation plan for any other State under part C of this subchapter ... to
protect visibility ." Here, it is undisputed that actual visibility conditions at Oklahoma's Wichita
Mountains already meet the 2018 progress goals determined to be reasonable by EPA, without
any further reductions from Texas sources. Thus, EPA has no authority to require any further
reductions or to disapprove Texas's SIP revision on the basis that it does not include those
further reductions, as the Supreme Court has recently held in interpreting this very same "good
neighbor" provision. See EPA v. EME Homer City Generation, 134 S. Ct. 1584, 1608 (2014)
("EPA cannot require a State to reduce its output of pollution by more than is necessary to
793
-------
[protect visibility] in every downwind State or at odds with the [reasonable progress] threshold
the Agency has set. ").
Response: We disagree that CAA section 110(a)(2)(D)(i)(II) specifically limits our authority to
require one state to adopt binding emission limits for the benefit of another state. In previous
actions for New Mexico and Oklahoma, we have promulgated emission limits for the visibility
protection benefit of other states (76 FR 52388, August 22, 2011; 76 FR 81728, December 28,
2011). Further, this action does not require Texas to reduce its output of pollution by more than
necessary to protect visibility in Oklahoma or at odds with reasonable progress, rather, EPA is
requiring Texas to reduce visibility pollutants by the amount sufficient to protect visibility in
Oklahoma's Class I area as well as its own Class I areas. We address monitoring data for the
Wichita Mountains in our responses to other comments.
Comment: Multiple commenters noted that §110(a)(2)(D)(i) is pollutant-specific.
[TCEQ/PUCT (0056) p. 18-19] The TCEQ noted that EPA states that because it is proposing the
need for additional SO2 controls on Texas sources to prevent interference with measures required
to be included in the Oklahoma Regional Haze SIP to protect visibility, the EPA must therefore
disapprove the §110(a)(2)(D)(i) submittals for the 1997 PM2.5, 2006 PM2.5, and 2010 SO2
NAAQS. The EPA fails to offer any support for this contention, or the inclusion of the PM2.5
standards in this list. The EPA has repeatedly stated that infrastructure requirements, including
transport requirements, are pollutant specific. Therefore, a requirement to increase SO2 controls
does not, without further explanation, necessarily include the requirements for PM2.5 Although
the EPA has taken other actions in conflict with its guidance on this issue, there is no rational
reason to continue to perpetuate this error.
[Luminant (0061) p. 142; GCLC (0063) p. 19] Luminant noted that EPA errs by ignoring the
"pollutant specific" nature of § 110(a)(2)(D)(i)(II).877 EPA claims to be disapproving Texas's
SIP submissions for different NAAQS (1997 PM2.5; 1997 ozone; 2006 PM2.5, 2008 ozone; 2010
NO2 NAAQS, and 2010 SO2) but EPA only finds that Texas's SIP is inadequate to protect
visibility because it does not contain certain limitations on SO2 emissions. 79 FR 74888. EPA
finds no deficiency with respect to PM2.5, NO2, or ozone, and thus EPA's reliance on those
NAAQS is unsupported and arbitrary and capricious.
[Luminant (0061) p. 142; GCLC (0063) p. 19] Luminant and GCLC noted that EPA refers to
the "protect visibility" requirement of § 110(a)(2)(D)(i) as "Prong 4" in its guidance on
infrastructure SIPS, interpreting " this prong to be pollutant-specific, such that the infrastructure
SIP submission need only address the potential for interference with protection of visibility
caused by the pollutant (including precursors) to which the new or revised NAAQS applies.1,74
Luminant noted that EPA claims to be disapproving Texas's SIP submissions for different
NAAQS (1997 PM2.5; 1997 ozone; 2006 PM2.5, 2008 ozone; 2010 NO2 NAAQS, and 2010 SO2)
but EPA only finds that Texas's SIP is inadequate to protect visibility because it does not contain
certain limitations on SO2 emissions. 79 FR 74888. Luminant and GCLC stated that EPA's
action to impose disapprovals with respect to the PM2.5, NO2, and ozone NAAQS is completely
unsupported and arbitrary and capricious given that EPA has found no error in Texas' program
regarding those pollutants or a visibility effect from those pollutants. GCLC concluded that
794
-------
since Texas has fully complied with the requirements of the CAA through its SIP submission,
while also complying with the infrastructure SIP requirements, EPA must approve of Texas' SIP
submittals.
Footnotes:.
74EPA, Guidance on Infrastructure State Implementation Plan (SIP) Elements under Clean Air Act Sections
110(a)(1) and 110(a)(2), at 33 (Sept. 2013).
Response: We did not ignore the pollutant specific nature of CAA section 110(a)(2)(D)(i)(II) as
we interpret the visibility protection portion of CAA 110(a)(2)(D)(i)(II) to be pollutant specific
such that the infrastructure SIP submission need only address the potential for interference with
protection of visibility caused by the pollutant (including precursors) to which the new or revised
NAAQS applies. See September 13, 2013 EPA memo "Guidance on Infrastructure State
Implementation Plan (SIP) Elements under Clean Air Act Sections 110(a)(1) and 110(a)(2)",
page 54-55
(www3.epa.gov/airquality/urbanair/sipstatus/docs/Guidance_on_Infrastructure_SIP_Elements_
Multipollutant_FINAL_Sept_2013.pdf). In its regional haze SIP, Texas relied on CAIRto
address regional haze requirements for controlling emissions of NOx and SO2. As CAIR is no
longer a valid rule, the Texas SIP cannot ensure that regional haze requirements for controlling
emissions of NOx and SO2 are met. NOx and SO2 emissions have the potential for interference
with protection of visibility. As (1) NOx is a precursor of ozone and PM2.5 and (2) NO2 is a
component of NOx, the potential for interference with visibility protection for the ozone, PM2.5
and NO2 NAAQS apply to NOx emissions. As SO2 is (1) a pollutant with its own NAAQS and
(2) a precursor of PM2.5, the potential for interference with visibility protection for the SO2 and
PM2.5 NAAQS apply to SO2 emissions. As the Texas SIP cannot ensure that Texas NOx and
SO2 emissions will not interfere with visibility protection measures required for any other State,
we are disapproving those portions of the SIP submittals for the ozone, PM2.5, NO2 and SO2
NAAQS.
Comment: [Nucor Steel (0058) p. 1-2] Nucor Steel disagreed with EPA's interpretation and
determination that the Regional Haze Rule and the Clean Air Act allow EPA to disapprove the
Texas SIP and to therefore issue a Federal Implementation Plan on the basis that the Texas SIP
does not include "measures to prohibit emissions that would interfere with the reasonable
progress goals set to protect Class I areas in other states. "79 FR 74820. The visibility
protection requirement in §110(a)(2)(D)(i)(II) of the Clean Air Act does not require a SIP to
contain provisions that prohibit emissions that will interfere with visibility "reasonable progress
goals" of other states; it requires adequate provisions prohibiting emissions that will interfere
with " measures required to be included in the applicable implementation plan for any other
state." The SIP requirements for visibility are different from the language in §110(a)(2)(D)(i)(I)
for protection of the NAAQS, which requires adequate SIP provisions to prohibit emissions that
contribute to nonattainment of or that interfere with another state's maintenance of a national
ambient air quality standard. The visibility protection requirement in §110(a)(2)(D)(i)(II) is
narrower and requires only provisions necessary to prevent interference with control measures
included in another state's plan to achieve a visibility standard. Reasonable progress goals,
projected deciview improvements and the like are standards; they are not "measures", i.e. they
795
-------
are not actions taken or enforced by a state to achieve a standard or goal. EPA's interpretation
ignores and improperly eliminates the differences in the language Congress used for NAAQS
SIP requirements, and the language used for visibility SIP requirements. To the extent that
EPA's disapproval of the Texas and Oklahoma SIPS is premised on the language in
§110(a)(2)(D)(i)(II), but is not based on direct interference with a specific control measure in
another state's regional haze SIP (as opposed to interference with a regional haze related goal in
or underlying another state's SIP), EPA's interpretation is contrary to the clear and express
language of section 110 of the Clean Air Act.
Response: We agree that the CAA requirement is to prohibit sources within a state from
emitting any air pollutants in amounts that will interfere with measures required to be included in
the SIP for any other State to protect visibility. We disagree that the visibility protection
requirement in § 110(a)(2)(D)(i)(II) "requires only provisions necessary to prevent interference
with control measures included in another state's plan to achieve a visibility standard." As we
noted in our proposal, (79 FR 74818, 74820), we interpret §110(a)(2)(D)(i)(II) as requiring states
to include in their SIPs measures to prohibit emissions that would interfere with the reasonable
progress goals set to protect Class I areas in other states. We also noted that this is consistent
with the requirements in the regional haze program that explicitly require each state to address its
share of the emission reductions needed to meet the reasonable progress goals for surrounding
Class I areas (64 FR 35714, 35735, July 1, 1999). We are disapproving the Texas interstate
transport SIP because (1) Texas may not rely on its regional haze SIP, which relied heavily upon
CAIR, to ensure that emissions from Texas do not interfere with measures to protect visibility in
nearby states and (2) additional control of SO2 emissions in Texas is needed to prevent
interference with measures required to be included in the Oklahoma SIP to protect visibility.
The SO2 controls found in our FIP address in part Texas emissions that interfere with reasonable
progress at the Wichita Mountains Class I area in Oklahoma.
As previously discussed, we have allowed states to rely on their approved regional haze plan to
meet the substantive requirements of the visibility component of 110(a)(2)(D)(i)(II) because the
regional haze plan achieved at least as much emissions reductions as projected by the RPO
modeling. As noted previously, the visibility component of interstate transport may be met if it
has an approved regional haze SIP that fully meets the requirements of 40 CFR 51.308 or
51.309. 40 CFR 51.308 and 51.309 specifically require that a state participating in a regional
planning process include all measures needed to achieve its apportionment of emission reduction
obligations agreed upon through that process. We are disapproving the Texas Regional Haze
SIP submittal because it does not include all measures needed to achieve Texas' apportionment
of emission reductions obligations. We are not disapproving the Texas interstate transport SIP
submittals because of our disapproval of the RPGs or projected deciview improvements. We are
disapproving it because we are disapproving the Texas' Long Term Strategy because the analysis
underlying it was flawed. Because of these flaws, Texas' SIP submittal does not include all the
measures necessary to secure Texas' apportionment of the emission reductions needed to meet
the progress goal that should account for all reasonable control measures for the Wichita
Mountains and the two Texas Class I areas. See elsewhere our other responses to comments
concerning consultation.
796
-------
Comment: [Anonymous (0046) p. 1] The commenter stated that throughout the proposed rule,
EPA Region 6 consistently refers to Clean Air Act Section 110(a)(2)(D)(i)(II) (prongs 3 and 4)
as the "good neighbor" provisions when addressing visibility requirements. This is inconsistent
with EPA's past informal naming convention that indicate Section 110(a)(2)(D)(i)(I) (prongs 1
and 2) is the "good neighbor" provision. Examples of this include EPA's own website,
www.epa.gov/airtransport, and EPA's Jan. 22, 2015 memo from Stephen Page, both describe the
"good neighbor" provision as (D)(i)(I) and not (D)(i)(II). The commenter suggested that EPA
Region 6 change the description throughout the final rule, when issued, so that only prongs 1 and
2 are the "good neighbor" provisions so as to eliminate any confusion between contribution to
nonattainment and interference with maintenance (prongs 1 and 2 - the "good neighbor"
provisions) and interference with PSD and visibility requirements (prongs 3 and 4).
Response: We agree that EPA Region 6 and other EPA offices have not been consistent when
referencing the Clean Air Act "good neighbor" provisions. In our August 15, 2006 guidance
memo "Guidance for State Implementation Plan (SIP) Submissions to Meet Current Outstanding
Obligations Under Section 110(a)(2)(D)(i) for the 8-Hour Ozone and PM2.5 National Ambient
Air Quality Standards" we referred to the "good neighbor" provisions as those covering all four
requirements (or prongs) in CAA sections (D)(i)(I) and (D)(i)(II)
(http://www.epa.gov/ttn/naaqs/aqmguide/collection/cp2/20060815_harnett_final_section_110(a)(
2)(D)(i)_guidance.pdf). The final rule does not use the term "good neighbor", but instead refers
to the interstate visibility transport requirement of CAA 110(a)(2)(D)(i)(II).
Comment: Transport [ODEQ (0079) p. 2] ODEQ supports EPA's proposed remedy to address
the impacts of anthropogenic emissions from Texas. The Texas and Oklahoma SIPs agree that
transport of air pollutants from Texas causes the majority of visibility impairment at Oklahoma's
Class I Area so the emission reductions from Texas sources should be the most effective means
to improve visibility in the Wichita Mountains. However, ODEQ noted that it is ironic that the
CAA and EPA regulations justify more stringent actions to protect visibility by reducing
transport than they do to protect public health under the ozone NAAQS by preventing the
transport of ozone and ozone precursors to our downwind state during ozone season. While all
areas in Oklahoma are currently designated attainment/unclassifiable, ODEQ special studies of
ozone concentrations in counties along the Texas border demonstrate those areas are
significantly impacted by transport from Texas.
Response: We appreciate ODEQ's support of our proposed remedy. ODEQ's comments
concerning requirements for the ozone NAAQS are outside the scope of this rulemaking.
17. Disapproval of the Oklahoma and Texas Reasonable Progress Goals
Comment: [ODEQ (0079) p. 1] The ODEQ stated that it understands and expects that no
emission reductions or other additional steps will be required of sources in Oklahoma under
EPA's proposed FIP. Oklahoma firmly believes that further additional reductions from sources
in Oklahoma are not needed, and if EPA makes such a determination, Oklahoma expects that
EPA will accept comment on such a finding in a proposed rule.
797
-------
Response: The commenter's understanding that no additional emissions reductions are required
from sources in Oklahoma under our proposed FIP is correct. Our final FIP does not impose
requirements on any sources in Oklahoma.
Comment: EPA Must Disapprove Oklahoma's RPGs. [Earthjustice (0067) p.25] Earthjustice
et al., stated that the EPA's proposed disapproval of Oklahoma's RPGs for Wichita Mountains,
see 79 Fed. Reg. at 74,864-73, is proper and required by the Clean Air Act. Oklahoma set RPGs
for the Wichita Mountains that do not reflect any reasonable emission reductions from Texas
beyond those that will be achieved by compliance with other requirements of the CAA. Under
Oklahoma's RPGs, the Wichita Mountains would not have achieved natural visibility until
2102-nearly four decades after the 2064 goal set forth in the Regional Haze Rule. Id. at 74,865.
This rate of progress cannot be characterized as "reasonable progress" toward the statutory goal
of remedying any existing visibility impairment, especially because the record is clear that
control measures satisfying the four reasonable progress factors are available for some of the
largest sources of visibility impairment at Wichita Mountains, yet Oklahoma's RPG does not
reflect installation and operation of these reasonable controls. Accordingly, EPA properly
disapproved Oklahoma's inadequate and unreasonably delayed progress goals.
Earthjustice et al., stated that Oklahoma's RPGs also failed to satisfy the Regional Haze Rule
because Oklahoma did not consult with Texas adequately. Texas coal plants, such as Big Bend
[sic] and Monticello, are the worst sources of visibility impairment at Wichita Mountains. The
visibility impairment caused by these Texas sources is far greater than the visibility impacts of
Oklahoma's own sources. Id. at 74,866-67. In order to engage in a meaningful consultation, an
upwind state such as Texas must provide other states with sufficient technical information
detailing the visibility impacts of individual sources and the feasibility and cost-effectiveness of
control measures on those sources. A downwind state such as Oklahoma must take a hard look
at this information and request that upwind states require the control measures that satisfy the
four factors laid out in the statute for making reasonable progress.32 As EPA explained, the
consultation process must "ensure that all states, including downwind states, take a hard look at
what measures are necessary for ensuring reasonable progress towards improving and
maintaining visibility at Class I areas." Id. at 74,867.
Earthjustice et al., stated that, because Texas did not provide Oklahoma adequate information,
EPA correctly concluded that Oklahoma should have requested that Texas further investigate
control measures on these sources and require appropriate control measures. Id. at 74,867,
74,871. Earthjustice et al., strongly supported EPA's conclusions as to what constitutes a proper
and meaningful consultation under the regional haze program. Given the regional, interstate
nature of regional haze, it is imperative that states substantively consult with each other and set
reasonable progress goals based on sources in other states installing control measures that meet
the four factors for reasonable progress.33 Oklahoma did not do so here and, therefore, EPA has
no choice but to disapprove the state's RPGs.
Footnotes:
798
-------
32 The reasonable progress goals must be based on measures that will result in natural visibility conditions by 2064,
unless the state demonstrates, based on a consideration of the four reasonable progress factors, that achieving the
2064 goal is unreasonable and the alternative goal is reasonable. 40 C.F.R § 51.308(d)(l)(ii).
33 As mentioned above, the reasonable progress goal for each area must ensure that natural visibility conditions are
achieved in 2064, unless the state demonstrates that this is unreasonable and an alternative goal that makes slower
progress is reasonable. 40 C.F.R 51.308(d)(l)(ii).
Response: We appreciate the commenter's support of our proposed disapproval of Oklahoma's
RPGs for the Wichita Mountains and of our proposed disapproval of the consultations between
Oklahoma and Texas to address the impact of emissions from sources in Texas at the Wichita
Mountains. We assume that the reference above to Big Bend in the sentence "Texas coal plants,
such as Big Bend and Monticello, are the worst sources of visibility impairment at Wichita
Mountains" actually refers to Big Brown.
Comment: [Luminant (0061) p. i and 1] Luminant stated that EPA should withdraw its proposal
and instead fully approve Texas's and Oklahoma's regional haze SIPs. Luminant stated that
there is no legal or technical basis for EPA's proposed FIP because the SIP submitted by Texas
fully complies with the statute and all regulatory standards. According to Luminant, the Clean
Air Act's regional haze program is about making reasonable incremental improvements to
visibility at national parks and certain other federal areas—it's not about what is purported to be
technologically possible or achieving alleged potential health benefits. Here, even though the
visibility goals EPA proposes for Texas and Oklahoma are already being met—as evidenced by
real-world monitoring data—EPA's proposal would require Texas to spend $2 billion for what
EPA projects would be no perceptible improvement in visibility. On every level, EPA's
proposal exceeds the agency's authority under the Clean Air Act and EPA's regional haze
regulations. As explained elsewhere in their comments, Luminant stated that EPA's proposal
suffers from many fatal legal and technical flaws, including:
1. EPA ignores state primacy
2. EPA is applying an unlawful standard to Texas
3. EPA's methodology is contrary to the Clean Air Act
4. EPA invents new requirements for Texas sources
5. EPA arbitrarily disapproves the consultation between Oklahoma and Texas
6. EPA's proposal is unnecessary and outside of its authority because EPA's visibility
goals have already been achieved
7. Texas sources have negligible impacts under EPA's own thresholds
8. EPA's proposal would achieve no detectable change in visibility
9. EPA treats Texas differently than every other CSAPR state
10. The costs of EPA's proposal are extreme and unjustified
11. EPA's proposal exceeds its regulatory authority
Response: We address these general comments in our responses to more specific comments
elsewhere in this document.
Comment: Luminant provided a summary of EPA's proposed disapproval of Texas's and
Oklahoma's RPGs. [Luminant (0061), p. 43] Based on its unprecedented and fatally flawed
799
-------
"additional analysis," EPA proposes to disapprove Texas's RPGs for Big Bend and Guadalupe
Mountains, Texas's long-term strategy, and Oklahoma's RPGs for Wichita Mountains. EPA
agrees with Texas's conclusion "that it is not reasonable to provide for rates of progress at...
Big Bend or the Guadalupe Mountains that would attain natural visibility conditions by 2064
(i.e., the URP)" and that a "slower rate of progress is reasonable."307 EPA also does not find
error in Texas's assessment of any of the four factors, but instead claims that "TCEQ's analysis
is insufficient to determine the visibility benefit of controlling the source or subset of sources
with the most effective controls for improving visibility conditions," although EPA fails to point
to any regulatory requirement to do so.308 As to Oklahoma, EPA does not take issue with
Oklahoma's four-factor analysis, but instead proposes to "reset Oklahoma's RPGs based on our
[Texas small group] analysis."309
As to the RPG for the Wichita Mountains, EPA finds that disapproval is necessary because "the
RPGs selected by the ODEQ for the Wichita Mountains do not include" the emission reductions
from the "SO2 BART FIP and the revised BART SIP for the AEP units" that have subsequently
been imposed.310 However, EPA's proposed FIP does not correct this error, nor does EPA
calculate or model visibility conditions in 2018 at Wichita Mountains with the BART controls in
place. EPA also finds that, even though Oklahoma "did not request that the TCEQ further
investigate" emission reductions from certain sources, Texas somehow "denied [Oklahoma] the
knowledge it needed—the visibility impacts of individual sources in Texas with the largest
potential to impact the visibility at the Wichita Mountains and the extent to which cost-effective
controls were available—in order to properly construct its RPG for the Wichita Mountains."311
EPA then attempts to "quantif[y]" new RPGs for the three Class I areas.312 EPA "quantifies]"
the new RPGs using "the results of our reasonable progress analysis of point sources as
described in detail in our FIP TSD."313 However, in doing so, EPA does not use the visibility
modeling from ENVIRON or its own unit-level quantification of benefits that were used to
justify the controls EPA determined to impose in its FIP. Nor does EPA model the visibility
conditions in light of the recently finalized BART requirements on Oklahoma sources. Instead,
EPA reverts back to the original CENRAP modeling used by Texas and Oklahoma to quantify
the states' original RPGs. EPA apparently subtracts from the 2018 CENRAP projection the so-
called visibility improvement that EPA separately estimated (under different modeling) from the
controls it would impose on Texas sources by 2018. EPA cites no other examples of states or
EPA calculating an RPG in this manner. Moreover, in their total aggregate effect, these so-
called improvements in visibility from the controls EPA is mandating are even smaller than
when viewed at the unit level. In other words, the individual visibility benefit that EPA modeled
for each controlled unit (which forms the basis for EPA's decision to regulate) overstates the
overall benefit that would actually be achieved in reality at each Class I area in 2018.
The differences between the states' RPGs and EPA's proposed RPGs are miniscule (0.03, 0.04,
and 0.14 dv for Big Bend, Guadalupe mountains, and Wichita mountains, respectively).314 EPA
only proposes new RPGs for the 20% worst days for these areas, but not for the 20% best
days.315
Footnotes:
307 Id. at 74,887.
308 Id. at 74,841.
800
-------
309 Id. at 74,889.
310 Id. at 74,871.
311 Id. at 74,871, 74,873.
312 Id. at 74,886.
313 Id.
314 Id. at 74,887, tbl.43.
315 See id. at 74,887.
Response: We acknowledge the background information provided by the commenter. We take
no position with respect to the commenter's synopsis. We address these comments in our
responses to more specific comments elsewhere in this document.
Comment: [TCEQ/PUCT (0056) p. 9] The TCEQ agreed with the EPA's proposal to find that
Texas' submission meets the requirements of §51.308(d)(l)(iv) regarding reasonable progress
goal minimum and state consultations for the two Texas Class I areas.
Response: We appreciate the TCEQ's support of our proposed determination that the regional
haze SIP submitted by Texas meets the reasonable progress goal minimum requirement of
section 51.308(d)(l)(vi) and the consultation requirement of section 51.308(d)(l)(iv). We are
finalizing this determination.
Comment: [TCEQ/PUCT (0056) p. 9-10] The TCEQ stated that the EPA's proposed
disapproval of Texas' RPGs and its substitution with new RPGs in the proposed FIP is based on
EPA's flawed interpretation of what the CAA requires for "reasonable progress goals." This
action is based on the EPA's conclusion that "reasonable progress" must be determined based on
source-specific cost of controls even though such a requirement did not exist in the statute, the
RHR, or the guidance available in 2009.
The TCEQ noted that the Texas 2009 RH SIP established RPGs for both Big Bend and
Guadalupe Mountains that provide for visibility improvement for the most impaired days over
the period of the SIP and ensure no degradation in visibility for the least impaired days over the
same period. The EPA agrees the SIP meets these requirements. The EPA also agrees that the
TCEQ considered the four statutory factors in establishing the RPGs for its Class I areas, in
accordance with the RHR. The RHR requires states to establish RPGs that" .... must provide
for an improvement in visibility for the most impaired days over the period of the
implementation plan and ensure no degradation in visibility for the least impaired days over the
same period"(§51.308(d)(1)). The four statutory factors in subparagraph (i) are factors the state
must consider in developing the RPGs. These factors in and of themselves do not determine the
reasonableness of the goals for the planning period. The RHR, in 40 CFR 51.308((d)(l)(iii),
requires the EPA to evaluate whether the state's goal for visibility improvement provides for
reasonable progress based on a demonstration of which the four statutory factors are only one
element.
Response: The CAA requires that the implementation plan for a State contain such emissions
limits, schedules of compliance, and other measures "as may be necessary to determine
reasonable progress." The CAA additionally dictates the factors that shall be taken into
801
-------
consideration in determining reasonable progress. The recitation of those factors does not make
approvable an unsound analysis and demonstration, which is the case here. To the extent the
comment means to assert that meeting the terms of 51.308(d)(1) can justify a flawed or weak
demonstration and analysis of reasonable progress under the statutory factors, we disagree. The
comment's citation to 40 CFR 51.308(d)(l)(iii) in fact reinforces our position that we have the
task of "evaluating the demonstrations developed by the State." In so doing, we have the
authority to disapprove any demonstration that is not approvable and fails to meet the
requirements, which again is the case here. The bases for partial disapproval of the SIP were
detailed in our proposal and have been elaborated on at length in the responses we offer to more
specific comments. We do not agree with the comment's assertion that our action is based on a
particular conclusion regarding "source specific costs of controls," rather than the SIP's failure to
meet CAA requirements, and we offer more detailed responses to comments on "source specific"
approaches and costs locatable in several sections of this document.
See our specific responses to the other comments elsewhere in this document.
Comment: [AEP (0055) p. 2] AEP stated that the EPA has provided new guidance with its
"interpretations" that fails to demonstrate that the Texas and Oklahoma SIPs do not represent
reasonable progress; especially given the length of time to achieve the goal by 2064. EPA also
fails to take into consideration recent information, the TCEQ's 2014 Five-Year Regional Haze
SIP Revision or the effects of early action or emission reduction accomplished or to be
accomplished by other EPA programs before imposing additional requirements beyond the state
submitted SIPs. Considering that the visibility improvements of these programs have not yet
been quantified, and the gradual progress anticipated in establishing such a long term goal, EPA
should be patient and not take such aggressive action in overriding reasonable state SIPs and
imposing additional controls. EPA's actions are not timely and will not be realized during the
progress planning period and are therefore unlawful.
Response: We discuss AEP's comment regarding "new guidance" and "interpretation" in this
document in the sections where we discuss our Clarified Interpretation of the Reasonable
Progress and Long-Term Strategy Requirements, as well as our final action. We disagree with
AEP's contention that we have failed to demonstrate that the Texas and Oklahoma SIPs do not
represent reasonable progress. Our rationale for why the RPGs established by Texas and
Oklahoma do not represent reasonable progress was presented in our proposed rule and is
discussed further in this document. AEP points to "the length of time to achieve the goal by
2064," in support of its contention that the Texas and Oklahoma SIPs represent reasonable
progress. It is not clear what point AEP intended to make with this statement, but in so far as it
is contending that 2064 is far enough into the future that there will be sufficient time to achieve
progress in future planning periods such that there is no need to require additional controls for
Texas sources in this planning period, we remind the commenter that under the Texas and
Oklahoma RH SIPs, Texas and Oklahoma Class I areas are not on the glide path to achieve the
national goal of natural visibility conditions by 2064 as calculated by the respective states.
Texas calculated that under the rate of progress it selected as reasonable, it would attain natural
visibility conditions at Big Bend in 2155 and at the Guadalupe Mountains in 2081. Oklahoma
calculated that under the rate of progress it selected as reasonable, it would attain natural
802
-------
visibility conditions at the Wichita Mountains in 2102. Therefore, we disagree that the amount
of time available to achieve the national goal of natural visibility conditions justifies full
approval of the Texas RH SIP and the remaining portions of the Oklahoma RH SIP.
With regard to the TCEQ's 2014 Five Year Progress Report SIP revision, we stated in our
proposal that the TCEQ submitted the first five-year report in March 2014, but we are not
including our analysis of that SIP revision within this action.641 The five-year progress report is a
requirement that is separate from the regional haze SIP required for the first planning period as
well as the subsequent planning periods, and it has separate content and criteria for us to review.
We therefore believe we are not obligated to consider or take action on the five year progress
report at the same time we take action on the regional haze SIP for a planning period, such as this
Texas regional haze SIP for the first planning period. Even so, we acknowledge that recent
monitoring data from IMPROVE monitors indicates that the five-year average measurements of
visibility extinction at Texas and Oklahoma Class I areas on the 20% worst days are lower (i.e.,
indicate better visibility conditions) than the numerical RPGs we are establishing for these Class
I areas. This is addressed in detail elsewhere in this document. Notwithstanding the greater than
anticipated visibility improvement that has taken place at Texas' and Oklahoma's Class I areas,
we continue to hold that the states' RPGs are not approvable because the states have not
demonstrated that their RPGs constitute reasonable progress based on reasonably considering the
four reasonable progress factors under section 51.308(d)(1). We remind the commenter that the
RPGs, unlike the emission limitations that apply to specific sources with BART requirements
and reasonable progress requirements, are not directly enforceable. Rather, the RPGs are an
analytical tool used by EPA and the states to evaluate whether measures in the SIP are sufficient
to achieve reasonable progress. As discussed in more detail in our proposal and in our final
action, Texas has not satisfied all the requirements under section 51.308(d)(1) and (3) in
determining whether additional controls for Texas sources are reasonable. We continue to hold
that the emission limitations we are establishing for specific sources in Texas in our final
action's FIP constitute reasonable progress for this planning period and were arrived at in
accordance with the CAA and the regional haze regulations. As discussed in our proposal and in
the TSD associated with it, we estimated the revised numerical RPGs for the 20% worst days in
2018 for Texas' two Class I areas and Oklahoma's Class I area to reflect the additional
anticipated visibility benefit from the controls in our FIP that will be in place by 2018.642 We did
not conduct additional photochemical modeling to calculate the revised RPGs, but instead we
adjusted the 2018 RPGs established by Texas and Oklahoma that were based on the CENRAP
modeling projections for 2018. We recognize that the RPGs we calculated are not as refined as
they would be had we conducted photochemical modeling to estimate them. We discuss this
issue in greater detail in the modeling responses elsewhere in this document.
We disagree with the commenter's contention that we should not impose additional controls on
Texas sources due to potential visibility improvements that have not yet been quantified resulting
from early actions and emission reductions accomplished or expected to be accomplished
through other EPA programs. If it is determined based on the demonstrations developed
pursuant to section 51.308(d)(l)(i) and (d)(l)(ii) that there are reasonable and cost-effective
controls available that would provide for reasonable progress, the statute and regional haze
641 79 FR 74864
642 79 FR 74886.
803
-------
regulations do not allow for a delay in requiring these controls to allow time for the
quantification and consideration of possible future visibility improvements.
Comments contending that our action is unlawful because it is not timely and the FIP
requirements will not be realized during the first planning period are addressed in the document
installation of controls beyond the first planning period.
Comment: [TCEQ/PUCT (0056) p. 4] The TCEQ stated that the Texas 2009 RH SIP, as
submitted, would ensure more than Texas' proportional contribution to progress toward
improved visibility conditions at Wichita Mountains through the first planning period that runs
through 2018.
The TCEQ stated that by 2018, Texas' 2009 RH SIP reduces Texas' apportioned contribution to
total visibility extinction at Wichita Mountains by more (26.1%) than the reduction from all
other states combined (24.5%). Also, Texas' 2009 RH SIP reduces Texas' visibility impairment
impact at Wichita Mountains by slightly more than its proportional share of the total baseline
visibility impact at Wichita Mountains. Additionally, the CENRAP states were in agreement
about the amount of progress that was reasonable at Wichita Mountains during the first planning
period.
The TCEQ stated that the EPA's proposed partial SIP disapproval and partial FIP undervalue the
effectiveness of the long-term strategy embodied in the Texas 2009 RH SIP. Without presenting
evidence, the EPA dismisses the progress made as being due to "meteorological conditions,
reduction in the impacts from SO2 emissions, and a reduction in the impacts from coarse
materials" (79 FR 74843). The EPA makes the meteorological assertion in spite of the fact that
2011 was one of the hottest and driest years in Texas history and there were unprecedented
wildfires that year. The current visibility conditions in Big Bend, Guadalupe Mountains, and
Wichita Mountains are already better than the respective state-established and the EPA-proposed
RPG for these three Class I areas.
Response: In support of Texas' 2009 RH SIP, the TCEQ contends that the SIP reduces Texas'
apportioned contribution to total visibility extinction at Wichita Mountains by 26.1%, while the
total reduction to the apportioned contribution from all other states is 24.5%. We do not believe
that the information presented by the TCEQ warrants full approval of Texas' 2009 RH SIP. We
discuss in our proposal, in our final rule, and elsewhere in this document why we are partially
disapproving the Texas RH SIP and promulgating a partial FIP. Furthermore, we remind the
TCEQ that other states individually have a very small baseline visibility impact on the Wichita
Mountains, especially when compared to Texas' impact. As a state's individual baseline
visibility impact on a Class I area becomes smaller, it is reasonable to expect that it will become
more difficult to find emissions reductions in that state that are both cost-effective and result in a
sizable improvement in visibility at the affected Class I area. Therefore, it is not unreasonable
for the total percent reduction in the apportioned contribution to total visibility extinction at
Wichita Mountains from all other states combined to be lower than the percent reduction in
Texas' apportioned contribution.
804
-------
As we discussed in our proposal, both states failed to meet the consultation requirements. See
the consultation discussion elsewhere in the document for more information. In summary,
Oklahoma had sufficient evidence from the CENRAP analyses to know it should have requested
that Texas further investigate controls at certain Texas sources or explicitly request Texas to
obtain additional reductions from Texas sources to address the impacts of emissions from these
sources at the Wichita Mountains. Because of the Texas flawed LTS analysis investigating
controls at certain Texas sources for Wichita Mountains, however, we believe that Oklahoma did
not have adequate information to reasonably establish its RPGs for the Wichita Mountains. The
record indicates that both Oklahoma and Texas acknowledged during the development of their
respective RH SIPs that based upon the CENRAP results, Texas point sources have a significant
visibility impact at the Wichita Mountains and that cost-effective controls were likely available
for these sources. However, this matter was not pursued further during the consultations
between the two states. We discussed in our proposal that there was a lack of development of
critical information related to Texas sources during Oklahoma's consultations with Texas and
other states.643 Texas did not provide the information necessary to identify reasonable reductions
from its sources, and inform consultations in order to develop coordinated management
strategies with Oklahoma. Considering the absence of the development of critical information
regarding Texas sources, the states did not have sufficient information to make a reasoned
decision about the amount of progress that was reasonable at Wichita Mountains during the first
planning period. Regardless of the agreement between the states, the basic intent of the
consultation requirements under the RH rule were not met.
Comments contending that current visibility conditions at the three Class I areas are already
better than the RPGs established by the states and the RPGs proposed by EPA are discussed in
the section where we respond to comments alleging we cannot disapprove the Texas SIP because
the RPGs are already being met. Comments contending that we dismiss the progress made as
being due to meteorological conditions, reduction in the impacts from SO2 emissions, and a
reduction in the impacts from coarse materials are addressed in the section of this document
where we discuss modeling issues.
Comment: No additional controls [on Limestone Units 1 and 2] are needed to assure
reasonable progress. [NRG (0078) p. 2] NRG stated that the EPA has proposed to require
substantially upgraded SO2 controls on NRG's Limestone Electric Generating Station Units 1
and 2 on the basis that these controls are necessary to assure reasonable progress in reducing
regional haze. NRG believes that such controls are unnecessary to achieve reasonable progress,
and are objectively unreasonable. Thus, EPA should not disapprove and replace the reasonable
progress goals or impose the additional controls on Limestone Units 1 and 2 contemplated by the
proposal.
Response: We discuss the issue of why we have disapproved Texas' reasonable progress and
long-term strategy demonstrations in our responses to more detailed comments elsewhere
throughout this document. In summary, we believe that Texas's reasonable progress and long-
term strategy demonstrations are flawed and must be disapproved. Our own analysis
demonstrates that scrubber upgrades to Limestone Units 1 and 2 are cost-effective, will result in
643 79 FR at 74867 and 74871.
805
-------
significant visibility improvement, and should be included in our FIP in order to address Texas'
reasonable progress and long-term strategy.
Comment: The Regional Haze Process Must Be Implemented Reasonably Going Forward.
[EEI (0076) p. 2-4] According to EEI, the CAA regional haze program tasks states with
determining what is reasonable progress toward elimination of man-made visibility impairment,
for which EPA has set a goal of 2064, along with specific progress milestones (10-year planning
and SIP revisions, with program reviews in the middle of the 10-year planning periods).2 The
regional haze program contemplates gradual visibility improvements along a "glide path" that
considers the 2064 goal, and does not require immediate reductions that exceed making
"reasonable progress", as determined by the state based on four statutory factors,3 in the first
planning period through 2018 or in any subsequent planning period. Thus, it neither requires nor
authorizes the frontloading of extensive control requirements. Instead, the regional haze
program should be implemented in a manner that allows states, through state environmental and
electric utility regulators and in conjunction with power companies, to plan the optimal timing of
emission control projects. This planning process is vital in order to minimize impacts on the cost
and reliable provision of electric power and to allow investment decisions to be made over
suitable planning horizons. It is additionally justified given the, at best, minimal visibility
benefits EPA claims would be achieved in the proposed rulemaking.
Thus, as EPA and the states begin to implement the next rounds of the regional haze program to
continue reasonable progress, EEI suggested that EPA should allow states to consider the timing
and scope of additional control activities, consistent with effective long-term utility planning.
EPA also must take into consideration the progress already made through the installation of
controls to satisfy Best Achievable Retrofit Technology (BART) requirements and BART-
equivalent measures such as the Cross-State Air Pollution Rule (CSAPR) and through other
CAA regulations that can result in reduced emissions that may contribute to visibility
impairment.
Further, EEI noted that it will be necessary to use the most up-to-date and accurate
implementation tools available as EPA and the states move forward with the regional haze
reasonable progress determinations. To that end, EPA should update both its atmospheric
modeling platforms as part of the upcoming Appendix W rewrite and the cost manual in order to
support reasonable future assessments of visibility impacts and appropriate control strategies.
The Agency should do this consistent with the Consolidated Appropriations Act of 2014. See
H.R. 3547, P.L. 113-76. EPA also should consider the latest available visibility monitoring
data for Class I areas when assessing potential controls needed to meet RPGs.
Response: EEI's contention that the regional haze program does not require immediate
reductions that exceed making reasonable progress as determined by the state based on the four
statutory factors is premised on the assumption that the emissions reductions that are part of the
state's long-term strategy and upon which its RPGs are based do in fact constitute reasonable
progress. We remind the commenter that the determination of what constitutes reasonable
progress must be made pursuant to section 51.308(d)(1). Based on its analyses under section
51.308(d)(1), a state (or EPA in the context of a FIP) may determine that a greater or lesser
806
-------
amount of visibility improvement than what is needed to get on the glide path is what constitutes
reasonable progress.644 As discussed in our proposal and our final action, we disagree with the set
of potential controls identified by the TCEQ as having the greatest impact on visibility on the
three Class I areas and how it analyzed and weighed the four reasonable progress factors in a
number of key areas.645 Therefore, we proposed to disapprove Texas' RPGs for its Class I areas
and conducted our own analysis of the four reasonable progress factors to fill in the regulatory
gap that would be created by our disapproval action. Based on our analyses under section
51.308(d)(1), we determined what level of control for Texas sources constitutes reasonable
progress and proposed this level of control in our FIP proposal. Therefore, we disagree that the
level of emissions reductions in the Texas regional haze SIP constitutes reasonable progress and
we disagree with the commenter's contention that we are frontloading extensive control
requirements for Texas sources, and we explain this in greater detail elsewhere in this document,
in our final action, and in our proposal. We are replacing Texas' flawed reasonable progress
analysis with our own and are finalizing the cost-effective reasonable progress controls we
proposed on the small number of Texas point sources that have the greatest visibility impacts on
the Class I areas of interest.
With regard to the comment that the regional haze program should be implemented in a manner
that allows for planning the optimal timing of emission control projects in order to minimize
impacts on the cost and reliable provision of electric power and to allow investment decisions to
be made over suitable planning horizons, the commenter does not specify how the regional haze
program should be implemented to allow for such planning. In so far as the commenter is
contending that delaying the requirement to install controls at electric utilities until a later
planning period would allow for planning the optimal timing of these emission control projects,
the CAA and regional haze regulations do not provide for delaying these requirements when it
has been demonstrated through our analysis that there are reasonable controls currently available
that would be cost-effective and would result in reasonable progress at the affected Class I areas.
We acknowledge the commenter's concerns and suggestions with regard to our and the states'
work on the RH SIPs for future planning periods. However, these types of comments are outside
the scope of this rulemaking. We agree with EEI that the states and we should always strive to
use the most up-to-date and accurate implementation tools available. We provide information
concerning this issue in the section of this document where we discuss stay of effective date and
the Consolidated Appropriations Act. Comments with regard to the latest available IMPROVE
monitor data for Big Bend, Guadalupe Mountains, and Wichita Mountains are addressed in the
section where we respond to comments alleging that we cannot disapprove the Texas SIP
because RPGs are already being met. Comments on updates to Appendix W and the cost manual
are addressed in our responses related to the Consolidated Appropriations Act.
Comment: EPA Failed to Conduct the Same Four Factor Reasonable Progress Analysis for
Oklahoma as it Did for Texas.
Luminant stated that EPA fails to even consider the four statutory factors with respect to non-
BART sources in Oklahoma that are impacting visibility at Wichita Mountains and to make a
644 64 FR 35732
645 79 FR 74838
807
-------
determination of whether additional controls are reasonable for those sources. In its SIP,
Oklahoma identified several sources in the states—including EGUs—that were not subject to
BART, but that are leading sources of SO2 emission in the state, including one of the top four
point source emitters of both SO2 and NOx (Grand River Dam Authority). Luminant submitted
the information in the following table:
Q/D Calculations for Three Oklahoma Sources Not
Included in EPA's Reasonable Progress Analysis
Facility
SO2 Emissions
(tpy, 2009)
NOx
Emissions
(tpy, 2009)
Distance to
Wichita
Mountains
(km)
Q/D for
Wichita
Mountains
Grand River Dam
Authority (GDRA)
(Oklahoma)
18,010.86
14,209.48
350
92.1
Western Farmers Electric
Cooperative—Hugo Plant
(Oklahoma)
9,709.60
3,207.01
321
40.2
Great Lakes Carbon
Corporation (Oklahoma)
3,929.26
558.35
214
21.0
EPA failed to consider the visibility benefit from imposing the same levels of control on these
(Oklahoma) sources as it is proposing to impose on the targeted Texas sources, and to determine
whether all existing and reasonable controls on Oklahoma sources, including BART, are
sufficient to attain a reasonable rate of progress for Wichita Mountains for the first planning
period. EPA apparently does not consider it reasonable to regulate these sources, yet their
visibility impact may the same as the Texas sources EPA does seek to regulate. According to
Luminant, EPA is applying a different standard to Texas sources than it is to sources in other
states.
EPA's analysis assumes, but does not demonstrate, that Texas sources are having greater impact
at Wichita Mountains than these Oklahoma sources. EPA offers no explanation of why it failed
to conduct the modeling and perform the statutory analysis that it would expect a state to conduct
in determining a reasonable progress goal. Had EPA not artificially limited its analysis to just a
few Texas sources, the likely outcome would be that Texas sources would show the same
minimal impact as these Oklahoma sources and other sources that EPA excluded—further
demonstrating that additional controls are unreasonable and unnecessary for all, as Oklahoma
and Texas concluded. The perceived error in Texas's long-term strategy hinges entirely on
EPA's "reset" of Oklahoma's RPG; there is no dispute that Texas's long-term strategy is
adequate to meet the RPG as established by Oklahoma.
EPA's "reset" RPG is unlawful. EPA has no basis for disapproving Oklahoma's RPG, no basis
for issuing a FIP with a substitute RPG for Wichita Mountains, no basis for disapproving Texas's
long-term strategy, and no basis for imposing additional SO2 limits on Texas sources.
Response:
808
-------
In previous rulemakings, we partially approved and partially disapproved portions of the Oklahoma
Regional Haze SIP.646 We concurrently approved portions of the SIP and promulgated a FIP that
established SO2 BART emission limits, which would result in emission reductions for six Oklahoma
sources,647 to remedy these deficiencies in the Oklahoma RH SIP. Subsequently, we replaced our
FIP for two of the sources with approval of Oklahoma's SIP Revision BART determinations for
those two sources. Implementation of the original FIP and the SIP Revision will result in greater
reasonable progress and greater emission reductions across these BART sources than was anticipated
in Oklahoma's February 19, 2010, Regional Haze SIP submittal.
We also reviewed Texas' analysis of its sources regarding its impacts on surrounding Class I
areas and found that Texas did not perform an adequate evaluation under the long-term strategy
requirements as required under the Regional Haze Rule (we discuss this in our proposal, TSDs,
and in other responses to comments in this document). In our December 16, 2014 proposal, we
agreed with Oklahoma that no additional reasonable progress measures for Oklahoma sources were
necessary during this first planning period, which is further discussed in this response.
In the February 19, 2010, Oklahoma Region Haze SIP, Oklahoma considered the four statutory
factors under Section 51.308(d)(l)(i)(A) in its evaluation of the potential for additional controls. The
ODEQ's analysis focused on moderate cost controls for sources likely to contribute to visibility
impairment at the Wichita Mountains.
The ODEQ evaluated the major sources of each visibility impairing pollutant within the State. In its
analyses of additional SO2 control, it noted that the three largest sources of sulfur emissions in the
State, OG&E Muskogee, OG&E Sooner, and AEP/PSO Northeastern, were subject to BART.
ODEQ also evaluated the non-BART EGU sources mentioned by Luminant, the commenter - GRDA
Units 1 and 2, Hugo Unit 1, and Great Lakes Carbon Corporation. Oklahoma considered these
sources and other sources for additional control under reasonable progress, but concluded that
retrofitting additional point sources of NOx and SO2 would impose unreasonable costs for small
visibility improvement, given that many point sources were already being controlled through BART,
consent decrees or other regulatory mechanisms; units already had adequate controls in place; or
units are located too far from the Wichita Mountains, and therefore have too little visibility impact to
646 We approved certain core elements of the Oklahoma Regional Haze SIP, including Oklahoma's: identification of
sources that are BART eligible and subject to BART; its determination of baseline and natural visibility conditions;
its coordination of regional haze and RAVI; monitoring strategy and other implementation requirements; its
coordination with States and federal land managers; and a number of the State's NOx, SO2, and PM BART
determinations. We disapproved Oklahoma's submitted SO2 BART determinations for Units 4 and 5 of the OG&E
Muskogee plant; Units 1 and 2 of the OG&E Sooner plant; and Units 3 and 4 of the AEP/PSO Northeastern plant.
We also disapproved the long-term strategy in Oklahoma's Regional Haze SIP because it failed to include
appropriate, enforceable controls for these six sources. To remedy these deficiencies in the Oklahoma Regional
Haze SIP, we promulgated a FIP. (Proposal: 76 FR 16188 (March 22, 2011); Final action: 76 FR 81728 (December
28, 2011)). We subsequently replaced our FIP for two of the sources in the Oklahoma FIP with approval of
Oklahoma's SIP revision BART determinations for those two sources. (Approval of Oklahoma's partial
replacement for FIP: 79 FR 12944 (March 7, 2014). Partial FIP withdrawal: 79 FR
12954 (March 7, 2014)).
647 76 FR 81728.
809
-------
justify the cost of retrofitting them. For instance, GDRA, by far the largest non-BART source648, is
already partially controlled for SO2 through flue gas desulfurization. We agreed with this analysis
and believe that Oklahoma provided a logical explanation for its approach and reasons for
excluding additional Oklahoma sources, including GRDA Units 1 and 2, Hugo Unit 1, and Great
Lakes Carbon Corporation. We also noted in our OK TSD that GRDA plans to install low NOx
burners and overfire air, and that Unit 1 (which is not scrubbed) is scheduled to be retired or
converted to natural gas.649
In our December 28, 2011 Oklahoma SIP approval and FIP, we proposed no action on whether
Oklahoma satisfied the reasonable progress requirements of our RH SIP requirements found at
section 51.308(d)(1). In this current action, we are finalizing our disapproval of Oklahoma's RH SIP
that addresses the requirements of Section 51.308(d)(1) to establish reasonable progress goals, except
for Section 51.308(d)(l)(vi).650 Oklahoma's lack of adequate information from Texas prevented it
from properly developing its reasonable progress goals for the Wichita Mountains, and we disagree
that we are applying a different standard to Texas sources than we are sources in other states. We
note that we were not required to do a four-factor analysis for Oklahoma's non-BART sources
because, as discussed in our proposal651 and OK TSD, we reviewed Oklahoma's four-factor analysis
for Oklahoma's non-BART sources, and agree with Oklahoma that it has demonstrated that it is not
reasonable to require additional emission reductions for those sources for this planning period. We
agree with Oklahoma's reasonable progress analysis for sources within Oklahoma and its assessment
that the Wichita Mountains would not meet the uniform rate of progress without significant
reductions from Texas sources. Because the reasonable progress goals Oklahoma established for the
Wichita Mountains does not include appropriate consideration of reductions at Texas sources, we
were required by the Regional Haze Rule to disapprove Oklahoma's reasonable progress goals.
We disagree that the visibility impact of the remaining sources in Oklahoma have the same
visibility impact as the Texas sources we seek to regulate under the proposal and also disagree
that the visibility benefit from imposing the same levels of control on these remaining Oklahoma
sources as we propose to impose on the targeted Texas sources is appropriate. The 20% worst
days at the Wichita Mountains are dominated by days impacted by emissions from sources in
Texas. The largest impacts from sources in Oklahoma rarely occur on the 20% worst days as
identified by the IMPROVE monitor data during the baseline period. Texas (all sources and
pollutants) is projected to contribute 27.5% of the visibility impairment at the Wichita Mountains
in 2018 based on CENRAP modeling that included projected reductions due to CAIR, compared
to 16.3% for Oklahoma sources. Texas point sources are projected to contribute 14% of the total
visibility impairment, including 7.7% impairment from EGUs across the state and 6.5% from
EGUs in the northeast portion of the state. This compares to only 3.9% of the total visibility
impairment from all Oklahoma point sources. As we discuss in the OK TSD, we note that the
total contribution from SO2 emission from Oklahoma sources not identified for control is only a
fraction of the 1.23 Mm"1 projected from all SO2 point sources, and none of those non-BART
648 GDRA has nearly twice the SO2 emissions the Western Farmers Electric Cooperative—Hugo Plant, and nearly
five times the SO2 emissions of Great Lakes Carbon Corporation.
Unit 1.
649 OK TSD at 26.
650 79 FR 74889.
651 79 FR 74871
810
-------
sources are located and have emission levels such that we would anticipate a significant fraction
of the 1.23 Mm-1 visibility impairment at the Wichita Mountains on the 20% worst days to be
reduced should any one source be controlled.652 We also note that compared to 1.23 Mm"1
impact from all Oklahoma point source SO2 emissions combined, the individual facility-level
impact from Monticello and Big Brown were modeled to be 1.275 Mm"1 and 1.169 Mm"1,
respectively based on 2018 CENRAP projected emissions.653 Therefore, controlling one facility
in Texas could achieve more visibility benefit on the 20% worst days than controlling all SO2
emissions from point sources in Oklahoma. Texas sources do not have a minimal impact,
compared to the Oklahoma sources regulated in the first planning period and other sources
outside of Oklahoma.
In reviewing Oklahoma's analysis of which sources to regulate and the level of control to require
of the units, we looked at whether the reductions from the Oklahoma sources were reasonable for
this, the first planning period. We note that Oklahoma gained emission reductions from BART
sources and those regulated under other CAA programs, and demonstrated that this addressed
visibility impacts from the largest emission sources in the state. We further reiterate that RPGs are
interim goals that represent incremental visibility improvement over time toward the goal of
natural background conditions. We calculate654, that for the planning period ending in 2018, our
FIP evaluated and addressed controls for 41% of Oklahoma's total 2002 SO2 budget of 170,021 tons
and 66% of Oklahoma's point source emissions.655 Considering the high control levels resulting
from the installation of scrubbers on the six EGU units in our FIP, we consider this level of control
adequate for reasonable progress for the first planning period.
Given the large contributions of visibility impairing emissions from sources outside of Oklahoma,
particularly from Texas EGUs, Oklahoma stated that it would be unreasonable to require severe or
over-control of Oklahoma sources to compensate and that to meet the URP it would have to consider
emission reduction measures available for those sources in Texas that contribute the most to visibility
impairment at Wichita Mountains. We agreed with this analysis, and the fact that Oklahoma
addressed the visibility impairment from its largest sources in the first planning period.
Note that, as described in other sections of this document, as the controls envisioned during this
planning period are adopted, the percentage impact from those facilities not controlled in
Oklahoma will become larger (on a percentage basis) and will be analyzed in future planning
periods. In other words, some of the facilities in Oklahoma that were not controlled in the first
planning period will become the largest impacting sources in the next planning period and should
be analyzed for additional reasonable controls. This methodology allows a consistent procedure
to identify facilities for additional control analysis in this and future planning periods and ensures
continuing progress towards the goal of natural visibility conditions.
In summary, we agree with Oklahoma's reasonable progress analysis for sources within Oklahoma
and its assessment that the Wichita Mountains would not meet reasonable progress and the URP
without significant reductions from Texas sources. For these reasons stated above, and others that
652 OK TSD at 26.
653 FIP TSD table A.4-la
654 See the file, "OK FIP percent EGU control.xlsx," in our docket.
655 Note that the actual reduction will be slightly lower than 41% because of the Oklahoma SIP revision which
changes the controls for two facilities slightly from the EPA Oklahoma SO2 FIP.
811
-------
EPA explored more fully in the TSD for this FIP proposal, we believe that Oklahoma has adequately
controlled its own sources for reasonable progress to the extent necessary for this planning period.
The SO2 reductions made by Oklahoma in the first planning period are substantial and the reasoning
underpinning that decision is valid. Reductions from Texas facilities are needed in order for
Oklahoma to make reasonable progress in the Wichita Mountains. We also note that there were
many sources in Texas that we did not require to be additionally controlled in the first planning
period. Thus, we disagree that we are applying a different standard to Texas sources than we are
sources in other states.
Our resetting the reasonable progress goal is lawful and legally necessary because we are
disapproving Oklahoma's RPG for the Wichita Mountains. Because Oklahoma's analysis in its
original SIP submittal did not take into consideration the emissions reductions from certain
BART facilities needed to satisfy the BART requirements under Section 51.308(e), the RPG
Oklahoma established in its RH SIP was flawed. We remedied this deficiency through setting
SO2 BART emission limits for six facilities in a final action dated December 28, 2011. In
addition, because we disapproved Texas' LTS evaluation, we were under a statutory obligation
to evaluate Texas sources and propose a FIP for those facilities where we determined that
reasonable emission controls could be installed for improved visibility benefit. Because the
reasonable progress goals Oklahoma established for the Wichita Mountains do not include
appropriate consideration of reductions at Texas sources, we were required by the Regional Haze
Rule to disapprove Oklahoma's reasonable progress goals. We recalculate new reasonable
progress goals for 2018 for the Wichita Mountains based on the results of our technical analysis
that additional controls at Texas sources were reasonable to meet the reasonable progress/long-
term strategy requirement for reasonable progress and accounting for the visibility benefit of the
required controls anticipated to be in place by 2018. We discuss our disapproval of the Texas
Regional Haze SIP in more depth elsewhere in this document. We discuss comments concerning
our calculation of the RPG in the section of this document where we respond to comments
concerning modeling issues.
18. International Emissions
Comment: [Luminant (0061) p. 42] Luminant provided background on the May 2006 Alpine
Geophysics control strategy analysis, CENRAP Regional Haze Control Strategy Analysis Plan 1.
Luminant noted that in February 2006, CENRAP contracted Alpine Geophysics, LLC ("Alpine")
to assist in developing and evaluating potential emission control strategies. Alpine was
specifically charged with developing a quantitative procedure to identify and prioritize potential
regional haze control strategies to be tested by CENRAP modelers. Alpine used a three-step
process: (1) assembling available information useful in quantifying the reductions in fine
particulate aerosol concentrations needed to satisfy CENRAP's preliminary regional haze
visibility projections; (2) developing Areas of Influence (AOIs) upwind of each Class I area
within which common "visibility precursor-Class I receptor" impacts could be aggregated into
similar groupings; and (3) synthesizing the first two steps together with information in the
estimated 2018 CENRAP emissions inventory and the cost-effectiveness of various controls, to
deduce a prioritized set of regional haze control strategies containing elements of both regional
emissions reductions and targeted reductions within the AOIs closest to those six CENRAP
812
-------
Class I areas for which positive visibility increments were estimated. As part of the analysis,
Alpine also developed a four-factor reasonable progress analysis for the states' consideration
Luminant stated that Alpine concluded that "as a result of the implementation of the list of
additional point and area source controls in the each primary AOI," Texas's Class I areas "will
be unable to achieve a level of emissions reduction necessary to bring these areas under the glide
slope by 2018 using the identified control strategy definitions." (2006 Alpine Report at 43)
Response: The EPA acknowledges the background information provided by the commenter.
We take no position with respect to the commenter's synopsis.
Comment: [Luminant (0061) p. 42, 100; CCP (0075) p. 2, 13-14; NRG (0078) p. 6; AECT
(0074) p. 8-9; Xcel Energy (0064) p. 14-21; Alpine (0078) p. 22] Multiple commenters stated
that EPA acknowledged, but arbitrarily and capriciously failed to account for international
sources of emissions that Texas cannot control and which render EPA's proposal ineffective in
improving visibility to meet the URP and 2064 goal. The commenters objected to EPA's
proposal that would require over-control of Texas sources to compensate for international
emissions.
Luminant noted that the 2006 Alpine Report developed for CENRAP concluded, "regardless of
the emissions reduction achieved by CENRAP with the available source category and technology
applications, there still is an emissions component which is directly out of their control." (2006
Alpine Report at 44)
Multiple commenters noted that, as recognized by EPA, CENRAP's Particulate Matter Source
Apportionment Technology (PSAT) modeling analysis indicates that emissions from Mexico and
other countries south of the U.S. contribute 52% of the visibility impairment at Big Bend and
25% of the visibility impairment at the Guadalupe Mountains.25 (Texas SIP ES-2; 2009 Texas
SIP Narrative at 10-9; 79 Fed. Reg. 74,843) In comparison, all Texas EGUs together
contributed only 4.5% at Big Bend26 and 4.1% at Guadalupe Mountains. (FIP TSD at A-29 to A-
30.) Luminant noted, accordingly, Texas found, the "national goal of visibility at these Class I
areas cannot be met without substantial reductions in emissions from outside the United States."
(2009 Texas SIP Narrative at 10-10.)
Several commenters noted that EPA concludes that "emissions and transport from Mexico and
other international sources will limit the rate of progress achievable" by Texas and concludes
that efforts to meet the goal of natural visibility by 2064 "would require further emissions
reductions not only within Texas, but also large emission reductions from international sources."
79 Fed. Reg. 74,843 (emphasis added).
Xcel Energy stated that if TCEQ cannot meet the glide path without "large emission reductions
from international sources," it is unreasonable for EPA to require additional controls from Texas
without making any effort to seek emissions reductions from international sources. EPA is
proposing to impose on American companies and American ratepayers the burdens of
813
-------
compliance with the FIP while allowing other countries to avoid any responsibility for their
contributions to visibility impairment in Class I areas.
Xcel Energy stated that, while EPA "agree[ d] with the TCEQ that a rate of improvement
necessary to attain natural visibility conditions by 2064 is not reasonable," EPA did not "believe
that the rate of improvement the [TCEQ] has selected is reasonable" and disagreed with TCEQ's
"analysis of emission measures needed to meet the URP." 79 Fed. Reg. at 74,843 (emphasis
added). In other words, EPA agreed that Texas reasonably concluded that it could not meet its
"glide path" for this planning period. Nonetheless, EPA disapproved of TCEQ's analysis for
failing to adequately consider additional controls to further increase its rate of improvement. Yet
EPA fails to account for the substantial contributions from international sources in making this
determination. And because Texas cannot be required to over-control in-state sources to make
up for international contributions, EPA's failure to adequately consider international sources in
its disapproval of Texas' RPG analysis is patently unreasonable.
Xcel Energy stated that over-control of in-state sources due to international contribution is
exactly what EPA proposes to require from Texas. EPA has stated that "[t]he glide path is not a
presumptive target, and States may establish a RPG that provides for greater, lesser, or
equivalent visibility improvement as that described by the glide path." RPG Guidance, at 1-3.
NRG and AECT stated that Texas sources' relative contributions are expected to shrink in the
future at the same time that the international contribution rises. Even as Texas SO2 emissions
decline in the next few years, EPA has projected that SO2 emissions from Mexico will increase
by 26% from 2012 to 2030.27 (2015 Alpine Report submitted with comment 0078 at 28-31).
NRG and AECT noted that, even if the contributions of the SO2 emissions from the Texas EGUs
to visibility impairment at the Class I areas were eliminated, visibility improvements in those
areas are not likely due to the significant contribution of emissions from sources in Mexico and
other countries south of the United States to visibility impairment at such areas. Thus, it is
unreasonable for EPA to ignore the impact of emissions from international sources and require
additional SO2 emissions controls for the identified Texas EGUs (such as Limestone Units 1 and
2) to achieve reasonable progress.
Footnotes:
25 "Technical Support Document for the Texas Regional Haze State Implementation Plan" (November, 2014), Table
15
26 Id., Figures 6 and 8
27 "Developing Mexico National Emissions Inventory Projections for the Future Years of 2008, 2012, and 2030"
(Martinus E. Wolf, et. al.)
Response: We agree with the commenters that international emissions significantly impact
visibility conditions at Big Bend and Guadalupe Mountains. However, as we discussed in the
preamble to the RHR, "the States should not consider the presence of emissions from foreign
sources as a reason not to strive to ensure reasonable progress in reducing any visibility
impairment caused by sources located within their jurisdiction."656 While the goal of the
regional haze program is to restore natural visibility conditions at Class I areas by 2064, the rule
requires only that reasonable progress be made towards the goal during each planning period,
656 64 FR 35755 (July 1, 1999)
814
-------
and in cases where it is not reasonable to meet the rate of progress needed to attain the goal in
2064, that the State demonstrate that it is not reasonable and that the selected rate of progress is
reasonable for that planning period. We recognize that it may not be possible to attain the goal
by 2064, or at all, because of impacts from new or persistent international emissions sources or
impacts from sources where reasonable controls are not available. However, states are still
required to demonstrate that they are establishing a reasonable rate of progress that includes
implementation of reasonable measures within the state to address visibility impairment in an
effort to make progress towards the natural visibility goal during each planning period.
The CAA has the goal of natural conditions, but does not have any date or requirement for plans
to demonstrate achievement of that goal, only reasonable progress towards it. The EPA rule
reiterates the CAA goal, and provides for the use of an analytical framework that compares the
rate of progress that will be achieved by a SIP to the rate of progress that if continued would
result in natural conditions in 2064. The EPA rule also reiterates the CAA requirement for
reasonable progress in reaching natural conditions, but it does not establish a requirement that
natural conditions be reached in 2064. We have approved a number of SIPs for the first planning
period that have projected that continued progress at the rate expected to be achieved during that
first period would not result in natural conditions until a date after 2064.
We disagree that our action requires "over control" of Texas sources. Our action requires the
control of particular Texas sources that due to their own emissions were impacting the visibility
at Texas and/or other state Class I areas. Our determination of reasonable controls is based on
consideration of the four-factors and the visibility benefit of controls in reducing the visibility
impairment from Texas sources for this first planning period. Comments concerning EPA's
efforts to address international emissions are addressed in a separate response to comment in this
section.
Nothing in the Regional Haze Rule or our FIP is calculated to hold Texas accountable for
emissions from Mexico. We agree those international emissions should be addressed to achieve
natural visibility, but our agreement on this point does not in any way relieve Texas of the
obligation to make reasonable progress, including through controls on its own sources, and
particularly through the emissions addressed with controls through our FIP. As we and the
commenters note, "emissions and transport from Mexico and other international sources will
limit the rate of progress achievable" by Texas and efforts to meet the goal of natural visibility
by 2064 "would require further emissions reductions not only within Texas, but also large
emission reductions from international sources." We agreed with Texas that it is not possible to
meet the rate of progress to attain natural visibility conditions by 2064, and that it was reasonable
to adopt a RPG that provides for a slower rate of improvement in visibility, in part due to the
significant impact from international emissions. However, Texas is required to demonstrate
based on the four-factors that the rate of progress adopted by the State is reasonable, and as we
stated further emission reductions within Texas are needed to make progress towards the goal of
natural visibility. As we discuss in depth elsewhere, Texas's four-factor analysis was flawed.
The size of the impact from international emissions in no way relieves Texas of the requirements
of the RHR to consider the four factors and evaluate controls for addressing visibility impairment
due to Texas sources and making reasonable progress towards the goal of natural visibility
conditions. In our action, we are requiring controls based on our evaluation of the four-factors
815
-------
and consideration of visibility benefits for those individual sources with the largest visibility
impacts at Texas' Class I areas and/or Wichita Mountains. We are not, as the commenter
suggests, requiring "over control" or requiring unreasonable controls in order to somehow
compensate for the impact from international sources.
Estimated international emissions from all sources are included in our modeling and the
modeling performed by CENRAP. The CENRAP source apportionment results provide
estimates of the contributions for Texas sources, sources in other States, and those international
sources outside of the jurisdiction of Texas and nearby states.657 This information was utilized in
Texas' and our consideration of whether or not it was reasonable to attain the URP, and in
consideration of the amount of visibility impairment that could potentially be addressed by each
contributing region. The CENRAP 2018 projections estimated that approximately 9% and 8% of
the total visibility impairment at Guadalupe Mountains and Big Bend, respectively, were due to
emissions from Texas point sources. Our additional analysis and modeling in support of this
action shows that a significant portion of that visibility impairment can be addressed by
controlling a small number of facilities through the use of cost-effective controls. For Big Bend,
the required controls address 1.88% of the total visibility impairment and 23.4% of impact from
all Texas point sources (based on 2018 CENRAP projected emission levels). Controls on Coleto
Creek alone addresses approximately 6% of the total Texas point source impact. For Guadalupe
Mountains, the required controls address 2.22% of total visibility impairment and 25.74%) of
impact from all Texas point sources (based on 2018 CENRAP projected emission levels).
Controls on Tolk alone addresses nearly 8%> of the total Texas point source impact. We also note
that the required controls address 5.8% of total visibility impairment and 41.4% of the impact
from all Texas point sources (based on 2018 CENRAP projected emission levels) at Wichita
Mountains. Each of the required controls result in significant visibility benefit at one or more
Class I areas.
The RHR addresses situations where increases in emissions from other countries prevent Class I
areas from achieving the established reasonable progress goals. The five-year progress report
requires states to assess the adequacy of their regional haze SIP.658 If the state determines at that
time "that the implementation plan is or may be inadequate to ensure reasonable progress due to
emissions from sources in another country, the State shall provide notification, along with
available information, to the Administrator."659 However, any anticipated increases in emissions
of international sources do not render it unnecessary for the state to implement reasonable
controls to address the visibility impairment due to their own sources.
Comment: [Luminant (0061) p. 100; CCP (0075) p. 2, 13-14; NRG (0078) p. 6; AECT (0074)
p. 8-9; Xcel Energy (0064) p. 14-21; AECOM (0061 and 0075) p. 5-1] Several commenters
argued that EPA has not upheld its duty to seek emissions reductions from international sources.
AECOM stated that international emissions play a significant role in visibility conditions for the
Texas Class I areas. As a result, TCEQ has requested assistance from EPA in addressing
657 See Figures 6, 7, 8 and 9 of the TX TSD for CENRAP source apportionments for all sources and point sources.
658 40 CFR 52.308(h)
659 40 CFR 51.308(h)(3)
816
-------
international emissions.86 Luminant noted that in 2009, "Texas requested] in its SIP that [EPA]
initiate and pursue federal efforts to reduce impacts from international transport," (79 FR 74844)
consistent with EPA's regulations. 40 CFR 51.308(h)(3). EPA acknowledges the substantial
impact of international sources and Texas's request but does not account for them in its review
and proposed disapproval, nor does EPA act on Texas's request from 2009. (79 FR 74842-44)
Luminant asserted that EPA's failure to consider this information in its analysis and account for
these international emissions in reviewing Texas's reasonable progress goals and long-term
strategy and in proposing its own goals is arbitrary and capricious.
Multiple commenters noted the EPA has explained in its regional haze regulations that "the
projected emissions from international sources will in some cases affect the ability of States to
meet reasonable progress goals." Thus, EPA specifically instructed that "EPA does not expect
States to restrict emissions from domestic sources to offset the impacts of international transport
of pollution." (64 FR 35714, 35736, July 1, 1999; and Additional Regional Haze Questions 19,
Sept. 27, 2006) Instead, EPA recognizes that it has a duty to help control emissions from
international sources. As the Agency noted in the 1999 regional haze rulemaking, "EPA will
work with the governments of Canada and Mexico to seek cooperative solutions on
transboundary pollution problems." (64 FR 35714, 35736).
Xcel Energy stated that EPA identified a process by which EPA would manage international
contributions: (1) a State would submit a "technical demonstration" showing that "international
emissions sources are responsible for a substantial increase in emissions affecting visibility
conditions in any Class I area;" (2) EPA would determine if it agreed with the finding; and (3) "If
EPA agrees with the State's finding, EPA will take appropriate action to address the international
emissions through available mechanisms." (64 FR 35747) Texas met all of those requirements.
Texas submitted modeling to EPA showing that Mexico's emissions were a significant
contributor to visibility impairment in Big Bend and Guadalupe Mountains. See Texas SIP, at
Section 10.6. EPA agrees that Mexico emissions are significant. See 79 Fed. Reg. at 74,843
("large emission reductions from international sources" would be required to meet natural
visibility goals). However, EPA has done nothing to meet its obligations to control emissions
from Mexico or other international sources. Thus, EPA, and not Texas, must re-evaluate what it
can do to help Texas meet its URP for Class I areas affected by international sources.
Footnote:
86 http://www.tceq.state.tx.us/assets/public/implementation/air/sip/haze/2SIP_ado_rev.pdf at ES-2.
Response: Our response to comments concerning how we considered international emissions
can be found above. Consistent with our earlier record statements, we agree that international
emissions, particularly emissions from Mexico, impact visibility conditions at Big Bend and
Guadalupe Mountains. Moreover, we acknowledge that Texas requests in its SIP that we initiate
and pursue federal efforts to reduce impacts from international transport. There are efforts
underway related to health issues along the United States- Mexico border. Given that emissions
contributing to health effects and those contributing to visibility impairment are generally the
same, the border studies and continuing emissions inventory development are tools to help
identify solutions for visibility impairment. The Border 2020 program aims to, among other
things, reduce air pollution to help meet national ambient air quality standards and reduce
emission through the use of energy efficiency and/or alternative/renewable energy projects. We
817
-------
expect that recent commitments from Mexico to reduce carbon dioxide and black carbon
emissions will have ancillary benefits to improve visibility at Class I areas in the future.
Comment: [Luminant (0061) p. 100; AECOM (0061 and 0075) p. 5-1] Luminant stated that
the EPA has repeatedly approved state reasonable progress goals that take into account
international emissions and do not require domestic sources to further reduce their emissions to
compensate for emissions over which the state has no control—including New Mexico's
reasonable progress analysis for the exact same monitor at Guadalupe Mountains. 657 AECOM
also noted that other state agencies have evaluated international contributions to regional haze
and determined its relevance to setting RPGs and URPs.87
Luminant asserted that EPA's failure to consider international emissions data is arbitrary and
capricious and violates EPA regional consistency regulations. International transport is a very
real and key factor to be considered, not just in evaluating whether the URP can be achieved, but
in setting the reasonable progress goal. Indeed, EPA concedes that "emissions and transport
from Mexico and other international sources will limit the rate of progress achievable on the
20% worst days . . . ,"658 And in fact, visibility impairment from Mexico is not just limited to
point sources but includes agricultural burning along the Texas-Mexico border.659 Yet, nowhere
in EPA's analysis does it account for these emissions. Luminant stated that EPA's use of a so-
called "natural conditions" or "clean" background for its Texas individual source modeling
wrongfully assumes that Texas can eliminate the impact of international emissions. In that way,
too, EPA's failure to account for international sources of visibility impairment is arbitrary and
capricious.
Footnotes:
87 http://www.ecy.wa.gov/programs/air/globalwarm_RegHaze/RH_SIP/Ch_9.pdf.
657 See, e.g., 77 Fed. Reg. at 70,701 (New Mexico); 77 Fed. Reg. at 30,256 (Idaho); 77 Fed. Reg. 3681, 3687 (Jan.
22, 2012) (Minnesota); 77 Fed. Reg. 76,174, 76,204 (Dec. 26, 2012) (Washington).
658 79 Fed. Reg. at 74,843 (emphasis added).
659 See, e.g., TCEQ, Today's Texas Air Quality Forecast (last modified Apr. 3, 2015),
https://www.tceq.texas.gov/airquality/monops/forecast_today.html (forecasting "[s]moke from agricultural burning
in Mexico and Central America" on March 25 and 26, 2015).
Response: We respond to Luminant's comment alleging inconsistency in the consistency
section of this document. We respond to Luminant's comment concerning the use of "natural
conditions" or "clean" background for individual source modeling in the modeling section of this
document. Our response to comments concerning how we considered international emissions
can be found elsewhere in this section of this document.
Comment: EPA ignores Mexico's contribution to visibility impairment and makes it
impossible for Texas to meet the goal of achieving natural visibility conditions.
[Xcel Energy (0064) p. 14-21]
Xcel Energy stated that it is not possible for Texas to achieve the URP because of the
overwhelming contribution from Mexican sources to visibility impairment in the Class I areas.
818
-------
Mexico's impacts dwarf the contribution of any and all point sources in Texas. In fact, Xcel
Energy's analysis (described below) shows that if every point source in Texas were shut down, it
would have only a marginal impact on visibility in the Guadalupe Mountains. It is simply
impossible for Texas to meet the URP and the 2064 visibility goal.
Xcel Energy conducted modeling to determine the impact of removing all Texas and U.S.
elevated point sources on Guadalupe Mountains. Following the procedures used by ENVIRON,
emissions processing was performed using the Sparse Matrix Operator Kernel Emissions
(SMOKE) processing system. For the purpose of this analysis, the point source emissions
inventories were updated so that no Texas or other United States elevated point sources would be
included in the CAMx modeling input. The SMOKE processing output data was further
processed in accordance with ENVIRON's procedures to create the final CAMx emissions input
files.
Xcel Energy stated that CAMx was run with the updated emissions input files containing no
Texas or other United States elevated point sources. The CAMx output files were post-
processed using the same scripts and utility programs used by ENVIRON. These intermediate
results were further processed using the EPA's Modeled Attainment Test Software. The final
Modeled Attainment Test Software output data were then analyzed using the glide path and
source contribution workbooks provided by the EPA. Xcel Energy used EPA's process systems
and emission input files so EPA can easily replicate this modeling. However, if EPA has any
difficulty conducting or confirming this modeling, Xcel Energy would be happy to share its
modeling files with EPA.
Figure 1 provided by Xcel Energy in comment 0064 (reproduced below) shows the glide path for
the Guadalupe Mountains, the most impacted Class I area for the Tolk Generating Station. The
uncontrolled haze index value of 16.36 dv is the modeled value determined in the 2018 baseline
scenario processed by ENVIRON for EPA. TCEQ did not propose additional reasonable
progress controls in its SIP, citing the growth and control projections already incorporated into
the 2018 baseline emissions data; therefore, the TCEQ SIP value is equivalent to the
uncontrolled 2018 baseline scenario value. The EPA FIP value of 16.21 dv corresponds to the
modeled haze index obtained from ENVIRON's high control efficiency modeling scenario. The
exclusion of all of Texas and other United States elevated point sources resulted in a modeled
haze index value of 14.88 dv, meaning that Mexican sources and natural contributions are
projected to account for 92%, or all but 1.48 deciviews, of visibility impairment in the
Guadalupe Mountains.
819
-------
•
EPAFIP, 16.21
17^19
16.49
""""
14-74
12.99
11.06
9.48
•
• • -.
5.87
7 7J 6.68
648
6 08
6.08
6.08
6.08
6.08
6.08 6.08
1 1— —»' * - ^ » r t t —i 11 r-—*—p l t T 1 l
Uniform Rate of Progress and 2018 Projected Progress
Guadalupe Mountains
20.00
18.00
16.00
14.00
12.00
3
"g 10.00
Si 8.00
ra
x
6.00
4.00
2.00
O.OO
/ ^ ^ ^ ^ ^ / # # -/ /
-------
by 2064 "would require further emissions reductions not only within Texas, but also large
emission reductions from international sources" (emphasis added).661
Furthermore, we find it necessary to note that Xcel's analysis incorrectly attributes 92% of
visibility impairment at Guadalupe Mountains to "Mexico sources and natural contributions."
Xcel's analysis examines the effect of removing Texas and U.S elevated point sources, but fails
to account for impacts from mobile and area sources within Texas and other states, and fails to
differentiate Mexican sources from other international sources. The analysis also fails to
consider that deciviews are a logarithmic function of extinction, resulting in the underestimation
of the percent contribution from Texas and U.S. point sources. Xcel's modeling estimates that
elimination of all U.S. and TX elevated point sources would result in a reduction of 1.48 dv,
from 16.36 dv to 14.88 dv. This is a reduction in extinction of 7.06 Mm"1, or a 13.76% reduction
in total extinction from 51.3459 Mm"1 (16.36 dv) to 44.2823 Mm"1 (14.88 dv).
CENRAP source apportionment for 2018 relied on by Texas and us in our analysis provides
estimates of the contribution from various source regions and source types. These results are
summarized below. We note that "BC" (boundary conditions) captures impacts from all sources
outside of the modeling domain.
CENRAP 2018 PS AT results
'or Guadalupe Mountains, all pollutants (extinction, Mm"1)
Category
US (excluding
Texas)
Texas
Gulf of
Mexico
Canada
Mexico
BC
Elevated
Point
4.497
0.004
0.297
3.210
Low-Level
Point
0.229
0.004
0.154
0.080
Natural
3.516
4.509
0.000
0.047
1.600
On-Road
Mobile
0.285
0.145
0.000
0.000
0.002
Non-Road
Mobile
0.435
0.232
0.001
0.060
0.492
Area
3.406
6.659
0.043
0.157
2.461
EGU Point
1.809
Non-EGU
Point
2.030
Boundary
Condition
0.000
0.000
0.000
0.000
0.000
4.343
Total
12.367
15.386
0.052
0.715
7.846
4.343
The CENRAP PSAT results are consistent with Xcel's modeling showing that the total
contribution from all U.S. and Texas point sources (elevated and low level) are approximately
8.56 Mm"1 compared to 7.06 Mm-1 estimated by Xcel for elevated point sources only. The
impacts from all Texas point sources (3.84 Mm-1) are larger than the estimated impacts from
661 79 FR 74843
821
-------
point sources in Mexico (the large portion of Mexico within the modeling domain) and are
comparable to the level of impacts from all sources outside of the modeling domain captured by
the boundary conditions. Overall impacts from all sources in Texas are larger than all sources in
Mexico and boundary conditions combined. As we discuss in our proposal and elsewhere in our
response to comments, we and Texas agreed that it was reasonable to focus on impacts from
point sources for this planning period. The impact from Texas point sources is significant, and
as our analysis shows, a significant portion of this impairment can be addressed by controlling a
small number of sources. Controls on just four units at Tolk and Big Brown are estimated to
reduce visibility impairment at the Guadalupe Mountains (estimated by CENRAP 2018
modeling) by approximately 0.49 Mm-1, or a 12.7% additional reduction in the total impact due
to Texas point sources. All required controls combined are estimated to reduce visibility
impairment at the Guadalupe Mountains by an additional 0.836 Mm"1, a 21.8% additional
reduction in impairment from Texas point sources beyond estimated visibility conditions
projected by the CENRAP 2018 modeling.
We address comments concerning our efforts to address visibility impairment from international
emissions in a separate response to comment above.
Comment: [Xcel Energy (0064) p. 14-21] Xcel Energy stated that even if all sources in the
state of Texas were to entirely cease operation, there still would be significant impairment at the
Big Bend and Guadalupe Mountains National Parks that would prevent EPA from meeting the
natural visibility goals prescribed by the Clean Air Act and the RHR. Xcel Energy stated that the
EPA has inaccurately and arbitrarily dismissed this in its evaluation. In fact, Figures 2 and 3 in
the Proposal do not accurately represent the effect of the emissions from Mexico because EPA
arbitrarily cut off the top contributors, without showing the full level of their contribution to
visibility impairment. 79 Fed. Reg. at 74,878-79. By not showing the visibility impairment to
scale, EPA's figures appear to indicate a greater impact from specific facilities in Texas, while
depicting a lesser impact from other sources. When the impacts are accurately portrayed, it
clearly shows how small a contribution Texas facilities make to visibility impairment.
Xcel Energy provided (Xcel Energy Figure 2) which shows the average source contribution to
the 2018 deciview haze index at the Guadalupe Mountains. The unaltered 2018 baseline
scenario resulted in a haze index of 17.19 dv while the exclusion of Texas and other United
States elevated point sources resulted in a value of 15.80 dv. The 38 selected facilities refer to
the facilities chosen by EPA for a haze index contribution analysis based on 2009 annual
emissions and the distances to the nearest Class I areas (Q/D analysis). The 21 selected sources
refer to the emissions sources chosen by EPA to be considered for proposed controls under the
FIP and these source collectively contribute only 0.18 dv to the visibility impairment at
Guadalupe.
(Xcel Energy Figure 3) displays the percent contribution to total 2018 light extinction at the
Guadalupe Mountains. These percentage contribution values were calculated based on the
collective contribution to light extinction by each source group as compared to the 2018 baseline
total extinction value. Here, again, the visibility effects of the EPA selected 21 sources are
miniscule in light of the contribution of international sources and natural contributions.
822
-------
Source Contribution to 2018 Deciview Haze Index at Guadalupe Mountains, W20 Group
(Figure 2 provided by Xeel Energy (comment 0064))
20.0000
18.0000
16.0000
14.0000
12.0000
¦g 10.0000
8,0000
6.0000
4.0000
2.0000
0.0000
17.19
0.42
2018 Baseline No TX or U.S. Elev 38 Selected
Pt. Sources Facilities
0.18
0.05
21 Selected Tolk Station
Sources
Total Percent Contribution to 2018 Total Light Extinction at Guadalupe Mountains, W20
Group (Figure 3 provided by Xcel Energy (comment 0064))
823
-------
Response: We address comments concerning how we considered international emissions in a
separate response above. We agree that impacts from international sources are significant and it
is not reasonable to achieve the URP and meet natural conditions in 2064. However, as was
stated in the preamble to the RIIR, "the States should not consider the presence of emissions
from foreign sources as a reason not to strive to ensure reasonable progress in reducing any
visibility impairment caused by sources located within their jurisdiction."662
We disagree with Xcel that Figures 2 and 3 in our proposal do not accurately represent the effect
of the emissions from Mexico because we "arbitrarily cut off the top contributors, without
showing the full level of their contribution to visibility impairment". Figures 1, 2 and 3 in the
proposal aim to highlight and compare the magnitude of impacts from the individual facilities in
Texas we evaluated for controls to impacts from other point sources in Texas and areas outside
of the state. The range of the chart was selected to allow the reader to observe the relative
differences between each facility's impact and compare that to impacts from nearby states and
regions. We provide additional discussion of these figures in the FIP TSD and note that the
Mexico contribution from point sources exceeds the range of the chart. We also provide the
682 64 FR 35755 ( Ju|y y 1999)
824
-------
percent contribution and the extinction due from Mexico point sources on the charts, so the
reader can accurately compare the impacts from each contributor shown.
Xcel suggests that because impacts from sources other than Texas and U.S. point sources are a
large portion of the total visibility impairment, no reductions from Texas point sources are
reasonable. As we discuss in our proposal and elsewhere in our response to comments, Texas
and we agreed that it was reasonable to focus the reasonable progress analysis on point source
emissions of SO2 and NOx, as the sources of these pollutants are the main anthropogenic
pollutants that affect visibility at Class I areas in Texas. Based on CENRAP 2018 source
apportionment results, Texas point sources are responsible for approximately 8.6% of the total
extinction at Guadalupe Mountains ( approximately Vi of the total impairment from all Texas
sources), compared to 7.4% from Mexico point sources and 10.6% from U.S. point sources
(excluding Texas). The largest contribution from all point sources combined in an individual
state is 1.16% from New Mexico point sources. The impact from Texas point sources is
significant and as our analysis shows, a significant portion of this impairment can be addressed
by controlling a small number of sources. Xcel's chart states that 0.64% of the total visibility
impairment is due to emissions from Tolk663. This represents nearly 8% of the total visibility
impairment from all of Texas point sources that can be addressed by controlling only one
facility. We note this is more than half of the contribution that would be addressed if New
Mexico or any other individual state controlled all point sources within its jurisdiction. In
considering the four-factors, including consideration of visibility benefit with cost, we found that
controls on Tolk were cost-effective and resulted in a significant reduction in visibility
impairment at Guadalupe Mountains. This is not inconsistent with the realization that significant
impacts from international emissions and other sources exist and should also be addressed. We
address comments concerning EPA's efforts to address visibility impairment from international
emissions in a separate response to comment above.
Comment: [AECOM (0061 and 0075) p. 5-1; CCP (0075) p. 13-14] AECOM stated that back
trajectories for 2011-2013 show that approximately 77% of the 20% worst day trajectories at
GUMO passed through Mexico. For BIBE, this percentage increases dramatically to about 96%.
These contributions were obtained using ArcGIS.
AECOM stated that Mexican point sources, particularly the coal-fired power plant facilities of
Carbon, are only about 230 km away from BIBE, while the nearest Texas facility with a
proposed new emission limit is about 500 km away. Emissions from these large power plants
are noteworthy - SO2 emissions from the draft EPA 2011 modeling platform cites 162,329 SO2
tons for Carbon II alone in 2008, which is an increase from 1997 (129,341 tons at Carbon II).88
In addition to international point sources, smoke plumes from agricultural fires in Central
America have been shown to travel northward into the U.S. further contributing to hazy
conditions. Agricultural burning helps return nutrients to the soil prior to the growing season.
This burning season usually occurs from March through May in Mexico.89 AECOM provided a
satellite image showing the widespread nature of agricultural burning. (See Figure 5-1 in
comment 0061/0075.) It is evident from this figure that the trajectory of the international
663 We note that Xcels's figure 2 above is not a good representation of the relative visibility impacts because
deciviews are a logarithmic function of extinction.
825
-------
emissions passes right over eastern Texas on their way to WIMO. Therefore, observed haze at
WIMO with a trajectory from eastern Texas could also have originated south of the border, with
substantial international haze contributions.
CCP stated that modeling shows that the sources that cause haze in Big Bend and Guadalupe
Mountains are rarely in the area where most of the emission sources targeted by EPA are located.
Modeling shows that during one of the worst droughts on record in 2011, some of the worst haze
days in Big Bend corresponded to "Fire - Mexico/Central America" flagged events. Looking at
trajectories from particular sources, Mexico's Carbon I and II facilities (only 165 km away from
Big Bend) heavily influenced Big Bend haze conditions, and SO2 emissions have actually
increased from 1999 from these plants.
AECOM and CCP stated that source apportionment modeling presented in the EPA FIP TSD
quantifies international contributions to the Texas Class I areas. Mexican point sources were
determined to contribute about 16.8% to BIBE's 20% worst days with a total contribution from
all emission sectors of over 25%.91 For Mexican point sources alone, this contribution is nearly
four times greater than all of the Texas EGU point source contributions, which were estimated to
be only 4.5% of the 20% worst haze days.92 AECOM provided figures indicating these source
apportionment results for both BIBE and GUMO. (See Figures 5-2 and 5-3 in comment
0061/0075.) Previous research such as the National Park Service 1999 study found that Mexico
contributes as much as 70% of the haze for the 20% worst days at BIBE.93
AECOM stated that impacts from international sources are even greater when boundary
conditions are analyzed. As part of the modeling performed by EPA for the base case of the
future year 2018, the source apportionment results of all 38 facilities in Texas modeled
separately by EPA, as well as contributions through the modeling boundary, were available. We
reviewed the modeling information for individual contributions to selected particulate matter
(PM) species over the entire 36 and 12-km computational domains. To assess the most
important species for purposes of this rulemaking, we extracted information for the contributions
to sulfate from the computational boundary conditions. The boundary conditions represent
hourly or monthly concentrations that are set by a global chemical transport model and then
transported through the boundaries of the CAMx computational domain by the same
meteorology used in the regional CAMx simulation. This analysis examined the monthly and
daily modeled contributions over the entire computational domain, as well as contributions at the
specific Class I area receptors.
AECOM stated that the monthly average sulfate concentrations were analyzed to identify any
periods throughout the year in which the contributions from sources beyond the model
boundaries are most evident. The largest contributions mostly originate in the southern boundary
(i.e., from Mexico or from countries further south). It also shows that the contributions from the
southern boundary are episodic with the likelihood of observing higher concentrations during the
winter and spring. AECOM Figure 5-4 in comment 0061/0075 displays monthly sulfate
contributions for April.
AECOM explained, in general, the combined contributions from all boundaries to sulfate near
the three Class I areas of interest for this rulemaking range from 0.125 [j,g/m3 to 0.38 [j,g/m3
826
-------
during the winter, from 0.125 [j,g/m3 to 0.5 [j,g/m3 during the spring and are less than 0.125 [j,g/m3
during the summer. It is evident that localized impacts, defined by large concentration gradients,
are present in Mexico and extend to southern Texas, particularly BIBE. It is clear from the
concentration gradient information that the international emissions south of the border are
responsible for this impact.
AECOM stated that a more detailed examination was conducted of episodic events during the
spring and summer. AECOM Figure 5-5 in comment 0061/0075 provides daily sulfate average
concentrations for a 3-day period (April 18 to April 20) as an illustration of one of the periods
when active transport from the boundary impacts the southern United States, especially states
near Mexico. We noted from the modeling results that the contribution from emissions beyond
the CAMx boundary were comparable to those shown in the previous figure for a much larger
period that started at the end of February and continued until mid-May.
AECOM stated that the source apportionment information also provides a means to compare the
relative contribution of the boundary emissions relative to any given emission source tagged by
the CAMx modeling conducted by EPA. AECOM Figure 5-6 in comment 0061/0075 provides a
time series of impacts that compares the sulfate daily average CAMx-predicted concentrations
that the WIMO, BIBE, and GUMO Class I areas in Texas experienced from both the boundary
emissions and a selected Texas power plant source (Big Brown). Coleto Creek would have a
smaller impact as it has less emissions. The time series clearly indicate that at these three Class I
areas, the contributions to sulfate from the boundary emissions are always far more significant
than the contributions from Big Brown. For instance, at WIMO, the annual average
concentration from the boundaries is 0.21 |ig/m\ while the contribution from Big Brown is an
order of magnitude lower, only 0.02 [j,g/m3. It is especially interesting to note that boundary
conditions appear to have a greater impact at WIMO than BIBE and GUMO through the
modeling year.
AECOM explained that, in general, it is obvious that the effect of controlling emissions at a plant
like Big Brown would be dwarfed by the massive impact of the international emissions. This
component of haze must be accounted for in regional haze SIPs in the development of RPGs
and/or natural conditions because these emissions from agricultural burns, power plants, or
wildfires from international sources are beyond the jurisdiction of state agencies.
Footnotes:
88http://vista.cira.colostate.edu/improve/studies/BRAVO/reports/FinalReport/BRAVO/A5_Kuhns2003EmissInv.pdf
at 4-5.
89 http://www.nasa.gov/mission pages/fires/main/world/20130503-mexico.html.
90 http://www.nasa.gov/vision/earth/environment/central_am_fires_prt.htm,
http://alg.umbc.edu/usaq/archives/2013 05 .html.
91 EPA FIP Technical Support Document at A-30.
92EPA FIP Technical Support Document at A-30.
93 http://www.epa.gov/visibilitv/pdfs/introvis.pdf.
Response: We address comments concerning back trajectory analysis in a separate response to
comment.
827
-------
We agree with AECOM that impacts from international emissions can be significant, including
impacts from fires in southern Mexico, and Central America. We also agree those international
emissions need to be addressed to achieve natural visibility, but our agreement on this point does
not in any way relieve Texas of the obligation to make reasonable progress, including through
controls on its own sources, and particularly through the emissions addressed with controls
through our FIP.
We disagree with AECOM that impacts from Coleto Creek would be smaller than impacts from
Big Brown because it has less emissions. However, AECOM fails to consider the location of the
source and the meteorology/transport conditions. Coleto Creek is closer to Big Bend and our
source apportionment modeling shows that the one unit at Coleto Creek has a larger impact on
the 20% worst days at Big Bend than the impact from two units at Big Brown.664
AECOM suggests that because impacts from international sources are a large portion of the total
visibility impairment, no reductions from Texas point sources are reasonable. As we discuss in
our proposal and elsewhere in our response to comments, Texas and we agreed that it was
reasonable to focus on impacts from point sources for this planning period. The impact from
Texas point sources is significant, and as our analysis shows, a significant portion of this
impairment can be addressed by controlling a small number of sources. We discuss the relative
significance of impacts from Texas point sources at Guadalupe Mountains in response to other
comments above. For Big Bend, Texas EGUs contribute 4.5% of the total visibility impairment.
The required controls in this action address 1.88% of the total visibility impairment, 41.8% of the
impact from all Texas EGUs, and 23.4% of impact from all Texas point sources (based on 2018
CENRAP projected emission levels). Controls on Coleto Creek alone address approximately 6%
of the total Texas point source impact. Controls on two units at Big Brown address another
approximately 6% of the total Texas point source impact.
Furthermore, AECOM apparently fails to appreciate that the comparison it cites should be
properly understood to be between the visibility impact from one facility to the visibility impact
from all sources around the world that lie outside of the modeling domain, including long range
transport from fires, windblown dust, and significant anthropogenic emissions. AECOM states
that annual average visibility impairment from Big Brown is approximately 10% of the annual
average contribution from those sources captured by the boundary conditions. This is a
significant fraction of the total visibility impairment that can be addressed through the
installation of controls on only two emission units. We also note that visibility impairment on the
20% worst days at each Class I area from Big Brown is larger; as can be seen by the data
submitted by AECOM, on some days, the visibility impairment due to Big Brown's emissions
approach or exceed that from all emissions sources captured by the boundary conditions. For
Wichita Mountains, controls on just Big Brown addresses almost 12% of the total visibility
impairment due to Texas point sources and 1.63% of the total visibility impairment from all
sources. In summary, the visibility impairment from the individual sources analyzed is
significant and controls on these sources provide for meaningful visibility improvement at one or
more Class I areas towards the goal of natural visibility conditions. This is not inconsistent with
the understanding that significant impacts from international emissions and other sources exist
and should also be addressed.
664 See Figure A.3-3 of the FIP TSD.
828
-------
Lastly, we agree with CCP that the sources it cites, Carbon I and Carbon II, are responsible for
significant levels of pollution. Carbon I, is a 1,200 MW power plant and Carbon II is a 1,400
MW coal-fired power plant. These two power plants, less than 1.5 miles apart, are less than 20
miles from the U.S.-Mexico border. Together, these power plants comprise one of the largest
uncontrolled sources of SO2 and NOx in North America.665 It has been demonstrated for some
time that they are significant contributors to visibility impairment at Big Bend.666 However,
addressing international emissions can be complex. For instance, Texas has recently issued
water discharge and mining permits to a coal mine in Maverick County, near the Texas border
town of Eagle Pass, to allow the Mexican company Dos Republicas to begin mining coal that
will reportedly be sent to these facilities.667 Prior to our delegation of the National Discharge
Elimination System (NPDES) permitting authority to Texas, we issued a NPDES permit for the
operation of this mine, and in the process issued an Environmental Impact Statement (EIS).668 In
our EIS, we stated that"... EPA does not have the authority to prohibit export of U.S. resources
which will cause the country environmental harm . . . EPA believes that the U.S. policy should
be to take actions which will generate the investment capital needed to directly solve the Carbon
I/II problem"669 Subsequent to that, we attempted to work with the government of Mexico
specifically on the problem of installing controls on these sources through a technical work
group composed of EPA and SEMARNAP (now SEMARNAT, the Mexican Environment and
Natural Resources Secretariat) staff. Unfortunately, these discussions did not result in any
control of Carbon I and II. However, EPA is committed to explore opportunities for further
discussions with Mexico concerning this subject.
Comment: Mexican Emissions Will Impact Results More Than Predicted
[Alpine (0078) p. 22]
In a 2015 report attached to NRG's comment (0078), EPA agreed with the TCEQ and based on
the CENRAP PSAT modeling, that emissions and transport from Mexico and other international
sources will limit the rate of progress achievable on the 20% worst days and that efforts to meet
the goal of natural visibility by 2064 would require further emissions reductions not only within
Texas, but also large emission reductions from international sources.
Alpine provided a table with summaries of more recent modeling studies utilized by EPA to
estimate the emissions from Mexico. (Table 7 of Alpine comment 0078, not reproduced here.)
According to Alpine, this table demonstrates Mexican point source emissions of SO2 have shown
an increase since 2008 and are projected to increase even more through 2030.
(http://www.epa.gov/ttnchiel/conference/eil8/session2/wolf.pdf) This increase has created a
665 Commission for Environmental Cooperation of North America, "North American Power Plant Air Emissions,"
http://www.cec.org/storage/56/4876_powerplant_airemission_en.pdf. TCEQ may keep this in consideration in
future studies on the impacts of sources from Mexico on Class I areas or otherwise.
666 Big Bend Regional Aerosol and Visibility Observational Study (BRAVO), Final Report, September 2004.
667 http://www.epbusinessjournal.com/2015/ll/dos-republicas-coal-partnership-coal-mine-expanded-water-
discharge-permit-application-to-be-heard-november-16th/.
668 Authorization to Discharge Under the National Pollutant Discharge Elimination System. Permit No.
TX0109011.
669 Final Environmental Impact Statement on Dos Republicas Resource Company, Inc.'s Proposed Eagle Pass Mine
in Maverick County, Texas, December 30, 1994. Page C-51.
829
-------
higher magnitude of transported visibility impairing emissions to the Class I areas as compared
to the CENRAP modeling estimates.
Additionally, Alpine stated that with demonstrated reductions in SO2 from Texas emission
sources, this international component most likely will increase the Mexican percentage of
contribution to visibility impairment at both Big Bend and Guadalupe Mountains and should be
considered when establishing rate-of-progress determinations.
Alpine stated that, based on these most current emission estimates and projects in use by EPA,
the impact of Mexican source emission projections should be considered, and with greater
weight, during additional regional haze modeling to better attribute non-domestic visibility
impairment at the Big Bend and Guadalupe Mountains Class I areas. Should these current
Mexican emissions and their increased regionally significant effect on sulfate concentrations
have been used by EPA in their modeling, the influence of domestically generated emissions and
relative impact of incremental controls within the State of Texas could be substantially less than
what EPA predicts.
Response: We agree that increases in emissions from Mexico, either actual increases or
increases due to updated emissions inventories, would decrease the relative impact from Texas
sources. However, the absolute impact remains the same and as we have demonstrated in our
analysis, reasonable controls can address a significant portion of the visibility impairment at the
Wichita Mountains (as well as the two Texas Class I areas) due to Texas sources, and in
particular the impact due to Texas point sources. Alpine suggests that if visibility conditions
were to worsen at a Class I area due to an increase in international emissions, controls on Texas
sources would become less reasonable. This is not correct and it runs counter to the goal of the
regional haze program. If it held weight, it would allow a State to avoid addressing its share of
the visibility impairment simply because outside forces are also affecting the amount of progress
that can be made. This is precisely why it is important to consider not only the percentage of
visibility impairment, but also the absolute reduction in extinction and the visibility benefits of
controls compared to "clean" background when evaluating the reasonableness of individual
controls or a control strategy. We discuss these concepts in more detail elsewhere in this
document.
The Regional Haze Rule addresses situations where increases in emissions from other countries
prevent Class I areas from achieving the established reasonable progress goals. The five-year
progress report requires states to assess the adequacy of their regional haze SIP.670 If the state
determines at that time "that the implementation plan is or may be inadequate to ensure
reasonable progress due to emissions from sources in another country, the State shall provide
notification, along with available information, to the Administrator."671 However, anticipated
increases in emissions of international sources do not relieve the state from implementing
reasonable controls to address the visibility impairment due to their own sources.
670 40 CFR 52.308(h)
671 40 CFR 51.308(h)(3)
830
-------
Comment: EPA Should Further Strengthen the Reasonable Progress Goals for Big Bend
and Guadalupe Mountains. [Earthjustice (0067) p.50]
Earthjustice et al., stated that reasonable progress goals must "provide for an improvement in
visibility for the most impaired days over the period of the implementation plan and ensure no
degradation in visibility for the least impaired days over the same period." 40 C.F.R. §
51.308(d)(1). In establishing a reasonable progress goal, EPA must:
(1) Consider the "costs of compliance, the time necessary for compliance, the energy and
non-air quality environmental impacts of compliance, and the remaining useful life of
any potentially affected sources," 40 C.F.R. § 51.308(d)(l)(i)(A); and
(2) "Analyze and determine the rate of progress needed to attain natural visibility
conditions by the year 2064. To calculate this rate of progress, the State must compare
baseline visibility conditions to natural visibility conditions in the mandatory Federal
Class I area and determine the uniform rate of visibility improvement (measured in
deciviews) that would need to be maintained during each implementation period in order
to attain natural visibility conditions by 2064. In establishing the reasonable progress
goal, the State must consider the uniform rate of improvement in visibility and the
emission reduction measures needed to achieve it for the period covered by the
implementation plan." 40 C.F.R. § 51.308(d)(l)(i)(B).
Earthjustice et al., stated that in establishing RPGs for Texas and Oklahoma Class I areas, EPA
concluded that progress goals that provide for natural visibility by 2064 are unreasonable
because "emissions and transport from Mexico and other international sources will limit the rate
of progress achievable on the 20% worst days and that efforts to meet the goal of natural
visibility by 2064 would require further emissions reductions within Texas and from
international sources." 79 Fed. Reg. at 74,843. EPA proposed progress goals that will not
achieve natural visibility conditions at all Texas and Oklahoma Class I areas for 82 to 173 years,
well past the 2064 goal. Id. at 74,887. Those progress goals are unreasonable, at a minimum, in
light of the agency's recognition of available, cost-effective measures at additional facilities,
which would achieve a greater rate of progress toward the 2064 natural visibility goal at Wichita
Mountains, Big Bend, and Guadalupe Mountains.
Earthjustice et al., stated that where an implementation plan "establishes a reasonable progress
goal that provides for a slower rate of improvement in visibility than the rate that would be
needed to attain natural conditions by 2064, the State must demonstrate, based on the factors in
paragraph (d)(l)(i)(A) of this section, that the rate of progress for the implementation plan to
attain natural conditions by 2064 is not reasonable; and that the progress goal adopted by the
State is reasonable." 40 C.F.R. § 51.308(d)(l)(ii) (emphasis added). The data upon which Texas
and EPA relied in determining the impact of Mexico emissions is more than 15 years old.61 To
the extent that EPA factors international emissions in establishing RPGs for Texas, EPA should
update its calculations for international emissions so that it is supported by technical
demonstrations and is based on "available monitoring information and appropriate data analysis
techniques." Id. § 51.308(d)(2)(iii); see also 64 Fed. Reg. at 35,746.
831
-------
Moreover, although some of the haze pollution in Texas and Oklahoma Class I areas may be
attributable to international sources, the presence of such emissions cannot be used thwart the
development of appropriate RPGs nor excuse forgoing reasonable, available controls on in-state
sources. Put differently, the fact that some portion of impairment is attributable to international
sources does not automatically make EPA's alternative reasonable progress goal reasonable. See
40C.F.R. § 51.308(d)(l)(i)-(ii).
Indeed, EPA has made clear that states cannot use emissions from foreign sources "as a basis to
ignore controls on in-state sources where such controls are clearly reasonable."62 Where
emissions from international sources may affect a state's ability to meet RPGs, EPA has
suggested that a three-prong approach is appropriate: (1) the state first needs to address sources
within its control; (2) the state should then request reductions from contributing states; and (3)
the state should then ask EPA to address international emissions.63 To the extent that EPA has
not undertaken each of these actions with respect to Texas, it should do so.
Earthjustice et al., concluded, given that recent and reliable data demonstrate that there are
additional, cost-effective measures in Texas that will achieve greater progress toward natural
visibility in Texas and Oklahoma Class I areas, EPA cannot relieve itself of the obligation to
address those sources by relying on decades-old data regarding emissions from Mexico.64 Thus,
EPA should revise the reasonable progress goals for Big Bend and Guadalupe Mountains to
reflect the installation and operation of the additional controls listed above at Welsh, Parish, and
other Texas sources.
Footnotes:
61 See App'x B to CENRAP Modeling TSD, TX166-011-08 (describing file names and data sources for emissions
inventory).
62 EPA, Approval and Promulgation of Implementation Plans; North Dakota, Regional Haze Implementation Plan;
Federal Implementation Plan for Interstate Transport of Pollution Affecting Visibility and Regional Haze, 77 Fed.
Reg. 20,894, 20,913 (Apr. 2, 2012); see also 64 Fed. Reg. at 35,755 (July 1, 1999) (noting that while "EPA will
not hold States responsible for developing strategies to 'compensate' for the effects of emissions from foreign
sources,. .. [s]tates should not consider the presence of emissions from foreign sources as a reason not to strive to
ensure reasonable progress in reducing any visibility impairment caused by sources located within their
jurisdiction"). 63 U.S. EPA Region 8, Comments on August 21, 2009 Draft Regional Haze SIP (FLM Consultation
Version), Enclosure 1 at 5 (comments on North Dakota Regional Haze SIP).
64 In any event, EPA has a duty to address international pollution transport concerns. 64 Fed. Reg. 35,755. If EPA
agrees with Texas's finding regarding interstate pollution impacts, then EPA should take appropriate action to
address the international emissions through available mechanisms. Id. In particular, EPA should actively encourage
Mexico to reduce emissions impairing visibility in Texas, Oklahoma, and elsewhere. EPA is well within its rights to
formally request reductions from Mexico where appropriate.
63 U.S. EPA Region 8, Comments on August 21, 2009 Draft Regional Haze SIP (FLM Consultation Version),
Enclosure 1 at 5 (comments on North Dakota Regional Haze SIP).
Response: We agree with Earthjustice that the presence of international emissions in no way
relieves Texas of the requirements of the Regional Haze Rule to consider the four factors and
evaluate controls for addressing visibility impairment due to Texas sources and making
reasonable progress towards the goal of natural visibility conditions. We disagree that the
progress goals we proposed are unreasonable because cost-effective measures at additional
facilities would provide for additional visibility benefit. As we discuss elsewhere in addressing
specific comments concerning additional controls, based on our evaluation of the four factors
832
-------
and consideration of visibility within the cost factor of the four factor reasonable progress
analysis, we determined that additional controls at Parish, Welsh and other sources were not
required for reasonable progress for the first planning period.
We utilized the emission inventories developed by CENRAP and used by Texas in our analysis,
including the emissions inventory for Mexico. We agree that there is uncertainty in these
emission inventories and that ongoing efforts to improve these inventories should be
incorporated into future regional haze planning evaluations. As we discuss above, the Regional
Haze Rule addresses situations where increases in international emissions prevent Class I areas
from achieving reasonable progress towards natural visibility conditions.
We address comments concerning how we considered international emissions in a separate
response above.
19. Grid Reliability
Comment: [TCEQ/PUCT (0056) p. 22] The TCEQ recommended that the EPA withdraw the
proposed FIP; however, if the EPA does finalize the FIP, the EPA should include an electric
reliability safety valve provision in the final rule.
The TCEQ maintained that its 2009 Regional Haze SIP is approvable as submitted and the EPA
should withdraw the proposed FIP. However, if the EPA does finalize the FIP then the final rule
should include a reliability safety valve provision. The EPA has not considered the potential
electric reliability implications of the proposed rule. A reliability safety valve provision in the
rule could be a provision that allows the EPA to grant an extension to the compliance dates in
situations where electric reliability is at risk, after consultation with the appropriate Independent
System Operator /Regional Transmission Organization.
[TCEQ/PUCT (0056) p. 20, and Appendix 1: ERCOT report] The TCEQ stated that the EPA
has not evaluated any potential impacts of the proposed FIP to reliability and prices of electricity
in Texas, as further discussed below. In 2014, the Electric Reliability Council of Texas (ERCOT
conducted a study of the impacts that environmental regulations have in the ERCOT Region.
The report, entitled "Impacts of Environmental Regulations in the ERCOT Region," was
finalized on December 16, 2014, and is included as Appendix 1 to the TCEQ's comments
(summarized below). While the report included a number of environmental regulations, such as
the MATS rule, Clean Power Plan, and CSAPR, ERCOT also included the EPA's proposed
Regional Haze FIP for Texas in its analysis. The TCEQ incorporates the ERCOT report into the
agency's comments and encourages the EPA to consider the findings of the ERCOT report.
ERCOT is the independent system operator (ISO) for the ERCOT Interconnection, which
encompasses approximately 90% of electric load in Texas. ERCOT is the independent
organization established by the Texas Legislature to be responsible for the reliable planning and
operation of the electric grid for the ERCOT Interconnection. Under the North American
Electric Reliability Corporation (NERC) reliability construct, ERCOT is designated as the
Reliability Coordinator, the Balancing Authority, and as a Transmission Operator for the
833
-------
ERCOT region. ERCOT is also registered for several other functions, including the Planning
Authority function.
ERCOT noted that there are several proposed or recently finalized EPA regulations that could
have an impact on grid reliability in ERCOT. These rules include the Mercury and Air Toxics
Standards (MATS), CSAPR, the Regional Haze program, the Cooling Water Intake Structures
rule, the Steam Electric Effluent Limitation Guidelines (ELG) rule, the Coal Combustion
Residuals (CCR) Disposal rule, and the Clean Power Plan. The ERCOT study assesses the
individual and cumulative impact of these regulations on generation resources in the ERCOT
region, and potential implications for grid reliability.
Resource owners in ERCOT will need to take actions to comply with these regulations in the
coming years, or else retire or mothball the units. Tables provided by ERCOT (Tables ES-1 and
Table ES-2 of comment 0056-A3 [not reproduced here]) show the potential compliance
requirements for coal and natural gas units, respectively, under these regulations.
According to the ERCOT report, coal units are the most affected by environmental regulations.
Without considering the Clean Power Plan, 3,000 MW to 8,500 MW of coal-fired capacity in
ERCOT can be considered to have a moderate to high risk of retirement - due primarily to the
costs of EPA's proposed requirements for the Regional Haze program. The results of this
analysis also suggest potential impacts from CSAPR in the short-term. By comparison, the other
regulations are not expected to have a significant system-wide impact, but could affect the
economics of a small number of units. The implementation and regulatory timeline of the Clean
Power Plan will impact decisions resource owners make about whether to retrofit or retire
impacted units. Additionally, the Clean Power Plan itself may cause unit retirements, due to the
need to meet stringent CO2 emissions limits on a state-wide basis. ERCOT's modeling analysis
suggests that the Clean Power Plan, in combination with the other regulations, will result in the
retirement of up to 8,700 MW of coal-fired capacity.
The results of the ERCOT study indicate that the Regional Haze requirements and the Clean
Power Plan will have significant impacts on the planning and operation of the ERCOT grid.
Both are likely to result in the retirement of coal-fired capacity in the ERCOT region. Currently,
resource owners are required to notify ERCOT no less than 90 days prior to the date that the unit
is retired or mothballed. Given the competitiveness of the ERCOT market and the current
uncertainty surrounding environmental regulations, it is unlikely that generators would notify
ERCOT of potential retirements or unit suspensions before the minimum notification deadline.
If ERCOT does not receive early notification of these retirements, and if multiple unit
retirements occur within a short timeframe, there could be periods of reduced system-wide
resource adequacy and localized transmission reliability issues due to the loss of generation
resources in and around major urban centers. Additionally, loss of the reliability services
provided by retiring units will strain ERCOT's ability to integrate new intermittent renewable
generation resources.
The Clean Power Plan will also result in increased wholesale and consumer energy costs in the
ERCOT region. Though the other regulations considered in this study will pose costs to owners
of generation resources, they are less likely to significantly impact costs for consumers.
834
-------
ERCOT undertook two parallel efforts for this study. First, in the summer of 2014, ERCOT
distributed a survey to fossil fuel-fired generators on the impacts of relevant environmental
regulations. The responses indicate the current compliance status of fossil fuel-fired resources in
the ERCOT region. Second, ERCOT conducted a modeling analysis of the impacts of CSAPR,
the Regional Haze program, and the Clean Power Plan on generation resources and energy costs
in the ERCOT region. The ERCOT report:
• Provides an overview of the environmental regulations evaluated in this study and
describes prior ERCOT analyses related to the potential impacts of environmental
regulations (Section 1);
• Discusses the requirements and associated costs of environmental regulations for
generation resources (Section 2);
• Presents the results of the generator survey which asked about planned retirements or
curtailments due to environmental regulations, currently installed control technologies,
and compliance strategies (Section 3 and Appendix A);
• Describes the methodology and results of ERCOT's modeling analysis performed using
the Energy Exemplar's PLEXOS Integrated Energy Model for different regulatory
scenarios (Section 4);
• Discusses the impacts of these regulations for grid reliability in the ERCOT region
(Section 5); and
• Presents a cost analysis of the relevant environmental regulations (Section 6)
The results of the ERCOT study indicate that the Regional Haze program and the Clean Power
Plan will both lead to the retirement of coal-fired capacity in ERCOT. EPA's proposed Regional
Haze FIP is likely to result in the retirement of coal units due to the costs associated with
upgrading and retrofitting scrubbers. ERCOT anticipates that 3,000 MW to 8,500 MW of coal-
fired capacity in ERCOT face a moderate to high risk of retirement due to these requirements. If
implemented as proposed, the Clean Power Plan will also result in coal unit retirements, due to
the need to meet stringent C02 emissions limits on a state-wide basis. ERCOT's analysis
suggests that the Clean Power Plan, in combination with other environmental regulations, will
result in the retirement of up to 8,700 MW of coal-fired capacity. By comparison, the other
regulations are not expected to have a significant system-wide impact, but could affect the
economics of a small number of units.
The retirement of existing capacity in ERCOT could result in localized transmission reliability
issues due to the loss of fossil fuel-fired generation resources in and around major urban centers,
and will strain ERCOT's ability to integrate new intermittent renewable generation resources. If
the expected retirement of coal resources were to occur over a short period of time, reserve
margins in the ERCOT region could reduce considerably, leading to increased risk of rotating
outages as a last resort to maintain operating balance between customer demand and available
generation. The need to maintain operational reliability (i.e., sufficient ramping capability) could
require the curtailment of renewable generation resources.
[GCLC (0063) p. 15-16] According to GCLC, EPA believes that energy concerns will not be
significant, because once again, by relying on BART Guidelines, EPA has only taken a narrow
835
-------
view of the energy penalties of controls at each individual unit. However, viewed on a larger
scale (i.e. statewide) the energy impacts of complying with the Proposed FIP may be significant,
particularly in the ERCOT region. EPA has failed to consider these broader impacts of its
proposal.
ERCOT is predicting that by 2020, ERCOT will have a reserve margin of 12.4%.68 By 2024, this
reserve margin will shrink to 7.3%, far below the 13.75% target margin for the region and the
margin that the NERC demands. The Proposed FIP will impact 10,131 MW of installed Texas
capacity either through forced scrubber upgrades or scrubber retrofits.69 If the ERCOT units
targeted with scrubber retrofits are forced to retire, this would take 2,972 MW off of the ERCOT
grid, reducing the reserve margin to 8.3% in 2020 and 3.3% in 2024, which places the grid in
substantial jeopardy when it comes to preserving reliability.
Importantly, GCLC noted that ERCOT's estimates regarding predicted capacity and demand do
not include the impacts of environmental rules. Therefore the predicted impacts of the Clean
Power Plan, the 316(b) Rule, the MATS rule and others are not reflected in this shortfall
calculation. Therefore, the predicted reserve margin will be even lower than stated above.
GCLC asserted that EPA's Proposed FIP could have a very significant effect on
energy/electricity availability in the state, which is a clear and convincing reason that EPA
should not continue with its Proposed FIP or impose the source-specific limitations it
contemplates. This was something that was not accounted for by EPA in its cost analysis, nor in
its analysis of energy impacts.
Footnotes:
68 ERCOT, Report on the Capacity, Demand, and Reserves in the ERCOT Region, 2015-2024, 9 (Dec. 1, 2014).
Available at:
http://www.ercot.com/content/gridinfo/resource/2014/adequacy/cdr/CapacityDemandandReserveReportDec2014.pd
f.
69 See U.S. Energy Information Administration, Form EIA-860 Annual Electric Generator Report, 2012 Data,
available at http://www.eia.doe.gov/cneaf/electricity/page/eia860.html.
Response: First, we note that controls achieving the level of control that we are requiring are
highly cost-effective, are in wide use in the industry, and thus should not require a source to shut
down to comply. In response to the TCEQ's comments, however, we contracted with Synapse
Energy Economics, Inc., a nationally recognized firm with particular expertise in the subject
area. (Synapse).672 Synapse assessed the information in the ERCOT report and we reproduce its
findings below:
1. ERCOT's perspective of market operations is short-sighted. ERCOT raises concerns that
reliability could be impacted if numerous coal units choose to retire simultaneously with
little notice to either ERCOT or other market participants. Unlike other competitive
market regions, ERCOT's rules do not require meaningful notice. ERCOT's charge as a
reliability coordinator may obligate it to implement rules requiring reasonable notice for
economic retirements.
672 Synapse's report, "ERCOT_Report_Review_Memo_20150908.pdf' is in our docket to this rulemaking action.
836
-------
2. ERCOT's assumptions about new gas turbine capacity are not realistic. While the FIP,
along with other environmental regulations ERCOT included in its study, will strain the
economic viability of coal plants and likely lead to less coal capacity, ERCOT has not
considered new resources that will be available to help address potential reliability
challenges. Specifically, ERCOT does not include approximately 4,500 MW of
additional gas-fired capacity coming online in Texas in the upcoming years. This
represents 7.5 percent of current gas capacity, and would double the modeled baseline
gas capacity additions through 2029.
3. The set of regulatory scenarios modeled is both incomplete and (now) outdated. Despite
an overall thorough analysis ERCOT excluded a critical scenario that would have
modeled the impact of the Regional Haze Program FIP by itself. This limits inferences
we can make about impacts. Additionally, since ERCOT finalized its study, EPA
finalized the Clean Power Plan. The final rule includes substantive changes that are
likely to affect all of the CO2 limit and price-inclusive scenario modeling results.
4. Electric Generating Unit owners' compliance "burdens" with the regional haze FIP may
be over-stated. Of the 15 coal-fired units subject to regional haze compliance
requirements, eight require upgrades to their existing scrubbers rather than new
scrubbers. ERCOT assumed that all of the scrubbers would be priced at the cost of a new
retrofit, thereby substantially increasing the cost of the regulation.
We reviewed and accept Synapse's findings and adopt its conclusion that ERCOT's report
contained significant flaws. In sum, ERCOT's report cannot support a determination that there is
likely to be any significant, adverse effect on the supply, distribution, or use of energy. During
our comment period, we received no, non-speculative information to validate claims that sources
would retire rather than install demonstrably cost-effective controls. Commenters who have
alleged grid reliability concerns in response to our proposed controls have not provided adequate
documentation for their assertions.
20. Determination of Nationwide Scope and Effect
Comment: [Earthjustice at al. (0067) p. 34] Earthjustice agreed that any petitions for review of
the rule must be filed in the D.C. Circuit Court of Appeals. Even if EPA were not clarifying its
interpretation of the existing haze rule in this action, EPA's rule takes action on two SIPs, the
review of which would normally occur in two separate circuits. For this reason alone, review of
the rule must occur in the D.C. Circuit.
Response: We appreciate the commenters' support, but note that we must make a determination
of nationwide scope or effect and publish that finding before judicial review of a local or
regionally applicable rule must occur in the D.C. Circuit. We have taken both steps here.
Comment: [TCEQ/PUCT (0056) p. 19] The TCEQ disagreed with the EPA's assertion that this
action is a rulemaking of nationwide scope and effect. Any appeal of the EPA's final action on
Texas' regional haze plan and FIP should be filed in the 5th Circuit Court of Appeals.
837
-------
The TCEQ Stated that the EPA argues that the proposed FIP and SIP disapproval actions for
Texas and Oklahoma have nationwide scope and effect and therefore, under CAA, §307(b)(l),
appeal must be to the D.C. Circuit. First, the TCEQ notes that the EPA has in fact taken the
opposite position in several final actions on regional haze plans in Oklahoma, New Mexico and
Arizona.17
The TCEQ stated that these EPA actions do not have nationwide scope and effect; they are not
nationally applicable, but apply only to two States. The EPA has provided no legal basis -
beyond a one sentence assertion- to support that its actions interpreting the RHR as they apply to
Texas and Oklahoma are of "nationwide scope and effect." This interpretation of the RHR as it
applies to Texas and Oklahoma Regional Haze SIPs is unsupported by the EPA's proposed
action. The action here specifically deals with plans adopted by Texas and Oklahoma to meet
the CAA and regional haze regulations as they apply in their respective jurisdictions. Each
regional haze plan submitted by the various States is unique, addressing visibility impairment at
Class I areas in those States and in surrounding States. The EPA's proposed partial disapproval
of Texas' plan and proposed imposition of a FIP does not rely solely on an interpretation of their
rules but rather on a review of the Texas plan's comportment with those rules. The EPA has
proposed determinations that Texas did not develop its natural visibility conditions and RPG
correctly.
The EPA then goes on to draft RPG controls for 15 Texas units and redo the natural visibility
estimates. This proposal is Texas-centric; it is not nationally applied.
The TCEQ Stated that the EPA then attempts to plug the obvious hole in its position by pointing
to congressional report language that allows the Administrator to determine its action has
nationwide scope and effect if the rulemaking extends to two judicial districts. This is not found
in the CAA. In fact, §307(b)(l) specifically States that "any implementation plan" or "any other
final action of the Administrator under this chapter .... which is locally or regionally applicable
may be filed only in the United States Court of Appeals for the appropriate circuit." The fact
that Oklahoma is in the Tenth Circuit and Texas is in the Fifth Circuit is immaterial to potential
petitions for review. The TCEQ's comments and any future actions it may or may not take in
court will be based on the EPAs action on Texas' SIP and any FIP the EPA has imposed on
Texas, not Oklahoma. As Stated previously, venue for regional haze plans in several
neighboring States, including Oklahoma, is already established in their respective circuits.
Footnote
17 See for example: 79 FR 12944, 12954 March 7, 2014; 77 FR 70693, 70705, Nov. 27, 2012; 78 FR 46142, 46174
July 13, 2013; 79 FR 52420, 52479, Sept. 3, 2014.
Response: We disagree with this comment. The commenter is conflating two distinct portions
of the CAA's judicial review provision. Under CAA Section 307(b)(1), "[a] petition for review
of. . . nationally applicable regulations promulgated, or final agency action taken, by the
Administrator . . . may be filed only in the United States Court of Appeals for the District of
Columbia." Contrary to the commenter's assertions, the EPA did not assert at proposal, nor do
we assert now, that our FIP for Texas and Oklahoma is a "nationally applicable" regulation.
CAA Section 307(b)(1) next provides that "[a] petition for review of the Administrator's action
in approving or promulgating any implementation plan under section 7410 ... or any other final
action of the Administrator . . . which is locally or regionally applicable may be filed only in the
838
-------
United States Court of Appeals for the appropriate circuit." The commenter cites this sentence,
but ignores the following sentence, which States "[notwithstanding the preceding sentence a
petition for review of any action referred to in such sentence may be filed only in the United
States Court of Appeals for the District of Columbia if such action is based on a determination of
nationwide scope or effect and if in taking such action the Administrator finds and publishes that
such action is based on such determination."
In other words, a final agency action that is locally or regionally applicable, such as a FIP, is
appealable only in the D.C. Circuit if two conditions are met: (1) the action is based on a
determination of nationwide scope or effect, and (2) EPA finds and publishes its determination.
Both conditions are met here. First, we proposed to find and have confirmed its finding in this
final rule that the Agency's action on the Texas and Oklahoma regional haze SIPs, which
includes the promulgation of a partial FIP for each State, is based on a determination of
nationwide scope and effect. Second, we have published that finding in the Federal Register.
While the CAA does not provide any guidance regarding the phrase "nationwide scope and
effect," the legislative history indicates that a determination of nationwide scope and effect is
appropriate if a local or regional action encompasses two or more judicial circuits, as is the case
with Texas and Oklahoma in this action. The commenter makes no effort to explain why this
legislative history should not be taken into account. Instead, the commenter cites to three other
EPA actions on regional haze SIPs where the Agency did not make a determination of
nationwide scope and effect. However, the commenter fails to mention that all of these actions
involved a single State and thus did not implicate multiple judicial circuits. We have routinely
made determinations of nationwide scope and effect when more than one circuit is involved.
Last year, for instance, we made a determination of nationwide scope and effect in a SIP
approval action that involved the States of Florida and North Carolina, which reside in separate
judicial circuits. See 79 FR 29362. We have made many other such determinations over the
years and similarly do so for this matter. Again, and as stated on our final action, the scope and
effect of this rulemaking extend to Texas and Oklahoma, which are located in two judicial
circuits.
We also determined that this action has nationwide scope and effect because at the core of this
rulemaking is our interpretation of the requirements of Sections 110(a)(2)D)(i)(II) and
169A(b)(2) of the CAA and multiple complex provisions of the Regional Haze Rule. Many
commenters disagreed with our interpretation of these provisions, with some providing
alternative interpretations that would substantially change the Regional Haze Rule, which is
implemented on a nationwide basis. Congress intended for such issues of national importance to
be decided by the D.C. Circuit. Therefore, it does not matter that the final rule's requirements
are largely Texas-centric, which reveals only that the final rule is locally or regionally applicable.
It says nothing of whether the final rule has nationwide scope and effect.
Comment: Any judicial challenge to a final rule must be heard in the U.S. Court of Appeals
for the Fifth Circuit. [CCP (0075) p. 15]
839
-------
CCP noted that the EPA States that the Texas and Oklahoma FIPs and interpretations therein are
of "nationwide scope and effect." See 79 Fed. Reg. 74,888. This is inconsistent with the
Congressional mandate in the CAA that rules with only local or regional effect should be heard
in the appropriate appellate court and not in the D.C. Circuit.
CCP stated that other FIP challenges were heard in other circuit courts of appeals for the
appropriate region and not in the D.C. Circuit. See Wyoming v. EPA, No. 14-9529 (10th Cir.)
(reviewing Wyoming SIP Disapproval/FIP); North Dakota v. EPA, 730 F. 3d 750 (8th Cir.
2013) (reviewing North Dakota SIP Disapproval/FIP); Oklahoma v. EPA, 723 F.3d 1201, 1213
n.7 (10th Cir. 2013) (reviewing Oklahoma SIP Disapproval/FIP). EPA fails to identify any
issue of sufficient "nationwide scope and effect" to require review in the D.C. Circuit.
Response: We disagree with this comment. Section 307(b)(1) reflects Congress's intent that
some local and regional actions should be reviewed in the D.C. Circuit only. The actions cited
to by the commenter all involved a single State in a single judicial circuit. In contrast, this final
rule involves two States in two separate judicial circuits. Moreover, we explained in detail why
the interpretations at the core of this rulemaking are nationally significant.
Comment: EPA appears to be attempting to dictate which appellate court should have
jurisdiction over any appeal of the final FIP. [Xcel Energy (0064) p. 25]
Xcel Energy stated that the EPA inappropriately suggests that the Proposal is unique in its
"nationwide scope and effect," which would site jurisdiction over any appeals of the Proposal in
the U.S. Court of Appeals for the D.C. Circuit ("D.C. Circuit"). See 79 Fed. Reg. at 74,888.
This is a blatant and improper attempt to override Congress' mandate in the CAA that rules with
only local or regional effect should be heard in the United States Court of Appeals for the
appropriate circuit and not in the D.C. Circuit. 42 U.S.C. § 7607(b)(1). EPAs final Regional
Haze FIP for Texas will determine what additional controls to impose on specific Texas sources
to achieve reasonable progress toward natural visibility goals for Class I areas located in Texas
and Oklahoma. It is an action of inherently local or regional effect that must be reviewed in the
U.S. Court of Appeals for the Fifth Circuit.
Xcel Energy Stated that the EPA has proposed interpretations in this Proposal that might have
applicability to other State regional haze FIPs. However, that does not make the Proposal of
"nationwide scope and effect." EPA also has proposed interpretations in other State regional
haze FIPs that might have applicability to Texas, but EPA has never claimed that those FIPs
were of nationwide scope and effect. In fact, most recently, on April 8, EPA proposed a regional
haze FIP for Arkansas. 80 Fed. Reg. 18,944 (Apr. 8,2015). Nowhere in the proposed
Arkansas FIP does EPA suggest that its interpretations therein are of nationwide scope and
effect. Instead, the proposed Arkansas FIP and other prior regional haze FIPs have been treated
as having "only local or regional effect." Indeed, other FIP challenges have already been heard
in other circuit courts of appeals for the appropriate region and not in the D.C. Circuit. See
Wyoming v. EPA, No. 14-9529 (10th Cir.) (reviewing Wyoming SIP Disapproval/PIP); North
Dakota v. EPA, 730 F.3d 750 (8th Cir. 2013) (reviewing North Dakota SIP Disapproval/PIP);
Oklahoma v. EPA, 723 F.3d 1201, 1213 n.7 (10th Cir. 2013) (reviewing Oklahoma SIP
840
-------
Disapproval/FIP). The final FIP for Texas should clearly State that any appeals of the FIP
should be heard in the U.S. Court of Appeals for the Fifth Circuit.
Response: We disagree with this comment. Section 307(b)(1) reflects Congress's intent that
some local and regional actions should be reviewed in the D.C. Circuit only. The actions cited
to by the commenter all involved a single State in a single judicial circuit. In contrast, this final
rule involves two States in two separate judicial circuits. It does not just involve the Fifth
Circuit, as the commenter suggests. Moreover, we explained in detail why the interpretations at
the core of this rulemaking are nationally significant.
Comment: EPA's action is reviewable in the 5th Circuit. [NRG (0078) p. 14]
NRG noted that the EPA has proposed to determine that its action is only reviewable in the D.C.
Circuit, on the basis that:
[T]his rule is based on a determination of nationwide scope and effect. The rule
discusses our interpretation of multiple provisions of the Regional Haze Rule and
explains how those provisions operate in the visibility-transport context. Our
interpretation of our regulations is applicable to all States, not just Texas and Oklahoma.
79 Fed. Reg. at 74,888.
NRG disagreed that this action is reviewable in the D.C. Circuit. The Clean Air Act provides
that a petition for review of "a locally or regionally applicable" EPA action "may be filed only in
the United States Court of Appeals for the appropriate circuit" unless it "is based on a
determination of nationwide scope or effect." 42 U.S.C. § 7607(b)(1). Further, if EPA's
proposed finding on judicial review were correct, no State implementation plan challenge would
ever lie outside the D.C. Circuit, despite the statute's role for other Circuits and the substantial
history of past decisions on State plans that have been determined in the various Circuits.
The proposal is locally or regionally applicable. NRG Stated that the "locally or regionally
applicable" nature of EPA's action is made clear by the fact that the proposal would only impose
new emission controls on a small number of facilities in Texas. A further indication that the
action is "locally or regionally applicable" is that it only regulates three out of over 100 areas
subject to the statutory provisions on regional haze See, e.g., htto://www.epa.gov/ttn/oaroq/tl/fr
notices/classimp.gif (last accessed April 2, 2015). Also notably, the proposed rule was signed by
EPA's Region 6 Regional Administrator Curry, based in Dallas, not by Administrator McCarthy
in Washington, D.C.
NRG noted that courts outside the D.C. Circuit have frequently exercised jurisdiction over
challenges to EPA regional haze rules that address specific States. For example:
In October 2014, the 10th Circuit rejected on the merits a challenge by environmental
groups to EPA's approval of a three-State emissions trading program to satisfy
regional haze rule requirements. WildEarth Guardians v. EPA, 770 F.3d 919 (10th
Cir. 2014).
841
-------
In July 2014, the 9th Circuit rejected on the merits a challenge by environmental
groups to EPA's approval of BART determinations for sources in Nevada. WildEarth
Guardians v. EPA, 759 F.3d 1064 (9th Cir. 2014)
In September 2013, the 8th Circuit issued an opinion on the merits on BART and
reasonable progress plan issues for North Dakota. North Dakota v. EPA, 730 F.3d
750 (8th Cir. 2013).
NRG Stated that the EPA's litigation position subsequent to the current proposal further
illustrates the distinction between this action, reviewable in the 5th Circuit, and other regional
haze actions for which judicial review may lie in the D.C. Circuit. On January 30, 2015, EPA
filed a brief with the 8th Circuit concerning a challenge to the Minnesota regional haze plan.
EPA's rationale as presented to the court was that the Minnesota lawsuit was a collateral attack
on the substance of a prior nationwide EPA rule finding that EPA-promulgated emissions trading
programs were superior to BART for regional haze purposes, as EPA had merely approved
Minnesota's reliance on the EPA emissions trading program. January 30, 2015 brief of EPA,
National Parks Conservation Association v. EPA, Nos. 12 2910, 12- 3481 (8th Cir.).
By contrast to the Minnesota lawsuit, NRG's objections to EPA's current proposal do not
contradict EPA's underlying regional haze rules. These objections are fact-specific to how EPA's
regulations should be applied to Texas sources. Thus, the current proposal is exemplary of the
type of "locally or regionally applicable" action for which judicial review lies in the 5th Circuit.
The proposal does not rely on a determination of nationwide scope and effect. NRG also
disagreed that the proposal relies on "a determination of nationwide scope and effect." Rather,
the proposal concerns the application of national regulations to fact-specific circumstances at a
small number of emissions sources and protected areas.
Response: We disagree with this comment. Section 307(b)(1) reflects Congress's intent that
some local and regional actions should be reviewed in the D.C. Circuit only. The actions cited
to by the commenter all involved a single State in a single judicial circuit. For example, not only
did we act separately on each of the four State plans at issue in WildEarth Guardians v. EPA,
770 F.3d 919 (10th Cir. 2014), but each State was in the Tenth Circuit. In contrast, this final
rule involves two States in two separate judicial circuits. It does not just involve the Fifth
Circuit, as the commenter suggests. Moreover, we explained in detail why the interpretations at
the core of this rulemaking are nationally significant.
We also note that the Regional Administrator is delegated with the authority to sign proposed
FIPs. Only the Administrator can sign this final rule.
Comment: EPA's Regional Haze FIP is not a rule of "nationwide scope and effect." [GCLC
(0063) p. 19-20]
According to GCLC, EPA declares in the preamble of the Proposed FIP that "this is a
rulemaking of nationwide scope or effect such that any petitions for review must be filed in the
U.S. Court of Appeals for the District of Columbia Circuit." 77 EPA's position is both factually
and legally incorrect.
842
-------
GCLC stated that the Proposed FIP addresses only two States' SIP submissions - Texas and
Oklahoma. Grouping these two States in a single SIP submission does not make a rule one of
"nationwide scope of effect." Further, in actual application, the only legal obligations imposed
by the Proposed FIP is limited to Texas and Texas generators; there are no substantive burdens
imposed on Oklahoma or its generations sources, so practically, the rule is limited solely to
Texas. EPA attempts to claim that "[its] interpretation of [its] regulations is applicable to all
States, not just Texas and Oklahoma."78 By this logic, any SIP rulemaking would be of
nationwide scope, because all SIP approvals and disapprovals require interpretation. This is
clearly contradictory to the purpose of Section 307(b)(1) 79 and if proven true, would effectively
read the judicial review provisions out of the CAA.
Footnotes:
77 Proposed FIP, 79 Fed. Reg. at 74888.
78 Proposed FIP, 79 Fed. Reg. at 74888.
79 42 USC §7607(b)(l).
Response: We disagree with this comment. Section 307(b)(1) reflects Congress's intent that
some local and regional actions should be reviewed in the D.C. Circuit only. This final rule
involves two States in two separate judicial circuits and, as we have explained in detail, includes
statutory and regulatory interpretations that are nationally significant. The commenter is
incorrect that under our logic, any SIP rulemaking would be of nationwide scope and effect. For
judicial review to lie in the D.C. Circuit, we must make an affirmative determination and
publish that finding. As many commenters have pointed out, we typically do not make such
findings in SIP rulemakings unless more than one judicial circuit is involved and interpretations
of national importance are at issue.
Comment: If EPA adopts its Proposal, judicial review of it will be proper in the Fifth
Circuit Court of Appeals [AECT (0074) p. 10]
AECT disagreed with EPA's assertion that any petitions for review must be filed in the U.S.
Court of Appeals for the D.C. Circuit because EPA's adoption of the Proposal would have
"nationwide scope or effect". EPA's assertion that its adoption would have "nationwide scope or
effect" is based on its claims that (i) the Proposal discusses EPA's interpretations of provisions in
its Regional Haze rules and such interpretations would be applicable to all States, and (ii) its
adoption would have "scope or effect beyond a single judicial circuit".31 AECT does not see how
either of those bases would support EPA's assertion that its adoption would have "nationwide
scope or effect."
AECT noted that EPA's interpretation of its Regional Haze rules in its Proposal cannot be
applicable to all States since EPA has already made determinations on the Regional Haze SIPs
under those rules for all States except for the two States that are covered by its Proposal.32
Notwithstanding that, it cannot be true that an EPA action regarding one State's SIP has
"nationwide scope or effect" merely because that action involved EPA interpreting its rules that
the State's SIP is addressing and such interpretations might be used in its evaluations of other
States' SIPs that address those rules. If that was true, AECT cannot conceive of an EPA action
843
-------
regarding a SIP that would not have "nationwide scope or effect," and, thus, would be subject to
judicial review in the applicable circuit court of appeals.
AECT Stated, in addition, if EPA was to adopt its Proposal, that action would not have "scope or
effect beyond a single judicial circuit". That action would only impose legal requirements on
EGUs in Texas. Accordingly, any legal challenge to that action would only relate to its impacts
on EGUs in Texas. Since Texas is only covered by a single judicial circuit— the Fifth Circuit —
EPA's action would not have "scope or effect beyond a single judicial circuit", and thus, would
not have "nationwide scope or effect".
Footnotes:
31 79 Fed. Reg. at 74888
32 79 Fed. Reg. at 74820 (stating that EPA has "acted on all of the States' regional haze SIPs for the first planning
period except for the Texas regional haze SIP and certain portions of the Oklahoma regional haze SIP")
Response: We disagree with this comment. Section 307(b)(1) reflects Congress's intent that
some local and regional actions should be reviewed in the D.C. Circuit only. Contrary to the
commenter's claim, this final rule involves two States in two separate judicial circuits. We have
partially approved and partially disapproved Texas' regional haze SIP and have promulgated a
FIP. We have also partially approved and partially disapproved Oklahoma's regional haze SIP
and promulgated a partial FIP. While the commenter is correct that the FIP imposes emission
limits on Texas sources but not Oklahoma sources, this does not mean that we have not taken
action on both States' SIPs. More importantly, our rationale for requiring SO2 reductions at
Texas EGUs is inextricably linked to our evaluation of the Oklahoma SIP.
Moreover, the EPA explained in detail why the interpretations at the core of this rulemaking are
nationally significant. The commenter is incorrect that under our logic, any SIP rulemaking
would be of nationwide scope and effect. For judicial review to lie in the D.C. Circuit, the EPA
must make an affirmative determination and publish that finding. As many commenters have
pointed out, the EPA typically does not make such findings in SIP rulemakings unless more than
one judicial circuit is involved and interpretations of national importance are at issue.
In summary, the commenter is incorrect that judicial review is appropriate in the Fifth Circuit.
The EPA has finalized and published its determination that judicial review can only be had in the
D.C. Circuit.
21. Reasonable Progress Determination for the Texas Class I Areas
Comment: Texas' reasonable progress analysis meets all statutory and regulatory
requirements, and EPA must approve it. Texas' Reasonable Progress Goals for Big Bend
and Guadalupe Mountains meet all statutory and regulatory requirements. [Luminant
(0061) p. 59]
Luminant stated that Texas established reasonable progress goals for 2018 for the two Class I
areas in the state—Big Bend and Guadalupe Mountains—that fully meet all statutory and
regulatory requirements, and, as such, EPA must approve them. For Big Bend and Guadalupe
844
-------
Mountains, Texas established goals for 2018, expressed in deciviews, that provide for an
improvement in visibility for the most impaired days and ensure no degradation in visibility for
the least impaired days over the period of the implementation plan (2008-2018). After
considering the four statutory factors, Texas determined that the uniform rate of progress was not
reasonably achievable by 2018, and thus Texas established interim goals for 2018 of 16.6
deciviews for Big Bend and a goal of 16.3 deciviews for Guadalupe Mountains, based on
CENRAP modeling.422 Of particular relevance to Texas' determination was CENRAP's
modeling that showed that the uniform rate of progress could not be achieved due to emissions
from outside the United States, which Texas could not control.423 CENRAP's analysis showed
that 52% of impairment at Big Bend and 25% of impairment at Guadalupe Mountains was due to
emissions from Mexico and further south.424
According to Luminant, in its proposal here, EPA agrees with Texas that the RPGs established
by Texas provide for "improvement in visibility for the most impaired days over the period of
the SIP and ensure no degradation in visibility for the least impaired days over the same
period."425 And EPA further agrees with Texas that RPGs based on the uniform rate of progress
are not achievable.426 EPA further agrees that international sources of emissions are the
reason.427 EPA's only point of departure from Texas' reasonable progress analysis is that the
2018 RPGs should be lowered by a miniscule (and undetectable) amount—from 16.6 to 16.57
deciviews for Big Bend and from 16.3 to 16.26 deciviews for Guadalupe Mountains.428
Luminant stated, in other words, the basis upon which EPA would second-guess Texas'
reasonable progress analysis—and for which it would impose billions of dollars in emission
control measures on a handful of select Texas sources—is stated in the hundredths of a deciview:
0.03 deciview improvement at Big Bend and 0.04 deciview improvement at Guadalupe
Mountains. And in terms of achieving the national goal of natural visibility conditions, EPA's
proposal would have no meaningful benefit, under even EPA's calculations. EPA projects that
its RPGs would achieve natural visibility at Big Bend in the year 2198 (compared to 2206 with
no further controls); at Guadalupe Mountains in the year 2163 (compared to 2169 with no
controls); and at Wichita Mountains in the year 2095 (compared to 2101 with no controls).429
Luminant asserted that the EPA has no authority to disapprove Texas' reasonable progress
analysis on this basis. Texas has not failed to consider any of the statutory factors, nor has EPA
identified any other requirement of the statute that Texas' SIP revision does not meet. Moreover,
as a matter of common sense, this difference is so negligible that it cannot support or justify
EPA's proposed disapproval or the costs EPA would impose through the FIP. In fact, the
additional "benefits" EPA projects are literally eliminated in the rounding. Under EPA's own
rounding convention, which rounds deciviews to the nearest tenth of a deciview for purpose of
reasonable progress goals,430 the RPGs that Texas established are exactly the same as the RPGs
that EPA now concludes are reasonable. EPA's disregard of its own rounding convention here is
unexplained and arbitrary and capricious.
Luminant stated that EPA's proposal to micro-engineer the outcome of Texas' reasonable
progress analysis is contrary to the statute and regulations and wholly unreasonable. EPA's
proposal rests on the incorrect assumption that EPA has the authority to review the substantive
outcome of a state's RPGs—down to the hundredth of a deciview. But the statute and
845
-------
regulations impose only an analytical requirement for states, which Texas has fully met. The
CAA and EPA's regulations do not establish required emission limitations or visibility
improvements, but instead impose only an "analytical" requirement to consider the four statutory
factors.431 Thus, as long as the state performs the required analysis (as Texas did here), EPA
must respect the state's choices and approve its reasonable progress goals—even where EPA
would have come to a different conclusion than the state in considering the factors.432 And even
though EPA believes that more reductions from Texas sources are "feasible,"433 such a standard
was specifically rejected by Congress in enacting the regional haze program and is not a lawful
basis for EPA's disapproval.434 EPA's proposal thus applied the wrong legal standard by which
to judge Texas' SIP revision.
Further, Luminant stated that there was no error in Texas' analysis. Texas' analysis followed the
regulations and EPA's own guidance in considering potential additional controls and assessing
the four statutory factors. And, as discussed in Section X and elsewhere in our comments, Texas
followed the same approach as other states, whose SIPs EPA approved without reservation 435
EPA finds no error in the data considered by Texas, nor does EPA contend that Texas failed to
conduct the four-factor analysis. Indeed, Texas relied on CENRAP modeling and control
strategy analysis with which EPA concurs.436 Additionally, Texas estimated and considered the
visibility improvements at multiple Class I areas (including the three at issue here) from the
potential controls it analyzed, noting that at each area the improvement was less than 0.5
deciviews and thus imperceptible.437 Based on this information—which is basically the same
information as EPA develops on its own and claims requires a different outcome—Texas
appropriately concluded that the additional controls were not reasonable.438
Footnotes:
422 2009 Texas SIP Narrative at 10-3.
423 Id. at 10-10.
424 Id.
425 79 Fed. Reg. at 74,834.
426 Id. at 74,887 ("We propose to find that it is not reasonable to provide for rates of progress at the Wichita
Mountains, Big Bend, or Guadalupe Mountains that would attain natural visibility conditions by 2064 (i.e., the
URP).").
427 Id. at 74,843.
428 Id. at 74,887.
429 See TX116-007-_33_Vis_modeling_summary ("2018 RPG calcs").
430 BART Alternative TSD at 24 n.24 ("Calculating visibility changes to the nearest tenth of a deciview (rather than
the nearest hundredth) is consistent with the practice for implementing the reasonable progress goals under the
Regional Haze rule.") (citing EPA Guidance for Tracking Progress).
431 40 C.F.R. § 51.308(d)(l)(i).
432 North Dakota, 730 F.3d at 768 ("[T]he CAA requires only that a state establish reasonable progress, not the most
reasonable progress."); 79 Fed. Reg. at 5205 (approving Wyoming's decision not to impose additional controls on
oil and gas source category because EPA concluded that the costs "were not so low that EPA could find it
necessarily unreasonable for the State to not have adopted them").
433 Technical Support Document for the Oklahoma Regional Haze State Implementation Plan and Federal
Implementation Plan (OK TSD) 29 ("Oklahoma TSD") (Nov. 2014).
434 In the conference committee that reconciled the House and Senate versions of the 1977 Clean Air Act
amendments, the term "maximum feasible progress" was specifically changed to "reasonable progress" in the final
legislation. See 1 Legislative History of the Clean Air Act Amendments 1977 Pub. L. No. 95-95 155 (1977) ("The
term 'maximum feasible progress' is changed to read 'reasonable progress' whenever it appears in the section.").
435 For example, EPA approved, without question, Washington's decision that the URP was not reasonably
achievable and that additional controls were not necessary during the first planning period, based on the same
846
-------
rationale Texas used in its SIP. 77 Fed. Reg. 76,174 (Dec. 26, 2012). As EPA explained: "Additional controls on
point sources or other source categories at this time is not likely to result in substantial visibility improvement in the
first planning period due to the significant contribution from emissions from natural fire, the Pacific offshore,
Canada, and outside the modeling domain." Id. at 76,204.
436 TX SIP TSD at 55 ("The CENRAP states' modeling, described in Section 8 of the Texas Regional Haze SIP, was
developed consistent with our guidance."); CENRAP Modeling TSD at 30 ("EPA concurred with the selection of
CAMx for the CENRAP regional haze modeling as it has been extensively used within the region and has been
proven to be an acceptable model.").
437 2009 Texas SIP Narrative at 10-6, tbl.10-6.
438 Id. at 10-7 ("At a total estimated cost exceeding $300 million and no perceptible visibility benefit, Texas has
determined that it is not reasonable to implement additional controls at this time.").
Response: We disagree with this comment on all issues raised. Contrary to Luminant's belief,
and as discussed elsewhere, our review of SIPs is not limited to a ministerial type of automatic
approval of a state's decisions. We must consider not only whether Texas considered the
appropriate factors but acted reasonably in doing so. In undertaking such a review, we do not
"usurp" the state's authority but ensure that such authority is reasonably exercised. As stated in
more detail elsewhere, Texas did not fully satisfy the requirements under Section 51.308(d)(1)
related to the evaluation of the four reasonable progress factors and establishment of the RPGs.
In particular, as we discussed in detail in our proposal and final, we disagree with the set of
potential controls identified by Texas and how it analyzed and weighed the four reasonable
progress factors under Section 51.308(d)(l)(i)(A). As stated previously, we disagreed with the
goals set by Texas and how the state analyzed and weighed the four reasonable progress factors.
As we stated in our proposal and within our responses to comments and final action, Texas did
not satisfy several of the requirements at Section 51.308(d)(1) with regard to setting RPGs, most
notably the requirement to reasonably consider the four statutory reasonable progress factors and
the requirement to adequately consider the emission reduction measures needed to meet the
URP. As we state in our proposal and are finalizing today:673
Because it only estimated the visibility benefit of all the controls together, the
TCEQ was not able to assess the potential benefit of controlling individual sources
with significant, and potentially cost-effective, visibility benefits. Also, we believe
that individual benefits were masked by the inclusion of those controls with little
visibility benefit that only served to increase the total cost figures.
We are finalizing our determination that Texas' analysis was deficient and not approvable because
the large control set they selected was not appropriately refined, targeted or focused on those
sources having significant and potentially cost-effective visibility benefits. Consistent with our
proposal, we conclude that control set was over-inclusive. It included controls on sources that
would increase total cost figures with little visibility benefit.
We, Texas and CENRAP all acknowledge that emissions from Mexico contribute to visibility
impairment at Texas' Class I Areas. We are not requiring Texas to over-control its sources in
order to make up the difference. We reject Luminant's conclusion that because these Class I
Areas are partially impacted by emissions from Mexico that no efforts from Texas to further
673 79 FR 74838.
847
-------
improve their visibility impacts by controlling Texas' share of the problem are warranted. A
reasoned four-factor analysis was still required.
Luminant claims that we agreed with Texas that the RPGs established by Texas provide for
"improvement in visibility for the most impaired days over the period of the SIP and ensure no
degradation in visibility for the least impaired days over the same period." However Luminant's
partial reproduction of our proposal misrepresents what we said. The full quote is reproduced
below: (footnote 674)
Although Texas' RPGs do provide for some improvement in visibility for the
most impaired days over the period of the SIP and ensure no degradation in
visibility for the least impaired days over the same period, we believe the overall
RPG goals that Texas established for its own Class I areas of Big Bend and
Guadalupe Mountains do not provide for reasonable progress based on the
four reasonable progress factors that a state is required to consider in selecting
a RPG under (d)(l)(i)(A). [emphasis added\.
Luminant holds up our conclusion that the rates of progress at the Wichita Mountains, Big Bend,
or Guadalupe Mountains that would attain natural visibility conditions by 2064 are not
reasonable as if that conclusion is somehow indicative of the approvability of Texas' regional
haze SIP. It is not. Our proposed acknowledgement of that fact has no bearing on our proposed
controls. Again, Luminant seems to believe that because perfect progress cannot be made, no
progress should be made. We reject this misplaced conclusion. Texas must meet the RP
statutory requirements in establishing their RPGs, and where we determine that these
requirements have not been satisfied, we have the authority to disapprove the State's RPGs and
indeed must disapprove it as not meeting the Federal requirements. Our disapproval is based on
the fact that Texas did not properly evaluate the four statutory factors in establishing its RPGs.
As further explained elsewhere in today's action, Congress directed in section 110 of the CAA
that states would take the lead in developing implementation plans, but balanced that decision by
requiring us to review the plans to determine whether a SIP meets the requirements of the CAA.
We must consider not only whether the State considered the appropriate factors in development
of its RPGs, but also whether the State acted reasonably in doing so. For the reasons explained
in this document and the preamble to this final rule, Texas failed to do so.
Luminant states we found no error in the data considered by Texas, nor we contend that Texas
failed to conduct the four-factor analysis. As we state in our proposal and in numerous responses
herein, we in fact found many errors in Texas' regional haze SIP. Luminant erroneously extends
our finding that CENRAP's modeling was acceptable to all aspects of the Texas Regional Haze
SIP. Also as we indicate many times in our response to comments, merely addressing our
regional haze requirements in some form or fashion does not equate with doing so properly or
reasonably.
We address Luminant's allegations of inconsistencies with our other actions in the consistency
section of this document. We disagree with comments that we applied the wrong legal standard
in evaluating the state's reasonable progress determinations because the state was not free to do
674 79 FR 74834
848
-------
make such determinations that were inconsistent with the CAA. Thus, while states have
discretion in establishing reasonable progress goals it must be reasonably exercised. Texas'
approach to reasonable progress was flawed and we properly rejected it. Please see our
responses in other more detailed comments regarding our authority to review SIPs and regarding
our authority to disapprove TX's reasonable progress goals, and consistency that includes and
explanation of Wyoming and Washington. We address comments concerning visibility benefits
of controls and the commenter's assertion that the benefits round to zero in a separate response to
comment where we address comments on cost versus visibility benefit.
Comment: [Associations (0059) p. 9-10] The Associations stated that Texas fully complied
with all statutory and regulatory requirements in developing its own reasonable progress goals
for the Big Bend and Guadalupe Mountain Class I areas. Specifically, Texas appropriately
considered the four statutory factors in 42 U.S.C. § 7479(g)(1) and established reasonable
progress goals after determining that the uniform rate of progress was not reasonably achievable
by 2018. EPA does not dispute that Texas' evaluation complied with the statutory and
regulatory requirements. EPA agrees that Texas' reasonable progress goals provide for
"improvement in visibility for the most impaired days during the period of the SIP and ensure no
degradation in visibility for the least impaired days over the same period" in accordance with 40
C.F.R. § 51.3108(d)(1). 79 Fed. Reg. at 74,834. EPA also agrees with Texas that the
reasonable progress goals based on the uniform rate of progress are not achievable due to
international emissions. Id. at 74,843, 87. Thus, because Texas has complied with these criteria,
EPA has a mandatory duty to approve the SIP pursuant to 42 U.S.C. § 7410(k)(3). See National
Ass'n of Home Builders, 551 U.S. 644 (EPA has no discretion when statutory language says the
agency "shall approve" a State program if statutory criteria are met). Nevertheless, EPA
arbitrarily proposes to second-guess Texas' 2018 reasonable progress goals by replacing them
with goals that would improve projected visibility at the Big Bend and Guadalupe Mountain
Class I areas by an imperceptible few hundredths of a deciview, but require enormous
expenditures by a handful of stationary sources.
The Associations stated that EPA's proposal is contrary to the cooperative federalism principles
on which the regional haze program is based. Under those principles, EPA must respect the
State's choices and approve its reasonable progress goals as long as the State performed the
required analysis. EPA does not claim that Texas failed to perform the required analysis.
Instead, it faults "how [Texas] analyzed and weighed the four reasonable progress factors." EPA,
Texas Technical Support Document 18. But, under the Clean Air Act, EPA must approve a
State's reasonable progress goals, even if EPA would have weighed the statutory factors
differently and reached a different result. See, North Dakota, 730 F.3d at 768 ("[T]he CAA
requires only that a state establish reasonable progress, not the most reasonable progress."); see
also Alaska Department of Environmental Conservation v. EPA, 540 U.S. 461, 490 (2004)
(recognizing that, when EPA reviews the reasonableness of state best available control
technology determinations, it must act with deference and cannot "second guess state
decisions"). Thus, it is unlawful and contrary to Congress' intent for EPA to disapprove of
Texas' reasonable progress goals simply because EPA, after second-guessing Texas' analysis,
concluded that additional reductions from a handful of additional sources are feasible. See also
Section II, infra.
849
-------
Response: We disagree with the Associations that "EPA does not dispute that Texas' evaluation
complied with the statutory and regulatory requirements." As we discuss in our responses to
more detailed comments herein, we disagree with the Associations that Texas fully complied
with all applicable requirements in developing its reasonable progress goals for Big Bend and
Guadalupe Mountains. As we note in our response to Luminant above, the Associations'
selection of discrete statements concerning the impact of international emissions, and Texas'
reasonable progress goals providing for "improvement in visibility for the most impaired days
during the period of the SIP and ensure no degradation in visibility for the least impaired days
over the same period," are taken out of context or misrepresent the actual text in our proposal.
Thus, Texas has not in fact complied with the applicable criteria and its regional haze SIP must
be disapproved. We further reject the Associations' characterization of our review as "second
guessing" Texas' reasonable progress goals. Our review was conducted according to the
authority granted to us by Congress, and as reflected in our Regional Haze Rule. We did no
second guessing, but rather conducted a thorough review and properly rejected Texas' flawed
analyses.
We do not agree that our proposed disapproval of the Texas RPG's is contrary to the CAA, the
RHR, or relevant case law. As detailed in our responses elsewhere, Congress crafted the CAA to
provide for states to take the lead in developing implementation plans, but balanced that decision
by requiring us to review the plans to determine whether a SIP meets the requirements of the
CAA. Our review of SIPs is not limited to a ministerial type of automatic approval of a state's
decisions. We must consider not only whether the State considered the appropriate factors but
acted reasonably in doing so. In undertaking such a review, we do not "usurp" the state's
authority but ensure that such authority is reasonably exercised. As discussed elsewhere in state
and federal roles under the Regional Haze program, the court in North Dakota held that the CAA
and the States operate under a framework of "cooperative federalism." Under this framework, the
court stated that the CAA left the individual states to make pollution restriction for particular
emitters within that state. "But, if a state fails to submit a SIP, submits an incomplete SIP, or
submits a SIP that does not meet the statutory requirements, we are obligated to implement our
own FIP to correct the deficiency in the SIP, unless the State can correct the deficiency itself and
we can approve that correction within two years. 42 U.S.C. 7410(c). This is commonly referred
to as cooperative federalism, and both Section 169A and Section 110 operate under this
framework." Thus, the CAA provides us with a critical oversight role in ensuring that SIPs
meet the CAA's requirements. As discussed elsewhere, the states' analyses and reasonable
progress determinations were flawed.
Comment: [AEP (0055) p. 2] AEP stated that Texas and Oklahoma submitted provisions that
satisfied the requirements of the Clean Air Act, the regional haze regulations and the EPA's
guidance for tracking reasonable progress. The requirements and guidance provide a process and
framework that directs states to develop a plan that demonstrates reasonable progress toward the
national goal. As noted in these requirements, EPA's role is to defer to the states in the plan
development and goal setting.
850
-------
Response: We disagree with the comment. Please see our response to Luminant above
concerning our SIP review authority as well as our responses to other, more detailed comments.
Comment: Texas' reasonable progress analysis and associated SIP submission complies
with all CAA requirements and must be approved. [GCLC (0063) p. 5]
GCLC stated that Texas' reasonable progress analysis complies with all statutory and regulatory
requirements. EPA's proposal to disapprove Texas' SIP submission regarding reasonable
progress is arbitrary, capricious, and not supported by the facts or law. As EPA itself has
admitted, EPA's "individual source" and "visibility benefit" approach for Texas is "without...
prior precedent." 18 Rather than deferring to Texas' reasoned judgment and its extensively
supported SIP, EPA has inappropriately decided to deny Texas' reasonable progress analysis
based on "how [Texas] analyzed and weighed the four reasonable progress factors," 19 and more
specifically, how Texas considered potential visibility benefits of the considered control
strategies. 20
Footnotes:
18 Declaration of Sam Coleman, Nat'! Parks Conservation Ass'nv. McCarthy, No. 11-01548, at 5 (D. D.C. 2014).
19 TX TSD at 18. (emphasis added).
20 Proposed FIP, 79 Fed. Reg. at 74838.
Response: We disagree with GCLC that Texas' reasonable progress analysis complies with all
statutory and regulatory requirements, as is detailed in our proposal, final action and in numerous
responses to comments contained herein. Similarly, please see our responses to visibility benefit,
Texas' four-factor reasonable progress analysis, and Texas' control strategy for more detailed
responses. We have addressed our review authority in more specificity elsewhere. Lastly, as
explained elsewhere, with regards to Sam Coleman's declaration - while we did say that there
was no prior precedent, this was in regards to the particular type of modeling undertaken. Sam
Coleman's statement is related to the additional modeling we determined was appropriate due to
the large distances involved and the large number of sources being analyzed. The cited portion
of this Declaration is thus taken out of context and does not represent the premise for which it
was cited.
Comment: EPA's Disapproval of Texas' Reasonable Progress Control Analysis Is
Proper. [Earthjustice (0067) p.21]
Earthjustice et al., stated that the cornerstone of the regional haze program is the requirement that
haze plans must make reasonable progress toward restoring natural visibility conditions at
national parks, wilderness areas, and other Class I areas. 42 U.S.C. § 7491(a)(1), (b)(2); 40
C.F.R. § 51.308(d)(l)(i)(B); see also Util. Air Regulatory Group, 471 F.3d at 1340 (agreeing
with EPA that reasonable progress is the ultimate measure of whether a haze plan complies with
"the regulatory scheme as a whole"); Ctr. for Energy & Econ. Dev. v. EPA, 398 F.3d 653, 660
(D.C. Cir. 2005) (the haze program's primary goal is to achieve reasonable progress toward
eliminating human-caused visibility impairment). A state is required to adopt "enforceable
emissions limitations, compliance schedules, and other measures as necessary to achieve the
851
-------
reasonable progress goals established by States having mandatory Class I Federal areas." 40
C.F.R. §51.308(d)(3). Moreover, because Texas causes or contributes to impairment in other
states' mandatory Class I areas, Texas "must demonstrate that it has included in its
implementation plan all measures necessary to obtain its share of the emission reductions needed
to meet the progress goal for the area." Id. § 51.308(d)(3)(ii).
In comments made at public hearings on the proposed rule, Texas and industry argued that Texas
is being treated differently than other states because EPA is imposing a novel requirement to
assess reasonable progress on a source-by-source basis. But EPA is using the same basic
framework for reviewing Texas' SIP that is established by the Clean Air Act: whether the plan
meets all applicable legal requirements. 42 U.S.C. § 7410(c), (k), (1). EPA's statutory
responsibility is to review each haze plan to ensure that it is "reasonably moored" to the Clean
Air Act and that the plan is based on a reasoned analysis. See Oklahoma v. EPA, 723 F.3d at
1206-12; North Dakota v. EPA, 730 F.3d at 761 (quoting from Alaska Dep't of Envtl.
Conservation v. EPA, 540 U.S. 461, 485 (2004)). This is precisely the approach that EPA has
taken in its review of the Texas plan
Earthjustice et al., stated that Texas and industry are inventing a straw man by claiming that EPA
is, for the first time, requiring that reasonable progress be analyzed on a source-by-source basis.
EPA has said no such thing. Indeed, EPA has not performed or required source-by-source
analysis of the vast majority of Texas' 1600+ point sources, let alone all of its individual area or
mobile source contributors. Instead, EPA has said that the analysis that Texas undertook failed
to comply with the statutory and regulatory requirements. EPA's conclusion that Texas' analysis
of the aggregate costs and benefits of reasonable progress controls was not approvable is not the
same thing as concluding that Texas was required to analyze each reasonable progress source
individually.
Earthjustice et al., stated that the EPA is properly proposing to disapprove Texas' reasonable
progress control analysis for at least two reasons. First, the aggregate visibility benefits of the
controls Texas considered are significant enough that it was unreasonable for Texas to dismiss
the controls out of hand. 30 Second, by considering costs only in the aggregate, and failing to
take a more fine-grained look at the costs of controls, Texas unreasonably rejected controls that
met the state's own cost-effectiveness threshold of $2,700 per ton.
Footnote 30: As discussed previously, visibility is not a fifth factor in reasonable progress analyses. Accordingly visibility
benefits may not be used to screen out reasonable progress controls. We consider Texas' evaluation of the "significance" of
visibility benefits afforded by its evaluated controls, and its weighing of costs versus benefits, only because this is the framework
of evaluation used by EPA.
Response: We take no position on the background statements or certain characterizations of our
action that are made by Earthjustice. For specific responses regarding comments on our
disapproval of reasonable progress goals, single source analysis, our review of Texas' Cost-
effectiveness methodology, and visibility benefit analysis, please see our responses to more
detailed comments concerning those issues.
852
-------
Comment: EPA does not provide any lawful basis for disapproving the RPGs for Big Bend
and Guadalupe or the Texas LTS. [UARG (0065) p. 15-21]
UARG stated that the hallmark of the CAA's regional haze program is state primacy in all
substantive decision-making, including in particular the evaluation of reasonable progress and
the establishment of RPGs. Indeed, EPA's "Guidance for Setting Reasonable Progress Goals
Under the Regional Haze Program" (June 1, 2007), Doc. ID No. EPA-R06-OAR-2014-0754-
0012, TX166-012-10 ("Reasonable Progress Guidance"), provides that the regional haze rule
"gives States wide latitude" to determine which measures to require pursuant to the Act's
reasonable progress provisions and provides that states "have considerable flexibility" in
deciding how to take the reasonable progress factors into consideration, id. at 4-2; see also id. at
5-1 ("In determining reasonable progress, CAA §169A(g)(l) requires States to take into
consideration a number of factors. However, you [i.e., states] have flexibility in how to take into
consideration these statutory factors and any other factors that you have determined to be
relevant.") (emphasis added); see also Am. Corn Growers Ass'n v. EPA, 291 F.3d 1, 5-9 (D.C.
Cir. 2002) (per curiam) (emphasizing state primacy). EPA nevertheless proposes to disregard
the deference that is due to Texas' reasonable progress analysis, its RPG determinations, and its
LTS. EPA proposes to disapprove Texas' RPGs for Big Bend and Guadalupe because "the state
has not demonstrated that its RPGs provide for reasonable progress towards meeting the national
visibility goal" on the grounds that, EPA contends, Texas failed "most notably ... to reasonably
consider the four statutory reasonable progress factors and the requirement to adequately justify
RPGs that are less stringent than the URP." 79 Fed. Reg. at 74,822. In connection with its
proposed disapproval of these RPGs, EPA also proposes disapproval of key elements of Texas'
LTS. Apart from EPA's rationale for disapproving Texas' SIP related to Wichita Mountains,
addressed above in Section III of these comments, EPA states it is disapproving the Texas LTS
because, in its view, Texas did not adequately consider the emissions limitations and schedules
for compliance needed for reasonable progress in Big Bend and Guadalupe. Id.
UARG stated that Texas' RPGs and the state's rationale for its determinations fully comport with
the CAA, the regional haze rule, and EPA's guidance. Indeed, EPA's statements in the proposed
rule illustrate that Texas did everything that was required of it under the regional haze rule. Id.
At 74,834 ("Based on the emission reductions from these measures, CENRAP modeled the
projected visibility conditions anticipated at each Class I area in 2018 and the TCEQ used these
results to establish its RPGs. The TCEQ states it developed its RPGs after considering the
regulatory factors required under Section 51,308(d)(l)(i)(A) ...."); id. at 74,835 ("TCEQ
developed a list of potential controls and costs associated with those controls to inform their
four-factor analysis. It used the control strategy analysis developed by CENRAP as the starting
point for its analysis."). EPA's discussion of Texas' evaluation of the reasonable progress
factors also demonstrates that the state conducted a lawful and appropriate reasonable progress
analysis. Id. at 74,837-38. Texas evaluated the costs of compliance and, in doing so, set cost
thresholds that it determined reasonably differentiated between reasonable and unreasonable
levels of expense. Id. at 74,837. It considered time necessary for compliance and remaining
useful life of affected sources and found that these factors were likely not determinative. Id. It
qualitatively assessed energy and non-air-quality environmental impacts, concluding that this
factor generally weighed against additional controls, but did not rest its decision on its
assessment of that factor. Id. Further, Texas identified the suite of emission controls that could
853
-------
be deployed at various Texas sources and conducted modeling to evaluate the degree of visibility
improvement that could be achieved at affected Class I areas. 79 Fed. Reg. at 74,837.
Reasonably - and certainly within the broad bounds of state discretion in this area - Texas
determined that the costs of additional controls were not warranted in part due to the minimal
degree of visibility improvement that those controls could achieve. Id. All of these
determinations comport with the CAA, the regional haze rule, and EPA's Reasonable Progress
Guidance.
UARG noted that the EPA rests its proposed disapproval on its disagreement with Texas'
determinations and on EPA's policy preferences for different choices. For instance, EPA states
the Agency "believe[s] the overall RPG goals [sic] that Texas established for its own Class I
areas of Big Bend and Guadalupe Mountains do not provide for reasonable progress based on the
four reasonable progress factors that a state is required to consider." Id. at 74,834. Similarly,
EPA states that it "disagree[s] with the set of potential controls identified by the TCEQ and how
it analyzed and weighed the four reasonable progress factors in a number of key areas." Id. At
74,838. These sort of "belie[fs]" and "disagree[ments]" are not lawful grounds for disapproving
a regional haze SIP.
UARG contended that the allegations of more specific purported flaws that EPA identifies do not
withstand scrutiny. For example, EPA states that, "[i]n general, the cost of compliance was the
key factor considered by the TCEQ," suggesting that Texas gave improper weight to this factor.
Id. at 74,837. The rest of EPA's discussion, however, shows that Texas gave appropriate weight
to each factor and, as noted above, that the state also considered the degree of visibility
improvement that would result from additional reasonable progress measures. Id. at 74,838
("The TCEQ's assessment of reasonable progress rested primarily on its calculation of the total
cost of the controls it analyzed versus the visibility benefits at the ten Class I areas it analyzed.").
UARG noted that the EPA also states that "because the TCEQ did not evaluate controls on a
source-by-source basis, source-specific factors related to the evaluation of the reasonable
progress four-factor analysis could not be considered," and for that reason, EPA proposes to
determine that Texas' analysis is "insufficient." Id. at 74,838, 74,841. States are not, however,
required to conduct reasonable progress assessments on a source-by-source basis. This is
confirmed by EPA itself in the Reasonable Progress Guidance, which states that the cost factor
can be evaluated "for individual sources or source categories." Reasonable Progress Guidance at
5-1 (emphasis added). State discretion to conduct a reasonable progress assessment without
analyzing controls on a source-by-source basis has been confirmed by the U.S. Court of Appeals
for the Tenth Circuit. WildEarth Guardians v. EPA, 770 F.3d 919, 944 (10th Cir. 2014)
("Neither the Clean Air Act nor the Regional Haze Rule requires source-specific analysis in the
determination of reasonable progress."). EPA fails to acknowledge this established law and cites
no support for its conclusion.
UARG stated that after conducting its own source-by-source analysis, all that EPA is able to
reach is a vague, open-ended conclusion:
These results also suggest that controlling a small number of sources will result in
visibility benefits at both Class I areas, and that rather than evaluating controls at all
854
-------
facilities identified by Texas combined, a subset of those facilities (and some additional
facilities not identified) may be reasonable. 79 Fed. Reg. at 74,839 (emphases added).
Despite the fact that this sentence appears to be incomplete, what is clear from it is that the most
EPA can even assert is that different regulatory determinations than the ones that Texas made in
its SIP might have been "reasonable" policy choices. Plainly, that is no basis for disapproving a
SIP.
UARG stated that the EPA also suggests that it must disapprove Texas' SIP because EPA
"believe[s] that in performing its control analysis, the TCEQ should have given greater
consideration to the flexibility in the CAIR trading program and the resulting uncertainty in the
projected emissions." Id. at 74,840. To explain its point, EPA highlights the difference in Texas'
Integrated Planning Model-based estimates of emission reductions due to CAIR at Big Brown
Unit 1 versus Big Brown Unit l's actual emissions in 2012. Id. EPA cannot fault Texas for
being unable to predict the future. The modeling that Texas relied on was the best information
available at the time it submitted the SIP, which it did long before 2012, and EPA cannot validly
judge a SIP unapprovable on the basis of information that became available for EPA's review
only after the SIP was submitted and only because EPA has missed its statutory deadline for
action on the submitted Texas SIP by nearly five years.
UARG noted that the EPA also takes issue with the cost threshold Texas used to exclude certain
controls from consideration. Texas chose $2,700 per ton, while EPA prefers $4,000 per ton to
$5,000 per ton. Id. at 74,838. This preference is not a reason for disapproving Texas' SIP.
Indeed, the only purported justification that EPA can muster - that reliance on the $2,700-per-ton
cost threshold EPA used in CAIR is inappropriate because "[a] state should look beyond BART
for additional reductions when developing its long-term strategy to achieve reasonable progress
at its Class I areas" - conflicts with EPA's own guidance indicating that BART may reasonably
be concluded to satisfy reasonable progress requirements in the first planning period. Id:,
Reasonable Progress Guidance at 4-2 to 4-3. EPA cites - and can cite - no basis in the CAA or
in the regional haze rule for disapproving Texas' cost threshold.4
Furthermore, EPA states that it proposes to disapprove Texas' URP analysis because "we do not
believe that the rate of improvement the TECQ [sic] has selected is reasonable, because we
disagree with its four-factor analysis and the analysis of emission measures needed to meet the
URP." 79 Fed. Reg. at 74,843. This is not a lawful basis for a SIP disapproval. Texas has
complied with the regional haze rule's requirements with respect to a URP analysis, and, for the
reasons stated above, EPA has not provided a valid justification for disapproving Texas'
reasonable progress analysis.
UARG stated that the EPA proposes to disapprove Texas' calculation of the URP as a result of
its proposal to disapprove the state's calculation of natural visibility conditions at Big Bend and
Guadalupe. Id. at 74,822. The primary reason for this proposed disapproval is Texas'
assumption that fine soil and coarse mass concentrations, and their resulting light extinction
effects, is entirely attributable to natural sources. Id. at 74,831. Texas provided a reasoned
justification for its decision in this regard, explaining that "to the extent its assumption that 100%
of coarse mass and fine soil is natural is an overestimate, it expects that its low organic carbon
855
-------
estimate will more than compensate for any errors in this assumption at this time." Id. Despite
the reasonableness of this finding, Texas, at the FLMs' request, performed a supplemental
analysis assuming that only 80 percent of this particulate matter was from natural sources. Id. In
the final analysis, the FLMs recognized that EPA's regulations at "40 CFR 51.308 give the State
[the] right" "to recalculate natural conditions for Big Bend [National Park] and Guadalupe
Mountains [National Park]" and agreed that "the basic approach used [by Texas] to adjust natural
conditions is reasonable, provided that the Proposed SIP address the uncertainty of the
assumption that all of the coarse mass and fine soil fraction on the worst 20 percent days is
natural."5 Texas' final SIP submission provides the basis for the assumption, as the FLMs
requested.6 In contrast, EPA, without any meaningful analysis, would impose use of default
values for this particulate matter provided for in the IMPROVE equation and proposes
disapproval of the Texas SIP on that basis. Id. at 74,832. EPA provides no basis for rejecting
Texas' reasoned explanation. Particularly in light of the FLM comments and Texas' rationale
for using 100 percent, EPA has a duty to fully explain and to provide an adequate rationale for
selecting the default values over Texas' determination. EPA failed to discharge that duty, and its
proposed action is improper for that reason as well.
Footnotes:
4Moreover, although the point is not directly relevant to this proceeding, EPA's statement in the preamble to the
proposed rule that "the URP does not establish a 'safe harbor' for the state in setting its progress goals," 79 Fed.
Reg. at 74,834, is wrong and should be corrected by EPA. The language in the preamble to the 1999 regional haze
rule that EPA quotes confirms that there is, in fact, such a safe harbor. EPA there made clear that if a state
determines that the progress the URP would require is reasonable, then "the State should identify this amount of
progress as its reasonable progress goal for the first long-term strategy." 64 Fed. Reg. 35,714, 35,732 (July 1, 1999)
(emphasis added). Only //the state instead "determines that additional progress is reasonable" should the state adopt
a more accelerated or ambitious RPG than the URP. Id. As this preamble language makes clear, the state has no
obligation to undertake additional analysis if it determines the URP is reasonable for its Class I area. The decision
whether to undertake such additional analysis is a matter reserved for the state's exercise of discretion.
5 U.S. Fish & Wildlife Serv. & Nat'l Park Serv., Comments on Texas Proposed Regional Haze Rule State
Implementation Plan at 2, 3 (Jan. 11, 2008), Doc. ID No. EPA-R06-OAR-2014-0754-0002, TX166-002-03.
6 See generally TCEQ, Revisions To The State Implementation Plan (SIP) Concerning Regional Haze at Appendix.
5.2 (adopted Feb. 25, 2009), Doc. ID No. EPA-R06-OAR-2014-0754-0002, TX166-002-05 ("2009 Texas SIP").
Response: We disagree with this comment on all issues raised. We take no position on UARG's
presentation of background information. As explained in greater detail elsewhere, we disagree
that states have "primacy" or that they are owed "deference" to the point that it would supplant
our authority to review the Texas Regional Haze SIP for conformance with federal requirements.
States have flexibility under the Regional Haze Rule to approach our requirements in different
ways, but they must still satisfy those requirements, and our role is to determine that they have
done so. We disagree that Texas reasonably established its reasonable progress goals. We
further disagree that our proposed disapproval was based on "beliefs" and "disagreements," but
in fact is founded on our own analysis that showed that a small number of Texas sources
significantly impacted the visibility at the Wichita Mountains, Big Bend, and the Guadalupe
Mountains; and that cost-effective controls are available to reduce those impacts. As stated
elsewhere, while states have discretion in establishing reasonable progress goals such discretion
must be reasonably exercised. We found that Texas analysis was flawed. As we stated in our
proposal, final and elsewhere in our responses to comments herein, we believe that Texas' four-
factor analysis for reasonable progress was flawed on several grounds and did meet federal
requirements.
856
-------
We disagree with UARG that our cost analysis indicates that Texas gave improper weight to cost
as a factor. We conducted our control cost analysis (which included a control efficiency
analysis) because we concluded that what Texas provided in its SIP was inadequate. We also
disagree that Texas properly considered visibility, principally because it obscured the visibility
impacts of individual sources (for which it partially conducted control efficiency and cost
analysis). See our other responses to comments concerning source category versus individual
source analysis for more information.
We disagree with UARG that after conducting our control cost analysis, our conclusion was
open-ended or vague. First, what UARG attributes to our analysis is not described via an
incomplete sentence and is actually our discussion of the CENRAP PSAT modeling and our
initial facility level modeling of sources identified through the Q/d analysis. Our point was that
in considering the results of the CENRAP PSAT analysis, and that was confirmed by our initial
modeling analysis, this should have suggested to Texas that controlling a smaller subset of
sources may be reasonable. The CENRAP PSAT modeling specifically highlighted the
importance of impacts from EGUs, particularly EGUs in the northeast portion of the states. Our
initial modeling confirmed this and further showed that on an individual basis, a small group of
sources were responsible for a large share of the visibility impacts from Texas point sources.
Instead of basing its analysis from that technical point, Texas proceeded with its own flawed
analysis. It was necessary for us to conduct additional modeling and analysis to determine if, in
fact cost-effective controls were available at these sources and whether such controls resulted in
significant visibility benefit improvement. Our own analysis was anything but open-ended or
vague, as we devoted two Technical Support Documents to it: our Cost TSD and our FIP TSD.
We refer UARG to those technical documents that are included in the docket to this action.
UARG takes issue with a statement in our proposal that the TCEQ should have given greater
consideration to the flexibility in the CAIR and the resulting uncertainty in the projected
emissions. Again, UARG misreads the text—we are not asking Texas to "predict the future."
As we explain in detail in the preceding two paragraphs, the TCEQ based its emission reductions
on IPM predictive modeling. In so doing, it assumed Big Brown's 2018 emission reduction of
approximately 1/3 (due to switching to a lower sulfur coal) would be an appropriate baseline for
use in a scrubber retrofit. We merely pointed out that were Big Brown to in fact install a SO2
scrubber, it could continue to burn its higher sulfur coal, thus greatly improving the scrubber's
Cost-effective ness ($/ton). We stated that the TCEQ's Cost-effective ness for Monticello was
similarly impacted. We summarized this situation by concluding that the TCEQ could [emphasis
added] have recognized that regional haze controls would likely not be in addition to the
anticipated reductions due to CAIR (the lower sulfur coal), but in place of them (retaining the
higher sulfur coal and installing a scrubber). Our use of "could" makes it clear this was not a
pivoting issue in our analysis of the Texas Regional Haze SIP. Texas' failure to properly
consider the visibility improvement and cost-effectiveness for scrubbers on Big Brown (and
other sources) was far more consequential.
We discuss UARG's allegation that we rejected Texas's $2,700/ton threshold in favor of a
$4,000 to $5,000 threshold in the section of this document where we respond to allegations of
inconsistency.
857
-------
UARG claims that our disapproval of Texas' URP analysis is not legal. As we explain in our
proposal:
However, for the reasons we have discussed above, although we agree with the
TCEQ that a rate of improvement necessary to attain natural visibility conditions
by 2064 is not reasonable, we do not believe that the rate of improvement the
TECQ has selected is reasonable, because we disagree with its four-factor
analysis and the analysis of emission measures needed to meet the URP.
Therefore we propose to disapprove the TCEQ's RPGs for Big Bend and the
Guadalupe Mountains under Section 51.308(d)(l)(ii). In so doing, we rely on the
specific directive in Section 51.308(d)(l)(iii): "In determining whether the State's
goal for visibility improvement provides for reasonable progress towards natural
visibility conditions, the Administrator will evaluate the demonstrations
developed by the State pursuant to paragraphs (d)(l)(i) and (d)(l)(ii) of this
section."
UARG provides no explanation other than its belief that Texas complied with the applicable
requirements and that that we did not provide a valid justification. As stated above, we disagree.
UARG also objects to our disapproval of Texas' URP analysis because we noted that the URP is
anchored by the natural conditions value in 2064, and because we proposed to disapprove Texas'
natural conditions estimate, we had to disapprove its URP calculations. UARG objects to our
proposed disapproval of Texas' natural conditions calculations and so believes Texas' URP
analysis should be approved. We disagree with UARG's assertion that we should have approved
Texas' natural conditions calculations and refer UARG to our responses to other more detailed
comments on that issue. As stated elsewhere in our response to comments, we also note other
reasons for disapproving Texas' URP analysis.
Please see our responses regarding our authority to review SIPs, our disapproval of Texas'
reasonable progress goals, our disapproval of Texas' URP, and our disapproval of Texas' long-
term strategy and consultation for more specific information.
Comment: TCEQ's four-factor analysis was reasonable and within its broad discretion,
and is supported by recent monitoring data showing RPGs will be met without additional
controls. [CCP (0075) p. 5-7]
CCP noted that the CAA and EPA regulations allow States to establish alternative RPGs to move
from "baseline" visibility conditions to the end goal of "natural conditions" by 2064. As detailed
in the SIP proposal and TCEQ's comments on the Proposed Rule incorporated here by reference,
TCEQ concluded that it would be unreasonable to meet the glidepath to meet the URP during the
2008-2018 planning period (the first planning period for regional haze programs) and that the
substantial reductions in haze pollutants that it anticipated via other regulatory programs would
be sufficient for this planning period. Therefore, Texas and Oklahoma established more
reasonable RPGs. In the Proposed Rule, EPA agrees that it is not reasonable to achieve the
URPs for Wichita Mountains, Big Bend and Guadalupe Mountains. Yet EPA proposes to
858
-------
disapprove of the State-established RPGs in favor of EPA-calculated RPGs that are purportedly
based on additional controls at Texas sources.
CCP stated that the EPA's regulations specify that "[f]or each mandatory Class I Federal area
located within the State, the State must establish goals (expressed in deciviews) that provide for
reasonable progress towards achieving natural visibility conditions." 40 C.F.R. § 51.308(d)(1)
(emphasis added). In other words, deciview goals should be set on an area-by-area basis, not on
an individual, source-specific basis. The United States Court of Appeals for the Tenth Circuit
recently confirmed that source-specific analysis is not required. See WildEarth Guardians v.
EPA, 770 F.3d 919, 944 (10th Cir. 2014) ("Neither the Clean Air Act nor the Regional Haze
Rule requires source-specific analysis in the determination of reasonable progress."). Moreover,
EPA's guidance supports TCEQ's approach. See "Guidance for Setting Reasonable Progress
Goals under the Regional Haze Program" at 5-1 (June 1, 2007) (explaining that each state has the
discretion to interpret the costs of compliance factor "to encompass the cost of compliance for . .
. source categories . . ."). That guidance also allows states to use EPA's BART guidelines which
provide states "flexibility in how they calculate costs" of compliance" and allow states to
"choose to apply the costs of compliance factor collectively for all of the sources in a source
category." See 70 Fed. Reg. 39,104, 29,127 (July 6, 2005). Other EPA guidance states that
"reasonable progress is not required to be demonstrated on a source-by source basis." See
"EPA's "Additional Regional Haze Questions" (Sept. 27, 2006 Revision). Accordingly, TCEQ
set RPGs based on a reasonable estimate of the visibility improvement needed to be reasonably
achieved in the Class I Federal area during the applicable planning period 2008-2018.
CCP stated that CAA Section 169A(g) requires the consideration of four-factors when setting
RPGs: (1) the cost of compliance, (2) the time for compliance, (3) the energy and non-air quality
impacts of compliance, and (4) the remaining useful life of any potentially affected sources. 42
U.S.C. §7491(g). In conducting the analysis, TCEQ properly focused its control strategy on
point source emissions of SO2 and NOx, as TCEQ determined that these are the main
anthropogenic pollutants that affect visibility at Class I areas in Texas and neighboring states.
TCEQ noted that it is already going beyond federal requirements by reducing NOx emissions
from mobile sources through the Texas Emissions Reduction Program, reinforcing the
appropriateness of focusing its control strategy on point sources. TCEQ then determined RPGs
reflecting visibility improvements from emissions reductions associated with the CAA, the Texas
Clean Air Act, Texas' ozone SIP revisions and rules, and agreements between EPA and oil
refineries for SO2 emissions reductions.
CCP stated that the TCEQ conducted a thorough analysis of the potential additional controls that
could be applied to help achieve even greater visibility at Class I areas affected by Texas
sources. See SIP Section 10 and Appendices 10-1, 10-2. In considering additional controls,
TCEQ worked with the Central Regional Air Planning Association ("CENRAP"). TCEQ used
the control strategy analysis prepared by CENRAP as the starting point for its analysis of
additional controls. The CENRAP analysis used the EPA AirControlNET tool to develop cost
per ton estimates for relevant pollutants. TCEQ reviewed this information and made changes
based on its knowledge of particular facilities—contrary to EPA's criticism that TCEQ did not
consider source-specific factors—and agency experience with implementing ozone control
strategies. TCEQ also added some individual sources that were not part of the EPA
859
-------
AirControlNET dataset that CENRAP relied on, in order to ensure that opportunities for cost-
effective visibility improvements were not overlooked. TCEQ used the CENRAP analysis to
determine an effectiveness ratio providing an estimate of improvement in visibility for every ton
of NOx and SO2 reduced and was able to develop an order-of-magnitude estimate of the likely
visibility improvements resulting from its point source control strategy. EPA's "Cost-effective
ness" analysis of controls, by contrast, is simply a statement of EPA's estimated cost per ton of
emissions reduced, without consideration of the visibility improvement that is likely to result.
CCP stated that the TCEQ considered aggregate visibility benefits from reductions in all sources
and determined that aggregate visibility benefits were not perceptible (1 dv is considered
perceptible to the human eye) and therefore not cost-effective. Specifically, TCEQ concluded
that substantial additional controls at an estimated annualized cost of $300 million were not
warranted for the . 16 dv benefit predicted by TCEQ at Big Bend, .22 dv benefit at Guadalupe
Mountains, and .36 dv benefit at Wichita Mountains. Reductions from a single source, e.g.
Coleto Creek Unit 1, could not therefore have a meaningful visibility benefit. Importantly,
visibility improvement is not one of the four statutory factors that a state is required to consider
when setting RPGs, and it should be given no more weight than TCEQ gave it here—as one
consideration in the cost of compliance determination. EPA instead considers visibility
improvement as the main factor in its analysis, to the exclusion of the four statutory factors. This
is inappropriate on the face of the statute, but becomes even more inappropriate when EPA bases
its proposed requirement on extra-statutory terms like "reasonable" and "significant" in
evaluating the visibility improvement that would result from the additional SO2 controls.
Response: We address elsewhere all the comments above, including the source specific
analyses comments, consideration of visibility, our cost-effectiveness analysis, perceptibility,
and the flaws in the Texas' technical analysis. In using the terms "reasonable" and "significant"
in evaluating the visibility improvement, please see our explanation in our final action under
general comments.
22. Deference
Comment: [Luminant (0061), p. i] Luminant stated that EPA's legal analysis ignores the
criteria that Congress included in the statute and relies instead on criteria that Congress did not
include. Ignoring its own regulations, EPA offers newly minted "interpretations" that bear no
resemblance to the plain language of the regulations themselves. Tellingly, EPA's proposal
begins not by offering a plausible reading of its regulations, but by explaining why it thinks this
new approach to Texas is beyond the review of a federal court. This is the opposite of reasoned
decision-making.
Response: We disagree with this comment. We never stated that our interpretations of our
regulations were "beyond the review of a federal court." We merely stated that our
interpretations were consistent with the statutory scheme and legislative history and thus entitled
to substantial deference.
860
-------
Comment: EPA's new interpretations of its regional haze regulations are plainly erroneous
and inconsistent with the plain meaning of the regulations [Luminant (0061) p. 84]
Luminant stated that EPA's unprecedented proposal for Texas rests on entirely new
interpretations of its regional haze regulations that are plainly erroneous and contrary to their
plain meaning, as well as inconsistent with prior EPA regional haze guidance interpreting those
regulations.570 EPA's interpretations are thus not a lawful basis for EPA's proposed action,571
and they are arbitrary and capricious because EPA's prior interpretations were relied on by Texas
in the multi-year development of its SIP and have "engendered serious reliance interests that
must be taken into account."572 EPA's new "interpretations" certainly are not entitled to judicial
deference as EPA asserts.573
Luminant noted, specifically, EPA rests its proposal to disapprove Texas' submission on at least
three new interpretations of 40 C.F.R. §§ 51.308(d)(3) that are contrary to the plain text of the
rules and EPA's prior practice under those rules. The first is that the phrase "progress goal" in
40 C.F.R. § 51.308(d)(3)(ii) does not mean a reasonable progress goal established by a state but
instead means a reasonable progress goal that is "approved or approvable" by EPA.574 EPA relies
on this erroneous interpretation to propose disapproval of Texas' long-term strategy, which fully
comports with the reasonable progress goal established by Oklahoma for the Wichita Mountains,
but which EPA claims does not comport with EPA's new proposed reasonable progress goal for
that area. The second is that 40 C.F.R. § 51,308(d)(3)(ii) and (iii) require a state to analyze the
four-factors in 42 U.S.C. § 7491(g)(1) and § 51.308(d)(l)(i) for "both their own Class I areas
and downwind Class I areas."575 The third is that a state may not meet the requirement in 40
C.F.R. § 51.308(d)(3)(iii) to "document the technical basis" for its long-term strategy by relying
on technical analyses developed by a regional planning organization ("RPO") and approved by
all participating states unless the RPO itself has conducted a four-factor analysis of the factors in
§ 51.308(d)(l)(i).576
Luminant asserted that all three of these interpretations are plainly erroneous and inconsistent
with EPA's own regulations and thus cannot form the basis for lawful action on Texas' and
Oklahoma's regional haze SIPs. EPA is wrong that its interpretation of the regulations is entitled
to deference based on the decision in Auer v. Robbins, under which courts defer to agency
interpretations of their regulations if certain conditions are met 577 Auer deference is entirely
unwarranted here. First, even under Auer, no deference is owed an "interpretation" that conflicts
with the plain language of the regulations.578 "[DJeference is warranted only when the language
of the regulations is ambiguous."579 Here, the language of the regulations is clear on its face and
plainly permissive of Texas' approach to meeting the reasonable progress and long-term strategy
requirements, and the regulations do not contain the requirements that EPA seeks to impose on
Texas and Oklahoma here.580 Second, even to the extent there was some ambiguity in EPA's
regional haze regulations, no deference is owed EPA here because its interpretation is contrary to
how EPA has consistently interpreted those regulations since they were enacted. The Supreme
Court has emphasized that Auer deference is inappropriate where an agency's new interpretation
of a regulation "would result in ... 'unfair surprise,"'581 particularly where "the agency's [new]
interpretation conflicts with a prior interpretation."582 EPA itself has admitted that its actions
here are "without. . . prior precedent," and surely this new approach results in "unfair surprise"
to Texas and the Texas sources targeted in EPA's proposal.
861
-------
Moreover, Luminant stated that EPA's plea for deference—at the outset of its proposal—only
confirms that even EPA acknowledges the fragility of its own position. Instead of providing the
best reading of its regulations at the rulemaking stage, EPA is providing a strained interpretation
of its regulations and seeking to justify that "interpretation" by invoking Auer.583 Auer's holding,
however, is subject to considerable doubt and is shaky ground upon which to base an entire
rulemaking that would require $2 billion dollars of expenditures. Indeed, as explained by the
concurring opinions in Perez v. Mortgage Bankers Ass'//, Auer is bad law and should be
overruled. First, Auer violates the separation of powers principle, because it permits the same
branch of government to both enact a rule and to interpret it.584 Second, Auer deprives the
judiciary of the power to "say what the law is," as required under Article III.585 EPA's proposal
here is a vivid illustration of both of these shortcomings in Auer.
According to Luminant, even under Auer, EPA's interpretations cannot be applied here. EPA
concedes that the new interpretations of its regulations that underpin its unique action here are
novel and have never been announced by the agency prior to this proposal.586 EPA's premise for
making these new interpretations— that "some uncertainty exists as to the respective roles and
responsibilities of upwind and downwind states in addressing visibility impairment"587—is itself
incorrect. As discussed above and throughout these comments, the respective roles of the states
are clear— they are to consult about potential impairment to each other's Class I areas through
regional planning organizations and seek to reach agreement on required emission reductions, as
Texas and Oklahoma did here. EPA points to no instance in which "uncertainty" has prevented
states from fully consulting under the regional haze program. EPA's attempt to manufacture
some "uncertainty" is a transparent effort to justify its new and novel approach under which EPA
would undercut the roles and responsibilities of the states and impose its own view of what is
"reasonable," in contravention of the plain language of its regulations.
Footnotes:
570 Id. at 74,828-30.
571 Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 414 (1945) (agency interpretation of its regulation is not
controlling where "it is plainly erroneous or inconsistent with the regulation"); see also Auer v. Robbins, 519 U.S.
452, 461 (1997) (same).
572 Perez v. Mortgage Bankers Ass'n, 135 S. Ct. 1199, 1209 (2015) (internal quotations omitted).
573 7 9 Fed. Reg. at 74,829.
574 Id.
575 Id.
576 Id.
577 Id. at 74,828-29.
578 Christopher v. SmithKline Beecham Corp., 132 S. Ct. 2156, 2166 (2012).
579 Christensen v. Harris Cnty., 529 U.S. 576, 588 (2000).
580 Id. ("The regulation in this case, however, is not ambiguous—it is plainly permissive.").
581 Christopher, 132 S. Ct. at 2167.
582 Id. at 2166.
583 Id.
584 See Perez v. Mortgage Bankers Ass'n, 135 S. Ct. 1199, 1213-14 (Mar. 9, 2015) (Scalia, J., concurring in the
judgment); see also Decker v. Nw. Envtl. Def. Ctr., 133 S. Ct. 1326, 1339-42 (2013) (Scalia, J., concurring in
part and dissenting in part).
585 See Perez, at 1222-23 (Thomas, J., concurring in the judgment).
586 79 Fed. Reg. at 74,823.
587 Id.
862
-------
Response: We disagree with this comment. For a discussion of why our clarified interpretations
of the quoted regulatory provisions are neither "plainly erroneous" nor "new," see our responses
to comments in Section l.g. In regards to reliance interests, the commenter's assertions do not
square with reality. Indeed, Texas' analyses and statements in its own SIP demonstrate that
Texas knew that the CENRAP analysis was merely a starting point that required supplementation
and that the Regional Haze Rule required States to conduct a four-factor analysis for both in-
state and downwind Class I areas. Thus, there was no "unfair surprise" as the commenter
suggests. Furthermore, we never stated that our actions were "without. . . prior precedent." The
full statement, included in a declaration from EPA Deputy Regional Administrator Sam Coleman
to the U.S. District Court for the District of Columbia Circuit, reads as follows:
EPA is undertaking the appropriate modeling. The additional modeling will enable
EPA to develop more specific information concerning the visibility impairment
caused by individual sources. Due to the large distances involved and the large
number of sources being analyzed, EPA is utilizing a different model than the
standard models used previously by EPA and States for Regional Haze SIPs. This
model is complicated and the results it generates are not directly comparable to the
modeling platform used in most other States' Regional Haze submittals.
Appropriate review of this model therefore requires thorough technical and policy
analysis and interpretation to ensure compliance with the Regional Haze Rule,
without the benefit of prior precedent to streamline the process.615
Thus, the actual quote, read in context, merely states that our review of the visibility model used
in the FIP did not have the benefit of prior precedent because it was not used in "most other
States' Regional Haze submittals." The commenter takes this quote grossly out of context to
support its position. Notably, the commenter again fails to mention that Texas itself used the
same visibility model when developing its regional haze SIP due to the large distances between
Texas sources and impacted Class I areas. We thoroughly explained the technical reasons why
we used this same model in our proposal and again in this final rule. Finally, the commenter's
attempt to invoke the dissent from Mortgage Bankers as if it were controlling law is without
merit. The deferential standard in Auer v. Robbins applies to an agency's interpretations of its
own regulations until a majority of the Supreme Court holds otherwise.
Comment: Earthjustice et al., stated that EPA's Interpretation of 40 C.F.R. Sections
51.308(d)(1) and (d)(3) is the Only Reasonable Reading of the Regulations, and EPA Need
Not Rely on Auer Deference to Support its Determination that Neither the Texas Nor
Oklahoma SIPs Provided for Reasonable Progress Toward Natural Visibility in Class I
Areas. [Earthjustice (0067) p.29, 31]
Earthjustice et al., stated that the Regional Haze Rule requires each state with a Class I area to
"establish goals . . . that provide for reasonable progress towards achieving natural visibility
conditions" {i.e., reasonable progress goals). 40 C.F.R. § 51.308(d)(1). In establishing control
675 Declaration of Sam Coleman, Nat'l Parks Conservation Ass'n v. McCarthy, No. 11-01548, at 5 (D.D.C. 2014)
(emphasis added).
863
-------
measures that will achieve those reasonable progress goals, the state must consider the four
statutory factors outlined in Section 169A(g)(l) of the CAA—"the costs of compliance, the time
necessary for compliance, the energy and non-air quality environmental impacts of compliance,
and the remaining useful life" of any potentially affected sources. 42 U.S.C. § 7491(g)(1).
Additionally, the state "must consult" with any other state which "may reasonably be anticipated
to cause or contribute to visibility impairment" at a Class I area within the state, and "develop
coordinated emission management strategies." 40 C.F.R. § 51.308(d)(l)(iv), (3)(i). Each state
must then "document the technical basis, including modeling, monitoring and emissions
information" to "demonstrate that it has included in its implementation plan all measures
necessary to obtain its share of the emission reductions needed to meet the progress goal for" the
affected Class I area. Id. § 51.308(d)(3)(ii)-(iii).
In disapproving the Texas and Oklahoma Regional Haze SIPs, EPA suggests that these
provisions do not explicitly address situations like this case:
where the control measures in an upwind state's long-term strategy are sufficient
to obtain its share of reductions needed to meet a RPG included in a downwind
state's SIP, but the goal itself is flawed precisely because the upwind state never
proposed sufficient control measures to ensure reasonable progress in the first
place. To prevent such situations, we interpret the term "progress goal" in Section
51.308(d)(3)(ii) as an approved or approvable progress goal.37
Earthjustice et al., stated that, according to EPA's interpretation, where states fail to consult and
properly assess whether any additional upwind state emission reductions are necessary to achieve
a reasonable progress goal at a downwind state Class I area, as Texas and Oklahoma failed to do
here, EPA must disapprove both SIPs.
While EPA invokes Auer deference in its proposed rule, EPA's regulatory interpretation does not
implicate the public participation concerns frequently associated with the rule of deference
articulated in Auer v. Robbins, 519 U.S. 452 (1997). In sharp contrast to the regulatory
interpretation offered in an amicus brief in Auer, or the policy guidance at issue in similar cases,
EPA has proposed to clarify the transport and consultation requirements of the Regional Haze
Rule and the Clean Air Act through an open and transparent notice and comment process, invited
public participation, and explained its reasoning. As a result, EPA need not rely on Auer
deference to support its interpretation of the existing haze regulations.38 There is an additional
reason that EPA need not invoke Auer deference here. EPA's interpretation of the Clean Air Act
and the Regional Haze Rule is not only the fairest and most natural reading of those provisions,
but is the only interpretation that is consistent with the regulation as a whole, the Clean Air Act,
and EPA's obligation to ensure that each state SIP is consistent with all applicable requirements
of the Act. See Decker v. Nw. Envtl. Def Ctr., 133 S. Ct. 1326, 1339-40 (2013) (Scalia, J.,
dissenting) (where an "agency's interpretation of the rule is also the fairest one," rather than an
"unnatural reading," deference is unnecessary to uphold decision). Additionally, EPA's decision
reflects a reasonable interpretation of the statute under Chevron v. Natural Res. Def. Council,
467 U.S. 837(1984).
Footnotes:
37 79 Fed. Reg. at 74,829 (emphasis in original).
864
-------
38 See, e.g., Perez v. Mortgage Bankers Ass 'n, 135 S. Ct. 1199 (2015) (Scalia, J., concurring)
(criticizing Auer deference because it encourages agencies to "write substantive rules more broadly and vaguely,
leaving plenty of gaps to be filled in later, using interpretive rules unchecked by notice and comment"); see also
Appalachian Power Co. v. EPA, 208 F.3d 1015, 1020 (D.C. Cir. 2000) (criticizing agency use of guidance
documents in the form of interpretive rules and policy statements, recognizing the potential problem that "[l]aw is
made, without notice and comment, without public participation, and without publication in the Federal Register or
the Code of Federal Regulations.").
Response: We generally agree with this comment, but take no position as to specific statements.
23. Reasonable Progress Goal Progress
Comment: TCEQ's four-factor analysis was reasonable and within its broad discretion,
and is supported by recent monitoring data showing RPGs will be met without additional
controls. [CCP (0075) p. 5-7]
CCP stated that actual emissions data supports the reasonableness of TCEQ's RPG
determinations. EPA acknowledges that States "may take credit in your long-term strategy for
emission reductions achieved after 2002." EPA RPG Guidance at 4-1. Indeed, even in the
Proposed Rule EPA notes that "more recent IMPROVE monitored data" showing "more
progress than anticipated by the CENRAP modeling has occurred." 79 Fed. Reg. 74,843. This
monitoring data, based on actual emissions, shows that SO2 emissions have decreased by nearly
50% from the baseline period through 2012. See Figure 1 provided by CCP. In fact, the most
recent five-year (2009-2013) averages of visibility monitoring data are 15.3 dv at Guadalupe
Mountains, 16.3 dv at Big Bend, and 21.2 dv at Wichita Mountains, each of which is lower than
EPA's proposed 2018 reasonable progress goals. See 79 Fed. Reg. 74,870. Thus, the three
Class I areas Big Bend, Guadalupe Mountains, and Wichita Mountains will meet the RPGs that
EPA proposes without further emission controls totaling more than $2 billion. Yet EPA wholly
ignores these actual conditions in developing its RPGs and in concluding that its RPGs are more
reasonable.
Texas SO2 Emission Trends (Figure 1 provided by CCP)
865
-------
1,000.000
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Year
CCP stated that a central tenet of EPA's FIP is that TCEQ's SIP is insufficient to meet RPGs to
meet the glide path toward natural visibility conditions by 2064. But EPA's Proposed Rule also
does nothing to accelerate the final goal of natural visibility levels. EPA's FIP purports to reach
natural visibility levels at Guadalupe Mountains only after 141 years and, at Big Bend, only after
173 years. This is not a significant improvement over Texas' date for obtaining natural
conditions, and demonstrates the absurdity of EPA's position.
CCP stated that, because TCEQ carefully considered the first of the factors, cost of compliance,
and determined that cost was so excessive relative to benefit, it determined that detailed analysis
of the other three factors was unnecessary even though it would have further supported a
determination that additional controls was unreasonable.
Response: We address consideration of recent monitoring data, natural visibility conditions and
URP, and Texas's four-factor analysis elsewhere in this document.
Comment: Multiple commenters noted that emission reduction impacts from other
regulations have improved visibility.
[EEI (0076) p. 2] EEI stated, as part of the ongoing transition to cleaner sources of generation,
the electric power sector generally - and Oklahoma and Texas generators in particular - have cut
emissions significantly in recent years and will continue to do so under many EPA regulations,
even without the regional haze FIP proposed by EPA here.
AECOM (0075) explained that many Texas programs have been in place in the state over the
years to achieve lower SO2 emissions and visibility improvements. These programs will work in
conjunction with the Regional Haze Rule to continue the documented reduction in SO2 emissions
866
-------
and other contributing factors to visibility impairment, and in reducing the need for additional
future controls on EGUs. A list of programs is provided in AECOM (0075).
AECOM (0075) stated that actual SO2 emissions from Texas have been significantly reduced
since the regional haze rule baseline period (2000-2004). As noted by the TCEQ,32 statewide
SO2 emissions had decreased by nearly 50% from the baseline period through 2012, as shown in
AECOM (0061/0075) Figure 2-1). The Texas 5-year regional haze progress report33 issued in
2014 indicates a projection of further reductions of haze-forming SO2 and NOx emissions from
point sources through 2018 (Figure 2-2 of AECOM 0061/0075). Therefore, it is expected that
visibility improvements observed through 2013, as discussed below for WIMO, GUMO, and
BIBE, will continue through the 2018 interim goal year.
[Alpine (0078) p. 22] Alpine concurred with EPA's findings that this observed improvement
from the baseline conditions is the result of reduction in the impacts from SO2 emissions and a
reduction in the impacts from coarse material but disagreed that it is related to meteorological
conditions more than emission changes. Alpine provided Figure 13 showing SO2 emissions in
Texas from 1999 through 2011 and noted that a reduction of 50% in SO2 emission has occurred
across all anthropogenic categories State-wide. A significant portion of this reduction is
attributed to electric utility coal fuel combustion (38%), as well as from industrial fuel
combustion and processes {10%). And according to TCEQ's point source emission inventory,
these values continue to decrease beyond 2011 levels for all point sources reporting to the State.7
Texas SO2 and NOx Emissions Trends
(AECOM Figures 2-1 and 2-2)
867
-------
Figure 2*1:
l.OOG ODD
TO COO
Texas SO2 Emission Trends
• "j
a
LA
5
c
&QG.OOO
7OQ0DO
600,000
soaooo
Texas Statewide Point Source SO, EmissionsTrends
£ 4GQODO
V
O 300,000
St
^ 20aceo
c
5 10Q 000
1997 193? 1999 2000 2001 200? 200S 200* 2005 2006 20C7 2008 2009 2010 2011 2012
Year
Figure 2-2; Projected Texas SQi and NOx EGU Emissions Trend through 2018M
Actual and Projected Emissions Trends for
Electric Power Generation
700,000
600,000
500,000
400,000
300,000
700,000
100,000
-S02
-NOx
-Linear (502 CENfWP Model)
oooooooooooooooo
«NiNiiNr
-------
¦ Electric Utility Coal Fuel Combustion ¦ Mobile Sources I Industrial Fuel Combustion & Processes BAN Others
Source: http://www.midwestozonegroup.com/files/AQTrendsSunmiary_Texas_3.ppt
[Luminant (0061) p. 96] Luminant stated that the progress under Texas's SIP is demonstrated
and undisputed. All record evidence indicates that the visibility improvement for these three
areas will persist and continue through 2018. In its five-year regional haze progress report,
Texas demonstrated that emissions of both SO2 and NOx from Texas generating units are
declining and will continue to decline under existing programs.631 As Texas explained in its
progress report: "Emissions of NOx decreased 44% from 255,556 tons in 2002 to 143,782 tons in
2011. Sulfur dioxide emissions decreased 23% from 560,860 tpy to 433,782 tpy during the same
period. Emissions have trended downward better than or as predicted in the CENRAP modeling
projections."632 And with current programs in place that will continue through 2018, including
CSAPR and MATS, there is no possibility that emissions will increase, and current visibility
conditions degrade, between now and 2018.
According to Luminant, SO2 emissions from Luminant's units that EPA would subject to new
requirements have steadily trended downward over the first planning period, further
underscoring the effectiveness of the measures relied on in Texas's SIP and the unreasonableness
of EPA's assumptions and proposed FIP. From 2009 to 2014, SO2 emissions from Luminant's
Big Brown, Martin Lake, Monticello, and Sandow Unit 4 were reduced by 27%.
(Table 11 and Figure 9 provided by Luminant (0061)) shows the quarterly and annual SO2
emissions for these units from 2009 to 2014, as well as the first quarter of 2015.633
869
-------
Quarterly and Annual SO2 Emissions from Luminant Units Proposed by EPA for Further
Controls (Table 11 and Figure 9 provided by Luminant (0061))
2009
2010
201 1
2012
2013
2014
*2015
Quarter 1
49,798
52,315
51,042
35,046
37,716
38,334
21,331
Quarter 2
45,429
49,136
51,967
37,669
38,880
32,041
Quarter 3
58,599
57,868
61,863
43,308
56,242
56,203
Quarter 4
57,423
52,674
47,291
41,710
36,544
26,987
Annual
Totals
21 1,249
211,994
212,164
157,733
169,382
153,565
21,331
40,000
¦ J
•2015 QI emissions subject to
change, pending final QA &
submission
2009 2010 2011 2012 2013
¦ Quarter I ¦ Quarter 2 ¦ Quarter 3 ¦ Quarter 4
2014
•2015
Luminant noted that first quarter 2015 SO2 emissions are sharply lower—approximately 57%
lower than the first quarter of 2009 and about 44% lower than the first quarter of 2014. Again,
the data unequivocally show that SO2 emissions are trending down, and thus there is no basis for
EPA's proposal. These data also underscore the arbitrary nature of EPA's approach of adjusting
ENVIRON's modeling based on three years of "actual" emissions from 2009 to 2013.634 As
explained by Sargent and Lundy in its report prepared in support of Luminant's comments,
EPA's selective use of data significantly understates the cost-effectiveness (that is, understates
the cost per ton of SO2 removed) and overstates the estimated benefits that EPA relies on for its
870
-------
proposal.635 A more representative 3-year period would be the most recent 3 years for which full
data is available—2012-2014.636
Footnotes:
7 https://www.tceq.texas.gov/airquality/point-source-ei/psei.html
32 http://www.tceq.state.tx.us/airquality/airsuccess/air-success-inventory.
33 https://www.tcea.texas.gov/assets/public/implementation/air/sip/haze/13012SIP ado.pdf.
34 http://www.tceq.state.tx.us/airquality/airsuccess/air-success-inventory.
35 https://www.tceq.texas.gov/assets/public/implementation/air/sip/haze/13012SIP_ado.pdf at 4-10.
631 See TCEQ, 2014 Five-Year Reasonable Progress Report 4-10, fig.4-2 (Feb. 26, 2014), available at
https://www.tceq.texas.gov/assets/public/implementation/air/sip/haze/13012SIP_ado.pdf ("Texas Five-Year
Progress Report").
632 Id. at 4-10.
633 The 2015 data is still being reviewed and is subject to change pending quality assurance and submission to
TCEQ.
634 See FIP TSD at A-59.
635 S&L Report at 11, 13,21,23.
636 Id. at 21.
Response: We acknowledge that actual SO2 and NOx emissions from Texas point sources have
decreased since the baseline period for the first planning period (2000-2004). As noted in our
proposed rulemaking, we also acknowledge that there is an observed reduction in the impacts
from SO2 emissions and a reduction in the impacts from coarse material at Big Bend, Guadalupe
Mountains, and Wichita Mountains.676 Some commenters also point to Texas' five-year regional
haze progress report as showing that SO2 and NOx emission from Texas EGUs are expected to
continue to decline through 2018, along with a continued trend in visibility improvement at the
three Class I areas. With regard to the assertion that Texas' five-year regional haze progress
report projects SO2 and NOx emissions from point sources to continue to decline through 2018
(with corresponding visibility improvement trends at the three Class I areas), Texas' five-year
regional haze progress report is pending evaluation as a SIP revision, and we intend to take
action on it in a future rulemaking. We note that the portion of the Texas' five-year regional
haze progress report referred to by the commenters677 compares actual annual emissions from
2002 through 2011 against a linear change between 2002 actual emissions and the 2018
CENRAP modeled emissions and concludes that emissions from 2002 to 2011 have trended
downward better than or as predicted in the CENRAP modeling projections. However, we noted
in our proposal that the CENRAP projected visibility impacts in 2018 from Texas point sources,
and EGUs in particular, are significant. As noted in our proposed rulemaking, based on
information provided by the TCEQ in materials other than the progress report, we do not expect
large additional emission reductions of SO2 in Texas between 2013 and 2018 under Federal
programs and the SIP as submitted.678 We have not seen evidence in support of something
different. Furthermore, emissions from some of the Texas EGUs that we are requiring controls
676 79 FR at 74843 and 74870.
677 2014 Texas Five-Year Reasonable Progress Report, p 4-10, figure 4-2.
678 Texas comments on Draft IPM modeling conducted by EPA for potential national rule making platform
provided on June 26, 2014. In this docket materials as "TCEQ comment letter to EPA on draft modeling platform
dated June 24, 2014. '2018 EMP signed.pdf.
871
-------
for and that impact visibility at the three Class I areas the most, are still above the emission level
projected in the 2018 CENRAP modeling. We are not aware of any upcoming controls or
changes in operation to suggest that future actual emissions at these specific sources will
decrease to those predicted levels. 679 We also note that our analysis uses recent emissions data
to identify those sources with the greatest potential to impact visibility and in assessing the
potential visibility benefits from controls. In addition, we considered recent emission reductions
by adjusting the 2018 modeled emission inventory to account for reductions at sources when
controls had been installed and relied on post-control actuals to support modeled emission
rates.680
We disagree with comments contending that emissions of both SO2 and NOx from Texas
generating units will continue to decline through 2018 under existing programs. As noted in our
proposed rulemaking, based on information provided by the TCEQ, we do not expect large
additional emission reductions of SO2 in Texas between 2013 and 2018 under federal programs
and the SIP as submitted.681 We have not seen evidence in support of something different. As we
explained in the FIP TSD682:
.. .based on recent comments from the TCEQ, it was also unclear if any further
SO2 or NOx reductions would occur due to these rules even if all litigation was
resolved. The TCEQ has provided extensive comments on recent emission
inventory indicating that further significant reductions in SO2 were not expected
due to CSAPR or MATS.683 We thought it was reasonable to continue to rely
upon the initial CENRAP 2018 modeling inventory initially and update the
information that we were more certain about as discussed above. We utilized
2009-2013 CEM data for EGUs in evaluation and selection of updated emission
levels to model.684 Comparison of recent CEM data with CAIR projections
indicated that the Texas EGUs were on track to meet the CAIR requirements
without further substantial reductions. For the ENVIRON modeling we did not
increase emissions for existing sources based on recent actuals but we did lower
emissions for some sources when controls had been installed and relied on post-
control actuals to support modeled emission rates. TCEQ in recent ozone
attainment modeling has also used recent CEM data to represent expected
emissions levels from Texas EGUs for future year of 2018 in recent Houston and
DFW area modeling.
With regard to the commenters' contention that the downward trend in emissions from Texas
point sources, especially Luminant's facilities, underscores the effectiveness of Texas' RH SIP,
we note that the annual and quarterly SO2 emissions data for Luminant's facilities for 2009-2015
679 79 FR 74842 and 74843.
680 Table 2 of ENVIRON 2018 Memorandum, Sept 16, 2013. Available in the docket for this action as TX166-010-
09 Memo_TXHAZE_2018CAMx 16Septl3
681 79 FR 74870.
682 FIP TSD at A-16
683 TCEQ comment letter to EPA on draft modeling platform dated June 24, 2014. '2018 EMP signed.pdf
684 Emission rates/data used in modeling are included in the report and electronic file
"Summary_emissions_for_39_selected_072913_ENV.xlsx" and CEM data included in file "TX Sources of Interest
Emissions 2007-2012 msf annualestimates.xls
872
-------
provided by the commenter demonstrates that although there has been an overall downward
trend in annual SO2 emissions during this time period, there has not been a downward trend in
SO2 emissions during Quarter 3 for the six year period for which full data is available. Except
for the years 2011 and 2012, when total SO2 emissions for Quarter 3 were either sizably higher
or lower compared to the other years during the 2009-2014 time period, emissions for Quarter 3
remained relatively unchanged during this six year period. This is significant because Quarter 3
corresponds to the summer months and many of the 20% worst days, which is what the RPGs are
based on, typically occur during the summer months. Emissions reductions during the fall and/or
winter months reduce annual emissions, but will not lead to improved visibility during the 20%
worst days. As shown in the figures below, the majority of the decline in annual SO2 emissions
is driven by seasonal operation of Monticello units 1 and 2.685 We also note, as discussed above,
NOx emissions for many of these units were updated in our modeling to better reflect the recent
actual emissions.
Big Brown Monthly Emissions (2009-2014)
Sum ol 502 Hons)
4900
4000
D«e r
Martin Lake Monthly Emissions (2009-2014)
685 See Luniinant CAMD emissions.xlsx in the docket for this action
873
-------
Sum of SQ2 (tons)
FacMy Nome -T
imto ~
-Martin Uko • 2
-Martin uKe-3
^ J? ji* rf^ rj? J? JJ'P rf? J^P (J^ jA1 jAN jAS ^ J^v* rf? jjN1' J jJ1 AV jAV rfj* jff1
Monticello Monthly Emissions (2009-2014)
Sum of 502 ftoini
1500
-MAntr»Ho-1
-Monflcello 2
- Mooterllci 3
\\ <$r ^ ^ Ns> ^v- «,v 0)V ^ «,v «\V o>v s^r Nv ^ -\v <$> N«y %v <$~ -\v # Vv -^v iv «>y
Sandow 4 Monthly Emissions (2009-2014)
874
-------
Sum of 5Q2 (tons)
Sandow - 4
J 000
IWO
Datf »
We also remind the commenters that even with the controls we are requiring for Texas EGUs
under our FIP, additional reductions would be needed for visibility conditions to meet or exceed
every URP goal in 2018 as calculated by us in our proposal. For example, current conditions at
Wichita Mountains (based on 2009-2013) is 21.2 dv, therefore additional reductions are needed
for the area to meet the URP goal of 20.01 dv in 2018. We have estimated that even with the
controls we are requiring (including those with an installation date beyond 2018), it is estimated
to take 82 years to reach natural visibility conditions at Wichita Mountains, 173 years at Big
Bend, and 141 years at Guadalupe Mountains, all falling short of the national goal of achieving
natural visibility conditions by 2064.686 We disagree with the commenter that our FIP "does
nothing to accelerate the final goal of natural visibility levels" and "is not a significant
improvement over Texas' date for obtaining natural conditions." We estimate that the required
controls are a significant acceleration of the final goal, reducing the number of years to meet
natural visibility by 25 to 30 years at Big Bend and Guadalupe Mountains. This is significant
progress towards the goal during this first planning period when compared to the estimates in the
Texas SIP. As we note in our proposal:
If the State determines that the amount of progress identified through the [URP]
analysis is reasonable based upon the statutory factors, the State should identify
this amount of progress as its reasonable progress goal for the first long-term
strategy, unless it determines that additional progress beyond this amount is also
reasonable. If the State determines that additional progress is reasonable based on
the statutory factors, the State should adopt that amount of progress as its goal for
the first long-term strategy [emphasis added].687
This determination must be based on the statutory factors. As we discuss in our proposal,688
Texas's reasonable progress four factor analysis was highly flawed, and thus could not serve as a
legitimate basis for rejecting additional progress including progress beyond the URP.
686 79 FR 74887
687 64 FR 35732 (July 1, 1999).
688 See discussion beginning on 79 FR 74838.
875
-------
Furthermore, we disagree that the actual observed reduction in SO2 and NOx emissions from
Texas point sources and any projected future reduction through 2018 makes our proposed FIP
unreasonable and unwarranted. As discussed in our proposal, in our final rule, and elsewhere in
this document, we have determined that the Texas and Oklahoma RH SIPs do not fully meet the
requirements of the RH rule. Oklahoma did not explicitly ask Texas for additional reductions
from Texas sources impacting Wichita Mountains. Texas performed a technically flawed
analysis for both establishing its Class I areas' RPGs and for determining under LTS its fair
share of emissions reductions for Wichita Mountains. We therefore were required to conduct
our own analysis under sections 51.308(d)(1) and (3). We conducted our own analysis in
accordance with the RH rule and the CAA and therefore, we believe our FIP is reasonable and
necessary to fully satisfy the regional haze requirements for Texas and Oklahoma. We identified
those sources with the greatest visibility impacts and determined based on consideration of the
four factors that for some of these sources controls were reasonable and provided for meaningful
visibility improvements towards the goal of natural visibility conditions for this first planning
period. In evaluating these specific sources, we considered recent emissions data.
We disagree with Luminant's contention that our use of three years of actual emissions from
2009 to 2013 is an arbitrary approach that understates the cost per ton of SO2 removed and
overstates the estimated benefits. We respond to that comment in the cost section of this
document.
Comment: Luminant provided background information on current visibility conditions in
the three relevant Class I areas and on the IMPROVE visibility monitoring network.
[Luminant (0061) p. 53]
Luminant noted that visibility conditions at Class I areas are monitored by a network of visibility
monitors coordinated and funded by EPA, other federal agencies, and multi-state air quality
organizations as part of the IMPROVE program. IMPROVE monitoring data are the backbone
of the regional haze program. The monitoring data from these sites is "used to establish baseline
conditions (for the 2000-2004 period) for each Class I area and to track progress toward goals
established in future SIPs."384 IMPROVE monitoring data is also used to validate modeling
results. As EPA has explained: "Model performance at IMPROVE monitors is of highest
importance, because these monitors are sited to be representative of the visibility conditions
impacting each Class I Area."385
Luminant noted that the IMPROVE network started in 1986 at two Class I areas and expanded to
30 sites in 1988.386 Among the new sites added in 1988 were the Big Bend (BIBE1) and
Guadalupe Mountains (GUMOl) monitors.387 After EPA's issuance of its regional haze
regulations in 1999, the network was upgraded and expanded to 110 sites nationally to cover all
of the 156 mandatory Class I areas.388
Luminant stated that the three Class I areas of interest in EPA's proposal here are each served by
a separate IMPROVE monitor—Big Bend's monitor is designated BIBE1; Guadalupe
Mountains' monitor is designated GUMOl; and Wichita Mountains' is designated WIMOl.389
However, some IMPROVE monitors in the network serve more than one Class I area. The 110
876
-------
current monitoring locations were selected to provide representative visibility monitoring at all
156 mandatory Federal Class I areas, except the Bering Sea Wilderness (which is too remote for
monitor deployment).390 As we explained:
Because of the broad spatial distributions of regional haze, and in order to use
monitoring resources efficiently, EPA determined, in conjunction with State and
Federal land managers, that some neighboring mandatory Federal Class I areas
could be represented by a single monitoring site . . . The EPA consulted with the
States in order to design a network that was as representative of all mandatory
Federal Class I areas as possible.391
Luminant noted that when the IMPROVE network was expanded after the 1999 regulations were
issued, the IMPROVE Steering Committee developed the plan for future network
configuration.392 As part of this planning process, the Committee concluded that "from a
technical perspective, monitoring sites that are relatively near to each other in remote areas can
be expected to collect similar data that might be considered redundant. . . Since regional haze
impacts are by definition those that are distributed over a broad geographic region, a
representative monitoring site does not necessarily need to be located very near to the [Class I
area] being represented."393
Luminant noted that, accordingly, the IMPROVE Steering Committee developed a "two-stage
process ... for selecting new monitoring sites."394 As relevant here, using this process, the
Committee concluded that the monitor for Guadalupe Mountains (GUMOl) would also serve as
the monitor for Carlsbad Caverns National Park, which is located in New Mexico and contiguous
with Guadalupe Mountains.395
Footnotes:
384 EPA, Guidance for Tracking Progress Under the Regional Haze Rule 1-4 (Sept. 2003) ("EPA Guidance for
Tracking Progress"). See also 64 Fed. Reg. at 35,728 (discussing use of IMPROVE data under Regional Haze
Rule).
385 CENRAP Modeling TSD at 34.
386 EPA Guidance for Tracking Progress at 1-12.
387 See IMPROVE, Spatial and Seasonal Patterns and Temporal Variability of Haze and Its Constituents in the
United States: Report III, at 1-6 (May 2000) ("IMPROVE 2000 Report"), available at
http://vista.cira.colostate.edu/improve/publications/reports/2000/2000.htm.
388 EPA Guidance for Tracking Progress at 1-3.
389 See Vista, IMPROVE/Protocol Network in Year 2002,
http://vista.cira.colostate.edu/improve/Overview/IMPROVENetworkExp.htm (last visited Mar. 22, 2015) (listing of
monitor locations).
390EPA Guidance for Tracking Progress at 1-3, 1-12.
391 Id. at 1-13 to 1-14.
392 IMPROVE 2000 Report at 1-7.
393 Id. at 1-7, 1-9.
394 Id. at 1-10.
395 Id.
Response: We acknowledge the historical information provided by the commenter. We take no
position with respect to the commenter's synopsis.
877
-------
Comment: Multiple commenters stated that EPA's proposal is unnecessary and outside of
its authority because EPA's 2018 visibility goals for Big Bend, Guadalupe Mountains, and
Wichita Mountains have already been achieved based on IMPROVE monitoring network
data without any further emission reductions or controls on Texas sources.
Luminant, EEI, AECOM and AECT provided the data below showing actual monitored visibility
compared to the goal that EPA claims must be achieved by 2018.
Monitored Visibility Compared to EPA's Proposed Goals
(deciviews, 20% worst days)
(Table provided by Luminant (0061), EET (0076), and AECOM (0061/0075))
Class I Area
Actual Observed
Conditions from
IMPROVE Monitors
(2(109-13) (5-yetir
average) in tleeivicws
EPA's Proposed KP(,
(2(!i.26
Wichita Mountains
21,2
21.33
Source: 79 FR 74843, 74870 and 74887, tbl.43.
[Luminant (0061) p. iv] Luminant stated that the Texas regional haze SIP has been in place
since 2009, and it is working. Texas emissions have decreased, and visibility has improved—
beyond what even EPA would require in this proposal. While EPA relies on computer modeling,
actual real-world data clearly and irrefutably demonstrate the effectiveness of the Texas plan.
These improvements are the result of on-going emission reductions that will continue under
Texas rules and national programs like the Cross-State Air Pollution Rule (CSAPR) and the
Mercury and Air Toxics Standard (MATS)—all without the additional $2 billion that EPA would
require. Luminant stated that the EPA ignores these data from its own monitors and even refuses
to account for CSAPR and MATS limitations in the modeling it uses to justify its proposal. But
the reality remains the same—Texas's plan is working and visibility is improving. EPA should
withdraw its FIP and approve the Texas SIP.
[Luminant (0061), p. 2, 55] Luminant stated that the EPA concedes that data from IMPROVE
monitors is "of highest importance,"9 yet EPA fails to account for the more recent, real-world
data from these monitors in its flawed modeling and proposed FIPs. Because the reasonable
progress goals have already been achieved, EPA has no authority to require further controls from
Texas sources.
[EEI (0076) p. 8-9] EEI concluded that the existing regulatory programs put into place by Texas
in its SIP are accomplishing in practice what EPA seeks to achieve with the imposition of more
878
-------
stringent control equipment on a regulatory timeline that will not deliver any benefits within the
first planning period.
[AECOM (0061/0075) p. 1-5] AECOM stated that actual visibility monitoring data indicates
faster progress toward the RPG and URP goals than EPA is assuming. Current monitoring data
from the IMPROVE25 network using the most recent 5-year average show that the three Class I
areas of interest already meet both the RPGs proposed by the States and the more stringent RPGs
proposed by EPA without any further emission controls. These current data also show that
WIMO is projected to meet the EPA approved URP for Oklahoma, and GUMO is projected to
meet the EPA-proposed URP by 2018, without the emission controls that EPA is proposing.
EPA notes that "more recent IMPROVE monitored data" at the three Class I areas "indicate
more progress than anticipated by the CENRAP modeling has occurred."27 EPA attributes this
improvement to "meteorological conditions, reduction in the impact of SO2 emissions, and a
reduction in the impacts from coarse material."28 Yet, EPA does not take this information into
account in its proposed URP and RPGs for the Class I areas and emission reductions for the 14
Texas sources.
[CCP (0075) p. 2] CCP expressed concern that actual visibility monitoring data shows that the
three Class I areas already meet the TCEQ- and EPA-proposed RPGs without any further
emission controls. CCP stated that significant reductions in SO2 emissions from Texas sources
since the baseline period (2000- 2004) have resulted in actual visibility improvements not
considered by EPA in proposing its RPGs and the required emission controls.
[AECT (0074) p. 6] AECT noted that the most recent five-year (2009-2013) averages of EPA's
IMPROVE visibility monitoring data show that EPA has not provided adequate support that any
of the additional SO2 emissions controls for the identified Texas EGUs is needed to meet EPA's
proposed 2018 reasonable progress goals for any of the three Class I areas.18 Further, visibility
monitoring data for 2018 should be even lower than it was in that five-year period since EPA's
IPM modeling predicts that visibility-impairing emissions in Texas will progressively decline
through 2018. Accordingly, AECT believes that EPA has not supported its proposal to require
additional SO2 emissions controls for any of the identified Texas EGUs.
Luminant (0054-3) stated that the EPA's proposal is based in large part on visibility goals for the
Wichita Mountains National Wildlife Refuge in Oklahoma and making sure that Texas sources
do not interfere with meeting those goals. The commenter stated that current data from EPA
monitors for the Wichita Mountains confirms that visibility conditions in that refuge have
improved steadily over the last 10 years, which EPA attributes to reductions in SO2 emissions
from Texas. In fact, current visibility monitoring data shows that the Wichita Mountains already
meet the 2018 goal that EPA has determined is reasonable. This progress is well ahead of the
2018 schedule, current regulatory programs will ensure this progress is maintained without the
additional cost that EPA's proposal would entail.
[UARG (0065) p. 30] As part of their argument that the proposed FIP is arbitrary, capricious,
and an abuse of discretion because it would have no perceptible effect on visibility conditions in
any Class I area, UARG noted that the proposed rule acknowledges that even in the absence of
the controls EPA now proposes, recent monitoring data at Wichita Mountains "indicate that more
879
-------
progress than anticipated by the CENRAP modeling has occurred." Id. at 74,870. EPA notes
that for 2009-2013 average conditions for the 20 percent worst days, Wichita Mountains
experienced 21.2-deciview visibility conditions, which is better than the 21.5-deciview average
that CENRAP projected for 2018. Id. EPA further notes that "[m]ore recent emission inventory
data show reductions in emissions in most states beyond what was projected in the 2018
modeling, including large reductions in emissions from the Eastern United States [and]
[ejmissions from non-EGU Texas point sources." Id. Given these facts admitted by EPA, the
Agency has failed to explain why the stringent emission reduction requirements in its proposed
rule are needed.
[TCEQ/PUCT (0056) p. 4] The TCEQ stated that the EPA's proposed partial SIP disapproval
and partial FIP undervalue the effectiveness of the long-term strategy embodied in the Texas
2009 RH SIP. Without presenting evidence, the EPA dismisses the progress made as being due
to "meteorological conditions, reduction in the impacts from SO2 emissions, and a reduction in
the impacts from coarse materials" (79 FR 74843). The EPA makes the meteorological assertion
in spite of the fact that 2011 was one of the hottest and driest years in Texas history and there
were unprecedented wildfires that year. The current visibility conditions in Big Bend,
Guadalupe Mountains, and Wichita Mountains are already better than the respective state-
established and the EPA-proposed RPG for these three Class I areas.
Footnotes:
9 EPA, CENRAP Modeling TSD: Review of Modeling and Emission Inventory Development for the Regional Haze
Implementation Plan for the State of Texas 34 (Nov. 2014) ("CENRAP Modeling TSD").
AECOM Footnotes:
25 Interagency Monitoring of Protected Visual Environments (IMPROVE) is a network of speciated particulate
monitors that obtain 24-hour average particulate concentrations. These concentrations are then converted into an
equivalent visibility metric (extinction) through the "IMPROVE equation" that is documented at
http://vista.cira.colostate.edu/improve/Publications/GrayLit/019_RevisedIMPROVEeq/RevisedIMPROVEAlgorith
m3.doc.
27 79 Fed. Reg. 74,818; 74,843; 74,870 (Dec. 16,2014).
2879 Fed. Reg. 74,843; 74,870.
ACET Footnotes:
18 Although EPA's Proposal suggests that weather (i.e., temperature and precipitation) may have caused the five-
year average of IMPROVE monitoring data to be abnormally low and, thus, be below EPA's proposed 2018
reasonable progress goals for the three Class I areas and be projected to meet their Uniform Rate of Progress for
2018 for those areas, AECT believes that weather did not cause the IMPROVE monitoring data to be abnormally
low. Available information on temperatures and precipitation suggests that other than 201 land 2012, none of the
years were unusually cool or wet, or otherwise conducive to abnormally low levels of haze. In fact, some of the
IMPROVE monitoring data is associated with one or more exceptional events that actually caused abnormally high
levels of haze. For example, an unusually intense series of wildfires in Mexico appear to be responsible for the
abnormally high levels of haze observed at Big Bend in 2011, and a dust storm appears to be responsible for the
abnormally high levels of haze observed at the Guadalupe Mountains in 2012.]
Response: We disagree with the commenters' contention that we ignore recent data from
IMPROVE monitors that shows actual observed visibility conditions at Big Bend, Guadalupe
Mountains, and Wichita Mountains are better than the numerical RPGs we proposed in our FIP
proposal. In our proposal, we acknowledge that the more recent data from IMPROVE monitors
at the three Class I areas indicate that more progress than anticipated by the 2018 CENRAP
880
-------
modeling has occurred and that we believe this observed improvement is the result of
meteorological conditions, reduction in the impact from fires, and reduction in the impacts from
SO2 emissions and coarse matter.689 However, we also determined that TCEQ's analyses do not
fully satisfy the requirements under section 51.308(d)(1) and (3). We discuss our evaluation of
recent meteorological conditions and visibility conditions in a separate response to comment
where we address modeling issues and back trajectory analysis.
We disagree with the comment that we should withdraw our FIP and approve the Texas regional
haze SIP because the Texas SIP is working and that this is evidenced by the decrease in Texas
emissions and the improvement in visibility beyond what we are requiring in our FIP. We do not
believe that the decrease in emissions from Texas sources and the fact that recent data from
IMPROVE monitors at the three Class I areas indicates visibility conditions are better than the
numeric RPGs we proposed, warrants full approval of the Texas RH SIP and withdrawal of our
FIP. Commenters making this contention have an incorrect understanding of the regional haze
requirements. We identified those sources that have the greatest visibility impacts at the Class I
areas of interest and through our analyses of the four factors, we found that cost-effective
controls that result in meaningful visibility improvement are available at some of these sources.
We also note that emissions at some of these sources that impact visibility are still above the
emission levels projected in the 2018 CENRAP modeling. The determination of whether an
RPG and the emission limitations and other control measures upon which it is based constitute
reasonable progress is made by conducting certain analyses and meeting the requirements under
section 51.308(d)(1). Additionally, section 51.308(d)(3)(i) requires that Texas consult with other
states if its emissions are reasonably anticipated to contribute to visibility impairment at that
state's Class I area(s).
As discussed in more detail in our proposal and in our final action, Texas did not fully satisfy the
requirements under section 51.308(d)(l)(i) and (ii) in determining whether additional controls for
Texas sources are reasonable and in establishing its RPGs. Therefore, we proposed to
disapprove those portions of the Texas regional haze SIP. Specifically, we disagreed with the set
of potential controls identified by the TCEQ and how it analyzed and weighed the four
reasonable progress factors in a number of key areas.690 In addition, we proposed to find that
Texas did not adequately address the requirement in section 51.308(d)(3)(i) to consult with
Oklahoma in order to develop coordinated emission management strategies to address its impacts
on the Wichita Mountains.691 Please see our proposal, our responses to other comments in this
document, and the TSDs for a detailed discussion of this. The decrease in emissions from Texas
sources along with recent data from IMPROVE monitors indicating that visibility conditions are
better than the levels anticipated in the CENRAP projections for 2018 and better than the RPGs
we proposed (which were based on adjusting the CENRAP 2018 projection to account for the
visibility benefit anticipated from controls required in this FIP), do not change our finding that
Texas did not fully satisfy the regional haze requirements under section 51.308(d)(1). The
required controls will result in a significant visibility improvement towards the goal of natural
visibility conditions for this planning period. Through our own analysis of the four reasonable
progress factors, we found that controlling a small number of Texas sources will result in
689 79 FR 74843 and 74870
690 79 FR 74838
691 79 FR 74856
881
-------
significant visibility benefits at the affected Class I areas, and that rather than evaluating controls
at all facilities identified by Texas combined, a subset of those facilities (and some additional
facilities not identified by Texas) is reasonable. In addition, we found that TCEQ did not
consider in its analysis the reasonableness of control upgrades or increased utilization of existing
controls. Through our own analysis, we found that there are a number of Texas sources with
older SO2 scrubbers that can upgraded and would result in cost-effective visibility benefits. We
continue to hold that the emission limitations we are establishing for sources in Texas constitute
reasonable progress and that we arrived at these emission limitations in accordance with the
CAA and the regional haze regulations.
We estimated the revised numerical RPGs for the 20% worst days in 2018 for Texas and
Oklahoma Class I areas to reflect the additional anticipated visibility benefit from the controls on
Texas sources in our FIP that will be in place by 20 1 8.692 We also assessed the additional benefit
from required controls that will not be in place by 2018 but did not include that benefit in the
proposed RPGs. While we recognize that recent IMPROVE monitoring data appear to indicate
that the three affected Class I areas are meeting the RPGs selected by the states as well as the
more stringent RPGs we have established in our FIP, we disagree that this means that we should
approve the Texas Regional Haze SIP. The SO2 emission limitations we are requiring in our FIP
for specific Texas sources are not required under the Texas RH SIP and result in significant
additional visibility benefit by achieving large reductions in pollutants at these sources. These
emission limitations along with the "on the book" reductions are what we have determined will
constitute reasonable progress at the affected Class I areas based on our analyses under section
51.308(d)(1). The RPGs we calculated in our proposal are intended to reflect the additional
visibility benefit that would result based on the emission limitations that would be in place by
2018 estimated by adjusting the CENRAP 2018 projected visibility conditions at the Class I
areas. We explained in our proposal why we cannot assume that the SO2 scrubber retrofits we
are requiring for certain sources will be installed and operational within this planning period,
which ends in 2018.693 We did not conduct photochemical modeling to quantify the revised
RPGs, but instead we adjusted the 2018 RPGs established by Texas and Oklahoma that were
based on the 2018 CENRAP modeling. As a result, the RPGs we quantified are not as refined as
they would be had we conducted additional photochemical modeling to estimate them and are
still based on the CENRAP 2018 projections. The estimated RPGs serve to demonstrate the
additional visibility benefit anticipated by the required controls anticipated to be in place in 2018
as compared to the RPGs established by Texas in the TX RH SIP. This helps explain why Texas
and Oklahoma Class I areas appear to already be meeting the revised RPGs we proposed.
However, we do not believe our method for calculating the revised RPGs presents a serious issue
because the RPGs themselves are not directly enforceable under the RH rule.694 The RPGs are an
analytical tool the state and we use to evaluate whether the measures in the implementation plan
are sufficient to achieve reasonable progress.695 Instead, what is enforceable under the RH rule
are the emission limitations and other control measures that apply to specific sources, and upon
which the RPGs are based. Since the emission limitations we are requiring in our FIP for
specific Texas sources (which is what our revised RPGs are based upon) are not included in the
692 79 FR 74886
693 79 FR 74874 and 74886
694 40 CFR 51.308(d)(l)(v)
695 See 51.308(d)(l)(v).
882
-------
Texas RH SIP and are not currently being achieved, we disagree that visibility at the Class I
areas has already improved beyond what we would require in our FIP and that our FIP is
therefore unjustified and unwarranted. The emission reductions required in this action will result
in significant visibility improvements at the Class I areas beyond what is currently being
achieved or observed.
With regard to comments that linear regression of the 2009-2013 observations from IMPROVE
monitors indicates the Wichita Mountains and Guadalupe Mountains are achieving a rate of
improvement greater than the URP for 2018, we discussed in our proposal that based on
information provided by the TCEQ we do not expect large additional emission reductions of SO2
in Texas between 2013 and 2018 under federal programs and the SIP as submitted.696 We have
not seen evidence in support of something different. Therefore, it is highly questionable whether
the rate of visibility improvement observed up until 2013 at the Class I areas will continue
through 2018 and whether the Class I areas will actually meet the URP for 2018. Furthermore,
as explained in more detail elsewhere in this document, meteorological conditions over the past
few years have deviated from the typical conditions and transport patterns. This helps explain
one factor why Texas and Oklahoma Class I areas appear to already be meeting the revised
RPGs we proposed and why we do not anticipate the rate of visibility improvement will
continue.
We disagree that the controls we are requiring in our FIP are projected to result in negligible
visibility benefit, without a perceptible effect on visibility conditions in any Class I areas
compared to the SIPs submitted by Texas and Oklahoma. We address elsewhere in this
document comments contending that our FIP provides negligible or no perceptible visibility
benefit.
With regard to the comment that we refused to account for CSAPR and MATS limitations in our
modeling analysis, as we discuss in depth in the FIP TSD and elsewhere in this response to
comments document, we have no reason to believe that significant additional reductions will
occur due to these programs. The TCEQ has provided extensive comments on recent emission
inventory indicating that further significant reductions in SO2 were not expected due to CSAPR
or MATS.697 In fact, Texas has used recent actual emissions in estimating emissions from EGUs
for future years (including 2018) in ozone attainment demonstration SIPs in DFW and HGB.698
We note that, as discussed in the FIP TSD, information available also indicates that SO2 credits
under the CSAPR program are much cheaper than originally projected and that Texas was
696 79 FR 74870
697 TCEQ comment letter to EPA on draft modeling platform dated June 24, 2014. '2018 EMP signed.pdf
698 HGB 1997 8-Hour Ozone standard attainment demonstration approved by EPA in 2013, see TSD materials for
2010 "Appendix B Emission Modeling for the HGB Attainment Demonstration SIP Revision for the 1997 Eight-
Hour Ozone Standard" on page B-78, "09017SIP_ado_Appendix_B.pdf'.; DFW 1997 8-Hour Ozone standard
attainment demonstration submitted to EPA, see TSD Appendix B: Emission Modeling for the DFW Attainment
Demonstration SIP Revision for the 1997 Eight-Hour Ozone Standard, Page B-39, "AppB_EI_ado.pdf'; DFW 2008
8-Hour Ozone standard attainment demonstration proposed for adoption Dec. 10, 2014 and posted October 2014,
see TSD materials "Appendix B Emissions Modeling for the Dallas-Fort Worth Attainment Demonstration State
Implementation Plan Revision for the 2008 Eight-Hour Ozone Standard" Starting Page 40.,DFWAD_SIP_Appendix
B.pdf
883
-------
originally projected to be a purchaser of credits rather than implementing significant controls.
Furthermore, none of the facilities required to install controls by this action have provided any
information on anticipated reductions at their facilities due to compliance with these programs.
We also note that we did consider recent actual emissions in our analysis and adjusted the 2018
emission inventory utilized by ENVIRON to account for reductions at sources when controls had
been installed and relied on post-control actuals to support modeled emission rates.699 Finally, as
discussed elsewhere, there is uncertainty in CSAPR budgets due to the recent CSAPR remand.
With regard to the comment that we dismissed the progress being made at the three Class I areas
as being due to meteorological conditions, reduction in the impacts from SO2 emissions, and a
reduction in the impacts from coarse materials without presenting evidence and in spite of the
fact that 2011 was one of the hottest and driest years in Texas history with unprecedented
wildfires and additional information provided by commenters on recent conditions and visibility,
we discuss our evaluation of recent meteorological conditions and visibility conditions in a
separate response to comment where we address modeling issues and back trajectory analysis.
As explained in more detail in that section, meteorological conditions over the past few years
(2011-2013) have deviated from the typical conditions and transport patterns.
Considering the above, we disagree that our FIP is unnecessary, arbitrary, capricious, and outside
our authority and we are finalizing our partial disapproval of the Texas and Oklahoma regional
haze SIPs and finalizing our Texas RP and LTS FIP and the Oklahoma RPG FIP.
Comment: EPA has no authority to disapprove Texas's SIP because all three Class I areas
will meet EPA's RPGs and URPs by 2018 without further reductions
[CCP (0075) p. 12-13] CCP stated that Texas's conclusion that additional controls are not
needed to meet the State's RPGs is further supported by the attached materials (AECOM
attachments to 0075) that show that the States' established RPGs have already been attained.
Current conditions based on more recent actual monitoring data account for the substantial
reduction in SO2 emissions. Table 3 provided by CCP in comment 0075 shows the URPs and
RPGs proposed by the States and EPA and the negligible difference between the State-proposed
RPGs and the proposed RPGs. Most importantly, the exhibit shows that TCEQ estimates of the
impact of current controls on emissions are reasonable and that no additional controls are needed
to meet both the TCEQ and EPA-proposed RPGs. Texas's insignificance determination should
be afforded deference by EPA. Conversely, EPA's proposal to require $2 billion in additional
controls when RPGs are already being met is unreasonable. EPA acknowledges the
improvement in current conditions but arbitrarily determined not to take this information into
account in establishing its revised goals and emission reductions for the 14 Texas sources.
URPs and RPGs Established by the States and Proposed by EPA and Current Conditions
(Table 3 provided by CCP in comment 0075)
699 Table 2 of ENVIRON 2018 Memorandum, Sept 16, 2013. Available in the docket for this action as TX166-010-
09 Memo_TXHAZE_2018CAMx 16Septl3
884
-------
Class I Area
State
Established
LRP (2018)
(20% worst
days)
Current
Conditions
State
Established
RPC (2018)
(20% worst
days)
EPA
Proposed
URP (2018)
(20% worst
days)
EPA
Proposed
RPC (2018)
(20% worst
days)
Difference
Between
State RPC
and EPA
Proposed
RPC
Big Bend
15.6
16.3
16.60
14.93
16.57
0.03
Guadalupe
Mountains
16.0
15.3
16.30
14.73
16.26
0.04
Wichita
Mountains
20.01
21.2
21.47
20.01
(approved)
21.33
0.14
[Luminant (0061) p. 92] Luminant stated that EPA has no authority to disapprove Texas's
regional haze SIP or to issue its proposed FIP because current visibility monitoring at the three
Class I areas demonstrates that all three areas already meet EPA's reasonable progress goals for
2018 and will even meet the uniform rate of progress (URP) by 2018, without the additional
controls that EPA would impose. Under EPA's regional haze regulations, Texas must only
demonstrate the reasonableness of its rate of progress where that rate "provides for a slower rate
of improvement in visibility than the rate that would be needed to attain natural conditions by
2064 [i.e., meet the URP] . . . "62° Here, Texas' SIP revision is achieving a rate of improvement
faster than the URP, and, indeed, visibility conditions in the three Class I areas already meet
EPA's own reasonable progress goals for 2018.
[AECOM (0061/0075) p. 2-1] AECOM noted that the EPA asserts in its proposal that the
projected rate of progress at the three Class I areas under discussion (WIMO, GUMO, and BIBE)
will fall short of the 2018 URP goals based upon various modeling projections.29 For example, in
its FIP Technical Support Document, EPA relies on modeling it commissioned from ENVIRON
that predicts the following 2018 base case visibility conditions at the three areas with current
controls and regulatory requirements in place: WIMO (21.61 dv); BIBE (16.80 dv); and GUMO
(16.36 dv).30 All of these modeled values are above the EPA-proposed URPs and RPGs
discussed above. In contrast, in the case of Caney Creek Wilderness Area in Arkansas, EPA
finds that "[b]ased on the 2018 CENRAP projections, Missouri and Arkansas established RPGs
for their Class I areas that provide for a slightly greater rate of improvement in visibility than
needed to attain the URP, and determined that the projected emission reductions included in the
model were adequate, and that it was not reasonable to request additional controls from Texas at
this time. We find these consultations acceptable."31
[AECOM (0065/0075), p. 2-1] AECOM (0061/0075) stated that recent monitoring data,
however, contradict EPA's modeling results and show WIMO, BIBE, and GUMO already meet
the 2018 EPA-proposed RPGs, and WIMO and GUMO are even projected to meet the URPs as
discussed below. When accounting for natural events and adjustments to natural conditions,
BIBE will be shown to meet the RPG and URP. Thus, there is no basis or need for EPA's
proposed additional emissions reductions from targeted Texas sources.
885
-------
To assess EPA's proposal, AECOM stated that it is critical to review actual haze data reported
by the Interagency Monitoring of Protected Visual Environments (IMPROVE) monitors from
WIMO, GUMO, and BIBE through the latest available year of data36 (2013) to determine the
actual rate of progress toward the URPs and RPG goals for 2018 that EPA is proposing. As EPA
has explained, "Model performance at IMPROVE monitors is of highest importance, because
these monitors are sited to be representative of the visibility conditions impacting each Class I
Area."37 AECOM and Luminant noted, given EPA's substantial delay in its action on Texas' SIP
(which was submitted to EPA in 2009), there has been a significant amount of actual data
collected at the IMPROVE monitors that demonstrate Texas's long-term strategy is working and
will achieve the URPs and RPGs for these three Class I areas. Luminant stated, in other words,
these areas are years ahead of the rate of progress that EPA itself has said is reasonable.
AECOM noted, to calculate current visibility conditions, EPA's regulations provide, "The period
for calculating current visibility conditions is the most recent five year period preceding the
required date of the implementation plan submittal for which data are available." AECOM
0061/0075 Table 2-140'41'42 shows the most recent 5-year average data of the 20% worst visibility
days from 2009-2013 for the three Class I areas in comparison to the EPA proposed RPGs for
2018. For these Class I areas, the recent 5-year average observations already show haze
measurements below the EPA-proposed 2018 RPGs. In other words, since the time Texas
submitted its SIP to EPA (2009), visibility in these areas has already improved more than EPA's
proposed target for 2018, without the additional reductions that EPA's proposed FIP would
impose.
According to AECOM and Luminant, Luminant Figures 5 through 7 and AECOM Figures 2-3
through 2-5 further illustrate this monitored data, as compared to EPA's RPGs for each Class I
area. These figures show recent monitored observations in deciviews from EPA's IMPROVE
monitoring network (blue diamonds); the most recent five-year average of these observations
(blue cross); and EPA's RPG for 2018 (purple circle). As these figures show, all three areas are
already below their EPA-proposed RPGs, based on the most recent five-year data, which EPA's
regulations provide as the basis for assessing reasonable progress.624
Actual IMPROVE Observations at WIMO Compared to EPA Proposed RPG
(Luminant (0061) Figure 5, AECOM (0061/0075) Figure 2-3)
886
-------
Reasonable Progress Goals for WIMO
30 -r
£ 15
u
V
a
10 —
~ Observations 2002-13
X 5-Year Obs. 2002-04, Obs. 2010-13
5 :
• EPA-proposed RPG
Linear (EPA-proposed RPG)
o H 1 1 1 1 1 1 1 1 1 1
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
Actual IMPROVE Observations at GUMO Compared to EPA Proposed RPG
(Luminant (0061) Figure 6, AECOM (0061/0075) Figure 2-4)
Reasonable Progress Goals for GUMO
20
18
16
14
5 12
41
:> io
u
dl
o 8
6
4
2
* ~ 17.2
< >
~
Vf ~ *
* r
16.3
* * * .
~
~
Observations 2000-13
X
5-Year Obs. 2000-2004, Obs. 2009-13
•
EPA-proposed RPG
¦Linear (EPA-proposed RPG)
1(11111
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
Actual IMPROVE Observations at BIBE Compared to EPA Proposed RPG
887
-------
(Luminant (0061) Figure 7, AECOM (0061/0075) Figure 2-5)
20
18
16
14
0
> 10
U
1
o *
Reasonable Progress Goals for BIBE
17.3
X-
~
~i5;3"
Observations 2001-13
X 5-Year Obs. 2001-04, 2009-13
EPA-proposed RPG
• Linear (EPA-proposed RPG)
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
2020
Luminant stated this substantial progress has been made, EPA concedes, based on "reduction in
the impacts from SO2 emissions."625 And EPA further concedes that these conditions will
continue, and not worsen, through 2018 based on emission reductions currently in place.626 There
is thus no reason or legal basis for requiring any further controls in the first planning period. Not
only have all three areas already met EPA's RPG, all three areas are projected to meet their URP
in 2018 based on the existing emission limitations, compliance schedules, and other measures
that Texas included in its SIP in 2009—without the additional SO2 controls proposed in EPA's
FIP and that EPA claims Texas and Oklahoma should have considered.
Luminant Figure 8 and AECOM Figure 2-6 shows the URP glide path for Wichita Mountains
(already approved by EPA627) as compared to recent IMPROVE monitoring data. With current
emission limitations in place, Wichita Mountains will meet the URP in 2018, and thus Luminant
concluded that there is no basis for disapproval of the Texas SIP or EPA's proposed FIP.
AECOM Figures 2-6 through 2-8 show the same IMPROVE data as compared to EPA's
approved URP for WIMO and its proposed URPs for BIBE and GUMO. With respect to the
URP and its associated glide path, WIMO and GUMO indicate they are currently measuring
conditions below the uniform glide path (green line). EPA recognizes this progress is greater
than anticipated by the CENRAP modeling performed in support of the regional haze SIP.43
BIBE does not indicate it is currently measuring below the uniform glide path; however, BIBE's
default natural conditions do not wholly represent the real impact of natural causes of hazy
conditions. Further, given the projected emission trends shown in AECOM Figure 2-2, a linear
regression44 of the annual IMPROVE observations (dashed blue line) can be used to show the
actual and projected rate of improvement for the 20% worst visibility conditions. Projecting this
observation-based regression to 2018, the data clearly show that both WIMO and GUMO are on
888
-------
track to meet the 2018 URP, while BIBE is within 0.5 deciview of making this target even
without the further refinements to natural conditions.
EPA Approved URP Glide Path for WIIYIO to 2018 and a Linear Regression Based on
IMPROVE Observations
(Luminant (0061) Figure 8, AECOM (0061/0075) Figure 2-6)
30
25
20
> 15
10
0
Uniform Rate of Progress for WIMO
23.8 ~
. . ~ 21.2
~ > 1 - y
20.0
* 19.1
•
EPA URP Glide Path, 20% Worst
*
X
Observations 2002-13
5-Year Obs. 2002-04, Obs. 2010-13
Linear (EPA URP Glide Path, 20% Worst)
Linear (Observations 2002-13)
i i i i i i i
i i i
2000 2002 2004 2006 2008 2010 2012 2014
2016 2018
2020
EPA Proposed URP Glide Path for GUMO to 2018 and a Linear Regression Based on
IMPROVE Observations (AECOM (0061/0075) Figure 2-7)
889
-------
Uniform Rate of Progress for GUMO
* ~ 17.2
< r — —
~
4 *
* * * ~ * < -
¦— . 14.7
i4.4
•
EPA U RP Glide Path, 20* Worst
~
Observations 200013
X
SYear Obi. 2000-2004. Obs. 2009-13
Linear (EPA URP Glide Path, 20% Worst)
- - -
Linear (Observations 2000-13)
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
EPA Proposed URP Glide Path for BIBE to 2018 and a Linear Regression Based on
IMPROVE Observations (AECOM (0061/0075) Figure 2-8)
Uniform Rate of Progress for BIBE
~ 17.3
f *
v tf f ~ 16.3
15.4
* 41 ~ -
| 14.9
•
EPA URP Glide Path, 20% Wot st
•
Ob servatloos 2001-13
X
5-Year Obs. 2001-04, 2009-13
Linear (EPA URP Glide Path, 20% Worst)
- - -
Linear [Ob&eivations 200113)
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
AECOM stated that the BIBE linear regression shows a reduction in haze by 2018 to below
EPA's proposed RPG, but the regression's projected 2018 end point misses the URP. This
appears to be the result of outlier observations in 2011 and not related to SO2 emissions from
890
-------
Texas point sources. A close examination of the BIBE observations in 2011 indicates that the
area was heavily influenced by natural conditions from windblown dust and wildfires. Texas
experienced one of its worst single-year droughts on record where, during May 2011 through
April 2012, the BIBE and GUMO Class I areas were designated as the most severe drought
category of "exceptional drought."45 Because 2011 observations at BIBE were largely influenced
by natural conditions, especially the highest visibility measurements of the year, and were left
unaccounted for based on the default methodology of calculating natural conditions estimates, an
outlier analysis of the 2011 annual average IMPROVE observation to recalculate the linear
regression is warranted.
Luminant stated that recent IMPROVE data demonstrate the same result for Guadalupe
Mountains and, with some adjustment for outliers,628 the same result for Big Bend—even
compared to EPA's proposed more stringent URPs that use default values for coarse mass and
soil (which is not a reasonable assumption by EPA).629 Using the natural conditions as calculated
by Texas, or even the refined approach calculated by AECOM, the result is even more
apparent—both Texas Class I areas are on track to meet the URP under the SIP proposed by
Texas.630
[NRG (0078) p. 3] NRG stated that EPA's own monitoring data show that haze levels in the
Class I areas of concern are at or below even EPA's proposed lower 2064 glide path. NRG stated
that, as explained in the attached report by Alpine Geophysics (0078), in a comparison of the air
quality monitoring data at the Class I areas in question to the Uniform Rate of Progress
calculated by EPA in the current proposal, almost all of the observed data are below the Uniform
Rate of Progress. To the extent that the observed monitoring data slightly exceed the Uniform
Rate of Progress for a particular year, the exceedance is associated with exceptional events that
are not representative of typical air quality. These same observational data are also below EPA's
projections for 2018 visibility impairment. NRG stated that these data show that further controls
are not needed to protect visibility in the Class I areas at issue in this action.
NRG stated that, to the extent that the observed monitoring data slightly exceed the Uniform
Rate of Progress for a particular year, the exceedance is associated with exceptional events that
are not representative of typical air quality. For example, an unusually intense series of wildfires
in Mexico appear to be responsible for the relatively high levels of haze observed at Big Bend in
2011, and a dust storm appears to be responsible for the haze observed at the Guadalupe
Mountains in 2012. Alpine Report at 8, 11-22.
[Alpine (0078) p.5, 7] Alpine stated that the EPA's proposal appears to ignore actual data
showing that the Big Bend, Guadalupe Mountains, and Wichita Mountains Class I areas are on a
glide slope to attaining natural visibility conditions by 2064.
Alpine stated, using both the most recent five-year (2009-2013) average conditions for the 20%
worst days and individual annual conditions for these same days, monitoring data indicates that
Big Bend and Wichita Mountains are currently observed to be below the EPA calculated uniform
rate of progress line and Guadalupe Mountains has recently maintained observations below EPA
modeling predictions. This further demonstrates that Texas is showing reasonable progress
towards meeting the national visibility goals at these Class I areas.
891
-------
It is further recognized that during the two recent highest years' of haze index observations
(2011 for Big Bend, Wichita Mountains and Guadalupe Mountains and 2013 for Guadalupe
Mountains), these monitors also had significantly high 'exceptional' event influences represented
with a high wildfire year reported in the southern U.S. and northern Mexico domains in 2011 and
episodic high dust storms reported during poor visibility days near Guadalupe Mountains in
2013, both contributing to the increased haze index observations, likely more than meteorology.
Alpine provided Figures 2 through 4 in Alpine comment 0078, showing these impacted W20%
observation calculations are highlighted as red dots in 2011 and 2013.
Observed and predicted W20% haze index values for Big Bend National Park.
(Figure 2 of Alpine (0078))
Uniform Rate of Progress and 2018 Projected Progress
Big Bend National Park
EPA Glide path (W20%) * EPA Modeled Prediction (W20%)
~ Observation (W20%) ~Texas G lide Path (W20%)
Observed and predicted W20% haze index values for Guadalupe Mountains.
(Figure 3 of Alpine (0078))
892
-------
Uniform Rate of Progress and 2018 Projected Progress
Guadalupe Mountains
EPA Gfide path (W20%) —EPA Modeled Prediction (W20%)
• Observation (W20%) -Texas. Glide Path(W20%)
Observed and predicted YV20% haze index values for Wichita Mountains Wilderness area.
(Figure 4 of Alpine (0078))
30.00
25.00
•y 20.00
x
01
¦a 15.00
o>
5 lo.oo
5.00
0.00
Uniform Rate of Progress and 2018 Projected Progress
Wichita Mountains Wilderness
n.&
21.61
23.81*
20.01
1730
14.31
11.87
9.16
7.53
^ ^ ^ ^ ^ ^ ^ # # J? #
EPA Glide path (W20%) » EPA Modeled Prediction (W?.0%) • Observation (W2Q%)
Shown in Table 1 and Figures 2 through 4 of comment 0078, recent observations (black dots in
Figures 1, 2, and 3) from these Class I areas indicate significant progress during the W20% days
893
-------
has been made and at levels significantly more pronounced than EPA's modeling predicts
(orange slope in Figures 1, 2, and 3).
Alpine noted, by EPA's own admission,
(w)e believe that this observed improvement from the baseline conditions is the
result of meteorological conditions, reduction in the impacts from SO2 emissions,
and a reduction in the impacts from coarse material. More recent emission
inventory data shows reductions in emissions in most states beyond what was
projected in the 2018 modeling, including large reductions in emissions from the
Eastern United States. Emissions from non-EGU Texas point sources are lower
than have been projected in the modeling. (79 FR 74843).
(w)e also note the more recent IMPROVE monitored data at the Big Bend and
Guadalupe Mountains indicate that more progress than anticipated by the
CENRAP modeling has occurred. (79 FR 74843)
Annual glide path and observed haze index (dv) at Big Bend (BIBE), Guadalupe
Mountains (GUMO), and Wichita Mountains Wilderness (WIMO) Class I areas for worst
20% visibility days. Highlighted values indicate exceptional event influence. (Table 1 of
Alpine (0078))
¦ • ¦ ¦ ¦ ¦ ¦ wI,:
C-^IA
(H
2«M
zoos
JU"U
2007
¦¦pun
2003
mm
IBti
ni2
201J
b:k
Observed
17.»
JfiSI
17,13
till?
16.96
16.87
IMS
I'm
16.62
16.10
i§„§§
I4.SS
... .1'
16.12
mat
IS. 95
15.70
15.*
15.29
gumo
CiiMe Path
11. If
1537
17.01
17.46
1
15.27
I' '»¦
16.75
16.31
1407
> 1 I
13,70
15.96
lb 33
15.12
15.61
17.25
WMMO
OJUe Putt
24,23
23, SI
23.27
2l«
22.72
22.45
n
21J1
21.17
'W if
•
2155
mm
21.89
2; ;.J
mw
*14
Furthermore, for Guadalupe Mountains and Wichita Mountains, observed condition trends for
the worst 20% visibility days fall below the EPA calculated uniform rate of progress line through
2013, not just the predicted and modeled visibility values.
It is recognized that Big Bend National Park, which while demonstrating an observational trend
lower than EPA's CAMx predicted values, exhibits a trend higher than EPA's calculated uniform
rate of progress line for the Class I area. This measured condition trend is considerably impacted
by the 2011 W20% haze index value and can be linked to significant smoke events resulting
from the record wildfire season in 2011 in the southwestern US and Mexico.
(http://www.ncdc.noaa.gov/sotc/fire/2011/13).
Alpine stated that when reviewing the observed visibility at Big Bend National Park and plotting
the W20% days, it is noted that many of these days have statistically high standard deviations
compared to other recent years of observations. Alpine provided a figure showing B20% and
W20% days for Big Bend National Park in 2011. (Figure 5 of Alpine comment 0078, not
894
-------
reproduced here). A closer review of the dates of these uncharacteristic observations tie many of
the W20% days to wildfire impact events during the 2011 calendar year.
Alpine explained that fires across the southern U.S. resulted in 2011 having the third most active
wildfire season since 2000 with respect to acres burned and sixth least active in terms of number
of fires, indicating a historically greater number of acres burned per fire than the twelve year
average. Texas had the most acres burned of any state during the year, with over 3.7 million
acres burned across the state during 2011, 43 percent of the national total and a State record in
acres burned since the Texas Forest Service started keeping records in 1985. During the three
month period of Mar-May in 2011, 20,100 fires burned over 3.2 million acres across the U.S.,
mostly across Texas, Arizona, and New Mexico. The acres burned were record high for the 3-
month period, surpassing the spring of 2008 when 1.5 million acres burned nationwide.
To demonstrate the impact that these events have on the observed conditions at each monitor,
Alpine presented Table 2 of Alpine comment 0078 with the individual days selected at Big Bend
in 2011 used to represent the W20% for the year. Highlighted dates and visibility observations
indicate those days found to be impacted by wildfire events using satellite imagery and
interpretation. It should be noted that this type of review is also applicable to the Guadalupe
Mountains and Wichita Mountains Wilderness Class I areas.
W20% days at Big Bend National Park in 2011. Highlighted values indicate days
with exceptional event influence. (Table 2 of Alpine (0078))
Date
Visibility (dv)
Date
Visibility (dv)
20-Feb-ll
18.63
12-May-l 1
17.05
16-Mar-l 1
16.05
27-May-l 1
18.02
19-Mar-l 1
17.60
30-May-l 1
22.84
2 8-Mar-11
17.25
8-Jun-l 1
23.86
3-Apr-11
17.25
11-Jun-l 1
16.45
6-Apr-l 1
17.47
14-Jun-ll
25.07
12-Apr-l 1
26.59
17-Jun-l 1
18.18
15-Apr-11
18.21
20-Jun-l1
16.21
21-Apr-11
23.26
18-Oct-ll
21.48
27-Apr-l 1
15.66
2-Nov-ll
23.31
30-Apr-l 1
15.66
5-Nov-11
15.66
6-May-l 1
18.81
2-Dec-ll
15.85
Alpine reviewed Satellite Smoke Text Product from the Satellite Services Division of the NOAA
National Environmental Satellite, Data, and Information Service (NESDIS) for each of the
W20% days at Big Bend National Park with fire impacted periods. Alpine provided a table
describing the smoke events (Table 3 of comment 0078, not reproduced here). Alpine also
provided satellite imagery of the domain collected from the NASA earth observatory website,
corroborating the textual descriptions of most of the smoke events, in Figures 6, 7, and 8 of
comment 0078 (not reproduced here).
895
-------
Alpine stated that when these wildfire impact days are removed from the W20% calculation, it
can be seen in Alpine Figure 9 that the observational trend line would fall well below the EPA
uniform rate of progress line.
Alpine stated that in 2013 as documented in other years, dust event impacts have been tied to
daily observations Class I areas in Texas.4 When removing dust impacted days from the W20%
calculations, there are visibility trend improvement results comparable to years when wildfire
impacts are removed, as shown in Alpine Table 4 and Figures 9 and 12 of comment 0078).
Alpine provided a figure showing the B20% and W20% days for Guadalupe Mountains in 2013
(Figure 10 of Alpine comment 0078, not reproduced here).
Alpine reviewed Satellite Smoke Text Product for each of the W20% days at Guadalupe
Mountain for dust impacted periods identified in Alpine Table 4. Further satellite imagery of the
domain was also collected from the NASA earth observatory website, corroborating the textual
descriptions of most of the dust events. Alpine provided example images (Figure 11 of Alpine
comment 0078, not reproduced here).
When these wildfire and dust event impacted days are removed from the W20% calculation and
new W20% calculations are incorporated, the observational trend line falls below both the EPA
and Texas calculated rate of progress lines for each of the Class I areas. The significantly active
wildfire year in 2011 and dust events in 2013 could justify removal of daily readings most
impacted by smoke and dust at each of the Class I areas and as demonstrated would show an
even greater attainment of the uniform rate of progress line.
Wildfire impact adjusted W20% haze index values for Big Bend National Park. (Figure 9
of Alpine (0078))
896
-------
Uniform Rate of Progress and 2018 Projected Progress
Big Bend National Park
20.00
2.00
0.00
/ / Z5 J? / / / / ^ # / / / # # # /
—•— EPA Glide path (W20%) —•— EPA Modeled Prediction (W20%)
• Observation (W20%) -•*--TexasGlide Path (W20%)
Linear (Observation (W20%))
W20% days at Guadalupe Mountains in 2013. Highlighted values indicate days
with exceptional event influence. (Table 4 of Alpine (0078))
Date
Visibility (dv)
Date
Visibility (dv)
9-Feb-13
26.93
12-Jun-13
15.53
24-Feb-13
19.55
18-Jun-13
14.75
5-Mar-13
16.40
24-Jun-13
15.04
I7-Mar-13
16.07
27-Jun-13
14.74
20-Mar-13
18.02
6-Jul-13
17.05
23-Mar-13
22.13
9-Jul-13
14.70
10-Apr-13
17.73
15-Jul-13
20.52
25-Apr-13
15.18
20-Aug-13
16.93
1-May-13
15.48
l-Sep-13
14.52
6-Jun-13
21.59
7-Sep-13
14.99
9-Jun-13
14.48
Dust impact adjusted W20% haze index values for Guadalupe Mountains. (Figure 12 of
Alpine (0078))
897
-------
20.00
18.00
16.00
14.00
12.00
10.00
8.00
6.00
4.00
2.00
0.00
Uniform Rate of Progress and 2018 Projected Progress
Guadalupe Mountains
6.68
EPA Glide path (W20%)
Observation (W2 0%)
Linear (Observation (W20%))
¦ EPA Modeled Prediction (W20%)
-**--TexasGlide Path(W20%)
Footnotes:
4https://www.tceq.texas.gov/assets/public/implementation/air/sip/bart/haze_sip-dust_storms.pdf
Luminant Footnotes:
620 40 C.F.R. § 51.308(d)(l)(ii). EPA's has elsewhere taken the position that states must affirmatively demonstrate
to EPA the reasonableness of their goal even if the URP will be achieved. 77 Fed. Reg. 14.604, 14,622 (Mar. 12,
2012). This interpretation, however, is based solely on EPA's view of prefatory language in the preamble to the
regional haze rule. Id. The regulation, which provides EPA's definitive interpretation of the statute, plainly says
otherwise and controls over EPA's statements in a preamble.
624 40 C.F.R. § 51.308(f)(1).
625 79 Fed. Reg. at 74,843, 74,870.
626 FIP TSD at A-45 ("Overall this information supports looking at recent actual emissions to represent future
emission levels in 2018.").
627 76 Fed Reg. 81,728 (Dec. 28,2011).
628 A full discussion and basis for excluding abnormal 2011 data from the Big Bend calculation is provided in
AECOM's report submitted as part of these comments.
629 See AECOM Report at 2-9.
630 See id. at 3-1 to 3-11.
AECOM Footnotes:
29 EPA, Technical Support Document for the Oklahoma and Texas Regional Haze Federal Implementation Plans
(FIP TSD), Nov. 2014.
so pip XSD, at A-20-A-26. ENVIRON's modeling, however, did not account for current emission limits under the
Cross State Air Pollution Rule, which became effective in 2015, but instead relied on prior emission limits under the
Clean Air Interstate Rule, which is no longer in effect.
3179 Fed. Reg. 78,456.
36 http://views.cira.colostate.edu/fed/DataWizard/Default.aspx.
33 CENRAP Modeling TSD at 34.
38 40 C.F.R. § 58.308(f)(1).
898
-------
39 Note: All values are reported in deciviews.
40 79 Fed. Reg. 74,843.
41 79 Fed. Reg. 74,870.
42 79 Fed. Reg. 74,887.
43 79 Fed. Reg. 74,843; 74,870.
44 A linear regression is a statistical method that calculates the best-fitting line for a set of observed data.
45National Public Radio (NPR), cited 2014: Dried Out: Confronting the Texas Drought. Web site:
http ://stateimpact. npr. org/texas/drought/.
Response: These comments are predicated on the assumption that: (1) if a Class I area meets its
URP, or (2) if subsequent monitoring shows a Class I area meets its RPG, it is automatically
relieved of any obligation to address the reasonable progress and long-term strategy requirements
in Sections 51.308(d)(1) and (d)(3). As we discuss below, this assumption is false.
Luminant states that we have no authority to disapprove Texas's regional haze SIP or to issue
our proposed FIP because current visibility monitoring at the three Class I areas demonstrates
that all three areas already meet our reasonable progress goals for 2018 and will even meet the
uniform rate of progress (URP) by 2018, without the additional controls our FIP would impose.
Luminant expresses its belief that under our regional haze regulations, Texas must only
demonstrate the reasonableness of its rate of progress where that rate "provides for a slower rate
of improvement in visibility than the rate that would be needed to attain natural conditions by
2064 [i.e., meet the URP] . . . Luminant acknowledges that we have previously refuted this
belief in our final action on the Arkansas SIP (77 FR 14622) but claims that our regulations in
section 51.308(d)(l)(ii) take precedence. We disagree there is any inconsistency between our
position on this issue in our Arkansas action and our regulations. As we state in the cited final
action:
While EPA agrees that the RHR requires states to consider the uniform rate of
improvement in visibility when formulating RPGs, we disagree that a state's
consideration of the URP and establishment of RPGs that provide for a slightly
greater rate of improvement in visibility than would be needed to attain the URP
is all that is needed to satisfy the RPG requirements in the RHR. EPA also
disagrees that the RHR only requires additional analysis when a state establishes
RPGs that provide for a slower rate of improvement than the URP. As explained
in our proposed rulemaking on the Arkansas RH SIP, in establishing its RPGs, the
State is required by CAA § 169A(g)(l) and 40 CFR 51.308(d)(l)(i)(A) to
"[cjonsider the costs of compliance, the time necessary for compliance, the
energy and non-air quality environmental impacts of compliance, and the
remaining useful life of any potentially affected sources, and include a
demonstration showing how these factors were taken into consideration in
selecting the goal."
The RHR states the following with regard to RPG requirements:
Today's final rule requires the States to determine the rate of
progress for remedying existing impairment that is reasonable,
899
-------
taking into consideration the statutory factors, and informed by
input from all stakeholders.700
An analysis of the four statutory factors is precisely the "further analysis" EPA
refers to in its proposed rulemaking on the Arkansas RH SIP.701 As explained
above, both the RHR and the CAA require states to undertake this analysis in
establishing its RPGs. Therefore, EPA disagrees that our proposed rulemaking on
the Arkansas RH SIP is arbitrary and capricious because it relies on factors, which
Congress has not intended it to consider. CAA section 169A(g)(l) clearly
requires states to consider these four factors in establishing their RPGs.
Accordingly, EPA's proposed disapproval of Arkansas's RPGs is consistent with
the RH regulations and the Act. Because the CAA section 169A(g)(l) and 40
CFR 51.308(d)(l)(i)(A) require that states consider the four statutory factors in
establishing their RPGs, a requirement which Arkansas has not satisfied, our
proposed disapproval of Arkansas's RPGs recognizes the purpose of the RPGs in
improving visibility impairment and is in keeping with the statutory requirements.
With regard to Luminant's claim that "Texas must only demonstrate the reasonableness of its
rate of progress where that rate "provides for a slower rate of improvement in visibility than the
rate that would be needed to attain natural conditions by 2064 [i.e., meet the URP]," we note that
the complete citation to section 51.308(d)(l)(ii) is the following:
For the period of the implementation plan, if the State establishes a reasonable
progress goal that provides for a slower rate of improvement in visibility than
the rate that would be needed to attain natural conditions by 2064, the State
must demonstrate, based on the factors in paragraph (d)(l)(i)(A) of this section,
that the rate of progress for the implementation plan to attain natural conditions
by 2064 is not reasonable; and that the goal adopted by the State is reasonable.
The State must provide to the public for review as part of its implementation
plan an assessment of the number of years it would take to attain natural
conditions if visibility improvement continues at the rate of progress selected by
the State as reasonable.
Section 51.308(d)(l)(ii) confers an additional obligation on states for Class I Areas that are not
meeting their URP. Luminant seems to believe that only states that are above the URP must
comply with section 51.308(d)(l)(i)(A). However, section 51.308(d)(l)(i)(A) itself does not
make such a distinction:
(i) In establishing a reasonable progress goal for any mandatory Class I Federal
area within the State, the State must:
(A) Consider the costs of compliance, the time necessary for compliance, the
energy and non-air quality environmental impacts of compliance, and the
remaining useful life of any potentially affected sources, and include a
700 64 FR 35731
701 76 FR 64195.
900
-------
demonstration showing how these factors were taken into consideration in
selecting the goal [emphasis added].
Clearly, a state's obligation to satisfy section 51.308(d)(l)(i)(A) applies in all cases, without
regard to its position on the URP. We see nothing in Luminant's claim that would cause us to
reconsider our position on these issues. Thus, we disagree with comments contending that given
the recent monitoring data from IMPROVE monitors, we have no authority to disapprove Texas'
regional haze SIP or to issue our FIP. Therefore, even when recent data from IMPROVE
monitors indicates that visibility conditions in the Class I area are better than the established
RPGs and/or that the area may be projected to meet the URP by 2018, Texas is not excused from
addressing the requirements under sections 51.308(d)(1) and 51.308(d)(3)(i) in evaluating
controls for additional sources and in establishing RPGs for its Class I areas.
The revised RPGs we are establishing in our FIP for Texas and Oklahoma Class I areas are based
on the additional improvement over the projected CENRAP 2018 visibility conditions projected
from the emission limitations on specific Texas sources as determined through our analyses
under section 51.308(d)(1). These emission limitations that apply to specific sources and not the
RPGs themselves are what is enforceable under the regional haze regulations. Therefore, we
disagree that the Texas RH SIP satisfies the reasonable progress requirements. We address
elsewhere in this document comments contending that there is a negligible difference between
the RPGs proposed by Texas and Oklahoma and those established by us in our FIP.
Luminant further contends that Texas and Oklahoma Class I areas are also projected to meet the
URPs for 2018 without the need for additional controls under our FIP. As an initial matter, we
question the contention that the Guadalupe Mountains and Wichita Mountains will meet the URP
by 2018 based on linear regression of the 2009-2013 annual observations from IMPROVE
monitors even without the additional controls that we propose in our FIP. We also question the
contention that Big Bend will meet the URP by 2018 when natural events and adjustments to
natural conditions are accounted for. We note that the URP for 2018 as calculated by us in our
proposal is 14.93 dv for Big Bend and 14.73 dv for Guadalupe Mountains, while the current
visibility conditions (2009-2013 average) are 16.30 dv at Big Bend and 15.3 dv at Guadalupe
Mountains. Although there has been greater visibility improvement up until 2013 than projected
in the 2018 CENRAP modeling, we noted in our proposal that based on information provided by
the TCEQ, we do not expect large additional emission reductions of SO2 in Texas between 2013
and 2018 under existing federal programs and the SIP as submitted.702 Therefore, it is
questionable whether the currently observed rate of visibility improvement at the Class I areas
will continue through 2018 and whether the Class I areas will actually meet the URP goal in
2018. Additional emission reductions are needed in order to further reduce visibility impairment
at these Class I areas. Furthermore, as explained in more detail elsewhere in this document, our
analysis of recent meteorological data and visibility conditions reveal that meteorological
conditions over the past few years have deviated from the typical conditions and transport
patterns. This helps explain one factor why Texas and Oklahoma Class I areas appear to already
be meeting the revised RPGs we proposed and why we do not anticipate this rate of visibility
improvement will continue.
702 79 FR 74870
901
-------
Nevertheless, we note that even when a Class I area is projected to meet the URP for 2018, this
is not sufficient reason to approve a state's RPGs when the state has not fully satisfied all the
requirements under section 51.308(d)(1) in establishing its RPGs and under section
51.308(d)(3)(i) to address its impact on other states' Class I areas. The preamble to the Regional
Haze Rule703 states that the URP does not establish a "safe harbor" for the state in setting its
progress goals:
If the State determines that the amount of progress identified through the [URP]
analysis is reasonable based upon the statutory factors, the State should identify
this amount of progress as its reasonable progress goal for the first long-term
strategy, unless it determines that additional progress beyond this amount is also
reasonable. If the State determines that additional progress is reasonable based on
the statutory factors, the State should adopt that amount of progress as its goal for
the first long-term strategy.
For instance, even though Arkansas established RPGs for its Class I areas that provide for a
slightly greater rate of improvement in visibility than needed to attain the URP, we disapproved
Arkansas' RPGs because the State did not satisfy the requirements under section 51.308(d)(l)(i)
and because some of the State's BART determinations did not meet the requirements of the CAA
and the RH rule.704 We proposed a FIP for Arkansas to address this deficiency that includes
controls to meet the BART requirements and additional controls on two units at one facility
under the reasonable progress requirements that will result in additional significant visibility
benefits. Similarly, we found that Texas did not fully satisfy the requirements under section
51.308(d)(1) and 51.308(d)(3)(i) in determining if controls on additional Texas sources would
provide for reasonable progress for the affected Class I areas. Therefore, we are disapproving
Texas' reasonable progress analysis and its RPGs.
We disagree with Luminant's comment alleging that we conceded in the TSD for our proposed
FIP that conditions will continue, and not worsen, through 2018 based on emission reductions
currently in place. The actual statement we made in our TSD was: "Overall this information
supports looking at recent actual emissions to represent future emission levels in 2018" (FIP
TSD at A-45)." That statement is specific to considering recent actual emissions at the EGUs we
analyzed for controls to represent future anticipated emissions at these specific units. In other
words, emission reductions are not anticipated at these specific units between now and 2018 and
so recent actual emissions can be expected to represent the emission levels in 2018 for these
units.
With regard to the comment that exceptional event influences in the form of high wildfires in
2011 in the southern U.S. and northern Mexico and episodic high dust storms near Guadalupe
Mountains in 2013 contributed more so than meteorology to the increased haze index
observations in 2011 for Big Bend, Wichita Mountains, and Guadalupe Mountains and in 2013
for Guadalupe Mountains, we recognize that there is year to year variability in monitored
703 64 FR 35732
704 76 FR 64186
902
-------
visibility conditions and therefore focus on a 5-yr average in assessing current or baseline
visibility conditions.
We disagree with Luminant and AECOM who contend the reason why Big Bend is projected to
miss the URP in 2018 is because the default natural conditions do not wholly represent the real
impact of natural causes of hazy conditions. As Alpine notes in its comments, Texas
experienced one of its worst single-year droughts on record from May 2011 through April 2012
and the area was heavily affected by windblown dust and wildfires in the southwestern US and
Mexico. We also agree with Alpine that if the influence of fire and dust were removed from the
monitoring data for Big Bend, the Guadalupe Mountains, and the Wichita Mountains, these Class
I areas would certainly be monitoring lower. However, as we discuss above, we disagree with
what Alpine seems to be implying that this relieves Texas of its responsibility to properly
analyze the four statutory factors under section 51.308(d)(l)(i)(A). If after analyzing these four
factors additional progress can be made, then the state's RPGs should incorporate that progress.
We also disagree that an outlier analysis of the 2011 annual average IMPROVE observation for
Big Bend to recalculate the linear regression is warranted because 2011 observations at Big Bend
were heavily influenced by natural conditions and this was unaccounted for based on the default
methodology of calculating natural conditions estimates.
We respond to comments concerning natural visibility conditions, and emission reductions from
CSAPR elsewhere in this document.
Comment: Methods for determining outliers in regional haze datasets
[AECOM (0061/0075) p. 2-8]
AECOM explained that methods for determining outliers in regional haze datasets were explored
in an EPA report by Science Applications International Corporation (SAIC) 46 The findings,
included in EPA guidance, define an outlier as a data point greater than two standard deviations
of the mean. The guidance states "the impact from a small number of days tends to average out
when the visibility is examined on a deciview scale over a 5-year period."47 However, on an
annual basis, we find the impacts of outliers to be significant due to the magnitude of the 2011
annual average in comparison to annual averages in recent years. Six outliers were identified
using a two standard deviation outlier cutoff for the 20% worst day measurements in deciviews.
These outliers correspond to the top 6 haziest days in 2011 which measured coarse matter,
organic mass, and soil contributing greater than 55% to the total aerosol extinction. With these
outliers removed and the 2011 20% worst haze days' average recomputed, the resulting
regression shows that BIBE would indeed meet the 2018 URP interim goal (14.8 vs. a goal of
14.9 deciviews) as illustrated in AECOM 0061/0075 Figure 2-9.
EPA Proposed URP Glide Path for BIBE to 2018 and a Linear Regression Based on
IMPROVE Observations Excluding 2011 Outliers
(AECOM (0061/0075) Figure 2-9)
903
-------
20
18
16
14
5
12
.Si
»
10
jj
O
8
6
4
2
0
Uniform Rate of Progressfor BIBE
Excluding 2011 Outliers
~ 17.3 .
«
>¦ -x . ~ ~
~ " t -l—
lb. 3
•
~ " 1 - ¦
14.9
" • 14.8
•
EPA URP Glide Path, 20% Worst
~
Obs. 2001-13 No 2011 Outliers
X
5-Year Obi. 200104, 2009 13
Linear (EPA URP Glide Path, 20% Worst)
Linear (Obs. 2001-13 No 2011 Outiers)
t 1 r
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
AECOM stated that excluding these outliers is further supported by EPA's regulatory treatment
of "exceptional events." EPA has supported the exclusion of certain air quality measurements if
it is known and proven to be associated with "exceptional events" when the measurements in
excess of one or more NAAQS are affected by natural and/or uncontrollable events.48
Exceptional events are not accounted for in regional haze data; however, they have been
proposed as a tool for regional haze natural conditions estimates.49 As such, these events are
important to consider as they relate to the outlier analysis. TCEQ submitted exceptional event
demonstration packages to EPA for seven days in 2011 regarding PM2.5 and PM10 measurements
impacted by high wind (i.e., windblown dust) and wildfire events.50 In addition to exceptional
events, many states add qualifier flags to air quality data when they believe it is influenced by an
event similar to an exceptional event. TCEQ applied the "Fire - Mexico/Central America"
qualifier flag to six additional days in 2011 of which two corresponded to the 2011 BIBE outlier
days.31 The outlier on April 12, 2011, the worst visibility day since 2007, was within three days
of a day with a fire-related qualifier flag (April 15, 2011). April 12th corresponded to a time
when widespread wildfires were burning in west Texas and Mexico.32 Furthermore, all six 2011
outliers were identified as naturally-caused days in the refined natural conditions analysis
provided by AECOM.
AECOM stated that, under EPA guidance, the monitoring data from IMPROVE sites is "used to
establish baseline conditions (for the 2000-2004 period) for each Class I area and to track
progress toward goals established in future SIPs,"53 In the case of WIMO, BIBE, and GUMO,
these data show that all three areas already meet the RPGs proposed by EPA and are on track to
meet the URPs. With SO2 emissions predicted to decline even further, there is thus no basis for
requiring further emission reductions beyond those already in place in order to meet the 2018
goals.
904
-------
Footnotes:
47 http://www.epa.gov/ttn/caaa/tl/memoranda/rh_tpurhr_gd.pdf at 3-14.
48 http://www.law.cornell.edu/cfr/text/40/50.14.
49 http://www.wrapair2.org/pdf/WRAP_NaturalConditionsReview_20130625.pdf at 14.
50 http://www.tceq.state.tx.us/airquality/monops/pm_flags.html.
51 https://ofmext.epa.gov/AQDMRS/aqdmrs.html.
52 http://www.theatlantic.com/photo/201 l/04/texas-wildfires/100050,
http://alg.umbc.edu/usaq/images/HMSFIRE_4_12_2011 .jpg.
53 http://www.epa.gov/ttn/caaa/tl/memoranda/rh_tpurhr_gd.pdf at 1-4.
Response: We disagree with AECOM that we should exclude data of the 20% worst day
measurements for the top 6 haziest days in 2011, which the commenter says are outliers caused
by high wind (i.e., windblown dust) and wildfire events. The guidance the commenter refers to
is our "Guidance for Tracking Progress Under the Regional Haze Rule," which is clear in that,
"[e]ach annual estimate of best and worst days should be based on all valid measured aerosol
concentrations during the calendar year. This includes high concentrations associated with
regional forest fires or other unusual events."705
Our guidance also provides that".. .events which result in apparent outliers in the data and do
have an impact on the regional visibility (e.g., forest fires) should be included in subsequent
trends analysis. The data should be flagged and explained, if possible, but should remain in the
data set."706 Further, our guidance also points out that ".. .the impact from a small number of
days tends to average out when the visibility is examined on a deciview scale over a 5-year
period. It is important to include these extreme concentrations in the estimates for 5-year
baseline and current visibility conditions because the impact from these events may be part of
natural background and is thus reflected in the estimate for the target visibility levels."707 We
remind AECOM that in assessing current visibility conditions at Class I areas, we look at the
average of the most recent 5-year period of IMPROVE monitor data, not at each individual year.
As a result, any extreme concentrations on a given year are expected to average out when
examining the average of the most recent 5-year period. Therefore, we disagree that apparent
outlier data for 2011 should be excluded from the dataset and that excluding these apparent
outliers is supported by EPA's regulatory treatment of "exceptional events" in assessing
compliance with the National Ambient Air Quality Standards (NAAQS).
We address elsewhere in this document comments contending that there is no basis for the
additional emissions reductions under our FIP because the Class I areas already meet the 2018
RPGs and because linear regression of the most recent IMPROVE monitor data indicates the
Class I areas are on track to meet the URPs. As we discuss in depth elsewhere in this document,
we disagree with the comment's assertion that SO2 emissions are predicted to decline further.
Comment: [Luminant (0061) p. 98] According to Luminant, the so-called "uncertainty" of
Texas EGU emission trends that EPA relies on as the basis for its disapproval does not exist and
is an arbitrary basis for EPA's proposal.637 EPA asserts, as the basis for its disapproval, that
705 EPA Guidance for Tracking Progress Under the Regional Haze Rule, September 2003 (at 3-14).
706 Id.
707 Id.
905
-------
"[w]e believe that in performing its control analysis, the TCEQ should have given greater
consideration to the flexibility in the CAIR trading program and the resulting uncertainty in the
projected emissions."638 EPA's "belief," however, is completely unfounded. The undisputed
evidence in the record is that EGU "[ejmissions have trended downward better than or as
predicted in the CENRAP modeling projections."639 Moreover, as Texas reported in its progress
report, actual EGU emissions of SO2 and NOx are below CAIR allowance allocations.640 And,
more fundamentally, there is always some "inherent amount of uncertainty in the assumed
emissions" used in modeling for regional haze, as EPA itself has recognized.641 Yet, as EPA has
also recognized, such "inherent uncertainty" "is not grounds for disapproving [a] SIP" that relies
on such modeling.642
Luminant stated that EPA has before it all of this information about current and expected
visibility conditions, yet chooses to ignore it, relying instead on "uncertainty" that has
indisputably been resolved by the passage of time and the availability of real-world data. This
claimed continued "uncertainty" surrounding CAIR is a phantom of EPA's creation. As of
January 1, 2015, CSAPR (EPA's replacement for CAIR) is in place with more stringent emission
budgets and, effective beginning January 1, 2017, stringent limitations on out-of-state trading
that CAIR did not impose. EPA's paltry explanation of why it does not credit or account for
CSAPR reductions in its analysis—that CSAPR SO2 allocations are "not much different than the
CAIR Cap for Texas"—is wholly unsupported.643 EPA also unreasonably and arbitrarily
discounts and does not consider NOx reductions from CSAPR and SO2 reductions from the
Mercury and Air Toxics Standards (MATS)—all of which will continue the downward trend in
Texas EGU emissions. EPA has compounded its error by failing to take into account the
additional emission reductions from Oklahoma BART sources that EPA and Oklahoma are
requiring so that those sources are BART-compliant like Texas sources. These measures will
lead to further large emission reductions and improvement in visibility at the Wichita Mountains
that EPA fails to consider.
Luminant noted that the simple fact is that these three Class I areas are improving faster than the
URP and EPA's RPGs and are thus on track to attain these goals in 2018 under the SIP revision
as proposed by Texas. EPA must therefore approve the SIP revision, and it certainly has no
authority or reasoned basis to impose any further reductions on Texas sources in a FIP. EPA's
statutory authority stops at the point that emission reductions are no longer necessary to make
progress toward achieving the national goal.644 That point has been reached, as the data clearly
show. And, as the Supreme Court has recently held, EPA certainly has no statutory authority to
require Texas's SIP to impose emission reductions on Texas sources so that visibility at
Oklahoma's Wichita Mountains can improve at a faster rate than the URP or more than EPA has
determined is reasonable.645
Footnotes:
637 79 Fed. Reg. at 74,837.
638 Id. at 74,840.
639 Texas Five-Year Progress Report at 4-10.
640 Id. at 2-11 to 2-12.
641 77 Fed. Reg. at 40,155.
642 Id. (emphasis added) (approving Nebraska regional haze SIP despite uncertainty in the assumed emissions and
discrepancies between the modeling and Nebraska SIP limits).
643 FIP TSD at A-45.
906
-------
644 42 U.S.C. § 7491(b)(2) (authorizing EPA to issue regulations to require states to adopt SIP emission limitations
"as may be necessary to make reasonable progress toward meeting the national goal" (emphasis added)).
645 See EPA v. EMEHomer City Generation, 134 S. Ct. 1584, 1609-10 (2014) ("EPA cannot require a State to
reduce its output of pollution by more than is necessary to achieve attainment in every downwind State or at odds
with the one-percent threshold the Agency has set."); see also 42 U.S.C. § 7410(a)(2)(D)(i)(II) (requiring only that
a SIP contain "adequate provisions . . . prohibiting . . . any source or other type of emissions activity within the State
from emitting any air pollutant in amounts which will— .. . interfere with measures required to be included in the
applicable implementation plan for any other State under part C of this subchapter ... to protect visibility"
(emphasis added)).
[UARG (0065) p. 18-19] As part of their argument that the EPA does not provide any lawful
basis for disapproving the RPGs for Big Bend and the Guadalupe Mountains or the Texas LTS,
UARG stated that the EPA suggests that it must disapprove Texas's SIP because EPA
"believe[s] that in performing its control analysis, the TCEQ should have given greater
consideration to the flexibility in the CAIR trading program and the resulting uncertainty in the
projected emissions." Id. at 74,840. To explain its point, EPA highlights the difference in
Texas's Integrated Planning Model-based estimates of emission reductions due to CAIR at Big
Brown Unit 1 versus Big Brown Unit l's actual emissions in 2012. Id. EPA cannot fault Texas
for being unable to predict the future. The modeling that Texas relied on was the best
information available at the time it submitted the SIP, which it did long before 2012, and EPA
cannot validly judge a SIP unapprovable on the basis of information that became available for
EPA's review only after the SIP was submitted and only because EPA has missed its statutory
deadline for action on the submitted Texas SIP by nearly five years.
Response: With regard to the comment that our explanation of why we did not account for
CSAPR reductions in our analysis is unsupported and that we unreasonably and arbitrarily did
not consider NOx reductions from CSAPR and SO2 reductions from MATS, our FIP TSD708
discusses in great detail that based on Texas' recent comments709 and other information, further
significant reductions in SO2 were not expected due to CSAPR or MATS, even if all related
litigation is resolved. We note that the CENRAP and our modeling includes estimated
reductions from CAIR, and we considered recent emission reductions by adjusting the 2018
modeled emission inventory to account for reductions at sources when controls had been
installed and relied on post-control actuals to support modeled emission rates, including NOx
reductions at some of the EGU sources evaluated in our reasonable progress/LTS analysis.710
Overall, this information supports looking at recent actual emissions to represent future emission
levels in 2018 for the specific EGUs evaluated in our reasonable progress/LST analysis.711
Contrary to the commenter's assertion, the evidence in the record does not suggest that emissions
from the sources we evaluated will be reduced in the near future, reducing their visibility impacts
on the Class I areas of interest. Furthermore, as discussed elsewhere, due to the recent CSAPR
remand there is ongoing uncertainty in the emission budgets. Comments that we failed to take
708 FIP TSD at A-45
709 Texas comments on Draft IPM modeling conducted by EPA for potential national rule making platform provided
on June 26, 2014. In this docket materials as "TCEQ comment letter to EPA on draft modeling platform dated June
24,2014. '2018EMPsigned.pdf.
710 Table 2 of ENVIRON 2018 Memorandum, Sept 16, 2013. Available in the docket for this action as TX166-010-
09 Memo TXHAZF,_2018CAMx 16Septl3
711 See FIP TSD, p. A-16, A-45 (found in the docket for this rulemaking).
907
-------
into account the additional emission reductions from Oklahoma BART sources are addressed
elsewhere.
As part of its arguments that emissions in Texas have trended downward, Luminant states,
"Texas reported in its progress report, actual EGU emissions of SO2 and NOx are below CAIR
allowance allocations." The commenter is incorrect. The data presented by TCEQ in the Texas
5-yr progress report SIP show facility allocations through the Title IV SO2 Trading Program and
a total statewide allowance allocation of 580,000 tpy of SO2. Compliance with CAIR was
established using these same allocations but at increasing ratios. The Title IV SO2 allowances
allocated for 2010-2014 were to be retired for compliance with CAIR at a ratio of two
allowances per ton of emissions. SO2 allowances allocated for 2015, and thereafter, were to be
retired for compliance at a ratio of 2.86 allowances per ton of emissions. The commenter
compares the Title IV SO2 program allowances to facility emissions but does not consider the
required allowance ratio to demonstrate compliance with CAIR or the amount of allowances
some facilities have purchased in order to demonstrate CAIR compliance. We note that the mere
fact that Texas EGUs have purchased sufficient allowances to be below the CAIR cap is only
evidence that these EGUS are in compliance with CAIR. Examining the cumulative statewide
emissions of SO2 for Texas EGUs presented in the Texas 5-yr progress report SIP shows
emissions exceed both the CAIR budgets of 320,946 tons SO2 per year for Phase I and 224, 662
tons of SO2 per year for Phase II.
We disagree with Luminant that the uncertainty in the projected emissions from the CAIR
trading program we pointed to in our proposal does not exist and is an arbitrary basis for our FIP.
Even Texas itself agreed with the uncertainty in visibility projections due to CAIR in its
Regional Haze SIP. As we discussed in our proposal, the TCEQ stated that it requested that key
EGUs in Texas review and comment on the predictions of the IPM model but no EGU made an
enforceable commitment to any particular pollution control strategy, preferring to retain the
flexibility offered by the CAIR program.712 Furthermore, TCEQ stated that because emission
allowances can be purchased by EGUs, visibility improvement may be less or more that that
predicted by the CENRAP's modeling. Despite the recent downward trend in SO2 and NOx
emissions from Texas EGUs, the flexibility to purchase emission allowances from out of state in
the future still remains along with the uncertainty in actual reductions from sources near the
affected Class I areas. Furthermore, we note that we are unaware, nor did any of the facilities
provide information on planned installation or upgrades of controls in the near future that would
result in emission reductions at these EGUs.
With regard to the comment that an inherent amount of uncertainty in the assumed emissions is
not grounds for disapproving a SIP that relies on modeling of such emissions, we disagree that
uncertainty in the projected emissions from CAIR is the basis for our disapproval of the Texas
RH SIP. Although we did note in our proposal that TCEQ should have given greater
consideration to the flexibility in the CAIR trading program and the resulting uncertainty in the
projected emissions, this did not form the sole basis of our partial disapproval of the Texas RH
SIP. As discussed in our proposal, the flaws in Texas' analyses under section 51.308(d)(1) and
in its consultations under section 51.308(d)(3)(i) formed the grounds for our disapproval of
Texas' reasonable progress analysis and RPGs for the Big Bend and Guadalupe Mountains and
712 79 FR 74838
908
-------
LTS for Wichita Mountains. Luminant does not present the quote concerning uncertainty in the
proper context. In stating that "We believe that in performing its control analysis, the TCEQ
should have given greater consideration to the flexibility in the CAIR trading program and the
resulting uncertainty in the projected emissions," we were addressing the fact that TCEQ's cost
analysis was flawed because the consideration of the tons reduced were calculated from a
starting point of the projected CAIR emissions rather than considering a baseline of actual
emissions. We provided an example of this in the Texas TSD (TX TSD at 22):
The TCEQ's cost-effectiveness calculation for post-combustion controls on Big
Brown Unit 1 was based on reducing that projected 2018 SO2 emission level of
23,142 tpy by 90%, resulting in a reduction of 20,828 tpy. This results in a cost
of $32,766,310/yr, or a cost-effectiveness calculation of $l,573/ton. However,
the installation of a scrubber would allow Big Brown flexibility in fuel choice
thus allowing the unit to continue to burn the higher average sulfur fuel it
currently burns, instead of moving to the low sulfur coal predicted by IPM.
Big Brown Unit l's SO2 emissions in 2012 were 32,100 tons. The issue of
scrubber efficiency aside, a reduction of 90% from these actual emission levels
would result in an SO2 reduction in the range of 29,000 to 31,000 tpy. While the
numerator ($) in the cost-effectiveness metric of $/ton will increase slightly
beyond what was estimated by Alpine Geophysics due to an increased sulfur
loading to the scrubber, the denominator (tons) would increase by roughly 50%,
thus improving (lowering) the overall cost-effectiveness of controlling Big Brown
Unit 1 significantly. Estimates for scrubbers at units at Monticello are similarly
impacted by the cost-methodology used by Texas in estimating cost-effectiveness
on a cost-per-ton basis.
This flaw resulted in an overestimate in the cost-effectiveness of controls as calculated by the
TCEQ in terms of $/ton. Therefore, contrary to Luminant's assertion that we are faulting Texas
for being unable to predict the future, we are actually faulting Texas' cost analysis for failing to
consider available emission data for these units at the time the cost-analysis was performed, and
failing to consider that "implementation of reasonable controls under the Regional Haze Rule
would likely not be in addition to anticipated reductions due to CAIR predicted by IPM, but
would replace or complement any controls predicted by IPM."
We address elsewhere in this document why recent improvement in visibility conditions at the
three Class I areas are not sufficient grounds for our approval of the Texas RH SIP.
We disagree that we have no authority or reasoned basis to impose any further reductions on
Texas sources in our FIP. The commenter references the CAA at 42 U.S.C. § 7491(b)(2) in
support of the statement that our authority stops at the point that emission reductions are no
longer necessary to make progress toward achieving the national goal. 42 U.S.C. § 7491(b)(2)
of the CAA directs EPA to promulgate regional haze regulations that"... require each
applicable implementation plan for a State in which any area listed by the Administrator under
subsection (a)(2) of this section is located (or for a State the emissions from which may
reasonably be anticipated to cause or contribute to any impairment of visibility in any such area)
909
-------
to contain such emission limits, schedules of compliance and other measures as may be
necessary to make reasonable progress toward meeting the national goal specified in subsection
(a) of this section..We continue to believe that our partial disapproval of the Texas Regional
Haze SIP and our disapproval of Oklahoma's RPGs and consultations under section 51.308(d)(1)
are in accordance with the CAA and our regional haze regulations. We remind Luminant that
the CAA at 42 U.S.C. §7941(g)(l) provides that"... in determining reasonable progress there
shall be taken into consideration the costs of compliance, the time necessary for compliance, and
the energy and nonair quality environmental impacts of compliance, and the remaining useful
life of any existing source subject to such requirements." Further, our regional haze regulations
at section 51.308(d)(l)(iii) contain the clear directive that "In determining whether the State's
goal for visibility improvement provides for reasonable progress towards natural visibility
conditions, the Administrator will evaluate the demonstrations developed by the State pursuant
to paragraphs (d)(l)(i) and (d)(l)(ii) of this section." As discussed extensively in our proposal,
we found that Texas' analyses do not fully satisfy the requirements under section 51.308(d)(l)(i)
and (ii) and (3), and we therefore determined that Texas' goals for visibility improvement do not
constitute reasonable progress towards natural visibility conditions. This is not changed by the
fact that recent data indicates that under the Texas Regional Haze SIP the Class I areas are
currently meeting the RPGs we are establishing or the fact that linear regression (which assumes
continuing emission reductions) of the most recent IMPROVE monitor data indicates they are
projected to meet the URP for 2018. The fundamental issue is that the analyses Texas conducted
that led it to determine that additional controls are not reasonable in the first planning period for
its two Class I areas and the Wichita Mountains area in Oklahoma and which it used in
determining its RPGs for Big Bend and Guadalupe Mountains do not fully meet the requirements
of the CAA and our regional haze regulations. Therefore, the Texas RH SIP does not
demonstrate that additional emissions reductions are not necessary to achieve reasonable
progress as defined under the CAA and our regional haze regulations. We also discuss
elsewhere in this document that the emission limitations and other control measures the RPGs
are based on, rather than the numerical value of the RPGs, are what is enforceable under the
regional haze regulations. The RPGs are an analytical tool we use to evaluate whether measures
in the implementation plan are sufficient to achieve reasonable progress. Therefore, to satisfy
the regional haze requirements it is not sufficient to meet the RPGs without also complying with
the emission limitations the RPGs are based on.
The comment that under EPA v. EME Homer City Generation, EPA has no statutory authority to
require emission reductions on Texas sources so that visibility at Oklahoma's Wichita Mountains
can improve at a faster rate than the URP or more than the RPGs is based on a fundamental
misunderstanding of the structure of the Regional Haze Rule. The language referenced by
Luminant in EME Homer City applies to the specific requirements in CAA section
110(a)(2)(D)(i) regarding the attainment and maintenance of national ambient air quality
standards (NAAQS). These air quality standards are specified by EPA at levels requisite to
protect the public health and welfare. Once EPA establishes a new NAAQS (or revises an
existing standard), EPA identifies those areas where the concentration of the regulated pollutant
exceeds the NAAQS. States with a nonattainment area are required to submit a plan that will,
among other things, bring the area into attainment with the applicable NAAQS.713 In addition,
SIPs must ensure that "contain adequate provisions . . . prohibiting . . . any source or other type
713 CAA Section 172(c)(1).
910
-------
of emissions activity within the State from emitting any air pollutant in amounts which will . . .
contribute significantly to nonattainment in, or interfere with maintenance by, any other State
with respect to any . . . [NAAQS]." EMEHomer City addresses this specific requirement related
to upwind States' contribution to downwind nonattainment of the NAAQS.
The role of RPGs (and the URP) in the visibility program is fundamentally different from that of
the NAAQS. Neither the RGPs nor the URP are enforceable standards such as the NAAQS. In
finalizing the Regional Haze Rule, we declined to set presumptive visibility targets but rather
concluded that States (or EPA in the context of a FIP) should have the flexibility in determining
reasonable progress goals based on consideration of the statutory factors.714 As we explained in
response to concerns that States would be subject to sanctions for failure to meet the RPGs:
.. .the reasonable progress goal is a goal and not a mandatory standard which must
be achieved by a particular date as is the case with the NAAQS... .the [RPG] itself
is not enforceable.715
We further explain that if the State fails to meet its RPGs, the State could either revise the
strategies in its SIP or revise its RPGs, but in either case the State's decision should be based on
appropriate analyses of the factors identified in 308(d)(1)(A) and (B). The RPGs set by a State
are accordingly based on the anticipated changes in emissions over the planning period of the
SIP, both from measures adopted specifically to address haze and from other requirements of the
CAA during the planning period. The RPGs, unlike the NAAQS, are not independent,
enforceable standards. The URP is yet another step further away from a standard such as the
NAAQS. The URP is an analytical requirement "to ensure that States consider the possibility of
setting an ambitious reasonable progress goal." We do not agree that the requirements in our FIP
will result in improvements in visibility at a faster rate than the URP, but regardless of whether
or not this is the case, there is nothing in the Regional Haze Rule that suggests that a State's
obligations to ensure reasonable progress can be met simply by meeting the URP 716
We discuss in the preceding paragraph and elsewhere in this document that the determination of
what constitutes reasonable progress is not based only on the numerical value of the RPG or
whether the state is on track to meet the URP. The determination of what constitutes reasonable
progress is based on the analyses conducted under section 51.308(d)(1) and under section
51.308(d)(3)(i) when a state impacts Class I areas in other states. We continue to hold that the
emission limitations we established based on our analyses under section 51.308(d)(1) constitute
reasonable progress for Texas and Oklahoma Class I areas. Therefore, we disagree that our FIP
imposes emission reductions on Texas sources so that visibility at Oklahoma's Wichita
Mountains can improve at a faster rate than the URP or more than what we have determined
would provide reasonable progress.
Comment: Two commenters supported EPA's proposal to not establish any additional
requirements on sources within Oklahoma.
714 64 FR at 36731.
715 64 FR at 35733.(emphasis in the original)
716 See 77 FR 14604, at 14629.
911
-------
[OG&E (0057) p. 3] OG&E stated that analysis by the Oklahoma Department of Environmental
Quality (ODEQ) shows that contributions to visibility impairment in the Wichita Mountains
come not only from in-state emission sources, but also sources upwind in Texas and the eastern
United States. (79 FR 74821) With regard to addressing contributions from in-state sources,
OG&E agreed with EPA's determination that no further reductions are required from emission
sources in Oklahoma in the planning period ending in 2018. (79 FR 74823) Regardless of the
outcome of the Proposal concerning the implementation of the FIP for Texas, OG&E stated that
the EPA should maintain this determination regarding Oklahoma in the final rule.
OG&E stated that in the proposal EPA noted improvements in visibility in the Wichita
Mountains as measured by the visibility monitoring system operated by the U.S. Fish and
Wildlife Service in the Wichita Mountains. Specifically, EPA noted that for the most recent 5-
year period for which data is available (2009-2013), average conditions for the 20% worst days
is 21.2 dv, below the 21.47 dv RPG developed by ODEQ for 2018.10 (79 FR 74870) Indeed,
annual summary data from the IMPROVE program website shows that 2012 and 2013 visibility
is very near the ODEQ URP goal for 2018 of 20.01 dv and better than ODEQ developed for the
2018 RPG. Visibility in 2012 was measured at an annual average of 20.17 dv and 2013 (the
most recent year for which finalized data is available) was measured at 20.15 dv.
OG&E stated that these data favor the EPA's determination that no additional emission
reductions for sources in Oklahoma are warranted in the planning period ending in 2018 and
beyond, especially when combined with the significant emissions reduction measures currently
being implemented.
[EEI (0076) p. 2, 10, 11] EEI noted that EPA proposes to partially disapprove Oklahoma's 2010
SIP regarding the RPG for the Wichita Mountains and replace that part of their plan with the
proposed FIP that relies upon the reductions in Texas to satisfy the 2018 milestone for the
Wichita Mountains. EPA states in the proposal that the 2018 targets for the Wichita Mountains
could not be achieved without emissions reductions from sources in Texas. See 79 Fed. Reg.
74,823. EPA's proposal appropriately "does not establish any additional requirements on sources
within Oklahoma." (79 FR 74818)
EEI supported EPA's proposal to not seek further reductions from Oklahoma facilities given the
recently finalized, litigated and currently in place FIP in Oklahoma being implemented by
Oklahoma utilities. EPA must include the reductions that will occur as a result of this Oklahoma
FIP and compliance with other air quality rules when determining whether additional reductions
are required to achieve the goals of the regional haze program in both Texas and Oklahoma.
EEI stated that the EPA's analysis does not take into account all emission reductions from
Oklahoma sources. Oklahoma sources are currently implementing the final 2012 SIP and FIP
provisions for the state's regional haze obligations and are in the process of retiring, converting
or installing expensive control equipment on affected units through 2019. EPA cannot ignore
these reductions when assessing whether additional reductions are necessary to achieve the goals
of the regional haze program. Given the current Oklahoma SIP and FIP in place for regional
haze and the fact that IMPROVE data show the Wichita Mountains are already attaining the
912
-------
2018 milestones, additional reductions are not required. Further, the Agency cannot ignore the
additional emissions reductions that are occurring within the state as a result of compliance with
other air quality rules, including the Mercury and Air Toxics Standards (MATS), which will
result in the conversion to natural gas or retirement of 60% of Oklahoma's coal-fired generation.
Response: We appreciate the commenters' support of our determination that no further
emissions reductions from Oklahoma sources are necessary in the first planning period.
We address elsewhere in this document where we address comments on modeling issues and
calculation of the RPGs, the comments contending that we must address in our FIP the additional
emissions reductions from sources in Oklahoma as a result of compliance with the Oklahoma
FIP we previously promulgated. We address comments concerning anticipated emission
reductions as a result of compliance with other air quality rules, including MATS elsewhere in
this document.
Comment: Recent IMPROVE Monitoring Data Do Not Support Texas and Industry's
Argument that the Proposed FIP is Unnecessary. [Earthjustice (0067) p.40]
Earthjustice et al., noted that, in their comments at public hearings on the proposed rule, Texas
and industry groups argued that the FIP is unnecessary because the affected Class I areas are
already on track to meet the RPGs that EPA is proposing. To the extent Texas and industry are
arguing that the current visibility conditions meet the RPGs EPA is proposing, that is largely a
result of the fact that EPA has not updated the majority of the 2018 projections that CENRAP
and Texas relied on. Goals based on the controls EPA has proposed and also on more updated
projections would likely be lower than the RPGs EPA is proposing. The recent improvement is
due to a variety of factors, which EPA discusses in the proposed rule, 79 Fed. Reg. at 74,843,
most of which are not enforceable limitations or are beyond the state's control and, therefore,
may be temporary.
According to Earthjustice et al., at best, this argument merely counsels in favor of EPA lowering
the RPGs. It does not show that the proposed controls themselves are unnecessary or
unreasonable. Further, the argument by Texas and industry reflects a misunderstanding of how
RPGs are set. RPGs are set to reflect controls that are reasonable; controls are not required in
order to meet pre-set RPGs. Congress defined reasonable progress as the amount of progress
that could be made after consideration of four factors. 42 U.S.C. § 7491(g)(1). After the four-
factor analysis defines reasonable progress, each haze SIP must include the enforceable measures
necessary to make reasonable progress. Id. § 7491(b)(2). The RPG for 2018 is calculated as the
baseline visibility condition minus the amount of reasonable progress (which is established based
on consideration of the four statutory factors).
Thus, Earthjustice et al., stated that RPGs are not comparable to the NAAQS. Under the
NAAQS, once the goal is set, if the goal is attained, no additional controls are necessary. Under
the haze program, the ultimate goal is set by the statute: elimination of all "man-made" haze. 42
U.S.C. § 7491(a)(1). Reasonable progress goals represent interim goals along the path toward
achieving the statutory goal of eliminating all anthropogenic haze. Even if it were true that the
913
-------
Class I areas already have achieved the RPGs that EPA has set, which is not the case, that would
mean merely that EPA should revise the RPGs to reflect the visibility conditions that will be
achieved once the proposed controls are in place. That is because the reasonable progress goals
are not the equivalent of the NAAQS that, once reached, satisfies a state's legal obligations.
Instead, the reasonable progress goals are merely steps along the way to achieving the statutory
goal of eliminating all man-made haze.
Moreover, under the haze provisions, Earthjustice et al., stated that the amount of reasonable
progress to be made is determined by reference to the uniform rate of progress and the emission
reductions achievable via a four-factor analysis. See 40 C.F.R. § 51.308(d)(l)(i)-(ii). RPGs
simply identify the visibility conditions that should result once the controls selected by the four-
factor analysis are implemented, provided that the state has analyzed the reductions necessary to
achieve the URP, and, if the goals are less stringent than the URP, the state has shown that it is
not reasonable to achieve the URP and its alternative goals are reasonable. See id. Texas's
argument omits the critical role that the four-factor analysis plays in defining reasonable progress
goals. Thus, even if visibility has improved faster than was previously projected, that does not in
any way call into question EPA's analysis showing that consideration of the four factors results
in the conclusion that the proposed controls are reasonable, and the reasonable progress goals
must reflect implementation of those reasonable controls.
Response: We appreciate the commenter's support of our FIP.
Comment: [Commenter 0053-1] The TCEQ stated that it shares the common goal of clear vistas
at the nation's national parks and wilderness areas. The TCEQ's 2009 regional haze SIP revision
moves us toward that goal in accordance with the federal Clean Air Act. It's important to focus
on the goal of visibility at federal Class I areas rather than any other goals that the EPA or others
may have in regards to this federal rule.
The TCEQ stated that Texas determined that current emissions reductions will be adequate for
the first control period of 2008 through 2018. This SIP revision was thoroughly vetted through
consultations with neighboring states, with the presence and input of the EPA, and it meets all
requirements of the federal Clean Air Act and EPA's regional haze rule for this 10-year period.
During the planning, proposal, and review stages between 1999 and mid-2013 the EPA never
stated that the methodology used by TCEQ was not approvable.
According to TCEQ, three primary issues illustrate some of the reasons why EPA should not
finalize the proposed federal implementation plan, which imposes significant costs with no
perceptible benefit. First, the EPA is unjustified in proposing additional controls since the three
Class I areas have already achieved better visibility than the goals EPA proposes to set for 2018.
The TCEQ stated that the most recent actual monitoring data on visibility improvement shows
that the Wichita Mountains, Guadalupe Mountains, and Big Bend Class I areas already have
better visibility than the reasonable progress goals EPA's proposing to set for 2018. This alone
makes EPA disapproval of the Texas SIP inappropriate. Using the EPA's own conventions for
914
-------
visibility, the benefits of EPA's proposed FIP by 2018 are zero at Big Bend and Guadalupe
Mountains and an imperceptible one-tenth of a deciview at Wichita Mountains.
Second, the TCEQ noted that the EPA inappropriately claims that Texas did not evaluate the
reasonable further progress goals four factor analysis correctly. The regional haze rule and the
EPA's guidance give the states flexibility in how to analyze additional controls using the
statutory factors, and TCEQ's analysis was consistent with EPA guidance and the CENRAP
analysis. Consistent with EPA's regional haze rule, Texas is not required to conduct the four
factor analysis for areas not in Texas such as Wichita Mountains.
Third, the TCEQ stated that the period covered by the Texas SIP submittal runs through 2018,
the end of the first required planning period. Some of the controls EPA proposes would not go
into effect until 2020 and are, therefore, inappropriate for consideration and irrelevant to the
approvability of Texas's regional haze SIP submittal.
Given these facts, the TCEQ concluded that the EPA should approve the Texas plan and
withdraw the proposed FIP, because the Texas 2009 regional haze SIP revision meets the
requirements of the federal Clean Air Act and EPA's regional haze rules.
Response: We disagree with the TCEQ that its 2009 Regional Haze SIP revision moves the
affected Class I areas toward that goal in accordance with the CAA. We discussed in our
proposal and elsewhere in this document why the TCEQ's 2009 Regional Haze SIP does not
fully satisfy the regional haze regulations and statutory requirements. The TCEQ implies that we
and other parties are focusing on goals other than visibility at federal Class I areas in
promulgating our FIP. However, it is not clear to what other goals the TCEQ refers. The
purpose of our FIP is to correct the inadequacies in the Texas and Oklahoma Regional Haze SIPs
and ensure that these states have plans in place that fully meet the CAA requirements and
regional haze regulations.
We address the comment asserting that we must disapprove Texas' RPGs despite the agreement
among states that participated in the consultation process that no additional emission reductions
were needed from Texas sources, in the section in this document addressing consultation
comments. Also, as the TCEQ notes, we did participate in some of Texas' consultations, and
provided input at certain stages in the planning and development of the SIP. This included
comments to the TCEQ, in which we were highly critical of its approach to satisfying key
aspects of the Regional Haze Rule, including some of the core inadequacies that led to our FIP.
These comments went largely unheeded. Regardless, we remind the TCEQ that there is no
requirement in the CAA that we must review, evaluate, and comment on a State's proposed SIP
revision. Our job under the CAA is to review a SIP submittal and determine if it meets the CAA
and rules, regardless of whether we commented or not on a State's proposed SIP during its State
rulemaking process.
We address elsewhere in this document comments contending that our FIP provides no
perceptible visibility benefit and that it is not justified and inappropriate because the three Class I
areas have already achieved better visibility than the RPGs EPA proposes to set for 2018. While
we acknowledge that the RH rule and the our guidance give the states a certain amount of
915
-------
flexibility in how to analyze additional controls using the four statutory factors, given the
circumstances of this case, we disagree with the set of potential controls identified by the TCEQ
and how it analyzed and weighed the four reasonable progress factors in a number of key areas.
This is discussed in detail in our proposal and elsewhere in this document.717
We disagree with the TCEQ that Texas was not required to consider and evaluate its own sources
to address its visibility impacts in Class I areas outside the state. In our proposal, we provided
the following explanation regarding this issue:
Finally, each state 'must document the technical basis, including modeling,
monitoring and emissions information, on which the State is relying to
determine its apportionment of emission reduction obligations necessary for
achieving reasonable progress in each mandatory Class I Federal area it
affects.' To reiterate, Section 169A(g)(l) of the CAA requires states to
determine 'reasonable progress' by considering the four statutory factors.
Therefore, this provision requires states to consider both their own Class I
areas and downwind Class I areas when they develop the technical basis
underlying their four-factor analyses. This documentation is necessary so that
the interstate consultation process can proceed on an informed basis and so
that downwind states can properly assess whether any additional upwind
emission reductions are necessary to achieve reasonable progress at their
Class I areas. The regulations further provide that, 'States may meet this
requirement by relying on technical analyses developed by the regional
planning organization and approved by all State participants.' Thus, states
have the option of meeting this requirement by relying on four-factor analyses
and associated technical documentation prepared by a regional planning
organization on behalf of its member states, to the extent that such analyses
and documentation were conducted. In situations where a regional planning
organization's analyses are limited, incomplete or do not adequately assess the
four factors, however, then states must fill in any remaining gaps to meet this
71 &
requirement.
Therefore, Texas is required to consider both its own Class I areas and downwind Class I areas
when it develops the technical basis underlying the four-factor analyses. Texas did not meet this
requirement.
We address issues concerning the installation of controls beyond the first planning period in
another the section of this document.
717 79 FR 74838
718 79 FR 74829
916
------- |