Technical Support Document (TSD)
for the Cross-State Air Pollution Rule Update for the 2008 Ozone NAAQS
Docket ID No. EPA-HQ-OAR-2015-0500

Ozone Transport Policy Analysis
Final Rule TSD

U.S. Environmental Protection Agency
Office of Air and Radiation
August 2016

1


-------
The analysis presented in this document supports the EPA's Final Cross-State Air Pollution
Rule Update for the 2008 Ozone National Ambient Air Quality Standards (CSAPR Update). This TSD
includes analysis to quantify upwind state emissions that significantly contribute to nonattainment or
interfere with maintenance of the 2008 ozone NAAQS in downwind states and quantification of
emission budgets (i.e., limits on emissions). The analysis is described in Section VI of the preamble to
the final rule. This TSD also broadly describes how the EPA used the Integrated Planning Model
(IPM) to inform air quality modeling, budget setting, and policy analysis aspects of this rule. This TSD
is organized as follows:

A.	Background on EPA's Analysis to Quantify Emissions that Significantly Contribute to
Nonattainment or Interfere with Maintenance of the 2008 Ozone NAAQS

B.	Using the Integrated Planning Model (IPM) to Assess Air Quality Modeling, Ozone-Season NOx
Cost Thresholds, and Policy Impacts

C.	Calculating Budgets From IPM Run and Historical Data

D.	Analysis of Air Quality Responses to Emission Changes Using an Ozone Air Quality Assessment
Tool (AQAT)

1.	Introduction: Development of the assessment tool

2.	Details on the construction of the assessment tool

3.	Description of analytical results

4.	Comparison between the air quality assessment tool estimates and CAMx air quality
modeling estimates

2


-------
A. Background on EPA's Analysis to Quantify Emissions that Significantly Contribute to
Nonattainment or Interfere with Maintenance of the 2008 Ozone NAAQS

In the preamble, we describe the four-step CSAPR framework that the EPA is applying to
identify upwind states' emissions that significantly contribute to nonattainment or interfere with
maintenance with respect to the 2008 ozone NAAQS in other states and to implement appropriate
emission reductions. This framework was also used in the original CSAPR rulemaking to address
interstate transport with respect to the 1997 ozone NAAQS and the 1997 and 2006 PM2.5 NAAQS.
See section IV of the preamble for an overview of the CSAPR framework.

The first step of the CSAPR framework uses air quality analysis to identify nonattainment and
maintenance receptors with respect to interstate transport for the 2008 ozone NAAQS. The second step
of the framework uses further air quality analysis to identify upwind states whose ozone pollution
contributions to these monitoring sites meet or exceed a specified threshold amount of 1% of the
NAAQS and therefore merit further analysis. See section V of the preamble for details on applying
these steps with respect to interstate emissions transport for the 2008 ozone NAAQS.

The third step in the framework quantifies upwind state emissions that significantly contribute
to nonattainment or interfere with maintenance of the 2008 ozone NAAQS at the downwind receptors,
and identifies the EGU NOx emission budgets for each state that represent the reduction of these
emissions levels. See section VI of the preamble with respect to interstate emissions transport for the
2008 ozone NAAQS. Finally, the fourth step of the framework implements the emission budgets in
each state via the CSAPR NOx ozone season allowance trading program. See section VII of the
preamble for details on implementation of this CSAPR trading program.

This TSD primarily addresses step three of the CSAPR framework, which itself consists of
several steps. In order to establish final EGU NOx emissions budgets for each linked upwind state, we
first identify levels of uniform levels NOx control stringency based on available EGU NOx control
strategies and represented by cost thresholds.1 These levels of uniform NOx control stringency are
modeled in IPM, as described in section B in this TSD for a discussion of this analysis. This data is
then combined with historic data in order to quantify a series of potential EGU NOx emission budgets
for each linked upwind state at each levels of uniform NOx control stringency. Next, we use the ozone
Air Quality Assessment Tool (AQAT) to estimate the air quality impacts of the upwind state EGU
NOx emission budgets on downwind ozone pollution levels for each of the assessed EGU NOx
emission budget levels. Specifically, we look at the magnitude of air quality improvement at each
receptor at each level of control, we examine whether receptors are "solved", and we look at the
individual contributions of each state to each of its receptors. See section D in this TSD for discussion
of the development and use of the ozone AQAT.

Finally, the EPA uses this air quality information in a multi-factor test, along with EGU NOx
reductions and cost, to select a particular level of uniform EGU NOx control stringency that addresses
each state's significant contribution to nonattainment and interference with maintenance (see Section
VI of the preamble for additional information).

In this TSD, we assessed the EGU NOx mitigation potential for all states in the contiguous U.S.
We also assessed the air quality impacts for all monitors in the contiguous U.S. However, in applying
the multi-factor test for purposes of identifying the appropriate level of control, the EPA only
evaluated EGU NOx reductions and air quality improvements from upwind states that were "linked" to
downwind receptors in step two of the CSAPR framework. These states are listed in Table A-l below.

1 See the EGU NOx Mitigation Strategies Final Rule TSD

3


-------
As described in preamble section VI, Delaware and the District of Columbia (D.C.) are "linked" to
downwind ozone problems but are not included in the final rule. Consequently, EPA did not include
Delaware or D.C. in applying the multi-factor test.

Table A-l. States Evaluated in the

Multi-factor Test

Ozone Season M)\

Alabama

Missouri

Arkansas

New Jersey

Illinois

New York

Indiana

Ohio

Iowa

Oklahoma

Kansas

Pennsylvania

Kentucky

Tennessee

Louisiana

Texas

Maryland

Virginia

Michigan

West Virginia

Mississippi

Wisconsin

4


-------
B. Using the Integrated Planning Model (IPM) to Project Impact of Ozone-Season NOx Cost
Thresholds, Budgets, and Policy Impacts

EPA used the Integrated Planning Model (IPM) v5.15 platform to inform air quality modeling
for the final rule. IPM was also used to analyze the ozone season NOx emission reductions available
from electric generating units (EGUs) at various uniform levels of NOx control stringency, represented
by cost, in each upwind state. Finally, IPM was used to evaluate illustrative compliance with the rule's
regulatory control alternatives (i.e., compliance with the finalized emission budgets, with a more
stringent alternative, and with a less stringent alternative).

IPM is a multiregional, dynamic, deterministic linear programming model of the U.S. electric
power sector that EPA uses to analyze cost and emissions impacts of environmental policies.2 All IPM
cases for this rule included representation of the Title IV SO2 cap and trade program; the NOx SIP
Call; the CSAPR regional cap and trade programs;3 consent decrees and settlements; and state and
federal rules as listed in the IPM documentation referenced above. The cases did not include the final
Clean Power Plan (CPP), as explained in Preamble section IV.B.

Application of the CSAPR 4-step framework requires robust data collection, IPM modeling,
and analysis and is in and of itself time consuming. Rather than freezing all IPM data sets at the outset
of EPA's analysis for the final rule, the EPA allowed for ongoing improvement of the relied upon EGU
data. As a result, each step of EPA's analysis for the final rule is informed by the best available data at
the time the analysis was conducted. For example, after the EPA began its analysis for the final rule,
Pennsylvania published its rule addressing requirements for Reasonable Available Control
Technologies (RACT) that would result in NOx emissions reductions during 2017. Rather than ignore
a significant new rule the EPA incorporated the rule into its analysis.4 One factor that enabled the EPA
to ensure that each step of its analysis is informed by the best available data is the agency's use of
AQAT (as described in section D of this TSD). Within a short period of time, AQAT allows the
agency to estimate the changes in air quality design values and air quality contributions as a result of
the changes in emissions. For example, EPA was able to estimate the changes in air quality as a result
of PA RACT and found that the 19 receptors identified in the final air quality modeling using CAMx

2	See "Documentation for EPA Base Case v.5.13 Using the Integrated Planning Model", "EPA Base Case v.5.14 Using
IPM Incremental Documentation. March 25, 2015", and "EPA Base Case v.5.15 Using IPM Incremental Documentation.
August, 2015," and "EPA v5.15 Supplemental Documentation for the Final CSAPR Update Rule" for further description of
the IPM model, available at https://www.epa.gov/airmarkets/power-sector-modeling

3	The D.C. Circuit issued its final decision in the litigation of CSAPR, remanding 11 states phase 2 NOx ozone season
budgets for reconsideration, finding that the budgets over-controlled. See EME Homer City Generation, L.P. v. EPA, 795
F.3d 118, 138 (2015) (EME Homer City II). In light of this remand, the projected base case for the final rule accounts for
compliance with the original CSAPR by including as constraints all original CSAPR emission budgets with the exception
of remanded phase 2 NOx ozone season emission budgets for 11 states. The EPA has also excluded the phase 2 NOx ozone
season emission budgets that were finalized in the original CSAPR supplemental rule for four additional states because
those state budgets would over-control in the same manner as the remanded budgets. Specifically, to reflect original
CSAPR ozone season NOx requirements, the modeling includes as constraints the original CSAPR NOx ozone season
emission budgets for 10 states: Alabama, Arkansas, Georgia, Illinois, Indiana, Kentucky, Louisiana, Mississippi, Missouri,
and Tennessee.

4	See the Memo to the Docket "Pennsylvania RACT memo to the docket for the final CSAPR Update" for details about this
rule.

5


-------
remained with design values greater than 76 ppb.5 EPA also found that each state included in the rule
maintained at least one linkage to one of those 19 receptors. EPA recognizes that AQAT is not the
equivalent of photochemical air quality modeling using tools such as CAMx. However, AQAT is
directly informed by robust CAMx data. Further, AQAT has evolved through iterative development
under the original CSAPR, the CSAPR Update proposal, and the final CSAPR Update. One such
evolution is its calibration to an emission reduction scenario that is very similar to the final rule NOx
control stringency. As a result, the version of AQAT used for the final CSAPR Update is a reliable
analytic tool that is specifically created and calibrated to the policy it is being used to evaluate. To
confirm its reliability for these purposes, the EPA used both AQAT and CAMx to evaluate the
illustrative policy analysis for the RIA, finding that AQAT provides results that are nearly identical to
CAMx. This assessment can be found in section D-4, below.

In the body of power sector modeling done for this rule, the EPA needed to quantify emissions
for three different analytic purposes. The first purpose was construct an Air Quality Modeling Base
Case to identify nonattainment and maintenance receptors and perform contribution analysis. This base
case was necessary given the long lead time required for air quality modeling. This base case
incorporated the most important fleet changes and retrofits identified through comments on the August
4, 2015 Notice of Data Availability (NOD A)6 and EPA's continuous review of the power sector.

The second purpose was to construct an illustrative, and internally consistent, base case and
control case to study the potential costs and benefits of this rulemaking, as described in the Regulatory
Impact Analysis, or RIA, for this rule. This set of cases is referred to as the "Illustrative Cases." To
allow time for air quality modeling and analysis of the policy, illustrative budgets were quantified from
the Illustrative Cost Threshold Runs. These Illustrative Cases incorporated additional comments from
the NODA and the proposal, as well as other modeling updates.

The third purpose was to quantify the final state emission budgets for the rule and to confirm
that the results of the illustrative analysis are representative of the final CSAPR Update. This set of
cases is referred to as the "Final Cases." For the Final Cases, the EPA conducted a new base case, cost
threshold cases, and policy case. This final set of analysis allowed EPA to incorporate the most
updated information possible for the calculation of emission budgets in the final regulation.

As a result, while EPA used the same budget quantification approach described in this
document in both the Illustrative and Final sets of analysis cases, the quantified results for budgets in
each track differ due to minor differences in the modeling projections within each track. Table B-l
below summarizes the various IPM runs conducted and Appendix A provides further details on each of
these scenarios.

5	For example, EPA was able to estimate the changes in air quality as a result of PA RACT and found that the 19 receptors
identified in the final air quality modeling using CAMx remained with design values greater than 76 ppb.

6	Notice of Availability of the Environmental Protection Agency's Updated Ozone Transport Modeling Data for the 2008
Ozone National Ambient Air Quality Standard (NAAQS), 80 FR 46271 (Aug. 4, 2015). available at
http://www.epa.gov/airtransport/pdfs/FR Version Transport NODA.pdf.

6


-------
Table B-l. Summary of Sets of Scenarios



Air Quality Modeling
Base Case

Illustrative Cases

Final Cases

Scenarios Run

Base Case

Base Case

Cost Threshold Cases
Policy Cases

Base Case

Cost Threshold Cases
Policy Cases

What Analysis Each
Set of Runs Inform

Base Case air quality
modeling (CAMx) to
identify nonattainment
and maintenance
receptors

Budgets used for the
Illustrative Policy
Cases

Air quality analysis
(CAMx and AQAT) of
the Illustrative Policy
Case

Illustrative Policy
Analysis for the RIA

Final Budgets

AQAT Analysis

Appended analysis
for the RIA

Updates Captured In
Each Set of Runs

Unit retirements,
additions and retrofits
flagged In NOD A
comments

Additional comments
on the NOD A and
Proposed Rule

Updated certain unit-
specific NOx rates
based on historical
trends

Emission rate of units
with SCRs is 0.081
lbs/mmBtu or lower

Updated additional
NOx rates based on
historical trends

Emission rate of units
with SCRs is 0.1
lbs/mmBtu or lower



See "NEEDS v.5.15
AQMCSAPR Update"
in the docket for full
set of unit level
characteristics

See "NEEDS v.5.15
Illustrative CSAPR
Update" in the docket
for full set of unit level
characteristics

See "NEEDS v.5.15
Final CSAPR
Update" in the docket
for full set of unit
level characteristics

For the Illustrative and Final Cases, the EPA modeled the emissions that would occur within
each state in a Base Case. The EPA then designed a series of IPM runs that imposed increasing cost
thresholds representing uniform levels of NOx controls and tabulated those projected emissions for
each state at each cost level. The EPA has referred to these runs as "Cost Threshold Runs" and these

7


-------
tabulations as "cost curves".7 The cost curves report the remaining emissions at each cost threshold
after the state has made emission reductions that are available up to the particular cost threshold
analyzed.

In each Cost Threshold run, the EPA applied ozone season cost thresholds to all fossil-fuel-
fired EGUs with a capacity greater than 25 MW in all states. As described in the EGU NOx Mitigation
Strategies Final Rule TSD, because of the time required to build advanced pollution controls, the
model was prevented from building any new post-combustion controls, such as SCR or SNCR, before
2020, in response to the cost thresholds.8 The modeling allows turning on idled existing SCR and
SNCR, optimization of existing SCR, shifting generation to lower-NOx emitting EGUs, and adding or
upgrading NOx combustion controls in 2017,9 such as state-of-the-art low NOx burners (LNB).

In these scenarios, EPA imposed cost thresholds of $800, $1,400, $3,400, $5,000, and $6,400
per ton of ozone season NOx. See Preamble Section VI for a discussion of how the cost thresholds
were determined. Table B-2 below summarizes the reduction measures that are broadly available at
various cost thresholds.

Table B-2. Reduction strategies available to EGUs at each cost threshold.

Cost Threshold ($ per
ton Ozone-Season NOx)

Reduction Options

$800

-Optimize extant operating SCRs
-generation shifting

$1,400

all above options and

- retrofitting state-of-the-art combustion controls
-Turning on and optimizing extant idled SCRs
-additional generation shifting

$3,400

all above options and

-Turning on extant idled SNCRs
-additional generation shifting

$5,000

all above options and additional generation shifting

$6,400

all above options and additional generation shifting

7	These projected state level emissions for each "cost threshold" run are presented in several formats. The IPM analysis
outputs available in the docket contain a "state emissions" file for each analysis. The file contains two worksheets. The
first is titled "all units" and shows aggregate emissions for all units in the state. The second is titled "all fossil > 25MW"
and shows emissions for a subset of these units that have a capacity greater than 25 MW. The emissions in the "all fossil >
25 MW" worksheet are used to derive the budgets for each upwind state at the cost thresholds, in an average year.

8	IPM results do include certain newly built post-combustion NOx control retrofits in base case modeling, cost curve runs,
and remedy runs. These pre-2020 retrofits do not reflect any controls installed in response to the rule, but instead represent
those that are already announced and/or under construction and expected to be online by 2018, or controls that were
projected to be built in the base case in response to existing consent decree or state rule requirements.

9	As described in Preamble Section IV.B, the EPA is aligning the analysis and implementation of this final rulemaking with
the 2017 ozone season in order to assist downwind states with the timely attainment of the 2008 ozone NAAQS. As
described in Preamble Section V.B.2., EPA adjusted the IPM v5.15 2018 run year results to account for differences
between 2017 and 2018 in the power sector, because IPM v5.15 does not have an output year of 2017.

8


-------
Within IPM, units with extant SCRs are defined as SCR equipped units with ozone season NOx
emission rates less than 0.2 lbs/mmbtu in the Base Case. These units had their emission rates lowered
to the lower of their mode 4 NOx rate in NEEDS10 and the "widely achievable" optimized emissions
rate of 0.081 lbs/mmbtu in the Illustrative Cost Threshold Cases and 0.10 lbs/mmBtu in the Final Cost
Threshold Cases.

Units equipped with SCRs with an emissions rate exceeding 0.20 lbs/mmBtu were considered
to have idled SCRs. These units had their emission rates lowered to the lower of their mode 4 NOx rate
in NEEDS and the "widely achievable" optimized emissions rate of 0.081 lbs/mmbtu in the Illustrative
Cost Threshold Cases of $1,400 per ton and higher, and 0.10 lbs/mmBtu in the Final Cost Threshold
Cases of $1,400 per ton and higher.

Units with idled SNCRs were identified as units equipped with SNCR and mode 4 NOx rates
in NEEDS greater than 0.30 lbs/mmbtu. These units were given NOx rates 25% lower to reflect reflect
SNCR operation in Cost Threshold cases of $3,400 per ton and higher.

Finally, unit combustion control configurations listed in NEEDS were compared against Table
3-11 in the Documentation for EPA Base Case v.5.13 Using the Integrated Planning Model IPM
v.5.13, which lists state-of-the-art combustion control configurations based on unit firing type. This
allowed EPA to identify units that would receive state-of-the-art combustion control upgrades in IPM.
EPA then followed the procedure in Attachment 3-1 of the documentation to calculate the each of
these unit's new NOx emission rate.

As described in preamble section VI, the EPA limited its assessment of shifting generation to
other EGUs within the same state as a proxy for the amount of generation shifting that could occur for
the 2017 ozone season. EPA did this by limiting state generation in each Cost Threshold case to the
level in its respective Base Case.

Section C.2 of this TSD describes how budgets were calculated based on each Cost Threshold
case. Once these budgets were calculated, EPA used the budgets for covered states to conduct IPM
runs to investigate the impact of compliance with the budgets calculated from the $800, $1,400, and
$3,400 per ton Cost Threshold Cases. These Cases are referred to as Policy Cases. Specifically, the
budgets calculated from the Illustrative $1400 per ton Cost Threshold case were used for the
Illustrative Policy Case, and the budgets calculated from the Illustrative $800 and $3400 per ton Cost
Threshold cases informed the Illustrative Less- and More-Stringent Policy Cases. Lastly, the budgets
calculated from the Final $1400 per ton Cost Threshold case were used for the Final Policy Case.

These Policy Cases were used to inform air quality impact analysis of the rule and inform the RIA.

To model the policy cases in IPM, EPA used the calculated budget and assurance levels to set
state and regional ozone-season NOx emissions. Additionally, EPA assumed a starting bank of
allowances equal to 21% of the sum of the state budgets. States could individually emit up to their
assurance levels in each run year, and collectively could not have emissions exceeding the sum of their
budget and banked allowances in each run year. In all policy cases, units with extant operating SCRs
were assumed to operate them at the lower of their mode 4 NOx rate in NEEDS and the "widely
achievable" emissions rate, as EPA determined this was a cost-effective mitigation strategy. For all of
the Policy Cases except the Illustrative Less Stringent Policy Case, all units with SCRs were assumed

10 The mode 4 NOx rate, as described in Chapter 3 of the Documentation for EPA Base Case v.5.13 Using Integrated
Planning Model, represents post-combustion controls operating and state-of-the-art combustion controls, where applicable.
For units determined to be operating their SCR, the rate is typically equal to the unit's rate reported in the 2011 ETS data.
For some units, as described in the EPA v.5.15 CSAPR Update Rule Base Cases Using IPM Incremental Documentation,
this data was updated. For units not operating their SCRs, the mode 4 rate is calculated as described in Attachment 3-1 of
the Documentation for EPA Base Case v.5.13 Using Integrated Planning Model.

9


-------
to operate them at the "widely achievable" emissions rate and units that did not already have state-of-
the-art combustion controls were assumed to retrofit them. In the Illustrative More Stringent Policy
Case, units with idled SNCRs were assumed to operate them. Finally, no state-level generation
constraints were imposed in the policy cases. While the EPA conservatively limited generation shifting
in developing the budgets, through use of state-level generation constraints, the EPA believes that
generation shifting may occur broadly among states and so removed that constraint for the IPM cases
modeling the policy.

10


-------
C. Calculating Budgets From IPM Run and Historical Data

1. Overview of the State Budget Formula

As described in preamble section VI, EPA developed state EGU NOx emissions budgets for
each of the cost thresholds. This section walks through the details of how each state emissions budget
was calculated for each cost threshold. As described in Section B of this TSD, the same process was
used to calculate two sets of budgets: an Illustrative for analysis of the rule and a Final set for
quantifying the final CSAPR Update budgets

As described in Preamble Section VI, the EPA determined it was appropriate to calculate
budgets by combining historical emissions and heat input data with projections from IPM. In the
proposed rule, the EPA calculated state budgets with the following formula:

State 2017 OS NOx Budget =

2015 State OS Heat Input * State 2017 IPM OS NOx Emissions Rate

(note: "OS" stands for Ozone Season, and is equivalent to the "summer" label in some IPM outputs)

EPA intended this formula to root the state emissions budgets in historical data by tying them to state-
level historic heat input as opposed to using emission projections that reflect interstate generation
patterns that occur in IPM. However, commenters raised a related concern that notwithstanding this
approach's use of state-level historic heat input, the model may still project a substantially cleaner
generation profile within the state than might be possible to achieve in the relatively short timeframe of
this rule. In other words, the proposal's application of an IPM-projected state-level emission rate to
historical state-level heat input data could still yield potentially insufficient tons for a state budget if
that state's EGUs were to maintain a similar total generation to 2015 but were unable to collectively
achieve that projected emission rate by the 2017 ozone season. To address this concern, EPA updated
the formula for the final rule to:

State 2017 OS NOx Budget = 2015 State OS Heat Input *

[2015 State OS NOx Emissions Rate —

(2017 IPM Base Case OS NOx Emissions Rate — 2017 IPM Cost Threshold OS NOx Emissions
Rate)]

This formula subtracts the change in emissions rate between the IPM Base Case and a Cost Threshold
Case from the actual 2015 emissions rate. This modified approach ensures that state budgets are
informed by IPM projections of state-level emission rate improvement (change from base case to cost
threshold case) while tying that improvement potential directly to observed emission rate performance
in 2015.

This change in analytic approach means that unit retirements and retrofits known to occur after
2015 but before 2017, which were automatically captured in the proposal's use of IPM-projected
emission rates, now need to be explicitly accounted for in the quantification of state budgets. In the
proposal methodology, these fleet changes were captured in the State 2017 IPM Emissions Rate. In the
formula above used to quantify budgets in this final rule, these fleet changes between 2015 and 2017
are reflected in both the IPM Base Case Rate and the Cost Threshold Case Rate, such that the effect of
these fleet changes on the state-level emission rates cancels itself out. In other words, the degree of

11


-------
state-level IPM-projected emission rate improvement represented in this formula only captures what
EGUs in that state can do to reduce emissions beyond the already-known retrofit and retirement
changes expected in that state between 2015 and 2017. Accordingly, EPA determined it was necessary
to adjust the 2015 State Emissions Rates to account for these known changes, so that the full degree of
emission reductions expected in the state by 2017 is captured in the budget. Therefore, the final budget
equation is:

State 2017 Budget = 2015 State OS Heat Input *

[Adjusted 2015 05 NOx State Emissions Rate —

(2017 IPM Base Case OS NOx Emissions Rate — 2017 IPM Cost Threshold OS NOx Emissions Rate)]

Finally, EPA notes that in rare instances, it is possible for a state's emission rate to increase in a
cost threshold case relative to its base case rate, even if a state's emissions decrease overall. This
situation could yield a result greater than the state's 2015 (unadjusted) emissions. This outcome can be
due to model-projected regional fuel prices and generation shifting among units. Therefore, EPA
assigned state budgets as the lower of the calculated state budget or the state's 2015 (unadjusted)
emissions.

2. Detailed Explanation of State Budget Calculations

Below is a detailed walk-through of how the formula for calculating state emissions budgets
was applied to each cost threshold run to generate a corresponding budget for each state. For
comparison and the purpose of AQAT modeling, this formula was also applied to the Base Case to
generate base case "budgets" (alternatively, this could be considered a $0 per ton ozone season NOx
cost threshold run). The budgets calculated from this process for the Final Cost Threshold cases appear
in table C-l. The detailed calculations for all Cost Threshold cases appear in Appendix E (Excel
spreadsheet).

First, the EPA tabulated each state's 2015 reported historical state-level ozone season heat
input and NOx emissions from affected sources. To capture the emissions impact of committed fleet
changes occurring before 2017, the EPA calculated an adjusted 2015 historical ozone-season NOx
emissions level for each state. For units with planned new state-of-the-art SCR retrofits to be in place
by the 2017 ozone season, heat input was assumed to match 2015 levels, but the units were given
emission rates of 0.075 lbs/mmBtu11 to reflect the control being in place for 2017 and in lieu of
whatever emission rate the unit reported in 2015 (before its SCR was in place). For units with planned
combustion control retrofits before 2017, the EPA recalculated emissions for the unit assuming heat
input matched 2015 levels and the emission rate was improved to the Mode 4 NOx Emissions Rate for
the unit listed in NEEDS.12 For units with coal-to-gas conversions planned to occur by the 2017 ozone
season, heat input was assumed to match 2015 levels, but the units were given an emissions rate equal
to half of its 2015 rate13 to reflect the NOx reductions associated with their conversion to gas for 2017.

11	This is a conservative estimate, based on the floor rates for new SCRs in the IPM documentation, ranging from 0.05 to
0.07 lbs/mmBtu, depending on coal type. See "Documentation for EPA Base Case v.5.13 Using the Integrated Planning
Model," table 5-5.

12	The Mode 4 NOx Emission Rate reflects the emission rate for a unit if state-of-the-art combustion controls were installed.
See NEEDS v5.15 Final CSAPR Update

13	This is consistent with NOx rate change used in IPM. See "Documentation for EPA Base Case v.5.13 Using the
Integrated Planning Model.", table 5-21.

12


-------
Lastly, the heat input and emissions for units retiring before 2017 was changed to zero. However, the
EPA assumed that the generation from the retiring unit would be replaced by other units. The displaced
heat input, which EPA used as a proxy for generation, from these units was assumed to be replaced by
generation within the state with an emission rate equal to the state's overall emission rate for the
remaining units.14 This heat input and associated emissions from replacement generation was then
added to the state's total, yielding adjusted 2015 historical ozone season heat input and NOx emissions.
With this data, EPA was also able to calculate each state's adjusted 2015 ozone season emissions rate.

Second, the EPA calculated each covered state's 2017 modeled ozone season NOx emission
rate for the base case and each cost threshold. To do this, EPA started with the IPM projected 2018
ozone season heat input and NOx emissions from affected sources for each state. Next, EPA added an
adjustment to account for differences in unit and SCR availability and operation between the IPM run
year of 2018 and the expected conditions applicable to calendar year 2017.15 Appendix C explains the
2017 adjustments and shows the adjustments made by model plant. Lastly, the state-level 2017
emissions rate was calculated as the total 2017 emissions from affected sources within the state,
divided by the total 2017 heat input from these sources.

Third, the EPA calculated the change in emissions rate between the IPM base case and each
cost threshold case.16 The EPA then subtracted this change in emissions rate from the adjusted 2015
emissions rate. This yielded state-level historically based emission rates reflecting modeled NOx
reduction potential.

Fourth, the EPA multiplied these rates by each state's adjusted 2015 heat input to yield
emission budgets for each cost threshold. In instances where this calculated budget was greater than
the state's 2015 (unadjusted) emissions, the state budget was set equal to the state's 2015 (unadjusted)
emissions. However, for the budgets finalized in this rule, all states had calculated budgets lower or
equal to their 2015 historical emissions. The state budgets for the Final Cost Threshold cases are
displayed in table C-l. Note that budgets are calculated for all states for the purpose of AQAT
analysis, as explained section D of this TSD, even if the state is not covered by the final CSAPR
Update Rule.

Finally, the EPA calculated the variability limits and assurance levels for each state based on
the calculated emission budgets. Each state's variability limit is 21% of its budget, and its assurance
level is the sum of its budget and variability limit (or 121% of its budget), shown in Table C-2. Under
the methodology established in the original CSAPR, the state-specific portion of the new unit set aside
(NUSA) (including the Indian Country NUSA) is calculated as the percentage equal to the projected
emissions from "planned units" divided by the state budget plus a base two percent. The calculated
existing unit allocation and NUSA, including the Indian Country NUSA, for the Final budgets is

14	Therefore, the 2015 adjusted heat input for each state is equal to its 2015 reported historical heat input.

15	Unlike the manner in which the EPA calculated state budgets that were finalized in this rule, the EPA did not include the
2017 emissions and heat input adjustments in the calculating the budgets that were included in the IPM modeling of the
Illustrative for Final Policy Cases. This is because the 2017 adjustments are done to account for emissions that are not
captured in the IPM 2018 run year, emissions that the model would not need allowances to cover. Including these 2017
adjustments in the budgets modeled in IPM would artificially inflate the state budgets and assurance levels in the model.
Therefore, analogous budgets without the 2017 adjustments were calculated for the purpose of modeling in IPM. For the
Final Policy Case, this only resulted in a total regional budget difference of 907 tons. The state budgets and assurance levels
used in IPM are shown in Appendix E.

16	The Base Case can be considered equivalent to a $0 per ton cost threshold run, with a corresponding "budget." To
calculate the equivalent "budget" for the Base Case using this process, the change in emissions rate for any state is zero.
Applied through the rest of the budget calculation process, this means that the base case "budgets" are equal to the adjusted
2015 historical emissions.

13


-------
shown in table C-3.17 The variability limits, assurance levels, New Unit Set-Asides and Indian
Country New Unit Set-Asides as further described in section VII of the preamble for the final CSAPR
Update.

A complicating factor of this analysis was the Pennsylvania RACT (PA RACT) Rule that was
finalized in April 2016. The PA RACT Rule will lead to reductions in ozone season NOx in
Pennsylvania, but not have an impact until 2017, the first year the Final CSAPR Rule would be in
effect. The EPA determined it was reasonable to not include it in its IPM modeling because it would
lead to a budget for Pennsylvania that does not reflect achievable emission reductions by way of
applying a regional uniform NOx control stringency for the 22 CSAPR Update states. In other words,
if EPA included PA RACT in the IPM modeling, then Pennsylvania's resulting budget would not be
commensurate with other state budgets for the final CSAPR Update. That is because the budget setting
methodology uses the change in emissions rate between a Base Case and a Cost Threshold case, and
2015 historical emissions data. Including the PA RACT in the Base Case for this rule would lead to a
small change in emissions rates between cases and apply that to historical data that does not reflect the
PA RACT Rule. Mixing 2015 historical data that does not reflect the impact of the PA RACT with
IPM cases that does reflect the PA RACT would yield budgets that were too large. However, as
explained in the Memo to the Docket "The Pennsylvania Additional RACT Requirements for Major
Sources of NOx and VOCs," the EPA found it reasonable to factor the PA RACT requirements into air
quality modeling using AQAT, and reflect its costs and emissions reductions in the Illustrative and
Final Base Cases and $800 per ton Cost Threshold Cases in the RIA.

As explained in the preamble, the EPA is promulgating EGU NOx ozone season emission
budgets reflecting the uniform cost threshold of $1,400 per ton to reduce significant contribution to
nonattainment and interference with maintenance. These budgets were calculated from the Final
$1,400 per ton Cost Threshold run.

For the RIA analysis, budgets calculated from the Illustrative $1,400 per ton Cost Threshold
run were used for the Illustrative Policy case. Additionally, the RIA includes analysis of the Illustrative
Less Stringent policy option, using the budgets from the Illustrative $800 per ton Cost Threshold case,
and an Illustrative More Stringent policy alternative, using the budgets from the Illustrative $3,400 per
ton Cost Threshold case. The EPA also included additional analysis of the Final Policy case in
Appendix 4-A of the RIA.

The IPM runs performed for the cost threshold analyses are listed in Appendix A of this TSD.
Table Appendix A-l lists the name of each IPM run next to a description of the run. The output files
of these model runs can be found in the rulemaking docket. Detailed budget calculations for all Cost
Threshold cases and the assurance levels used for Policy Cases can be found in Appendix E.

17 See '03 NAAQS CSAPR Update ~ NUSA Calculations' (Excel spreadsheet) in the docket.

14


-------
Table C-l. Calculated Budgets for States for Final Cost Threshold Runs.



State















Covered















by Final



Final

Final

Final

Final

Final



CSAPR

Final

$800/ton

$1400/ton

$3400/ton

$5000/ton

$6400/ton



Update

Base

Cost

Cost

Cost

Cost

Cost

State

Rule?

Case

Threshold

Threshold

Threshold

Threshold

Threshold

Alabama

Y

15,179

14,332

13,211

12,620

11,928

11,573

Arkansas18

Y

12,560

12,048

9,210

9,048

8,518

8,050

Arizona



10,710

10,680

9,810

9,788

9,710

9,486

California



1,905

1,905

1,905

1,905

1,810

1,810

Colorado



14,010

14,008

13,994

13,645

13,495

12,950

Connecticut



605

584

558

558

554

554

Delaware



497

497

497

494

494

494

Florida



22,779

21,611

17,123

16,631

16,481

16,375

Georgia



8,762

8,495

8,481

8,525

8,532

7,764

Iowa

Y

11,478

11,477

11,272

11,065

10,891

10,491

Idaho



152

152

152

152

152

152

Illinois

Y

14,850

14,682

14,601

14,515

14,248

14,054

Indiana

Y

31,382

28,960

23,303

21,634

19,990

18,720

Kansas

Y

8,031

8,030

8,027

7,975

7,962

7,767

Kentucky

Y

26,318

24,052

21,115

21,007

20,273

19,496

Louisiana

Y

19,101

19,096

18,639

18,452

18,442

18,426

Massachusetts



1,119

1,119

1,112

1,098

1,071

1,072

Maryland

Y

3,871

3,870

3,828

3,308

2,938

2,926

Maine



109

109

109

109

109

109

Michigan

Y

19,811

19,558

16,545

15,298

12,616

12,115

Minnesota



7,068

7,068

6,864

6,761

6,651

6,451

Missouri

Y

18,443

17,250

15,780

15,299

14,673

14,555

Mississippi

Y

6,438

6,438

6,315

6,243

6,203

6,205

Montana



6,540

6,540

6,540

6,535

6,535

6,535

North Carolina



17,419

14,424

14,326

13,189

12,460

12,207

North Dakota



18,738

18,398

18,016

17,655

17,497

17,435

Nebraska



9,737

9,678

8,970

7,197

6,467

6,448

New Hampshire



416

416

416

416

415

415

New Jersey

Y

2,114

2,100

2,062

2,008

1,867

1,879

New Mexico



9,443

9,443

9,055

8,921

8,746

8,650

Nevada



2,405

2,301

2,241

2,112

1,559

886

New York

Y

5,531

5,220

5,135

5,006

4,746

4,594

Ohio

Y

27,382

23,659

19,522

19,165

18,561

18,348

Oklahoma

Y

13,747

13,746

11,641

9,174

8,790

8,439

Oregon



—

—

—

—

—

—

Pennsylvania

Y

35,607

30,852

17,952

17,928

17,621

17,374

Rhode Island



283

283

283

283

283

283

South Carolina



5,486

5,288

5,288

5,293

5,300

5,318

South Dakota



853

853

853

853

853

853

Tennessee

Y

7,779

7,736

7,736

7,735

7,724

7,729

Texas

Y

54,839

54,521

52,301

50,011

48,795

47,994

Utah



16,949

16,949

14,149

13,592

11,356

11,324

Virginia

Y

9,367

9,365

9,223

8,754

8,619

8,416

Vermont



52

52

52

52

52

52

Washington



3,085

3,085

3,085

3,085

3,085

3,085

Wisconsin

Y

7,939

7,924

7,915

7,790

7,435

7,023

West Virginia

Y

26,874

25,984

17,815

17,380

17,388

17,373

Wyoming



16,005

15,828

14,691

13,728

12,371

12,121

CSAPR State Total

378,641

360,900

313,148

301,415

290,228

283,547

18 The calculated budgets shown in this table for Arkansas correspond to its budget for 2018 and subsequent control
periods. As discussed in Preamble Section VI.D, Arkansas's 2017 is 12,048 tons.

15


-------
Table C.2. State Budgets, Variability Limits, and Assurance Levels for the CSAPR Upda

State

Budget

Variability Limit

Assurance Level

Alabama

13,211

2,774

15,985

Arkansas19

12,048 (2017)
9,210 (2018+)

2,530 (2017)
1,934 (2018+)

11,144 (2017)
14,478 (2018+)

Iowa

11,272

2,367

13,639

Illinois

14,601

3,066

17,667

Indiana

23,303

4,894

28,197

Kansas

8,027

1,686

9,713

Kentucky

21,115

4,434

25,549

Louisiana

18,639

3,914

22,553

Maryland

3,828

804

4,632

Michigan

16,545

3,474

20,019

Missouri

15,780

3,314

19,094

Mississippi

6,315

1,326

7,641

New Jersey

2,062

433

2,495

New York

5,135

1,078

6,213

Ohio

19,522

4,100

23,622

Oklahoma

11,641

2,445

14,086

Pennsylvania

17,952

3,770

21,722

Tennessee

7,736

1,625

9,361

Texas

52,301

10,983

63,284

Virginia

9,223

1,937

11,160

Wisconsin

7,915

1,662

9,577

West Virginia

17,815

3,741

21,556

CSAPR Update
Region Total

313,986 (2017)
313,148 (2018+)

N/A

N/A

e Rule.

19 As discussed in Preamble Section VI.D, Arkansas's 2017 is 12,048 tons and 9,210 tons in 2018 and subsequent control
periods.

16


-------
Table C-3. Existing unit allocation and NUSA, including the Indian Country NUSA, for the Final
CSAPR Update Budgets.				

State

Final 2017* EGU

New unit set-

New unit set-aside

Indian country



NOx Emission

aside amount

amount (tons)1

new unit set-aside



Budgets (tons)

(percent)



amount (tons)

Alabama

13,211

2

255

13

Arkansas*

12,048/9,210

2/2

240/185



Illinois

14,601

2

302



Indiana

23,303

2

468



Iowa

11,272

3

324

11

Kansas

8,027

2

148

8

Kentucky

21,115

2

426



Louisiana

18,639

2

352

19

Maryland

3,828

4

152



Michigan

16,545

4

643

17

Mississippi

6,315

2

120

6

Missouri

15,780

2

324



New Jersey

2,062

9

192



New York

5,135

5

252

5

Ohio

19,522

2

401



Oklahoma

11,641

2

221

12

Pennsylvania

17,952

3

541



Tennessee

7,736

2

156



Texas

52,301

2

998

52

Virginia

9,223

6

562



West Virginia

17,815

2

356



Wisconsin

7,915

2

151

8

1 New-unit set-aside amount (tons) does not include the Indian country new unit set-aside amount (tons).
*The EPA is finalizing CSAPR EGU NOx ozone season emission budgets for Arkansas of 12,048 tons for
2017 and 9,210 tons for 2018 and subsequent control periods.

3. Assessing the Impact of Limiting Inter-State Generation Shifting

As described in preamble section VI, the EPA limited its assessment of shifting generation to
other EGUs within the same state as a proxy for the amount of generation shifting that could occur for
the 2017 ozone season. The amount of NOx reductions between the IPM cases that were used to
quantify the emission budgets under this assumption can be found in Appendix F. To determine the
impact of this assumption, the EPA conducted a separate analysis similar to the $1,400 per ton cost
threshold scenario, except without limiting IPM's ability to shift generation between states. EPA
calculated budgets from this alternative case and compared them to the final budgets, also shown in
Appendix F. This analysis showed that budgets would only be 1,101 tons lower in the CSAPR region
compared to the final budgets if the EPA assumed generation shifting was unrestricted between the
states, or 0.35% of the sum of the final CSAPR state budgets.

17


-------
At the state level, 19 of the 22 states had budgets less than 1% different and two additional
states had changes of less than 2%, when comparing this alternative case to the approach used in the
final rule. One state, Maryland, had an alternative budget 20% lower than its final budget. These
results show two things. First, overall, this constraint had minimal impacts on state budgets. Second, in
one case, the constraint did significantly impact a state budget, lowering Maryland's budget 20% as
generation was shifted out of state. This evaluation supports EPA's determination that it was
reasonable to use a conservative estimate of the potential for emissions reductions from generation
shifting in the relatively short timeframe of this particular rule by limiting IPM's ability to shift
generation among states in the modeling projections to inform state budget quantification.

18


-------
D. Analysis of Air Quality Responses to Emission Changes Using an Ozone Air Quality
Assessment Tool (AQAT)

EPA has defined each linked upwind state's significant contribution to nonattainment and
interference with maintenance of downwind air quality using a multi-factor test (described in the
preamble at section VI in step three of the CSAPR framework) which is based on cost, emissions, and
air quality factors. A key quantitative input for determining the amount of each state's emission
reduction obligation is the predicted downwind ambient air quality impacts of upwind EGU emission
reductions under the budgets at various levels of NOx emission control (see in section C of this TSD).
The emission reductions under the various levels of emission budgets can result in air quality
improvements such that individual receptors drop below the level of the NAAQS based on the
cumulative air quality improvement from the states, as well as decrease each upwind state's
contributions such that they possibly drop below the 1% threshold that was used to identify the states
for further analysis in step 2 of the CSAPR framework.

Direct simulation of air quality in CAMx would be the optimal way to examine these questions
at each level of emission budgets. However, time and resource limitations (in particular the amount of
time needed to set up, run the CAMx model, and analyze the results for a single model run) precluded
the use of full air quality modeling for all but a few emissions scenarios. Therefore, in order to
estimate the air quality impacts for the various levels of emission budgets and for the illustrative
control alternative, EPA used a simplified air quality assessment tool (AQAT).20 The simplified tool
allows the Agency to analyze many more NOx emission budget levels than would otherwise be
possible. The inputs and outputs of the tool can be found in the "AQAT final calibrated.xlsx" excel
workbook.

The remainder of section D of this document will:

•	Present an introduction and overview of the ozone AQAT;

•	Describe the construction of the ozone AQAT; and

•	Provide the results of the NOx emission budget level analyses.

1. Introduction: Development of the ozone AQAT

The ozone AQAT was developed for use in the rule's step three air quality analysis as part of
the multi-factor test. Specifically, the AQAT was designed to evaluate air quality changes in response
to emissions changes in order to quantify necessary emission reductions under the good neighbor
provision and to evaluate potential budgets for over-control as to either the 1% threshold or the
downwind receptor status. EPA described and used a similar tool in the original CSAPR to evaluate
good neighbor obligations with respect to the fine particulate matter (PM2.5) NAAQS and in the
proposed CSAPR Update to evaluate good neighbor obligations with respect to ozone. For the CSAPR
Update, EPA refined both the construction and application of the assessment tool for use in estimating
changes in ozone concentrations in response to changes in NOx emissions. One important change

20 EPA used CAMx to model both the base case (i.e., to determine the receptors and the contributions of each state to those
receptors) and the final budget policy scenario. The air quality estimates in AQAT were comparable to design values in
CAMx for the final illustrative control scenario, suggesting that the air quality estimates for alternative control scenarios
performed using AQAT are also reasonable.

19


-------
between the original CSAPR and this effort is to use AQAT to examine changes in ozone rather than
PM2.5. We followed the methodology developed in the original CSAPR rulemaking where we
calibrate the response of a pollutant using two CAMx simulations at different emission levels.21'22
A critical factor in the assessment tool is the establishment of a relationship between ozone
season NOx emission reductions and reductions in ozone. Within AQAT, we assume that the reduction
of a ton of emissions of NOx from the upwind state results in a particular level of improvement in air
quality downwind.23 For the purposes of developing and using an assessment tool to compare the air
quality impacts of NOx emission reductions under various emission budget levels, we determine the
relationship between changes in emissions and changes in ozone contributions on a receptor-by-
receptor basis. Specifically, EPA assumed that, within the range of total NOx emissions being
considered (as defined by the emission budget levels), a change in ozone season NOx emissions leads
to a proportional change in downwind ozone contributions.24 This proportional relationship was then
modified using calibration factors created using the 2017 base case contribution air quality modeling
and the 2017 illustrative control case from the proposal to account for the majority of the nonlinearity
between emissions and ozone concentrations.25 For example, we assume that a 20% decrease in the
upwind state's emissions leads to a 20% decrease in its downwind ozone contribution in the

21	In CSAPR, we estimated changes in sulfate using changes in SO2 emissions.

22	In this rule, we used CAMx to calibrate the assessment tool's predicted change in ozone concentrations to changes in
NOx emissions. This calibration is receptor-specific and is based on the changes in NOx emissions and resulting ozone
concentrations between the 2017 base case and the modeled "illustrative control case" in 2017 from the modeling
conducted for the proposed rule. This "illustrative control case" was created during the development of the assessment tool
for the proposed rule and is an EGU NOx ozone-season emission budget sensitivity scenario at $ 1,300/ton for ozone-season
NOx, reflecting emission reductions from sources in the 23 eastern states that the EPA proposed to regulate under the rule.
One intent of this control scenario was to create a calibration point within the range of all emission reductions for the
geography examined by EPA using the AQAT. This calibration point was used to create site-specific calibration factors so
that the response of ozone concentrations to upwind NOx emission changes would more-closely align with ozone estimates
from CAMx.

23	This downwind air quality improvement is assumed to be indifferent to the source sector or the location of the particular
emission source within the state where the ton was reduced. For example, reducing one ton of NOx emissions from the
power sector is assumed to have the same downwind ozone reduction as reducing one ton of NOx emissions from the
mobile source sector. Because the emission reductions from base case to the 2017 illustrative control case at proposal and
the resulting air quality improvements per ton of reduction occur exclusively from the power sector, the calibration factor
and thereby calibrated ozone AQAT is tuned to changes in emissions in the power sector.

24As noted by EPA at proposal and as stated by commenters, the relationship between NOx emissions and ozone
concentrations is known to be non-linear when examined over large ranges of NOx emissions (e.g., J.H. Seinfeld and S.N.
Pandis, Atmospheric Chemistry and Physics From Air Pollution to Climate Change, 2nd Edition, John Wiley and Sons,
2006, Hoboken, NJ, pp 236-237). However, for some ranges of NOx, VOC, and meteorological conditions, the relationship
may be reasonably linear. In this assessment tool, we are assuming a linear relationship between NOx emissions and ozone
concentrations calibrated between two CAMx simulations. To the extent that the changes in concentrations are the result of
small changes in emissions, EPA disagrees with commenters assertions that these relationships are highly nonlinear. The
nonlinearities are evident over tens of ppb of ozone changes with tens of percent changes in the overall emission
inventories. For most states examined here, under the various control scenarios, most changes in the emission inventory are
on the order of a few percent and most air quality changes are on the order of a fraction of a ppb. A significant portion of
the nonlinearity is accounted for by using the calibration factor and having the air quality estimates occur at levels of
emissions around the base case and the illustrative control case (which were both modeled in CAMx). In the EPA's air
quality estimates using the calibrated AQAT, to the extent that uncertainties and non-linearities are present, they are more-
likely to be the result of assuming uniform percent changes in statewide emissions, rather than changes in emissions at
particular sources.

25 See the Ozone Transport Policy Analysis TSD, EPA-HQ-OAR-2015-0500-0186 and the Ozone AQAT ,EPA-HQ-OAR-
2015-0500-0069, both from the prosed rule, for the values and a description of the calibration factors from proposal.

20


-------
"uncalibrated" ozone AQAT, while it may only decrease by 10% decrease in "calibrated" AQAT
(where the calibration factor is 0.5).

The creation of the calibration factors is described in detail in section D.2.c (1).

In summary, because the tool is only being used over a fairly narrow emissions range (for
which a calibration factor has been developed), and because other options such as using CAMx to
model all scenarios is cost and time-prohibitive, EPA used ozone AQAT to estimate the downwind
ozone reductions due to upwind NOx emission reductions for the air quality input to the multi-factor
test for this rule. Other options, such as directly scaling the results (i.e., an "uncalibrated ozone
AQAT") will likely greatly overestimate the air quality impacts of emission reductions. The
successful comparison of the AQAT estimates with the CAMx results for the $l,400/ton final
illustrative control scenario demonstrates that, for the purposes here, the AQAT can sufficiently
estimate the air quality impacts of relatively small emission changes.

Section D.2, below, is a technical explanation of the construction of the ozone AQAT. Readers
who prefer to access the results of the analysis using the ozone AQAT are directed to section D.3.

2. Details on the construction of the ozone AQAT

(a) Overview of the ozone AQAT

This section describes the step-by-step development process for the ozone AQAT. All of the
input and output data can be found in the Excel worksheets described in Appendix B. In the ozone
AQAT, EPA links state-by-state NOx emission reductions (derived from the IPM EGU modeling) with
CAMx modeled ozone contributions in order to predict ozone concentrations at different levels of
emission budgets at monitoring sites. The reduction in ozone contributions at each level of emissions
budgets and the resulting air quality improvement at monitoring sites with projected nonattainment
and/or maintenance problems in the 2017 base case were then considered in a multi-factor test for
identifying the level of emissions reductions that define significant contribution to nonattainment and
interference with maintenance.

In applying AQAT to analyze air quality improvements at a given receptor, emissions were
reduced in only those upwind states that were "linked" to that receptor in step 2 of the CSAPR
framework (i.e., those states that contributed an air quality impact at or above the 1 percent of the
NAAQS). Emissions were also reduced in the state that contained that receptor (regardless of the level
of that state's contribution) at a level of control stringency consistent with the budget level applied in
upwind states.

Specifically, the key estimates from the ozone AQAT for each receptor are:

•	The ozone contribution as a function of emissions at each budget level, for each upwind
state that is contributing above the 1 percent air quality threshold and the state containing
the receptor.

•	The ozone contribution under base case NOx emissions (i.e., the adjusted historic IPM 5.15
base case, or $0 per ton emission budget), for each upwind state that is not above the 1
percent air quality threshold for that receptor.

21


-------
• The non-anthropogenic (i.e., background, boundary, biogenic, and wildfire) ozone

concentrations. These are assumed to be constant and equal to the contributions from the
2017 final base case source apportionment modeling.

The results of the ozone AQAT analysis for each emission budget level can be found in section D.3 of
this document.

(b) Data used to construct the ozone AQAT for this rule

Several air quality modeling and emissions inventory sources were used to construct the
calibrated ozone AQAT for this rule. Several data sources from proposal provide the necessary
information to construct the calibration factors.26 Using the calibration factors, EPA modulated the
final 2017 CAMx ozone season contributions for each upwind state to each downwind receptor. For
each scenario, EPA multiplied each state's percent change in emissions at each emission budget level
relative to the 2017 base case ozone season NOx emission inventories from all source sectors used in
the source apportionment CAMx air quality modeling (this includes all anthropogenic sources and
excludes biogenic sources and wildfires) by the receptor-specific calibration factor and the state's base
case contribution. Note that the change in emissions for each emission budget level is compared with
the IPM emission estimates from the base case that was modeled in CAMx. The base case emission
inventories for the 2017 base case and the CAMx 2017 base case source apportionment air quality
modeling results are discussed in the Air Quality Modeling Technical Support Document for the Final
Cross-State Air Pollution Rule Update. The ozone season NOx EGU emissions for each emission
budget level including the base case as modeled in AQAT, are listed in Table D-3 and described in
section C of this TSD.

As described in the Air Quality Modeling Technical Support Document for the Final Cross-
State Air Pollution Rule Update and the preamble at section V, the air quality contributions and
emissions were modeled for all states in the contiguous United States and the District of Columbia.
Thus, in the ozone AQAT, any emission differences between the air quality modeling base case and
the base case emissions budget scenario or the emission budget cost levels (for linked states) would
result in changes in air quality contributions and ozone concentrations at the downwind monitors.27

26EPA used the proposal calibration factors, rather than calibration factors based on the final modeling because we found
them to have more-representative emission reductions throughout the 22 state geography. The final budget control scenario
has smaller emission reductions compared with the proposal budget control scenarios. In addition, for the final modeling,
the geographic distribution of emissions changed between the air quality modeling base and $l,400/ton illustrative control
cases, leading to calibration factors that seemed to be unusually large or small. The datasets required to construct the
calibration factors were: the proposed 2017 base case ozone-season NOx emission inventories from all source sectors used
in the source apportionment CAMx air quality modeling (this includes all anthropogenic sources and excludes biogenic
sources and wildfires); the proposed CAMx 2017 ozone-season contributions for each upwind state to each downwind
receptor; and the proposed 2017 illustrative control case ozone-season NOx emissions inventories from all source sectors.
An additional dataset, the proposed 2017 ozone concentrations from CAMx for the illustrative control case, was used to
compare the ozone AQAT-estimated ozone concentrations for this scenario to the corresponding air quality modeling
results, and develop calibration factors to align the response of ozone to changes in NOx emissions in the ozone AQAT
with the response predicted by CAMx. See the Ozone AQAT ,EPA-HQ-OAR-2015-0500-0069 from the proposed rule for
the data and results.

27 Because the illustrative control case does not include emission changes in some upwind states (e.g., states in the western
portion of the domain), calibration factors developed for monitors in these states, and any resulting changes in air quality
projected by AQAT, may not be representative.

22


-------
(c) Detailed outline of the process for constructing and utilizing the ozone AO AT

The ozone AQAT was created and used in a multi-step process. First, using the data sets and
the AQAT from the proposed CSAPR Update, calibration factors were created. Next, a calibrated
ozone AQAT was created using the contributions and emission inventory from the 2017 base case air
quality modeling for the final rule. For each emissions budget scenario evaluated, for each state, EPA
identified the percent change in anthropogenic NOx emissions relative to the 2017 base case and
multiplied this by the receptor-specific calibration factor as well as by the state- and receptor-specific
contribution. This resulted in a state- and receptor-specific "change in contribution" relative to the
2017 base case. Each state's change in contribution value was then added to its 2017 base case
contribution and the results summed for all states for each receptor. To this total of each state's
contribution to each receptor, the receptor-specific base case contributions from the other source-
categories was added, resulting in an estimated design value for each receptor.28 The calibrated ozone
AQAT was used to project the ozone concentrations for each NOx emission budget level on a receptor-
by-receptor basis for every monitor throughout the domain.

In order to facilitate understanding of the calibration process, EPA describes below a
demonstrative example used at proposal: monitor number 240251001 in Harford County, Maryland,
with a 2017 base case projected ozone average design value of 81.3 ppb and maximum design value of
84.0 ppb, at proposal.

(1) Create the calibration factors

The process for creating the calibration factors remains unchanged from the proposal.
Furthermore, EPA used those data sets and retained the same calibration factors used at proposal (i.e.,
in the remainder of this subsection D.2.c.(l), we refer to data sets from the proposal). To create the
calibration factors, EPA used emissions and contributions from proposal to estimate the change in
predicted ozone due to NOx emission reductions under the proposed illustrative control case relative to
the proposed 2017 base case.

First, EPA calculated ozone season state-level 2017 base case total NOx emissions from all
source sectors from the proposal. These emissions estimates were used for the proposed rule's CAMx
2017 source apportionment modeling. This emissions data is divided into multiple source sectors for
the purposes of air quality modeling: power sector point, non-power sector point, non-point, onroad,
nonroad, C3 marine, aim, and fires (see the Emissions Inventory TSD from the proposed rule for
additional details on the emissions inventories used in the CAMx air quality modeling).29 The state-
level total NOx emissions are the sum of emissions from all these source sectors. Next, EPA
calculated the ozone season 2017 total NOx emissions across all source sectors for the illustrative
control case. EPA calculated the ratio of the emissions for the illustrative control case to the total
emissions for the base case for each state modeled in CAMx. More information on the emissions
inventories can be found in the preamble to the proposed rule and in the August 4, 2015 Notice of Data

28	Details on procedures for calculating average and maximum design values can be found in the Air Quality Modeling
Technical Support Document for the Final Cross-State Air Pollution Rule Update.

29	"Technical Support Document (TSD) Preparation of Emissions Inventories for the Version 6.2, 2011 Emissions
Modeling Platform", available at

www.epa.gov/ttn/chief/emch/201 lv6/201 lv6_2_2017_2025_EmisMod_TSD_aug2015 .pdf

23


-------
Availability, or NODA.6 The total emissions data and resulting ratios from the proposal can be found
in Table D-l.

For each monitor, the "uncalibrated" change in concentration from proposal was found by
multiplying the 2017 base case ozone contribution by the difference in the ratio of emissions. The
difference in the ratio of emissions was calculated as the difference in total ozone season NOx
emissions between the illustrative control case and the 2017 base case scenario divided by the 2017
base case emission. Thus, when the illustrative control case had smaller emissions than the base case,
the net result was a negative number. The change in concentrations summed across all states was the
total "uncalibrated" change in concentration.

24


-------
Table D-l. From the Proposed Rule, the 2017 Base Case and 2017 Illustrative Control Case
Ozone Contributions (ppb) for Monitor Number 240251001 in Harford County, Maryland, as
well as







2017

Ratio of Illustrative

Difference between the Illustrative

Estimated 2017



2017 Base

2017 Base

Illustrative

Control Case

Control Case Emissions and Base

Contribution of Ozone



Case Ozone

Case NOx

Control Case

Emissions to Base

Case Emissions as a Fraction of

fUncalibrated Ozone

State/Source

Contributions

Emissions

NOx Emissions

Case Emissions

Base Case Emissions

AOAT)

AL

0.4053

88,805

85,721

0.97

-0.03

-0.01

AZ

0.0958

71,906

71,906

1.00

0.00

0.00

AR

0.2264

69,737

69,039

0.99

-0.01

0.00

CA

0.1106

236,322

236,322

1.00

0.00

0.00

CO

0.1942

90,756

90,756

1.00

0.00

0.00

CT

0.011

17,672

17,672

1.00

0.00

0.00

DE

0.1559

7,786

7,786

1.00

0.00

0.00

DC

0.7334

2,252

2,252

1.00

0.00

0.00

FL

0.1141

177,514

177,513

1.00

0.00

0.00

GA

0.3035

103,536

103,526

1.00

0.00

0.00

ID

0.0349

27,893

27,893

1.00

0.00

0.00

[L

0.672

148,178

147,770

1.00

0.00

0.00

IN

1.8904

139,133

127,487

0.92

-0.08

-0.16

[A

0.1933

70,467

70,045

0.99

-0.01

0.00

KS

0.285

79,939

79,513

0.99

-0.01

0.00

KY

1.973

106,830

97,311

0.91

-0.09

-0.18

LA

0.2597

173,330

172,886

1.00

0.00

0.00

ME

0.0005

17,576

17,576

1.00

0.00

0.00

MD

24.619

46,029

45,312

0.98

-0.02

-0.38

MA

0.0037

35,369

35,369

1.00

0.00

0.00

MI

0.8339

131,486

124,374

0.95

-0.05

-0.05

MN

0.1142

89,328

89,332

1.00

0.00

0.00

MS

0.1596

54,832

54,706

1.00

0.00

0.00

MO

0.5299

101,035

99,736

0.99

-0.01

-0.01

MT

0.0688

38,504

38,504

1.00

0.00

0.00

NE

0.1569

70,005

70,005

1.00

0.00

0.00

NV

0.0279

28,192

28,192

1.00

0.00

0.00

NH

0.0009

8,932

8,932

1.00

0.00

0.00

NJ

0.4374

52,743

52,031

0.99

-0.01

-0.01

NM

0.1688

65,263

65,263

1.00

0.00

0.00

NY

0.4009

109,910

107,416

0.98

-0.02

-0.01

NC

0.4684

98,064

91,850

0.94

-0.06

-0.03

ND

0.0848

74,118

74,118

1.00

0.00

0.00

OH

4.0022

160,110

150,516

0.94

-0.06

-0.24

OK

0.4683

131,763

129,215

0.98

-0.02

-0.01

OR

0.0232

40,507

40,507

1.00

0.00

0.00

PA

6.0769

174,664

147,166

0.84

-0.16

-0.96

RI

0.0006

5,845

5,844

1.00

0.00

0.00

SC

0.1097

55,897

55,846

1.00

0.00

0.00

SD

0.0587

22,192

22,192

1.00

0.00

0.00

TN

0.7044

85,759

85,693

1.00

0.00

0.00

rx

1.0563

467,245

465,179

1.00

0.00

0.00

UT

0.0942

66,486

66,486

1.00

0.00

0.00

VT

0.0015

5,473

5,473

1.00

0.00

0.00

VA

5.3016

87,754

87,514

1.00

0.00

-0.01

WA

0.0327

75,833

75,833

1.00

0.00

0.00

wv

2.9988

64,839

53,954

0.83

-0.17

-0.50

WI

0.2178

75,047

75,035

1.00

0.00

0.00

WY

0.2063

68,864

68,864

1.00

0.00

0.00

TRIBAL

0.0436

26,717

26,717

1.00

0.00

0.00

CNMX

0.7368





1.00

0.00

0.00

OFFSHORE

0.4494





1.00

0.00

0.00

FIRE

0.3074





1.00

0.00

0.00

ICBC

16.652





1.00

0.00

0.00

BIOG

6.0915





1.00

0.00

0.00

25


-------
Next, the estimate of the monitor specific ozone responses under the illustrative control case
was used to calibrate the ozone AQAT to CAMx and to derive the calibration factor. First, the changes
in ozone predicted by the ozone AQAT and CAMx for the average design values were calculated for
each monitor for the illustrative control case relative to the 2017 base case concentrations. The change
in ozone predicted by CAMx was then divided by the change in ozone predicted by the uncalibrated
AQAT, resulting in a monitor-specific calibration factor (see Table D-2 for an example calculation).
The calculation of these monitor-specific calibration factors provided EPA with the ability to align the
ozone response predicted by the ozone AQAT to the ozone response predicted by CAMx at a level of
NOx reductions that EPA expected to be close to the range of all emission reductions examined by
EPA for the final rule.

The ozone AQAT and CAMx concentration differences from proposal can be found in the
"ozone AQAT.xlsx" excel workbook on worksheet "2017 contributions uncalibrated" in columns BN
and BO, respectively.25 The resulting calibration factor from the proposal can be found in column BP
of the aforementioned excel worksheet.

Following the completion of the AQAT analysis and modeling of the $l,400/ton level of the
budget control scenario in CAMx, EPA was able to calculate updated calibration factors using the
differences in concentrations and emissions between the base case and budget control cases. EPA
examined all of the AQAT estimates for receptors using the final calibration factors and did not
observe differences that would have resulted in substantive changes in the application of the multi-
factor test (i.e., the status of receptors remained unchanged, either attainment, nonattainment, or
maintenance at each emission budget level).

A comparison of the calibrations is shown in Figure 1, for the receptors in the rule (left panel)
and overall, for all receptors in the eastern US (right panel). During the course of this analysis, EPA
noticed, for some monitors (generally ones that were not nonattainment and/or maintenance receptors
in the base case and often were located in the western US) that the final calibration factors led to
nonintuitive results. In part, this may be a result of changes in the EGU emission estimates between
the initial base case CAMx modeling and the final budget scenarios (particularly for states that are not
in the geography of the final rule). Consequently, EPA elected to continue its use of the calibration
factors from the proposal.26 Using the calibration factors from the proposal in the creation of the final
calibrated AQAT also enabled EPA to evaluate the AQAT estimates for the final rule by comparing
them to the final air quality modeling of the illustrative budget control scenario since this air quality
modeling was not used in the creation of AQAT (see section D.4 of this document for details of this
comparison).

Calibration Factor for Update Rule Receptors	Cumulative distribution of calibration factors in

c	the eastern US

O

2 800

jD

"to i" 700

V

ro ™ 600



y = 0.9589x

•



R2 = 0.852





• 		







0





•





5

500

£ I 400
300
200
100 i
0 1



itittt t

•

•H* 			

s

• final



•



•

•		

• proposed



	

~

•

0.4	0.6

2	3

Proposed Rule	Calibration factor value

Figure 1. (left panel) A comparison of the calibration factors from the proposed and final rule, (left
panel) A cumulative distribution, showing the number of monitors that have calibration factors below a
particular value.

26


-------
Table D-2. From the Proposal, Ozone Contributions in the 2017 Base Case and 2017 Illustrative
Control Case Calibration Scenario from CAMx and Uncalibrated Ozone AQAT for Monitor
Number 240251001 in Harford County, Maryland (See Table D-l). These Values are then Used
to Create a Calibration Factor.



2017 Base Case

Ozone
Concentration
(PPb)

Estimated 2017
Illustrative Control
Case Calibration
Scenario Ozone
Concentration (ppb)

Estimated
Change in
Concentration

CAMx

81.369

80.469

-0.900

Ozone AQAT

81.369

78.803

-2.566

Calibration Factor - Change in
Concentration from CAMx Divided
by Change in Concentration from the
Ozone AQAT





0.3508

(2) Create a calibrated version of the ozone AQAT for emission budget analysis for the final rule

Next, EPA used 2017 base case emissions and 2017 base case air quality ozone contributions
from the final rule air quality modeling along with the calibration factors from the proposal to create a
"calibrated" AQAT for the final rule. EPA examined the changes in the final rule contributions from
changes in emissions relative to the final rule base case emissions (while using the calibration factor).
This calibrated AQAT was then used to estimate the change in predicted ozone due to NOx emission
reductions under each emission budget level evaluated.

First, as described above in section C of this TSD, EPA identified various levels of emissions
budgets based on projected changes in emissions rates and adjusted historic data. For each state, the
EGU emissions examined in AQAT are presented in Table D-3 (some additional columns as well as
some minor differences in emissions can be found compared with the budgets in Tables C.l and C.2).
In all emission budget simulations, the contributions for all states were adjusted to the adjusted
historic level or to an emission budget level using IPM.30 For this assessment, because the emissions
from all other sectors are constant, the EPA focused only on the differences in EGU emissions between
each cost threshold31 and the 2017 base case used in the modeling (see Table D-4 for the emission
differences).32 Finally, for each emission budget level, EPA calculated the ratio of the emission
differences to the total NOx emissions for the 2017 base case used in the air quality modeling for each
state modeled in CAMx (see Table D-5).33

30	For Pennsylvania, this included an adjustment to reflect the estimated impact of the PA RACT Rule, as described in the
Memo to the Docket "The Pennsylvania Additional RACT Requirements for Major Sources of NOx and VOCs."

31	For Pennsylvania, the EPA used cost thresholds adjusted to reflect the PA RACT.

32	We note that the total ozone-season NOx emissions from the IPM outputs used in the assessment tool air quality analysis
and the EGU emissions used in the CAMx air quality modeling were slightly different (i.e., some EGU emissions are
apportioned to different sectors in the emission inventory used in CAMx). However, within ozone AQAT, because the
difference in emissions were consistently calculated using IPM's "all fossil >25 MW", the resulting air quality estimates are
not affected.

33The total emissions from all anthropogenic sources (excluding Biogenics and Fires), coinciding with the emissions that
were "tagged" in the source-apportionment modeling.

27


-------
For each emission budget level analyzed, on a receptor-by-receptor basis, the emissions change
for each upwind state is associated with one of two emission budget levels (either the adjusted historic
base case emission level or the particular threshold cost level) depending on whether the upwind state
is "linked" to that receptor or if the receptor is located within the state. States that are contributing
above the air quality threshold (i.e., greater than or equal to 1 percent of the NAAQS) to the monitor,
as well as the state containing the monitor, make NOx emission reductions available at the particular
emission budget level. The emissions for all other states are adjusted to the adjusted historic base case
level.

For the $l,400/ton final illustrative control alternative for the RIA, all states were adjusted to
the emission levels in the illustrative control case, regardless of whether the state was "linked." These
scenarios examine the emission results when budgets have been applied to the geography. This
scenario was modeled in CAMx allowing a comparison with the AQAT estimates.

For each monitor, the predicted 2017 change in contribution of ozone from each state is
calculated by multiplying the state-specific 2017 base case ozone contributions from the final air
quality modeling by the calibration factor as well as by the ratio of the change in emissions (Table D-5,
for either the emission budget level or the adjusted historic base case emission budget level adjusted
for PA RACT depending on whether the state is linked, to the total 2017 base case emissions for all
sectors used in the air quality modeling (see Table D-4 for the emission differences)).34 This calibrated
change in ozone is then added to the ozone contribution from the 2017 base case final air quality
modeling. The result is the state and receptor specific "calibrated" total ozone contribution after
implementation of the emission budgets at a particular level of control.

For each monitor, these state-level "calibrated" contributions are then summed to estimate total
ozone contribution from the states to a particular receptor in the CAMx modeling domain. Finally,
"other" modeled ozone contributions ("TRIBAL", "CNMX", "OFFSHORE", "FIRE", "ICBC", and
"BIOG") are added from the 2017 base case final air quality modeling to the state contributions to
account for other sources of ozone affecting the modeling domain. The total ozone from all the states
and "other" contributions equals the average design values estimated in the assessment tool. The
maximum design values were estimated by multiplying the estimated average design values by the
ratio of the modeled 2017 base case maximum to average design values.

Generally, as the emission budget stringency increased, the estimated average and maximum
design values at each receptor decreased. In the assessment tool, the estimated value of the average
design value was used to estimate whether the location will be out of attainment, while the estimated
maximum design value was used to estimate whether the location will have problems maintaining the
NAAQS. The area was noted as having a nonattainment or maintenance issue if either estimated air
quality level was greater than or equal to 76 ppb.

34 The change in concentration can be positive or negative, depending on whether the state's emissions are larger or smaller
than the 2017 air quality modeling base case emission level.

28


-------
Table D-3. 2017 Ozone Season EGU NOx Emissions (Tons) for Each State at Various Emission







Final



















Base

















Final

Case

















Base

$0/ton

















Case

Emissio

















$0/ton

n











Final



Air

Emissi

Budgets

Final

Final

Final

Final

Final

$l,400/ton



Quality

on

with

$800/ton

$1400/ton

$3400/ton

$5000/ton

$6400/ton

Illustrative



Modeling

Budget

PA

Emission

Emission

Emission

Emission

Emission

Scenario



Base Case

s

RACT

Budgets

Budgets

Budgets

Budgets

Budgets

AQ Model

Alabama

10,902

15,179

15,179

14,332

13,211

12,620

11,928

11,573

8,521

Arkansas

9,890

12,560

12,560

12,048

9,210

9,048

8,518

8,050

6,862

Arizona

8,328

10,710

10,710

10,680

9,810

9,788

9,710

9,486

7,766

California

1,624

1,905

1,928

1,905

1,905

1,905

1,810

1,810

1,626

Colorado

13,426

14,010

14,010

14,008

13,994

13,645

13,495

12,950

13,434

Connecticut

387

605

605

584

558

558

554

554

386

Delaware

155

497

497

497

497

494

494

494

155

Florida

24,617

22,779

22,779

21,611

17,123

16,631

16,481

16,375

24,374

Georgia

11,120

8,762

8,762

8,495

8,481

8,525

8,532

7,764

8,840

Iowa

11,127

11,478

11,478

11,477

11,272

11,065

10,891

10,491

10,729

Idaho

15

152

152

152

152

152

152

152

16

Illinois

13,252

14,850

14,850

14,682

14,601

14,515

14,248

14,054

14,130

Indiana

40,223

31,382

31,382

28,960

23,303

21,634

19,990

18,720

26,047

Kansas

11,257

8,031

8,031

8,030

8,027

7,975

7,962

7,767

9,416

Kentucky

27,500

26,318

26,318

24,052

21,115

21,007

20,273

19,496

18,535

Louisiana

9,824

19,101

19,101

19,096

18,639

18,452

18,442

18,426

9,877

Massachusetts

939

1,119

1,119

1,119

1,112

1,098

1,071

1,072

961

Maryland

3,218

3,871

3,871

3,870

3,828

3,308

2,938

2,926

3,509

Maine

188

109

109

109

109

109

109

109

188

Michigan

21,415

19,811

19,811

19,558

17,023

15,782

13,110

12,612

17,524

Minnesota

9,710

7,068

7,068

7,068

6,864

6,761

6,651

6,451

8,596

Missouri

15,836

18,443

18,443

17,250

15,780

15,299

14,673

14,555

18,125

Mississippi

7,793

6,438

6,438

6,438

6,315

6,243

6,203

6,205

7,498

Montana

7,151

6,540

6,540

6,540

6,540

6,535

6,535

6,535

7,212

North Carolina

19,713

17,419

17,419

14,424

14,326

13,189

12,460

12,207

19,469

North Dakota

14,392

18,738

18,738

18,398

18,016

17,655

17,497

17,435

11,557

Nebraska

12,196

9,737

9,737

9,678

8,970

7,197

6,467

6,448

12,195

New Hampshire

140

416

416

416

416

416

415

415

140

New Jersey

1,776

2,114

2,114

2,100

2,062

2,008

1,867

1,879

1,731

New Mexico

5,626

9,443

9,443

9,443

8,834

8,633

8,367

8,219

5,735

Nevada

3,597

2,405

2,405

2,301

2,241

2,112

1,559

886

3,339

New York

4,275

5,531

5,531

5,220

5,135

5,006

4,746

4,594

3,783

Ohio

27,038

27,382

27,384

23,659

19,522

19,165

18,561

18,348

18,434

Oklahoma

16,718

13,747

13,747

13,746

11,641

9,174

8,790

8,439

14,642

Oregon

-

—

0

—

—

—

—

—

-

Pennsylvania

39,987

35,607

20,200

17,514

15,452

15,428

15,121

14,874

14,461

Rhode Island

180

283

283

283

283

283

283

283

180

South Carolina

5,839

5,486

5,486

5,288

5,288

5,293

5,300

5,318

5,824

South Dakota

537

853

853

853

853

853

853

853

537

Tennessee

6,944

7,779

7,779

7,736

7,736

7,735

7,724

7,729

7,129

Texas

56,331

54,839

54,839

54,521

52,301

50,011

48,795

47,994

54,345

Utah

21,618

16,949

16,949

16,949

14,149

13,592

11,356

11,324

21,616

Virginia

2,586

9,367

9,367

9,365

9,223

8,754

8,619

8,416

2,858

Vermont

0

52

52

52

52

52

52

52

0

Washington

136

3,085

3,085

3,085

3,085

3,085

3,085

3,085

136

Wisconsin

6,488

7,939

7,939

7,924

7,915

7,790

7,435

7,023

6,548

West Virginia

26,110

26,874

26,874

25,984

17,815

17,380

17,388

17,373

16,298

Wyoming

11,216

16,005

16,005

15,828

14,691

13,728

12,371

12,121

11,290

* EPA analyzed all segments of the PA RACT Rule. For the EGU sector, this was primarily a cap on emission rates at coal EGUs with SCRs (12,848
tons), but also included a number of less impactful provisions on EGU emission rates (59 tons). This results in 12,907 tons of reduction from EGU PA
RACT. These EGU PA Ract reductions are only included in the PA estimates for the final base case $0/ton emission budgets with PA RACT and the final
$800/ton emissions budgets cases. A separate 2,500 tons of reductions from non-EGU PA RACT is included in the PA estimates for all cases except the
air quality modeling base case, the final base case, and the final $l,400/ton illustrative scenario AQ model.

29


-------
Table D-4. 2017 Ozone Season EGU NOx Emission Differences (Thousand Tons) for Each State

Relative to the Air Qua

ity Modeling Base Case Emission

^evel as

Modeled in AQAr

•









Final Base



















Final Base

Case $0/ton











Final



Total



Case

Emission

Final

Final

Final

Final

Final

$l,400/ton



Antro-

Final

$0/ton

Budgets

$800/ton

$1400/ton

$3400/ton

$5000/ton

$6400/ton

Illustrative



pogenic

Base

Emission

with

Emission

Emission

Emission

Emission

Emission

Scenario



NOx

Case

Budgets

PA RACT

Budgets

Budgets

Budgets

Budgets

Budgets

AQ Model

Alabama

88.3

-0.5

4.3

4.3

3.4

2.3

1.7

1.0

0.7

-2.4

Arizona

67.3

-2.4

2.4

2.4

2.4

1.5

1.5

1.4

1.2

-0.6

Arkansas

68.1

0.0

2.7

2.7

2.2

-0.7

-0.8

-1.4

-1.8

-3.0

California

210.4

0.0

0.3

0.3

0.3

0.3

0.3

0.2

0.2

0.0

Colorado

89.3

0.0

0.6

0.6

0.6

0.6

0.2

0.1

-0.5

0.0

Connecticut

17.5

0.0

0.2

0.2

0.2

0.2

0.2

0.2

0.2

0.0

Delaware

7.7

0.0

0.3

0.3

0.3

0.3

0.3

0.3

0.3

0.0

District of Columbia

2.3

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

Florida

173.9

-0.2

-1.8

-1.8

-3.0

-7.5

-8.0

-8.1

-8.2

-0.2

Georgia

105.2

-2.3

-2.4

-2.4

-2.6

-2.6

-2.6

-2.6

-3.4

-2.3

Idaho

28.0

0.0

0.1

0.1

0.1

0.1

0.1

0.1

0.1

0.0

Illinois

146.0

1.3

1.6

1.6

1.4

1.3

1.3

1.0

0.8

0.9

Indiana

136.8

2.0

-8.8

-8.8

-11.3

-16.9

-18.6

-20.2

-21.5

-14.2

Iowa

73.4

-0.2

0.4

0.4

0.3

0.1

-0.1

-0.2

-0.6

-0.4

Kansas

109.0

-0.4

-3.2

-3.2

-3.2

-3.2

-3.3

-3.3

-3.5

-1.8

Kentucky

95.9

-4.2

-1.2

-1.2

-3.4

-6.4

-6.5

-7.2

-8.0

-9.0

Louisiana

168.9

0.2

9.3

9.3

9.3

8.8

8.6

8.6

8.6

0.1

Maine

17.8

0.0

-0.1

-0.1

-0.1

-0.1

-0.1

-0.1

-0.1

0.0

Maryland

44.7

0.4

0.7

0.7

0.7

0.6

0.1

-0.3

-0.3

0.3

Massachusetts

35.4

0.0

0.2

0.2

0.2

0.2

0.2

0.1

0.1

0.0

Michigan

121.9

-0.9

-1.6

-1.6

-1.9

-4.4

-5.6

-8.3

-8.8

-3.9

Minnesota

89.5

-1.1

-2.6

-2.6

-2.6

-2.8

-2.9

-3.1

-3.3

-1.1

Mississippi

54.4

-0.2

-1.4

-1.4

-1.4

-1.5

-1.6

-1.6

-1.6

-0.3

Missouri

100.9

4.8

2.6

2.6

1.4

-0.1

-0.5

-1.2

-1.3

2.3

Montana

37.4

0.0

-0.6

-0.6

-0.6

-0.6

-0.6

-0.6

-0.6

0.1

Nebraska

68.2

0.0

-2.5

-2.5

-2.5

-3.2

-5.0

-5.7

-5.7

0.0

Nevada

27.9

-0.2

-1.2

-1.2

-1.3

-1.4

-1.5

-2.0

-2.7

-0.3

New Hampshire

8.9

0.0

0.3

0.3

0.3

0.3

0.3

0.3

0.3

0.0

New Jersey

49.8

0.0

0.3

0.3

0.3

0.3

0.2

0.1

0.1

0.0

New Mexico

66.2

0.0

3.8

3.8

3.8

3.2

3.0

2.7

2.6

0.1

New York

106.3

-0.1

1.3

1.3

0.9

0.9

0.7

0.5

0.3

-0.5

North Carolina

97.0

0.7

-2.3

-2.3

-5.3

-5.4

-6.5

-7.3

-7.5

-0.2

North Dakota

65.8

-2.8

4.3

4.3

4.0

3.6

3.3

3.1

3.0

-2.8

Ohio

148.7

3.1

0.3

0.3

-3.4

-7.5

-7.9

-8.5

-8.7

-8.6

Oklahoma

124.7

-0.2

-3.0

-3.0

-3.0

-5.1

-7.5

-7.9

-8.3

-2.1

Oregon

40.3

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

Pennsylvania

165.2

-5.0

-4.4

-19.8

-22.5

-24.5

-24.6

-24.9

-25.1

-25.5

Rhode Island

5.9

0.0

0.1

0.1

0.1

0.1

0.1

0.1

0.1

0.0

South Carolina

56.4

-0.1

-0.4

-0.4

-0.6

-0.6

-0.5

-0.5

-0.5

0.0

South Dakota

22.3

0.0

0.3

0.3

0.3

0.3

0.3

0.3

0.3

0.0

Tennessee

87.5

0.4

0.8

0.8

0.8

0.8

0.8

0.8

0.8

0.2

Texas

435.5

-0.4

-1.5

-1.5

-1.8

-4.0

-6.3

-7.5

-8.3

-2.0

Utah

62.8

0.0

-4.7

-4.7

-4.7

-7.5

-8.0

-10.3

-10.3

0.0

Vermont

5.4

0.0

0.1

0.1

0.1

0.1

0.1

0.1

0.1

0.0

Virginia

81.6

0.2

6.8

6.8

6.8

6.6

6.2

6.0

5.8

0.3

Washington

76.1

0.0

2.9

2.9

2.9

2.9

2.9

2.9

2.9

0.0

West Virginia

65.9

0.3

0.8

0.8

-0.1

-8.3

-8.7

-8.7

-8.7

-9.8

Wisconsin

74.2

0.1

1.5

1.5

1.4

1.4

1.3

0.9

0.5

0.1

Wyoming

65.8

0.1

4.8

4.8

4.6

3.5

2.5

1.2

0.9

0.1

* EPA analyzed all segments of the PA RACT Rule. For the EGU sector, this was primarily a cap on emission rates at coal EGUs with SCRs (12,848
tons), but also included a number of less impactful provisions on EGU emission rates (59 tons). This results in 12,907 tons of reduction from EGU PA
RACT. These EGU PA Ract reductions are only included in the PA estimates for the final base case $0/ton emission budgets with PA RACT and the final
$800/ton emissions budgets cases. A separate 2,500 tons of reductions from non-EGU PA RACT is included in the PA estimates for all cases except the
air quality modeling base case, the final base case, and the final $l,400/ton illustrative scenario AQ model.

30


-------
Table D-5. 2017 Ozone Season EGU NOx Emission Fractions for Each State Relative to the Air

Quality Modeling Base

Case Emission Level as Mode

ed in AC

~AT.





Final Base

Final Base











Final





Case

Case $0/ton

Final

Final

Final

Final

Final

$l,400/ton



Final

$0/ton

Emission

$800/ton

$1400/ton

$3400/ton

$5000/ton

$6400/ton

Illustrative



Base

Emission

Budgets with

Emission

Emission

Emission

Emission

Emission

Scenario AQ



Case

Budgets

PA RACT

Budgets

Budgets

Budgets

Budgets

Budgets

Model

Alabama

-0.0056

0.0485

0.0485

0.0389

0.0262

0.0195

0.0116

0.0076

-0.0270

Arizona

-0.0351

0.0354

0.0354

0.0349

0.0220

0.0217

0.0205

0.0172

-0.0084

Arkansas

0.0000

0.0392

0.0392

0.0317

-0.0100

-0.0124

-0.0201

-0.0270

-0.0445

California

0.0000

0.0013

0.0014

0.0013

0.0013

0.0013

0.0009

0.0009

0.0000

Colorado

0.0001

0.0065

0.0065

0.0065

0.0064

0.0024

0.0008

-0.0053

0.0001

Connecticut

0.0000

0.0125

0.0125

0.0113

0.0098

0.0098

0.0096

0.0096

0.0000

Delaware

0.0000

0.0445

0.0445

0.0445

0.0445

0.0441

0.0441

0.0441

0.0000

District of Columbia

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

Florida

-0.0013

-0.0106

-0.0106

-0.0173

-0.0431

-0.0459

-0.0468

-0.0474

-0.0014

Georgia

-0.0214

-0.0224

-0.0224

-0.0250

-0.0251

-0.0247

-0.0246

-0.0319

-0.0217

Idaho

0.0000

0.0049

0.0049

0.0049

0.0049

0.0049

0.0049

0.0049

0.0000

Illinois

0.0089

0.0109

0.0109

0.0098

0.0092

0.0087

0.0068

0.0055

0.0060

Indiana

0.0143

-0.0646

-0.0646

-0.0824

-0.1237

-0.1359

-0.1479

-0.1572

-0.1036

Iowa

-0.0025

0.0048

0.0048

0.0048

0.0020

-0.0008

-0.0032

-0.0087

-0.0054

Kansas

-0.0032

-0.0296

-0.0296

-0.0296

-0.0296

-0.0301

-0.0302

-0.0320

-0.0169

Kentucky

-0.0440

-0.0123

-0.0123

-0.0360

-0.0666

-0.0677

-0.0754

-0.0835

-0.0935

Louisiana

0.0011

0.0549

0.0549

0.0549

0.0522

0.0511

0.0510

0.0509

0.0003

Maine

0.0000

-0.0044

-0.0044

-0.0044

-0.0044

-0.0044

-0.0044

-0.0044

0.0000

Maryland

0.0081

0.0146

0.0146

0.0146

0.0136

0.0020

-0.0063

-0.0065

0.0065

Massachusetts

0.0005

0.0051

0.0051

0.0051

0.0049

0.0045

0.0037

0.0038

0.0006

Michigan

-0.0075

-0.0132

-0.0132

-0.0152

-0.0360

-0.0462

-0.0681

-0.0722

-0.0319

Minnesota

-0.0124

-0.0295

-0.0295

-0.0295

-0.0318

-0.0329

-0.0342

-0.0364

-0.0124

Mississippi

-0.0031

-0.0249

-0.0249

-0.0249

-0.0272

-0.0285

-0.0292

-0.0292

-0.0054

Missouri

0.0472

0.0258

0.0258

0.0140

-0.0006

-0.0053

-0.0115

-0.0127

0.0227

Montana

0.0006

-0.0164

-0.0164

-0.0164

-0.0164

-0.0165

-0.0165

-0.0165

0.0016

Nebraska

0.0000

-0.0361

-0.0361

-0.0369

-0.0473

-0.0733

-0.0840

-0.0843

0.0000

Nevada

-0.0072

-0.0428

-0.0428

-0.0465

-0.0487

-0.0533

-0.0732

-0.0973

-0.0092

New Hampshire

0.0000

0.0309

0.0309

0.0309

0.0309

0.0309

0.0308

0.0308

0.0000

New Jersey

-0.0001

0.0068

0.0068

0.0065

0.0057

0.0047

0.0018

0.0021

-0.0009

New Mexico

0.0004

0.0576

0.0576

0.0576

0.0484

0.0454

0.0414

0.0392

0.0016

New York

-0.0011

0.0118

0.0118

0.0089

0.0081

0.0069

0.0044

0.0030

-0.0046

North Carolina

0.0072

-0.0236

-0.0236

-0.0545

-0.0555

-0.0672

-0.0748

-0.0774

-0.0025

North Dakota

-0.0422

0.0660

0.0660

0.0609

0.0551

0.0496

0.0472

0.0462

-0.0431

Ohio

0.0210

0.0023

0.0023

-0.0227

-0.0505

-0.0529

-0.0570

-0.0584

-0.0579

Oklahoma

-0.0017

-0.0238

-0.0238

-0.0238

-0.0407

-0.0605

-0.0636

-0.0664

-0.0167

Oregon

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

Pennsylvania

-0.0306

-0.0265

-0.1198

-0.1360

-0.1485

-0.1486

-0.1505

-0.1520

-0.1545

Rhode Island

0.0000

0.0174

0.0174

0.0174

0.0174

0.0174

0.0174

0.0174

0.0000

South Carolina

-0.0016

-0.0063

-0.0063

-0.0098

-0.0098

-0.0097

-0.0096

-0.0092

-0.0003

South Dakota

0.0000

0.0142

0.0142

0.0142

0.0142

0.0142

0.0142

0.0142

0.0000

Tennessee

0.0050

0.0095

0.0095

0.0091

0.0091

0.0090

0.0089

0.0090

0.0021

Texas

-0.0009

-0.0034

-0.0034

-0.0042

-0.0093

-0.0145

-0.0173

-0.0191

-0.0046

Utah

0.0000

-0.0744

-0.0744

-0.0744

-0.1190

-0.1279

-0.1635

-0.1640

0.0000

Vermont

0.0000

0.0096

0.0095

0.0096

0.0096

0.0096

0.0096

0.0096

0.0000

Virginia

0.0026

0.0831

0.0831

0.0831

0.0813

0.0756

0.0739

0.0715

0.0033

Washington

0.0000

0.0388

0.0388

0.0388

0.0388

0.0388

0.0388

0.0388

0.0000

West Virginia

0.0048

0.0116

0.0116

-0.0019

-0.1259

-0.1325

-0.1324

-0.1327

-0.1490

Wisconsin

0.0009

0.0195

0.0195

0.0193

0.0192

0.0175

0.0128

0.0072

0.0008

Wyoming

0.0009

0.0728

0.0728

0.0701

0.0528

0.0382

0.0176

0.0138

0.0011

* EPA analyzed all segments of the PA RACT Rule. For the EGU sector, this was primarily a cap on emission rates at coal EGUs with SCRs (12,848
tons), but also included a number of less impactful provisions on EGU emission rates (59 tons). This results in 12,907 tons of reduction from EGU PA
RACT. These EGU PA Ract reductions are only included in the PA estimates for the final base case $0/ton emission budgets with PA RACT and the final
$800/ton emissions budgets cases. A separate 2,500 tons of reductions from non-EGU PA RACT is included in the PA estimates for all cases except the
air quality modeling base case, the final base case, and the final $l,400/ton illustrative scenario AQ model.

31


-------
3. Description of the results of the analysis using the assessment tool for the approach.

EPA used the ozone AQAT to estimate improvements in downwind air quality at base case
levels, then $800 per ton and, then, at higher emission budget levels . At each emission budget level,
using AQAT, EPA examined whether the average and maximum design values for each of the
receptors decreased to values below 76 ppb at which point their nonattainment and maintenance issues
would be considered resolved. EPA also examined each states' air quality contributions at each
emission budget level, assessing whether a state maintained at least one linkage to a receptor that was
estimated to remain in nonattainment and/or maintenance. EPA examined emission budget levels of
$0/ton, $0/ton including adjustments due to PA RACT, $800/ton, $l,400/ton, $3,400/ton, $5,000/ton,
and $6,400/ton. PA RACT was included in all of the scenarios that were more stringent than the
$0/ton including adjustments due to PA RACT scenario. The preamble explains at section VI.D how
EPA considered the results of the air quality analyses described in this TSD to determine the
appropriate set of emission budgets for reducing significant contribution to nonattainment and
interference with maintenance.

The average and maximum design values (ppb) estimated using the assessment tool for each
identified receptor for each emission budget level have been truncated to a tenth of a ppb and can be
found in Tables D-6 and D-7, respectively. The monitors are in alphabetical order by state. Notably,
no monitors are projected to have either their average or maximum design values drop below 76 ppb in
the transition from the air quality modeling base case to the adjusted historic base cases (i.e., $0/ton
emissions budget with or without PA RACT). At each of the NOx emission budget levels examined,
we found that only one additional monitor is projected to have resolved its average design value
problems (i.e., nonattainment). We project that monitor 211110067 in Kentucky is resolved at the
$l,400/ton threshold, meaning that we estimate its average design value will drop below 76 ppb.

Many monitors are projected to have maintenance issues at all emission budget levels.
However, three monitors have their maintenance issues solved at various emission budget levels. At
the $800/ton emission budget level, we estimate that the maintenance problems at monitor 421010024
in Pennsylvania to be resolved (when PA RACT is also considered). Delaware is solely linked to this
receptor (see the discussion of Delaware in preamble section VI). At the $l,400/ton emission budget
level, two additional monitors are projected to have their maintenance concerns resolved. These are
monitors 211110067 in Kentucky and 390610006 in Ohio. Tennessee is uniquely linked to the latter
receptor in Ohio following the resolution of its other linkage (monitor 421010024 in Pennsylvania)
which is projected to occur at the less stringent $800/ton emission budget. See the discussion of
Tennessee in preamble section VI for additional details.

We observe no additional change in receptor status between the $3,400/ton and $6,400/ton
emission budget levels.

In the assessment of air quality using the calibrated assessment tool, we are able to estimate the
change in the air quality contributions of each upwind state to each receptor (see the description of the
state and receptor-specific contributions in section D.2.c.(2)) in order to determine whether any state's
contribution is below the 1 percent threshold used in step 2 of the CSAPR framework to identify
"linked" upwind states. For this over-control assessment, we compared each state's adjusted ozone
concentration against the 1% air quality threshold at each of the emission budgets levels up to
$6,400/ton at each remaining receptor, using AQAT (see the "links" worksheets in the AQAT

32


-------
workbook, referred to in Appendix B). For Delaware at $800/ton and Tennessee at $l,400/ton, where
their final monitors have all of their nonattainment and maintenance issues resolved, the reductions
required by the emissions budgets would not reduce their contributions below the 1% threshold. For
all other linked states, we did not see instances where a state's contributions dropped below 1% of the
NAAQS for all of its linkages to downwind receptors. This is not a surprising result because, for a
linkage to be resolved by emission reductions of just a few percent, the contribution would need to be
within a few percent of the threshold. As a hypothetical example, if the state is making a 6% emission
reduction in its overall anthropogenic ozone season NOx emissions, and the calibration factor was 0.5,
its base case maximum contribution to a remaining unresolved nonattainment and/or maintenance
receptor would need to be just under 1.03% of the NAAQS or 0.77 ppb, to drop below the 0.75 ppb
linkage threshold.

Lastly, once the budgets for the rule were established (based on the results of the multi-factor
test) and IPM was used to model compliance with the rule, it was possible to estimate air quality
concentrations at each downwind receptor using the ozone AQAT for the $1,400 final illustrative
control air quality case. This scenario was directly modeled in CAMx. The average and maximum
design value estimates from AQAT and CAMx can be found in Table D-8. The results are described
in the following section (D.4). The design value results from AQAT (i.e., which receptors are
estimated to have nonattainment and/or maintenance problems) for the final emissions budgets
scenario are similar to that of the $l,400/ton emission budget level.

33


-------
Table D-6. Average Ozone DVs (ppb) for NOx Emissions Budget Levels ($/ton) Assessed Using

Monitor
Identification
Number

State

County

CAMx
2017 Base
Case (ppb)

Assessment Tool Average Ozone Design Values (ppb).

Adjusted
Historic Base
Case

Adjusted Historic
Base Case w/ PA
RACT

$800

$1,400

$3,400

$5,000

$6,400

90010017

Connecticut

Fairfield

74.1

74.1

73.9

73.9

73.8

73.8

73.8

73.8

90013007

Connecticut

Fairfield

75.5

75.5

75.3

75.3

75.2

75.2

75.2

75.2

90019003

Connecticut

Fairfield

76.5

76.5

76.3

76.3

76.2

76.2

76.2

76.2

90099002

Connecticut

New Flaven

76.2

76.2

76.1

76.1

76.0

76.0

76.0

76.0

211110067

Kentucky

Jefferson

76.9

76.4

76.4

76.0

75.4

75.3

75.1

75.0

240251001

Maryland

Flarford

78.8

79.0

78.8

78.7

78.4

78.3

78.2

78.2

260050003

Michigan

Allegan

74.7

74.6

74.6

74.6

74.4

74.4

74.3

74.3

360850067

New York

Richmond

75.8

75.7

75.3

75.2

75.0

74.9

74.9

74.9

361030002

New York

Suffolk

76.8

76.8

76.6

76.6

76.5

76.5

76.5

76.5

390610006

Ohio

Flamilton

74.6

74.1

74.1

73.3

72.1

71.9

71.7

71.5

421010024

Pennsylvania

Philadelphia

73.6

73.4

72.7

72.5

72.2

72.1

72.1

72.0

480391004

Texas

Brazoria

79.9

79.9

79.9

79.9

79.8

79.7

79.7

79.7

481210034

Texas

Denton

75.0

74.9

74.9

74.9

74.9

74.8

74.8

74.8

482010024

Texas

Flarris

75.4

75.4

75.4

75.3

75.3

75.3

75.3

75.3

482011034

Texas

Flarris

75.7

75.7

75.7

75.7

75.6

75.6

75.6

75.5

482011039

Texas

Flarris

76.9

76.9

76.9

76.9

76.8

76.8

76.8

76.7

484392003

Texas

Tarrant

77.3

77.2

77.2

77.2

77.2

77.1

77.1

77.0

484393009

Texas

Tarrant

76.4

76.3

76.3

76.3

76.3

76.2

76.2

76.2

551170006

Wisconsin

Sheboygan

76.2

76.2

76.2

76.1

76.1

76.0

76.0

76.0

Table D-7. Maximum Ozone DVs (ppb) for NOx Emissions Budget Levels ($/ton) Assessed Using
the Ozone AQAT for all Nineteen Nonattainment and Maintenance Receptors.	

Monitor
Identification
Number

State

County

CAMx 2017
Base Case
(PPb)

Assessment Tool Maximum Ozone Design Values (ppb).

Adjusted
Historic Base
Case

Adjusted Historic
Base Case w/ PA
RACT

$800

$1,400

$3,400

$5,000

$6,400

90010017

Connecticut

Fairfield

76.6

76.6

76.4

76.4

76.3

76.3

76.3

76.3

90013007

Connecticut

Fairfield

79.7

79.7

79.5

79.5

79.4

79.4

79.4

79.3

90019003

Connecticut

Fairfield

79.5

79.5

79.3

79.3

79.2

79.2

79.2

79.2

90099002

Connecticut

New Flaven

79.2

79.1

79.0

79.0

79.0

79.0

78.9

78.9

211110067

Kentucky

Jefferson

76.9

76.4

76.4

76.0

75.4

75.3

75.1

75.0

240251001

Maryland

Flarford

81.4

81.6

81.4

81.3

81.0

80.9

80.8

80.8

260050003

Michigan

Allegan

77.7

77.6

77.6

77.5

77.4

77.3

77.3

77.2

360850067

New York

Richmond

77.4

77.3

76.9

76.7

76.5

76.5

76.5

76.4

361030002

New York

Suffolk

78.4

78.3

78.2

78.2

78.1

78.1

78.1

78.1

390610006

Ohio

Flamilton

77.4

76.8

76.8

76.0

74.7

74.6

74.3

74.1

421010024

Pennsylvania

Philadelphia

76.9

76.7

76.0

75.7

75.4

75.3

75.3

75.2

480391004

Texas

Brazoria

80.8

80.8

80.8

80.8

80.7

80.7

80.6

80.6

481210034

Texas

Denton

77.4

77.3

77.3

77.3

77.3

77.2

77.2

77.2

482010024

Texas

Flarris

77.9

77.9

77.9

77.9

77.9

77.8

77.8

77.8

482011034

Texas

Flarris

76.6

76.6

76.6

76.6

76.5

76.5

76.5

76.5

482011039

Texas

Flarris

78.8

78.7

78.7

78.7

78.7

78.7

78.6

78.6

484392003

Texas

Tarrant

79.7

79.6

79.6

79.6

79.5

79.5

79.4

79.4

484393009

Texas

Tarrant

76.4

76.3

76.3

76.3

76.3

76.2

76.2

76.2

551170006

Wisconsin

Sheboygan

78.7

78.6

78.6

78.6

78.5

78.5

78.4

78.4

34


-------
4. Comparison between the air quality assessment tool estimates and CAMx air quality modeling
estimates.

As described earlier, because the AQAT was calibrated using data from the proposal, it was
possible to evaluate the estimates from the tool with the independent model design value predictions
from CAMx for the final modeling 2017 illustrative control scenario. The average and maximum
design values from AQAT and CAMx, as well as the differences, are shown in Table D-8. The AQAT
values and the differences in the table have been truncated to a tenth of a ppb.

Strong correlations are observed (nearly one to one) between the estimated average and
maximum design values from AQAT and CAMx (Figure 2). The slopes of the least-squares linear
regression lines are almost exactly equal to 1, and the regression coefficients are also nearly 1. The
differences in estimates are also small, averaging less than 0.1 ppb and reaching a maximum of about
0.2 ppb for several receptors. These small differences suggest that each of the state contributions are
being reasonably approximated. The results of this demonstrate that, considering the constraints faced
by the EPA, the AQAT provides reasonable estimates of air quality concentrations for each receptor,
can provide reasonable inputs for the multi-factor assessment, and can serve as a method to test for
linkages dropping below the threshold.

Table D-8. Average and Maximum DVs (ppb) in the 2017 Budget Control Scenario as Modeled
in CAMx and as Estimated in Calibrated AQAT, for Receptors with Maximum DVs Greater
than or Equal to 76 ppb in the 2017 Base Case Modeled in CAMx.	

Monitor

















Identification

State

County





2017 Budget Control Scenario



Numbe































Difference,

Difference,







CAMx

CAMx

AQAT

AQAT

Avg. DV

Max. DV







Avg. DV

Max. DV

Avg. DV

Max. DV

(CAMx-
AQAT)

(CAMx-
AQAT)

90010017

Connecticut

Fairfield

73.7

76.2

73.7

76.1

0.0

0.1

90013007

Connecticut

Fairfield

75.0

79.2

75.1

79.3

-0.1

-0.1

90019003

Connecticut

Fairfield

76.0

79.0

76.1

79.1

-0.1

-0.1

90099002

Connecticut

New Haven

76.0

78.9

75.9

78.9

0.1

0.0

211110067

Kentucky

Jefferson

75.1

75.1

75.1

75.1

0.0

0.0

240251001

Maryland

Harford

78.0

80.6

78.1

80.8

-0.1

-0.2

260050003

Michigan

Allegan

74.4

77.4

74.4

77.4

0.0

0.0

360850067

New York

Richmond

74.8

76.4

74.8

76.3

0.0

0.1

361030002

New York

Suffolk

76.5

78.0

76.4

78.0

0.1

0.0

390610006

Ohio

Hamilton

71.7

74.3

71.7

74.3

0.0

0.0

421010024

Pennsylvania

Philadelphia

72.1

75.3

71.9

75.1

0.2

0.2

480391004

Texas

Brazoria

79.8

80.7

79.8

80.7

0.0

0.0

481210034

Texas

Denton

74.9

77.3

74.9

77.3

0.0

0.0

482010024

Texas

Harris

75.4

77.9

75.3

77.8

0.1

0.1

482011034

Texas

Harris

75.6

76.5

75.6

76.5

0.0

0.0

482011039

Texas

Harris

76.8

78.7

76.8

78.7

0.0

0.0

484392003

Texas

Tarrant

77.1

79.5

77.1

79.5

0.0

0.0

484393009

Texas

Tarrant

76.3

76.3

76.3

76.3

0.0

0.0

551170006

Wisconsin

Sheboygan

76.0

78.5

76.0

78.4

0.0

0.1

35


-------
Average DVs	Maximum DVs

81.0
80.0
79.0
78.0

Q.

3 77.0
q 76.0

i 75 0 I	f

y = l.OOOlx





ft2 = 0.9984

















74.0

73.0	...•*•

72.0 	ml'""'	

71.0 I	

71.0 72.0 73.0 74.0 75.0 76.0 77.0 78.0 79.0 80.0 81.0	73.0 74.0 75.0 76.0 77.0 78.0 79.0

AQAT DV (ppb)	AQAT DV (ppb)



82.0



81.0



80.0

-Q

CL

79.0

Q.



>

78.0

o



X

77.0





<

76.0

u





75.0



74.0



73.0









y = lioOOlx





R2 = 0.9977 I	— . -







	*7*	









































Figure 2. Least squares linear regression plots showing correlations between CAMx and calibrated
AQAT for the 2017 illustrative budget control scenario base case for estimated average and maximum
design values (ppb) in the left and right panels, respectively.

36


-------
Appendix A: IPM Runs Used in Transport Rule Significant

Contribution Analysis

37


-------
Table A-l lists IPM runs used in analysis for this rule. The IPM runs can be found in the
docket for this rulemaking under the IPM File Name listed in square brackets in the table below

Table Appendix A-l. IPM Runs Used in Transport Rule Significant Contribution Analysis

Run Name
[IPM File Namel

Description

Air Quality Modeling Base Case

[5.15_OS_NOx_AQM_Base_Case]

Model run used for the air quality modeling base case, which
includes the national Title IV S02 cap-and-trade program; NOx
SIP Call; the Cross-State Air Pollution trading programs, and
settlements and state rules. This is based on AEO estimates from
2015 and excludes the final Clean Power Plan. It also includes key
fleet updates regarding new units, retired units, and control
retrofits, particularly those raised in comments to the agency on
the August 4, 2015 NOD A6.

Illustrative Base Case

[5.15_OS_NOx_Illustrative_Base_Case]

Model run used as the base case for the Illustrative Analysis of the
final rule. Based upon the 5.15_OS_NOx_AQM_Base_Case case,
but incorporating comments the agency received on the model,
particularly those that affect NOx emissions and NOx rates.

Illustrative $800 per ton Cost Threshold

[5.15 OS NOx Illustrative 800 CT]

Imposes a marginal cost of $800 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation".

Illustrative $1,400 per ton Cost Threshold

[5.15_OS_NOx_Illustrative_1400_CT]

Imposes a marginal cost of $1,400 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls.

Illustrative $3,400 per ton Cost Threshold

[5.15_OS_NOx_Illustrative_3400_CT]

Imposes a marginal cost of $3,400 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls. Units
with SNCR "fully operate" those controls.

Illustrative $5,000 per ton Cost Threshold

[5.15_OS_NOx_Illustrative_5000_CT]

Imposes a marginal cost of $5,000 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls. Units
with SNCR "fully operate" those controls.

Illustrative $6,400 per ton Cost Threshold

[5.15_OS_NOx_Illustrative_6400_CT]

Imposes a marginal cost of $6,400 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls. Units
with SNCR "fully operate" those controls.

Illustrative Less Stringent Policy Case

[5.15_OS_NOx_Illustrative_Less_Stringent]

Imposes the budgets with variability limits derived from the $800
per ton of NOx case were applied to states covered by the final
rule. Units in covered states with extant and operating SCRs
operate them at "full operation." This run is also called the
"Illustrative Less Stringent Case"

38


-------
Illustrative Policy Case

[5.15_OS_NOx_Illustrative_Policy]

Imposes the budgets with variability limits derived from the
$1,400 per ton of NOx case were applied to states covered by the
final rule. Units in covered states with SCRs operate them at "full
operation" and units with SCRs that are not operating return them
to "full operation". Units without SOA combustion controls
upgraded to SOA combustion controls. This run is also called the
"Illustrative Policy Case". This case was used for the Illustrative
Policy Air Quality Modeling analysis.

Illustrative More Stringent Policy Case

[5.15_OS_NOx_Illustrative_More_Stringent]

Imposes the budgets with variability limits derived from the
$3,400 per ton of NOx case were applied to states covered by the
final rule. Unit in covered states with SCRs operate them at "full
operation" and units with SCRs that are not operating return them
to "full operation". Units without SOA combustion controls
upgraded to SOA combustion controls. Units with SNCR "fully
operate" those controls. This run is also called the "Illustrative
More Stringent Case"

Final Base Case

[5.15_OS_NOx_Final_Base_Case]

Model run used as the base case for setting the final state budgets
for this rule. Based upon the

5.15_OS_NOx_Illustrative_Base_Case case, incorporates updated
NOx rates for units that altered behavior significantly in 2015 as
compared to 2011.

Final $800 per ton Cost Threshold

[5.15 OS NOx Final 800 CT]

Imposes a marginal cost of $800 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation".

Final $1,400 per ton Cost Threshold

[5.15_OS_NOx_Final_1400_CT]

Imposes a marginal cost of $1,400 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls.

Final $3,400 per ton Cost Threshold

[5.15_OS_NOx_Final_3400_CT]

Imposes a marginal cost of $3,400 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls. Units
with SNCR "fully operate" those controls.

Final $5,000 per ton Cost Threshold

[5.15_OS_NOx_Final_5000_CT]

Imposes a marginal cost of $5,000 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls. Units
with SNCR "fully operate" those controls.

Final $6,400 per ton Cost Threshold

[5.15_OS_NOx_Final_6400_CT]

Imposes a marginal cost of $6,400 per ton of ozone season NOx in
all states starting in 2017. Also forces all extant and currently
operating SCR to operate at "full operation" and all non-operating
SCR are returned to "full operation". Units without SOA
combustion controls upgraded to SOA combustion controls. Units
with SNCR "fully operate" those controls.

39


-------
Final Policy Case

[5.15 OS NOx Final Policy]

Imposes the budgets with variability limits derived from the
$1,400 per ton of NOx case were applied to states covered by the
final rule. Units in covered states with SCRs operate them at "full
operation" and units with SCRs that are not operating return them
to "full operation". Units without SOA combustion controls
upgraded to SOA combustion controls. This run is also called the
"Final Policy Case". This case was used for the Illustrative Policy
Air Quality Modeling analysis.

Final $1,400 per ton Cost Threshold With
No State Generation Limits

[5.15_OS_NOx_Final_1400_CT]

Similar to the 5.15 OS-NOx Finfal 1400 CT run, but without
limits on state generation levels

Notes:

1.	For the "Illustrative" cases, "fully operating" SCRs means units achieve an emission rate of 0.081
lbs/mmBtu or lower. For the "Final" cases, it means these units achieve an emission rate of 0.10
lbs/mmBtu or lower.

2.	Delaware was included as a covered state and given a budget in the Illustrative Policy Case and the
Less and More Stringent Policy Cases. This budget did not limit Delaware's emissions in those cases
and therefore did not impact the model results. Delaware is not covered by this final rule.

40


-------
Appendix B: Description of Excel Spreadsheet Data Files Used in the

AQAT

41


-------
EPA placed the AQATfinalcalibrated.xlsx Excel workbook file in the Transport Rule docket
that contains all of the emission and CAMx air quality modeling inputs and resulting air quality
estimates from the AQAT. The following bullets describe the contents of various worksheets within
the AQAT workbook:

State-level emission totals

•	"2017ek (base)" contains state and source-sector specific ozone-season NOx emission totals for
the 5.15 base case modeled in CAMx. Column D, "TOTAL w/o beis, fires" is an input in the
AQAT.

•	"201 lek (base)" contains state and source-sector specific ozone-season NOx emission totals
for the 5.15 base case for 2011. This is not used in the AQAT.

•	"state_sector_summer" contains the emissions used to create the "2017ek (base)" and "201 lek
(base)" worksheets.

•	"IPM TBtu and NOx_policy case" contains the IPM 5.15 estimates for each of the cases in the
air quality modeling.

•	"Budget Calcs (2017)" contains each emission budget level.

•	"State Level Emissions" are the total ozone-season NOx emissions for the various base,
emissions budgets (EB), and $l,400/ton illustrative control scenario. The results reflect
emissions from all fossil units greater than 25 MW for 2017. This sheet also includes the
calculations to transform the emission estimates to emission differences as fractions of each
state's 2017 base case emission inventory used in the CAMx source apportionment modeling.

Air quality modeling design values from CAMx

•	"2017ek-ctrl-ozone-DV-3x3 - Ozon" contains design values for three scenarios (the 2011 case,
the 2017 final air quality modeling base case, and the 2017 $1,400 final illustrative control
case. The average and maximum design values in 2017 are shown using one decimal place and
to four decimal places.

State-level ozone contributions

•	"2017 contributions (orig)" includes the original contributions with five decimal places of
resolution. The truncated shortened version of these contributions equal the truncated base case
average design value. See the Air Quality Modeling Technical Support Document for the Final
Cross-State Air Pollution Rule Update and the preamble for details about the contributions.

•	"2017 contributions" is a second copy of the contributions. In column BK, the contributions
are summed.

Air quality estimates

•	"Summary DVs" contains the average and maximum design value estimates (truncated to one
decimal place) for receptors that were nonattainment or maintenance in the 2017 base case air
quality modeling. Monitors that are at or above 76.0 ppb are shaded.

•	"2017 calibration". Contains the unadjusted estimated change in concentration resulting from
the difference in emissions between the 2017 base case and the 2017 final budget control
scenario case. The calibration factor from the proposal is found in columns BN and BQ. The
calibration factor from the final rule modeling is calculated in column CA. The calibration
factor used is in column CB. The ratio of the maximum to average design value for the base
case is used on all worksheets to estimate the maximum design value.

42


-------
•	"0 IPM" contains the contributions and design values for the $0/ton final base case. These are
the emission estimates directly from IPM.

•	"0 eng EB" contains the contributions and design values for the $0/ton final emissions budgets
base case) cost threshold analysis (where non-linked states were adjusted to the final emissions
budgets base case level). These are the emission estimates after the final base case emission
estimates from IPM have been converted using the adjusted historical emission values.

•	"0 eng EBwith PA" contains the contributions and design values for the $0/ton final emissions
budgets base case including PA RACT) cost threshold analysis (where non-linked states were
also adjusted to the final emissions budgets base case including PA RACT emission level).

•	"800 eng EB" contains the contributions and design values for the $800/ton emissions budgets
analysis (where non-linked states were adjusted to the final emissions budgets base case
including PA RACT emission level).

•	"1400 eng EB" contains the contributions and design values for the $l,400/ton emissions
budgets analysis (where non-linked states were adjusted to the final emissions budgets base
case including PA RACT emission level).

•	"3400 eng EB" contains the contributions and design values for the $3,400/ton emissions
budgets analysis (where non-linked states were adjusted to the final emissions budgets base
case including PA RACT emission level).

•	"5000 eng EB" contains the contributions and design values for the $5,000/ton emissions
budgets analysis (where non-linked states were adjusted to the final emissions budgets base
case including PA RACT emission level).

•	"6400 eng EB" contains the contributions and design values for the $6,400/ton emissions
budgets analysis (where non-linked states were adjusted to the final emissions budgets base
case including PA RACT emission level).

•	"1400_control" contains the estimated state-by-state and receptor-by-receptor air quality
contributions and design values for the $l,400/ton illustrative policy case emissions. All states
are adjusted to this emission level regardless of whether they are "linked" to a specific monitor.
These values can be compared to the CAMx values.

•	The "6400 eng EB links", "5000 eng EB links", "3400 eng EB links", "1400 eng EB links",
"800 eng EB links", "0 eng EB links", "0 eng EB with PA links", "0 IPM links", "0 eng EB w
PA links all", and "1400_control links" worksheets assess the linkages for the 1% threshold. A
contribution is set to zero if the maximum design value at that monitor is less than 76.0 ppb or
if it is a contribution from the state containing the monitor (i.e., "home" state). Compare rows
4 and 5 to look for linkages that affect whether a state is no longer linked to a monitor that
continues to have air quality issues. A value of 1 indicates that the state is "linked". Note that,
for the over-control assessment, we are particularly interested in states where there is a value of
1 in row 4 and no value in row 5.

43


-------
Appendix C: Description of 2017 Adjustments to 2018 IPM EGU
Ozone-Season NOx Emissions Data

44


-------
As described in Preamble Section IV.B, the EPA is aligning the analysis and implementation of this
final rulemaking with the 2017 ozone season in order to assist downwind states with the timely
attainment of the 2008 ozone NAAQS. As described in Preamble Section V.B.2., adjusted the IPM
v5.15 2018 run year results to account for differences between 2017 and 2018 in the power sector,
because IPM v5.15 does not have an output year of 2017.

To calculate the 2017 emissions for the base case, uniform NOx cost threshold cases, proposed remedy
and alternative cases, and produce a flat files for air quality modeling, EPA started with the 2018 Base
Case results and made modifications to emissions of units in three categories as described in the table
below.

Table Appendix C-l. Descri

)tion of 2017 ozone-season NOx Adjustment Calculation

2017 ozone-season NOx
Adjustment Case

How 2017 Adjustments Were Calculated

SCR Operation/Installation

For units that had an SCR in 2018 but were assumed to not operate
(or be installed) in 2017, EPA recalculated the NOx emissions for
the unit with the 2018 heat input and the 2016 emissions rate

Retirement

For units projected to retire in 2018, emissions from the 2016 run
year were included in the 2017 emissions

For uniform NOx cost threshold cases and policy alternative cases,
emissions from units with SCRs were determined by multiplying
their heat input in the 2016 run year by their optimized NOx
removal rate.

Coal-To-Gas

For units that had implemented coal-to-gas retrofit options in 2018
and had not dispatched, emissions from the 2016 run year were
incorporated. However, if the coal-to-gas retrofit options had
dispatched in 2018, then the NOx emissions were calculated based
on the 2018 fuel use and 2016 NOx rate.

vfote: 2017 adjustment calculations are based on data from the retrofitted or retired model plant, and therefore
may not match unit data in the parsed file, which is not a direct IPM output but rather a post-processed file

The Excel workbook, "Appendix C-2: Incremental ozone season NOx emissions and heat input for
2017 Adjustments" show the units that were affected by these changes. The workbook lists by unit the
reason for the change, the incremental ozone season NOx and the incremental heat input for 2017 that
was calculated. Units may not have added emissions and heat input in every case. These adjustments
are summarized at the state level in Appendix E.

45


-------
Appendix D: Ozone-Season NOx Emissions Budgets for IPM Modeling

46


-------
To best model the three regulatory control alternatives in IPM, EPA did not include the 2017
budget adjustments in the state budgets, as those adjustments are accounting for things not
reflected in the model. Table Appendix D-l shows the proposed 2017 state budgets and the
equivalent 2018 transport region and state emission constraints used for IPM analysis in the IPM
v5.15 model platform. The calculations of these numbers are included in Appendix E.

Table Appendix D-l. Ozone-Season NOx Emissions Budgets For IPM v5.15 Modeling



Illustrative Policy
Case

Less Stringent Policy
Case

More Stringent Policy
Case

Budget

Assurance
Level

Budget

Assurance
Level

Budget

Assurance
Level

Alabama

12,599

15,245

13,548

16,393

11,406

13,801

Arkansas

9,211

11,145

12,060

14,593

9,041

10,940

Delaware

497

601

497

601

494

598

Iowa

11,272

13,639

11,477

13,887

11,065

13,389

Illinois

14,588

17,651

14,632

17,705

14,464

17,501

Indiana

21,527

26,048

26,419

31,967

19,804

23,963

Kansas

7,782

9,416

7,785

9,420

7,730

9,353

Kentucky

19,675

23,807

23,030

27,866

19,475

23,565

Louisiana

18,636

22,550

19,087

23,095

18,470

22,349

Maryland

3,457

4,183

3,795

4,592

2,838

3,434

Michigan

16,483

19,944

18,630

22,542

15,222

18,419

Missouri

15,085

18,253

16,628

20,120

14,604

17,671

Mississippi

6,315

7,641

6,350

7,684

6,191

7,491

New Jersey

2,057

2,489

2,063

2,496

2,061

2,494

New York

5,050

6,111

5,129

6,206

4,928

5,963

Ohio

18,763

22,703

22,372

27,070

18,599

22,505

Oklahoma

11,742

14,208

13,871

16,784

9,254

11,197

Pennsylvania

19,554

23,660

29,875

36,149

19,479

23,570

Tennessee

9,115

11,029

9,115

11,029

9,115

11,029

Texas

51,931

62,837

54,544

65,998

50,022

60,527

Virginia

9,224

11,161

9,357

11,322

8,758

10,597

Wisconsin

7,862

9,513

7,922

9,586

7,791

9,427

West Virginia

18,152

21,964

25,730

31,133

17,706

21,424

CSAPR Update
Region Total

310,577

N/A

353,916

N/A

298,517

N/A

Final Policy Case

Budget

Assurance
Level

13,210

15,984

9,210

11,144

N/A

N/A

11,272

13,639

14,587

17,650

23,303

28,197

8,027

9,713

20,782

25,146

18,639

22,553

3,820

4,622

16,545

20,019

15,780

19,094

6,315

7,641

2,061

2,494

5,135

6,213

19,522

23,622

11,619

14,059

17,946

21,715

7,693

9,309

52,300

63,283

9,223

11,160

7,915

9,577

17,815

21,556

312,719

N/A

47


-------
Appendix E: Detailed Budget Calculations

48


-------
See the spreadsheet "Ozone Transport Policy Analysis Final Rule TSD Appendix E" for detailed
calculations of state budgets and assurance levels.

49


-------
Appendix F: State Generation Constraint Analysis

50


-------
Table F-l. Tons of EGU NOx reduction potential from shifting generation (with state level
generation constraints), compared to adjusted historic emissions.35





Reductions from





2015 Adjusted
Historical
Emissions

Generation Shifting

Reductions From



between the Final Base

Generation Shifting as

Row Labels

Case and the Final
$1,400 per ton Cost
Threshold Case

a Percentage of 2015
Adjusted Historical
Emissions

Alabama

15,179

-474

-3%

Arkansas

12,560

-348

-3%

Illinois

14,850

-241

-2%

Indiana

31,382

-1,205

-4%

Iowa

11,478

-10

0%

Kansas

8,031

-8

0%

Kentucky

26,318

-690

-3%

Louisiana

19,101

-318

-2%

Maryland

3,871

-45

-1%

Michigan

19,811

-357

-2%

Mississippi

6,438

-29

0%

Missouri

18,443

-430

-2%

New Jersey

2,114

-34

-2%

New York

5,531

-295

-5%

Ohio

27,382

-252

-1%

Oklahoma

13,747

-457

-3%

Pennsylvania

35,607

-353

-1%

Tennessee

7,779

-34

0%

Texas

54,839

-441

-1%

Virginia

9,367

-83

-1%

West Virginia

26,874

33

0%

Wisconsin

7,939

-34

0%

TOTAL

378,641

-6,107

-2%

35 See "Reductions From Generation Shifting in the Final $1400 Per Ton Cost Threshold Case" in the docket for this
rule

51


-------
Table F-2. Comparison of state level ozone season NOx emission reductions from affected
sources and resulting state budgets between the $1,400 per ton cost threshold cases with
and without state level generation constraints.36 Note that the change in budget in the table
would be additional reductions (negative numbers) of fewer reductions (positive numbers) due to
allowing generation to shift among states. This change would be incremental to NOx reductions
from generation shifting identified in table F-l.

State

Final Budget

Budget

Change in

Percent



(with state

without state

Budget

Change



generation

generation

(tons)





limits) (tons)

limits (tons)





Alabama

13,211

13,097

-114

-1%

Arkansas

9,210

9,290

80

1%

Iowa

11,272

11,230

-42

0%

Illinois

14,601

14,608

7

0%

Indiana

23,303

23,399

96

0%

Kansas

8,027

7,969

-58

-1%

Kentucky

21,115

20,931

-184

-1%

Louisiana

18,639

18,572

-67

0%

Maryland

3,828

3,066

-762

-20%

Michigan

16,545

16,615

70

0%

Missouri

15,780

15,776

-4

0%

Mississippi

6,315

6,438

123

2%

New Jersey

2,062

2,074

12

1%

New York

5,135

5,146

11

0%

Ohio

19,522

19,516

-6

0%

Oklahoma

11,641

11,450

-191

-2%

Pennsylvania

17,952

18,038

86

0%

Tennessee

7,736

7,736

0

0%

Texas

52,301

52,098

-203

0%

Virginia

9,223

9,169

-54

-1%

Wisconsin

7,915

7,889

-26

0%

West Virginia

17,815

17,849

34

0%

Total

313,148

312,047

-1,192

0%

36 See "Budget Calculations for Final $1400 per ton Cost Threshold Case with No State Generation Limits" in the
docket for calculations.

52


-------