Technical Support Document (TSD)
for the Final Federal Good Neighbor Plan for the 2015 Ozone National Ambient Air Quality
Standards
Docket ID No. EPA-HQ-OAR-2021-0668
Ozone Transport Policy Analysis
Final Rule TSD
U.S. Environmental Protection Agency
Office of Air and Radiation
March 2023
1
-------
The analysis presented in this document supports the EPA's final Federal Good Neighbor
Plan for the 2015 Ozone National Ambient Air Quality Standards. This TSD includes analysis to
help quantify upwind state emissions that significantly contribute to nonattainment or interfere
with maintenance of the 2015 ozone NAAQS in downwind states and quantification of EGU
emission budgets and the resulting effects on air quality of the EGU and non-EGU strategies
included in the final rule. The analysis is described in Sections V and VI of the preamble to the
rule. This TSD also describes how the EPA used historical data and the Integrated Planning
Model (IPM) to inform air quality modeling, budget setting, and policy analysis aspects of this
rule for EGUs, as well as describing analysis of the non-EGU policy scenarios, including for
overcontrol. Finally, this TSD includes an assessment on the effects of ozone concentrations on
forest health. This TSD is organized as follows:
A. Using Engineering Analytics and the Integrated Planning Model (IPM) in the Step
3 Assessment of Significant Contribution to Nonattainment and Interference with
Maintenance 3
B. Calculating Step 4 EGU Emission Budgets from Historical Data 6
1. Calculating 2023-2029 Engineering Baseline Heat Input and Emissions 6
2. Estimating impacts of combustion and post combustion controls on state-level
emission rates 9
3. Variability Limits 35
4. Calculating Dynamic Budgets Starting in 2026 35
C. Analysis of Air Quality Responses to Emission Changes Using an Ozone Air
Quality Assessment Tool (AQAT) 41
1. Introduction 43
2. Details on the construction of the ozone AQAT for this rule 46
3. Description of the analytic results using the primary approach for the Step 3
AQAT configuration 61
4. Comparison between the air quality assessment tool estimates using the
primary and alternative calibration factors 72
5. Assumptions made in the air quality assessment tool 77
D. Selection of Backstop Emission Rate 81
1. Observations of fleet operation for well-controlled units 81
2. Creating "comparably stringent" emission rates using the 2014 1-hour S02
concepts 84
3. Accommodating startup and shutdown emissions using a 50-ton buffer 87
E. Preliminary Environmental Justice Screening Analysis for EGUs 88
F. Assessment of the Effects of Ozone on Forest Health 92
Appendix A: State Emission Budget Calculations and Engineering Analytics 95
Appendix B: Description of Excel Spreadsheet Data Files Used in the AQAT 96
Appendix C: IPM Runs Used in Transport Rule Significant Contribution Analysis 101
Appendix D: Description of the Analytic Results using the Primary Approach for the
"Full Geography" AQAT Configuration in 2026 103
Appendix E: Feasibility Assessment for Engineering Analytics Baseline 106
Appendix F: Preset State Emission Budgets 110
Appendix G: Comparison of CSAPR 2012 Budgets to Actual 2012 Emissions Ill
Appendix H: Sensitivity for order of emissions reductions from EGUs and nonEGUs 115
2
-------
Appendix I: Figures Related to Preamble Section V and Section VI
Appendix J: Additional Sensitivity Examining the AQAT Calibration Factors
Appendix K: Additional AQAT sensitivity including the IRA
116
118
123
A. Using Engineering Analytics and the Integrated Planning Model (IPM) in the Step 3
Assessment of Significant Contribution to Nonattainment and Interference with
Maintenance
In order to establish EGU NOx emissions control stringencies for each linked upwind
state, EPA first identifies various possible uniform levels of NOx control stringency based on
available EGU NOx control strategies and represented by cost thresholds.1 The EGU emission
reductions pertaining to each level of control stringency are derived using historical data,
engineering analyses, and the Integrated Planning Model (IPM) for the power sector as described
in section B of this TSD. A similar assessment for one scenario was done for non-EGUs. Next,
EPA uses the ozone Air Quality Assessment Tool (AQAT) to estimate the air quality impacts of
the upwind state emissions reductions on downwind ozone pollution levels for each of the
assessed cost threshold levels. Specifically, EPA looks at the magnitude of air quality
improvement at each receptor at each level of control; it also examines whether receptors change
status (shifting from either nonattainment to maintenance, or from maintenance to attainment),
and looks at the individual contributions of each state to each of its receptors. See section C in
this TSD for discussion of the development and use of the ozone AQAT.
In this TSD, EPA assesses the EGU NOx mitigation potential for all states in the
contiguous U.S. EPA assessed the air quality impacts from emission reductions for all monitors
in the contiguous U.S. for which air quality contribution estimates were available. In applying
the multi-factor test for purposes of identifying the appropriate level of control, the EPA
evaluated NOx reductions and air quality improvements at the receptors determined to have a
transport problem (see section IV.F. of the Preamble), and the 23 upwind states2 that were linked
to downwind receptors3 in step two of the 4-Step Good Neighbor Framework. These states are
listed in Table A-l below. Since California EGUs are not covered in this final rule, this TSD's
references to "affected states" or "states covered by this rule" in EGU-related emissions
materials does not include California.4
1 See the EGU NOx Mitigation Strategies Final Rule TSD.
2 Note that 4 of the 23 upwind states are also states with non-attainment or maintenance receptors, or "home states."
3 Monitor 490570002 in Weber County, UT ceased operation in 2019 and is no longer considered to be a receptor in
this final rule. Including this monitoring site in the analysis for Step 3 is not determinative for the final results of this
analysis.
4 EPA notes that there are two receptors on tribal lands in California. The regulatory ozone monitor located on the
Morongo Band of Mission Indians ("Morongo") reservation is a projected downwind receptor in 2023 and the
Temecula, California regulatory ozone monitor is a projected downwind receptor in 2023 (and in past regulatory
actions has been deemed representative of air quality on the Pechanga Band of Luiseno Indians ("Pechanga")
reservation). As California EGUs are not covered in this action (and no other state would be linked to these
receptors), EPA does not include these receptors when discussing receptors impacted by EGU reductions. However,
these receptors and their corresponding design value change due to both EGU reductions and non-EGU reductions
elsewhere and in California and are shown in the accompanying AQAT file. See Ozone AQAT Final.xlsx for
results.
3
-------
Table A-l. Upwind States Evaluated in the Multi-factor Test
Alabama+
Nevada
Arkansas
New Jersey
California*
New York
Illinois
Ohio
Indiana
Oklahoma
Kentucky
Pennsylvania
Louisiana
Texas
Maryland
Utah"
Michigan
Virginia
Minnesota+
West Virginia
Mississippi
Wisconsin+
Missouri
*California EGUs are not covered by this rule.
+Linkages for Alabama, Minnesota, and Wisconsin are projected to resolve before 2026. Therefore, those
states have a lower level of emission control stringency compared to states that are projected to be linked
in 2026.
AIn recognition of Utah's lack of state jurisdiction over an existing EGU in the Uintah and Ouray
Reservation, the effects of the rule for that facility are presented independently from Utah in this document
and fall under the descriptor "tribal" or "tribal data. "
Similar to the CSAPR Update and the Revised CSAPR Update, EPA relied on adjusted
historical data (engineering analytics) as part of the process to identify emissions control
stringencies to eliminate significant contribution at step three within the 4-Step Good Neighbor
Framework. Historical data were adjusted through the engineering analytics tool to analyze the
ozone season NOx emission reductions available from EGUs at various uniform levels of NOx
control stringency, represented by cost per ton, in each upwind state. Finally, IPM was used to
evaluate compliance with the rule and the rule's regulatory control alternatives (i.e., compliance
with the emission budgets, with a more stringent alternative, and with a less stringent
alternative). In order to examine the impact of the recently passed Inflation Reduction Act (IRA),
EPA also performed two additional scenarios, namely an updated baseline scenario that included
key provisions of the IRA, and a run that included both the final rule and key IRA provisions.
EPA also used its engineering analytics tool and IPM projections to perform air quality
assessment and sensitivity analysis as part of step 3.
The engineering analysis tool uses 2021 ozone-season data as representative historical
emissions and operating data reported under 40 CFR part 75 by covered units 4. It is a tool that
builds estimates of future unit-level and state-level emissions based on exogenous changes to
historical heat input and emissions data reflecting fleet changes that will occur subsequent to the
last year of available data. See Section B. Calculating Step 4 EGU Emission Budgets from
Historical Data for a detailed description of the engineering analytics tool.
4
-------
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
All IPM cases for this rule included representation of the Title IV SO2 cap and trade program;
the NOx SIP Call; the CSAPR, Update, and Revised CSAPR Update regional cap and trade
programs; consent decrees and settlements; and state and federal rules as listed in the IPM
documentation referenced above. For details on which measures are endogenously modeled
within IPM and which are not, please see Appendix Table C-l.
Table A-2 below summarizes the reduction measures that are broadly available at various
cost thresholds for EGUs.
Table A-2. Reduction strategies available to EGUs at each cost threshold.
Cost Threshold ($ per
ton Ozone-Season NOx)
Reduction Options
$1,800
-Retrofitting state-of-the-art combustion controls;
-Optimizing idled SCRs;
-Optimizing operating SNCRs
$11,000
-All options above and;
-Installing SCR and SNCR on coal and oil/gas steam units
greater than 100 MW and lacking post combustion
controls.
For the Engineering Analytics:
At $l,800/ton:
o If 2021 adjusted baseline rate was greater than 0.08 lb/MMBtu for SCR controlled
coal units, that rate and corresponding emissions were adjusted down to 0.08
lb/MMBtu starting in 2023;
o for SCR controlled oil/gas units, if the adjusted historical rate was greater than
0.03 lb/MMBtu then the rate was adjusted downwards to 0.03 lb/MMBtu starting
in 2023;
o for SCR controlled combined cycle units, if the adjusted historical rate was
greated than 0.012 lb/MMBtu then the rate was adjusted downwards to 0.012
lb/MMBtu in 2023;
o for SCR controlled combustion turbine units, if the adjusted historical rate was
greated than 0.03 lb/MMBtu then the rate was adjusted downwards to 0.03
lb/MMBtu in 2023; and
o for units with LNB upgrade potential and an adjusted historical rate greater than
0.199 lb/MMBtu, their rates were adjusted downwards to 0.199 lb/MMBtu
starting in 2023.
5 See "Documentation for EPA's Power Sector Modeling Platform v6 using Updated Summer 2021 Reference
Case". Available at https://www. https://www.epa.gov/airmarkets/documentation-epas-power-sector-modeling-
platform-v6-summer-2021 -reference-case. See also the "Updated Summer 2021 Reference Case Incremental
Documentation for the 2015 Ozone NAAQS Actions." https://www.epa.gov/power-sector-modeling/supporting-
documentation-2015-ozone-naaqs-actions
5
-------
o Starting in 2023 units with SNCRs were given their mode 2 NOx rates6 if they
were not already operating at that level or better in 2019.
At $11,000/ton:
o Same as $l,800/ton; additionally:
o Coal units greater than or equal 100 MW and lacking a SCR were given an
emission rate equal to 0.05 lb/MMBtu reflecting SCR installation starting in 2026.
Oil/gas steam units greater than or equal 100 MW and with a three year (2019-
2021) average of ozone season emissions of at least 150 tons were given an
emission rate of 0.03 lb/MMBtu reflecting SCR installation starting in 2026.
B. Calculating Step 4 EGU Emission Budgets from Historical Data
1. Calculating 2023-2029 Engineering Baseline Heat Input and Emissions
The underlying data and calculations described below can be found in the workbook titled
(Appendix A - Final Rule State Emission Budget Calculations and Engineering Analytics). They
are also available in the docket and on the EPA website.
EPA starts with 2021 reported, seasonal, historical NOx emissions and heat input data for each
unit.7 This reflects the latest representative owner/operator reported data available at the time of
EPA analysis.8 The NOx emissions data for units that report data to EPA under the Acid Rain
Program (ARP), Cross-State Air Pollution Rule (CSAPR), CSAPR Update, and Revised CSAPR
Update are aggregated to the summer/ozone season period (May-September). Because the unit-
level NOx emissions for the summer/ozone-season period are relevant to determining ozone-
season emissions budgets, those files are shown in the "Unit 2023" through "Unit 2029" sheets
in the "Appendix A: Final Rule State Emission Budget Calculations and Engineering Analytics"
file accompanying this document.9 In that file, unit-level details such as facility name, unit ID,
etc. are shown in columns A through H of the "Unit 2023" through "Unit 2029" worksheets.
Reported historical data for these units such as unit type, capacity, fuel, existing post combustion
controls, historical emissions, heat input, generation, etc. are shown in columns I through U. The
2021 historical emissions value is in column R. The assumed future year baseline emissions
estimate (e.g., 2023-2029) is shown in column AD, and reflects either the same emissions level
6 For a unit with an existing post-combustion control, mode 1 reflects the existing post-combustion control not
operating and mode 2 the existing post-combustion control operating. For details, please see Chapter 3.10 of the
IPM documentation available at: https://www.epa.gov/airmarkets/documentation-epas-power-sector-modeling-
platform-v6-summer-2021 -reference-case.
7 "Seasonal" refers to the ozone-season program months of May through September.
8 As explained in VLB.4 of the preamble, at the end of this procedure EPA is able to evaluate, as part of its quality
assurance and quality check, whether the use of the most recent historical final data (e.g., 2021) is representative of
the baseline heat input and emissions for each state and make any adjustments if needed.
9 The EPA notes that historical unit-level ozone season EGU NOx emission rates are publicly available and quality
assured data. The emissions are monitored using continuous emissions monitors (CEMs) or other monitoring
approaches available to qualifying units under 40 CFR part 75 and are reported to the EPA directly by power sector
sources.
6
-------
as that observed in 2021, or a modification of that value based on changes expected to the
operational or pollution control status of that unit.10 These modifications are made due to:
a. Retirements - Emissions from units with upcoming confirmed retirement dates are
adjusted to zero for ozone seasons subsequent to that retirement date. Retirement
dates are identified through a combination of sources including EIA Form 860,
utility-announced retirements,11 and stakeholder feedback provided to EPA, as
reflected in the National Electricity Energy Data System (NEEDS) February 2023
file. For the purpose of the engineering analysis, when companies have announced
they will either sell a unit or retire it by a certain date, the EPA assumed that the unit
would retire unless there is news of a specific potential buyer. Retirement dates are
shown in columns J and K and the impact of retirements on emissions is shown in
column V. The retiring units are flagged in column W.12-13
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
0MMBtu x 0.2 lb/MMBtu = 0 ton
b. Coal to Gas Conversion - Emissions from coal units with scheduled conversions to
natural gas fuel use are adjusted to reflect reduced emission rates associated with
natural gas for years subsequent to that conversion date. To reflect a given unit's
conversion to gas, that unit's future emission rates for NOx are assumed to be half of
its 2021 coal-fired emission rates while utilization levels are assumed to remain the
same.14 Therefore, the future year estimated emissions for these converting units are
expected to be half of 2021 levels for NOx. Units expected to convert to gas are
flagged using EIA Form 860, utility announcements, and stakeholder feedback, as
reflected in NEEDS February 2023. For the purpose of the engineering analysis,
when units have a requirement to either convert to gas or retire (i.e., cease burning
coal) but there has been no indication which option a unit will take, EPA assumed
that the unit would convert to gas. The impact of coal to gas conversion for the future
111 Based on data and changes known at time of analysis.
11 Starting with the June 2022 version of NEEDS, EPA has begun including announced retirements as that
represents the most likely future behavior for the unit, unless compelling information suggests such retirement may
not happen or may be delayed. EPA also determined that including announced retirements in the engineering
analysis would be helpful in establishing pre-set budgets, particularly beyond 2024, as that would help ensure state
emission budgets are reflective of the best information on the power sector's operating profile in future years. It has
been EPA's experience that in recent years, units' announced retirements tend to be moved forward rather than
pushed back in time, making the inclusion of announced retirements reasonable. For cases beyond 2024 where unit
retirements may be pushed back, the calculation of the dynamic budgets would capture those delayed retirements
and would adjust accordingly (i.e., they would continue to reflect the operation of the unit in question). Since states
would receive the higher of the pre-set and dynamic budgets from 2026 through 2029, this would prevent states
from being under-budgeted because of changes in projected retirements used to establish the preset budgets.
12 EPA updated its inventory of units flagged as retiring in column N based on stakeholder input, including on
previous rulemakings and the latest data from EIA 860 and the PJM retirement tracker.
13 Units that are to retire by the start of the a year's ozone season are considered retired for that year in the
engineering analysis. Units that will operate for at least part of the ozone season of a given year will not be
considered retired until the following year for the engineering analysis.
14 This is consistent with NOx rate change used in IPM. See "Documentation for EPA's Power Sector Modeling
Platform v6 using Summer 2021 Reference Case." table 5-18.
7
-------
year is shown in column Z, flagged in column AA. The example below pertains to
NOx emission estimates. For any control decisions after the point of conversion, the
unit is treated as an O/G Steam unit, shown in column I.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtu xO.l lb/MMBtu = 0.5 ton
c. Retrofits - Emissions from units with scheduled SCR or SNCR retrofits are adjusted
to reflect the emission rates expected with new SCR installation (0.05 lb/MMBtu of
NOx for a coal unit, and 0.03 lb/MMBtu for an oil/gas steam unit) and new SNCR
(25% decrease in previously reported emission rate for all boilers except circulating
fluidized bed boilers that receive a 50% decrease in previously reported emission
rate) and are assumed to operate at the same 2021 utilization levels.15 These emission
rates were multiplied by the affected unit's 2021 heat input to estimate the future year
emission level. The impact of post-combustion control retrofits on future year
emissions assumptions is shown in column AB, flagged in column AC.
For SNCR:
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtu x 0.15 lb MMBtu = 0.75 ton
For SCR:
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtu x 0.050 lb/MMBtu = 0.25 ton
d. Other - EPA also made several unit-specific adjustments to 2021 emission levels to
reflect forthcoming emission or emission rate requirements specified in consent
decrees, BART requirements, state RACT rules, and/or other revised permit limits.
The impacts for future year emission assumptions are shown in column AD, flagged
in column AE.16
e. New Units - Emissions for new units are identified in the "New units" worksheet.
They reflect under-construction and/or permitted units greater than 25 MW that are
expected to be in commercial operation by the designated future year. These assumed
emission values for new units are reflected in column F and the online years are in
column I. To obtain these emissions, EPA identified all new fossil-fired EGUs
coming online after 2021 according to EIA Form 860 and stakeholder comments, as
reflected in NEEDS v6 October 2022. EPA then identified the heat rate and capacity
values for these units using EIA Form 860, as reflected in NEEDS v6 October 2022,
and stakeholder-provided data. Next, EPA identified the 2019 average seasonal
capacity factor for similar units that came online between 2015-2019. EPA used these
15 Ibid.
16 EPA checked its inventory of units impacted by consent decrees based on input provided stakeholders and
comments on previous rulemakings. No units were determined to be impacted as described in the Allowance
Allocation under the Final Rule TSD.
8
-------
seasonal capacity factors (e.g., 65% for natural gas combined cycle units and 10% for
combustion turbines), the unit's capacity, the unit's heat rate, and the unit's estimated
NOx rate to estimate future year emissions (capacity x capacity factor x number of
hours in ozone season x heat rate x NOx emission rate = NOx emissions).17
Additionally, for approximately fifteen additional units that are not new units but
which have not previously reported data to EPA under 40 CFR part 75 and for
purposes of the emissions budgets established under this rule are treated as new units
starting in 2024, EIA data sources are used to obtain the necessary data.
2021
Future Year (e.g., 2023)
Unit x
0 MMBtu xO.O lb/MMBtu = 0 ton
100 MW *0.65 *(153x24) *8000 Btu/KWh
*0.01 lb/MMBtu = 9 tons
After completing these steps, EPA has unit-level future year baselines that originate from the
most recently reported representative data (2021) and incorporate known EGU fleet changes.
The state-level file reflects a summation of the unit-level values..
2. Estimating impacts of combustion and post combustion controls on state-level emission rates
Next, EPA evaluates the impact of the different combustion and post-combustion controls.
Similar to the methodology above, EPA continued to adjust the historical data to reflect a future
year with specific uniform control assumptions. However, these adjustments were to capture
changes incremental to the baseline reflecting different uniform control measures. EPA applied
these adjustments for analytical purposes to all states, but only the affected states' adjustments
are relevant for emission budgets in this rule. Each of these adjustments is shown incrementally
for the relevant mitigation technology in the "Unit 2023" through "Unit 2029" worksheets.
a. SCR optimization - Emissions from units with existing SCRs, but that operated at an
emission rate greater than a fuel and unit type optimized level (0.08 lb/MMBtu for coal
steam, 0.03 for oil/gas steam, 0.03 for combustion turbine, and 0.012 for combined cycle)
in 2021, were adjusted downwards to reflect expected emissions when the SCR is
operated to the applicable optimized emission rate. The applicable optimized emission
rate is multiplied by the baseline heat input level to arrive at the future year emissions
estimate for a given unit. The impact on future year emission assumptions is shown in
column AF and flagged in column AG of the "Unit 2023" through "Unit 2029"
worksheets. EPA notes this assumption only applies to ozone-season NOx as that is the
season in which this rule would likely incentivize such operation. In the rule, EPA also
incorporated a flag in column AG for units with SCRs and a shared stack. For units with
an SCR that share a stack with a unit(s) that does not have SCR,, EPA did not assume
potential emission reductions attributable to existing SCR optimization as the reported
split of emissions between units may not reflect the actual split of emissions. Though
some commenters provided their own emission splits or emission rates for each unit
17 Emission rate data is informed by historical data, as reflected in NEEDS, for like units coming online in the last
five years. See "2019 and 2020 new NGCC Data" worksheet in the "EGU Power Sector 2019 and 2020 data" file in
the docket. EPA-HQ-OAR-2021-0668-0142
9
-------
sharing a stack, the EPA chose to consistently use the verified reported data. The EPA
notes that in some cases, the adjustments to NOx rates suggested by commenters would
result in lower budget because either: the impled emissions rate for the non-SCR unit
would be pushed above a 0.199 lb/MMBtu emissions rate and be eligible for a rate
commensurate with a state-of-the-art combustion control upgrade; or because starting in
2026-2027 the implied higher emissions and emissions rate at the non-SCR unit would be
reduced to the 0.05 lb/MMBtu commensurate with retrofitting a new SCR rather than
higher 0.08 lb/MMBtu rate commensurate with optimizing an existing SCR.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtu x 0.08 lb/MMBtu = 0.4 ton
b. State-of-the-art combustion controls - Emissions from units that were operating in 2021
without state-of-the-art combustion controls were adjusted downwards to reflect assumed
installation of, or upgrade to, these controls and their expected emission rate impact. EPA
assumed a future year emission rate of 0.199 lb/MMBtu for units expected to
install/upgrade combustion controls. This emission rate was multiplied by each eligible
unit's future year baseline heat input to estimate its future emission level. Details of
EPA's assessment of state-of-the-art NOx combustion controls and corresponding
emission rates are provided in the EGU NOx Mitigation Strategies Final Rule TSD. The
impact of state-of-the-art combustion controls on future year emission assumptions is
shown in column AH and flagged in column AI of the "Unit 2023" through "Unit 2029"
worksheets. EPA also incorporated a flag in column AI, based on stakeholder input, for
units with a shared stack. For these units, based on stakeholder provided data, EPA did
not assume potential emission reductions attributable to state-of-the-art combustion
controls as explained in preamble section V.B. Note, these assumptions apply emissions
adjustments throughout the entire year as the controls operate continuously once
installed.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.4 lb/MMBtu = 2 ton
10,000 MMBtu x 0.199lb/MMBtu = ~1 ton
c. SNCR optimization - Emissions from units with existing SNCRs, but that operated at an
emission rate greater than the SNCR optimization rate, were adjusted downwards to
reflect expected emissions when the SNCR is optimized. This emission rate was
identified specific to each unit based on historical data and is described in the EGU NOx
Mitigation Strategy Final Rule TSD. The optimized emission rate is multiplied by future
year baseline heat input levels to arrive at the future year emissions estimate. For the
units affected by this adjustment, the impact on future year emission assumptions is
shown in column AJ and flagged in column AK of the "Unit 2023" through "Unit 2029"
worksheets. Note, this assumption only applies to ozone-season NOx as that is the season
in which this rule's program would likely incentivize such operation.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtu x 0.15 lb MMBtu = 0.75 ton
10
-------
Post Combustion Control Retrofits (SNCR and SCR): Emissions for eligible coal and
oil/gas steam units were adjusted to reflect expected emission reductions from the retrofit
of either an SCR or SNCR. Table B-l shows the eligibility of units assumed to receive
each type of retrofit in the engineering analysis. Uncontrolled units at coal facilities that
share a stack with an existing SCR but are also eligible to receive a new retrofit SCR are
given an emission rate assuming an optimized new SCR in years for which this control
measure is available. For more information on the retrofit assumptions, see section V.B
of the Preamble.
SNCR retrofit- Emissions from coal steam units less than 100 MW without post-
combustion controls as well as coal-fired circulating fluidized bed (CFB) boilers of any
size without post-combustion controls were adjusted downwards to reflect expected
emissions if an SNCR were to be retrofitted on the unit. The emission rate was identified
as the higher of 75% of the unit's baseline emission rate level (i.e., reflecting a 25%
reduction from the technology) or 0.08 lb/MMBtu (i.e., an emission rate floor for
SNCR).18 For CFB units, the emission rate was identified as the higher of 50% of the
unit's baseline emission rate level or 0.08 lb/MMBtu. The adjusted emission rate is
multiplied by future year baseline heat input levels to arrive at the future year emissions
estimate for that technology. For the units affected by this adjustment, the impact on
future year emission assumptions is shown in column AO and flagged in column AP of
the "Unit 2023" through "Unit 2029" worksheets.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtu x 0.15 lb/MMBtu = 0.75 ton
18 See https://www.epa.gov/airmarkets/retrofit-cost-analvzer for the "Retrofit Cost Analyzer (Update 1-26-2022)"
Excel tool (EPA-HQ-OAR-2021-0668-0118) and for the documentation of the underlying equations in "IPM Model
- Updates to Cost and Performance for APC Technologies: SNCR Cost Development Methodology for Coal-fired
Boilers" (February 2023).
11
-------
77. SCR retrofit- Emissions from 1) coal units greater than 100 MW without SCR controls and
2) oil/gas steam units greater than 100 MW without an SCR and a three year (2019-2021)
average of ozone season emissions of at least 150 tons were adjusted downwards to
reflect expected emissions if an SCR were to be retrofitted on the unit.19 The emission
rate was identified as the higher of 10% of the unit's baseline emission rate or 0.05
lb/MMBtu for coal steam units and 0.03 lb/MMBtu for oil/gas steam units (i.e., a 90%
reduction with an emission rate floor of 0.05 or 0.03 lb/MMBtu). 20 The adjusted
emission rate is multiplied by future year baseline heat input levels to arrive at the future
year emissions estimate for that technology. For the units affected by this adjustment, the
impact on future year emission assumptions is shown in column AO and flagged in
column AP of the "Unit 2023" through "Unit 2029" worksheets. Note, this assumption
only applies to ozone-season NOx. To inform quantification of state budgets for the 2026
ozone season control period as explained in preamble section VI.A.2.a, the EPA also
quantifies an intermediate point halfway between the pre- and post-SCR rate is shown as
"SCR (Half)" in column AN. For units with an SCR that share a stack with a unit(s) that
does not have SCR an intermediate point halfway between pre- and post-SCR
optimization is also shown in this column, mirroring the half-way phase in for SCR
retrofits.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtu x 0.05 lb/MMBtu = 0.25 ton
Table B-l. Post-Combustion Control Retrofit Assumptions for Coal and Oil/Gas Steam
Units in the Engineering Analysis.
Fuel
Unit Type
Capacity
(MW)
Average of 2019 to
2021 Ozone Season
NOx (tons)
Retrofit
Type
Emission
Rate
(lb/MMBtu)
Coal
not CFB
>=100
All
SCR
0.05
Coal
not CFB
<100
All
SNCR
25% reduction
Coal
CFB
All
All
SNCR
50% reduction
Oil/Gas
All
>=100
>=150
SCR
0.03
With all of these unit-level adjustments applied, the resulting unit-level heat input and
unit-level emissions are summed up to the state level. New units' emissions and generation and
19 The EPA used a 3-year average of 2019-2021 reported ozone season emissions to derive a tons per ozone season
value representative for each covered oil/gas steam unit. This three year period includes a variety of circumstances
for the economy and demand for electricity and using the average avoids including or excluding units because of a
single anomalous year of generation and emissions.
211 "IPM Model - Updates to Cost and Performance for APC Technologies: SCR Cost Development Methodology
for Coal-fired Boilers" (February 2023);
"IPM Model - Updates to Cost and Performance for APC Technologies: SCR Cost Development Methodology for
Oil/Gas-fired Boilers" (February 2023)
12
-------
other state level budget adjustments21 are added after this step to inform the state-level totals. ;
these state-level emissions are visible in the worksheets titled "State 2023" through "State 2029"
in the Appendix A: Final Rule State Emission Budget Calculations and Engineering Analytics
workbook accompanying this document.22
Finally, the EPA identified the column in each "state" tab that corresponds to the control
stringency identified for that state and that year as described in Section V of the preamble. These
values constitute the preset state emission budgets and are shown in column Q. Emission levels
at each control stringency are shown in Tables B-2 through B-8 for all states in the contiguous
United States, regardless of whether they were covered in the program. The preset state budgets
for covered states are displayed in Tables B-9 through B-15.
21 The state level budget adjustment is described in Section VI.B.4.a. of the Preamble.
22 Appendix A: Proposed Final Rule State Emission Budget Calculations and Engineering Analytics shows the unit-
level details and calculations described in sections B. 1 and B.2 of this TSD, before aggregating those values to use
at the state and regional level. The unit-level values inform the state-level budgets and are not a prediction of how
each unit will operate in the future. Although anchored in historical data, EPA recognizes at the unit-level some
units will overperform and some units will underperform the unit-level values.
13
-------
Table B-2. 2023 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2023
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
Alabama
6,412
6,379
6,379
6,379
Arizona
7,723
7,639
7,570
7,439
Arkansas
8,955
8,927
8,927
8,927
California
1,731
1,340
1,340
1,340
Colorado
6,470
6,393
6,393
6,393
Connecticut
381
355
355
355
Delaware
423
388
388
384
Florida
13,541
11,000
11,000
11,000
Georgia
5,191
5,179
5,179
5,172
Idaho
240
240
240
240
Illinois
7,721
7,652
7,652
7,474
Indiana
13,298
12,442
12,442
12,440
Iowa
9,867
9,867
9,813
9,752
Kansas
6,231
5,484
5,484
5,484
Kentucky
13,900
13,601
12,999
12,999
Louisiana
9,974
9,459
9,459
9,363
Maine
108
86
86
86
Maryland
1,214
1,214
1,214
1,206
Massachusetts
297
265
265
265
Michigan
10,746
10,742
10,742
10,727
Minnesota
5,643
5,544
5,544
5,504
Mississippi
6,283
6,210
5,299
5,299
Missouri
20,094
12,755
12,755
12,598
Montana
3,071
3,071
3,071
3,071
Nebraska
8,931
8,894
8,381
8,381
Nevada
2,372
2,368
2,368
2,368
New
Hampshire
330
267
267
267
New Jersey
915
773
773
773
New Mexico
2,289
2,259
2,259
2,259
New York
3,977
3,912
3,912
3,912
North
Carolina
12,355
9,209
9,209
9,180
North Dakota
12,246
12,246
12,246
11,436
Ohio
10,264
9,110
9,110
9,110
Oklahoma
10,470
10,271
9,580
9,580
Oregon
342
292
292
292
Pennsylvania
8,573
8,238
8,238
8,138
14
-------
State
2023
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
Rhode Island
279
148
148
148
South
Carolina
4,273
3,531
3,531
3,531
South Dakota
521
521
521
521
Tennessee
4,319
4,209
4,209
4,209
Texas
41,276
40,367
40,367
40,134
Utah
15,762
15,755
15,755
15,755
Vermont
54
54
54
54
Virginia
3,329
3,165
3,087
3,065
Washington
1,999
1,729
1,729
1,729
West Virginia
14,686
14,132
13,586
13,306
Wisconsin
6,321
6,315
6,315
6,295
Wyoming
11,643
11,561
10,966
10,953
Total
337,041
315,557
311,498
309,292
Note: All states are included solely for illustrative purposes. Grayed out
states are not covered by the program.
15
-------
Table B-3. 2024 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2024
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
Alabama
6,522
6,489
6,489
6,489
Arizona
7,723
7,639
7,570
7,439
Arkansas
8,955
8,927
8,927
8,927
California
1,673
1,283
1,283
1,283
Colorado
6,470
6,393
6,393
6,393
Connecticut
381
355
355
355
Delaware
423
388
388
384
Florida
12,868
10,381
10,381
10,381
Georgia
5,191
5,179
5,179
5,172
Idaho
240
240
240
240
Illinois
7,555
7,486
7,486
7,325
Indiana
12,218
11,415
11,415
11,413
Iowa
9,867
9,867
9,813
9,752
Kansas
5,510
4,763
4,763
4,763
Kentucky
13,900
13,601
12,999
12,999
Louisiana
9,974
9,459
9,459
9,363
Maine
108
86
86
86
Maryland
1,214
1,214
1,214
1,206
Massachusetts
297
265
265
265
Michigan
10,294
10,290
10,290
10,275
Minnesota
4,197
4,099
4,099
4,058
Mississippi
6,042
5,969
5,058
5,058
Missouri
18,612
11,273
11,273
11,116
Montana
3,071
3,071
3,071
3,071
Nebraska
8,931
8,894
8,381
8,381
Nevada
2,592
2,589
2,589
2,589
New
Hampshire
330
267
267
267
New Jersey
915
773
773
773
New Mexico
2,289
2,259
2,259
2,259
New York
3,977
3,912
3,912
3,912
North
Carolina
12,355
9,209
9,209
9,180
North Dakota
12,246
12,246
12,246
11,436
Ohio
9,083
7,929
7,929
7,929
Oklahoma
10,274
10,075
9,384
9,384
Oregon
342
292
292
292
Pennsylvania
8,573
8,238
8,238
8,138
16
-------
State
2024
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
Rhode Island
279
148
148
148
South
Carolina
4,273
3,531
3,531
3,531
South Dakota
521
521
521
521
Tennessee
4,064
3,983
3,983
3,983
Texas
41,276
40,367
40,367
40,134
Utah
15,924
15,917
15,917
15,917
Vermont
54
54
54
54
Virginia
3,019
2,855
2,778
2,756
Washington
1,999
1,729
1,729
1,729
West Virginia
13,185
12,784
12,239
11,958
Wisconsin
6,321
6,315
6,315
6,295
Wyoming
11,643
11,561
10,966
10,953
Total
327,773
306,578
302,519
300,330
Note: All states are included solely for illustrative purposes. Grayed out
states are not covered by the program.
17
-------
Table B-4. 2025 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2025
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
Alabama
6,522
6,489
6,489
6,489
Arizona
7,723
7,639
7,570
7,439
Arkansas
8,955
8,927
8,927
8,927
California
1,672
1,282
1,282
1,282
Colorado
6,470
6,393
6,393
6,393
Connecticut
381
355
355
355
Delaware
423
388
388
384
Florida
12,913
10,426
10,426
10,426
Georgia
5,191
5,179
5,179
5,172
Idaho
240
240
240
240
Illinois
7,555
7,486
7,486
7,325
Indiana
12,218
11,415
11,415
11,413
Iowa
9,867
9,867
9,813
9,752
Kansas
5,510
4,763
4,763
4,763
Kentucky
13,211
12,911
12,472
12,472
Louisiana
9,717
9,203
9,203
9,107
Maine
108
86
86
86
Maryland
1,214
1,214
1,214
1,206
Massachusetts
288
256
256
256
Michigan
10,294
10,290
10,290
10,275
Minnesota
4,197
4,099
4,099
4,058
Mississippi
6,022
5,949
5,037
5,037
Missouri
18,612
11,273
11,273
11,116
Montana
3,071
3,071
3,071
3,071
Nebraska
8,931
8,894
8,381
8,381
Nevada
2,549
2,545
2,545
2,545
New
Hampshire
330
267
267
267
New Jersey
915
773
773
773
New Mexico
2,232
2,201
2,201
2,201
New York
3,977
3,912
3,912
3,912
North
Carolina
12,270
9,124
9,124
9,114
North Dakota
12,246
12,246
12,246
11,436
Ohio
9,083
7,929
7,929
7,929
Oklahoma
10,266
10,068
9,376
9,376
Oregon
350
300
300
300
18
-------
State
2025
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
Pennsylvania
8,573
8,238
8,238
8,138
Rhode Island
279
148
148
148
South
Carolina
4,273
3,531
3,531
3,531
South Dakota
521
521
521
521
Tennessee
4,064
3,983
3,983
3,983
Texas
39,684
38,775
38,775
38,542
Utah
15,924
15,917
15,917
15,917
Vermont
54
54
54
54
Virginia
3,019
2,855
2,778
2,756
Washington
1,999
1,729
1,729
1,729
West Virginia
13,185
12,784
12,239
11,958
Wisconsin
6,014
6,008
6,008
5,988
Wyoming
10,429
10,347
9,752
9,739
Total
323,543
302,348
298,451
296,282
Note: All states are included solely for illustrative purposes. Grayed out
states are not covered by the program.
19
-------
Table B-5. 2026 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2026
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR
(Half)/SNCR
Retrofit
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Alabama
6,371
6,339
6,339
6,339
6,053
5,767
Arizona
5,342
5,258
5,188
5,058
4,157
3,256
Arkansas
8,728
8,700
8,700
8,700
6,365
4,031
California
1,672
1,282
1,282
1,282
1,282
1,282
Colorado
4,483
4,405
4,405
4,405
3,731
3,058
Connecticut
381
355
355
355
355
355
Delaware
423
388
388
384
384
384
Florida
11,298
8,811
8,811
8,811
8,111
7,411
Georgia
5,191
5,179
5,179
5,172
5,089
5,007
Idaho
240
240
240
240
240
240
Illinois
6,644
6,575
6,575
6,415
5,889
5,363
Indiana
9,468
8,700
8,700
8,698
8,410
8,135
Iowa
9,773
9,773
9,773
9,713
6,790
4,026
Kansas
5,510
4,763
4,763
4,763
3,938
3,112
Kentucky
13,211
12,911
12,472
12,472
10,190
7,908
Louisiana
9,704
9,189
9,189
9,093
6,370
3,810
Maine
108
86
86
86
86
86
Maryland
901
850
850
842
842
842
Massachusetts
287
256
256
256
256
256
Michigan
7,790
7,786
7,786
7,771
6,743
5,831
Minnesota
4,197
4,099
4,099
4,058
3,321
2,584
Mississippi
6,022
5,949
5,037
5,037
3,484
2,084
Missouri
18,612
11,273
11,273
11,116
9,248
7,381
Montana
3,071
3,071
3,071
3,071
2,124
1,177
Nebraska
8,931
8,894
8,381
8,381
5,672
3,070
Nevada
1,146
1,142
1,142
1,142
1,142
1,142
New
Hampshire
330
267
267
267
267
267
New Jersey
915
773
773
773
773
773
New Mexico
2,029
1,998
1,998
1,998
1,833
1,668
New York
3,977
3,912
3,912
3,912
3,650
3,388
North
Carolina
11,700
8,847
8,847
8,837
7,490
6,142
North Dakota
12,246
12,246
12,246
11,436
7,181
2,927
Ohio
9,083
7,929
7,929
7,929
7,929
7,929
Oklahoma
10,259
10,061
9,369
9,369
6,631
4,291
Oregon
350
300
300
300
300
300
20
-------
State
2026
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR
(Half)/SNCR
Retrofit
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Pennsylvania
8,362
8,010
8,010
7,910
7,512
7,158
Rhode Island
279
148
148
148
148
148
South
Carolina
4,273
3,531
3,531
3,531
3,531
3,531
South Dakota
509
509
509
509
509
509
Tennessee
4,064
3,983
3,983
3,983
3,983
3,983
Texas
39,684
38,775
38,775
38,542
31,123
23,704
Utah
9,930
9,923
9,923
9,923
6,258
2,593
Vermont
54
54
54
54
54
54
Virginia
3,019
2,855
2,778
2,756
2,565
2,373
Washington
527
257
257
257
257
257
West Virginia
13,185
12,784
12,239
11,958
10,818
9,678
Wisconsin
5,016
5,010
5,010
4,990
4,692
4,394
Wyoming
9,174
9,093
8,499
8,486
6,149
3,811
Total
298,470
277,538
273,697
271,528
223,923
177,473
Note: All states are included solely for illustrative purposes. Grayed out
states are not covered by the program.
21
-------
Table B-6. 2027 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2027
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Alabama
6,268
6,236
6,236
6,236
5,741
Arizona
5,342
5,258
5,188
5,058
3,256
Arkansas
8,728
8,700
8,700
8,700
4,031
California
1,672
1,282
1,282
1,282
1,282
Colorado
4,285
4,208
4,208
4,208
2,860
Connecticut
381
355
355
355
355
Delaware
339
312
312
308
308
Florida
11,297
8,810
8,810
8,810
7,410
Georgia
5,191
5,179
5,179
5,172
5,007
Idaho
240
240
240
240
240
Illinois
6,644
6,575
6,575
6,415
5,363
Indiana
9,468
8,700
8,700
8,698
8,135
Iowa
9,773
9,773
9,773
9,713
4,026
Kansas
5,510
4,763
4,763
4,763
3,112
Kentucky
13,211
12,911
12,472
12,472
7,908
Louisiana
9,628
9,113
9,113
9,017
3,792
Maine
108
86
86
86
86
Maryland
901
850
850
842
842
Massachusetts
287
256
256
256
256
Michigan
7,097
7,094
7,094
7,078
5,691
Minnesota
3,044
2,945
2,945
2,905
1,990
Mississippi
6,022
5,949
5,037
5,037
2,084
Missouri
18,559
11,220
11,220
11,063
7,329
Montana
3,071
3,071
3,071
3,071
1,177
Nebraska
8,247
8,210
8,177
8,177
2,974
Nevada
1,115
1,113
1,113
1,113
1,113
New
Hampshire
330
267
267
267
267
New Jersey
915
773
773
773
773
New Mexico
2,029
1,998
1,998
1,998
1,668
New York
3,977
3,912
3,912
3,912
3,388
North
Carolina
11,700
8,847
8,847
8,837
6,142
North Dakota
12,246
12,246
12,246
11,436
2,927
Ohio
9,083
7,929
7,929
7,929
7,929
Oklahoma
9,317
9,119
8,427
8,427
3,917
Oregon
350
300
300
300
300
Pennsylvania
8,362
8,010
8,010
7,910
7,158
22
-------
State
2027
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Rhode Island
279
148
148
148
148
South
Carolina
4,273
3,531
3,531
3,531
3,531
South Dakota
509
509
509
509
509
Tennessee
2,747
2,666
2,666
2,666
2,666
Texas
37,261
36,352
36,352
36,119
23,009
Utah
9,930
9,923
9,923
9,923
2,593
Vermont
54
54
54
54
54
Virginia
3,019
2,855
2,778
2,756
2,373
Washington
527
257
257
257
257
West Virginia
13,185
12,784
12,239
11,958
9,678
Wisconsin
3,442
3,436
3,436
3,416
3,416
Wyoming
9,174
9,093
8,499
8,486
3,811
Total
289,138
268,216
264,855
262,686
172,878
Note: All states are included solely for illustrative purposes. Grayed out states
are not covered by the program.
23
-------
Table B-7. 2028 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2028
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Alabama
6,268
6,236
6,236
6,236
5,741
Arizona
5,117
5,033
4,964
4,834
3,193
Arkansas
8,728
8,700
8,700
8,700
4,031
California
1,672
1,282
1,282
1,282
1,282
Colorado
3,867
3,790
3,790
3,790
2,577
Connecticut
381
355
355
355
355
Delaware
339
312
312
308
308
Florida
10,863
8,489
8,489
8,489
7,089
Georgia
5,191
5,179
5,179
5,172
5,007
Idaho
240
240
240
240
240
Illinois
5,215
5,145
5,145
4,985
4,555
Indiana
8,613
7,845
7,845
7,843
7,280
Iowa
9,773
9,773
9,773
9,713
4,026
Kansas
5,510
4,763
4,763
4,763
3,112
Kentucky
12,839
12,540
12,189
12,189
7,837
Louisiana
9,628
9,113
9,113
9,017
3,792
Maine
108
86
86
86
86
Maryland
901
850
850
842
842
Massachusetts
287
256
256
256
256
Michigan
7,097
7,094
7,094
7,078
5,691
Minnesota
3,044
2,945
2,945
2,905
1,990
Mississippi
4,076
4,003
3,716
3,716
1,752
Missouri
18,559
11,220
11,220
11,063
7,329
Montana
3,071
3,071
3,071
3,071
1,177
Nebraska
8,247
8,210
8,177
8,177
2,974
Nevada
1,115
1,113
1,113
1,113
1,113
New
Hampshire
330
267
267
267
267
New Jersey
915
773
773
773
773
New Mexico
2,029
1,998
1,998
1,998
1,668
New York
3,977
3,912
3,912
3,912
3,388
North
Carolina
11,700
8,847
8,847
8,837
6,142
North Dakota
12,246
12,246
12,246
11,436
2,927
Ohio
8,047
6,911
6,911
6,911
6,911
Oklahoma
9,317
9,119
8,427
8,427
3,917
Oregon
350
300
300
300
300
Pennsylvania
8,362
8,010
8,010
7,910
7,158
24
-------
State
2028
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Rhode Island
279
148
148
148
148
South
Carolina
4,273
3,531
3,531
3,531
3,531
South Dakota
509
509
509
509
509
Tennessee
2,212
2,130
2,130
2,130
2,130
Texas
33,189
32,280
32,280
32,047
21,623
Utah
9,930
9,923
9,923
9,923
2,593
Vermont
54
54
54
54
54
Virginia
3,019
2,855
2,778
2,756
2,373
Washington
527
257
257
257
257
West Virginia
13,185
12,784
12,239
11,958
9,678
Wisconsin
3,442
3,436
3,436
3,416
3,416
Wyoming
6,722
6,640
6,640
6,627
3,294
Total
275,363
254,572
252,518
250,349
166,688
Note: All states are included solely for illustrative purposes. Grayed out
states are not covered by the program.
25
-------
Table B-8. 2029 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2029
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Alabama
5,210
5,105
5,105
5,105
4,610
Arizona
5,117
5,033
4,964
4,834
3,193
Arkansas
7,001
6,974
6,974
6,974
3,582
California
1,672
1,282
1,282
1,282
1,282
Colorado
3,348
3,270
3,270
3,270
2,057
Connecticut
381
355
355
355
355
Delaware
339
312
312
308
308
Florida
10,863
8,489
8,489
8,489
7,089
Georgia
3,849
3,837
3,837
3,830
3,665
Idaho
240
240
240
240
240
Illinois
4,170
4,101
4,101
4,050
4,050
Indiana
7,062
6,374
6,374
6,371
5,808
Iowa
9,138
9,138
9,138
9,077
3,549
Kansas
5,510
4,763
4,763
4,763
3,112
Kentucky
11,520
11,221
10,870
10,870
7,392
Louisiana
8,897
8,383
8,383
8,286
3,639
Maine
108
86
86
86
86
Maryland
901
850
850
842
842
Massachusetts
287
256
256
256
256
Michigan
6,063
6,059
6,059
6,044
4,656
Minnesota
2,654
2,618
2,618
2,578
1,663
Mississippi
4,076
4,003
3,716
3,716
1,752
Missouri
18,559
11,220
11,220
11,063
7,329
Montana
3,071
3,071
3,071
3,071
1,177
Nebraska
8,247
8,210
8,177
8,177
2,974
Nevada
882
880
880
880
880
New
Hampshire
330
267
267
267
267
New Jersey
915
773
773
773
773
New Mexico
2,029
1,998
1,998
1,998
1,668
New York
3,977
3,912
3,912
3,912
3,388
North
Carolina
9,088
6,588
6,588
6,588
5,139
North Dakota
12,246
12,246
12,246
11,436
2,927
Ohio
7,545
6,409
6,409
6,409
6,409
Oklahoma
9,317
9,119
8,427
8,427
3,917
Oregon
350
300
300
300
300
Pennsylvania
6,032
5,680
5,680
5,580
4,828
26
-------
State
2029
Baseline
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit
Rhode Island
279
148
148
148
148
South
Carolina
3,031
2,804
2,804
2,804
2,804
South Dakota
509
509
509
509
509
Tennessee
1,198
1,198
1,198
1,198
1,198
Texas
30,134
29,225
29,225
28,992
20,635
Utah
9,930
9,923
9,923
9,923
2,593
Vermont
54
54
54
54
54
Virginia
2,578
2,414
2,337
2,334
1,951
Washington
527
257
257
257
257
West Virginia
13,185
12,784
12,239
11,958
9,678
Wisconsin
3,442
3,436
3,436
3,416
3,416
Wyoming
6,722
6,640
6,640
6,627
3,294
Total
252,584
232,812
230,758
228,728
151,697
Note: All states are included solely for illustrative purposes. Grayed out
states are not covered by the program.
As described in Section V of the Preamble, EPA identified $11,000/ton as the level of control
stringency for determining significant contribution from EGUs under the Step 3 multifactor test.
However, EPA determined that retrofitting post-combustion could not be widely accomplished
until the 2026 ozone season. Therefore, Section VI of the Preamble explains that EPA applied
the reductions available at the $l,800/ton representative cost threshold for years 2023-2025 to
arrive at a budget estimate for those years. Then, starting in 2026, EPA applied the reductions
available at the $11,000/ton representative cost threshold to arrive at a budget estimate for that
year, though for the 2026 budgets only, EPA used the "SCR (half)" rate for applicable units
rather than the rate commensurate with SCR retrofits, as discussed in section VI. A.2.a of the
Preamble. Those state-level emissions budgets for the affected states along with the
corresponding percent reduction relative to 2021 and the state's baseline emissions for that year
are shown below in Tables B-9 through B-15.23
23 A table providing state emission budgets for these linked states is provided in Appendix F
27
-------
Table B-9. OS NOx: 2023 Emissions I
2016
2021
Baseline
2023
Budget
(tons)
%
Reduction
from 2021
%
State
OS
NOx
(tons)
OS
NOx
(tons)
2023
OS NOx
(tons)
Reduction
from 2023
Baseline
Alabama
11,612
6,648
6,412
6,379
4%
1%
Arkansas
13,223
8,955
8,955
8,927
0%
0%
Illinois
14,550
11,335
7,721
7,474
34%
3%
Indiana
34,670
14,162
13,298
12,440
12%
6%
Kentucky
25,403
14,571
13,900
13,601
7%
2%
Louisiana
19,615
11,391
9,974
9,363
18%
6%
Maryland
4,471
1,428
1,214
1,206
16%
1%
Michigan
17,632
13,555
10,746
10,727
21%
0%
Minnesota
7,587
5,652
5,643
5,504
3%
2%
Mississippi
7,325
5,790
6,283
6,210
-7%
1%
Missouri
25,255
20,388
20,094
12,598
38%
37%
Nevada
2,275
2,457
2,372
2,368
4%
0%
New Jersey
2,463
1,324
915
773
42%
16%
New York
6,534
3,997
3,977
3,912
2%
2%
Ohio
24,205
11,697
10,264
9,110
22%
11%
Oklahoma
12,761
10,470
10,470
10,271
2%
2%
Pennsylvania
31,896
12,785
8,573
8,138
36%
5%
Texas
54,668
42,746
41,276
40,134
6%
3%
Utah
12,955
15,762
15,762
15,755
0%
0%
Virginia
9,833
3,329
3,329
3,143
6%
6%
West
Virginia
21,178
14,686
14,686
13,791
6%
6%
Wisconsin
7,946
6,321
6,321
6,295
0%
0%
Total
368,055
239,450
222,184
208,119
13%
6%
udget and % Reduction
28
-------
Table B-10. OS NOx: 2024 Emissions Budget and % Reduction
2016
2021
Baseline
2024
Budget
(tons)
%
Reduction
from 2021
%
State
OS
NOx
(tons)
OS
NOx
(tons)
2024
OS NOx
(tons)
Reduction
from 2024
Baseline
Alabama
11,612
6,648
6,522
6,489
2%
0%
Arkansas
13,223
8,955
8,955
8,927
0%
0%
Illinois
14,550
11,335
7,555
7,325
35%
3%
Indiana
34,670
14,162
12,218
11,413
19%
7%
Kentucky
25,403
14,571
13,900
12,999
11%
6%
Louisiana
19,615
11,391
9,974
9,363
18%
6%
Maryland
4,471
1,428
1,214
1,206
16%
1%
Michigan
17,632
13,555
10,294
10,275
24%
0%
Minnesota
7,587
5,652
4,197
4,058
28%
3%
Mississippi
7,325
5,790
6,042
5,058
13%
16%
Missouri
25,255
20,388
18,612
11,116
45%
40%
Nevada
2,275
2,457
2,592
2,589
-5%
0%
New Jersey
2,463
1,324
915
773
42%
16%
New York
6,534
3,997
3,977
3,912
2%
2%
Ohio
24,205
11,697
9,083
7,929
32%
13%
Oklahoma
12,761
10,470
10,274
9,384
10%
9%
Pennsylvania
31,896
12,785
8,573
8,138
36%
5%
Texas
54,668
42,746
41,276
40,134
6%
3%
Utah
12,955
15,762
15,924
15,917
-1%
0%
Virginia
9,833
3,329
3,019
2,756
17%
9%
West
Virginia
21,178
14,686
13,185
11,958
19%
9%
Wisconsin
7,946
6,321
6,321
6,295
0%
0%
Total
368,055
239,450
214,624
198,014
17%
8%
29
-------
Table B-ll. OS NOx: 2025 Emissions Budget and % Reduction
2016
2021
Baseline
2025
Budget
(tons)
%
Reduction
from 2021
%
State
OS
NOx
(tons)
OS
NOx
(tons)
2025
OS NOx
(tons)
Reduction
from 2025
Baseline
Alabama
11,612
6,648
6,522
6,489
2%
0%
Arkansas
13,223
8,955
8,955
8,927
0%
0%
Illinois
14,550
11,335
7,555
7,325
35%
3%
Indiana
34,670
14,162
12,218
11,413
19%
7%
Kentucky
25,403
14,571
13,211
12,472
14%
6%
Louisiana
19,615
11,391
9,717
9,107
20%
6%
Maryland
4,471
1,428
1,214
1,206
16%
1%
Michigan
17,632
13,555
10,294
10,275
24%
0%
Minnesota
7,587
5,652
4,197
4,058
28%
3%
Mississippi
7,325
5,790
6,022
5,037
13%
16%
Missouri
25,255
20,388
18,612
11,116
45%
40%
Nevada
2,275
2,457
2,549
2,545
-4%
0%
New Jersey
2,463
1,324
915
773
42%
16%
New York
6,534
3,997
3,977
3,912
2%
2%
Ohio
24,205
11,697
9,083
7,929
32%
13%
Oklahoma
12,761
10,470
10,266
9,376
10%
9%
Pennsylvania
31,896
12,785
8,573
8,138
36%
5%
Texas
54,668
42,746
39,684
38,542
10%
3%
Utah
12,955
15,762
15,924
15,917
-1%
0%
Virginia
9,833
3,329
3,019
2,756
17%
9%
West
Virginia
21,178
14,686
13,185
11,958
19%
9%
Wisconsin
7,946
6,321
6,014
5,988
5%
0%
Total
368,055
239,450
211,707
195,259
18%
8%
30
-------
Table B-12. OS NOx: Preset 2026 Emissions Budget and % Reduction
2016
2021
Baseline
Preset
%
Reduction
from 2021
%
State
OS
OS
2026
2026
Reduction
NOx
NOx
OS NOx
Budget
from 2026
(tons)
(tons)
(tons)
(tons)
Baseline
Alabama
11,612
6,648
6,371
6,339
5%
1%
Arkansas
13,223
8,955
8,728
6,365
29%
27%
Illinois
14,550
11,335
6,644
5,889
48%
11%
Indiana
34,670
14,162
9,468
8,410
41%
11%
Kentucky
25,403
14,571
13,211
10,190
30%
23%
Louisiana
19,615
11,391
9,704
6,370
44%
34%
Maryland
4,471
1,428
901
842
41%
7%
Michigan
17,632
13,555
7,790
6,743
50%
13%
Minnesota
7,587
5,652
4,197
4,058
28%
3%
Mississippi
7,325
5,790
6,022
3,484
40%
42%
Missouri
25,255
20,388
18,612
9,248
55%
50%
Nevada
2,275
2,457
1,146
1,142
54%
0%
New Jersey
2,463
1,324
915
773
42%
16%
New York
6,534
3,997
3,977
3,650
9%
8%
Ohio
24,205
11,697
9,083
7,929
32%
13%
Oklahoma
12,761
10,470
10,259
6,631
37%
35%
Pennsylvania
31,896
12,785
8,362
7,512
41%
10%
Texas
54,668
42,746
39,684
31,123
27%
22%
Utah
12,955
15,762
9,930
6,258
60%
37%
Virginia
9,833
3,329
3,019
2,565
23%
15%
West
21,178
14,686
13,185
10,818
26%
18%
Virginia
Wisconsin
7,946
6,321
5,016
4,990
21%
1%
Total
368,055
239,450
196,225
151,329
37%
23%
31
-------
Table B-13. OS NOx: Preset 2027 Emissions Budget and % Reduction
2016
2021
Baseline
Preset
%
Reduction
from 2021
%
State
OS
OS
2027
2027
Reduction
NOx
NOx
OS NOx
Budget
from 2027
(tons)
(tons)
(tons)
(tons)
Baseline
Alabama
11,612
6,648
6,268
6,236
6%
1%
Arkansas
13,223
8,955
8,728
4,031
55%
54%
Illinois
14,550
11,335
6,644
5,363
53%
19%
Indiana
34,670
14,162
9,468
8,135
43%
14%
Kentucky
25,403
14,571
13,211
7,908
46%
40%
Louisiana
19,615
11,391
9,628
3,792
67%
61%
Maryland
4,471
1,428
901
842
41%
7%
Michigan
17,632
13,555
7,097
5,691
58%
20%
Minnesota
7,587
5,652
3,044
2,905
49%
5%
Mississippi
7,325
5,790
6,022
2,084
64%
65%
Missouri
25,255
20,388
18,559
7,329
64%
61%
Nevada
2,275
2,457
1,115
1,113
55%
0%
New Jersey
2,463
1,324
915
773
42%
16%
New York
6,534
3,997
3,977
3,388
15%
15%
Ohio
24,205
11,697
9,083
7,929
32%
13%
Oklahoma
12,761
10,470
9,317
3,917
63%
58%
Pennsylvania
31,896
12,785
8,362
7,158
44%
14%
Texas
54,668
42,746
37,261
23,009
46%
38%
Utah
12,955
15,762
9,930
2,593
84%
74%
Virginia
9,833
3,329
3,019
2,373
29%
21%
West
21,178
14,686
13,185
9,678
34%
27%
Virginia
Wisconsin
7,946
6,321
3,442
3,416
46%
1%
Total
368,055
239,450
189,177
119,663
50%
37%
32
-------
Table B-14. OS NOx: Preset 2028 Emissions Budget and % Reduction
2016
2021
Baseline
Preset
%
Reduction
from 2021
%
State
OS
OS
2028
2028
Reduction
NOx
NOx
OS NOx
Budget
from 2028
(tons)
(tons)
(tons)
(tons)
Baseline
Alabama
11,612
6,648
6,268
6,236
6%
1%
Arkansas
13,223
8,955
8,728
4,031
55%
54%
Illinois
14,550
11,335
5,215
4,555
60%
13%
Indiana
34,670
14,162
8,613
7,280
49%
15%
Kentucky
25,403
14,571
12,839
7,837
46%
39%
Louisiana
19,615
11,391
9,628
3,792
67%
61%
Maryland
4,471
1,428
901
842
41%
7%
Michigan
17,632
13,555
7,097
5,691
58%
20%
Minnesota
7,587
5,652
3,044
2,905
49%
5%
Mississippi
7,325
5,790
4,076
1,752
70%
57%
Missouri
25,255
20,388
18,559
7,329
64%
61%
Nevada
2,275
2,457
1,115
1,113
55%
0%
New Jersey
2,463
1,324
915
773
42%
16%
New York
6,534
3,997
3,977
3,388
15%
15%
Ohio
24,205
11,697
8,047
6,911
41%
14%
Oklahoma
12,761
10,470
9,317
3,917
63%
58%
Pennsylvania
31,896
12,785
8,362
7,158
44%
14%
Texas
54,668
42,746
33,189
21,623
49%
35%
Utah
12,955
15,762
9,930
2,593
84%
74%
Virginia
9,833
3,329
3,019
2,373
29%
21%
West
21,178
14,686
13,185
9,678
34%
27%
Virginia
Wisconsin
7,946
6,321
3,442
3,416
46%
1%
Total
368,055
239,450
179,467
115,193
52%
36%
33
-------
Table B-15. OS NOx: Preset 2029 Emissions Budget and % Reduction
2016
2021
Baseline
Preset
%
Reduction
from 2021
%
State
OS
OS
2029
2029
Reduction
NOx
NOx
OS NOx
Budget
from 2029
(tons)
(tons)
(tons)
(tons)
Baseline
Alabama
11,612
6,648
5,210
5,105
23%
2%
Arkansas
13,223
8,955
7,001
3,582
60%
49%
Illinois
14,550
11,335
4,170
4,050
64%
3%
Indiana
34,670
14,162
7,062
5,808
59%
18%
Kentucky
25,403
14,571
11,520
7,392
49%
36%
Louisiana
19,615
11,391
8,897
3,639
68%
59%
Maryland
4,471
1,428
901
842
41%
7%
Michigan
17,632
13,555
6,063
4,656
66%
23%
Minnesota
7,587
5,652
2,654
2,578
54%
3%
Mississippi
7,325
5,790
4,076
1,752
70%
57%
Missouri
25,255
20,388
18,559
7,329
64%
61%
Nevada
2,275
2,457
882
880
64%
0%
New Jersey
2,463
1,324
915
773
42%
16%
New York
6,534
3,997
3,977
3,388
15%
15%
Ohio
24,205
11,697
7,545
6,409
45%
15%
Oklahoma
12,761
10,470
9,317
3,917
63%
58%
Pennsylvania
31,896
12,785
6,032
4,828
62%
20%
Texas
54,668
42,746
30,134
20,635
52%
32%
Utah
12,955
15,762
9,930
2,593
84%
74%
Virginia
9,833
3,329
2,578
1,951
41%
24%
West
21,178
14,686
13,185
9,678
34%
27%
Virginia
Wisconsin
7,946
6,321
3,442
3,416
46%
1%
Total
368,055
239,450
164,053
105,201
56%
36%
34
-------
3. Variability Limits
Once EPA determined state-emission budgets representative of the control stringency,
EPA calculated the minimum variability limits and assurance levels for each state based on the
calculated emission budgets. Each state's minimum variability limit is calculated as 21% of its
budget, and its assurance level is the sum of its budget and variability limit (or 121% of its
budget)24 The minimum variability limits and assurance levels are further described and shown
in section VI of the preamble for this rule. (In a control period where a state's emissions budget
is the dynamic budget rather than the preset budget, the variability limit will be computed as a
percentage of the dynamic budget rather than a percentage of the preset budget.)
4. Calculating Dynamic Budgets Starting in 2026
The dynamic budgets methodology for 2026 and subsequent years begins with the data reported
to CAMD, similar to the engineering analysis used to determine the preset 2023 through 2029
preset state budgets. Dynamic budgets utilize predetermined emission rates (relying on the same
historical data and methodology described for the preset emission budgets) for each unit. The
dynamic budget methodology differs from the methodology used to determine preset emission
budgets in that the dynamic methodology takes that emission rate and multiplies it by heat-input
values reported and calculated from the most recent data at the time of calculation {i.e., data not
yet available) instead of the most recent data available at time of rule promulgation (e.g., 2021
heat input data) to estimate unit and state emissions (i.e., state emission budgets). Preamble
Section VI.B.4.b describes how EPA uses a rolling, multi-year heat input data set to derive a
normalized unit-level heat input value. This updating heat input value is the dynamic variable
which makes the state emissions budgets dynamic. The dynamic heat inputs are multiplied by
preset unit-level emission rates prescribed for each year in the dynamic budget templates in
Appendix A: State Emissions Budget Calculations and Engineering Analytics to get an emissions
amount for each unit, and the resulting unit-level emissions amounts for all the units in a state are
summed to determine the dynamic state-level budget for the year. That Appendix has a
worksheet titled "Dynamic Budget 2026 Template", and a second titled "Dynamic Budget 2027+
Template". These worksheets don't show the dynamic budgets for those future years, but they
provide the unit-level NOx rates and the heat input fields to be populated with future data that
EPA will use to calculate dynamic budgets for each future year. These worksheets reflect the
initial inventory of EGUs used to derive the dynamic ozone season state emissions budget for
each control period in 2026 and thereafter.
Inventory of EGUs for determining dynamic budget
The unit name and corresponding facility detail such as state, ORIS, Boiler, Plant Type
are listed in columns A through Q of the "dynamic budget 2026" and "dynamic budget
2027+" worksheets.
24 As described in Section VI of the Preamble for this rule, the EPA is finalizing a minimum variability limit of 21%.
Starting in the 2023 control period, the variability limit would be the higher of 21 percent or the percentage (if any)
by which the total reported heat input of the state's affected EGUs in the control period exceeds the total reported
heat input of the state's affected EGUs as reflected in the state's emissions budget for the control period. EPA
expects that the minimum 21 percent value would apply in almost all instances.
35
-------
The inventory of units in these worksheets reflects EPA's assessment of the future
inventory based on current data. It is not an applicability determination, and the eventual
inventory of units comprising the dynamic budgets may be slightly expanded (e.g.,
reflecting new units that come online) or slightly reduced (e.g., reflecting units that have
ceased operation) at the time of issuing the dynamic budgets.
The anticipated inventory of units used to calculate the dynamic budget for each control
period is identified as follows:
o Units that, to the best of EPA's knowledge, are affected under the rule, that
reported heat input for the historical control period two years before the year of
control period for which the dynamic budget is being calculated (e.g., for
calculation of the 2026 budgets, heat input was reported in 2024); and that had a
deadline for certification of monitoring systems under § 97.1030(b) by May 1 of
that historical control period (e.g., by May 1st of 2024 for the 2026 state budget
calculation) will be included in the dynamic budget calculations.25
o New units will be included in the dynamic budget calculations starting with the
first control period for which the units have reported a full control period of data
following their monitor certification deadlines. For example, a unit with a
deadline for certification of monitoring systems under § 97.1030(b) by May 1st of
2024 that reports heat input during the 2024 control period will be included in the
2026 dynamic state budget calculation. EPA will rely on reported CAMD Power
Sector Emissions data to identify these units.
Unit-level emission rate, heat input, and emissions data for dynamic budget
For each of the units identified in the above inventory, EPA populates a pre-determined
emission rate. Where available, this rate comes directly from the Engineering Analytic
unit-files described above and used in preset budget calculations. EPA applies the
emission rate reflecting the selected control stringency. For the "dynamic budget 2026"
worksheet, these emission rates come from the "unit 2026" worksheet, and are calculated
by dividing the unit-level emissions value from column AN into the unit-level heat input
value from column X in the "unit 2026" worksheet. These unit-level emission rate
reflects the control stringency identified in EPA's determination of significant
contribution applied to these units in 2026. For the "dynamic budget 2027+" worksheet,
these emission rates come from column AR in the "unit 2027" worksheet, which are
calculated by dividing the unit-level emissions value from column AO into the unit-level
heat input value from column X in the "unit 2027" worksheet. The "unit 2026" and "unit
2027" worksheets reflect lower emission rates for some units where post-combustion
25 For the 2026 budget calculation, this will generally be the same inventory of units included in the "unit 2026 file"
for Group 3 states, except that a unit that actually operates in the 2024 control period will be included in calculating
the state's 2026 dynamic budget even if, for purposes of calculating the 2026 preset budgets in this rulemaking, the
unit was assumed to be retired in 2026.
36
-------
control retrofit potential is identified.26 2027 reflects full implementation of EPA
identified stringency measures, so the rates identified in the "Dynamic Budget 2027+
worksheet will not change to reflect any further stringency level, consequently it will be
utilized for each dynamic budget year after 2027 as well.
There are two types of units (new units, and 2021 non-operating units) for which the
above step would not yield an assumed emission rate. Therefore, EPA populates an
assumed emission rate based on the following:
o For new units, EPA applies the following assumed emission rates for well
controlled units identified for each generation type as discussed in the EGU NOx
Mitigation Strategies Final Rule TSD27:
Applied New Unit Emission Rates for Dynamic Budgets
Unit Type
Assumed NOx Emission Rate
(lb/MMBtu)
Coal Steam
0.05
Oil/Gas Steam
0.03
Combustion Turbine
0.011
Combined Cycle
0.011
All other fossil
0.05
o For 2021 non-operating units (thus lacking any identified emission rate in the
"unit 2024" file), EPA applies an emission rate based on that unit's last year in
which it had ozone season operating data prior to 2021. These units are flagged as
having "substitute data" in the dynamic budget templates. If that rate exceeds the
assumed step 3 technology in effect for that year (e.g., SCR optimization in 2026
for a coal steam unit with an existing SCR), then the emission rate will be
adjusted down to that level (e.g., 0.08 lb/MMBtu). If these units have no operating
data from a prior ozone season, than they would be assigned rates according to the
table above.
These corresponding emission rates for all units are shown in column R of the "dynamic
budget 2026", and "dynamic budget 2027+" worksheet.
Columns T through X in the "dynamic budget" worksheets will reflect the updated heat
input for the units as it becomes available. This is the dynamic variable, and it will be
populated through future ministerial actions. For instance, these columns would be
populated with heat input values from 2020-2024 for the 2026 dynamic budget
26 The emission rate for Alabama, Minnesota, and Wisconsin continue to be identified by column AQ at this step as
those states are not subject to the post-combustion control stringency assumptions. For any expected unit-level coal-
to-gas switch identified in the "Unit 2026" worksheet or later years, the emission rates in the dynamic budget
worksheet reflects their expected plant type as of 2025.
27 Combined cycle and combustion turbines with SCR retrofits can achieve emission rates as low as 0.002
lb/MMBtu (see "Combustion Turbine NOX Technology Memo" (January 2022) EPA-HQ-OAR-2021-0668-0085),
although EPA assumes a floor rate of 0.011 lb/MMBtu for this analysis, marching the assumed floor rate used in
IPM.
37
-------
calculation. For the 2027 dynamic budget" worksheet, these columns will be populated
with heat input values from 2021-2025, and so forth, and so forth.
Column Y reflects the average heat input from the highest three heat input values from
the five year baseline captured in columns T through X (this is the representative unit-
level heat input).
Column Z reflects the representative unit level heat input from column W divided by the
state total of representative unit-level heat inputs.
Column AA-AC reflect the state's heat input over the last three available and column AD
reflects the average of these three years (this is the Representative State Level Heat Input
value).28
Column AE reflects the unit's normalized unit-level heat input obtained by multiplying
the representative unit-level percent of state total (column Z) by the representative state
level heat input (column AD).29
Column AF reflects the unit-level assumed emissions for the purposes of state emissions
budget quantification. This value will be obtained by multiplying the emission rate (in
column R) by the normalized unit-level heat input value (column AE). The product is
divided by 2,000 to convert from pounds to short tons.
Summation of the unit-level emission estimates to derive the given year's dynamic budget
After completing the above steps, the unit-level emission values that will be identified in
column AF of each "dynamic budget" worksheet are summed to the state level. These states
(those 22 covered for EGU Group 3 under this action) and state-level values (in tons) are
displayed in columns AH and AI of the same "dynamic budget" worksheet. These tonnage
values in column AI reflect the state dynamic budgets for the given year (starting in 2026). At
this step, a rounding function is applied to express the values to the nearest ton. These state
dynamic budgets will be calculated and made public approximately 1 year prior to the beginning
of the control period for that vintage year (e.g., 2026 dynamic budgets will be announced in
summer of 2025) through the schedule identified in Section VI.A of the preamble.
The procedure for computing a state's dynamic emissions budget for a control period can be
expressed in terms of the following formula:
28 For the 2022 and 2023 state heat input totals, the EPA incorporated heat input adders at this step for Utah and
Nevada to reflect the total estimated heat input and emissions from fifteen units that are likely to be considered
existing units for purposes of the dynamic budget calculations starting with the 2026 control period but that do not
report data under the Acid Rain Program and consequently did not report data for the 2022 control period and are
not expected to report data for the 2023 control period. The units and the amounts of ozone season heat input
assumed for each unit are listed in preamble Table VLB.3-1.
29This value is left blank for unit that reports no heat input in the year two years before the year of the control period
for which the dynamic trading budget is being calculated.
38
-------
71
dbp = Y(
Avg HIi
x Avg HIS x ERt)
Ia=iAvg HIi
Where:
DBp = the dynamic emissions budget for a state for control period "p" in pounds;
Avg His = the average of the sum of the total control period heat input values reported
under 40 CFR part 75 for all affected units in the state for the control periods in the years
two, three, and four years before control period "p" (whether or not the units operated
during the control period two years before control period "p") (This is referred to as the
"Representative State-Level Heat Input");
Avg HI; = the average of the three highest of the five total control period heat input
values reported under 40 CFR part 75 for unit "i" for the control periods in the years two,
three, four, five, and six years before control period "p" (excluding any control period
that commenced before the unit's first deadline to begin reporting heat input under 40
CFR part 75 under any regulatory program), or if there are fewer than three non-zero
values for the unit from the five control periods, the average of all the non-zero values
(This is referred to as the "Representative Unit-Level Heat Input");
ERi = the NOx emissions rate shown for unit "i" and control period "p" in the document
"Unit-Specific Ozone Season NOx Emissions Rates for Dynamic Budget Calculations"
posted at www.regulations.gov in docket EPA-HQ-OAR-2021-0668 or, for a unit not
listed in that document, the NOx emissions rate identified according to the type of unit
and (where applicable) the type of fuel combusted by the unit during the control period
containing the unit's deadline for certification of monitoring systems for the Group 3
trading program under 40 CFR 97.1030(b) as follows:
0.011 lb/MMBtu, for a simple cycle combustion turbine or a combined cycle
combustion turbine other than an integrated coal gasification combined cycle unit;
0.030 lb/MMBtu, for a boiler combusting only fuel oil or gaseous fuel (other than
coal-derived fuel) during such control period; or
0.050 lb/MMBtu, for a boiler combusting any amount of coal or coal-derived fuel
during such control period or any other unit not covered by the two preceding
paragraphs;
p = designator for the control period in a given year;
i = designator for an individual affected unit in the state whose first deadline to begin
reporting heat input under 40 CFR part 75 under any regulatory program was on or
before May 1 of the control period two years before control period "p" and that reported
heat input under 40 CFR part 75 during the control period two years before control period
p"; and
39
-------
n = number of affected units in the state whose first deadline to begin reporting heat input
under 40 CFR part 75 under any regulatory program was on or before May 1 of the
control period two years before control period "p" and that reported heat input under 40
CFR part 75 for the control period two years before control period "p".
40
-------
C. 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 V. A-D applying Step 3 of the 4-Step Good Neighbor Framework) which is
based on cost, emissions, and air quality factors. A key quantitative input for this analysis is the
predicted downwind ambient air quality impacts at various levels of NOx emission control
assessed for upwind EGU and non-EGU sources. The emission reductions associated with the
various cost thresholds analyzed for this rule are expected to result in different amounts of air
quality improvement at the downwind receptors. The downwind air quality impacts are also used
to inform EPA's assessment of potential overcontrol, as discussed in more detail below.
Air quality modeling would be the optimal way to estimate the air quality impacts at each
cost threshold level from EGU and non-EGU emissions reductions. However, due to time and
resource limitations EPA was unable to use photochemical 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 reductions and to ensure that each step of its analysis is informed by the evolving
emissions data, EPA used a simplified air quality assessment tool (AQAT) to interpolate between
existing photochemical modeling cases.30 The simplified tool allows the Agency to analyze
many more levels of NOx control stringency than would otherwise be possible.31 EPA
recognizes that AQAT is not the equivalent of photochemical air quality modeling but in the
Agency's view is adequate to this purpose. AQAT is built using air quality modeling data and
facilitates the use of existing photochemical air quality modeling estimates.
The use of AQAT to generate "appropriately reliable projections of air qualtiy conditions
and contributions" when there is limited time to conduct full-scale photochemical grid modeling
was upheld by the D.C. Circuit in MOG v. EPA, No. 21-1146 (D.C. Cir. March 3, 2023):,
Based on the record before us, EPA appears to have chosen analytical techniques
rationally connected to the Revised Rule and appropriately explained its use of the linear
interpolation and subsequent methods for establishing the Revised Rule. In addition,
EPA's methodology did also incorporate photochemical modeling, [petitioner's]
preferred technique, as the "foundation for its projections" and "merely layered an
additional mathematical function, linear interpolation" over the original projected data to
generate 2021 ozone concentrations. EPA then performed further data analysis by
checking its 2021 interpolated projection against both a sensitivity analysis and
engineering analytics approach.
[...] EPA also was cognizant of the CAA's statutory directive that emissions reductions
should be done "as expeditiously as practicable." [CAA section 181(a)(1)], Given the
30 EPA used CAMx to model several base cases (i.e., one of 2016, one of 2023, and one of 2026). The EPA
calculated air quality contributions for each state for both the 2023 and 2026 cases. In addition, EPA modeled with
source apportionment the 2026 final policy control case. At proposal, EPA also modeled the 2026 base case and a
2026 case with air quality contributions where EGU and non-EGU emissions were uniformly reduced by 30%.
31 As an example, each AQAT estimate under the Step 3 methodology focuses on the specific air quality linkages for
an individual receptor and the air quality effects of emission reductions from those specific states. Consequently, for
-700 receptors, each with a specific pattern of states contributing greater than or equal to the 1% threshold, and 6
levels of stringency, this would entail 4,200 individual photochemical air quality modeling simulations to replicate.
41
-------
limited amount of time EPA had to complete the rulemaking for the Revised Rule, we
discern that EPA reasonably chose to use existing air quality modeling and contribution
information to derive an appropriately reliable projection of air quality conditions and
contributions in 2021. . . . [I]n the context of the deferential standard afforded EPA,
[petitioner] has not established that EPA's linear interpolation method is oversimplified
or that the agency has produced unreasonable results.
Midwest Ozone Group v. EPA, No. 21-1146 (D.C. Cir.), Slip Op. at 11 (internal cites omitted).
See id. (quoting Appalachian Power Co. v. EPA, 135 F.3d 791, 802 (D.C. Cir. 1998)) ("[S]o long
as EPA 'acted within its delegated statutory authority, ... we will not interfere with its
conclusion." (quoting Ethyl Corp. v. EPA, 51 F.3d 1053, 1064 (D.C. Cir. 1995)).
In this rulemaking, as in the Revised CSAPR Update, the Agency also determines there is
utility in the AQAT methodology for estimating downwind air quality impacts for various NOx
emission reduction strategies, particularly in light of the timing considerations explained in the
preamble in section IV.A. As explained above, assessing downind air quality impacts using
CAMx photochemical air quality modeling would require running hundreds, if not thousands, of
time-and resource-intensive simulations. In comparison to the AQAT tool used to support
Revised CSPAR Update, the EPA has updated the AQAT tool using the most recent air quality
modeling available and improved the tool by making it more state-specific as explained in more
detail section C.2 of this TSD. And, using AQAT, the EPA conducted the same types of
sensitivity analyses generated to support the Revised CSPAR Update (sections C.3, C.4, and
Appendix J) as well as some additional sensitivity analyses (Appendices H and K). The results of
these sensitivity analyses confirm the reliability of EPA's assessment of downwind air quality
impacts using the AQAT tool for this rulemaking.
AQAT has evolved through iterative development under the original CSAPR, the CSAPR
Update, and the Revised CSAPR Update. One evolution was incorporating a second source
apportionment photochemical modeling emissions case in order to improve the interpolation.
This was done by aligning the change in air quality concentration with the change in emissions
using a calibration factor. This creates a specific calibration factor for each state for each
receptor, rather than a single calibration factor uniformly to states for each receptor. EPA
examined several emissions scenarios for the year 2026 using two different calibration factors as
a mechanism to estimate the range of results.
The inputs and outputs of the tool can be found in the "Ozone=AQAT=Final.xlsx" excel
workbook.32
The remainder of section C 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 emissions cost threshold analyses.
32 The AQAT estimates in the workbook are based on EGU emission estimates completed on Jan 20, 2023 and may
not represent the final emission estimates used in the rule.
42
-------
1. Introduction
The ozone AQAT was developed for use in the Step 3 air quality analysis as part of the
multi-factor test. Specifically, the AQAT was designed to evaluate air quality improvements in
response to emissions changes, allowing evaluation of total air quality improvement at each
receptor, an assessment of whether each receptor is above or below the NAAQS, and an
assessment of each state's air quality contribution relative to the linkage threshold. EPA
described and used a similar tool in the original CSAPR to evaluate good neighbor obligations
with respect to the ozone and fine particulate matter (PM2.5) NAAQS and in both the CSAPR
Update and final Revised CSAPR Update to evaluate good neighbor obligations with respect to
ozone. For the CSAPR Update, EPA refined the construction and application of the assessment
tool to improve estimates of changes in ozone concentrations in response to changes in NOx
emissions. This methodology was used again in the Revised CSAPR Update. Here, we extend
the methodology developed in the CSAPR Update rulemaking to calibrate the response of a
pollutant using two CAMx simulations at different emission levels where we have full sets of
state level emissions and contribution data.33'34
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, on a state-by-
state and receptor-by-receptor basis, 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.35 For
the purposes of developing and using an assessment tool to compare the air quality impacts of
NOx emission reductions under various emission reduction cost threshold scenarios, we
determine the relationship between changes in emissions and changes in ozone contributions on
a state-by-state and receptor-by-receptor basis. Specifically, EPA assumed that, within the range
of total NOx emissions being considered (as defined by the cost threshold emission scenarios), a
change in ozone season NOx emissions leads to a proportional change in downwind ozone
contributions.36 This proportional relationship was then modified using calibration factors based
on state-specific source apportionment (i.e., contribution) air quality modeling from proposal
(the 2026 base case and the sector-specific reduction scenario where the 2026 base case EGU
and non-EGU NOx emissions were reduced by 30% in each state). At final, the air quality
contributions from these two air quality modeling scenarios from proposal were reassessed, with
the contributions recalculated based on the contribution days identified in the 2023 final rule
contribution modeling. These "primary" calibration factors were used for all scenarios at final.
33 In CSAPR, we estimated changes in sulfate using changes in SO2 emissions.
34 In this rule, we used CAMx to calibrate the assessment tool's predicted change in ozone concentrations to changes
in NOx emissions. This primary calibration is state and receptor-specific and is derived using air quality modeling
from the proposed rule based on the changes in NOx emissions and resulting ozone concentrations between the 2026
base case and a 2026 control scenario where EGU and non-EGU emissions were simultaneously reduced by 30%.
As a sensitivity assessment, we used the alternative state and receptor-specific calibrations using the state and
receptor specific differences in air quality contributions and emissions between the 2026 base case and the 2023
base case.
35 As discussed in more detail in section C.5 of this TSD.
36As discussed in more detail in section C.4 of this TSD.
43
-------
Additionally, the 2023 and 2026 base case contribution modeling results from the final
rule were utilized to create an independent set of "alternative" calibration factors that were used,
in turn, to assess the AQAT results created using the primary calibration factors (see section C.4
of this TSD for more details of this assessment). Since these "alternative" calibration factors are
based on reductions from multiple source sectors that changed between 2023 and 2026 (e.g.,
mobile sources and all other anthropogenic sources of NOx), an AQAT using these "alternative"
factors could be used to evaluate cases that included emissions reductions outside of the EGU
and non-EGU sectors. Since the primary calibration factors are based exclusively on emissions
reductions from the source sectors being regulated in this rule and exclude emission reductions
from sectors that are not being regulated (and which may have different emissions patterns and
emissions release heights), EPA elected to use the primary calibration factors for the AQAT-
based assessment for Step 3 and for its overcontrol assessment. Section C.5 describes the factors
and assumptions that affect the calibration factors.
The calibration factors are designed for the purpose of adjusting the ozone response in
response to emissions changes in order to reflect the non-linear, non-one-to-one proportional
relationship between changes in NOx emissions and the associated changes in ozone. For
example, given a particular state and receptor in 2026, we could assume that a 20% decrease in
an upwind state's emissions leads to a 20% decrease in its downwind ozone contribution in the
"uncalibrated" ozone AQAT, while following the application of the primary calibration factor
the downwind ozone contribution may only decrease by 10% in "calibrated" AQAT (where the
calibration factor is 0.5). Typically, the calibration factors were substantially less than one, often
to the order of 0.3, for the downwind states containing the receptors, (thus, a 10% decrease in
emissions from a particular state would result in a 3% decrease in the ozone contribution from
that state), while the calibration factors for upwind states farther from the downwind receptor
increased to values around 1 (where a 10% reduction in emissions would result in a 10%
decrease in ozone contribution from the emitting state). Consequently, in a relative sense (i.e., on
a percentage basis), emission reductions from farther away states are more-effective than states
near the receptor in reducing that state's contribution,37 The reason for this relationship is the
difference in the chemical state of the emissions as they cycle between NOx and ozone due to
encounters with various oxidative/reductive chemical regimes and meteorological conditions
during transport. The creation of the calibration factors is described in detail in section C.2.c (1)
of this TSD.
Section C.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 C.3.
In summary, EPA conducted a variety of AQAT scenarios38 summarized in the table
below to inform its primary Step 3 evaluation. The results discussed in the remainder of the
document pertain to the scenarios described in Table C-l, which reflect alternative views of
future emissions. Each of these scenarios was examined using two configurations of AQAT
where the patterns of reductions were adjusted between a single-receptor oriented "Step 3"
37 The CAMx photochemical modeling used to create the state- and receptor-specific calibration factors (that was
developed in this rule) allows EPA to make this observation.
38 EPA uses the word scenario and case interchangeably, referring to a cost threshold level of OS NOx emissions
reductions from EGUs and non-EGUs.
44
-------
configuration (the approach used in Step 3) and a full geography control configuration (where
the overall effects of the rule are applied to all receptors). The "Full Geography" configuration
results are shown in Appendix D. Next, we examined the results when a separate calibration
approach was applied.
Table C-l - Summary of Scenarios Evaluated with AQAT
Scenario
Summary
$0
Baseline
$1,600
Baseline + SCR optimize
$1,600
Baseline +SCR optimize + SOA CC
$1,800
Baseline +SCR/SNCR optimize
$1,800
Baseline +SCR/SNCR optimize + SOA CC
$11,000 (i.e.,
"Full Step 3,
EGU only")
Baseline +SCR/SNCR optimize + SOA CC + SCR Retrofit
$11,000+ non-
EGUs (i.e., "Full
Step 3")
Baseline +SCR/SNCR optimize + SOA CC + SCR Retrofit + non-EGUs
$1,800+ non-
EGUs
Baseline +SCR/SNCR optimize + SOA CC + non-EGUs
CAMx AQ
Modeling Final
Rule Policy
Control
Emission levels associated with the CAMx photochemical AQ modeling of the
final rule policy control scenario.
$0 w/IRA
Baseline + delta in emissions between IPM base and IPM base w/IRA
$11,000 w/IRA
Baseline +SCR/SNCR optimize + SOA CC + SCR Retrofit + delta in
emissions between IPM final policy and IPM final policy w/IRA
$11,000+ non-
EGUs w/IRA
Baseline +SCR/SNCR optimize + SOA CC + SCR Retrofit + non-EGUs +
delta in emissions between IPM final policy and IPM final policy w/IRA
*A11 "baseline" references entail Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs.
All non-EGU scenarios were only evaluated in 2026.
"Non-EGUs" in the context of this TSD refer to the suite of emissions controls and emissions reductions identified
at Step 3 for all of the non-EGU industries.
Configuration and Calibration Factor Sensitivities: For each scenario above, EPA ensured the
robustness of its finding by doing the analysis with its "Primary" calibration approach as well an
"Alternative" AQAT calibration approach.
Primary Calibration - state- and monitor-specific calibrations created using the
relationships between NOx emissions reductions and air quality improvements derived
using the 2026 base case and 2026 reduction case (where EGUs and non-EGUs had their
emissions reduced by 30%). Both of these model runs were done at proposal.
Alternative AQAT Calibration - state- and monitor-specific calibrations created using the
relationships between NOx emissions reductions and air quality improvements derived
45
-------
using the 2023 base case and 2026 base cases (where source sectors across the emissions
inventory made reductions). Results from this calibration are discussed in Section C.4.
We also performed sensitivities for each of the rows in Table C-l reflecting two different
approaches to assessing the effects of the rule, which we will refer to as "configurations." These
approaches are summarized here and further discussed in section C.2.(c).2 below.
Step 3 Configuration - For the "Step 3" configurations, all states that contributed at or
above 1% of the NAAQS to a particular monitor in the air quality modeling base case for
the year being analyzed (either 2023 or 2026), as well as the state containing the monitor
were simultaneously adjusted to the emission levels for each of the scenarios in Table C-
1. At that particular monitor all other states were adjusted to the engineering base case
level. This approach forms our primary analysis, the results of which are discussed in the
preamble of the final rule.
Full Geography Configuration - For the "Full Geography" configuration, all states that
were linked to any receptor in the 2023 or 2026 base cases (i.e., only states included in
the rule), but no other states39, were simultaneously adjusted to the emission levels for
each of the scenarios in Table C-l. This approach presents an alternative way of thinking
about the effect of the rule, in a more holistic way, but this approach introduces a "who
goes first" problem and the potential for capturing incidental overcontrol resulting from
emissions reductions in states not linked to a particular receptor above 1% of the
NAAQS. The results of the "full geography" configuration are shown in Appendix D.
2. Details on the construction of the ozone AO AT for this rule
(a) Overview of the ozone AO AT
This section describes the step-by-step development process for the ozone AQAT. All 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
photochemical model, the non-EGU assessment and/or the IPM EGU modeling combined with
the EGU engineering assessment) with 2026 CAMx modeled ozone contributions in order to
estimate ozone concentrations at monitoring sites associated with different levels of emissions
control for each of the scenarios described in Table C-l.
In applying AQAT to analyze air quality improvements at a given receptor for the Step 3
configuration analyzing each of the cost-threshold scenarios, emissions were reduced in only
those upwind states that were "linked" to that receptor in step 2 of the Good Neighbor
Framework (i.e., those states that contributed an air quality impact at or above 1 percent of the
NAAQS). Emissions were also reduced in the state that contained that receptor (regardless of the
39 For the purposes of the AQAT "Full Geography" estimates, we included California as being included in the rule
and making any available reductions. See the preamble section I for how this state is treated in the rule.
46
-------
level of that state's contribution or whether that state was linked to another state) at a level of
control stringency consistent with the stringency level applied in upwind states.40
Specifically, the key estimates from the ozone AQAT for each receptor are:
The ozone contribution as a function of emissions at each cost threshold scenario, for
each upwind state contributing above the 1 percent air quality threshold and the state
containing the receptor.
The ozone contribution under engineering analysis base case NOx emissions in the
various years, for each upwind state that is not above the 1 percent air quality
threshold for that receptor.
The non-anthropogenic (i.e., background, boundary, biogenic, and wildfire) ozone
concentrations. These are assumed to vary linearly in direct proportion to the total
anthropogenic contribution change relative to the total change in these components
between the 2026 final base case source apportionment modeling and the 2023 final
base case source apportionment modeling scenario.41
The results of the ozone AQAT Step 3 analysis for each emissions scenario can be found in
section C.3 of this document. The results for the "full geography" configuration can be found in
Appendix D.
(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. As described in the Air Quality Modeling TSD, EPA
performed contribution modeling for 2023 and 2026 using base case emissions to quantify the
amount of ozone formed from several source "tags." In the modeling for 2023 and 2026, EPA
tagged anthropogenic emissions from each state individually as well as total anthropogenic
emissions from Canada and Mexico combined, offshore drilling platforms and shipping, wild
and prescribed fires, lightning, biogenic sources, and initial/boundary conditions (which
represent the net contribution from all sources outside the modeling domain). In addition, at
proposal, EPA also performed state-specific contribution modeling for a 2026 scenario in which
EGU and Non-EGU NOx emissions were reduced by 30 percent. Note that the 2026 base case
emissions for air quality modeling at proposal used IPM emission estimates. In the ozone AQAT,
any emission differences between the 2026 air quality modeling base case and a scenario would
result in changes in air quality contributions and ozone concentrations at the downwind monitors.
The emission inventories used in the air quality modeling for the 2023 and 2026 base case are
40In this Step 3 configuration, EPA assumes that the downwind state will implement (if it has not already) an
emissions control strategy for their sources that is of the same stringency as each upwind control strategy examined
in the scenario. Under this approach, EPA accounts for what may be considered the downwind state's "fair share."
41 In previous versions of AQAT, EPA has held these components constant at the base case levels. The emissions are
held constant in the photochemical modeling for the various cases, so changes in the resulting contributions are a
result of changing chemistry. In the photochemical modeling, we observe that these AQ contributions change in
response to changing chemistry in response to changes in anthropogenic emissions and contributions from the states.
In other words, the anthropogenic emission changes result in slightly different chemistry that affects the
nonanthropogenic contributions. The impact of the change is usually a small fraction of a ppb.
47
-------
discussed in the Final Rule Emissions Modeling TSD Preparation of Emissions Inventories for
the 2016v3 North American Emissions Modeling Platform42 while the inventories from proposal
are discussed in the TSD Preparation of Emissions Inventories for 2016v2 North American
Emissions Modeling Platform and in the Air Quality Modeling TSD used at proposal (Docket
ID: EPA-HQ-OAR-2021-0668-0099). Finally, for each of the EGU and non-EGU scenarios
examined with the AQAT, the EGU and non-EGU emissions were created from the engineering
analysis emission inventory described in section B. The ozone season NOx EGU and non-EGU
emissions for each emission scenario including the base case as modeled in AQAT are described
in section C of this TSD.
(c) Detailed outline of the process for constructing and utilizing the ozone AQAT
The ozone AQAT was created and used in a multi-step process. In brief, the ozone
AQAT was created using the contributions and emissions inventory from the 2023 and 2026 base
case air quality modeling from the final rule as well as the relationships between emissions
reductions and air quality improvements derived using the 2026 base case and 2026 30% NOx
reduction cases from proposal. This primary-calibration AQAT was used to evaluate all policy
scenarios listed in Table C-l. As a first step, EPA developed calibration factors to account for the
nonlinear response of ozone to NOx reductions. To calculate the expected change in ozone for
each emissions cost threshold scenario evaluated, EPA identified the fractional change in
anthropogenic NOx emissions relative to the 2026 base case in each state from the final rule and
then multiplied this fractional change by the state and receptor-specific primary calibration factor
as well as by the state- and receptor-specific contribution from the final rule. This resulted in a
state- and receptor-specific "calibrated change in contribution" relative to the 2026 base case
from the final rule. Each state's change in contribution value was then added to its 2026 base
case contribution and the results summed for all states for each receptor.43 Next, the receptor-
specific base case contributions from the other source-categories44 were added to the sum of each
state's contribution. Note that the contributions from these other source categories were modified
according to the ratio of the total change in anthropogenic contribution from the 2026 base. This
was accomplished by taking the ratio of the change in nonstate contribution to the change in state
contribution between the 2026 base and the 2023 scenario and multiplying it (the ratio) by the
expected change in total state contribution. This accounted for the interaction between changes in
US anthropogenic emissions and ozone, principally formed from these other categories.
Summing up all the contributions, the net result of these calculations is an estimated average
design value for each receptor that reflects the emissions changes associated with each scenario
evaluated.45
This primary-calibrated ozone AQAT was used to project the ozone concentrations for
each level of NOx control stringency as implemented through emission budgets on a state-by-
42 https://www.epa.gov/air-emissions-modeling/2016-version-3-technical-support-document
43 In some cases (where emissions are lower than modeled in the 2026 base case) the change in contribution can be
negative.
44 The other source categories include contributions from anthropogenic emission from Canada and Mexico,
offshore drilling platforms and shipping, wild and prescribed fires, biogenic emissions, lightning, and
initial/boundary conditions which represent the net contribution from all sources outside the modeling domain.
45 Details on procedures for calculating average and maximum design values can be found in the Air Quality
Modeling TSD.
48
-------
state and receptor-by-receptor basis for every monitor throughout the modeling domain. EPA
conducted these runs using both the Step 3 configuration approach and the Full Geography
configuration approach. The results using the primary calibration approach for the Step 3 Cases
are presenting in Section C.3 of this document. The results of the primary calibration approach
for the Full Geography Cases are in Appendix D.
(1) Steps to create the primary calibration factors
The process for creating the calibration factors follows the basic premise of the approach
used in the CSAPR Update and Revised CSAPR Update, but is updated to make the factors state
as well as receptor specific.
For the primary approach, for each state, EPA summed the ozone season total
anthropogenic NOx emissions across all relevant source sectors for the 2026 base case and 2026
30% EGU and non-EGU NOx reduction case from proposal. For each state, EPA calculated the
"fractional reduction ratio" as the ratio of the difference in anthropogenic emissions relative to
the total anthropogenic emissions for its 2026 base case. In other words, the difference in
emissions in the fractional reduction ratio consists of OS anthropogenic NOx emissions in the
2026 30% NOx reduction case from proposal minus the OS NOx in the 2026 base case from
proposal. This difference in tons is then divided by the 2026 base case emissions from proposal,
resulting in a "fractional reduction" for the 30% NOx reduction case. The total anthropogenic
emissions data and resulting fractional reduction ratios can be found in Table C-2 and in the
ozone AQAT worksheet titled "calib emiss f' in the "OzoneAQATfinal.xlsx" workbook.
In order to facilitate understanding the next steps of the calibration process for the
primary approach, EPA describes below a demonstrative example: the Westport monitor number
090019003 in Fairfield County, Connecticut, with a 2026 base case projected ozone average
design value of 74.6 parts per billion (ppb) and maximum design value of 74.8 ppb. The air
quality modeling contributions for this receptor for the various modeled cases are included in
Table C-2.
For each monitor, the "uncalibrated" change in contribution from each upwind state
(Table C-2 for Westport) was found by multiplying each state's 2026 base case ozone
contribution by the reduction fraction ratio (i.e., the difference in emissions as a fraction of the
2026 base case emissions). The equation for these calculations is shown in equation 1.
Uncalibrated ozone change in air quality contribution = 2026 base case contribution from
proposal x ((2026 30 NOx case anthropogenic emissions from proposal 2026 base case
anthropogenic emissions from proposal)/2026 base case anthropogenic emissions from
proposal) Eqn C-l
Thus, when the 2026 30% NOx reduction case had lower emissions than the 2026 base
case, the net result was a negative number. Then, each state's fractional change in emissions ratio
was multiplied by its 2026 base case contribution to get a state-specific change in contribution
(Table C-2). For each state, this change in concentration reflects its total "uncalibrated" change.
49
-------
Table C-2. The Primary Approach for Creating Calibration Factors Illustrated Using Air Quality Modeling
from Proposal for the Westport Monitor Number 090019003 in Fairfield County, Connecticut.
A
B
C
D
E
F
G
H
State
Modeled
Modeled
2026
Westport
Westport
Uncalibrated
Modeled
Calibration
2026 Base
2026 30%
Fractional
2026 Base
2026 30%
AQAT Ozone
Ozone
Factor for
Case NOx
EGU/non-
Reduction in
Case Ozone
NOx Cut
Change
Change
EGUs and
Emissions
EGU
Emissions
Contributions
Ozone
(Column C*
(Column E -
non-EGUs
Reduction
Ratio
Contributions
Column D)
Column D)
(Column
NOx
((Column B-
G/Column
Emissions
Column A)/
Column A)
F)
Alabama
61,759
52,853
-0.14
0.105
0.095
-0.015
-0.010
0.67
Arizona
33,463
32,313
-0.03
0.012
0.012
0.000
0.000
0.60
Arkansas
39,488
35,333
-0.11
0.137
0.127
-0.014
-0.010
0.69
California
133,629
127,270
-0.05
0.032
0.031
-0.002
-0.001
0.67
Colorado
49,825
45,877
-0.08
0.051
0.048
-0.004
-0.003
0.85
Connecticut
10,887
10,256
-0.06
2.762
2.777
-0.160
0.015
-0.09
Delaware
6,447
6,135
-0.05
0.421
0.408
-0.020
-0.012
0.61
District of
-0.002
-0.001
0.53
Columbia
1,302
1,245
-0.04
0.037
0.036
Florida
92,166
84,786
-0.08
0.063
0.058
-0.005
-0.004
0.88
Georgia
60,266
55,302
-0.08
0.140
0.133
-0.012
-0.007
0.61
Idaho
17,321
16,296
-0.06
0.023
0.023
-0.001
-0.001
0.58
Illinois
91,069
83,536
-0.08
0.634
0.611
-0.052
-0.023
0.44
Indiana
68,291
59,091
-0.13
0.930
0.875
-0.125
-0.054
0.43
Iowa
41,049
36,033
-0.12
0.119
0.110
-0.014
-0.009
0.59
Kansas
59,107
53,798
-0.09
0.091
0.087
-0.008
-0.005
0.56
Kentucky
50,887
43,739
-0.14
0.847
0.762
-0.119
-0.085
0.72
Louisiana
100,361
86,348
-0.14
0.250
0.226
-0.035
-0.024
0.70
Maine
12,918
11,982
-0.07
0.006
0.006
0.000
0.000
0.57
Maryland
23,671
22,513
-0.05
1.089
1.064
-0.053
-0.024
0.45
Massachusetts
26,353
25,321
-0.04
0.064
0.063
-0.003
-0.002
0.69
Michigan
75,940
66,736
-0.12
1.339
1.254
-0.162
-0.085
0.52
Minnesota
55,972
49,439
-0.12
0.158
0.144
-0.018
-0.014
0.76
Mississippi
33,156
29,336
-0.12
0.096
0.088
-0.011
-0.007
0.65
Missouri
67,664
60,958
-0.10
0.288
0.268
-0.029
-0.020
0.71
Montana
25,642
23,333
-0.09
0.064
0.059
-0.006
-0.005
0.91
Nebraska
38,322
34,126
-0.11
0.057
0.054
-0.006
-0.004
0.59
Nevada
16,178
14,980
-0.07
0.010
0.010
-0.001
-0.001
0.72
New
0.000
0.000
0.14
Hampshire
6,719
6,596
-0.02
0.016
0.016
New Jersey
31,805
30,607
-0.04
8.023
8.079
-0.302
0.057
-0.19
New Mexico
62,210
58,527
-0.06
0.045
0.043
-0.003
-0.002
0.75
New York
65,642
61,970
-0.06
13.288
13.198
-0.743
-0.090
0.12
North
-0.043
-0.029
0.68
Carolina
51,986
46,303
-0.11
0.389
0.360
North Dakota
55,294
52,126
-0.06
0.077
0.074
-0.004
-0.003
0.75
Ohio
78,681
70,003
-0.11
1.947
1.814
-0.215
-0.133
0.62
Oklahoma
83,411
76,046
-0.09
0.139
0.131
-0.012
-0.008
0.66
Oregon
29,345
27,680
-0.06
0.024
0.023
-0.001
-0.001
0.69
Pennsylvania
103,565
95,081
-0.08
6.581
6.211
-0.539
-0.370
0.69
Rhode Island
4,187
4,011
-0.04
0.008
0.008
0.000
0.000
0.59
South
-0.016
-0.010
0.62
Carolina
38,939
34,839
-0.11
0.154
0.144
South Dakota
11,084
10,494
-0.05
0.036
0.035
-0.002
-0.001
0.40
Tennessee
47,475
43,303
-0.09
0.254
0.243
-0.022
-0.011
0.51
Texas
280,717
261,613
-0.07
0.490
0.469
-0.033
-0.021
0.62
Utah
29,762
26,807
-0.10
0.026
0.025
-0.003
-0.002
0.69
Vermont
3,378
3,363
0.00
0.011
0.011
0.000
0.000
-2.51
Virginia
46,496
43,302
-0.07
1.135
1.097
-0.078
-0.038
0.49
Washington
47,754
45,338
-0.05
0.043
0.042
-0.002
-0.001
0.45
West Virginia
39,500
35,285
-0.11
1.236
1.139
-0.132
-0.098
0.74
Wisconsin
41,032
37,456
-0.09
0.176
0.167
-0.015
-0.008
0.54
Wyoming
32,928
28,322
-0.14
0.061
0.054
-0.009
-0.007
0.79
Tribal Data
4,052
3,352
-0.17
0.002
0.002
0.000
0.000
0.99
50
-------
Table C-3. The Total Anthropogenic NOx Emissions (tons) used in the CAMx Photochemical Modeling for
the Final 2026 and 2023 Base Cases.
State
Modeled 2026
Modeled 2023
Base Case NOx
Base Case NOx
Emissions
Emissions
(final)
(final)
Alabama
56,096
62,236
Arizona
35,514
45,689
Arkansas
44,639
48,316
California
137,932
143,158
Colorado
49,742
53,682
Connecticut
10,201
11,320
Delaware
6,492
7,001
District of
Columbia
1,057
1,158
Florida
88,786
99,464
Georgia
61,626
74,320
Idaho
17,024
19,977
Illinois
84,913
93,730
Indiana
70,963
80,266
Iowa
46,523
51,561
Kansas
56,844
62,841
Kentucky
49,829
54,497
Louisiana
98,585
105,825
Maine
13,617
15,739
Maryland
23,023
25,546
Massachusetts
28,194
30,375
Michigan
69,697
74,659
Minnesota
55,848
63,850
Mississippi
32,407
37,544
Missouri
68,407
78,783
Montana
25,336
28,391
Nebraska
42,355
47,930
Nevada
18,043
23,066
New
Hampshire
6,830
7,514
New Jersey
31,368
34,030
New Mexico
70,923
73,072
New York
64,616
69,157
North
Carolina
55,518
65,920
North Dakota
69,173
73,341
Ohio
75,421
81,856
Oklahoma
77,225
85,520
Oregon
28,271
31,783
Pennsylvania
87,453
100,143
Rhode Island
4,172
4,601
South
Carolina
40,161
44,381
South Dakota
12,372
14,390
Tennessee
46,637
55,463
Texas
299,134
332,363
Utah
31,387
40,748
Vermont
3,447
3,960
Virginia
45,636
51,041
Washington
46,143
52,545
West Virginia
45,466
47,380
Wisconsin
41,877
49,713
Wyoming
35,517
41,055
Tribal Data
5,522
5,976
51
-------
Next, the state- specific ozone responses under the 2026 30% NOx reduction case from
the CAMx modeling from proposal was used to derive the primary calibration factors and
calibrate the ozone AQAT. The calibration factors were calculated by taking the change in
modeled ozone from CAMx and dividing by the change in ozone predicted by the uncalibrated
AQAT (Eqn. C-l). This resulted in state- specific calibration factors (see Table C-2 for an
example calculation of the primary calibration factors using the Westport CT monitor 090019003
in Fairfield County). This procedure was separately repeated for each monitor, with the result
being state- and monitor-specific calibration factors.
The use of these state- and monitor-specific calibration factors provided EPA with the
ability to align the ozone response predicted by the "uncalibrated" ozone AQAT to the ozone
response predicted by CAMx. In other words, this provides EPA with a method to systematically
interpret the existing CAMx air quality modeling data. Following the creation of the "primary"
calibration factors, EPA created a set of "alternative" set of calibration factors using the source
apportionment modeling of the 2023 and 2026 base cases from the final rule following the
procedure outlined here (section C.4 for results comparing the primary and alternative
approaches for select scenarios).
The ozone AQAT calibration factors for all monitors can be found in the
"Ozone AQAT Final.xlsx" excel workbook in columns I through BF, on worksheets
"primary calibration" and "alternative calibration" for the primary and "alternative" calibration
scenarios, respectively. The calibration factor, when multiplied by an "uncalibrated" air quality
change results in a "calibrated" change in air quality contribution. The "uncalibrated" air quality
change is calculated by taking the fractional change in emissions ratio for a scenario and
multiplying that by the state-specific air quality contributions.
The final step in the creation of a calibrated AQAT is to develop an adjustment approach
for the non-anthropogenic air quality contributions that are not being directly varied within the
AQAT - and that generally have constant emissions for all cases. While the emissions are
constant, the air quality contributions from these sources do vary slightly as the chemistry
throughout the domain changes in response to anthropogenic emissions changes from the states.
The adjustment approach affects the air quality contributions from Canada and Mexico, offshore
drilling platforms and shipping, wild and prescribed fires, lightning, biogenic sources, and
initial/boundary conditions (i.e., "all other" contributions). In previous versions of AQAT, these
contributions were held fixed at the base case values. For this final rule, because we have full
source apportionment estimates for both cases used in the calibration process, we are able to
adjust these contributions by relating their change to a change in the anthropogenic contributions.
We do this based on multiplying the change in the total anthropogenic contributions from the
states between the scenario and the base case by the ratio of the change from the sum of the "all
other" contributions divided by the change in the total anthropogenic contribution from the base
and calibration cases. For example, at the Westport CT receptor, the difference between the 2026
base case and the 2023 base case was -0.113 ppb for "all other" contributions and 2.113 ppb for
the anthropogenic contributions, resulting in a ratio of -0.053. In other words, a 1 ppb increase in
the anthropogenic contribution could be expected to result in a 0.053 ppb decrease in the
contribution from the "all other" emissions (even though these emissions have not changed).
As an example application of this adjustment, in the 2026 engineering base case using the
primary version of AQAT, the total anthropogenic contribution was 42.5145 (compared to a
2026 modeled base case value of 42.22 ppb). The difference between the engineering base case
52
-------
of the total anthropogenic emissions and the modeled base case values were then multiplied by
the resultant ratio above to get a calibrated change in the "all other" contributions of -0.0157 ppb.
Thus, the "all other" contribution changed from the 2026 modeled base case value of 29.08054
ppb to an engineering base case value of 29.0648 ppb.
53
-------
(2) Create a calibrated version of the ozone AO AT for emission control stringency level analysis
for the rule
EPA examined the changes in the 2026 air quality contributions due to changes in EGU
and non-EGU emissions for various scenarios relative to the final 2026 base case emissions
(while using the calibration factors). The AQAT, as calibrated above, was used for each
emissions cost threshold level evaluated for EGUs and non-EGUs (see Table C-l for the list and
description of the scenarios). For 2023 simulations, EPA calculated a calibrated change in
contribution that was then applied to the 2023 contributions. In 2023, the calibrated change in
contribution was found by taking the change in emissions from the 2023 final base case to the
2023 cost threshold level and dividing this emissions change by the 2026 base case emission
level. The emissions for 2023 and 2026 photochemical modeling base cases can be found in
Table C-3. This fractional emission change was then multiplied by the 2026 contribution and the
calibration factor.
For each scenario in AQAT, we assembled a complete NOx emission inventory
representing all anthropogenic sources for each state for each year. This inventory is composed
of the EGU inventory and the remaining portion of the inventory. As described in sections A and
B of this TSD regarding an important component of the total EGU emission inventory, EPA
identified various cost threshold levels of emissions (i.e., scenarios) based on potential changes
in emissions rates and adjusted historical data. The remaining portion of the total anthropogenic
NOx emission inventory (excluding the EGU emissions) are presented in Table C-4 for each
state and year.
The total EGU point emissions inventory is composed of emissions from units that report
emissions to EPA's Clean Air Markets Division (CAMD) under 40 CFR Part 75 (most emissions
from these sources are measured by CEMS) and units that are typically included in EPA's power
sector modeling using the Integrated Planning Model (IPM) but that do not report to CAMD and
typically lack CEMS (i.e., the non-CEM units). Within the air quality modeling platform,
different approaches are taken to create the total EGU point inventory depending on whether an
emissions inventory for EGUs is created using IPM or engineering analysis, with each have a
different non-CEM emission component. The non-CEM component for the 2016 base case air
quality model platform using EGU emissions based on CEMS is comparable to that needed for
engineering analysis. The non-CEM component for the 2023 and 2026 air quality modeling cases
are based on IPM EGU emissions. All three non-CEM emission values are shown in Table C-4.
In AQAT, for each engineering analysis based scenario, the 2016-based non-CEM component
was added to the engineering analysis EGU emissions.
For each scenario in AQAT, we assembled a complete emission inventory representing
all anthropogenic sources for each state. In other words, we combine the year-specific
anthropogenic emissions from Table C-4 (where the EGU point emissions have been removed),
with a replacement EGU point inventory comprised of the relevant EGU non-CEM component
from Table C-4, and one of the engineering analysis EGU estimates from Section B of this TSD.
The complete anthropogenic emission inventory totals for each state, including the non-
CEM components, are compared to the final 2026 base case that was included in the air quality
modeling. For each state, for each emissions scenario, EPA calculated the ratio of the emission
differences from the scenario and the final 2026 air quality modeling base case to the total NOx
emissions for the final 2026 air quality modeling base case (see Tables C-5 and C-6). Scenarios
that are not viable, for technical or policy reasons, have been grayed out in these tables.
54
-------
In Tables C-4 and C-6, respectively, we examined the emission reduction potential for
the non-EGUs, and then included these emission reductions along with the emission reductions
from EGUs where new post-combustion controls have been applied and where all EGU
emissions have been applied except new post-combustion controls. We, then, calculated the ratio
of the emission difference relative to the 2026 air quality modeling base case.46
Once the reduction ratios were calculated, they could be applied to a particular state's air
quality contribution at a particular monitor along with the calibration factor to get a calibrated
change in concentration. These changes were then applied to the original air quality contribution
to get an adjusted contribution.
As described above, two AQAT estimates were created for each of the scenarios based on
the "Step 3" configuration and the "Full Geography" configuration. These apply different
patterns of emission reductions to the states at various monitors. For each scenario analyzed
using the Step 3 configuration, on a receptor-by-receptor basis, the emissions change for each
upwind state is associated with one of two emission levels (either the engineering base case
emission level for that year or the particular cost threshold level) depending on whether the
upwind state is contributing at or above 1% of the NAAQS in the air quality modeling base case
to that receptor or if the receptor is located within the state.47 In these scenario assessments using
the Step 3 configuration, each monitor is treated completely independently, and the
modifications are applied regardless of whether the state is included in the rule and regardless of
whether the monitor is considered a receptor for the rule. In other words, states that are
contributing above the air quality threshold (i.e., greater than or equal to 1 percent of the
NAAQS) to that specific monitor, as well as the state containing the monitor (regardless of
whether that state is included in the rule or not (e.g., for Colorado and Connecticut), make NOx
emission reductions that are available at the particular cost threshold level for that year. The
emissions for all other states are adjusted to the engineering base case level for that year
regardless of whether they are linked to another receptor. Consequently, for the Step 3
configuration for a single scenario (where there are 730 monitors), there are potentially 730
individual patterns of linked and unlinked states, and, thus, 730 potential AQAT simulations.
When we assess the maximum air quality contributions to remaining receptors, we limit the
analysis to those receptors originally identified using the photochemical air quality modeling in
the base case.
For the scenarios assessed using the "Full Geography" configuration, all states that were
linked to any receptor in the 2023 or 2026 base cases (i.e., only states included in the rule) were
simultaneously adjusted to one of the cost threshold levels shown in Table C-l, regardless of
whether (or not) the state was "contributing at or above the 1% of the NAAQS in the base case
air quality modeling to a particular receptor. In other words, all states that were included in the
rule were adjusted for each receptor, while all other states were adjusted to the base case. In
46 With the EPA 2026 AQAT analysis, EPA looked at full implementation of SCR retrofit potential in 2026 when
examining that mitigation strategy (recognizing that program implementation and compliance allows some
flexibility to realize a portion of these reductions in 2027). This ensures an appropriate analysis of the effects of the
rule with respect to the determination of "significant contribution" and overcontrol analysis, See Section V.D of the
preamble for further discussion. It ensures all Step 3 related reductions are tested for overcontrol, regardless of any
timing flexibility offered during implementation regarding the 2026/2027 phase in or the backstop rate extension up
to 2030.
47 For purposes of AQAT analysis, tribal EGU emissions are adjusted based on linkages using either the tribal
contribution or the contribution from Utah. In this way, for the Colorado receptors to which Utah is linked, we make
sure we account for emission reductions from tribal EGUs located within the borders of Utah.
55
-------
these scenarios using the "full geography" configuration, the emissions of the state containing
the monitor were adjusted only if it was linked to a monitor in another state. So, for example,
Connecticut was adjusted to engineering analysis base case levels since the state is not "linked"
to a receptor in another state and is not included in the final rule. The scenarios assessed using
the "full geography" configuration examine the air quality results when emission reductions have
been applied to the final rule geography. EPA views this analysis as not appropriate for Step 3
because it introduces the problem of allowing linked states to potentially free ride on reductions
from non-linked states (i.e., EPA views this situation as having the potential to display potential
overcontrol that is only incidental). It therefore introduces an issue where the order of individual
states making emissions reductions could affect the results (i.e., a "who goes first" problem).
Nonetheless, this analysis can be used to show thateven if this approach were acceptable or for
some reason legally requiredemission reductions made for states that are not specifically
linked at or above 1% of the NAAQS to a monitor are not anticipated to affect the air quality at
that monitor to a degree that would change any results in the Step 3 analysis.
As described above, for each monitor, the predicted change in contribution of ozone from
each state is calculated by multiplying the state-specific 2026 base case ozone contributions from
the air quality modeling by the state- and receptor-specific calibration factor as well as by the
ratio of the change in emissions (Tables C-5 or C-6 for either the emissions cost threshold level
or the engineering base case emission level depending on whether the state is linked in 2023 or
2026).48 This state- and receptor-specific calibrated change in ozone is then added to the ozone
contribution from either the 2023 or 2026 base case air quality modeling, depending on whether
the scenario is for 2023 or 2026. The result is the state- and receptor- specific "calibrated" total
ozone contribution taking into account the emissions remaining at a particular emission reduction
cost threshold level.
For each monitor, these state-level "calibrated" contributions are then summed to
estimate total ozone contribution from all states to a particular receptor. "Other" ozone
contributions, as described above in section C.2.(b), are added to the state contributions to
account for other sources of ozone affecting the monitor. The change in concentration from the
"other" nonanthropogenic ozone categories are found by multiplying the change in the total
anthropogenic concentration, between the scenario and the base case, by the "nonState"
calibration factors (calculated as the ratio of the change from these "all other" contributions
divided by the change in the total anthropogenic contribution from the 2026 base case to the
2023 case).49 This change in the "other" contribution is then added to the base case value to get
the total "other" contribution for the scenario. 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 2026 base case maximum and average design values.
Generally, as the emissions cost threshold stringency increased, the estimated average
and maximum design values at each receptor decreased. In the assessment tool, the estimated
average design value was used to further estimate whether the location will be out of attainment.
Meanwhile, the estimated maximum design value was used to further estimate whether the
48 The change in concentration can be positive or negative, depending on whether the state's total anthropogenic
ozone season NOx emissions for the scenario are larger or smaller than the air quality modeling base case emission
level for that year.
49 See column BV in "2023_Scenario_primary" or "2026_Scenario_primary" in the Ozone AQAT Final Rule Excel
file
56
-------
location will have problems maintaining compliance with the NAAQS. An area was noted as
having a nonattainment or maintenance issue if either estimated air quality level was greater than
or equal to 71 ppb.
57
-------
Table C-4. Ozone Season Anthropogenic NOx Emissions (Tons) without the EGU Point
Inventory for Each State for 2023 and 2026, the non-CEM EGU Emissions from 2016,
State
2023 OS
2026 OS
2016 non-
2023 IPM
2026 IPM
2026 non-
NOx
NOx
CEM EGU
non-CEM
non-CEM
EGU
Emission
Emissions
Emissions
EGU
EGU
Emission
s w/out
w/out
(tons)
Emissions
Emissions
Reductions
EGUs
(tons)
EGUs
(tons)
(tons)
(tons)
(tons)
Alabama
56,301
50,689
482
409
375
-
Arizona
37,767
32,429
684
413
430
-
Arkansas
37,601
33,911
144
125
151
1,546
California
137,682
131,712
1,855
4,900
5,472
1,600
Colorado
45,691
42,286
333
2,195
2,600
-
Connecticut
10,057
9,047
1,272
1,001
898
-
Delaware
6,808
6,289
80
58
59
-
District of Columbia
1,143
1,042
0
15
16
-
Florida
86,562
77,663
5,803
5,615
5,176
-
Georgia
65,100
57,396
1,614
1,286
608
-
Idaho
19,538
16,745
509
370
112
-
Illinois
87,678
80,699
55
598
705
2,311
Indiana
64,377
58,607
611
764
865
1,976
Iowa
40,960
36,510
635
475
575
-
Kansas
56,535
51,888
103
224
334
-
Kentucky
41,631
37,915
1
314
417
2,665
Louisiana
94,803
89,607
3,885
790
1,316
7,142
Maine
13,531
12,333
1,972
1,735
1,030
-
Maryland
24,165
21,885
901
1,044
1,098
157
Massachusetts
27,843
25,766
1,949
2,097
2,079
-
Michigan
63,275
58,837
1,367
1,327
1,523
2,985
Minnesota
55,212
50,422
1,740
1,549
722
-
Mississippi
33,233
30,560
1,663
299
426
2,499
Missouri
62,434
54,563
469
124
152
2,065
Montana
24,264
21,361
933
208
58
-
Nebraska
36,217
32,360
665
574
579
-
Nevada
19,169
16,592
155
689
1,209
-
New Hampshire
7,263
6,578
327
205
205
-
New Jersey
32,305
29,546
1,064
1,212
1,206
242
New Mexico
72,061
70,090
98
56
91
-
New York
63,581
59,425
1,989
3,145
3,129
958
North Carolina
49,369
43,878
739
1,439
1,206
-
North Dakota
58,778
55,705
156
0
16
-
Ohio
69,906
63,465
722
1,350
1,472
3,105
Oklahoma
76,860
70,318
1
185
357
4,388
Oregon
31,284
27,178
704
495
1,086
-
Pennsylvania
89,024
82,296
2,005
2,843
3,192
2,184
Rhode Island
4,327
3,908
35
252
243
-
South Carolina
36,183
32,417
643
758
555
-
South Dakota
13,820
11,803
14
0
3
-
Tennessee
49,954
44,362
6
233
287
-
Texas
290,799
271,630
1,996
2,118
2,078
4,691
Utah
25,768
22,990
561
134
410
252
Vermont
3,911
3,436
41
49
11
-
Virginia
46,978
41,933
2,995
2,707
2,551
2,200
Washington
51,605
45,280
1,536
940
862
-
West Virginia
33,465
32,071
1
6
6
1,649
Wisconsin
43,533
39,136
61
523
596
-
Wyoming
31,006
29,366
11
2
4
-
Tribal Data
3,096
2,979
57
44
0
-
58
-------
Table C-5. 2023 Fractional Di
'ference in Emissions for each Scenario.50
State
Engineering
Baseline
Optimize
SCR
Optimize S£R^
+ soApe^
Optimize
SNCR+ SCR
Optimize
SNCR-KSCR +
so^-ee
New SCR/SNCIJ^-*f
Optimize R+
SCR>SfJACC("Full
_J8t6p3 - EGU only")
Alabama
0.02
0.02
0.02
0.02
0.02
0.00
Arizona
0.01
0.01
0.01
0.01
0.00
-0.11
Arkansas
-0.04
-0.04
-0.04
-0.04
-0.04
-0.15
California
-0.01
-0.02
-0.02
-0.02
-0.02
-0.02
Colorado
-0.02
-0.03
-0.03
-0.03
-0.03
-0.08
Connecticut
0.04
0.04
0.04
0.04
0.04
0.04
Delaware
0.05
0.04
0.04
0.04
0.04
0.04
District of Columbia
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Florida
0.07
0.04
0.04
0.04
0.04
0.01
Georgia
-0.04
-0.04
-0.04
-0.04
-0.04
-0.04
Idaho
0.02
0.02
0.02
0.02
0.02
0.02
Illinois
0.02
0.02
0.02
0.02
0.02
0.01
Indiana
-0.03
-0.04
-0.04
-0.04
-0.04
-0.07
Iowa
0.00
0.00
0.00
0.00
0.00
-0.13
Kansas
0.00
-0.01
-0.01
-0.01
-0.01
-0.05
Kentucky
0.02
0.01
0.00
0.01
0.00
-0.10
Louisiana
0.03
0.02
0.02
0.02
0.02
-0.03
Maine
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Maryland
0.03
0.03
0.03
0.03
0.03
0.03
Massachusetts
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Michigan
0.01
0.01
0.01
0.01
0.01
-0.04
Minnesota
-0.02
-0.02
-0.02
-0.02
-0.02
-0.05
Mississippi
0.11
0.11
0.08
0.11
0.08
-0.02
Missouri
0.06
-0.05
-0.05
-0.05
-0.05
-0.11
Montana
0.00
0.00
0.00
0.00
0.00
-0.08
Nebraska
-0.05
-0.05
-0.06
-0.05
-0.06
-0.19
Nevada
-0.08
-0.08
-0.08
-0.08
-0.08
-0.14
New Hampshire
0.06
0.05
0.05
0.05
0.05
0.05
New Jersey
0.01
0.00
0.00
0.00
0.00
0.00
New Mexico
0.00
0.00
0.00
0.00
0.00
0.00
New York
0.01
0.01
0.01
0.01
0.01
0.00
North Carolina
-0.06
-0.12
-0.12
-0.12
-0.12
-0.17
North Dakota
-0.03
-0.03
-0.03
-0.04
-0.04
-0.17
Ohio
-0.01
-0.03
-0.03
-0.03
-0.03
-0.03
Oklahoma
0.02
0.02
0.01
0.02
0.01
-0.06
Oregon
0.02
0.02
0.02
0.02
0.02
0.02
Pennsylvania
-0.01
-0.01
-0.01
-0.01
-0.01
-0.02
Rhode Island
0.01
-0.02
-0.02
-0.02
-0.02
-0.02
South Carolina
-0.08
-0.10
-0.10
-0.10
-0.10
-0.10
South Dakota
0.00
0.00
0.00
0.00
0.00
0.00
Tennessee
-0.03
-0.03
-0.03
-0.03
-0.03
-0.03
Texas
0.01
0.00
0.00
0.00
0.00
-0.05
Utah
-0.02
-0.02
-0.02
-0.02
-0.02
-0.36
Vermont
0.01
0.01
0.01
0.01
0.01
0.01
Virginia
0.05
0.05
0.04
0.05
0.04
0.04
Washington
0.06
0.05
0.05
0.05
0.05
0.03
West Virginia
0.02
0.00
-0.01
0.00
-0.01
-0.06
Wisconsin
0.00
0.00
0.00
0.00
0.00
-0.01
Wyoming
0.05
0.04
0.03
0.04
0.03
-0.15
Tribal Data
0.04
0.04
0.04
0.04
0.04
-0.22
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
511 The fractional changes in emissions are essentially "percent changes" in emissions. These fractions are changes
relative to the 2026 air quality modeling base emission inventory for each state. Negative numbers indicate emission
decreases, while positive numbers indicate emission increases.
59
-------
Table C-6. 2026 Fractional Di
Terence in
Emissions for each Scenario.51
State
Engineering
Baseline
Optimize
SCR
Optimize
SCR + SOA
CC
Optimize
SNCR+ SCR
Optimize
SNCR+ SCR
+ SOA CC
New
SCR/SNCR +
Optimize
SNCR+ SCR
+ SOA CC
("Full Step 3
- EGU only")
non-EGU
+New
SCR/SNCR +
Optimize
SNCR+ SCR
+ SOA CC
("Full Step
3")
Alabama
0.03
0.03
0.03
0.03
0.03
0.02
0.02
Arizona
0.08
0.08
0.07
0.07
0.07
0.02
0.02
Arkansas
-0.04
-0.04
-0.04
-0.04
-0.04
-0.15
-0.18
California
-0.02
-0.02
-0.02
-0.02
-0.02
-0.02
-0.03
Colorado
-0.05
-0.05
-0.05
-0.05
-0.05
-0.08
-0.08
Connecticut
0.05
0.05
0.05
0.05
0.05
0.05
0.05
Delaware
0.05
0.04
0.04
0.04
0.04
0.04
0.04
District of Columbia
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Florida
0.07
0.04
0.04
0.04
0.04
0.02
0.02
Georgia
0.04
0.04
0.04
0.04
0.04
0.04
0.04
Idaho
0.03
0.03
0.03
0.03
0.03
0.03
0.03
Illinois
0.03
0.03
0.03
0.03
0.03
0.01
-0.01
Indiana
-0.03
-0.04
-0.04
-0.04
-0.04
-0.05
-0.08
Iowa
0.01
0.01
0.01
0.01
0.01
-0.12
-0.12
Kansas
0.01
0.00
0.00
0.00
0.00
-0.03
-0.03
Kentucky
0.03
0.02
0.01
0.02
0.01
-0.08
-0.13
Louisiana
0.05
0.04
0.04
0.04
0.04
-0.01
-0.09
Maine
0.06
0.06
0.06
0.06
0.06
0.06
0.06
Maryland
0.03
0.03
0.03
0.03
0.03
0.03
0.02
Massachusetts
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Michigan
-0.02
-0.02
-0.02
-0.02
-0.02
-0.05
-0.10
Minnesota
0.01
0.01
0.01
0.01
0.01
-0.02
-0.02
Mississippi
0.18
0.18
0.15
0.18
0.15
0.06
-0.02
Missouri
0.08
-0.03
-0.03
-0.03
-0.03
-0.09
-0.12
Montana
0.00
0.00
0.00
0.00
0.00
-0.07
-0.07
Nebraska
-0.01
-0.01
-0.02
-0.01
-0.02
-0.15
-0.15
Nevada
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
New Hampshire
0.06
0.05
0.05
0.05
0.05
0.05
0.05
New Jersey
0.01
0.00
0.00
0.00
0.00
0.00
-0.01
New Mexico
0.00
0.00
0.00
0.00
0.00
0.00
0.00
New York
0.01
0.01
0.01
0.01
0.01
0.00
-0.01
North Carolina
0.01
-0.04
-0.04
-0.04
-0.04
-0.08
-0.08
North Dakota
-0.02
-0.02
-0.02
-0.03
-0.03
-0.15
-0.15
Ohio
-0.03
-0.04
-0.04
-0.04
-0.04
-0.04
-0.08
Oklahoma
0.04
0.04
0.03
0.04
0.03
-0.03
-0.09
Oregon
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Pennsylvania
0.06
0.06
0.06
0.05
0.05
0.05
0.02
Rhode Island
0.01
-0.02
-0.02
-0.02
-0.02
-0.02
-0.02
South Carolina
-0.07
-0.09
-0.09
-0.09
-0.09
-0.09
-0.09
South Dakota
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Tennessee
0.04
0.04
0.04
0.04
0.04
0.04
0.04
Texas
0.05
0.04
0.04
0.04
0.04
-0.01
-0.02
Utah
0.01
0.01
0.01
0.01
0.01
-0.18
-0.19
Vermont
0.02
0.02
0.02
0.02
0.02
0.02
0.02
Virginia
0.05
0.05
0.05
0.05
0.04
0.04
-0.01
Washington
0.03
0.02
0.02
0.02
0.02
0.02
0.02
West Virginia
0.00
-0.01
-0.03
-0.02
-0.03
-0.08
-0.12
Wisconsin
0.04
0.04
0.04
0.04
0.04
0.04
0.04
Wyoming
0.09
0.08
0.07
0.08
0.07
-0.06
-0.06
Tribal Data
0.10
0.10
0.10
0.10
0.10
-0.16
-0.16
51 The fractional changes in emissions are essentially "percent changes" in emissions. These fractions are changes
relative to the 2026 air quality modeling base emission inventory for each state. Negative numbers indicate emission
decreases, while positive numbers indicate emission increases.
60
-------
3. Description of the analytic results using the primary approach for the Step 3 AO AT
configuration.
For each year, 2023 and 2026, EPA used the ozone AQAT to estimate improvements in
downwind air quality at base case levels and at each of the cost threshold scenarios. For each
scenario, EPA examined the average and maximum design values for each of the receptors. EPA
evaluated the degree of change in ozone concentration and assessed whether it decreased the
average or maximum design values to below 71 ppb (at which point their nonattainment and
maintenance issues, respectively, would be considered resolved). In each scenario, EPA also
examined each state's air quality contributions, assessing whether a state maintained at least one
linkage (i.e., greater than or equal to 1% (0.70 ppb) to a receptor located in a downwind state that
was estimated to remain in nonattainment and/or maintenance. EPA examined incrementally the
engineering base case, and all of the mitigation steps described in Section V of the preamble and
calculated in the engineering analysis (with the exception of the "half SCR" scenario) (see
section B and Table C-l of this TSD for details and a list, respectively). EPA also assessed
changes in air quality for the non-EGU mitigation potential for 2026.
The key findings of this analysis are 1) no states have their contribution to a receptor
identified in the base case CAMx air quality modeling drop below 1% at any mitigation level
assessed for as long as that receptor remained in nonattainment or maintenance, and 2) all
covered states remain linked to a downwind problematic receptor up through the penultimate
mitigation step. These findings affirm EPA's identification of the final rule control stringency
and also verify that the final stringency level does not constitute overcontrol. These findings held
through EPA's alternative assessments as well (i.e., using the Alternative AQAT Calibration
factor and the Full Geography Configuration). The preamble explains how EPA considered the
results of the air quality analyses described in this TSD to determine the appropriate emission
levels for eliminating significant contribution to nonattainment and interference with
maintenance. Additional details on receptor impacts are described in the remainder of this
section below.
There are 31 receptors outside California in 2023 and 17 receptors in 2026 that are
projected to be in nonattainment or maintenance status according to the base case CAMx air
quality modeling results (see the Air Quality Modeling TSD for details). In other words, we did
not include monitors whose average or maximum concentrations increased to 71 ppb or higher
when we assessed any of the emissions scenarios (e.g., the engineering analysis base case
scenario).
For each year, using the Step 3 configuration of AQAT with the primary calibration, the
average and maximum design values (in ppb) were estimated. Air quality values for each
identified receptor and cost threshold level can be found in Tables C-l through C-10. The values
have been rounded to hundredths of a ppb. Scenarios that have been deemed nonviable are
grayed out in these tables.
In 2023, we observe that all monitors consistently have their average and/or maximum
design values at or above 71 ppb for all viable scenarios (Tables C-l and C-8). We observe that
there is air quality improvement at increasing cost threshold levels. In 2023 (but also for 2026)
61
-------
we observe that receptors 350151005 and 350250008 in Eddy County and Lea County New
Mexico, respectively, do not have calibration factors based on the "primary" approach.52
In 2026, of the 17 receptors, two receptors have their average design values drop below
71 ppb when going from the engineering analysis base case to a scenario reflecting full
implementation of identified Step 3 EGU mitigation measures. The average design values for
receptor 090013007 in Fairfield County Connecticut and receptor 481671034 in Galveston Texas
drop below 71 ppb in this scenario reflecting all EGU reductions through SCR retrofit (inclusive
of comparable reductions in Connecticut for the former, which is not linked to a receptor in
another state). The change in these two receptors from attainment to maintenance does not
completely resolve these receptors and does not resolve any upwind states' linkage to a
downwind state due to remaining linkages at these or other receptors.
The maximum design value for monitor 080690011 in Larimer County Colorado drops
below 71 ppb when EGU emission reductions associated with new SCRs are applied (inclusive
of comparable reductions in Colorado, which is not linked to a receptor in another state). The
maximum design values for receptors 480391004 in Brazoria County Texas and receptor
481671034 in Galveston Texas have their maximum design values drop below 71 ppb when the
"Full Step 3" Scenario is applied. See Table C-10 for the values.
In regards to upwind contributions, we are able to use the calibrated AQAT 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 C.2.c.(2)) in order to
determine whether any state's contribution is below the 1 percent threshold used in step 2 of the
4-Step Good Neighbor Framework to identify "linked" upwind states. For this assessment, we
compared each state's adjusted ozone concentration against the 1% air quality threshold at each
of the cost threshold levels at each remaining receptor, using the Step 3 configuration of AQAT
using the primary calibration factor. For 2023 and 2026, these results are shown in Tables C-l 1
and C-l2, respectively.
To see static air quality contributions and design value estimates for the receptors of
interest for each year and cost level scenario, see the individual worksheets (labeled in Appendix
B). For interactive worksheets, refer to the "202X_scenario_primary" worksheets after setting
the desired scenario in the "summary_DVs_202X" worksheet. In the summary_DVs worksheet,
adjust cells II and 12 to match the desired scenario of interest. The numbering for the various
scenarios is shown in Table C-13. For a cost threshold scenario estimate, cell II would be a value
of 0 through 8 (note that 6, and 7 are invalid), while cell 12 should be fixed with a value of 0.
Generally, for all linked states, in all years, across all cost level scenarios, we did not see
instances where all of the state's contributions dropped below 1% of the NAAQS assessed across
all its linkages to remaining downwind receptors. That is, for a single receptor, if a state was
linked to that receptor in the base case for that year the state almost always remained linked with
a contribution greater than or equal to 1% of the NAAQS in all scenarios. This is not a surprising
result because, for a linkage to be resolved by emission reductions of just a few percent, the
original base 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
52 In the air quality modeling for the proposal, we do not have air quality contributions for these monitors for either
(or both) the 2026 base case and the 2026 case where EGU and non-EGU emissions have been reduced by 30%.
Consequently, using the "primary" approach in AQAT, we also do not have design value or contribution
calculations for these receptors. Using the "alternative" approach, we have estimates for these receptors (see
"Ozone AQAT Final.xlsx" for the values.
62
-------
ozone season NOx emissions, and the calibration factor was 0.5, its original 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 0.72 ppb, to drop below the 0.70 ppb linkage threshold.
Note that, for Wyoming, the 2023 air quality modeling base case air quality contribution is
below 0.70 ppb. Consequently, in AQAT when the adjustment is made to states with air quality
modeling contributions above the linkage threshold, Wyoming is designated as "below the
threshold" and is assigned the engineering analysis base case value (which raises its contribution
above the linkage threshold). The result is that the contribution remains constant, appearing to be
"linked" at progressively higher cost level scenarios. A similar situation is present for Alabama
in 2026, where the contribution remains constant at the engineering analysis base case value. We
would expect that if emission reductions for these two states were made from the engineering
analysis base case level to the level used in the air quality photochemical modeling (which
incorporate projected fleet turnover from IPM in addition to the known fleet turnover used in the
engineering analysis as described in section C.2 and preamble Section IV.C.2 would result in a
lower total EGU point emission value), it would result in the air contributions dropping below
the 0.70 ppb linkage threshold in AQAT.
In this final rule, using the Step 3 configuration of AQAT using the primary calibration
factor, there are some instances where the maximum remaining contribution to a remaining
receptor that has a maximum design value at or above 71 drops below the contribution threshold.
In all cases where this happens, it is due to particular receptors dropping below the NAAQS,
rather than changes to the contributions to an individual monitor. In 2026, when emissions
reductions from new SCR and non-EGUs are applied, the highest AQAT-estimated contributions
for Arkansas, Mississippi, and Oklahoma drop below the linkage threshold of 0.70 ppb. The
change in violating monitors, described above, and the shift in contributions between receptors,
explains the large changes in contributions that occurs for these states (Table C-12). In some
cases, for individual linkages, a state drops below the contribution threshold. However, aside
from the instances noted above, in all such cases the state remained linked above the threshold to
at least one other receptor (Table C-12). In the scenario where emissions reductions from new
SCR and non-EGUs are applied, we observe that Oklahoma's contribution to Galveston Texas
drops below the linkage threshold at the same time the cumulative air quality improvements from
other states cause the receptor to have its maintenance problem resolved.
As explained in section V.D.4 of the preamble, using the Step 3 configuration of AQAT
using the primary calibration factor, EPA performed the overcontrol test at Step 3 using an
identical methodology to that used in prior CSAPR Rules. That analysis indicated that there was
no overcontrol at full implementation of the mitigation strategies in 2026 identified in this action.
Even with full implementation of EGU and non-EGU reductions, nonattainment/maintenance
receptors and corresponding linkages persisted for most of the covered states. The exceptions
were the Brazoria and Galveston receptors in Texas. These receptors were projected to be in
attainment in 2026 at full implementation, and this was the case in AQAT using the primary
calibration factors as well as in the CAMx modeling of the final rule.53 There are three states
with downwind linkages only to one or both of these receptors (Oklahoma, Mississippi, and
Arkansas). Therefore, at the Step 3 overcontrol evaluation, the EPA specifically evaluated
53 EPA notes that using the Step 3 configuration of AQAT using the alternative calibration factor, that the maximum
design value for the Galveston, Texas receptor remains above 71 ppb and Arkansas, Mississippi, and Oklahoma
have contributions that are greater than or equal to 0.70 ppb at the full implementation of EGU and non-EGU
emissions reductions. See Appendix D for details.
63
-------
whether a less stringent policy prior to full implementation of the finalized EGU and non-EGU
stringencies would have shifted these receptors into projected attainment and/or resolved the
upwind air quality contributions (i.e., Step 2 linkages) at this less-stringent control level. Neither
of these conditions occurred, and therefore the EPA concluded that there is no evidence for
overcontrol at the final rule's control level, and, in light of the otherwise applicable Step 3
determinations regarding the appropriate level of emissions control to eliminate significant
contribution, there is evidence for undercontrol if these states were subject to a lesser stringency.
Consequently, as discussed in the preamble, the EPA concludes that the uniform control
stringencies identified at Step 3 applied for all other states linked in 2026 also represent the
appropriate level of control for the states linked to the two Texas receptors.
A review of the larger context for the projections used in conducting our analysis lends
further support for our conclusion that the full suite of emissions controls for 2026 is appropriate,
given the need to balance both overcontrol and undercontrol concerns in the complex arena of
forecasting interstate ozone transport. Even with full implementation of the final rule, based on
the CAMx photochemical modeling of the Final Rule Policy Control Scenario, these two
receptors are only projected to come into attainment by a relatively small degree, and these
projections reflect a combination of both this rule's requirements and anticipated but
unenforceable economic and meteorological projections for 2026. Moreover, the form of
implementation of this rule for both EGUs and non-EGUs, as discussed in Section VI of the
preamble, is designed to ensure a certain degree of emissions control performance (as determined
at Step 3) without dictating the operational levels of any facility. The form of implementation
does not place an enforceable cap on total emissions such that the total estimated emissions
reductions from the rule that inform our overcontrol analysis can be considered to be absolutely
certain or legally enforceable. Under these circumstances, attempting to parse out some lesser
stringency of control for any state whose linkage just barely resolves in 2026 under the rule
would go beyond the Agency's obligation to avoid "over-control" and impinges the equally
compelling imperative to avoid "under-control." The projected resolution of an air quality
receptor to just barely achieving attainment should generally be considered a positive result of
the EPA's good neighbor rulemakings, not a result to be avoided.
64
-------
Table C-7. 2023 Average Ozone DVs (ppb) for NOx Emissions Cost Threshold Levels
($/ton) Assessed Using the Ozone AQAT for All Receptors. _
Site
state
county
Engineering
Analysis
Base
SCR
Optimize
SCR /
Optimize/
SOAOC
SCR
Optimize +
SNCR
Optimize
SCR /
Optimize/
soa ca+
SNOR
Optimize
SCR /
Optimize y
soaco4-
SNCR
Optinmze +
SCpi/SNCR
/Retrofit
r Full Step 3
/- EGU only")
40278011
Arizona
Yuma
70.36
70.35
70.34
70.34
70.34
70.30
80350004
Colorado
Douglas
71.12
71.10
71.10
71.10
71.10
70.34
80590006
Colorado
Jefferson
72.63
72.61
72.61
72.61
72.61
71.99
80590011
Colorado
Jefferson
73.29
73.27
73.27
73.27
73.27
72.42
80690011
Colorado
Larimer
70.79
70.78
70.78
70.78
70.78
70.25
90010017
Connecticut
Fairfield
71.62
71.58
71.57
71.57
71.56
71.42
90013007
Connecticut
Fairfield
72.99
72.93
72.91
72.91
72.90
72.68
90019003
Connecticut
Fairfield
73.32
73.28
73.26
73.27
73.25
73.05
90099002
Connecticut
New Haven
70.61
70.54
70.52
70.53
70.51
70.30
170310001
Illinois
Cook
68.13
68.11
68.11
68.11
68.11
67.92
170314201
Illinois
Cook
67.92
67.88
67.88
67.88
67.88
67.76
170317002
Illinois
Cook
68.47
68.38
68.38
68.37
68.37
68.22
350130021
New Mexico
Dona Ana
70.83
70.82
70.82
70.82
70.82
70.61
350130022
New Mexico
Dona Ana
69.73
69.72
69.72
69.72
69.72
69.51
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
70.59
70.53
70.53
70.52
70.52
69.61
481210034
Texas
Denton
69.93
69.90
69.88
69.89
69.88
69.35
481410037
Texas
El Paso
69.82
69.82
69.81
69.81
69.81
69.57
481671034
Texas
Galveston
71.82
71.75
71.72
71.73
71.70
70.49
482010024
Texas
Harris
75.33
75.27
75.27
75.25
75.25
74.30
482010055
Texas
Harris
71.19
71.13
71.11
71.12
71.10
70.07
482011034
Texas
Harris
70.32
70.26
70.26
70.25
70.25
69.31
482011035
Texas
Harris
68.01
67.95
67.95
67.94
67.94
67.06
490110004
Utah
Davis
71.88
71.87
71.87
71.87
71.87
70.79
490353006
Utah
Salt Lake
72.48
72.47
72.47
72.47
72.47
71.44
490353013
Utah
Salt Lake
73.21
73.20
73.20
73.20
73.20
72.32
550590019
Wisconsin
Kenosha
70.75
70.65
70.65
70.65
70.65
70.42
551010020
Wisconsin
Racine
69.59
69.46
69.46
69.46
69.46
69.25
551170006
Wisconsin
Sheboygan
72.64
72.46
72.46
72.46
72.46
72.19
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
65
-------
Table C-8. 2023 Maximum Ozone DVs (ppb) for NOx Emissions Cost Threshold Levels
($/ton) Assessed Using the Ozone AQAT for All Receptors.
Site
state
county
Engineering
Analysis
Base
SCR
Optimize
SCR /
Optimize/
SOAOC
SCR
Optimize +
SNCR
Optimize
SCR /
Optimize/
SOA CO+
SNCR
Optimize
SCR /
Optimize y
SOA CCA
SNCR
OptijlMze +
SCpSNCR
/Retrofit
("Full Step 3
/- ECU only")
40278011
Arizona
Yuma
72.05
72.04
72.04
72.04
72.04
71.99
80350004
Colorado
Douglas
71.71
71.70
71.70
71.70
71.70
70.93
80590006
Colorado
Jefferson
73.32
73.31
73.31
73.31
73.31
72.68
80590011
Colorado
Jefferson
73.89
73.87
73.87
73.87
73.87
73.01
80690011
Colorado
Larimer
71.99
71.98
71.98
71.98
71.98
71.44
90010017
Connecticut
Fairfield
72.22
72.18
72.17
72.17
72.16
72.02
90013007
Connecticut
Fairfield
73.89
73.83
73.81
73.81
73.80
73.57
90019003
Connecticut
Fairfield
73.62
73.58
73.56
73.57
73.55
73.35
90099002
Connecticut
New Haven
72.71
72.65
72.62
72.63
72.61
72.39
170310001
Illinois
Cook
71.82
71.80
71.80
71.80
71.80
71.61
170314201
Illinois
Cook
71.41
71.37
71.37
71.37
71.37
71.24
170317002
Illinois
Cook
71.27
71.17
71.17
71.17
71.17
71.00
350130021
New Mexico
Dona Ana
72.13
72.12
72.12
72.12
72.12
71.91
350130022
New Mexico
Dona Ana
72.43
72.42
72.42
72.42
72.42
72.20
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
72.69
72.63
72.63
72.62
72.62
71.69
481210034
Texas
Denton
71.73
71.70
71.68
71.69
71.68
71.14
481410037
Texas
El Paso
71.43
71.42
71.41
71.41
71.41
71.16
481671034
Texas
Galveston
73.13
73.05
73.02
73.03
73.01
71.77
482010024
Texas
Harris
76.93
76.87
76.87
76.85
76.85
75.88
482010055
Texas
Harris
72.20
72.13
72.12
72.12
72.10
71.06
482011034
Texas
Harris
71.52
71.46
71.46
71.45
71.45
70.49
482011035
Texas
Harris
71.52
71.46
71.46
71.45
71.45
70.52
490110004
Utah
Davis
74.08
74.07
74.07
74.07
74.07
72.96
490353006
Utah
Salt Lake
74.07
74.06
74.06
74.06
74.06
73.02
490353013
Utah
Salt Lake
73.71
73.70
73.70
73.70
73.70
72.81
550590019
Wisconsin
Kenosha
71.65
71.55
71.55
71.55
71.55
71.32
551010020
Wisconsin
Racine
71.39
71.25
71.25
71.25
71.25
71.04
551170006
Wisconsin
Sheboygan
73.54
73.36
73.36
73.36
73.36
73.08
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
66
-------
Table C-9. 2026 Average Ozone DVs (ppb) for NOx Emissions Cost Threshold Levels
($/ton) Assessed Using the Ozone AQAT for All Receptors.
Site
state
county
Engineering
Analysis
Base
SCR
Optimize
SCR
Optimize +
SOACC
SCR
Optimize +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit
("Full Step
3-EGU
only")
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
non-EGU
("Full Step
3")
40278011
Arizona
Yuma
69.87
69.86
69.86
69.86
69.86
69.84
69.80
80590006
Colorado
Jefferson
71.70
71.69
71.69
71.69
71.69
71.36
71.34
80590011
Colorado
Jefferson
72.06
72.05
72.05
72.05
72.05
71.59
71.57
80690011
Colorado
Larimer
69.84
69.83
69.83
69.83
69.83
69.54
69.53
90013007
Connecticut
Fairfield
71.25
71.20
71.18
71.18
71.17
70.98
70.66
90019003
Connecticut
Fairfield
71.58
71.53
71.52
71.52
71.51
71.34
71.06
350130021
New Mexico
Dona Ana
70.06
70.05
70.05
70.05
70.05
69.89
69.86
350130022
New Mexico
Dona Ana
69.17
69.16
69.15
69.15
69.15
69.00
68.96
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
69.89
69.84
69.84
69.82
69.82
68.96
68.50
481671034
Texas
Galveston
71.29
71.22
71.19
71.20
71.17
70.02
69.28
482010024
Texas
Flarris
74.83
74.77
74.77
74.76
74.76
73.86
73.39
490110004
Utah
Davis
69.90
69.90
69.90
69.90
69.90
69.34
69.28
490353006
Utah
Salt Lake
70.50
70.49
70.49
70.49
70.49
69.96
69.91
490353013
Utah
Salt Lake
71.91
71.91
71.91
71.91
71.91
71.45
71.40
551170006
Wisconsin
Sheboygan
70.83
70.66
70.66
70.65
70.65
70.51
70.27
67
-------
Table C-10. 2026 Maximum Ozone DVs (ppb) for NOx Emissions Cost Threshold Levels
($/ton) Assessed Using the Ozone AQAT for All Receptors.
Site
state
county
Engineering
Analysis
Base
SCR
Optimize
SCR
Optimize +
SOACC
SCR
Optimize +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit
("Full Step
3-EGU
only")
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
non-EGU
("Full Step
3")
40278011
Arizona
Yuma
71.47
71.46
71.46
71.46
71.46
71.44
71.40
80590006
Colorado
Jefferson
72.30
72.29
72.29
72.29
72.29
71.95
71.93
80590011
Colorado
Jefferson
72.66
72.65
72.65
72.65
72.65
72.19
72.16
80690011
Colorado
Larimer
71.04
71.03
71.03
71.03
71.03
70.73
70.72
90013007
Connecticut
Fairfield
72.06
72.00
71.98
71.99
71.97
71.78
71.46
90019003
Connecticut
Fairfield
71.78
71.73
71.72
71.72
71.71
71.54
71.26
350130021
New Mexico
Dona Ana
71.36
71.35
71.35
71.35
71.35
71.19
71.16
350130022
New Mexico
Dona Ana
71.77
71.76
71.76
71.76
71.76
71.60
71.56
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
72.02
71.96
71.96
71.95
71.95
71.06
70.58
481671034
Texas
Galveston
72.51
72.44
72.41
72.42
72.39
71.22
70.47
482010024
Texas
Flarris
76.45
76.39
76.39
76.38
76.38
75.46
74.98
490110004
Utah
Davis
72.10
72.10
72.10
72.10
72.10
71.52
71.46
490353006
Utah
Salt Lake
72.10
72.09
72.09
72.09
72.09
71.55
71.50
490353013
Utah
Salt Lake
72.31
72.31
72.31
72.31
72.31
71.84
71.80
551170006
Wisconsin
Sheboygan
71.73
71.55
71.55
71.55
71.55
71.41
71.17
68
-------
Table C-ll. 2023 Maximum Air Quality Contribution (ppb) to a Remaining Receptor.
54
state
Engineering
Analysis Base
SCR Optimize
SCR /
Optimize y
SOA Cy
SCR Optimize
+ SNCR
Optimize
SCR /
Optimize y
SOA CCA
SNCIt
Optimize
SCR /
Optimize y
SOACCA
SNCIt
Optimize +
SCR/SNCR
Retrofit
("full Step 3
/ - EGU
/ only")
Alabama
0.77
0.77
0.77
0.77
0.77
0.76
Arkansas
1.18
1.18
1.18
1.18
1.18
1.06
California
6.27
6.26
6.26
6.26
6.26
6.26
Illinois
19.08
19.08
19.08
19.08
19.08
19.09
Indiana
9.88
9.82
9.82
9.82
9.82
9.66
Kentucky
0.85
0.85
0.84
0.85
0.84
0.79
Louisiana
9.70
9.66
9.66
9.66
9.66
9.30
Maryland
1.31
1.31
1.31
1.31
1.31
1.31
Michigan
1.60
1.60
1.60
1.60
1.60
1.58
Minnesota
0.85
0.85
0.85
0.85
0.85
0.84
Mississippi
1.42
1.42
1.39
1.42
1.39
1.31
Missouri
1.95
1.82
1.82
1.82
1.82
1.74
Nevada
1.05
1.05
1.05
1.05
1.05
0.99
New Jersey
8.37
8.38
8.38
8.38
8.38
8.38
New York
16.12
16.12
16.12
16.12
16.12
16.10
Ohio
2.04
2.02
2.02
2.02
2.02
2.02
Oklahoma
1.03
1.02
1.02
1.02
1.02
0.98
Pennsylvania
5.99
5.97
5.97
5.97
5.97
5.94
Texas
4.75
4.75
4.75
4.75
4.75
4.64
Utah
1.29
1.29
1.29
1.29
1.29
0.93
Virginia
1.82
1.81
1.81
1.81
1.81
1.80
West Virginia
1.52
1.50
1.49
1.50
1.48
1.43
Wisconsin
2.87
2.87
2.87
2.87
2.87
2.85
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
54 Values greater than or equal to 0.70 ppb indicate the state remains linked to a remaining downwind receptor.
69
-------
Table C-12. 2026 Maximum Air Quality Contribution
(ppb) to a Remaining Receptor
State
Engineering
SCR
SCR
SCR
SCR
SCR Optimize
SCR Optimize +
Analysis Base
Optimize
Optimize +
Optimize +
Optimize +
+ SOA CC +
SOA CC +
SOACC
SNCR
Optimize
SOACC +
SNCR
Optimize
SNCR
Optimize +
SCR/SNCR
Retrofit ("Full
Step 3 - EGU
only")
SNCR Optimize
+ SCR/SNCR
Retrofit + non-
EGU ("Full
Step 3")
Arkansas
1.12
1.12
1.12
1.12
1.12
1.01
0.57
California
6.09
6.08
6.08
6.08
6.08
6.08
6.04
Illinois
13.60
13.60
13.60
13.60
13.60
13.59
13.57
Indiana
8.34
8.27
8.27
8.27
8.27
8.22
8.05
Kentucky
0.81
0.80
0.80
0.80
0.80
0.75
0.72
Louisiana
9.67
9.64
9.64
9.63
9.63
9.29
4.30
Maryland
1.08
1.08
1.08
1.08
1.08
1.08
1.08
Michigan
1.47
1.47
1.47
1.47
1.47
1.46
1.45
Mississippi
1.32
1.32
1.29
1.32
1.29
1.21
0.35
Missouri
1.78
1.65
1.65
1.65
1.65
1.59
1.55
Nevada
0.90
0.90
0.90
0.90
0.90
0.90
0.90
New Jersey
8.09
8.10
8.10
8.10
8.10
8.10
8.11
New York
12.68
12.67
12.67
12.67
12.67
12.66
12.64
Ohio
1.92
1.90
1.90
1.90
1.90
1.90
1.85
Oklahoma
0.77
0.77
0.77
0.77
0.77
0.72
0.61
Pennsylvania
5.70
5.68
5.68
5.68
5.68
5.65
5.55
Texas
4.44
4.44
4.44
4.43
4.43
4.34
4.30
Utah
1.07
1.07
1.07
1.07
1.07
0.89
0.88
Virginia
1.14
1.14
1.14
1.14
1.13
1.13
1.10
West Virginia
1.36
1.35
1.34
1.34
1.33
1.28
1.24
55 Values greater than or equal to 0.70 ppb indicate the state remains linked to a remaining downwind receptor.
70
-------
Table C-13. Description of the Various Scenarios Evaluated in AQAT.
Scenario
Cost Threshold
Level
Description
0
$0
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
1
$1,600
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs +SCR
optimize
2
$1,600
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs +SCR
optimize + SOA CC
3
$1,800
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
+SCR/SNCR optimize
4
$1,800
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
+SCR/SNCR optimize + SOA CC
5
$11,000
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
+SCR/SNCR optimize + SOA CC + SCR Retrofit
8
$11,000+ non-
EGUs
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
+SCR/SNCR optimize + SOA CC + SCR Retrofit + non-EGUs
9
$1,800 + non-
EGUs
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
+SCR/SNCR optimize + SOA CC + non-EGUs
10
AQ Modeling
Control Scenario
Emission levels associated with the AQ modeling of the control scenario.
14
$0 w/IRA
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs + delta
in emissions between IPM base and IPM base w/IRA
15
$11,000 w/IRA
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
+SCR/SNCR optimize + SOA CC + SCR Retrofit + delta in emissions between
IPM final policy and IPM final policy w/IRA
16
$11,000+ non-
EGUs w/IRA
Baseline Engineering Analysis 202x OS NOx + engineering non-CEMs
+SCR/SNCR optimize + SOA CC + SCR Retrofit + non-EGUs + delta in
emissions between IPM final policy and IPM final policy w/IRA
71
-------
4. Comparison between the air quality assessment tool estimates using the primary and
alternative calibration factors
As described earlier, the "primary" version of AQAT was calibrated using modeled
ozone data from the proposed rule using a 2026 case where EGUs and non-EGUs were reduced
by 30%. Since the primary calibration factors were developed by modulating the sectors being
regulated in this rulemaking, we conclude that these calibration factors were the most appropriate
ones to use within the Step 3 methodology. However, we also created a second set of
"alternative" calibration factors, reflecting changes between the 2023 and 2026 base cases using
AQ modeling from the final rule. Each of these sets of calibration factors represents a different
assessment of a linear relationship between emissions reductions and changes in air quality based
on the different emission levels and reductions from various sectors. Thus, it was possible to
produce air quality estimates from the tool for emissions scenarios using the "primary"
calibration factors as well as similar results using the "alternative" calibration factors.
Comparing those results, we are able to assess the importance of the particular calibration factor
(i.e., linearity assumption assumed) on the conclusions. The two calibration factors implicitly
have different assumptions about the spatial distribution of the emissions reductions and of the
sectors being reduced. While EPA believes its primary version is the most appropriate calibration
approach, the use of alternatives calibration factors for sensitivity analysis allows EPA to ensure
its findings are consistent and robust across a range of assumptions regarding source, location,
and degree of emission changes.
The two calibration scenarios bracket the policy range explored using the AQAT. In this
section, we assessed the effects of the calibration factors, focusing on two separate policy-
relevant emissions scenarios. Appendix J presents an additional comparison using the two
calibration factors - comparing against the CAMx Final Rule Policy Control scenario.
Using the primary and alternative calibration factors for the Step 3 configuration of
AQAT, we assessed the maximum design values for two policy-relevant scenarios: the 2026
engineering analysis base case scenario and the "Full Step 3" Scenario. For each of these
scenarios, EPA looked at the difference in maximum design values using the primary calibration
and the alternative calibration. The results are shown in Tables C-14 and C-15, respectively. The
AQAT values and the differences in the tables have been rounded to a hundredth of a ppb. For
these two scenarios, the differences are moderate between the two AQAT calibrations, with a
largest difference of 0.59 ppb for the engineering analysis base. The largest difference was 0.57
ppb for the "Full Step 3" Scenario. This largest difference occurred at the Galveston Texas
receptor.
In this assessment, most receptors maintain the same attainment condition (i.e., showing
average and/or maximum design values either above or below the level of the NAAQS)
regardless of the calibration factor utilized. This indicates EPA's air quality findings are robust
to the remaining nonlinearity in ozone chemistry and uncertainties in the geographical
distribution of the sources (after accounting for the majority of this nonlinearity using the
calibration factor). Specifically, by using multiple calibration factors that arrive at the same
conclusions regarding linkages and overcontrol, this analysis illustrates that the nonlinearity of
the ozone chemistry that is not accounted for using a single linear calibration factors across the
range of emission reductions assessed here and/or the difference in spatial location and intensity
of the sources and/or differences in the sectors are not affecting the conclusions about whether
receptors are resolved and whether states continue to have contributions above the linkage
72
-------
threshold to those receptors. For the engineering base case, all receptors had maximum design
values at or above the NAAQS using both calibration factors. This tends to confirm the air
quality and contribution modeling using CAMx that the states linked to these receptors are
appropriately included in the rule.
In the "Full Step 3" Scenario, there are some differences in the receptor status between
primary and alternative calibration factors. However, none of these differences would impact
EPA's overcontrol finding. Evidence of overcontrol would, at a minimum, require 1) all
receptors to which an upwind state is linked to drop below the NAAQS at both the full
implementation of mitigation measures (i.e., "Full Step 3") and in the scenario where the last
increment of reductions is removed (i.e., "Full Step 3 - EGU only") or 2) show a state's
contribution to drop below 1% in both cases. These conditions are not met under either primary
or alternative calibration factors. For example, the Galveston Texas receptor is estimated to be
resolved using the version of AQAT with the primary calibration factor but is estimated to
remain above the NAAQS using the alternative calibration factors. However, there is no
difference in the receptor status in the penultimate Step 3 increment (i.e., "Full Step 3 - EGU
only") using either calibration factor (and thus no evidence of overcontrol). One other difference
in regulatory status for a receptor occurs using the alternative calibration factors: the maximum
design value for the Salt Lake Utah receptor (490353006) remains above the NAAQS using the
version of AQAT with the primary calibration factor but is just barely below 71 ppb (less than
0.01 ppb) using the alternative calibration factors. However, this change in status has no impact
in terms of eliminating all of any upwind state's linkages. This assessment, again, suggests that
the control level selected in Step 3 is appropriate.
Finally, using the alternative calibration factor, we examined the maximum contribution
to the highest remaining receptor for each upwind state (Table C-16). In this case, all states
remain linked when the emissions reductions from the "Full Step 3" scenario are applied. This
further affirms no overcontrol for upwind states only linked to the Galveston Texas receptor. For
instance, Oklahoma presents possibly the closest case for analysis. Under the primary calibration
approach, analyzing the "Full Step 3" scenario of the final rule, the air quality contribution for
Oklahoma drops below the 1% contribution threshold to the Galveston Texas receptor, and the
receptor's maximum design value also drops below 71 ppb, but there is no overcontrol as no
such conditions occurred in the penultimate step (i.e., "Full Step 3 - EGU only") as described
above. Under the alternative calibration scenario, however, Oklahoma's contribution remains
above the linkage threshold to this receptor in the "Full Step 3" scenario (and the receptor also
remains above 71 ppb), putting even more distance between Oklahoma and any potential
overcontrol.56
In the past, some opponents of EPA's transport regulatory actions have misconstrued the
overcontrol test to require that EPA should investigate hypothetical ever-more-thinly-sliced
"stopping points" within the emissions control program on the mistaken premise that regulators
can somehow stop on a dime where not one pound of emissions reduction more than is
purportedly necessary would be required of that state. Neither the EPA nor the Supreme Court of
the United States endorse this perspective as an appropriate understanding of the overcontrol test.
56 EPA notes that the Galveston Texas receptor is estimated to be in attainment and maintenance in both the CAMx
Final Rule Policy Control scenario as well as the AQAT estimates using both the "primary" and "alternative"
calibration factors. This "CAMx Final Rule Policy Control" emissions scenario is different than the "Full Step 3"
emissions scenario used in Step 3, where in the "Alternative" version of AQAT the receptor's maintenance issues
remain unresolved.
73
-------
However, the alternative calibration factor analysis presents a plausible alternative method of
assessing the rule's effects in AQAT, and under this method, the debate over that hypothetical
concept of a perfectly precise stopping point would be moot. Since the alternative method
indicates that the state's linkage does not resolve even in the full emissions control scenario of
the final rule, it cannot be established with sufficient certainty based on the present record that
there is any overcontrol with respect to Oklahoma. In short, these findings from the use of the
alternative calibration approach support the conclusions in the preamble that there is no
overcontrol.
The results of this comparison, which are relatively similar, demonstrate that the AQAT
provides reasonable estimates of air quality concentrations for each receptor. Considering the
time and resource constraints faced by the EPA, AQAT can provide reasonable inputs for the
multi-factor and overcontrol assessments.
74
-------
Table C-14. 2026 Maximum Ozone DVs (ppb) for the Engineering Analysis Base Scenario
Using Two Calibration Factors.
Site
state
county
Primary
Calibration
Alternative
Calibration
Delta AQ
between
Calibration
Approaches
40278011
Arizona
Yuma
71.47
71.51
-0.04
80590006
Colorado
Jefferson
72.30
72.46
-0.16
80590011
Colorado
Jefferson
72.66
72.76
-0.10
80690011
Colorado
Larimer
71.04
71.02
0.02
90013007
Connecticut
Fairfield
72.06
72.12
-0.06
90019003
Connecticut
Fairfield
71.78
71.89
-0.11
350130021
New Mexico
Dona Ana
71.36
71.45
-0.09
350130022
New Mexico
Dona Ana
71.77
71.79
-0.02
350151005
New Mexico
Eddy
350250008
New Mexico
Lea
480391004
Texas
Brazoria
72.02
71.73
0.28
481671034
Texas
Galveston
72.51
71.92
0.59
482010024
Texas
Flarris
76.45
76.03
0.42
490110004
Utah
Davis
72.10
72.16
-0.06
490353006
Utah
Salt Lake
72.10
72.11
-0.01
490353013
Utah
Salt Lake
72.31
72.36
-0.05
551170006
Wisconsin
Sheboygan
71.73
71.93
-0.20
75
-------
Table C-15. 2026 Maximum Ozone DVs (ppb) for the "Full Step 3" Scenario Using Two
Calibration Factors.
Site
state
county
Primary
Calibration
Alternative
Calibration
Delta AQ
between
Calibration
Approaches
40278011
Arizona
Yuma
71.40
71.40
0.00
80590006
Colorado
Jefferson
71.93
72.20
-0.26
80590011
Colorado
Jefferson
72.16
72.38
-0.22
80690011
Colorado
Larimer
70.72
70.73
-0.01
90013007
Connecticut
Fairfield
71.46
71.57
-0.11
90019003
Connecticut
Fairfield
71.26
71.31
-0.05
350130021
New Mexico
Dona Ana
71.16
71.13
0.03
350130022
New Mexico
Dona Ana
71.56
71.54
0.01
350151005
New Mexico
Eddy
350250008
New Mexico
Lea
480391004
Texas
Brazoria
70.58
70.89
-0.30
481671034
Texas
Galveston
70.47
71.04
-0.57
482010024
Texas
Flarris
74.98
75.25
-0.27
490110004
Utah
Davis
71.46
71.03
0.44
490353006
Utah
Salt Lake
71.50
70.99
0.51
490353013
Utah
Salt Lake
71.80
71.65
0.15
551170006
Wisconsin
Sheboygan
71.17
71.33
-0.16
76
-------
Table C-16. 2026 Maximum Air Quality Contribution (ppb) to a Remaining Receptor
Using the Alternative Calibration.57
State
Engineering
Analysis Base
SCR
Optimize
SCR
Optimize +
SOACC
SCR
Optimize +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize
SCR Optimize
+ SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit ("Full
Step 3 - EGU
only")
SCR Optimize +
SOA CC +
SNCR Optimize
+ SCR/SNCR
Retrofit + non-
EGU ("Full
Step 3")
Arkansas
1.14
1.14
1.14
1.14
1.14
1.08
0.76
California
6.09
6.07
6.07
6.07
6.07
6.07
6.03
Illinois
13.67
13.67
13.67
13.66
13.66
13.62
13.54
Indiana
8.44
8.41
8.41
8.41
8.41
8.39
8.31
Kentucky
0.81
0.80
0.80
0.80
0.80
0.75
0.72
Louisiana
9.46
9.45
9.45
9.45
9.45
9.35
9.21
Maryland
1.08
1.08
1.08
1.08
1.08
1.08
1.08
Michigan
1.44
1.44
1.44
1.44
1.44
1.39
1.32
Mississippi
1.34
1.34
1.31
1.34
1.31
1.22
1.14
Missouri
1.78
1.65
1.65
1.65
1.65
1.58
1.54
Nevada
0.90
0.90
0.90
0.90
0.90
0.90
0.83
New Jersey
8.12
8.10
8.10
8.10
8.10
8.10
8.08
New York
12.71
12.71
12.71
12.71
12.71
12.67
12.60
Ohio
1.92
1.90
1.90
1.90
1.90
1.90
1.85
Oklahoma
0.77
0.77
0.76
0.77
0.76
0.73
0.70
Pennsylvania
5.69
5.68
5.68
5.67
5.67
5.64
5.55
Texas
4.51
4.50
4.50
4.50
4.50
4.33
4.27
Utah
1.07
1.07
1.07
1.07
1.07
0.93
0.92
Virginia
1.13
1.12
1.12
1.12
1.12
1.12
1.10
West Virginia
1.36
1.36
1.35
1.35
1.35
1.33
1.33
5. Assumptions made in the air quality assessment tool
There are some key assumptions about the relationship between emission and air quality
within the AQAT. In particular, we assume that the downwind air quality improvement is
indifferent to the geographic location and to the physical characteristics of the particular
emission source within the state where a particular ton was reduced. We also assume that the
emissions are reduced in a proportional way across the ozone-season and are not preferentially
eliminated on particular days or at particular hours. We also assume that the air quality impact is
indifferent to height of release or to the particular source sector from which it 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 non-EGU source
sector. Note that, in this particular AQAT, the emissions reductions assessed under various
scenarios in the rule are exclusively from the EGU and non-EGU sectors and these sectors match
the sectors on which the calibration factors are based. Though, the distribution of sources may be
different. As described in the section on the construction of AQAT, the calibration factors are
built using the pattern of emission reduction and the resulting air quality changes between the
two photochemical modeling runs (from the proposal).
In actuality, emission reductions will be concentrated at individual sources. The resulting
air quality improvements from these emission reductions will be larger in the immediate vicinity
of the source. At larger downwind distances, the unit-by-unit variations in emission behavior
57 Values greater than or equal to 0.70 ppb indicate the state remains linked to a remaining downwind receptor.
77
-------
(relative to the calibration scenario) will be substantially less important as transport and
dispersion reduces the gradients in concentration. The closer the distribution of sources and the
magnitudes of reductions at those sources match the pattern of reductions used to construct the
calibration factor, the less uncertainty there will be in the results.
One additional source of uncertainty within AQAT is the relationship between NOX
emissions and ozone concentrations. This relationship 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). Figure C-l is an adaptation of this figure, where we have
isolated the ozone isopleths at 70 and 80 ppb. One can readily find examples in the scientific
literature (e.g., Kinosian, 1982; Luo et al., 2021; and Koplitz et al., 2021) where similar figures
are presented.58,59,60
200
_Q
Q-
Q_
160
120
80
40
70 ppb/ /80 ppb
(b)
1
t
(a)
/)\
/
^ 80 ppb
70 ppb
400 800 1200
Initial VOC, ppbC
1600
2000
Figure C-l. An adaptation of the ozone isopleth diagram from Seinfeld and Pandis
(2006). The ozone isopleths show nonlinear relationships between NOX and VOC emissions. At
locations a, b, and c, the isopleth lines are parallel to each other, suggesting a linear relationship
at each of those emissions regimes.
58 S. Koplitz, H. Simon, B. Henderson, J. Liljegren, G. Tonnesen, A. Whitehill, B. Wells
Changes in ozone Chemical Sensitivity in the United States from 2007 to 2016
ACS Environ. Au(2021), 10.1021/acsenvironau.lc00029
lc00029. https://pubs.acs.org/doi/full/10.1021/acsenvironau. Ic00029
59 J.R.Kinosian. 1982. Ozone-Precursor relationships fromEKMA Diagrams. Environ. Sci. Technol., Vol. 16, No.
12, 1982. https://pubs.acs.Org/doi/pdf/10.1021/es00106a011
611H. Luo, K. Zhao, Z.Yuan, L.Yang, J. Zheng, Z. Huang, X. Huang. 2021. Emission source-based ozone isopleth
and isosurface diagrams and their significance in ozone pollution control strategies. Journal of Enviromnental
Sciences, Volume 105, July 2021, Pages 138-149. https://doi.Org/10.1016/j.jes.2020.12.033
78
-------
This nonlinearity can be seen by following one of the ozone isopleth lines and observing
that there are various combinations of NOX and VOC that result in a constant level of ozone,
that the lines are not straight over the entire emissions regime. For example, there are particular
levels of VOC emissions with different level of NOX emissions that can result in the same ozone
concentration (Figure C-l). The nonlinearities are evident over tens of percent changes in the
overall emission inventories and tens of ppb of ozone changes. Focusing in on small areas in the
figure (see, for example, locations a, b, and c), one can observe that the isopleths are often
parallel to each other (when looking at some smaller range of NOX and/or VOC changes). This
suggests that, for that particular emissions and ozone regime, that one could expect a linear
relationship between emissions change and concentration change (assuming that the meteorology
is held constant). The linearity would be present even with simultaneous VOC emission changes
(particularly if they vary in proportion to the NOx emission changes). In some cases the linear
relationship between NOx emission change and ozone change can be positive (i.e., emission
reductions result in decreases in ozone (see for example location c in Figure C-l)) while in other
cases it is negative (i.e., emission reductions result in increases in ozone, see for example,
locations a and b in Figure C-l). The relationship between emissions and ozone concentration
depends on the levels and composition of the NOx and VOC emissions as well as on the
particular meteorology in that area. For a particular location, the relationship can vary from one
day to the next as the emissions and meteorology change. As described in the Air Quality
Modeling TSD, in this action, the air quality modeling average and maximum design values and
state contributions are based on averaging multiple days together. So, the relationship between
NOx, VOC, and the resulting concentration change in the contributions is also based on
averaging the response over these days.
Relationships between emissions and ozone concentrations comparable to that shown in
Figure C-l are usually created for particular locations and focus on local relationships, but the
general principles can apply for each of the chemical constituents including those transported to
the location. As described in the Air Quality Modeling TSD, during transport, the emissions
form ozone (which can undergo additional transformations as it passes from one chemical
regime to the next). Consequently, pollution from one upwind state may be in the form of ozone
or NOx, for example, as it encounters the downwind area, while local emissions of NOx or
VOCs may still be in the process of transforming into ozone. In both cases, we would expect a
linear relationship between emissions changes and changes in concentration. But those
relationships could be different. The relationships for a particular receptor and state can be seen
in the calibration factor for that receptor and state. The calibration factors range from positive
values to negative values, though most are positive (and they tend to go toward a value of 1 (or
higher) for states that are farther away from particular monitors indicating that a particular
percent change in NOx emissions would result in the same percent change in ozone contribution
from that state. For the state containing the monitor, the values tend to be lower (meaning the
monitor is less responsive to emission changes from that state on a percentage basis).
For the states evaluated here, under the various control scenarios, the changes in the
emission inventory are on the order of a few percent and the resulting air quality changes are on
the order of a fraction of a ppb. Consequently, as described above, the changes in air quality in
response to emissions changes are likely to be linear over this small range. In this assessment
tool, we are assuming a linear relationship between NOx emissions and ozone concentrations,
but this relationship is calibrated using two CAMx simulations (basically giving us known points
79
-------
on the figure (conceptually similar to where the parallel lines cross the ozone isopleths at
locations a, b, or c in Figure C-l). Note that the emissions differences and the resulting changes
in air quality between the two CAMx simulations is less than 10 ppb (making it even more likely
that the relationship is linear). This relationship should hold for emission reductions around the
area that calibration factor was created for (both in the emission regime between those two
CAMx simulations and the area immediately above and below those modeled emission levels).
Thus, while emissions and ozone are demonstrably nonlinear, CAMx photochemical modeling
allows us to identify an area on the emissions and ozone curve and describe it using a linear
relationship. Errors and uncertainty in the linear calibration approach will occur if the reduction
between the two air quality model simulations is too large, or if the two simulations are too close
together (i.e., with little emission change between the scenarios).
Using an earlier version of the tool, EPA had the tool and methodology peer-reviewed
(see AQAT Review Summary Memo included in the docket). This review focused on applying
the methodology to S02 emissions and sulfate concentrations for estimating PM2.5, highlighting
some of the primary assumptions that were made in that version of the tool and offering
suggestions. In the case of this tool, a number of the improvements (such as individual state and
receptor calibration factors, calibration factors based on emission changes from a particular
source sector (and corresponding heights of emissions release), and holding the days used in the
creation of the average contributions) conform to suggestions made by the reviewers.
Finally, as done in the earlier section of this TSD (Section C.4), we can assess the effects
of the uncertainty resulting from the assumptions within AQAT (including nonlinearity in the
emissions to ozone relationship, variation in geographic location of the sources, time and
magnitude of emission release, and source sector) by using an alternative set of calibration
factors created using another emissions scenario modeled in CAMx. As described, above, this
comparison confirms the results.
80
-------
D. Selection of Backstop Emission Rate
For the reasons described in the preamble, EPA is complementing the longer-term mass-
based trading program (premised on seasonal emission rate performance) with a short-term
"backstop" emission rate for some units. This section discusses how that rate was set. At
proposal, EPA considered hourly, 24-hour, 7-day and 30-day periods as potentially appropriate
averaging lengths for the rate. While all these time periods would likely provide appropriate
assurance for post-combustion controls to operate on an hourly and daily basis, including during
ozone episodes, as described in the preamble, EPA is finalizing the daily (e.g., 24-hr) period as
an appropriate length of averaging time for the backstop rate.
As described in the preamble, in implementing the daily backstop emission rates, the
EPA is accounting for emissions during start-up and shutdown where the emission rate may
exceed the daily limit by including a 50-ton buffer.
As described in the preamble, in establishing the appropriate rate, EPA evaluated several
methods and data sets. These are:
1. EPA evaluated daily emission patterns for units that have SCRs with seasonal rates in the
range of the average seasonal emission rates identified in the rulemaking (i.e., at 0.08
lb/MMBtu or below).
2. EPA applied the concept of "comparable stringency" developed in the 2014 1-hr S02
attainment area guidance for converting emission rates so they provide comparable
stringency over different time frames. In this case, we convert longer-term emission rate
assumptions (e.g., seasonal and monthly rates at 0.08 lb/MMBtu to daily rates at 0.14
lb/MMBtu)
3. EPA evaluated start up and shut down events and identified a 50-ton threshold before any
additional 2 allowances per ton surrender requirement is triggered in an effort to
accommodate these events.
Each of these methods is discussed in more detail, below.
1. Observations of fleet operation for well-controlled units
EPA examined the daily operation of coal-fired units with SCR in 2021, comparing the
daily rate to the seasonal average rate. We counted the number of days that had values higher
than particular values (e.g., 0.12 lb/MMBtu, 0.14 lb/MMBtu, and 0.16 lb/MMBtu) as a function
of the seasonal average emission rate. Knowing that there is variation in emission rate, with
values above and below the seasonal average, we wanted to identify the frequency and
magnitude of some of the higher emission rate values for units that typically had low seasonal
rates. A low seasonal rate suggests that the post-combustion controls on the unit are well-
designed and modern and are being well-run and well-maintained. The results are shown in
Figure D-l. As an example, for a unit with a seasonal rate of 0.08, we could expect, on average,
about 4.7% of the daily rate values to be higher than 0.14 lb/MMBtu.
81
-------
Focusing on the 0.14 lb/MMBtu rate, EPA identified 164 units that had ozone season
rates at or below 0.08 lb/MMBtu in the 2021 ozone season.61 As described above, daily emission
rates from these units rarely exceeded 0.14 lb/MMBtu. On the days that the rate did exceed, it
was frequently close to the 0.14 lb/MMBtu rate. There were a total of 572 tons of "excess"
emissions (i.e., emissions above what would have been emitted had the emission rate been
capped at 0.14 lb/MMBtu on those days). This compares with 60,339 tons of total seasonal
emissions from those units. Thus, these "excess" emissions are about 0.9% of their seasonal
emissions.
61 See the Excel workbook, Daily Backstop rate for existing SCRs - accommodating startup shutdown.xlsx for
details.
82
-------
A
1.2
E CD
~ £ 0.8
0.6
0.4
0.2
V.
^
0.05 0.1 0.15 0.2 0.25
OS NOx rate (Ib/mmBtu)
0.3
0.35
0.4
0.15 0.2 0.25
OS NOx rate (Ib/mmBtu)
0.1 0.15 0.2 0.25 0.3
OS NOx rate (Ib/mmBtu)
0.4
Figure D-l. Examination of the fraction of operating time where the daily rate was higher than
0.12, 0.14, or 0.16 lb/MMBtu in 2021 (in A, B, and C, respectively) as a function of the average
ozone season emission rate for the unit.
83
-------
2. Creating "comparably stringent" emission rates using the 2014 1-hour S02 concepts
a. Background
In the 2014 Guidance for 1- Hour S02 Nonattainment Area SIP Submissions, EPA
introduced concepts and methods for ensuring that NAAQS violations of the 1-hr S02 NAAQS
do not occur.62'63 For example, the 2014 1-hr S02 Guidance defined a "critical emission value"
to refer to the hourly emission rate that an air quality model predicts would result in the 5-year
average of the annual 99th percentile of daily maximum hourly concentrations at the level of the
1-hour NAAQS, given representative meteorological data for the area. In the guidance EPA
explained that, for that standard, establishing 1-hour limits at the critical emission value is an
approach to developing a control strategy that ensures that NAAQS violations do not occur.
Consequently, the EPA recommended that approach in the September 2011 draft guidance, as it
was consistent with the EPA's longstanding S02 policy that source emission limits should match
the averaging time of the relevant S02 NAAQS. However, EPA notes that different averaging
time-based limits require a case-by-case analysis of specific facts and data, and "comparable
stringency" is not an assumed approvable result.
The EPA continues to consider that approach to be acceptable. As discussed in the
subsequent 2014 Guidance, in order to provide adequate assurance that the NAAQS will be met,
the EPA noted that any emissions limits based on averaging periods longer than 1 hour should be
designed to have comparable stringency to a 1-hour average limit at the critical emission value.
A limit based on the 30-day average of hourly emissions levels, for example, at a given numeric
level is likely to be a less stringent limit than a 1-hour limit at the same numeric level since the
control level needed to meet a 1-hour limit every hour is likely to be greater than the control
level needed to achieve the same limit on a 30-day average basis. Therefore, as a general matter,
the EPA expects that any emission rates with a longer averaging time would reflect a lower
numeric emission rate and emission rates with shorter averaging time would reflect a higher
numeric emission rate. Although the emission rate values are different numerically, they are of
comparable stringency when the averaging time is applied.
b. Application
In this rule, EPA is looking to ensure that emission reductions achieved are
commensurate with the installation and operation of post-combustion control devices for
portions of the fossil EGU fleet. Consistent with the 8-hour ozone NAAQS time frame, EPA is
meeting its statutory obligation to eliminate significant contribution from upwind states, in part,
by ensuring the operation of these post-combustion controls (or commensurate reductions) every
day of the ozone season when the units are operating. To achieve this, EPA converts its seasonal
emission rate performance assumptions for such post-combustion control technology (used to
determine seasonal state mass limits) to a daily emission rate of comparable stringency. EPA
does this by utilizing the concepts applied in the 2014 1-hour S02 Guidance. That Guidance was
developed for a similar purpose, to identify "comparably stringent" emissions limits over
62 Docket ID: EPA-HQ-OAR-2021-0668-0123, https://www.epa.gov/sites/default/files/2016-
06/documents/20140423guidance_nonattainment_sip.pdf
63 We note that given the form of the emission rate metric, the emissions and operational data used in the
calculation, as well as the NAAQS being addressed are important to consider when setting an emission rate and that
procedures that may be applicable for one NAAQS (i.e., the 2015 8-hr Ozone) would not necessarily be applicable
for another (e.g., 1-hour S02).
84
-------
different time periods. EPA notes that concept could be applied to help identify daily (e.g., 24-
hour) rates that are comparably stringent to rates based on longer averaging times. In other
words, because we have clear definitions of longer-term (e.g., seasonal) emissions rates that
eliminate significant contribution, we could use the 1-hr S02 methodology to identify
complementary short-term rates that are "comparably stringent" that would ensure control
operation on a daily basis. In this case, we are not looking for 1-hr emission limits, nor are we
looking to limit emissions on a pounds per hour basis to match a modeled "critical emissions
value." Rather, we have seasonal emission rates of 0.08 lb/MMBtu (demonstrating full SCR
operation for units with this existing technology) which can be converted to 24-hour rates in a
pound per unit of heat input rather than a pound per hour framework. As with the 1-hr S02 limit,
we expect that the longer-term rates would be lower than 24-hour rates that would be adjusted
higher to accommodate the variation in operation, demand for electricity, variation in fuel, and
other technical and engineering limitations.
The EPA issued the 2014 1-hr S02 guidance based on consideration of the statistical
nature of the NAAQS and based on analyses of selected cases suggesting that comparably
stringent short term average limits can commonly be expected to provide adequate assurance of
control operation.
Here, EPA expects that an emission rate established for a source with an averaging time
shorter than 30-day or seasonal would be set at a higher level, yet would provide a comparable
degree of stringency as the longer-term emission rate assumption (that would provide assurance
that significant contribution and interference with maintenance are being eliminated). In theory,
the longer-term emission rate assumptions would allow occasional emission spikes, but this
longer-term emission rate (or comparable mass limit implemented in the trading program) would
also require emissions to be lower for most of the averaging period than they would be required
to be with a short-term emission rate (i.e., 24-hour). Here, the EPA envisions that meeting both
the short-term rate and longer-term emission rate assumption in practice would require similar
emission control levels and would commonly result in similar emission patterns, yet having the
short-term backstop rate provides additional assurance that sources will reliably operate their
SCRs each day throughout the ozone season.
In the 2014 1-hour S02 guidance Appendix C presented example calculations in which
the level of the longer-term emission rate is derived from a statistical analysis of a set of data that
reflect the emissions variability that the controlled source is expected to exhibit. The analysis
underlying those example calculations compared the set of emission values averaged over the
longer averaging time against the set of 1-hour emission values from which the longer-term
averages were derived.64 The example calculations in Appendix C reflected a comparison of 99th
percentile values of the sets of 30-day averages and 1-hour averages. Alternative averaging times
were also explored, including 24-hr time-periods. In applying the 1-hour S02 guidance concepts,
here, we envision that the control strategy needed to meet a comparably stringent longer term
emission rate would be essentially the same as the control strategy needed to meet a daily rate,
specifically the operation of SCR post-combustion controls.
64 In the 2014 1-hour S02 guidance, EPA suggested that hourly data for at least 3 to 5 years of stable operation (i.e.,
without changes that significantly alter emissions variability) may be needed to obtain a suitably reliable analysis.
For EGUs such data sets are widely available, as required by 40 CFR part 75 and reported to the EPA. Similar
emissions monitoring is required for a few additional source types under 40 CFR part 51, Appendix P, though these
hourly data are not commonly made publicly available.
85
-------
c. Methods and Results
Starting with the coal-fired EGUs that are currently equipped with SCRs, EPA followed
the methodology laid out in the guidance evaluating daily, 7-day, and 30-day variability on a
lb/MMBtu basis (Table D-l).65'66 We show the estimated rates using the ratios for a seasonal rate
at 0.08 lb/MMBtu. In all cases, we assume a daily emission rate of 0.14 lb/MMBtu (i.e., the
value assumed across the coal steam fleet) is appropriate given that fuel mix does not appear to
substantially change the values.
To convert between the various rates, we can use the ratios of the 99th percentile values
for the various time-periods. As an example, under the 2014 guidance, if we wanted to calculate
a 30-day average rate that was comparably stringent to an hourly rate, we would take the ratio of
the 99th percentile values (the 30-day value divided by the hourly value). This "adjustment
factor" would then be multiplied by the hourly value that we want to convert (usually the hourly
critical emission value, or CEV). Similarly, if we wanted to calculate a daily value, we would
multiply the ratio of the 99th percentile values (the daily value divided by the hourly value) by
the hourly critical emission value.
Comparably stringent 30-day rate = Hourly CEV*Ratio of 30-Day to hourly 99th Percentiles
Comparably stringent Daily rate = Hourly CEV*Ratio of Daily to hourly 99th Percentiles
Combining these two equations, by rearranging both to have the hourly CEV equal in
both, and then solving for the comparably stringent daily rate:
Comparably stringent daily rate =
30-day rate * Ratio of Daily to hourly 99th Percentiles/ Ratio of 30-Day to hourly 99th
Percentiles
EPA computed the following ratios or adjustment factors using the same data procedures
used in creating the ratios in the 2014 guidance. The resulting unit-level 99th percentile ratios for
various averaging times as well as various fleet-wide averages are shown in the excel file
(Units_daily_rate_conversions_proposal.xlsx) included in the docket for the rule. Summary
values are included in Table D-l. Substituting values from Table D-l into the above equations
0.08 lb/MMBtu (a seasonal value taken to be equal to the 30-day rate)*0.97/0.56 = 0.14
lb/MMBtu. Thus, here, following the methodology that EPA outlined in the 2014 guidance, EPA
concludes that a long-term rate of 0.08 lb/MMBtu could be considered to be comparably
stringent to a short-term rate of 0.14 lb/MMBtu. The graphs in Figure D-l show that for units
fully operating their controls (i.e., achieving the 0.08 lb/MMBtu seasonal rate), the daily rate is
unlikely to necessitate any change in performance or behavior.
65 Because of the method for calculating the rate, which is the sum of the daily emissions divided by the daily heat
input utilized, hours where the unit does not operate will not impact the calculation.
66 For this assessment, we assume that the 30-day and seasonal rates would be at comparable levels. Typically, a 30-
day rate would have a larger variability than a seasonal rate inclusive of those particular 30 days, but this should be
relatively small since a seasonal value would include roughly one fifth of the values in the 30-day rate. Here, with
just a few ozone seasons included, EPA did not believe it could reasonably estimate a 99th percentile variability in
seasonal values.
86
-------
Table D-l. Ratios to convert between various time-averages, applied to a 0.08 lb/MMBtu
seasonal rate.
Unit
Plant Type
Fuel
Ratio of
NOx OS
99th
Percentiles
(30 Day
Over
Hour)
Ratio of
NOx OS
99th
Percentiles
(Day Over
Hour)
Ratio of
NOx OS
99th
Percentiles
(Hour
Over
Hour)
Conversion of
Default
Seasonal SCR
Rate to a
Comparably
Stringent Day
Rate
(lb/MMBtu)
coal steam
Fleet avg
0.56
0.97
1
0.14
coal steam
Bituminous
0.53
0.93
1
0.14
coal steam
Bituminous,
Subbituminous
0.56
0.99
1
0.14
coal steam
Lignite
0.73
1.14
1
0.12
coal steam
Subbituminous
0.64
1.01
1
0.13
3. Accommodating startup and shutdown emissions using a 50-ton buffer
EPA examined units with SCR controls at coal fired units that operated during the 2021
ozone season with a seasonal average NOx rate under 0.08 lb/MMBtu. We identified 164 coal
units nationwide with SCRs operating in this way - during a time period for which there was not
a daily rate applied to those EGUs by the CSAPR program in effect at that time. As described in
section D. 1 of this TSD, for these units we found that only 0.9% (572 of 60,350 tons) of their
emissions occurred above the 0.14 lb/MMBtu emission rate that we are finalizing as the backstop
rate under this rule.67 These 572 tons were widely distributed across the 164 coal units, such that
only two units had over 30 tons of such emissions and none had over 50 tons of such emissions.
In 2021, there were 124 coal SCR units that had ozone season NOx rates above 0.08 lb/MMBtu.
121 of these units had a total of 18,629 tons of "excess" emissions (above the 0.14 lb/MMBTU
daily backstop rate), representing 23.0% of their total ozone-season emissions, ranging from
under 1 ton to 3,623 tons of excess emissions at the individual EGU level. Even if 50 tons were
excluded for each unit, there would still be 15,374 excess tons subject to a 3-for-l allowance
surrender ratio, and thus this relatively poor performance would still be disincentivized.
For these 164 units in 2021 that had emission rates below 0.08 lb/MMBTU, we also
examined their emissions in the 2022 ozone-season relative to the 0.14 daily backstop rate.
Again, we found that all units would not have issues when a 50-ton buffer was applied, with the
closest being a unit having 47.5 tons of "excess" emissions. See section VI.B of the preamble for
further discussion on this 50-ton buffer's incorporation into the final rule.
67 See the Excel workbook, Daily Backstop rate for existing SCRs - accommodating startup shutdown.xlsx for
details.
87
-------
E. Preliminary Environmental Justice Screening Analysis for EGUs
EPA conducted a screening analysis regarding potential environmental justice concerns
associated with emissions from EGUs.68 This analysis, discussed in this section, is distinct from
the EJ impacts analysis for the full rule in Chapter 7 of the RIA. EPA's EJ Technical Guidance69
states that: "A regulatory action may involve potential environmental justice concerns if it could:
(1) create new disproportionate impacts on minority populations, low-income populations, and/or
indigenous peoples; (2) exacerbate existing disproportionate impacts on minority populations,
low-income populations, and/or indigenous peoples; or (3) present opportunities to address
existing disproportionate impacts on minority populations, low-income populations, and/or
indigenous peoples through the action under development." In this TSD, EPA uses a screening
analysis to identify the potential for coal-fired EGUs to contribute to air pollution in areas with
potential EJ concerns.
This initial screening analysis examines two groups of coal-fired EGUs within the
geography: those EGUs with existing SCRs that will receive a backstop rate in 2024, and those
EGUs currently lacking SCRs that will receive a backstop rate by no later than 2030. It considers
whether each group demonstrates a greater potential to expose areas of potential EJ concern to
air pollution, relative to the national coal-fired EGU fleet. This screening-level analysis helped
EPA identify potential EJ concerns during the process of rule development, while subsequent
analysis presented in the RIA provides an evaluation of the distributional impacts of the
requirements finalized in this action. These two sets of analyses are distinct but complementary -
the screening analysis presented in this TSD evaluates the potential for environmental justice
concerns associated particularly with EGUs, and the environmental justice analyses presented in
the RIA estimate the ultimate impacts of the final rule.
Based on this screening analysis, both groups of EGUs demonstrated relatively high
potential to expose areas of potential EJ concern to further pollution. While this screening
analysis does not identify all potentially impacted downwind areas or quantify the downwind air
quality impacts, exposures, and potential health effects of these sources (the aggregate impact of
which is evaluated and discussed in the RIA), it does demonstrate that a relatively high potential
exists for the sources in these two groups to affect areas facing pre-existing disproportionate
susceptibility to exposure. Ultimately, all final rule determinations are justified under the EPA's
interstate transport framework for implementing the good neighbor provision for the 2015 ozone
NAAQS. This analysis indicates whether two groups of EGUs receiving backstop rates under the
final rule exhibit a relatively high potential to expose areas of potential EJ concern to further
pollution. An overview of the methodology is described below.
68 A potential EJ concern is defined as "the actual or potential lack of fair treatment or meaningful involvement of
minority populations, low-income populations, tribes, and indigenous peoples in the development, implementation
and enforcement of environmental laws, regulations and policies" (U.S. EPA, 2015). For analytic purposes, this
concept refers more specifically to "disproportionate impacts on minority populations, low-income populations,
and/or indigenous peoples that may exist prior to or that may be created by the proposed regulatory action" (U.S.
EPA, 2015).
69 U.S. Environmental Protection Agency (EPA), 2015. Guidance on Considering Environmental Justice During the
Development of Regulatory Actions, Docket ID: EPA-HQ-OAR-2021-0668-0087
88
-------
Methodology
The screening assessment in this TSD is based on EPA's peer-reviewed70 Power Plant
Screening Methodology (PPSM) and is carried out in three parts. First, to estimate which census
block groups have some potential to be exposed by emissions from each EGU, EPA used
NOAA's Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model to generate
forward trajectories for large coal-fired EGUs located in linked upwind states under this final
rule/71* A forward trajectory is a modeled parcel of air that moves forward (i.e., downwind) due
to winds and other meteorological factors. For each EGU, we used the HYSPLIT model to
simulate the downwind path of air parcels passing individual EGUs four times per day12:00
AM, 6:00 AM, 12:00 PM, and 6:00 PM (local standard time). For simplicity, EPA limited the
modeling to the period June 1 to August 31 (the period over which ozone concentrations are the
most likely to be elevated) for the years 2018 to 2020. In addition, EPA ran each trajectory for
only 24 hours. While the horizontal spatial resolution of the HYSPLIT model is based on 12-km
meteorology (in some respects limiting our ability to resolve spatial differences less than 12
kilometers), we ran model simulations over 1,100 times for each facility (4 runs a day across 92
ozone season days for 3 years). These trajectories reflect a modeled air parcel's coordinates and
elevation at every hour downwind of each EGU stack.72 For simplicity in this initial screen, we
limit our evaluation to coordinates of those trajectories that are within the contiguous United
States. While the 24-hour transport time used in this screening analysis identifies many of the
near-source areas that are most frequently impacted, emissions can travel over larger distances
and longer times and have substantive air quality impacts downwind, particularly when
contributions from individual sources from geographically distinct areas (each of which could be
relatively small) are aggregated to have a larger collective impact. Those collective air quality
impacts are analyzed using photochemical air quality modeling in this final rule's RIA.73
It is important to note that unlike the other models used to quantify downwind ozone
concentrations related to this rule, the HYSPLIT model is not a photochemical model - the
model does not include chemical transformation and does not provide estimates of downwind
pollutant concentrations.74 We are using HYSPLIT trajectories in a qualitative way to examine
70 The Peer Review Summary Report and EPA's Response will be available on EPA's website.
71 The HYSPLIT model determines the pathway of a modeled parcel of air using the NOAA's National Center for
Environmental Information North American Mesoscale Forecast System 12 kilometer forecast gridded meteorology
dataset (NAM-12) (https://www.ncei.noaa.gov/access/metadata/landing-page/bin/iso?id=gov.noaa.ncdc:C00630).
The horizontal resolution of the NAM-12 dataset is 12.191 kilometers, the vertical resolution is 26-layers from 1000
to 50 hecto Pascals, and the temporal resolution is 3-hours. (Stein etal.. 2015. Draxler and Hess. 1998).
72 The EPA uploaded into an Oracle database the HYSPLIT model output results for each forward trajectory,
including the originating EGU, the coordinates and elevation above ground for each hour of the trajectory, and the
trajectory elapsed time since release from the EGU. Within the Oracle database, the trajectory coordinates are used
to construct line segments that can be displayed within a geographic information system (GIS) software package to
overlay each modeled forward trajectory. The use of GIS allows a user to overlay HYSPLIT trajectories over census
blocks of interest display the likely path that EGU emissions may travel in the absence of atmospheric residence
time, chemical dispersion, or atmospheric deposition.
73 For example, in 2016, the EPA used HYSPLIT to examine 96-hour trajectories and altitudes up to 1,500 meters in
a corollary analysis to the source apportionment air quality modeling to corroborate upwind state-to-downwind
linkages. Details of this analysis can be found in Appendix E ("Back Trajectory Analysis of Transport Patterns") of
the Air Quality Modeling Technical Support Document for the Final Cross State Air Pollution Rule Update, which
is available at: https://www.epa.gov/sites/default/files/2017-05/documents/aq_modeling_tsd_final_csapr_update.pdf
74 The HYSPLIT model is run assuming the air parcel is neutrally buoyant and inert (i.e., without any dispersion,
deposition velocity, or atmospheric residence time constraints).
89
-------
the spatial patterns of pollutant transport from EGUs.75 The model results simply simulate the
path that the wind would carry a modeled parcel of air from the stack(s) of each EGU.76
Next, EPA screened each of the downwind areas that intersected with a HYSPLIT
trajectory to identify census block groups with potential environmental justice concerns. The
intent of this screen in this application is to generally identify areas of potentially higher
susceptibility to environmental factors such as air pollution. The screen was performed using
data from EPA's EJScreen, an environmental justice mapping and screening tool that includes 11
different environmental indicators and 6 different demographic indicators.77 For this analysis,
EPA evaluated the available information at the census block group level and calculated the
average of the following four socioeconomic indicators found in EJScreen: low-income,
unemployment rate, limited English speaking, and less than high school education. This average,
converted to a percentile, is similar to the supplemental demographic index in EJScreen.
However, unlike the supplemental demographic index, the index used in this screen does not
include low-life expectancy, which was not available at the time the assessment was conducted.
Note that the index used in this screen does not consider the exposure and vulnerability of
communities to multiple environmental burdens and their cumulative impacts, nor does it
quantify ozone-specific health risks. Rather, this aggregate indicator offers a general look at the
relative potential susceptibility of each block group to environmental exposure. For further
discussion of these indicators and the other indicators currently available in the EJScreen tool,
see the EJScreen Technical Documentation available at https://www.epa.gov/ejscreen.
In the final step of the screening analysis, EPA combined the results of the previous two
steps by layering the modeled HYSPLIT trajectories over census block groups and associated
combined socioeconomic values to produce a relative score for each EGU that considers the
population-weighted average combined socioeconomic value of the population that is potentially
affected by that EGU. This score is calculated for each EGU by identifying each block that
intersects with each trajectory originating from that EGU, summing the product of each block
group's combined socioeconomic value and its population, and then dividing that aggregated
total by the total population of all those intersected block groups. The resulting value is
converted to a percentile relative to the scores generated for the entire coal steam fleet. Higher
scores are assigned to EGUs with trajectories that intersect areas with higher population
weighted average combined socioeconomic values. The intent of this approach is to highlight
EGUs with the potential to affect areas where people who might be more vulnerable on average
might live. While these values are useful in a screening context to identify relative differences
across the EGU fleet, they do not provide any absolute or relative measure of exposure or risk.
EPA compared the relative scores across each group of EGUs to the fleet to determine
whether the groups exhibit a higher potential to expose areas of EJ concern than the fleet on
average. The scores for the fleet are distributed such that half of the EGUs score above the 50th
percentile, and half score below the 50th percentile. For each of the two groups of EGUs screened
in this analysis, more than half score higher than the 50th percentile. This distribution suggests
that each of these two groups demonstrates a higher relative potential to expose people who
75 In general, pollutant concentrations are the result of transport, dispersion, and transformation. As noted, this
analysis does not consider photochemical transformations.
76 Consistent with the intent of this screening analysis, this model provides information about where non-reactive
pollutants might initially travel from each EGU over a limited 24-hour period but does not quantify the magnitude of
impact at any given location.
77 U.S. Environmental Protection Agency (EPA), 2022. EJSCREEN Technical Documentation and EJScreen
Technical Document Appendix.
90
-------
might be more susceptible to air pollution, on average, compared with the EGU fleet assessed
across the entire contiguous United States.
Furthermore, EPA found that each group contained many individual EGUs with scores
above the 80th percentile (20 EGUs with existing SCRs and 9 EGUs lacking SCRs). This means
that these EGUs rank among the top 20% of EGUs in the country based on the scoring approach
described above. The 80th percentile threshold has been identified by the Agency in early
applications of EJScreen as an initial screening filter and has been used in past screening
experience to identify areas that may warrant further review, analysis, or outreach.78
The findings of this screening analysis suggest that this rule's imposition of a backstop
emissions rate on the EGUs included in these two groups may benefit areas of potential
environmental justice concern.
78 U.S. Environmental Protection Agency (EPA), 2022. EJSCREEN Technical Documentation.
91
-------
F. Assessment of the Effects of Ozone on Forest Health
Air pollution can impact the environment and affect ecological systems, leading to
changes in the ecological community and influencing the diversity, health, and vigor of
individual plant species. When ozone is present in the environment, it enters the plant through
the stomata and can interfere with carbon gain (photosynthesis) and allocation of carbon within
the plant, making fewer carbohydrates available for plant growth, reproduction, and/or yield
(2020 PA, section 4.3.1 and 2013 ISA, p. 1-15).79,:80 Ozone can impact a variety of commercial
and ecologically important species throughout the United States. These include forest tree and
herbaceous species as well as crops. Such effects at the plant scale can also be linked to an array
of effects at larger spatial scales and higher levels of biological organization, causing impacts to
ecosystem productivity, water cycling, ecosystem community composition and alteration of
below-ground biogeochemical cycles (2020 PA, section 4.3.1 and 2013 ISA, p. 1-15)..81 With the
data sets available to the Agency, here, we focus on selected forest tree species.
Assessing the impact of ozone on forests in the United States involves understanding the
risk to tree species from ozone concentrations in ambient air and accounting for the prevalence
of those species within the forest. Across several reviews of the ozone NAAQS and based on
longstanding body of scientific evidence, EPA has evaluated concentration-response functions
which relate ozone exposure to growth-related effects in order to consider the risk of ozone-
related growth impacts on forest trees (2020 PA, section 4.3.3, 2013 ISA and 2020 ISA). For this
purpose, EPA has focused on cumulative, concentration-weighted indices of exposure, such as
the W126-based cumulative exposure index (2020 PA, section 4.3.3.1.1, 2020 ISA, section
ES.3). Measured ozone concentrations in ambient air of the United States are used to calculate
the W126-based index as the annual maximum 3-month sum of daytime hourly weighted ozone
concentrations, averaged over 3 consecutive years. The sensitivity of different trees species
varies about the growth impacts of ozone exposure. Based on well-studied datasets relating
W126 index to reduced growth, exposure response functions have been developed for 11 tree
species (2020 PA, section 4.3.3.1.2 and Figure 4-3 and 2013 ISA, section 9.6). For these species,
the impact from ozone exposure has been determined by exposing seedlings to different levels of
ozone concentrations over one or more seasons (which have been summarized in terms of W126
index) and measuring reductions in growth (which are then summarized as "relative biomass
loss"). The magnitude of ozone impact on a forest community will depend on the prevalence of
different tree species of relatively more versus less sensitivity to ozone and the abundance in the
community.
79 U.S. EPA (2020). Policy Assessment for the Review of the Ozone National Ambient Air Quality Standards. U.S.
Environmental Protection Agency. Office of Air Quality Planning and Standards, Health and Environmental Impacts
Division, Research Triangle Park, NC. EPA-452/R-20-001.
Available https://www.epa.gov/sites/production/files/2020-05/documents/o3-final_pa-05-29-20compressed.pdf.
Docket ID: EPA-HQ-OAR-2021-0668-0164
80 U.S. EPA (2020). Integrated Science Assessment for Ozone and Related Photochemical Oxidants. U.S.
Environmental Protection Agency. Washington, DC. Office of Research 3A-35 and Development. EPA/600/R-
20/012. Available at: https://www.epa.gov/isa/integrated-science-assessment-isa-ozone-and-related-photochemical-
oxidants. Docket ID: EPA-HQ-OAR-2021-0668-0078
81 U.S. EPA (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants (Final Report).
Office of Research and Development, National Center for Environmental Assessment. Research Triangle Park, NC.
U.S. EPA. EPA-600/R-10-076F. February 2013. Available
at: https://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=P100KETF.txt. Docket ID: EPA-HQ-OAR-2021-0668-0075
92
-------
Some of the most common tree species in the eastern United States, where the benefits
from this rule will be most pronounced, are black cherry (Primus serotina), yellow or tulip-
poplar (Liriodendron tulipifera), sugar maple (Acer saccharum), eastern white pine (Pinus
strobus), Virginia Pine (Pinus virginiana), red maple (Acer rubrum), and quaking aspen
(Populus trenuloides). Since 2008, EPA has assessed the impact of ozone on these tree species
within the eastern United States for the period from 2000 to 2021 as part of the Clean Air Market
Division (CAMD) annual power sector programs progress report.82 Over this time period ozone
concentrations have improved substantially because of various emission reduction programs,
such as NBP, CAIR, CSAPR, CSAPR Update, Revised CSAPR Update, and other local and
mobile source reductions such as Tier2 and Tier3 rules. Past EPA assessments have shown that
the improvements in ozone are evident both for the regulatory metric, 3-year average of 4th
highest 8-hr daily maximum ozone concentration, and for the W126 metric.83 In forests where
certain sensitive species dominate the forest community, the estimates of relative biomass loss
from ozone have decreased substantially. However, for the period from 2019-2021, the eastern
United States still has areas where the species-weighted relative biomass loss estimated from
ozone for the seven common trees listed above is up to 11.0% (Figure F-l)84.
Ozone levels are expected to continue to decrease through 2026 based on model
projection of the impacts on ozone concentrations resulting from baseline "on the books" control
programs as well as by emission reductions under this rule. In a past analysis, as ozone declines,
estimates of relative biomass loss of these trees' species will also decline as they have from 2000
to 2021, indicating this proposed rule would result in increased protection of forest ecosystems
and resources. Under this rule, ozone concentrations are expected to decline faster than without
the rule (e.g., under the base case). While EPA does not have the tools to quantify the expected
level of improvement at this time, based on the previous relationships between ozone design
values and W126 determined as part of the review of the 2020 ozone NAAQS (2020 PA, section
4D.3.2.3 and Table 4D-12), W126 values are expected to improve as design values decrease. As
described in the preamble, the rule is expected to improve air quality as controls are optimized
and installed between 2023 and 2026.
82 See the annual progress reports for several recent years at
https://www3.epa.gov/airmarkets/progress/reports/index.html.
https://www3.epa.gov/airmarkets/progress/reports/pdfs/2020 fall report.pdf [Docket ID: Docket ID EPA-HQ-
OAR-2021 -0668-0170],
https ://www3. epa. gov/airmarkets/pro gress/reports/pdfs/2019 full report.pdf [Docket ID EPA-HQ-OAR-2021-
0668-0077], and https://www3 .epa. gov/airmarkets/progress/reports/pdfs/2018 fall report.pdf [Docket ID EPA-HQ-
OAR-2021-0668-0076]
83 U.S. EPA (2020). Policy Assessment for the Review of the Ozone National Ambient Air Quality Standards. U.S.
Environmental Protection Agency. Office of Air Quality Planning and Standards, Health and Environmental Impacts
Division, Research Triangle Park, NC. EPA-452/R-20-001.
Available https://www.epa.gov/sites/production/files/2020-05/documents/o3-final_pa-05-29-20compressed.pdf.
Docket ID: EPA-HQ-OAR-2021-0668-0164
84 To estimate the biomass loss for forest ecosystems across the eastern United States, the biomass loss for each of
the seven tree species was calculated using the three-month, 12-hour W126 exposure metric at each location, along
with each tree's individual C-R functions. The W126 exposure metric was calculated using monitored ozone data
from CASTNET and AQS sites, and a three-year average was used to minimize the effect of variations in
meteorological and soil moisture conditions. The biomass loss estimate for each species was then multiplied by its
prevalence in the forest community using the U.S. Department of Agriculture (USDA) Forest Service IV index of
tree abundance calculated from Forest Inventory and Analysis (FIA) measurements.
93
-------
The reductions from this rule are likely to provide further protection to natural forest
ecosystems by reducing the potential for ozone-related impacts.
Figure F-l: Estimated Black Cheny, Yellow Poplar, Sugar Maple, Eastern White Pine, Virginia
Pine, Red Maple, and Quaking Aspen Biomass Loss due to Ozone Exposure for 2019-2021.
Biomass (% Loss)
> 1%
1 to 3%
3 to 6%
6 to 9%
>9%
Max = 11%
See the annual progress reports at https://www3.epa.gov/airmarkets/progress/reports/index.html
94
-------
Appendix A: State Emission Budget Calculations and Engineering Analytics
See Excel workbook titled "Final Rule State Emission Budget Calculations and Engineering
Analytics" on EPA's website and in the docket for this rulemaking
95
-------
Appendix B: Description of Excel Spreadsheet Data Files Used in the AQAT
96
-------
EPA placed the Ozone=AQAT=Final.xlsx Excel workbook file in the docket that contains
all 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 emissions
"2026_EA" and "2023_EA" contain EGU emissions measurements and estimates for
each state. Various columns contain the 2021 OS measured emissions, and then
emissions for the engineering base along with each of the cost thresholds.
"NOx_non-CEM" has a breakdown of the point EGU non-CEM emission inventory
component used in the air quality modeling.
"non-EGU emiss" has the total anthropogenic emission reductions by state.
"2026_OS NOx" and "2023_OS NOx" each of these worksheets reconstructs total
anthropogenic emissions for the year, with various EGU emission inventories for
different cost threshold (including the engineering base case). The total anthropogenic
emissions can be found for each state in columns AG through AL. These totals are then
compared to the 2026gf emission level (column Y on the "2026_OS NOx" worksheet) to
make a fractional change in emissions in columns AV through BA. For 2026, Non-EGU
emissions change and fractional change) are found in columns BC through BF.
Air quality modeling design values and contributions from CAMx
"2023gf_All" contains average and maximum design values as well as state by state
contributions for the 2023gf base case modeled in CAMx.
"2026gf_All" contains average and maximum design values as well as state by state
contributions for the 2026gf base case modeled in CAMx.
"23gf_days.2026gf_cntl" contains average and maximum design values as well as state
by state contributions for the 2026gf final policy control case modeled in CAMx.
"2026fj_All_proposal_calib" contains average and maximum design values as well as
state by state contributions for the 20261] base case modeled in CAMx from proposal.
"2026fj_30NOx_proposal_calib" contains average and maximum design values as well
as state by state contributions for the case modeled in CAMx where EGU and non-EGU
emissions were reduced by 30% from proposal.
"receptor list" contains a list of the receptors whose average and/or maximum design
values are greater than or equal to 71 ppb in 2023 and 2026 in the final base case air
quality modeling.
Calibration factor creation and assessment
"primary calibration" includes the state-by-state and receptor-by-receptor calculation of
the calibration factors based on the 2026 base and 2026 air quality modeling where EGU
and non-EGU NOx emissions were reduced by 30% from proposal. The calibration
factors can be found in columns I through BF.
"alternative calibration" includes the state-by-state and receptor-by-receptor calculation
of the calibration factors based on the 2026 base and 2023 base contributions and
97
-------
emissions using the air quality modeling from the final rule. The calibration factors can
be found in columns I through BF.
Air quality estimates
" summary _DVs_2026" contains the average and maximum design value estimates
(rounded to two decimal places) for receptors that were nonattainment or maintenance in
the 2026 air quality modeling base case. Values using the Step 3 configuration and
primary calibration factor for each cost threshold level are shown starting in column L.
Under this approach, the maximum contribution to remaining receptors is shown in
columns AG through AR. Furthermore, a set of design value estimates are shown
(columns AT through BG) for the full geography configuration scenarios, where all states
that are originally linked in the base make adjustments to different cost levels.
Adjustment to cells II and 12 will result in interactive adjustment for the other worksheets
and will adjust the design values in columns I (the Step 3 configuration) and J (a "full
geography" configuration where the geography remains fixed) and the maximum
contributions to remaining linkages in column AE. The alternative calibration factor
simulation results are shown in columns BJ through CC. .
"summary_DVs_2023" contains the average and maximum design value estimates
(rounded to two decimal places) for receptors that were nonattainment or maintenance in
the 2023 air quality modeling base case. Values using the Step 3 configuration and
primary calibration factor for each cost threshold level are shown starting in column L.
Under this approach, the maximum contribution to remaining receptors is shown in
columns AF through AM. Furthermore, a set of design value estimates are shown
(columns AO through BE) for the full geography configuration scenarios, where all states
that are originally linked in the base make adjustments to different cost levels.
Adjustment to cells II and 12 will result in interactive adjustment for the other worksheets
and will adjust the design values in columns I (the Step 3 configuration) and J (a "full
geography" configuration where the geography remains fixed) and the maximum
contributions to remaining linkages in column AD.
"2023_scenario_primary"and "2026_scenario_primary" contains the average and
maximum design value estimates (as well as the individual state's air quality
contributions) for a particular scenario identified in cells H2 and H3 using the primary
AQAT calibration factor. The fractional emission changes for each of the linked and
unlinked states are shown in rows 2 and 3.
"2023_scenario_primary_links" and "2026_scenario_primary_links" contains the
individual state's air quality contributions for a particular receptors that remain at or
above 71 ppb for the scenario identified in cells II and 12.
"2026_full_geo_primary" and "2023_full_geo_primary" contains the average and
maximum design value estimates (as well as the individual state's air quality
contributions) for a particular scenario identified in cells H2 and H3. States that are
"linked" to any receptor in the geography are assigned the values in row 2 while
nonlinked states are assigned the values in row 3. Note that, only the "home" states, that
are linked to receptors in other states are assigned the "linked" state values in row 2.
"2026_scenario_alt" contains the average and maximum design value estimates (as well
as the individual state's air quality contributions) for a particular scenario identified in
cells H2 and H3. The fractional emission changes for each of the linked and unlinked
98
-------
states are shown in rows 2 and 3. This uses the "alternative" calibration factor based on
the 2023 air quality modeling, rather than the "primary" calibration factor based on the
proposal 2026 air quality modeling with the 30% reduction from EGUs and non-EGUs.
The individual scenario worksheets labeled:
o "2023_ step3_base",
o "2023_ step3_SCRopt",
o "2023_ step3_SCRoptwCC",
o "2023_ step3_SNCRopt",
o "2023_ step3_SNCRoptwCC",
o "2023_ step3_newSCR",
o "2026_ step3_base",
o "2026_ step3_SCRopt",
o "2026_ step3_SCRoptwCC",
o "2026_ step3_SNCRopt",
o "2026_ step3_SNCRoptwCC",
o "2026_ step3_newSCR",
o "2026_ step3_nonEGU",
o "2023_full_geo_base",
o "2023_full_geo_SCRopt",
o "2023_full_geo_SCRoptwCC",
o "2023_full_geo_SNCRopt",
o "2023_full_geo_SNCRoptwCC",
o "2023_full_geo_newSCR",
o "2026_full_geo_base",
o "2026_full_geo_SCRopt",
o "2026_full_geo_SCRoptwCC",
o "2026_full_geo_SNCRopt",
o "2026_full_geo_SNCRoptwCC",
o "2026_full_geo_newSCR",
o "2026_full_geo_nonEGU",
o "2026_full_geo_nonEGU_l st",
o "2023_ step3_base_wIRA",
o "2023_ step3_newSCR_wIRA",
o "2026_ step3_base_wIRA",
o "2026_ step3_newSCR_wIRA",
o "2026_ step3_nonEGU_wIRA",
o "2026_ step3_nonEGU_lst",
o "2026_AQ_Model_Policy_Control"
o "2026_ step3_base_alt",
o "2026_ step3_SCRopt_alt",
o "2026_ step3_SCRoptwCC_alt",
o "2026_ step3_SNCRopt_alt",
o "2026_ step3_SNCRoptwCC_alt",
o "2026_ step3_newSCR_alt",
o "2026_ step3_nonEGU_alt",
o "2026_AQ_Model_Policy_Contr_alt"
99
-------
o "2026_ step3_nonEGU_lst_alt",
o contain static air quality contributions and design value estimates for all monitors
for the particular year and scenario.
100
-------
Appendix C: IPM Runs Used in Transport Rule Significant Contribution Analysis
101
-------
Table Appendix C-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 C-l. IPM Runs Used in Transport Rule Significant Contribution Analysis
Run Name
[IPM File Name]
Description
Air Quality Modeling Base Case
[EPA620_TR_14c]
Model run used for the air quality modeling base case at steps 1
and 2, 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. It also includes key
fleet updates regarding new units, retired units, and control
retrofits that were known by Summer of 2022.
Illustrative Final Rule
[EPA620 TR21 ]
Model run used for 2026 air quality analysis of the Final rule.
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. It also includes key fleet updates
regarding new units, retired units, and control retrofits that were
known by Summer of 2022. Includes the illustrative final rule.
For details, please see Chapter 4 of the RIA.
Air Quality Modeling Base Case + IRA
[EPA620 TR19]
Model run used for the air quality modeling base case sensitivity
analysis in the presence of the IRA at steps 1 and 2, which
includes all information from the Air Quality Modeling Base
Case [EPA620_TR_14c] as well as parameters reflecting the key
provisions of the Inflation Reduction Act of 2022. For details
please see Appendix 4A of the RIA for this rulemaking.
102
-------
Appendix D: Description of the Analytic Results using the Primary Approach for the "Full
Geography" AQAT Configuration in 2026
As an alternative assessment, it was possible to estimate air quality concentrations in
what we call a "full geography" configuration at each downwind receptor using the ozone
AQAT. Here, we apply an approach where all states covered by the rule (regardless of whether
they are linked to a particular receptor or to a different receptor in the geography) have the same
cost threshold scenario "full geography" estimates.85 We also kept the states containing the
receptor (such as Colorado and Connecticut) that are not linked to receptors in other states at the
base case emission levels (rather than modulate them up to the same cost threshold level as the
linked upwind states). This allows us to assess the effects of the rule as a whole, and only the
rule, in that year on the receptors. In this assessment, we used the primary calibration factor for
all scenarios.
In general, assessed across the scenarios, the receptor difference between the Step 3
configuration and the "full geography" configuration are relatively small. For the "Full Step 3"
scenario in which non-EGU controls are applied, we observe a difference in status for the
Sheboygan County, Wisconsin receptor. In this scenario in the "Step 3" configuration, the
receptor remains maintenance, while in the "full geography" configuration, the receptor's
maintenance status is resolved to a very marginal degree, at 70.96 ppb. Even if EPA were to rely
on this "full geography scenario" for its overcontrol analysis (which we do not think appropriate
for reasons explained in section C of this TSD), it would not change the outcome of our
overcontrol finding, because 1) states still remain linked to one or more problematic receptors,
and/or 2) the penultimate increment of reductions (i.e., "Full Step 3 - EGU only" scenario)
shows the maintenance status persists - suggesting that an earlier stopping point would be
undercontrol. The average and maximum design values for 2026 are shown in Tables Appendix
D-l and Appendix D-2.
85 For the purposes of the AQAT "Full Geography" estimates, we included California as being included in the rule
and making any available reductions. See the preamble section I for how this state is treated in the rule.
103
-------
Table Appendix D-l. 2026 Average Ozone DVs (ppb) for Each Scenario Assessed using the
"Full Geography" AQAT Configuration.
Site
state
county
Engineering
Analysis
Base
SCR
Optimize
SCR
Optimize +
SOACC
SCR
Optimize +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit
("Full Step
3-EGU
only")
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
non-EGU
("Full Step
3")
40278011
Arizona
Yuma
69.87
69.86
69.86
69.86
69.86
69.84
69.80
80590006
Colorado
Jefferson
71.70
71.70
71.70
71.70
71.70
71.54
71.52
80590011
Colorado
Jefferson
72.06
72.06
72.06
72.06
72.06
71.81
71.78
80690011
Colorado
Larimer
69.84
69.84
69.84
69.84
69.84
69.69
69.67
90013007
Connecticut
Fairfield
71.25
71.17
71.15
71.16
71.14
70.89
70.52
90019003
Connecticut
Fairfield
71.58
71.51
71.49
71.50
71.48
71.25
70.93
350130021
New Mexico
Dona Ana
70.06
70.05
70.05
70.05
70.05
69.91
69.87
350130022
New Mexico
Dona Ana
69.17
69.16
69.16
69.16
69.16
69.01
68.97
350151005
New Mexico
Eddy
350250008
New Mexico
Lea
480391004
Texas
Brazoria
69.89
69.81
69.80
69.80
69.79
68.85
68.32
481671034
Texas
Galveston
71.29
71.19
71.16
71.18
71.15
69.95
69.17
482010024
Texas
Flarris
74.83
74.76
74.75
74.75
74.74
73.74
73.22
490110004
Utah
Davis
69.90
69.90
69.90
69.90
69.90
69.34
69.28
490353006
Utah
Salt Lake
70.50
70.49
70.49
70.49
70.49
69.96
69.91
490353013
Utah
Salt Lake
71.91
71.90
71.90
71.90
71.90
71.44
71.40
551170006
Wisconsin
Sheboygan
70.83
70.65
70.64
70.64
70.64
70.39
70.07
104
-------
Table Appendix D-2. 2026 Maximum Ozone DVs (ppb) for Each Scenario Assessed using
the "Full Geography" AQAT
Configural
tion.
Site
state
county
Engineering
Analysis
Base
SCR
Optimize
SCR
Optimize +
SOACC
SCR
Optimize +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit
("Full Step
3-EGU
only")
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
non-EGU
("Full Step
3")
40278011
Arizona
Yuma
71.47
71.46
71.46
71.46
71.46
71.44
71.40
80590006
Colorado
Jefferson
72.30
72.30
72.29
72.30
72.29
72.14
72.11
80590011
Colorado
Jefferson
72.66
72.66
72.66
72.66
72.66
72.40
72.38
80690011
Colorado
Larimer
71.04
71.04
71.04
71.04
71.04
70.88
70.87
90013007
Connecticut
Fairfield
72.06
71.97
71.95
71.96
71.94
71.69
71.31
90019003
Connecticut
Fairfield
71.78
71.71
71.69
71.70
71.68
71.45
71.13
350130021
New Mexico
Dona Ana
71.36
71.36
71.36
71.36
71.35
71.21
71.17
350130022
New Mexico
Dona Ana
71.77
71.76
71.76
71.76
71.76
71.62
71.57
350151005
New Mexico
Eddy
350250008
New Mexico
Lea
480391004
Texas
Brazoria
72.02
71.94
71.92
71.92
71.91
70.94
70.39
481671034
Texas
Galveston
72.51
72.41
72.38
72.39
72.36
71.15
70.36
482010024
Texas
Flarris
76.45
76.38
76.37
76.36
76.35
75.33
74.80
490110004
Utah
Davis
72.10
72.10
72.10
72.10
72.10
71.52
71.46
490353006
Utah
Salt Lake
72.10
72.09
72.09
72.09
72.09
71.55
71.49
490353013
Utah
Salt Lake
72.31
72.30
72.30
72.30
72.30
71.84
71.79
551170006
Wisconsin
Sheboygan
71.73
71.55
71.54
71.54
71.53
71.29
70.96
105
-------
Appendix E: Feasibility Assessment for Engineering Analytics Baseline
Similar to the Revised CSAPR Update Final Rule, EPA analyzed and confirmed that the
assumed power sector fleet operations in its baseline emissions and emission control stringency
control levels as implemented through estimated budgets were compatible with future load
requirements by verifying that new units in addition to the existing fleet would provide enough
generation, assuming technology-specific capacity factors, to replace the retiring generation that
is assumed to occur in years 2023 through 2027. EPA assessed generation adequacy specific to
the states covered under this action. EPA uses these observations to determine whether any
assumed replacement generation from the existing fleet is necessary to offset the announced
retirements and continue to satisfy electricity load. Additionally, EPA looked at whether the
combination of new units (both fossil and non-fossil) provide sufficient new generation to
replace retiring generation. In this case, EPA found that the new unit generation from fossil and
renewable generation would exceed the generation from retiring units in all three scenarios
examined, indicating that no further replacement generation from existing units is needed.
Moreover, EPA found the change in generation from the covered fossil units to be within the
observed historical trend.
EPA first identified the collective Engineering Analytics baseline heat input and
generation for 2023-2027 from the states covered in this action and compared it to
historical trends between 2017-2021 for these same states (Scenario 1). This illustrated
that the assumed heat input and generation from fleet turnover reflected in the
Engineering Analytics was well within with recent historical trends (see tables Appendix
E-l, and Appendix E-2 below).
EPA then compared the collective baseline heat input and generation from the states
covered in this action to a scenario where fossil generation remains at 2021 levels instead
of continuing to decline (Scenario 2).
Finally, EPA identified the 2022 Energy Information Administration's Annual Energy
Outlook (EIA AEO) annual growth projections from 2021 through 2027 total electricity
demand levels (0.7%) from its reference case and estimated an upper bound future year
scenario where covered fossil generation grew at levels matching this fleet-wide total
growth rate (Scenario 3).86
EPA's assessment illustrates the amount of generation in its Engineering Analytics
baseline, factoring in retirements and new fossil units, is more than sufficient to
accommodate all three scenarios.87 For instance, generation from fossil sources in these
states has dropped at an average rate of 2% per year between 2018 and 2021 (799 TWh to
750 TWh). However, EPA's assumed baseline generation from covered fossil sources for
the states reflects a rate of decline of 1.7% per year between 2023 and 2027. See Table
Appendix E-2.
86 Department of Energy, Annual Energy Outlook 2022. Available at
https://www.eia.gov/outlooks/aeo/data/browser/#/?id=62-AE02022&cases=ref2022&sourcekey=0
87 Based on historical trends, modeling, and company statements, EPA expects levels similar to scenario 1 and
scenario 2 to be most likely.
106
-------
EPA then identified new RE capacity under construction, testing, or in site prep by 2022.
For years beyond 2022, EPA also identified new RE capacity that was planned but with
regulatory approvals pending for years 2023 and beyond (as this capacity is unlikely to
have yet started construction).88
EPA calculated and added the RE generation values to the fossil baseline to estimate
future year generation in the state (see Table Appendix E-2). EPA used a capacity factor
of 42.7% for wind, 21.6% for solar, and 65% forNGCC.
Using these technology-specific capacity factors based on past performance and IPM
documentation, EPA anticipated over 36 TWh from new non-fossil generation already
under construction or being planned with regulatory approval received. This level of
expected new generation combined with the baseline generation from existing units
exceeds the expected load for the states under all three scenarios.89
Not only is the future baseline generation level assumed in EPA's engineering analysis
well within the recent historical fossil generation trend (See Table Appendix E-2) on its
own (which illustrates no need for replacement generation), but when added to the
amount of potential new generation from RE (over 36 TWh), exceeds the generation
assuming no change (scenario 2) and the upper bound analysis for future covered fossil
generation that assumes 0.7% growth from the existing fossil fleet (scenario 3). This
indicates that available capacity and generation assumed would serve load requirements
in this upper bound scenario.
Not included in the tables below nor in EPA's baseline, but listed in the latest EIA 860m
is even more planned NGCC combined cycle for years 2023 and 2024 that is pending
regulatory approval. Assuming some of this generation becomes available in the outer
years, that constitutes additional generation that further exceeds EPA's upper bound
generation levels below - further bolstering the observation that no replacement
generation from existing units needs to be assumed to fill generation from retiring units.
88 Department of Energy, EIA Form 860, Generator Form 3-1. 2020. Available at
https://www.eia.gov/electricity/data/eia860/
89 While EPA notes the baseline generation exceeds the covered fossil load in all three scenarios in Table F-3, EPA
anticipates scenarios 1 and 2 being more representative of likely covered fossil load based on historical trends,
future modeling, and utility resource plans.
107
-------
Table Appendix E-l: Heat Input (TBtu) Change Due to Fleet Turnover (Historical and
Future)
Values for 2018-2021 reflect reported data, while 2023-2026 reflects assumed heat input.
Region
2018
2019
2020
2021
2023
2024
2025
2026
2027
Alabama
388
352
327
323
313
313
313
310
309
Arkansas
220
203
160
193
193
193
193
191
191
Illinois
397
332
283
334
256
250
250
217
217
Indiana
479
404
371
411
356
330
330
302
302
Kentucky
354
316
270
303
301
301
296
296
296
Louisiana
312
318
282
281
271
271
269
269
268
Maryland
105
92
82
88
71
71
71
71
71
Michigan
349
326
283
309
273
258
258
217
217
Minnesota
144
132
108
129
129
108
108
108
94
Mississippi
218
211
224
190
184
180
180
180
180
Missouri
313
269
254
288
284
249
249
249
248
Nevada
108
98
100
103
103
103
103
94
89
New Jersey
152
146
119
120
112
112
112
112
112
New York
238
202
234
240
233
233
233
233
233
Ohio
405
402
395
400
364
338
338
338
338
Oklahoma
276
235
232
213
213
211
211
211
196
Pennsylvania
487
509
535
565
535
535
535
535
535
Texas
1,530
1,501
1,355
1,403
1,385
1,385
1,375
1,375
1,347
Utah
144
133
132
165
165
165
165
125
125
Virginia
251
249
261
215
203
195
194
194
194
West
Virginia
309
295
268
313
307
273
273
273
273
Wisconsin
222
192
195
221
221
221
213
185
151
Total
7,397
6,915
6,472
6,806
6,471
6,294
6,269
6,085
5,986
108
-------
Appendix E-2: Assumed Baseline OS Generation and Expected New Build Generation from
Covered Fossil Units (TWh)
2023
2024
2025
2026
2027
Scenario 1 - Generation Levels (with
continued pace of 2.7% decline)
707
687
669
650
632
Scenario 2 - Generation Levels (no change
from 2021)
747
747
747
747
747
Scenario 3 - Generation Levels (0.7% growth
from covered fossil)
758
763
768
774
779
Assumed Baseline Fossil Generation with
Reported Fossil Retirement and Reported New
Build
729
712
709
690
681
New Build (Non-Fossil)
59
87
90
93
107
Total Baseline Generation
788
798
799
784
788
109
-------
Appendix F: Preset State Emission Budgets
2023 Illustrative
Emission
2026 Preset
2027 Preset
2028 Preset
2029 Preset
Budgets Before
2024 Emission
2025 Emission
Emission
Emission
Emission
Emission
State
Prorating (tons)
Budgets (tons)
Budgets (tons)
Budgets (tons)
Budgets (tons)
Budgets (tons)
Budgets (tons)
Alabama
6,379
6,489
6,489
6,339
6,236
6,236
5,105
Arkansas
8,927
8,927
8,927
6,365
4,031
4,031
3,582
Illinois
7,474
7,325
7,325
5,889
5,363
4,555
4,050
Indiana
12,440
11,413
11,413
8,410
8,135
7,280
5,808
Kentucky
13,601
12,999
12,472
10,190
7,908
7,837
7,392
Louisiana
9,363
9,363
9,107
6,370
3,792
3,792
3,639
Maryland
1,206
1,206
1,206
842
842
842
842
Michigan
10,727
10,275
10,275
6,743
5,691
5,691
4,656
Minnesota
5,504
4,058
4,058
4,058
2,905
2,905
2,578
Mississippi
6,210
5,058
5,037
3,484
2,084
1,752
1,752
Missouri
12,598
11,116
11,116
9,248
7,329
7,329
7,329
Nevada
2,368
2,589
2,545
1,142
1,113
1,113
880
New Jersey
773
773
773
773
773
773
773
New York
3,912
3,912
3,912
3,650
3,388
3,388
3,388
Ohio
9,110
7,929
7,929
7,929
7,929
6,911
6,409
Oklahoma
10,271
9,384
9,376
6,631
3,917
3,917
3,917
Pennsylvania
8,138
8,138
8,138
7,512
7,158
7,158
4,828
Texas
40,134
40,134
38,542
31,123
23,009
21,623
20,635
Utah
15,755
15,917
15,917
6,258
2,593
2,593
2,593
Virginia
3,143
2,756
2,756
2,565
2,373
2,373
1,951
West Virginia
13,791
11,958
11,958
10,818
9,678
9,678
9,678
Wisconsin
6,295
6,295
5,988
4,990
3,416
3,416
3,416
110
-------
Appendix G: Comparison of CSAPR 2012 Budgets to Actual 2012 Emissions
This appendix provides a comparison of the budgets for the first year of the four original CSAPR
trading programs90 to actual emissions in the year when those budgets were originally scheduled
to be implemented. Specifically, it compares the state emissions budgets originally planned for
2012, which were not actually implemented until 2015 because of a judicial stay, to the
respective states' actual emissions for 2012.
This comparison shows that for all four trading programs, even without the implementation of
CSAPR, the affected region as a whole had 2012 emissions lower than the sum of the state
budgets that would have applied in that year had the programs' implementation not been
delayed. As shown in the tables below, in each of the four trading programs, the affected EGUs
in all of the covered states collectively emitted below the sum of the state budgets for the
program. Furthermore, the analysis shows that the affected EGUs in most covered states, even
without the rule in place, collectively emitted below their individual state budgets in 2012.
The collective 2012 emissions from a given state's affected EGUs exceeded the state's intended
2012 budget by more than what would later have been the state's variability limit in only four
instances: Illinois for annual NOx, Louisiana for OS NOx, and Missouri for both annual and OS
NOx.91 However, further analysis indicates a strong possibility that even these few exceedances
would not have occurred had the rule actually been in place. EGUs in Missouri, for example,
emitted 34,275 tons of NOx in the 2012 ozone season, exceeding their OS NOx budget of 22,788
tons by 11,487 tons (see table). During this same 2012 ozone season, New Madrid and Thomas
Hill, two facilities located in Missouri, emitted 16,449 tons of NOx. All five units at these two
facilities had SCRs. If these five units had run their SCRs so as to achieve average NOx
emissions rates of 0.12 lb/MMBtu, they would have emitted 12,297 fewer tons of NOx, and
Missouri's EGUs collectively would have emitted less than the state's 2012 OS NOx emissions
budget. As another example, Kincaid units 1 and 2 and Marion unit 4 in Illinois are all coal units
with SCR controls. In 2012, these units achieved annual average emissions rates of 0.40, 0.33,
and 0.23 lb/MMBtu, but in the 2009 ozone season the units ran their SCRs so as to achieve much
lower NOx emissions rates of 0.06, 0.06, and 0.12 lb/MMBtu, respectively. If these three units
had run their SCRs in 2012 so as to achieve the same average emissions rates the same units
achieved during the 2009 ozone season, their emissions would have dropped by 9,633 tons, very
close to the 9,812 tons by which Illinois EGUs' collective 2012 annual NOx emissions exceeded
the state's 2012 annual NOx budget.
90 Original CSAPR, 76 FR 48208 (August 8, 2011), including the changes to the budgets by the Final February and
Final June Revision Rules. 77 FR 10324 (Feb. 21, 2012); 77 FR 34830 (June 12, 2012).
91 The CSAPR trading programs include variability limits of 18% for SO2 and annual NOx emissions and 21% for
ozone season NOx emissions. The programs' assurance provisions generally require additional allowance surrenders
when a state's emissions exceed the state's emissions budget by more than the variability limit. While the assurance
provisions did not apply for the first two years of the CSAPR programs - so the 2012 exceedances shown in the
tables would not have triggered any extra allowance surrenders - the variability limits still serve as a useful metric
for the degree of state-level emissions variability that would generally be accommodated by the programs' design.
Ill
-------
Note: CSAPR Budgets shown here include the Final February Revisions Rule and Final June
Revisions Rule, where applicable.
Table Appendix G-l. Pre-stay 2012 Annual CSAPR SO2 Budgets, 2012 Annual SO2
Emissions, and Percent Emitted Difference Between the Budgets and Actual Emission in
2012 by State
% Emitted Above
or Below State
Pre-stay 2012
Annual S02
Sum of 2012
Budget (compare
to variability
limit of 18%
State
Budget (short
tons)
S02 Mass
(short tons)
starting two years
later)
Alabama
216,033
128,828
-40.4%
Georgia
158,527
101,072
-36.2%
Illinois
234,889
152,172
-35.2%
Indiana
290,762
273,628
-5.9%
Iowa
107,085
81,368
-24.0%
Kansas
41,980
32,947
-21.5%
Kentucky
232,662
186,180
-20.0%
Maryland
30,120
22,884
-24.0%
Michigan
229,303
194,702
-15.1%
Minnesota
41,981
25,286
-39.8%
Missouri
207,466
138,833
-33.1%
Nebraska
68,162
62,389
-8.5%
New Jersey
7,670
3,661
-52.3%
New York
36,296
17,637
-51.4%
North Carolina
136,881
58,295
-57.4%
Ohio
315,393
323,977
2.7%
Pennsylvania
278,651
249,716
-10.4%
South Carolina
96,633
44,973
-53.5%
Tennessee
148,150
66,258
-55.3%
Texas
294,471
339,160
15.2%
Virginia
70,820
31,488
-55.5%
West Virginia
146,174
83,265
-43.0%
Wisconsin
79,480
61,565
-22.5%
Total
3,469,589
2,680,283
-22.7%
S02 Group 1
2,551,802
1,945,627
-23.8%
S02 Group 2
917,787
734,656
-20.0%
112
-------
Table Appendix G-2. Pre-stay 2012 Annual CSAPR NOx Budgets, 2012 Annual NOx
Emissions, and Percent Emitted Difference Between the Budgets and Actual Emission in
2012 by State
% Emitted Above
or Below State
Pre-Stay 2012
Sum of 2012
Budget (compare
to variability
limit of 18%
State
Annual NOx
Budget
NOx Mass
(short tons)
starting two years
later)
Alabama
72,691
48,781
-32.9%
Georgia
62,010
34,892
-43.7%
Illinois
47,872
57,684
20.5%
Indiana
109,726
105,713
-3.7%
Iowa
38,335
34,827
-9.2%
Kansas
31,354
33,295
6.2%
Kentucky
85,086
80,299
-5.6%
Maryland
16,633
18,334
10.2%
Michigan
65,421
66,810
2.1%
Minnesota
29,572
24,353
-17.6%
Missouri
52,400
69,814
33.2%
Nebraska
30,039
26,906
-10.4%
New Jersey
8,218
6,300
-23.3%
New York
21,722
24,823
14.3%
North Carolina
50,587
51,057
0.9%
Ohio
95,468
84,281
-11.7%
Pennsylvania
119,986
132,094
10.1%
South Carolina
32,498
19,066
-41.3%
Tennessee
35,703
26,182
-26.7%
Texas
137,701
129,367
-6.1%
Virginia
33,242
26,219
-21.1%
West Virginia
59,472
52,783
-11.2%
Wisconsin
34,101
24,850
-27.1%
Total
1,269,837
1,178,729
-7.2%
113
-------
Table Appendix G-3. Pre-stay 2012 Ozone Season CSAPR NOx Budgets, 2012 Ozone
Season NOx Emissions, and Percent Emitted Difference Between the Budgets and Actual
Emission in 2012 by State
% Emitted Above
or Below State
Sum of 2012
Budget (compare
to variability
limit of 21%
State
Pre-stay 2012
OS NOx Budget
NOx OS Mass
(short tons)
starting two years
later)
Alabama
31,746
24,963
-21.4%
Arkansas
15,110
16,407
8.6%
Florida
28,644
30,764
7.4%
Georgia
27,944
14,957
-46.5%
Illinois
21,208
23,526
10.9%
Indiana
46,876
45,007
-4.0%
Iowa
16,532
15,550
-5.9%
Kentucky
36,167
35,982
-0.5%
Louisiana
18,115
22,084
21.9%
Maryland
7,179
8,298
15.6%
Michigan
28,041
30,161
7.6%
Mississippi
12,429
10,713
-13.8%
Missouri
22,788
34,275
50.4%
New Jersey
4,128
3,650
-11.6%
New York
10,369
12,364
19.2%
North Carolina
22,168
25,021
12.9%
Ohio
41,284
40,277
-2.4%
Oklahoma
36,567
31,242
-14.6%
Pennsylvania
52,201
62,916
20.5%
South Carolina
13,909
9,747
-29.9%
Tennessee
14,908
14,388
-3.5%
Texas
65,560
61,292
-6.5%
Virginia
14,452
13,106
-9.3%
West Virginia
25,283
24,314
-3.8%
Wisconsin
14,784
11,851
-19.8%
Total
628,392
622,855
-0.9%
114
-------
Appendix H: Sensitivity for order of emissions reductions from EGUs and nonEGUs
This appendix provides a comparison of the AQAT estimates using the Step 3 configuration
approach where we examine the effects of including EGU SCR retrofit emissions reductions
prior to or after the non-EGU emission reductions. The average and maximum design values in
2026 are shown in Table Appendix H-l. In essence, if non-EGU emission reductions occur prior
to EGU SNCR and SCR retrofits, all of the monitors (with the exception of Larimer, Colorado)
maintain the same status (either in nonattainment and/or maintenance) with their average or
maximum design values greater than or equal to 71 ppb. In the case of Larimer, the monitor is
estimated to have a maintenance issue if the non-EGU emission reductions occur with less-
stringent EGU emission reductions (consisting of optimizing existing SCR and SNCR and
installing SOA CC). Alternatively, if only EGU emission reductions occur (consisting of
optimizing existing SCR and SNCR, installing SOA CC, and retrofitting SCRs and/or SNCRs)
and not non-EGU emission reductions, the maximum design value drops below 71 ppb
indicating that it no longer would have a maintenance issue at this level of stringency. However,
no states have their last remaining linkage to this receptor. Consequently, the order of the EGU
and non-EGU emission reductions make no difference to the conclusions in this final rule about
overcontrol.
Table Appendix H-l. 2026 Average and Maximum Ozone DVs (ppb) for the AQAT Step 3
Scenarios Assessed for All Receptors.
Site
state
county
Engineering
Analysis
Base
(Avg. DV)
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit
(Avg. DV)
SCR
Optimize +
SOACC +
SNCR
Optimize +
non-EGU
(Avg. DV)
Engineering
Analysis
Base
(Max. DV)
SCR
Optimize +
SOACC +
SNCR
Optimize +
SCR/SNCR
Retrofit
(Max. DV)
SCR
Optimize +
SOACC +
SNCR
Optimize +
non-EGU
(Max. DV)
40278011
Arizona
Yuma
69.87
69.84
69.82
71.47
71.44
71.42
80590006
Colorado
Jefferson
71.70
71.36
71.67
72.30
71.95
72.26
80590011
Colorado
Jefferson
72.06
71.59
72.02
72.66
72.19
72.62
80690011
Colorado
Larimer
69.84
69.54
69.82
71.04
70.73
71.01
90013007
Connecticut
Fairfield
71.25
70.98
70.85
72.06
71.78
71.65
90019003
Connecticut
Fairfield
71.58
71.34
71.23
71.78
71.54
71.43
350130021
New Mexico
Dona Ana
70.06
69.89
70.01
71.36
71.19
71.32
350130022
New Mexico
Dona Ana
69.17
69.00
69.12
71.77
71.60
71.72
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
69.89
68.96
69.36
72.02
71.06
71.47
481671034
Texas
Galveston
71.29
70.02
70.43
72.51
71.22
71.64
482010024
Texas
Harris
74.83
73.86
74.29
76.45
75.46
75.90
490110004
Utah
Davis
69.90
69.34
69.84
72.10
71.52
72.04
490353006
Utah
Salt Lake
70.50
69.96
70.43
72.10
71.55
72.03
490353013
Utah
Salt Lake
71.91
71.45
71.86
72.31
71.84
72.26
551170006
Wisconsin
Sheboygan
70.83
70.51
70.41
71.73
71.41
71.31
115
-------
Appendix I: Figures Related to Preamble Section V and Section VI
Figure 1 to Section V.D.I - EGU Ozone Season NOx Reduction Potential in 22 Linked States
and Corresponding Total Reductions in Downwind Ozone Concentration at Nonattainment and
Maintenance Receptors for Each Cost Threshold Level Evaluated (2023)
0.06
Ozone improvement
MOx reduction
. , .potential. , . .
$500
$1,000
Cost per Ton
$1,500
18,000
16,000
14,000
12,000
10,000
8,000
6,000
4,000
2,000
$2,000
3
T3
1)
C£
C
O oo
on o
£ t
^ re
c ^
o c
CO QJ
(0 4->
<1) O
m q_
q_
$2,000 $4,000 $6,000 $8,000 $10,000 $12,000
Cost per Ton
116
-------
Figure 3 to Section V.D.I: : EGU Ozone Season NOx Reduction Potential in 19 Linked States
and Corresponding Total Reductions in Downwind Ozone Concentration at Nonattainment and
Maintenance Receptors for Each Cost Threshold Level Evaluated and Illustrative Evaluation of
Cost Thresholds beyond Identified Technology Breakpoints (2026)92
_Q
Q.
Q.
0,5
0.45
£ 0.4
o
03
Q.
I 0.25
>-
02
(TJ
a 0.15
i
< 0.1
< o
r
/
/
/
ft
[t
Ozone Improvement
- NOx Reduction Potential
t
90,000
80,000
70,000
60,000
50,000
40,000
30,000
20,000
10
0
E ~
c
o
V)
ra
$20,000 $40,000 $60,000 $80,000 $100,000 $120,000
92 For the evaluation of air quality impacts for the cost levels beyond our technology breakpoints (i.e., beyond
$11,000 per ton), the EPA relies on an average air quality per ton reduction factor derived from its AQAT analysis.
The EPA notes that these illustrative points (those beyond $11,000 per ton) reflect SCRs on steam units less than
100 MW and oil/gas steam units <150 tons per season, combustion control upgrade on combustion turbines, and
SCRs on combustion turbines > 100 MW respectively. Although, not shown above, EPA also observes that we
evaluated SCR on combined cycle unit and identified higher cost and higher resource intensity (i.e., higher ratio of
retrofit projects per ton reduced). These mitigation measures and costs are further discussed in the EGU NOx
Mitigation Strategies Final Rule TSD.
117
-------
Appendix J: Additional Sensitivity Examining the AQAT Calibration Factors
This appendix describes another sensitivity evaluating the primary and alternative
calibration factors used in the Step 3 configuration of AQAT. As described in section C.2, the
AQAT was calibrated using modeled ozone data from the proposed rule using a 2026 case where
EGUs and non-EGUs were reduced by 30%. We refer to this as the "primary calibration"
AQAT. As discussed in section C.4, we also evaluated an alternative set of calibration factors,
reflecting changes between the 2023 and 2026 base cases using AQ modeling from the final rule.
That analysis tends to confirm that the regulatory conclusions presented in the preamble are
robust to alternative approaches to calculating the air quality effects of the rule.
As described in the AQ Modeling TSD, EPA conducted photochemical air quality
modeling for the control scenario of the final rule (CAMx 2026 Final Rule Policy Control Case).
The emissions and the emission changes projected in this modeling and the emission reduction
fractions (relative to the 2026 photochemical modeling base case) are shown in Table Appendix
J-l. This additional photochemical air quality modeling offered us another opportunity to
evaluate the performance of AQAT.
As described in section C.2 and evaluated in C.4, each of the two calibrated AQATs
represent a different assessment of a linear relationship between emissions reductions and
changes in air quality based on the different emission levels and reductions from various sectors.
Using the primary and alternative calibration approaches, the average and maximum design
values from AQAT for the control scenario for the air quality modeling along with the CAMx
Final Rule Policy Case results are shown in Tables Appendix J-2 and J-3, respectively. The
CAMx Final Rule Policy Control Case design values, the AQAT design values using both
calibration factors, and the differences between the CAMx design value and each of the AQAT
values in the tables have been rounded to a hundredth of a ppb. For this scenario, the differences
in the average design values between the CAMx modeling and AQAT are moderate, with a
maximum value of 0.30 for the primary approach and 0.68 ppb for the alternative approach (both
for Davis, Utah receptor) (Table Appendix J-2). Most monitors had difference much lower than
those values.93 In response to comment, EPA performed further statistical evaluation of AQAT
consistency with CAMx. Averaged across all air quality monitors, the mean bias was -0.01 ppb
(-0.02%) and -0.03 ppb (-0.05%) using the primary and alternative calibration factors,
respectively.94 Focusing on the 2026 receptors that are at or above 71 ppb in the air quality
modeling base case (outside of California), the mean bias was -0.07 ppb (-0.1%) and -0.06 ppb (-
0.08%>) using the primary and alternative calibration factors, respectively. Collectively, these
comparisons against an independent photochemical air quality modeling simulation further
affirmed that a calibrated AQAT can create reasonable estimates of air quality concentrations for
each receptor.
In this assessment, all receptors had the same condition for the average design value (i.e.,
showing values either above or below the level of the NAAQS) regardless of the calibration
93 Additional evaluation values and metrics (e.g., mean bias and root mean square error) can be found in the
"AQAT ozone final.xlsx" results worksheets "2026_AQ_Model_Policy Control" and
"2026_AQ_Model_Policy_Contr_alt" using the primary and alternative calibration factors, respectively.
94 These metrics (and the others presented in the Excel file) compare favorably with those found by researchers. See,
for example, K.W. Appel, A.B. Gilliland, G. Sarwar, R.C. Gilliam. Evaluation of the Community Multiscale Air
Quality (CMAQ) model version 4.5: sensitivities impacting model performance: part I-ozone. Atmos. Environ., 41
(40) (2007), pp. 9603-9615, 10.1016/j.atmosenv.2007.08.044.
118
-------
factor utilized. When examining the maximum design values, in the CAMx Final Rule Policy
Control Case the maximum design value for the Larimer Colorado receptor dropped below 71
ppb, while it remained above 71 ppb for both the primary and alternative calibration approach.
For the Fairfield Connecticut receptor, the maximum design value remained above 71 ppb in the
CAMx Final Rule Policy Control Case and for the primary calibration approach but dropped
below 71 ppb (to 70.99 ppb) for the alternative calibration approach. These potential changes in
status for these two monitors (i.e., for Larimer Colorado or for Fairfield Connecticut) did not
affect the linkage status of any state. This assessment, again, indicates that the uncertainties
created by the nonlinearity of the ozone chemistry that is not accounted for by using the linear
calibration factors across the range of emission reductions assessed here and/or the difference in
spatial location and intensity of the sources and/or differences in the sectors usually do not affect
the conclusions about whether receptors are resolved and whether states continue to have
contributions above the linkage threshold to those receptors. In other words, the regulatory
conclusions set out in the preamble are robust to the particular calibration factors used in AQAT.
119
-------
Table Appendix J-l. The Total Anthropogenic NOx Emissions Used in the 2026 Base and Final Rule Policy
State
Modeled
Modeled
2026 NOx
2026 Base
2026
Reduction vs
Case NOx
Control
2026 Base
Emissions
Case NOx
Case
(final)
Emissions
(final)
Fractional
Reduction in
Emissions
Alabama
56,096
55,912
-0.003
Arizona
35,514
35,260
-0.007
Arkansas
44,639
37,449
-0.161
California
137,932
136,266
-0.012
Colorado
49,742
49,802
0.001
Connecticut
10,201
10,212
0.001
Delaware
6,492
6,494
0.000
District of Columbia
1,057
1,057
0.000
Florida
88,786
88,782
0.000
Georgia
61,626
61,674
0.001
Idaho
17,024
17,078
0.003
Illinois
84,913
82,914
-0.024
Indiana
70,963
68,035
-0.041
Iowa
46,523
46,862
0.007
Kansas
56,844
57,227
0.007
Kentucky
49,829
43,968
-0.118
Louisiana
98,585
87,536
-0.112
Maine
13,617
13,617
0.000
Maryland
23,023
22,872
-0.007
Massachusetts
28,194
28,197
0.000
Michigan
69,697
65,956
-0.054
Minnesota
55,848
54,685
-0.021
Mississippi
32,407
29,740
-0.082
Missouri
68,407
61,594
-0.100
Montana
25,336
25,338
0.000
Nebraska
42,355
42,407
0.001
Nevada
18,043
18,014
-0.002
New Hampshire
6,830
6,839
0.001
New Jersey
31,368
31,053
-0.010
New Mexico
70,923
70,933
0.000
New York
64,616
63,446
-0.018
North Carolina
55,518
55,889
0.007
North Dakota
69,173
69,262
0.001
Ohio
75,421
70,764
-0.062
Oklahoma
77,225
69,864
-0.095
Oregon
28,271
28,271
0.000
Pennsylvania
87,453
85,354
-0.024
Rhode Island
4,172
4,164
-0.002
South Carolina
40,161
40,332
0.004
South Dakota
12,372
12,392
0.002
Tennessee
46,637
46,648
0.000
Texas
299,134
293,557
-0.019
Utah
31,387
26,472
-0.157
Vermont
3,447
3,448
0.000
Virginia
45,636
44,741
-0.020
Washington
46,143
46,143
0.000
West Virginia
45,466
42,167
-0.073
Wisconsin
41,877
41,995
0.003
Wyoming
35,517
36,054
0.015
Tribal Data
5,522
4,200
-0.239
120
-------
Table Appendix J-2. 2026 Average Ozone DVs (ppb) for the CAMx AQ Modeling of the
Final Rule Policy Control Case Using the Two Calibration Factors.
Site
state
county
AQ
Modeling
AQAT
Estimate
using
Primary
Calibration
Factor
AQAT
Estimate
using
Alternative
Calibration
Factor
Delta AQ
between
Primary
Calibration
Approach
and AQ
Modeling
Delta AQ
between
Alternative
Calibration
Approach
and AQ
Modeling
40278011
Arizona
Yuma
69.80
69.84
69.82
-0.04
-0.02
80590006
Colorado
Jefferson
71.80
71.86
71.88
-0.06
-0.08
80590011
Colorado
Jefferson
72.30
72.18
72.24
0.12
0.06
80690011
Colorado
Larimer
69.70
69.87
69.91
-0.17
-0.21
90013007
Connecticut
Fairfield
70.40
70.39
70.41
0.01
-0.01
90019003
Connecticut
Fairfield
70.80
70.85
70.79
-0.05
0.01
350130021
New Mexico
Dona Ana
69.90
69.82
69.79
0.08
0.11
350130022
New Mexico
Dona Ana
68.90
68.92
68.92
-0.02
-0.02
350151005
New Mexico
Eddy
69.10
69.05
0.05
350250008
New Mexico
Lea
69.20
69.17
0.03
480391004
Texas
Brazoria
68.20
68.25
68.54
-0.05
-0.34
481671034
Texas
Galveston
69.20
69.01
69.63
0.19
-0.43
482010024
Texas
Flarris
73.20
73.13
73.38
0.07
-0.18
490110004
Utah
Davis
69.70
69.40
69.02
0.30
0.68
490353006
Utah
Salt Lake
70.30
70.02
69.62
0.28
0.68
490353013
Utah
Salt Lake
71.70
71.49
71.33
0.21
0.37
551170006
Wisconsin
Sheboygan
70.50
70.25
70.26
0.25
0.24
121
-------
Table Appendix J-3. 2026 Maximum Ozone DVs (ppb) for the CAMx AQ Modeling Final
Rule Policy Control Scenario Using the Two Calibration Factors.
Site
state
county
CAMx
Modeling
AQAT
Estimate
using
Primary
Calibration
Factor
AQAT
Estimate
using
Atlernative
Calibration
Factor
Delta AQ
between
Primary
Calibration
Approach
and CAMx
Modeling
Delta AQ
between
Alternative
Calibration
Approach
and CAMx
Modeling
40278011
Arizona
Yuma
71.40
71.44
71.42
-0.04
-0.02
80590006
Colorado
Jefferson
72.50
72.46
72.47
0.04
0.03
80590011
Colorado
Jefferson
72.90
72.78
72.84
0.12
0.06
80690011
Colorado
Larimer
70.90
71.07
71.10
-0.17
-0.20
90013007
Connecticut
Fairfield
71.30
71.19
71.20
0.11
0.10
90019003
Connecticut
Fairfield
71.10
71.05
70.99
0.05
0.11
350130021
New Mexico
Dona Ana
71.10
71.12
71.08
-0.02
0.02
350130022
New Mexico
Dona Ana
71.50
71.52
71.52
-0.02
-0.02
350151005
New Mexico
Eddy
73.30
73.35
-0.05
350250008
New Mexico
Lea
71.60
71.57
0.03
480391004
Texas
Brazoria
70.30
70.32
70.62
-0.02
-0.32
481671034
Texas
Galveston
70.40
70.19
70.82
0.21
-0.42
482010024
Texas
Flarris
74.80
74.72
74.97
0.08
-0.17
490110004
Utah
Davis
71.80
71.58
71.19
0.22
0.61
490353006
Utah
Salt Lake
71.80
71.61
71.20
0.19
0.60
490353013
Utah
Salt Lake
72.20
71.89
71.72
0.31
0.48
551170006
Wisconsin
Sheboygan
71.40
71.14
71.15
0.26
0.25
122
-------
Appendix K: Additional AQAT sensitivity including the IRA
As described in preamble section V.D, we assessed the effects of including the Inflation
Reduction Act (IRA) on the emissions projections. EPA then assessed the effects of these
potential IRA-related emissions changes on air quality using AQAT to verify it did not alter
EPA's geographic or overcontrol findings. EPA evaluated air quality contributions and receptor
status for the base case in 2023, for the base case in 2026, the "Full Step 3" scenario in 2026, and
the "Full Step 3 - EGU only" scenario in 2026 using the Step 3 configuration of AQAT with the
primary calibration factor. These are the four scenarios that are most relevant for the construction
of the policy. For these scenarios, EPA accounted for the effects of the IRA by calculating the
emission differences (i.e., deltas) for each state between the IPM case without the IRA and then
with the same IPM case but including the IRA. It then applied this delta to the respective AQAT
scenario. See the worksheet "IRAcases" in the ozone AQAT final.xlsx to see the calculations
of how these emissions differences were applied. In short, we took the difference in expected
emissions (an IPM case with and without the IRA). To create the engineering analysis base
including the IRA, we subtracted the state emission deltas (from the IPM base case with and
without the IRA) from the engineering analysis base emissions for that state. For the penultimate
and final cost threshold cases (i.e., "Full Step 3 - EGU only" and "Full Step 3" Scenarios,
respectively), the emission difference was similarly obtained by identifying the difference
between the IPM Final Policy Case with and without the IRA.
The air quality contributions for the four scenarios incorporating the IRA are shown in
Table Appendix K-l. Comparing these values with the respective policy case (without the IRA)
from Tables C-l 1 and C-12, we observe that while there are minor differences in contributions
there are no differences in which states remain linked in 2023 or 2026. Comparing the 2023
average and maximum design values for the base cases with and without IRA using Tables C-7,
C-8, and Appendix K-2, we can observe that there are no changes in receptor status.95 Next,
comparing the 2026 average and maximum design values for the base cases, from the "Full Step
3 - EGU only," or from the "Full Step 3" cases with and without the IRA using Tables C-9, C-
10, and Appendix K-3 and Appendix K-4, we can observe that, again, there are no changes in
receptor status (i.e., the receptor is consistently above or below 71 ppb comparing the with- and
without-IRA cases). Consequently, EPA concludes that even factoring in the projected effects of
the IRA the conclusions in the final rule regarding geographic scope and overcontrol remain
valid.
95 We also examined the hypothetical Step 3 case for 2023 where SCRs are retrofit, both with and without RIA (i.e.,
the 2023 "Full Step 3 - EGU only" scenario). In this case, we see no changes in linkage status. All states continue to
remain linked at or above 1% of the NAAQS to a remaining nonattainment or maintenance receptor.
123
-------
Table Appendix K-l. 2023 and 2026 Maximum Air Quality Contribution (ppb) to a
Remaining Receptor.96
State
2023 Base
2026 Base
2026 "Full Step
2026 "Full
Case w/
Case w/ IRA
3 - EGU only"
Step 3"
IRA
Case w/ IRA
Casew/ IRA
Alabama
0.77
Arkansas
1.18
1.12
1.01
0.57
California
6.27
6.10
6.09
6.05
Illinois
19.08
13.60
13.59
13.56
Indiana
9.90
8.31
8.22
8.05
Kentucky
0.86
0.82
0.75
0.72
Louisiana
9.68
9.64
9.29
4.30
Maryland
1.31
1.09
1.09
1.09
Michigan
1.60
1.46
1.46
1.45
Minnesota
0.85
Mississippi
1.41
1.32
1.21
0.35
Missouri
1.94
1.78
1.59
1.55
Nevada
1.06
0.90
0.90
0.90
New Jersey
8.37
8.09
8.10
8.11
New York
16.12
12.68
12.66
12.64
Ohio
2.04
1.90
1.90
1.85
Oklahoma
1.02
0.77
0.72
0.61
Pennsylvania
5.93
5.66
5.61
5.52
Texas
4.75
4.45
4.34
4.31
Utah
1.29
1.07
0.90
0.89
Virginia
1.83
1.14
1.13
1.10
West Virginia
1.51
1.35
1.28
1.24
Wisconsin
2.88
96 Values greater than or equal to 0.70 ppb indicate the state remains linked to a remaining downwind receptor.
124
-------
Table Appendix K-2. 2023 Average and Maximum Ozone DVs (ppb) for the Engineering
for All Receptors.
Site
state
county
2023
Engineering
Analysis
Base Case
(Avg. DV)
2023
Engineering
Analysis
Base Case
w/IRA (Avg.
DV)
2023
Engineering
Analysis
Base Case
(Max. DV)
2023
Engineering
Analysis
Base Case
w/IRA
(Max. DV)
40278011
Arizona
Yuma
70.36
70.36
72.05
72.06
80350004
Colorado
Douglas
71.12
71.17
71.71
71.77
80590006
Colorado
Jefferson
72.63
72.67
73.32
73.37
80590011
Colorado
Jefferson
73.29
73.35
73.89
73.95
80690011
Colorado
Larimer
70.79
70.83
71.99
72.02
90010017
Connecticut
Fairfield
71.62
71.57
72.22
72.17
90013007
Connecticut
Fairfield
72.99
72.95
73.89
73.85
90019003
Connecticut
Fairfield
73.32
73.28
73.62
73.58
90099002
Connecticut
New Haven
70.61
70.59
72.71
72.69
170310001
Illinois
Cook
68.13
68.14
71.82
71.83
170314201
Illinois
Cook
67.92
67.93
71.41
71.42
170317002
Illinois
Cook
68.47
68.47
71.27
71.27
350130021
New Mexico
Dona Ana
70.83
70.83
72.13
72.13
350130022
New Mexico
Dona Ana
69.73
69.73
72.43
72.43
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
70.59
70.56
72.69
72.67
481210034
Texas
Denton
69.93
69.91
71.73
71.72
481410037
Texas
El Paso
69.82
69.82
71.43
71.42
481671034
Texas
Galveston
71.82
71.79
73.13
73.09
482010024
Texas
Harris
75.33
75.30
76.93
76.91
482010055
Texas
Harris
71.19
71.16
72.20
72.17
482011034
Texas
Harris
70.32
70.29
71.52
71.50
482011035
Texas
Harris
68.01
67.98
71.52
71.49
490110004
Utah
Davis
71.88
71.90
74.08
74.10
490353006
Utah
Salt Lake
72.48
72.50
74.07
74.10
490353013
Utah
Salt Lake
73.21
73.23
73.71
73.73
550590019
Wisconsin
Kenosha
70.75
70.75
71.65
71.65
551010020
Wisconsin
Racine
69.59
69.61
71.39
71.40
551170006
Wisconsin
Sheboygan
72.64
72.65
73.54
73.55
125
-------
Table Appendix K-3. 2026 Average Ozone DVs (ppb) for the Base, "Full Step 3 - EGU
only", and "Full Step 3" Cases with and without the IRA Assessed Using the Ozone AQAT
for All Receptors.
Site
state
county
2026
Engineering
Analysis
Base Case
(Avg. DV)
2026
Engineering
Analysis
Base Case
w/ IRA
(Avg. DV)
2026 "Full
Step 3 -
EGU only"
Case (Avg.
DV)
2026 "Full
Step 3 -
EGU only"
Case w/ IRA
(Avg. DV)
2026 "Full
Step 3"
Case (Avg.
DV)
2026 "Full
Step 3"
Case w/ IRA
(Avg. DV)
40278011
Arizona
Yuma
69.87
69.89
69.84
69.85
69.80
69.81
80590006
Colorado
Jefferson
71.70
71.73
71.36
71.40
71.34
71.38
80590011
Colorado
Jefferson
72.06
72.10
71.59
71.64
71.57
71.62
80690011
Colorado
Larimer
69.84
69.87
69.54
69.58
69.53
69.56
90013007
Connecticut
Fairfield
71.25
71.18
70.98
70.95
70.66
70.63
90019003
Connecticut
Fairfield
71.58
71.51
71.34
71.31
71.06
71.03
350130021
New Mexico
Dona Ana
70.06
70.08
69.89
69.91
69.86
69.88
350130022
New Mexico
Dona Ana
69.17
69.19
69.00
69.02
68.96
68.98
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
69.89
69.90
68.96
69.01
68.50
68.54
481671034
Texas
Galveston
71.29
71.28
70.02
70.07
69.28
69.33
482010024
Texas
Flarris
74.83
74.85
73.86
73.91
73.39
73.45
490110004
Utah
Davis
69.90
69.91
69.34
69.39
69.28
69.33
490353006
Utah
Salt Lake
70.50
70.50
69.96
70.01
69.91
69.95
490353013
Utah
Salt Lake
71.91
71.92
71.45
71.48
71.40
71.44
551170006
Wisconsin
Sheboygan
70.83
70.80
70.51
70.51
70.27
70.27
Table Appendix K-4. 2026 Maximum Ozone DVs (ppb) for the Base, "Full Step 3 - EGU
only", and "Full Step 3" Cases with and without the IRA Assessed Using the Ozone AQAT
for All Receptors.
Site
state
county
2026
Engineering
Analysis
Base Case
(Max. DV)
2026
Engineering
Analysis
Base Case
w/ IRA
(Max. DV)
2026 "Full
Step 3 -
EGU only"
Case (Max.
DV)
2026 "Full
Step 3 -
EGU only"
Case w/ IRA
(Max. DV)
2026 "Full
Step 3"
Case (Max.
DV)
2026 "Full
Step 3"
Case w/ IRA
(Max. DV)
40278011
Arizona
Yuma
71.47
71.49
71.44
71.45
71.40
71.41
80590006
Colorado
Jefferson
72.30
72.33
71.95
71.99
71.93
71.97
80590011
Colorado
Jefferson
72.66
72.70
72.19
72.23
72.16
72.21
80690011
Colorado
Larimer
71.04
71.07
70.73
70.77
70.72
70.76
90013007
Connecticut
Fairfield
72.06
71.99
71.78
71.75
71.46
71.42
90019003
Connecticut
Fairfield
71.78
71.71
71.54
71.51
71.26
71.23
350130021
New Mexico
Dona Ana
71.36
71.38
71.19
71.21
71.16
71.18
350130022
New Mexico
Dona Ana
71.77
71.79
71.60
71.62
71.56
71.58
350151005
New Mexico
Eddy
0.00
0.00
0.00
0.00
0.00
0.00
350250008
New Mexico
Lea
0.00
0.00
0.00
0.00
0.00
0.00
480391004
Texas
Brazoria
72.02
72.02
71.06
71.10
70.58
70.63
481671034
Texas
Galveston
72.51
72.50
71.22
71.27
70.47
70.52
482010024
Texas
Flarris
76.45
76.47
75.46
75.51
74.98
75.04
490110004
Utah
Davis
72.10
72.11
71.52
71.57
71.46
71.51
490353006
Utah
Salt Lake
72.10
72.10
71.55
71.60
71.50
71.54
490353013
Utah
Salt Lake
72.31
72.32
71.84
71.88
71.80
71.84
551170006
Wisconsin
Sheboygan
71.73
71.70
71.41
71.41
71.17
71.17
126
------- |