Technical Support Document (TSD)
for the Proposed Federal Implementation Plan Addressing Regional Ozone Transport for the
2015 Ozone National Ambient Air Quality Standard
Docket ID No. EPA-HQ-OAR-2021-0668
Ozone Transport Policy Analysis
Proposed Rule TSD
U.S. Environmental Protection Agency
Office of Air and Radiation
February 2022
1
-------
The analysis presented in this document supports the EPA's proposed Federal
Implementation Plan Addressing Regional Ozone Transport for the 2015 Ozone National
Ambient Air Quality Standard (Cross-State Air Pollution Rule for the 2015 Ozone NAAQS).
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 emission budgets (i.e., limits on emissions) and the resulting effects on air
quality primarily focused on EGUs. The analysis is described in Sections VI and VII of the
preamble to the rule. This TSD also broadly 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 some limited analysis for
overcontrol of the non-EGU policy scenarios. 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 Integrated Planning Model (IPM) in Step 3
Assessment of Significant Contribution to Nonattainment and Interference with
Maintenance 3
B. Calculating Step 4 EGU Emission Budgets from Historical Data and IPM
Analysis 7
1. Calculating 2023-2026 Engineering Baseline Heat Input 8
2. Estimating impacts of combustion and post combustion controls on state-level
emission rates 10
3. Estimating Emission Reduction Potential from Generation Shifting 13
4. Variability Limits 28
5. Calculating Dynamic Budgets Starting in 2025 28
C. Analysis of Air Quality Responses to Emission Changes Using an Ozone Air
Quality Assessment Tool (AQAT) 31
1. Introduction 32
2. Details on the construction of the ozone AQAT for this proposed rule 34
3. Description of the analytic results 47
4. Comparison between the air quality assessment tool estimates 58
D. Selection of Short-term Rate Limits 60
1. Observations of fleet operation for well-controlled units 60
2. Creating "comparably stringent" emission rates using the 2014 1-hour S02
concepts 62
E. Preliminary Environmental Justice Screening Analysis 67
F. Assessment of the Effects of Ozone on Forest Health 71
Appendix A: State Emission Budget Calculations and Engineering Analytics 74
Appendix B: Description of Excel Spreadsheet Data Files Used in the AQAT 75
Appendix C: IPM Runs Used in Transport Rule Significant Contribution Analysis 80
Appendix D: Generation Shifting Analysis 82
Appendix E: Feasibility Assessment for Engineering Analytics Baseline 83
Appendix F: State Emission Budgets and Variability Limits 87
Appendix G: Figures Related to Preamble Section VI and Section VII 88
2
-------
A. Using Engineering Analytics and Integrated Planning Model (IPM) in 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 sections B and C 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 D 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 29 receptors from the 9 home
states, excluding California and its receptors, and the 26 upwind2 that were linked to downwind
receptors 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 proposed rule, this TSD's references
to "affected states" or "states covered by this rule" in EGU-related material does not include
California.3
Table A-l. Upwind States Evaluated in the Multi-factor Test
Alabama+
Nevada
Arkansas
New Jersey
California*
New York
Delaware+
Ohio
1 See the EGU NOx Mitigation Strategies Proposed Rule TSD.
2 Note that 7 of the 26 upwind states are also states with non-attainment or maintenance receptors, or "home states."
Colorado and Connecticut are home states, but do not significantly contribute to a downwind state non-attainment or
maintenance receptor.
3 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 (in non-California states)
and non-EGU reductions elsewhere and in California and are shown in the accompanying AQAT file. See
Ozone AQAT Proposal.xlsx for results.
3
-------
Illinois
Oklahoma
Indiana
Pennsylvania
Kentucky
Tennessee+
Louisiana
Texas
Maryland
Utah"
Michigan
Virginia
Minnesota
West Virginia
Mississippi
Wisconsin
Missouri
Wyoming
*California EGUs are not covered by this rule.
+Linkages for Alabama, Delaware, and Tennessee are resolved before 2026. Therefore, those states have
a lower level of emission control stringency compared to states that continue to be linked in 2026.
A In recognition of Utah's lack of state jurisdiction over an existing EGU in the Uintah and Ouray
Reservation, that reservation was evaluated separately from the rest of the land within Utah's borders.
Similar to the CSAPR Update and the Revised CSAPR Update, EPA relied on adjusted
historical data (engineering analytics) and its power sector modeling platform using IPM 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 and used along with IPM 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). 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 analytics tool uses the latest historical representative emissions and
operating data reported under 40 CFR part 75 by covered units (which were 2021 ozone-season
data at the time of this analysis). 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
C. Calculating Budgets from Historical Data and IPM Analysis for a detailed description of the
engineering analytics tool.
IPM is a multiregional, dynamic, deterministic linear programming model of the U.S.
electric power sector that EPA uses to analyze cost and emissions impacts of environmental
policies.4 All IPM cases for this rule included representation of the Title IV SO2 cap and trade
program; the NOx SIP Call; the CSAPR and CSAPR Update regional cap and trade programs;
consent decrees and settlements; and state and federal rules as listed in the IPM documentation
referenced above.
To quantify the emission reduction potential of generation shifting correlated to each
control stringency representing different pollution control technologies, EPA conducted a set of
modeling runs referred to as the "Cost Threshold Cases." EPA first adjusted the model to reflect
4 See "Documentation for EPA's Power Sector Modeling Platform v6 using Summer 2021 Reference Case".
Available at https://www. https://www.epa.gov/airmarkets/documentation-epas-power-sector-modeling-platform-
v6-summer-2021 -reference-case
4
-------
the relevant control technologies being considered (referred to as the "Adjusted Base Case" for
each stringency level) and then imposed a dollar per ton price constraint (e.g., $l,800/ton, or
$10,000/ton) to project the additional reductions expected from generation shifting
commensurate with the estimated representative technology cost at that control stringency level.
For the "Cost Threshold" IPM runs, the EPA designed a series of IPM runs that imposed
increasing cost thresholds representing uniform levels of NOx controls and tabulated those
projected emissions for each state at each cost level. These tabulations, when combined with
adjusted historical data, are described as "cost curves."5 The cost curves report the remaining
emissions at each cost threshold for each state after EGUs have made emission reductions that
are available up to the particular cost threshold analyzed, inclusive of the pollution reduction
technologies available in that control stringency as well as emission reductions from generation
shifting at a commensurate representative cost per ton.
In each Cost Threshold run, the EPA applied the applicable ozone-season cost level to all
fossil-fuel-fired EGUs with a capacity greater than 25 MW in all states, though only the
estimates for the nonattainment and maintenance receptors, the "home states" for those receptors,
and the affected states with proposed EGU reductions affect the results in step 3. As described in
the EGU NOx Mitigation Strategies Proposed Rule TSD, because of the time required to build
advanced pollution controls, the model was prevented from building any new post-combustion
controls, such as SCR or SNCR, before the 2025 run year,6 in response to the cost thresholds.7
Similarly, the model was not enabled to build incremental new units in that time frame. In
response to the ozone-season NOx cost, the modeling assumes turning on idled existing SCR and
SNCR, optimization of existing SCR, adding or upgrading NOx combustion controls (such as
state-of-the-art low NOx burners (LNB)) in 2023/2024, and projects shifting generation to lower-
NOx emitting EGUs. In this TSD, we sometimes refer to state-of-the-art combustion controls, or
SOA CC, generally, as combustion controls. For details on which measures are endogenously
modeled within IPM and which are not, please see Appendix Table C-l.
In these scenarios, EPA imposed cost thresholds of $1,800 and $11,000/ton of ozone
season NOx.8 See Preamble Section VI for a discussion of how the cost thresholds were
5 These projected state level emissions and heat input for each "cost threshold" run are presented in several formats.
The IPM analysis outputs available in the docket contain a "state emissions" file for each analysis. The file contains
two worksheets. The first is titled "all units" and shows aggregate emissions for all units in the state. The second is
titled "all fossil > 25MW" and shows emissions for a subset of these units that have a capacity greater than 25 MW.
The 2023 and 2025 emissions and heat input in the "all fossil > 25 MW" worksheet is used to derive the generation
shifting component of the state emission budgets for each upwind state at level of emission control stringency.
6 IPM uses model years to represent the full planning horizon being modeled. By mapping multiple calendar years to
a run year, the model size is kept manageable. For this analysis, IPM maps the calendar year 2023 to run year 2023,
calendar years 2024-2026 to run year 2025 and calendar years 2027-2029 to run year 2028. For model details, please
see Chapter 2 of the IPM documentation, available at:
https://www.epa.gov/system/files/documents/2021-09/epa-platform-v6-summer-2021-reference-case-09-ll-21-
v6.pdf
7 IPM results do include certain newly built post-combustion NOx control retrofits in base case modeling, cost curve
runs, and remedy runs. These pre-2023 retrofits do not reflect any controls installed in response to the rule, but
instead represent those that are already announced and/or under construction and expected to be online by 2023, or
controls that were projected to be built in the base case in response to existing consent decree or state rule
requirements.
8 The $11,000/ton cost threshold run is named such to clarify it is linked to that level NOx Mitigation stringency
measures. Because the run was conducted before the $11,000/ton representative price was calculated, the run only
imposes a NOx price of $10,000/ton. Since that NOx price did not induce significant amounts of generation shifting,
5
-------
determined. Table A-2 below summarizes the reduction measures that are broadly available at
various cost thresholds.
Table A-2. Reduction strategies available to EGUs at each cost threshold.
Cost Threshold ($ per
ton Ozone-Season NOx)
Reduction Options
$1,800
-Generation Shifting;
-Retrofitting state-of-the-art combustion controls;
-Optimizing idled SCRs;
-Optimizing operating SNCRs9
$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 both Engineering analytics and IPM:
• At $l,800/ton:
o Engineering Analytics
¦ 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;
¦ 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;
¦ 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;
¦ 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
¦ 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.
¦ Starting in 2023 units with SNCRs were given their mode 2 NOx rates10 if
they were not already operating at that level or better in 2019.
o IPM - cost of $l,800/ton applied to EGUs > 25 MW; units with existing SCRs
have their emission rates lowered to the lower of their mode 4 NOx rate in
given the other mitigation strategies included in the model run, EPA does not believe that the results would have
changed appreciably if a $ll,000/ton price on NOx was included instead.
9 As explained in the preamble section VLB, EPA notes that this technology becomes widely available at
$l,800/ton. For purposes of assessing generation shifting available at this technology level's commensurate cost,
EPA relies on its $l,800/ton IPM analysis.
10 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.
6
-------
NEEDS and the "widely achievable" optimized emissions rate consistent with the
rates used in the Engineering Analysis. 11
• At $11,000/ton:
o Engineering Analytics - Same as $l,800/ton; additionally, coal units greater than
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 100
MW and operating at an average 20% capacity factor or higher were given an
emission rate of 0.03 lb/MMBtu reflecting SCR installation starting in 2026.
o IPM - Cost of $10,000/ton applied to EGUs > 25 MW;12 in addition to the
emission rate adjustments noted in the $l,800/ton scenario, coal units greater than
100 MW and lacking SCR were assigned an emission rate of 0.05 lb/MMBtu
reflecting SCR installation starting in model run year 2025. Oil/gas steam units
greater than 100 MW were given an emission rate of 0.03 lb/MMBtu reflecting
SCR installation in model year 2025 (to which calendar year 2026 is mapped).
As described in preamble section VLB, the EPA limited its assessment of generation
shifting to reflect shifting only to other EGUs within the same state as a proxy for generation
shifting that could occur during the near-term implementation timeframe of the rule. EPA did
this by establishing a minimum level of required generation in each state in each Cost Threshold
run equal to its respective Base Case generation level. EPA also prohibited the model from
constructing any new (unplanned) capacity built in response to the price signal in the near term
as it was interested in capturing generation shifting among the existing fleet.
B. Calculating Step 4 EGU Emission Budgets from Historical Data and IPM Analysis
In this proposed rule the EPA calculated state budgets with the following formula:
2023 State OS NOx Budget =
2023 State OS Baseline Heat Input *[2023 State OS NOx Emissions Rate —
(2023 IPM Base Case OS NOx Emissions Rate — 2023 IPM Cost Threshold
OS NOx EmissionsRate)]13
The first two variables in the equation are derived from historical data and are the primary
determinants of states' emissions budgets. They are described in sections B.l and B.2 below.
11 The mode 4 NOx rate, as described in Chapter 3 of the Documentation for EPA Base Case v.6 Using Integrated
Planning Model, represents post-combustion controls operating and state-of-the-art combustion controls, where
applicable. For units determined to be operating their SCR, the rate is typically equal to the unit's rate reported in
previous year ETS data. For units not operating their SCRs, the mode 4 rate is calculated as described in Attachment
3-1 of Documentation for EPA's Power Sector Modeling Platform v6 using Summer 2021 Reference Case available
at https ://www. epa. gov/airmarkets/documentation-epas-power-sector-modeling-platform-v6-summer-2021 -
reference-case.
12 See footnote 8 for explanation of why $10,000/ton was used in the IPM modeling.
13 The year in the formula changes for each year of budget calculation.
7
-------
The last two variables are identified through IPM analysis and described in section B.3 below.14
In section B.4, EPA discusses variability limits.
1. Calculating 2023-2026 Engineering Baseline Heat Input
The underlying data and calculations described below can be found in the workbook titled
(Appendix A - Proposed 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.15 This reflects the latest representative owner/operator reported data available at the time of
EPA analysis. 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 2026" sheets in
the "Appendix A: Proposed Rule State Emission Budget Calculations and Engineering
Analytics" file accompanying this document.16 In that file, unit-level details such as facility
name, unit ID, unit type, capacity, etc. are shown in columns A through H of the "unit 2023"
through "unit 2026" worksheets. Reported historical data for these units such as unit type, fuel,
existing post combustion controls, historical emissions, heat input, generation, etc. are shown in
columns I through W. For approximately twenty additional units that have not reported to EPA
but which are included in this proposal, EIA data sources are used to obtain the necessary data.
The 2021 historical emissions value is in column Q. The assumed future year baseline emissions
estimate (e.g., 2023-2026) is shown in column AF, and reflects either the same emissions level
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.17 These modifications are made due to:
a. Retirements - Emissions from units with upcoming confirmed retirement dates are
adjusted to zero for years subsequent to that retirement date. Retirement dates are
identified through a combination of sources including EIA Form 860, utility-
announced retirements, stakeholder feedback provided to EPA, and the National
Electricity Energy Data System (NEEDS) October 2021 file. The impact of
retirements on emissions is shown in column X. The retiring units are flagged in
column Y.18
14 Given the proximity of the first implementation year to the analytics for this rulemaking and its promulgation,
EPA determined the use of this approach to develop budgets to implement the chosen level of emission control
stringency provided the most precision and expediency for this rulemaking.
15 "Seasonal" refers to the ozone-season program months of May through September.
16 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.
17 Based on data and changes known at time of analysis.
18 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.
8
-------
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.19 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, NEEDS October 2021, and stakeholder feedback. The
impact of coal to gas conversion for the future year is shown in column AB, flagged
in column AC. The example below pertains to NOx emission estimates.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtJ/ x 0.1 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) 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.20 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 AD,
flagged in column AE.
For SNCR:
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtJ/ 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 MMBtJ/ 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, and/or other revised permit limits. The impacts for
future year emission assumptions are shown in column AF, flagged in column AG.21
19 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.
211 Ibid.
21 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 Proposed Rule TSD.
9
-------
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 H. To obtain these emissions, EPA identified all new fossil-fired EGUs
coming online after 2021 according to EIA Form 860 and in NEEDSv.6 October
2021. EPA then identified the heat rate and capacity values for these units using EIA
Form 860, NEEDSv.6 October 2021 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 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).22
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 and state-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 and provides the
state-level heat input value used as the first variable in the emissions budget formula below.23
^—2023 State OS NCb^Budget =
^023 State OS Baseline HeatTnptit *[2023 State OS NOx Emissions Rate —
' (zuzi IPM Base~Case OS NOx Emissions Rate — 2023 IPM Cost Threshold OS NOx Emissions
Rate)]
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 to
determine the second variable in the equation above. 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
22 Emission rate data is informed by the NEEDS data and historical data 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.
23 EPA also created a future year baseline for 1) NOx and SO: emission from EGUs not currently covered under
existing EPA programs that require emissions monitoring and reporting under 40 CFR part 75, and for other
pollutants for all grid connected EGUs (e.g., PM2.5, P.M10. CO). These data points were used in some of the air
quality analysis and in some of the system impacts estimates for the RIA. In the appendix to this TSD, the EPA
evaluates whether the assumed aggregate heat input changes given retirements and new builds are consistent with
trends observed historically in the fleet and with new planned units identified in EIA Form 860.
10
-------
proposed in this rule. Each of these adjustments is shown incrementally for the relevant
mitigation technology in the "unit 2023" through "unit 2026" 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 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 AH and flagged in column AI of the "unit 2023" through "unit 2026"
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 proposed rule,
EPA also incorporated a flag in column AI for units with SCRs and a shared stack. For
these units, EPA did not assume potential emission reductions attributable to existing
SCR optimization as explained in preamble section VLB.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtJ/ 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 Proposed Rule TSD.
The impact of state-of-the-art combustion controls on future year emission assumptions is
shown in column AJ and flagged in column AK of the "unit 2023" through "unit 2026"
worksheets. EPA also incorporated a flag in column AK, 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 VLB. 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 MMBtJ/ xO 0.199Ib/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 Proposed 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
11
-------
shown in column AL and flagged in column AM of the "unit 2023" through "unit 2026"
worksheets. Note, this assumption only applies to ozone-season NOx as that is the season
in which this proposal's programwould likely incentivize such operation.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtJ/ x 0.15 lb/MMBtu = 0.75 ton
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 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 an the retrofit assumptions, see section VLB
of the Preamble.
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
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).24 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 AP and flagged in column AQ of
the "unit 2023" through "unit 2026" worksheets.
24 See https://www.epa.gov/aimiarkets/retrofit-cost-analYzer for the "Retrofit Cost Analyzer (Update 1-26-2022)"
Excel tool 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" (August 2021).
12
-------
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtJ/ x 0.15 lb/MMBtu = 0.75 ton
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. 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). 25 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 AP and flagged in column AQ of
the "unit 2023" through "unit 2026" worksheets. Note, this assumption only applies to
ozone-season NOx.
2021
Future Year (e.g., 2023)
Unit x
10,000 MMBtu x 0.2 lb/MMBtu = 1 ton
10,000 MMBtJ/ x 0.05 lb MMBtu = 0.25 ton
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. This state emissions total is dividied by the
state heat input total to derive the state emission rate in the formula below. EPA notes, this
emission rate for any given uniform control level times the baseline heat input would provide
state-level emissions before generation shifting is incorporated; these state-level emissions are
visible in the worksheets titled "State 2023" through "State 2026" in the Appendix A: Proposed
Ride State Emission Budget Calculations and Engineering Analytics workbook accompanying
this document.26
State 2023 OS NOx Budget =
2023 State OS Baseline Heat Input^j2023 State OSNOx Emissions~RM^-
(2023 IPM Base Case OSN(T^hilWsbiuia Rule — 202J HTCfCost Threshold OS NOx Emissions
Rate)]
3. Estimating Emission Reduction Potential from Generation Shifting
The last two variables in the equation relate to emission reductions from generation shifting.
Here, as in the Revised CSAPR Update, EPA uses the Integrated Planning Model (IPM) to
25 By comparison, in the IPM Summer 2021 Reference Case, EPA assumes new SCRs on coal steam units can
achieve a 90% reduction in emission with floor rates of 0.05 to 0.07 lb/MMBtu, depending on coal type, and an 80%
reduction, with no floor rate, for oil gas steam units. See "Documentation for EPA's Power Sector Modeling
Platform v6 using Summer 2021 Reference Case". Available at
https://www.https://www.epa.gov/ainnarkets/documentation-epas-power-sector-modeling-platfonn-v6-summer-
2021 -reference-case
26 EPA makes these illustrative unit-level details described in B. 1 and B.2 available, before aggregating those
values to use at the state and regional level. The illustrative unit-level values are meant to be a tool to inform a state-
level estimate, 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 overperfonn and some units will underperfonn the unit-level
illustrative values.
13
-------
capture the change in heat input weighted average emission rate in a state's fossil-fuel fired
power fleet while holding everything else equal and applying a given dollar per ton marginal cost
to ozone-season NOx emissions.27 To derive this value, EPA first prepares an adjusted base case
that reflects all the combustion or post-combustion mitigation measures discussed above for a
given cost threshold. These adjusted base cases are specific to the uniform mitigation scenario.
For instance, for the $l,800/ton scenario EPA adjusts its base case to reflect the optimization of
SCRs, SNCRs and combustion control upgrades by adjusting the emission rates to the levels
discussed above for relevant units not already achieving that level. EPA then executes an IPM
run with these new exogenous assumptions and observes the state-level emission rate for fossil-
fuel fired units greater than 25 MW. This is the third variable in the emissions budget formula.
Next, EPA performs cost threshold scenarios where, for each cost threshold run, EPA applies the
same set of assumptions in the corresponding mitigation measures scenario but layers on a
commensurate marginal cost price signal (e.g., $l,800/ton). In addition to the mitigation
measures assumed, the entire fossil-fuel fired EGU fleet greater than 25 MW in the state is
subjected to a cost-per-ton price associated with those mitigation measures. The model solves
for least-cost dispatch given this additional marginal cost for seasonal ozone emissions. In its
cost threshold modeling, EPA imposed a minimum generation level in each state covering all
EGUs equal to their projected generation level in the IPM base case, such that EPA would not
include emission reduction potential for a given state related to increased electricity imported
from out-of-state generators.
EPA observes the state-level emission rate for fossil-fuel fired units greater than 25 MW
in the applicable cost threshold scenario.28 This data point becomes the fourth variable in the
state-emissions budget formula. The difference between the third and fourth variables reflects
the change in emission rate due solely to generation shifting at a given dollar per ton level.
State 2023 OS NOx Budget =
2023 State OS Baseline Heat Input *[2023 State OS NOx Emissions Rate —
tlase Case OS NOx Emissions Rate — 2023 IPM Cost Threshold uS /Wjffrmissicms
This difference in the state-level emission rate between the two IPM cases is shown in
columns B and C in the worksheet titled "Generation Shifting".30 These values are in the
Appendix A: Proposed Rule State Emission Budget Calculations and Engineering Analytics
workbook accompanying this document. Column B provides the "2023" generation shifting
emission rate delta pertaining to the $l,800/ton threshold that corresponds to mitigation
27 EPA relies on IPM for this analysis as generation shifting occurs on a cost continuum and is a function of least-
cost dispatch under different constraints.
28 In each cost threshold run, EPA quantified generation-shifting emission rate changes from the IPM 2023 run-year
to avoid capturing generation shifting attributable to model-projected new builds in later years.
29 The year in the formula changes for each year of budget calculation.
30 If the state's assumed emission rate reductions from generation shifting were greater than 10% of the IPM
baseline, or its adjusted historical baseline for that year was less than 90% of the IPM baseline, then no additional
reductions were assumed from generation shifting at the cost threshold of $l,800/ton in EPA's 2023 analysis. If the
state's assumed emission rate reductions from generation shifting were greater than 10% of the IPM baseline, or its
adjusted historical baseline for that year was less than 90% of the IPM baseline, then reductions consistent with the
results from the $l,800/ton analysis were assumed from generation shifting at higher cost thresholds of $10,000/ton
($11,000/ton cost threshold run) in EPA's 2026 analysis.
14
-------
technologies available in 2023, 2024, and/or 2025. Column C provides the generation shifting
emission rate delta pertaining to the $11,000/ton threshold that corresponds to technologies in
2026 and later years. Therefore, column B value is used for state emission calculations in the
"2023", "2024", and "2025" state worksheets. Column C value is used in the "2026"
worksheet.31
Once EPA calculated the change in emissions rate between the IPM adjusted base case
and each cost threshold case, the EPA then subtracted this IPM-projected change in emissions
rate from the engineering analytics-derived state OS NOx emission rate (the second variable in
the formula). This computation yields state-level, historically-anchored emission rates reflecting
NOx reduction potential for a given control stringency, inclusive of generation shifting at a
commensurate representative cost level.
Finally, the EPA multiplied these rates by each state's adjusted heat input (historical heat
input adjusted for retirements and new builds identified in variable one of the formula) to yield
emission budgets for each cost threshold. The state budgets for the different cost thresholds are
displayed in Tables B-2 through B-5.
In addition to being shown below, the state-level emission budgets are calculated in the
far right-hand side columns of each "State" worksheet for each mitigation technology scenario
available that year. These budgets reflect an application of the formula described above to the
data in the spreadsheet. These state-emission budgets reflect the inclusion of generation shifting.
31 EPA notes the "2025" and "2026" worksheets showing state-level emisison estimates subject to different
technologies are illustrative only. The "dynamic budget" worksheet for each year 2025 and beyond is the worksheet
used to calculate state-emission budgets for covered states in those future years.
15
-------
Table B-2. 2023 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios*
State
2023
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR Optimization
+ SOA CC
SCR Optimization
+ SOA CC + SNCR
Optimization
SCR Optimization
+ SOA CC +
SNCR
Optimization +
Generation
Shifting
Alabama
6,648
6,616
6,492
6,492
6,261
Arizona
7,723
7,639
7,570
7,439
7,570
Arkansas
8,955
8,927
8,927
8,927
8,889
California
1,606
1,216
1,216
1,216
1,216
Colorado
6,467
6,389
6,389
6,389
6,389
Connecticut
381
355
355
355
355
Delaware
423
388
388
384
388
Florida
13,770
11,339
11,339
11,339
11,339
Georgia
5,514
5,497
5,497
5,490
5,497
Idaho
240
240
240
240
240
Illinois
7,662
7,592
7,592
7,415
7,542
Indiana
12,351
11,495
11,495
11,486
11,160
Iowa
9,072
9,072
9,018
8,958
9,018
Kansas
6,231
5,484
5,484
5,484
5,484
Kentucky
13,900
13,454
12,853
12,853
11,640
Louisiana
9,987
9,408
9,408
9,312
9,408
Maine
108
86
86
86
86
Maryland
1,208
1,208
1,208
1,200
1,195
Massachusetts
297
265
265
265
265
Michigan
10,737
10,733
10,733
10,718
10,733
Minnesota
4,207
4,109
4,109
4,068
3,961
Mississippi
5,097
5,024
4,400
4,400
4,400
Missouri
20,094
12,749
12,749
12,525
12,081
Montana
3,071
3,071
3,071
3,071
3,071
Nebraska
8,931
8,894
8,381
8,381
8,381
Nevada
2,346
2,280
2,280
2,280
2,280
New Hampshire
247
184
184
184
184
New Jersey
915
810
810
810
799
New Mexico
2,289
2,259
2,259
2,259
2,259
New York
3,927
3,863
3,863
3,863
3,763
North Carolina
12,354
9,298
9,298
9,268
9,298
North Dakota
12,246
12,246
12,246
11,436
12,246
Ohio
10,295
9,134
9,134
9,134
8,369
Oklahoma
10,463
10,265
9,573
9,573
9,573
Oregon
337
288
288
288
289
Pennsylvania
12,242
9,364
9,364
9,264
8,955
Rhode Island
279
148
148
148
148
South Carolina
4,273
3,531
3,531
3,531
3,531
South Dakota
568
568
568
568
568
Tennessee
4,319
4,209
4,209
4,209
4,234
Texas
40,860
39,938
39,938
39,706
38,516
16
-------
State
2023
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR Optimization
+ SOA CC
SCR Optimization
+ SOA CC + SNCR
Optimization
SCR Optimization
+ SOA CC +
SNCR
Optimization +
Generation
Shifting
Utah
15,500
15,493
15,493
15,493
14,981
Vermont
54
54
54
54
54
Virginia
3,415
3,251
3,174
3,120
3,144
Washington
1,999
1,729
1,729
1,729
1,729
West Virginia
14,686
14,132
13,586
13,306
12,759
Wisconsin
5,933
5,927
5,927
5,907
5,983
Wyoming
10,191
10,110
9,514
9,501
8,543
Total
334,421
310,331
306,436
304,124
298,774
Linked in 2023
238,306
221,983
218,724
217,450
211,062
Linked in 2026
226,916
210,771
207,635
206,365
200,179
17
-------
Table B-3. 2024 Ozone Season NOx Emissions for States at Di
State
2024
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization +
SOA CC +
SNCR
Optimization
SCR Optimization +
SOA CC + SNCR
Optimization +
Generation Shifting
Alabama
6,701
6,668
6,545
6,545
6,306
Arizona
7,723
7,639
7,570
7,439
7,570
Arkansas
8,955
8,927
8,927
8,927
8,889
California
1,589
1,199
1,199
1,199
1,199
Colorado
5,877
5,799
5,799
5,799
5,799
Connecticut
381
355
355
355
355
Delaware
473
438
438
434
438
Florida
13,097
10,720
10,720
10,720
10,720
Georgia
5,514
5,497
5,497
5,490
5,497
Idaho
240
240
240
240
240
Illinois
7,763
7,694
7,694
7,516
7,640
Indiana
10,525
9,712
9,712
9,703
9,400
Iowa
9,072
9,072
9,018
8,958
9,018
Kansas
6,231
5,484
5,484
5,484
5,484
Kentucky
13,900
13,454
12,853
12,853
11,640
Louisiana
9,987
9,408
9,408
9,312
9,408
Maine
108
86
86
86
86
Maryland
1,208
1,208
1,208
1,200
1,195
Massachusetts
297
265
265
265
265
Michigan
10,737
10,733
10,733
10,718
10,733
Minnesota
4,207
4,109
4,109
4,068
3,961
Mississippi
5,097
5,024
4,400
4,400
4,400
Missouri
20,094
12,749
12,749
12,525
12,081
Montana
3,071
3,071
3,071
3,071
3,071
Nebraska
8,931
8,894
8,381
8,381
8,381
Nevada
2,438
2,372
2,372
2,372
2,372
New Hampshire
247
184
184
184
184
New Jersey
915
810
810
810
799
New Mexico
2,289
2,259
2,259
2,259
2,259
New York
3,927
3,863
3,863
3,863
3,763
North Carolina
12,354
9,298
9,298
9,268
9,298
North Dakota
12,246
12,246
12,246
11,436
12,246
Ohio
10,295
9,134
9,134
9,134
8,369
Oklahoma
10,463
10,265
9,573
9,573
9,573
Oregon
337
288
288
288
289
Pennsylvania
12,242
9,364
9,364
9,264
8,955
Rhode Island
279
148
148
148
148
South Carolina
4,273
3,531
3,531
3,531
3,531
South Dakota
568
568
568
568
568
Tennessee
4,319
4,209
4,209
4,209
4,234
Texas
40,860
39,938
39,938
39,706
38,516
Utah
15,673
15,666
15,666
15,666
15,146
Vermont
54
54
54
54
54
ferent Uniform Control Scenarios
18
-------
State
2024
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization +
SOA CC +
SNCR
Optimization
SCR Optimization +
SOA CC + SNCR
Optimization +
Generation Shifting
Virginia
3,106
2,942
2,865
2,843
2,836
Washington
1,999
1,729
1,729
1,729
1,729
West Virginia
14,686
14,132
13,586
13,306
12,759
Wisconsin
5,029
5,023
5,023
5,003
5,077
Wyoming
10,249
10,167
9,572
9,559
8,586
Total
330,627
306,634
302,739
300,459
295,067
Linked in 2023
235,776
219,497
216,237
214,995
208,564
Linked in 2026
224,283
208,181
205,045
203,808
197,586
19
-------
Table B-4. 2025 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2025
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR Optimization
+ SOA CC + SNCR
Optimization +
Generation
Shifting
Alabama
6,701
6,668
6,545
6,545
6,306
Arizona
7,723
7,639
7,570
7,439
7,570
Arkansas
8,955
8,927
8,927
8,927
8,889
California
1,547
1,157
1,157
1,157
1,157
Colorado
5,877
5,799
5,799
5,799
5,799
Connecticut
381
355
355
355
355
Delaware
473
438
438
434
438
Florida
13,142
10,765
10,765
10,765
10,765
Georgia
5,514
5,497
5,497
5,490
5,497
Idaho
240
240
240
240
240
Illinois
7,763
7,694
7,694
7,516
7,640
Indiana
9,737
9,017
9,017
9,008
8,723
Iowa
9,072
9,072
9,018
8,958
9,018
Kansas
6,231
5,484
5,484
5,484
5,484
Kentucky
13,211
12,765
12,325
12,325
11,134
Louisiana
9,854
9,275
9,275
9,179
9,275
Maine
108
86
86
86
86
Maryland
1,208
1,208
1,208
1,200
1,195
Massachusetts
288
256
256
256
256
Michigan
10,778
10,774
10,774
10,759
10,774
Minnesota
4,197
4,099
4,099
4,058
3,951
Mississippi
5,097
5,024
4,400
4,400
4,400
Missouri
18,610
11,265
11,265
11,041
10,679
Montana
3,071
3,071
3,071
3,071
3,071
Nebraska
8,247
8,210
8,177
8,177
8,177
Nevada
2,438
2,372
2,372
2,372
2,372
New Hampshire
247
184
184
184
184
New Jersey
915
810
810
810
799
New Mexico
2,232
2,201
2,201
2,201
2,201
New York
3,927
3,863
3,863
3,863
3,763
North Carolina
12,228
9,172
9,172
9,162
9,172
North Dakota
12,246
12,246
12,246
11,436
12,246
Ohio
10,295
9,134
9,134
9,134
8,369
Oklahoma
10,283
10,084
9,393
9,393
9,393
Oregon
345
296
296
296
297
Pennsylvania
12,242
9,364
9,364
9,264
8,955
Rhode Island
279
148
148
148
148
South Carolina
4,273
3,531
3,531
3,531
3,531
South Dakota
568
568
568
568
568
Tennessee
4,064
3,983
3,983
3,983
4,008
Texas
39,186
38,265
38,265
38,032
36,851
Utah
15,673
15,666
15,666
15,666
15,146
20
-------
State
2025
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization
+ SOA CC +
SNCR
Optimization
SCR Optimization
+ SOA CC + SNCR
Optimization +
Generation
Shifting
Vermont
54
54
54
54
54
Virginia
3,243
3,079
3,001
2,980
2,970
Washington
1,999
1,729
1,729
1,729
1,729
West Virginia
14,686
14,132
13,586
13,306
12,759
Wisconsin
4,178
4,171
4,171
4,152
4,217
Wyoming
10,249
10,167
9,572
9,559
8,586
Total
323,874
300,004
296,750
294,490
289,197
Linked in 2023
229,853
213,697
210,599
209,357
203,046
Linked in 2026
218,615
202,607
199,632
198,395
192,294
21
-------
Table B-5. 2026 Ozone Season NOx Emissions for States at Different Uniform Control Scenarios
State
2026
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization +
SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC +
SNCR
Optimization +
SCR/SNCR
Retrofit
SCR Optimization
+ SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit +
Generation
Shifting
Alabama
6,701
6,668
6,545
6,545
5,785
5,785
Arizona
5,237
5,153
5,084
4,954
3,152
3,152
Arkansas
8,728
8,700
8,700
8,700
4,031
3,923
California
1,547
1,157
1,157
1,157
1,157
1,157
Colorado
5,877
5,799
5,799
5,799
3,482
3,482
Connecticut
381
355
355
355
355
355
Delaware
473
438
438
434
434
434
Florida
13,142
10,765
10,765
10,765
8,041
8,041
Georgia
5,514
5,497
5,497
5,490
5,325
5,325
Idaho
240
240
240
240
240
240
Illinois
7,763
7,694
7,694
7,516
6,465
6,115
Indiana
9,737
9,017
9,017
9,008
7,997
7,791
Iowa
9,072
9,072
9,018
8,958
3,556
3,556
Kansas
6,231
5,484
5,484
5,484
3,394
3,394
Kentucky
13,211
12,765
12,325
12,325
7,761
7,573
Louisiana
9,854
9,275
9,275
9,179
3,752
3,752
Maine
108
86
86
86
86
86
Maryland
1,208
1,208
1,208
1,200
1,200
1,189
Massachusetts
287
256
256
256
256
256
Michigan
9,129
9,125
9,125
9,110
6,170
6,114
Minnesota
4,197
4,099
4,099
4,058
2,584
2,536
Mississippi
5,077
5,004
4,379
4,379
1,913
1,914
Missouri
18,610
11,265
11,265
11,041
7,373
7,246
Montana
3,071
3,071
3,071
3,071
1,177
1,177
Nebraska
8,247
8,210
8,177
8,177
2,974
2,974
Nevada
2,438
2,372
2,372
2,372
1,211
1,211
New Hampshire
247
184
184
184
184
184
New Jersey
915
810
810
810
810
799
New Mexico
2,232
2,201
2,201
2,201
1,712
1,712
New York
3,927
3,863
3,863
3,863
3,338
3,238
North Carolina
12,228
9,172
9,172
9,162
6,467
6,467
North Dakota
12,246
12,246
12,246
11,436
2,927
2,927
Ohio
10,295
9,134
9,134
9,134
8,941
8,586
Oklahoma
10,283
10,084
9,393
9,393
4,315
4,275
Oregon
345
296
296
296
296
304
Pennsylvania
11,738
9,000
9,000
8,901
7,228
6,819
Rhode Island
279
148
148
148
148
148
South Carolina
4,273
3,531
3,531
3,531
3,531
3,531
South Dakota
568
568
568
568
568
568
Tennessee
4,064
3,983
3,983
3,983
3,983
3,983
Texas
39,186
38,265
38,265
38,032
23,369
21,946
22
-------
State
2026
Baseline
(Engineering
Analysis)
SCR
Optimization
SCR
Optimization
+ SOA CC
SCR
Optimization +
SOA CC +
SNCR
Optimization
SCR
Optimization +
SOA CC +
SNCR
Optimization +
SCR/SNCR
Retrofit
SCR Optimization
+ SOA CC + SNCR
Optimization +
SCR/SNCR
Retrofit +
Generation
Shifting
Utah
9,679
9,672
9,672
9,672
2,604
2,620
Vermont
54
54
54
54
54
54
Virginia
3,243
3,079
3,001
2,980
2,597
2,567
Washington
1,999
1,729
1,729
1,729
639
639
West Virginia
14,686
14,132
13,586
13,306
11,026
10,597
Wisconsin
3,628
3,622
3,622
3,602
3,575
3,473
Wyoming
10,249
10,167
9,572
9,559
4,580
4,490
Total
312,443
288,714
285,461
283,201
182,758
178,705
Linked in 2023
220,909
204,893
201,795
200,554
134,492
130,437
Linked in 2026
209,670
193,803
190,829
189,591
124,290
120,235
As described in Section VI 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 VII 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. 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-6
through B-10.32
32 A table providing state emission budgets and associated variability limits for these linked states is provided in
Appendix F
23
-------
Table B-6. OS NOx: 2023 Emissions Budget, and % Reduction
State
2016 OS
NOx
(tons)
2021 OS
NOx
(tons)
Baseline
2023 OS
NOx
(tons)
2023
Budget
(tons)
%
Reduction
from 2021
%
Reduction
from 2023
Baseline
Alabama
11,612
6,648
6,648
6,364
4%
4%
Arkansas
13,223
8,955
8,955
8,889
1%
1%
Delaware
551
423
423
384
9%
9%
Illinois
14,550
11,276
7,662
7,364
35%
4%
Indiana
34,670
14,162
12,351
11,151
21%
10%
Kentucky
25,403
14,571
13,900
11,640
20%
16%
Louisiana
19,615
11,456
9,987
9,312
19%
7%
Maryland
4,471
1,422
1,208
1,187
17%
2%
Michigan
17,632
13,554
10,737
10,718
21%
0%
Minnesota
7,587
5,652
4,207
3,921
31%
7%
Mississippi
7,325
5,790
5,097
5,024
13%
1%
Missouri
25,255
20,388
20,094
11,857
42%
41%
Nevada
2,275
2,457
2,346
2,280
7%
3%
New Jersey
2,463
1,322
915
799
40%
13%
New York
6,534
3,997
3,927
3,763
6%
4%
Ohio
24,205
11,728
10,295
8,369
29%
19%
Oklahoma
12,761
10,470
10,463
10,265
2%
2%
Pennsylvania
31,896
12,792
12,242
8,855
31%
28%
Tennessee
9,759
4,319
4,319
4,234
2%
2%
Texas
54,668
42,760
40,860
38,284
10%
6%
Utah
12,955
15,762
15,500
14,981
5%
3%
Virginia
9,833
3,329
3,415
3,090
7%
10%
West Virginia
21,178
14,686
14,686
12,478
15%
15%
Wisconsin
7,946
6,307
5,933
5,963
5%
0%
Wyoming
15,664
11,643
10,191
9,125
22%
10%
Total
394,029
255,868
236,363
210,297
18%
11%
24
-------
Table B-7. OS NOx: 2024 Emissions Budget, and % Reduction
State
2016
2021
Baseline
2024
%
%
OS
OS
2024
Budget
Reduction
Reduction
NOx
NOx
OS
(tons)
from 2021
from 2024
(tons)
(tons)
NOx
Baseline
(tons)
Alabama
11,612
6,648
6,701
6,306
5%
6%
Arkansas
13,223
8,955
8,955
8,889
1%
1%
Delaware
551
423
473
434
-3%
8%
Illinois
14,550
11,276
7,763
7,463
34%
4%
Indiana
34,670
14,162
10,525
9,391
34%
11%
Kentucky
25,403
14,571
13,900
11,640
20%
16%
Louisiana
19,615
11,456
9,987
9,312
19%
7%
Maryland
4,471
1,422
1,208
1,187
17%
2%
Michigan
17,632
13,554
10,737
10,718
21%
0%
Minnesota
7,587
5,652
4,207
3,921
31%
7%
Mississippi
7,325
5,790
5,097
4,400
24%
14%
Missouri
25,255
20,388
20,094
11,857
42%
41%
Nevada
2,275
2,457
2,438
2,372
3%
3%
New Jersey
2,463
1,322
915
799
40%
13%
New York
6,534
3,997
3,927
3,763
6%
4%
Ohio
24,205
11,728
10,295
8,369
29%
19%
Oklahoma
12,761
10,470
10,463
9,573
9%
9%
Pennsylvania
31,896
12,792
12,242
8,855
31%
28%
Tennessee
9,759
4,319
4,319
4,234
2%
2%
Texas
54,668
42,760
40,860
38,284
10%
6%
Utah
12,955
15,762
15,673
15,146
4%
3%
Virginia
9,833
3,329
3,106
2,814
15%
9%
West Virginia
21,178
14,686
14,686
12,478
15%
15%
Wisconsin
7,946
6,307
5,029
5,057
20%
-1%
Wyoming
15,664
11,643
10,249
8,573
26%
16%
Total
394,029
255,868
233,849
205,835
20%
12%
25
-------
Table B-8. OS NOx: Illustrative 2025 Emissions Budget, and % Reduction
State
2016
2021
Baseline
Illustrative
%
%
OS
OS
2025
2025
Reduction
Reduction
NOx
NOx
OS
Budget
from 2021
from 2025
(tons)
(tons)
NOx
(tons)
Baseline
(tons)
Alabama
11,612
6,648
6,701
6,306
5%
6%
Arkansas
13,223
8,955
8,955
8,889
1%
1%
Delaware
551
423
473
434
-3%
8%
Illinois
14,550
11,276
7,763
7,463
34%
4%
Indiana
34,670
14,162
9,737
8,714
38%
11%
Kentucky
25,403
14,571
13,211
11,134
24%
16%
Louisiana
19,615
11,456
9,854
9,179
20%
7%
Maryland
4,471
1,422
1,208
1,187
17%
2%
Michigan
17,632
13,554
10,778
10,759
21%
0%
Minnesota
7,587
5,652
4,197
3,910
31%
7%
Mississippi
7,325
5,790
5,097
4,400
24%
14%
Missouri
25,255
20,388
18,610
10,456
49%
44%
Nevada
2,275
2,457
2,438
2,372
3%
3%
New Jersey
2,463
1,322
915
799
40%
13%
New York
6,534
3,997
3,927
3,763
6%
4%
Ohio
24,205
11,728
10,295
8,369
29%
19%
Oklahoma
12,761
10,470
10,283
9,393
10%
9%
Pennsylvania
31,896
12,792
12,242
8,855
31%
28%
Tennessee
9,759
4,319
4,064
4,008
7%
1%
Texas
54,668
42,760
39,186
36,619
14%
7%
Utah
12,955
15,762
15,673
15,146
4%
3%
Virginia
9,833
3,329
3,243
2,948
11%
9%
West Virginia
21,178
14,686
14,686
12,478
15%
15%
Wisconsin
7,946
6,307
4,178
4,198
33%
0%
Wyoming
15,664
11,643
10,249
8,573
26%
16%
Total
394,029
255,868
227,962
200,352
22%
12%
26
-------
Table B-9. OS NOx: Illustrative 2026 Emissions Budget, and % Reduction
State
2016
2021
Baseline
Illustrative
%
%
OS
OS
2026
2026
Reduction
Reduction
NOx
NOx
OS
Budget
from 2021
from 2026
(tons)
(tons)
NOx
(tons)
Baseline
(tons)
Alabama
11,612
6,648
6,701
6,306
5%
6%
Arkansas
13,223
8,955
8,728
3,923
56%
55%
Delaware
551
423
473
434
-3%
8%
Illinois
14,550
11,276
7,763
6,115
46%
21%
Indiana
34,670
14,162
9,737
7,791
45%
20%
Kentucky
25,403
14,571
13,211
7,573
48%
43%
Louisiana
19,615
11,456
9,854
3,752
67%
62%
Maryland
4,471
1,422
1,208
1,189
16%
2%
Michigan
17,632
13,554
9,129
6,114
55%
33%
Minnesota
7,587
5,652
4,197
2,536
55%
40%
Mississippi
7,325
5,790
5,077
1,914
67%
62%
Missouri
25,255
20,388
18,610
7,246
64%
61%
Nevada
2,275
2,457
2,438
1,211
51%
50%
New Jersey
2,463
1,322
915
799
40%
13%
New York
6,534
3,997
3,927
3,238
19%
18%
Ohio
24,205
11,728
10,295
8,586
27%
17%
Oklahoma
12,761
10,470
10,283
4,275
59%
58%
Pennsylvania
31,896
12,792
11,738
6,819
47%
42%
Tennessee
9,759
4,319
4,064
4,008
7%
1%
Texas
54,668
42,760
39,186
21,946
49%
44%
Utah
12,955
15,762
9,679
2,620
83%
73%
Virginia
9,833
3,329
3,243
2,567
23%
21%
West Virginia
21,178
14,686
14,686
10,597
28%
28%
Wisconsin
7,946
6,307
3,628
3,473
45%
4%
Wyoming
15,664
11,643
10,249
4,490
61%
56%
Total
394,029
255,868
219,017
129,522
49%
41%
27
-------
Table B-10. Emission Reduction Attributable to Generation Shifting (2025 and 2026 are
illustrative).
Baseline
OS NOx
Budget
Without
Gen
Shifting
Budget
With Gen.
Shifting
% Reduction from
Generation Shifting
as a Percentage of
Baseline
2023
236,363
217,961
210,297
3%
2024
233,849
213,509
205,835
3%
2025
227,962
207,906
200,352
3%
2026
219,017
133,802
129,522
2%
4. Variability Limits
Once EPA determined state-emission budgets representative of the proposed control
stringency, EPA calculated the variability limits and assurance levels for each state based on the
calculated emission budgets. Each state's variability limit is was assumed to be 21% of its
budget, and its assurance level is the sum of its budget and variability limit (or 121% of its
budget).33 The variability limits and assurance levels are further described and shown in section
VII of the preamble for this rule and shown in Table Appendix F-l.
5. Calculating Dynamic Budgets Starting in 2025
The dynamic budgets methodology for 2025 and subsequent years begins with the
engineering analysis used to determine the preset 2024 state budgets and the illustrative 2026
state emissions budgets described above. There are three substantive changes made to the budget
calculation. First, the inventory of existing units in the group 3 program is updated to reflect new
units not known at the time of final rule. Second, the heat input value for individual units is
updated to reflect the latest reported data. Whereas the illustrative budgets rely on 2021 heat
input data as its basis for estimating future EGU operation levels in future years, the dynamic
budget would substitute in the most recent reported heat input data (e.g., 2023 would be used for
2025 budgets). Finally, the dynamic budget calculation would omit any estimation of generation
shifting based reductions as that would be captured through the incorporation of new heat input
data (and corresponding dynamic budget calculations). The methodology to derive the dynamic
budgets is explained below.34
33 As described in Section VII of the Preamble for this rule, the EPA is proposing a variability limit of 21% for 2023
and 2024. Starting in 2025, 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.
34 Emission reductions derived from generation shifting will be captured in the dynamic budgets in all cases. For the
pre-set budget years it is estimated and incorporated through an additional calculation step. For dynamic budget
years, it is directly incorporated through the inclusion of updated heat input data reflecting observed, post-
compliance generation shifting - therefore the need for an "estimation" is mooted.
28
-------
Appendix A: State Emissions Budget Calculations and Engineering Analytics has a worksheet
titled "Dynamic Budget 2025 Template" and another titled "Dynamic Budget 2026+ Template".
These worksheets don't show budgets for those future years, but provide the mechanics and data
fields(some of which are prepopulated if the data point is fixed, some are left blank if to be
populated with future data) to demonstrate how EPA intends to calculate dynamic budgets for a
future year. These worksheets reflect: 1) the initial inventory of EGUs used to derive the ozone
season state emissions budget for each year 2025 and thereafter, 2) the prepopulated unit-level
emission rate and entry space for future heat input data used to estimate unit-level emissions, and
3) the template for summation of the unit-level emission estimates to identify the states dynamic
budget for a future year (omitting any additional generation shifting assumption used in the
illustrative budgets).
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 P of the "dynamic budget 2025" and "dynamic budget
2026+" worksheet
• The inventory of units is comprised of:
o The inventory of units included in the "unit 2024 file" for Group 3 states. These
are all of the existing units assumed during rule promulgation at the time of the
last preset budget year (i.e., 2024). (Note - any unit that subsequently retires is
effectively nulified in the calculations as its heat input value is adjusted to zero in
steps below)
o New units that were not included in the "unit 2024 file", but that commenced
operation and had a deadline for certification of monitoring systems under
§97.1030(b) by Maylst of the latest year of historical data (e.g., by May 1st of
2023 for the 2025 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 unit-files described
above and used in the pre-determined and illustrative budget calculations. EPA applies
the emission rate reflecting the selected control stringency identified and applied for
those illustrative state budgets. For the "dynamic budget 2025" worksheet, these emission
rates come from column AR in the "unit 2024" worksheet, which are calculated by
dividing the unit-level emissions value from column AN into the unit-level heat input
value from column Z in the "unit 2024" worksheet.35 The use of the "unit 2024 file"
emission rate value is consistent with the notion that no additional mitigation measures
are assumed in 2025. These unit-level emission rate reflects the control stringency
identified in EPA's determination of significant contribution applied to these units in
35 This emissions value is multiplied by 2000 to convert tons to pounds. Therefore, the emission rate is expressed in
a lb/MMBtu metric.
29
-------
2025. For the "dynamic budget 2026+" worksheet, these emission rates come from
column AS in the "unit 2026" worksheet, which are calculated by dividing the unit-level
emissions value from column AP into the unit-level heat input value from column Z in
the "unit 2026" worksheet. This value is also shown in column AS in the "unit 2026"
worksheet. The "unit 2026" worksheet reflects the lower emission rate for some units
where post-combustion control retrofit potential is identified.36
• 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 Proposed Rule TSD:
Applied New Unit Emission Rates for Dynamic Budgets
Unit Type
Assumed Emission Rate (lb/MMBtu)
Coal Steam
0.05
Oil/Gas steam
0.03
Combustion Turbine
0.03
Combined Cycle
0.012
All other fossil
0.05
o For 2021 non-operating units (and thus no 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. If that rate exceeds the assumed
step 3 technology in effect for that year (e.g., SCR optimization in 2025 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 Q of the "dynamic
budget 2025" and "dynamic budget 2026+" worksheet.
o This step is completed at the time of promulgation of this rulemaking, and
therefore these rates (reflecting the removal of significant contribution) are
determined and published in the rule's promulgation. This variable is not
dynamic.
• Column R in the "dynamic budget" worksheet 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, this column would be populated with
36 The emission rate for Alabama, Delaware, and Tennessee continue to be identified by column AR, rather than AS,
at this step as those states are not subject to the post-combustion control stringency assumptions
30
-------
reported 2023 heat input for the "dynamic budget 2025" worksheet (as 2023 data will be
the latest available data at the time of deriving the 2025 budget). For the "dynamic budget
2026+" worksheet, this column will be populated with the latest reported heat input value
for the identified unit for each budget year 2026 and beyond. When applied to derive
2026 budgets, this column would be populated with reported 2024 heat input data, when
applied to derive 2027 budgets it would be populated with reported 2025 heat input data,
and so forth.
o Any unit included in the inventories identified for "dynamic budget 2025" and
"dynamic budget 2026+" worksheets for which reported heat input data from the
most recent historical year is not available due to the fact that the unit was not yet
monitoring and reporting at the start of that data year (e.g., 2023), then EPA will
continue to rely on the same heat input value used in the "unit 2024" worksheet.
• Column S reflects the unit-level assumed emissions. This value will be obtained by
multiplying the emission rate (in column Q) by the heat input value (column R). 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 S of each "dynamic budget" worksheet are summed to the state level. These states
(those 25 covered for EGU Group 3 under this action) and state-level values (in tons) are
displayed in columns Y and Z of the same "dynamic budget" worksheet. These tonnage
values in column Z reflect the state budgets for the given year (starting in 2025). At this step,
a rounding fuction is applied to express the values to the nearest ton. These state tonnage
totals (i.e., budgets) are made public and implemented approximately 1 year prior to their
vintage year (e.g., 2025 budgets will be announced prior to the summer of 2024) through the
schedule identified in Section VII of the preamble.
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 VI.B-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 determining the
amount of each state's emission reduction obligation is the predicted downwind ambient air
quality impacts of the various levels of NOx emission control assessed for upwind EGU and
non-EGU sources. See sections A and B of this TSD for information regarding EGUs and see
preamble section VLB.2 and VI.C.2 and the Non-EGU Screening Assessment TSD for
information about non-EGUs. The emission reductions associated with the various cost
thresholds analyzed for this proposed rule are expected to result in different amounts of air
quality improvement at the downwind receptors. The downwind air quality impacts are used to
inform EPA's assessment of potential overcontrol, as discussed in more detail below.
31
-------
Air quality modeling would be the optimal way to estimate the air quality impacts at each
cost threshold level from EGUs and non-EGUs 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).37 The simplified tool
allows the Agency to analyze many more levels of NOx control stringency as implemented
through emission budgets than would otherwise be possible. 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 directly informed by air quality modeling data. Further, AQAT has evolved
through iterative development under the original CSAPR, the CSAPR Update, and the Revised
CSAPR Update. One such evolution is its calibration of the change in air quality based on air
quality modeling of a particular emission reduction scenario. Here, EPA continues the
development of the AQAT to make state and receptor specific calibration factors, rather than just
receptor specific calibration factors. EPA examined one of the cost threshold 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=Proposal.xlsx"
excel workbook.38
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 emission cost threshold analyses.
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 changes in
response to emissions changes in order to quantify necessary emission reductions under the good
neighbor provision and to evaluate potential levels of emission control stringency as
implemented through budgets for over-control as to either the 1% threshold or the downwind
receptor status. EPA described and used a similar tool in the original CSAPR to evaluate good
neighbor obligations with respect to the fine particulate matter (PM2.5) NAAQS and in both the
CSAPR Update and final Revised CSAPR Update to evaluate good neighbor obligations with
respect to ozone. For the CSAPR Update, EPA refined both the construction and application of
the assessment tool for use in estimating changes in ozone concentrations in response to changes
in NOx emissions. This methodology was reapplied in the Revised CSAPR Update. Here, we
extend the methodology developed in the CSAPR Update rulemaking and calibrate the response
37 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 2028 cases. EPA also modeled a 2026 case
with air quality contributions where EGU and non-EGU emissions were uniformly reduced by 30%.
38 The AQAT estimates in the workbook are based on EGU emission estimates completed on December 7, 2021 and
may not represent the final emission estimates used in the rule.
32
-------
of a pollutant using two CAMx simulations at different emission levels where we have full sets
of state level emissions and contribution data.39'40
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.41 For
the purposes of developing and using an assessment tool to compare the air quality impacts of
NOx emission reductions under various emission cost threshold emission levels, 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 levels or changes
from year-to-year), a change in ozone season NOx emissions leads to a proportional change in
downwind ozone contributions.42 This proportional relationship was then modified using
calibration factors created based on state-specific source apportionment (i.e., contribution) air
quality modeling of 2023 and 2026 base case emissions and a sensitivity scenario in which 2026
base case EGU and non-EGU NOx emissions were reduced by 30% in each state. The
contributions from the 2026 30% NOx reduction case were applied for cases that examine EGU
or non-EGU emissions reductions, whereas, the 2023 and 2026 base case contribution modeling
results were applied for estimating ozone design values for additional future years, as necessary,
that were not modeled explicitly. The calibration factors are designed to adjust the response of
ozone to emissions changes to reflect the non-linear, non-one-to-one proportional relationship
between changes in NOx emissions and the associated changes in ozone. For example, for a
particular state and receptor in 2026, we could assume that a 20% decrease in the 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 calibration factor (based on the change to
39 In CSAPR, we estimated changes in sulfate using changes in SO2 emissions.
40 In this rule, we used CAMx to calibrate the assessment tool's predicted change in ozone concentrations to changes
in NOx emissions. This calibration is state and receptor-specific and is 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%. For time periods before or after 2026, we used the an 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.
41 This downwind air quality improvement is assumed to be indifferent to the source sector or the location of the
particular emission source within the state where the ton was reduced. For example, reducing one ton of NOx
emissions from the power sector is assumed to have the same downwind ozone reduction as reducing one ton of
NOx emissions from the non-EGU source sector. Similarly, when we are using the alternative calibration factors we
assume that reducing a ton of emissions from the power sector has the same effect as reducing a ton of emissions
from the mobile source sector.
42The relationship between NOx emissions and ozone concentrations is known to be non-linear when examined over
large ranges of NOx emissions (e.g., J.H. Seinfeld and S.N. Pandis, Atmospheric Chemistry and Physics From Air
Pollution to Climate Change, 2nd Edition, John Wiley and Sons, 2006, Hoboken, NJ, pp 236-237). However, for
smaller ranges of NOx and VOC emissions, while meteorological conditions are held constant, the relationship may
be reasonably linear. The nonlinearities are evident over tens of ppb of ozone changes with tens of percent changes
in the overall emission inventories. For most states examined here, under the various control scenarios, most
changes in the emission inventory are on the order of a few percent and most air quality changes are on the order of
a fraction of a ppb. In this assessment tool, we are assuming a linear relationship between NOx emissions and ozone
concentrations calibrated between two CAMx simulations. A significant portion of the nonlinearity is accounted for
by using the calibration factors and having the air quality estimates occur at levels of emissions between the 2026
base case and the other case used in the calibration (which were both modeled in CAMx).
33
-------
the 2026 30% reduction from EGU and non-EGU sources) it may only decrease by 10% in
"calibrated" AQAT (where the calibration factor is 0.5). Typically, the calibration factors were
substantially less than one for the state containing the receptor, often on the order of 0.3 (thus, a
10%) decrease in emissions from a particular state would result in a 3% decrease in the ozone
contribution from that state) and then increased with states that are farther upwind to values
around 1 (where a 10%> reduction in emissions would result in a 10%> decrease in ozone
contribution from the particular state). The reason for this relationship is the difference in
chemical state for the emissions as they cycle between NOx and ozone as they encounter various
oxidative/reductive chemical regimes and meteorological conditions as they are transported. 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.
2. Details on the construction of the ozone AQAT for this proposed rule
(a) Overview of the ozone AQAT
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
predict ozone concentrations at different levels of emission levels at monitoring sites. The
reduction in state-by-state ozone contributions for each year at each cost threshold level and the
resulting air quality improvement at monitoring sites with projected nonattainment and/or
maintenance problems were then considered in a multi-factor test for identifying the level of
emissions reductions that define significant contribution to nonattainment and interference with
maintenance.
In applying AQAT to analyze air quality improvements at a given receptor for 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 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.43
43Here, 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 here. Under this approach,
EPA accounts for what may be considered the downwind state's "fair share." As discussed in the preamble, Section
VII.D, EPA no longer believes it is a necessary part of the "overcontrol" analysis to account for the downwind
state's "fair share." In this regard, we present results in this TSD both with emissions reductions in unlinked home
states (called the "scenario" estimates) and without this assumption (called the "control" estimates). At each receptor
under the "scenario" estimates we only consider the impact of emissions reductions from upwind states that were
linked to that particular receptor while for the "control" estimates we consider the impact of emissions reductions
that are linked to any receptor.
34
-------
Specifically, the key estimates from the ozone AQAT for each receptor are:
• The ozone contribution as a function of emissions at each cost threshold level, for
each upwind state that is contributing above the 1 percent air quality threshold and the
state containing the receptor.
• The ozone contribution under base case NOx emissions 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 base case source apportionment modeling and the 2026 30% EGU
and non-EGU source apportionment modeling scenario.
The results of the ozone AQAT analysis for each emission cost threshold level for EGUs and
non-EGUs can be found in section C.3 of this document.
(b) Data used to construct the ozone AQAT for this rule
Several air quality modeling and emissions inventory sources were used to construct the
calibrated ozone AQAT for this rule. 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 both 2023 and 2026,
EPA tagged anthropogenic emission in each state individually as well as total anthropogenic
emissions in Canada and Mexico combined, emissions from offshore drilling platforms and
shipping, emissions from wild and prescribed fires, biogenic emissions, and boundary conditions
which represent the net contribution from all sources outside the modeling domain. In addition,
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 used IPM emission estimates while the 2016 base year used EGU
continuous emissions monitoring system (CEMS) data. In the ozone AQAT, any emission
differences between the 2026 air quality modeling base case and the 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
discussed in the Preparation of Emissions Inventories for 2016v2 North American Emissions
Modeling Platform TSD. An additional emission scenario in which 2026 base case EGU and
non-EGU emissions were reduced by 30% was also modeled with state-by-state source
apportionment (see the Air Quality Modeling TSD for details). Finally, for each of the EGU and
non-EGU scenarios examined with AQAT, the EGU and non-EGU emissions that were modeled
were replaced with a 2026 EGU and non-EGU emission inventory used within Step 3. 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
35
-------
The ozone AQAT was created and used in a multi-step process. In brief, ozone AQAT
was created using the contributions and emission inventory from the 2023 and 2026 base case air
quality modeling as well as the 2026 30% NOx reduction case to evaluate all policy scenarios.
As a first step, EPA developed calibration factors to (1) estimate ozone concentrations in future
years that were not simulated with air quality modeling and (2) account for the nonlinear
response of ozone to NOx reductions. Ozone concentrations for alternative years were, while not
evaluated at proposal would be based on calibration factors based on the change in ozone
concentrations and contributions between the 2026 base case and the 2023 base case. These
calibration factors are included as a sensitivity analysis (described later). 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
and then multiplied this fractional change by the state and receptor-specific calibration factor as
well as by the state- and receptor-specific contribution. This resulted in a state- and receptor-
specific "calibrated change in contribution" relative to the 2026 base case. Each state's change
in contribution value was then added its 2026 base case contribution and the results summed for
all states for each receptor.44 Next, the receptor-specific base case contributions from the other
source-categories45 were added to the sum of each state's contribution. Note that the
contributions from these other source categories were modified by the ratio of the total change in
anthropogenic contribution from the 2026 base by the total difference between the 2026 base and
the 2026 30% reduction scenario to account for the interaction between changes in US
anthropogenic emissions and ozone principally formed from these other categories. The net
result of these calculations is an estimated design value for each receptor that reflects the
emissions changes associated with each scenario evaluated.46
The 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-state and
receptor-by-receptor basis for every monitor throughout the modeling domain.
(1) Steps to create the 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.
EPA summed the ozone season total anthropogenic NOx emissions across all relevant
source sectors for both the 2026 30% EGU and non-EGU NOx reduction case and the 2023 base
case. EPA calculated the ratio of the anthropogenic emissions for each of these two cases to the
total anthropogenic emissions for the 2026 base case for each state modeled in CAMx. More
information on the emissions inventories can be found in the preamble to the proposed rule. The
total anthropogenic emissions data and resulting fractional reduction ratios can be found in Table
C-l and in the ozone AQAT worksheet "calib emiss". The difference in emissions in the
44 In some cases (where emissions are lower than modeled in the 2026 base case) the change in contribution can be
negative.
45 The other source categories include contributions from anthropogenic emission in Canada and Mexico, emissions
from offshore drilling platforms and shipping, emissions from wild and prescribed fires, biogenic emissions, and
boundary conditions which represent the net contribution from all sources outside the modeling domain.
46 Details on procedures for calculating average and maximum design values can be found in the Air Quality
Modeling TSD.
36
-------
fractional reduction ratio is the OS anthropogenic NOx emissions in the 2026 30% NOx
reduction case minus the OS NOx in the 2026 base case. This difference in tons is then divided
by the 2026 base case emissions, resulting in a "fractional reduction" for the 30% NOx reduction
case (Table C-l). A similar procedure was used to get the fractional reduction ratio for the 2023
base case (except the 30% NOx reduction anthropogenic emissions were replaced by the 2023
base case anthropogenic emissions).
In order to facilitate understanding the next steps of the calibration process, 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-l.
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 (Table C-l for Westport) by the reduction fraction ratio (i.e., thedifference in
emissions as a fraction of the 2026 base case emissions). The fractional reduction ratios are
found in Table C-l. The equation for these calculations is shown in equation 1 for the 30% NOx
case. Equation 1 was also used for developing calibration factors based on 2023, except that the
2023 base case emissions were used instead of the 2026 NOx reduction emissions.
Uncalibrated delta contribution = 2026 contribution x ((2026 30 NOx case anthropogenic
emissions - 2026 base case anthro emissions)/2026 base case anthropogenic emissions) Eqn C-l
Thus, when the 2026 30% NOx reduction case or 2023 base case had lower emissions
than the 2026 base case, the net result was a negative number. Each state's reduction fractional
change in emissions was multiplied by its 2026 base case contribution to get a state and receptor-
specific change in contribution (Table C-2). For each state for each monitor, this change in
concentration is total "uncalibrated" change in concentration.
37
-------
Table C-l. The Total Anthropogenic 2026 Base Case, 2026 w/30% EGU and non-EGU Reduction, and 2023
Base Case NOx Emissions used in the Modeling and Ozone Contributions (ppb) for the Westport Monitor
Number 090019003 in Fairfield County, Connecticut. i
State
Modeled
Modeled
Modeled
2026 30% NOx
2023 Base Case vs
Westport
Westport
Westport
2026 Base
2026 30%
2023 Base
Reduction vs 2026
2026 Base Case
2026 Base
2026 30%
2023 Base
Case NOx
EGU/non-
Case NOx
Base Case
Fractional
Case Ozone
NOx Cut
Case Ozone
Emissions
EGU
Reduction
NOx
Emissions
Emissions
Fractional
Reduction in
Emissions
Reduction in
Emissions
Contributions
Ozone
Contributions
Contributions
Alabama
61,759
52,853
66,312
-0.14
0.07
0.106
0.095
0.111
Arizona
33,463
32,313
38,612
-0.03
0.15
0.013
0.013
0.015
Arkansas
39,488
35,333
43,202
-0.11
0.09
0.136
0.126
0.148
California
133,629
127,270
139,593
-0.05
0.04
0.033
0.032
0.034
Colorado
49,825
45,877
53,121
-0.08
0.07
0.055
0.052
0.058
Connecticut
10,887
10,256
11,820
-0.06
0.09
2.861
2.879
2.959
Delaware
6,447
6,135
6,878
-0.05
0.07
0.423
0.410
0.431
District of
1,302
1,245
1,390
-0.04
0.07
0.037
0.036
0.038
Columbia
Florida
92,166
84,786
100,080
-0.08
0.09
0.063
0.058
0.067
Georgia
60,266
55,302
67,589
-0.08
0.12
0.140
0.133
0.154
Idaho
17,321
16,296
19,622
-0.06
0.13
0.027
0.026
0.030
Illinois
91,069
83,536
97,086
-0.08
0.07
0.512
0.490
0.530
Indiana
68,291
59,091
73,491
-0.13
0.08
0.716
0.671
0.760
Iowa
41,049
36,033
46,836
-0.12
0.14
0.109
0.101
0.122
Kansas
59,107
53,798
62,587
-0.09
0.06
0.095
0.090
0.099
Kentucky
50,887
43,739
54,506
-0.14
0.07
0.802
0.721
0.830
Louisiana
100,361
86,348
103,038
-0.14
0.03
0.250
0.225
0.256
Maine
12,918
11,982
14,097
-0.07
0.09
0.016
0.016
0.017
Maryland
23,671
22,513
25,735
-0.05
0.09
1.088
1.063
1.140
Massachusetts
26,353
25,321
28,105
-0.04
0.07
0.298
0.293
0.308
Michigan
75,940
66,736
80,760
-0.12
0.06
0.881
0.815
0.922
Minnesota
55,972
49,439
62,656
-0.12
0.12
0.137
0.124
0.148
Mississippi
33,156
29,336
34,435
-0.12
0.04
0.095
0.088
0.100
Missouri
67,664
60,958
76,251
-0.10
0.13
0.284
0.264
0.312
Montana
25,642
23,333
28,408
-0.09
0.11
0.074
0.068
0.081
Nebraska
38,322
34,126
43,826
-0.11
0.14
0.059
0.055
0.066
Nevada
16,178
14,980
18,286
-0.07
0.13
0.011
0.010
0.012
New
6,719
6,596
7,287
-0.02
0.08
0.096
0.096
0.103
Hampshire
New Jersey
31,805
30,607
34,476
-0.04
0.08
8.550
8.609
8.855
New Mexico
62,210
58,527
65,186
-0.06
0.05
0.048
0.046
0.050
New York
65,642
61,970
69,960
-0.06
0.07
14.186
14.100
14.365
North Carolina
51,986
46,303
58,908
-0.11
0.13
0.388
0.359
0.438
North Dakota
55,294
52,126
59,167
-0.06
0.07
0.098
0.094
0.103
Ohio
78,681
70,003
85,480
-0.11
0.09
1.787
1.663
1.901
Oklahoma
83,411
76,046
90,114
-0.09
0.08
0.146
0.137
0.154
Oregon
29,345
27,680
33,155
-0.06
0.13
0.028
0.027
0.031
Pennsylvania
103,565
95,081
107,022
-0.08
0.03
6.829
6.450
6.905
Rhode Island
4,187
4,011
4,559
-0.04
0.09
0.043
0.042
0.045
South Carolina
38,939
34,839
43,650
-0.11
0.12
0.154
0.144
0.169
South Dakota
11,084
10,494
12,972
-0.05
0.17
0.037
0.036
0.043
Tennessee
47,475
43,303
52,389
-0.09
0.10
0.253
0.241
0.275
Texas
280,717
261,613
305,019
-0.07
0.09
0.496
0.475
0.536
Utah
29,762
26,807
35,692
-0.10
0.20
0.029
0.027
0.034
Vermont
3,378
3,363
3,853
0.00
0.14
0.025
0.025
0.028
Virginia
46,496
43,302
50,590
-0.07
0.09
1.131
1.092
1.194
Washington
47,754
45,338
53,412
-0.05
0.12
0.047
0.046
0.053
West Virginia
39,500
35,285
43,830
-0.11
0.11
1.233
1.134
1.342
Wisconsin
41,032
37,456
45,503
-0.09
0.11
0.154
0.146
0.165
Wyoming
32,928
28,322
34,211
-0.14
0.04
0.068
0.061
0.070
Tribal Data
4,052
3,352
4,057
-0.17
0.00
0.002
0.002
0.002
38
-------
Table C-2. The Uncalibrated Ozone Change (ppb) between the 2026 Base Case and the 2026 w/30% EGU
and non-EGU Reduction Case and the 2023 Base Case, Along with the Change in Ozone (ppb) from the Air
Quality Modeling, as well as the Resulting State-specific Calibration Factors for the Westport Monitor
State
Uncalibrated
Uncalibrated
Modeled
Modeled Ozone
Calibration Factor for EGUs and non-
Calibration Factor for
OzoneChange
Ozone
Ozone
Change
EGUs
Adjusting Years
(2026 to 2026
Change (2026
Change
(2026 to 2023
(Ratio of Modeled Ozone Change to
(Ratio of Modeled Ozone
w/ 30% Cut)
to 2023 Base)
(2026 to
2026 w/
30%
Cut)
Base)
Uncalibrated Ozone Change 2026 to
2026 w/ 30% Cut)
Change to Uncalibrated
Ozone Change 2026 to
2023 Base)
Alabama
-0.015
0.008
-0.010
0.005
0.67
0.69
Arizona
0.000
0.002
0.000
0.002
0.60
0.81
Arkansas
-0.014
0.013
-0.010
0.011
0.70
0.89
California
-0.002
0.001
-0.001
0.001
0.68
0.92
Colorado
-0.004
0.004
-0.004
0.003
0.85
0.84
Connecticut
-0.166
0.245
0.018
0.098
-0.11
0.40
Delaware
-0.020
0.028
-0.013
0.008
0.63
0.27
District of
-0.002
0.003
-0.001
0.001
0.56
0.26
Columbia
Florida
-0.005
0.005
-0.005
0.004
0.89
0.75
Georgia
-0.012
0.017
-0.007
0.013
0.62
0.78
Idaho
-0.002
0.004
-0.001
0.003
0.59
0.90
Illinois
-0.042
0.034
-0.022
0.018
0.52
0.53
Indiana
-0.096
0.055
-0.045
0.044
0.47
0.81
Iowa
-0.013
0.015
-0.009
0.013
0.64
0.83
Kansas
-0.009
0.006
-0.005
0.004
0.58
0.64
Kentucky
-0.113
0.057
-0.082
0.028
0.73
0.48
Louisiana
-0.035
0.007
-0.025
0.006
0.71
0.89
Maine
-0.001
0.001
-0.001
0.001
0.52
0.74
Maryland
-0.053
0.095
-0.025
0.051
0.47
0.54
Massachusetts
-0.012
0.020
-0.005
0.010
0.41
0.50
Michigan
-0.107
0.056
-0.066
0.040
0.62
0.72
Minnesota
-0.016
0.016
-0.013
0.010
0.80
0.63
Mississippi
-0.011
0.004
-0.007
0.005
0.66
1.24
Missouri
-0.028
0.036
-0.020
0.028
0.70
0.78
Montana
-0.007
0.008
-0.006
0.007
0.92
0.90
Nebraska
-0.006
0.009
-0.004
0.007
0.61
0.78
Nevada
-0.001
0.001
-0.001
0.001
0.72
0.86
New
-0.002
0.008
0.000
0.006
0.03
0.80
Hampshire
New Jersey
-0.322
0.718
0.060
0.305
-0.18
0.43
New Mexico
-0.003
0.002
-0.002
0.002
0.75
0.87
New York
-0.794
0.933
-0.086
0.179
0.11
0.19
North Carolina
-0.042
0.052
-0.029
0.050
0.69
0.96
North Dakota
-0.006
0.007
-0.004
0.005
0.66
0.71
Ohio
-0.197
0.154
-0.124
0.114
0.63
0.74
Oklahoma
-0.013
0.012
-0.009
0.008
0.67
0.69
Oregon
-0.002
0.004
-0.001
0.003
0.75
0.79
Pennsylvania
-0.559
0.228
-0.378
0.076
0.68
0.33
Rhode Island
-0.002
0.004
-0.001
0.002
0.46
0.49
South Carolina
-0.016
0.019
-0.010
0.015
0.63
0.81
South Dakota
-0.002
0.006
-0.001
0.006
0.43
0.94
Tennessee
-0.022
0.026
-0.012
0.022
0.53
0.86
Texas
-0.034
0.043
-0.021
0.040
0.62
0.93
Utah
-0.003
0.006
-0.002
0.005
0.70
0.79
Vermont
0.000
0.003
0.000
0.003
-2.37
0.80
Virginia
-0.078
0.100
-0.039
0.063
0.50
0.64
Washington
-0.002
0.006
-0.001
0.005
0.50
0.94
West Virginia
-0.132
0.135
-0.098
0.110
0.75
0.81
Wisconsin
-0.013
0.017
-0.008
0.011
0.57
0.68
Wyoming
-0.010
0.003
-0.008
0.002
0.79
0.85
Tribal Data
0.000
0.000
0.000
0.000
1.01
-10.30
39
-------
Next, the estimate of the state and monitor specific ozone responses under the 2026 30%
NOx reduction case (or the 2023 base case) was used to calibrate the ozone AQAT to CAMx and
to derive the calibration factors. One set of factors was created using the 2026 30% NOx
reduction case and is applied to all scenarios where EGU and/or non-EGUs were reduced, the
other set of factors was created using the 2023 base and is applied to estimate base case ozone
contributions in other alternative years (as well as for a sensitivity study). First, the changes in
ozone predicted by the ozone AQAT and CAMx for the average design values were calculated
for each state and each monitor for the 2026 30% NOx reduction case or the 2023 base case air
quality contributions relative to the 2026 base case concentrations. The change in modeled
ozone (i.e., the difference between the 2023 and 2026 base case state-specifc contributions) was
then divided by the change in ozone predicted by the uncalibrated AQAT, resulting in state and
monitor-specific calibration factors (see Table C-2 for an example calculation using the two
cases for the Westport CT monitor 090019003 in Fairfield County). The calculation of these
state and monitor-specific calibration factors provided EPA with the ability to align the ozone
response predicted by the ozone AQAT to the ozone response predicted by CAMx for EGUs and
non-EGUs (based on the factors for the 30% reduction scenario) and to translate the base to
alternative years (based on the factors for the 2023 base case scenario)47.
The ozone AQAT calibration factors for all monitors can be found in the
"OzoneAQATProposal.xlsx" excel workbook in columns I through BF, on worksheets
"2026to2026w30_calib_(rec, stat)" and "2026to2023_calib_(rec, state)" for the two cases. The
calibration factor, multiplied by the fractional change in emissions (relative to the 2026) base and
multiplied by the 2026 base air quality contribution, results in the fractional change in air quality
contribution for any alternative scenario.
The final step in the creation of the calibration factors is to make an adjustment to all the
other air quality source apportionment categories that are not being directly varied within the
tool. This includes contributions from anthropogenic emission in Canada and Mexico, emissions
from offshore drilling platforms and shipping, emissions from wild and prescribed fires, biogenic
emissions, and boundary conditions which represent the net contribution from all sources outside
the modeling domain. In previous versions of AQAT, these contributions were held fixed at the
base case values. For this proposed rule, because we have full source apportionment estimates
for both calibration cases, we are able to adjust these contributions. We do this based on
multiplying the change in the total anthropogenic contribution between the scenario and the base
case by the ratio of the change from the sum of all other contributions divided by the change in
the total anthropogenic contribution. For example, for the Westport CT receptor the difference
between the 2026 and the 2026 30% reduction case was 0.24275 ppb for the all other
contributions and -1.14287 ppb for the anthropogenic contributions, resulting in a ratio of -
0.2124. In the 2026 engineering base, the total anthropogenic total was 45.4428 (compared to a
2026 modeled base value of 45.15215 ppb). The difference between these values was multiplied
by the ratio to get a calibrated change in the "all other" contributions of -0.0617 ppb. Thus, the
"all other" contribution changed from 29.44759 ppb to 29.3859 ppb.
Noting that EPA did not use these calibration factors since EPA only evaluated 2023 and 2026.
40
-------
(2) Create a calibrated version of the ozone AO AT for emission control stringency level and
associated emissions budget analysis for the proposed rule
Next, EPA examined the changes in the 2026 air quality contributions from changes in
EGU and non-EGU emissions for various scenarios relative to the 2026 base case emissions
(while using the calibration factors). This calibrated AQAT was used for each emission cost
threshold level evaluated for EGUs and non-EGUs. For 2023 simulations, EPA started with the
2023 contributions and adjusted them using the 2026 calibration factors with the 30% NOx
reduction from EGUs and non-EGUs, the change in emissions relative to the 2026 base
emissions which would be applied to the 2026 base case contributions.48
First, as described in sections A and B of this TSD for EGUs, EPA identified various cost
threshold levels of emissions based on projected changes in emissions rates and adjusted
historical data. For each state, for each year, the total anthropogenic NOx emissions (excluding
the EGU emissions) are presented in Table C-3.
The EGU point 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 nonCEM units). Within the air quality modeling platform,
different approaches are taken depending on whether an emissions inventory for EGUs is created
using an IPM-based emission estimates or an engineering analysis based platform. The nonCEM
components for the 2016 base air quality model platform using EGU emissions based on CEMS,
and the 2023 and 2026 air quality modeling cases based on IPM EGU emissions are shown in
Table C-3. For each cost threshold engineering analysis based estimate examined in AQAT, a
constant engineering-based nonCEM point EGU component was created from the 2016 air
quality modeling platform and added to the engineering analysis cost threshold values. For
scenarios where we would directly use IPM results, we would apply either the 2023 or 2026
nonCEM component from the air quality model platform. For 2023 and 2026, we show EGU
emissions for units with CEMS as a function emissions control stringency level (see Tables B-l
through B-5 for the years 2023 through 2026, respectively). These levels include:
• Engineering Baseline,
• Optimize SCR,
• Optimize SCR + State-of-the-Art Combustion Controls (referred to as SOA CC),
• Optimize SNCR+ SCR ,
• Optimize SNCR+ SCR + SOA CC ,
• New SCR/SNCR + Optimize SNCR+ SCR + SOA CC.
In the construction of AQAT, for each scenario, we assembled an emission inventory
from all anthropogenic sources for each state. In other words, we combine the year-specific
48 For other years outside of 2023 and 2026, while not examined in this proposed rule, EPA
would first use the 2026 base to 2023 base calibration factors to adjust to a different base year,
then layer on additional adjustments using the 2026 based EGU and non-EGU AQAT changes as
was done for 2023.
41
-------
anthropogenic emissions from Table C-3, with the relevant EGU nonCEM component from C-3,
and one of the EGU CEM estimates from Tables B-l through B-5.
Finally, these emission totals are compared to the 2026 base case that was included in the
air quality modeling. For each emission cost threshold level, EPA calculated the ratio of the
emission differences from the scenario and the 2026 air quality modeling base case to the total
NOx emissions for the 2026 air quality modeling base case used in the air quality modeling for
each state (see Tables C-4 and C-5). Scenarios that are not viable, for technical or policy reasons,
have been grayed out in these tables.
In Tables C-3 and C-5, respectively, we examined the emission reduction for non-EGUs
in Tier 1 and Tier 2, and then estimated the ratio of the emission difference relative to the 2026
air quality modeling base case.
For each scenario analyzed, 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 "linked" to that receptor or if the receptor is located within the state. States that
are contributing above the air quality threshold (i.e., greater than or equal to 1 percent of the
NAAQS) to the monitor, as well as the state containing the monitor (regardless of whether that
state linked to another monitor and regardless of whether it contributions equal to 1 percent of
the NAAQS or not), make NOx emission reductions 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.
For the control case scenarios, all states that were linked to any receptor in 2023 or in
2026 were simultaneously adjusted to the emission levels in the control case, regardless of
whether (or not) the state was "linked" to a particular receptor. In these control scenarios, the
state containing the monitor was only adjusted if it was linked to a monitor in another state. This
scenario examines the emission results when budgets have been applied to the geography and
can be used to show that emission reductions made for states that are not linked 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. 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-4 or C-5 for either the emission cost threshold level or the
engineering base case emission level depending on whether the state is linked in 2023 or 2026).49
This 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 after
implementation of the emission at a particular cost threshold level.
For each monitor, these state-level "calibrated" contributions are then summed to
estimate total ozone contribution from the states to a particular receptor. Finally, "other" ozone
contributions, as described in section C.2.(b), above are added to the state contributions to
account for other sources of ozone affecting the receptor. The change in the "other" ozone
concentrations are estimated by multiplying the change in the anthropogenic total between the
scenario and the base case by the "nonState" calibration factors (calculated as the ratio of the
49 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.
42
-------
change from these all other contributions divided by the change in the total anthropogenic
contribution from the 2026 base case to the 2026 30% reduction case).50 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 to average design values.
Generally, as the emission cost threshold stringency increased, the estimated average and
maximum design values at each receptor decreased. In the assessment tool, the estimated value
of the average design value was used to estimate whether the location will be out of attainment,
while the estimated maximum design value was used to estimate whether the location will have
problems maintaining the NAAQS. The area was noted as having a nonattainment or
maintenance issue if either estimated air quality level was greater than or equal to 71 ppb.
50 See column CB in "Scenario_2023" or "Scenario_2026" in the Ozone AQAT Proposed Rule Excel file
43
-------
Table C-3. Ozone Season Anthropogenic NOx Emissions (Tons) without EGUs for Each
State for 2023 and 2026, the nonCEM EGU Emissions from 2016, 2023, and 2026, and non-
EGU Tier 1 and
State
2023
2026
2016
2023 IPM
2026 IPM
non-EGU
non-EGU
nonCEM
nonCEM
nonCEM
Tier 1
Tier 1+Tier
EGU
EGU
EGU
(tons)
2 (tons)
Emissions
Emissions
Emissions
(tons)
(tons)
(tons)
Alabama
60,935
55,559
482
447
454
-
-
Arizona
37,335
32,124
367
712
771
1,158
1,158
Arkansas
37,177
33,905
141
744
764
922
1,654
California
133,627
127,011
2,059
5,425
5,989
1,598
1,666
Colorado
47,331
43,944
334
1,883
1,919
605
605
Connecticut
10,117
9,215
1,272
1,542
1,529
-
-
Delaware
6,696
6,243
80
108
128
-
-
District of Columbia
1,372
1,283
0
18
18
-
-
Florida
88,929
80,635
5,810
6,964
7,007
-
-
Georgia
63,965
57,183
1,620
779
860
-
-
Idaho
19,258
16,946
528
118
118
-
-
Illinois
89,028
82,830
55
2,505
2,614
2,452
2,452
Indiana
62,476
57,227
611
1,234
1,061
2,787
3,175
Iowa
37,064
33,062
635
833
879
-
-
Kansas
59,950
55,935
109
706
1,162
-
-
Kentucky
42,436
38,993
1
437
495
2,291
2,291
Louisiana
93,619
89,483
3,885
2,943
3,056
4,121
6,769
Maine
12,706
11,693
1,972
1,167
1,131
-
-
Maryland
24,204
22,185
901
1,287
1,400
45
45
Massachusetts
25,975
24,237
2,566
1,960
1,963
-
-
Michigan
68,824
64,256
1,367
4,044
4,262
2,731
2,731
Minnesota
56,445
51,785
1,740
1,821
1,822
673
673
Mississippi
31,505
29,533
1,726
830
833
1,577
1,761
Missouri
64,300
57,595
471
355
481
3,103
3,103
Montana
24,522
21,735
933
105
115
-
-
Nebraska
33,201
29,654
665
549
547
-
-
Nevada
16,753
14,604
155
1,222
1,209
-
-
New Hampshire
7,041
6,496
374
236
222
-
-
New Jersey
32,531
29,836
1,083
1,667
1,572
-
-
New Mexico
64,011
60,945
98
201
211
-
-
New York
63,577
59,403
1,996
4,297
4,348
389
500
North Carolina
49,950
44,822
740
1,837
1,863
-
-
North Dakota
50,581
47,243
156
296
309
-
-
Ohio
75,112
69,485
722
1,029
1,134
2,611
2,790
Oklahoma
87,314
81,005
1
791
834
3,575
3,575
Oregon
31,977
28,226
712
1,174
1,115
-
-
Pennsylvania
96,364
91,144
2,187
5,188
6,484
3,132
3,284
Rhode Island
4,324
3,954
35
212
210
-
-
South Carolina
39,072
35,678
604
961
914
-
-
South Dakota
12,519
10,606
30
46
68
-
-
Tennessee
51,273
46,686
7
441
516
-
-
Texas
279,623
264,173
1,996
5,603
5,385
4,440
4,440
Utah
28,574
26,189
561
604
679
757
757
Vermont
3,807
3,376
61
46
2
-
-
Virginia
47,506
42,891
2,995
2,196
2,371
1,465
1,563
Washington
52,406
46,560
1,536
1,006
1,193
-
-
West Virginia
32,640
32,020
1
591
524
982
982
Wisconsin
42,620
38,936
61
723
712
677
2,150
Wyoming
29,310
27,910
11
904
81
826
826
Tribal Data
2,728
2,730
50
0
0
-
-
44
-------
Table C-4. 2023 Fractional Difference in Emissions Relative to 2026 Air Quality Modeling
Base Case for Each State.
State
Engineering
Baseline
Optimize
SCR
Optimize SCJc
+ SOAC£/
Optimize
SNCR+SCR
Optimize /
SNCR+ SPR +
SOACT
New
SCR/SNCfT +
Optimize
SN0i+ SCR +
/SDA CC
Alabama
0.03
0.02
0.02
0.02
0.02
0.01
Arizona
0.20
0.20
0.20
0.20
0.19
0.08
Arkansas
0.08
0.08
0.08
0.08
0.08
-0.05
California
-0.02
-0.02
-0.02
-0.02
-0.02
-0.02
Colorado
0.02
0.02
0.02
0.02
0.02
-0.04
Connecticut
0.00
-0.01
-0.01
-0.01
-0.01
-0.01
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
0.05
0.05
0.05
0.05
0.03
Georgia
0.06
0.06
0.06
0.06
0.06
0.06
Idaho
0.02
0.02
0.02
0.02
0.02
0.02
Illinois
0.00
0.00
0.00
-0.01
-0.01
-0.02
Indiana
0.03
0.02
0.02
0.02
0.02
0.00
Iowa
-0.01
-0.01
-0.01
-0.01
-0.01
-0.14
Kansas
0.06
0.05
0.05
0.05
0.05
0.01
Kentucky
0.04
-0.01
-0.02
-0.01
-0.02
-0.09
Louisiana
0.09
0.04
0.04
0.04
0.04
-0.02
Maine
0.05
0.05
0.05
0.05
0.05
0.05
Maryland
0.03
0.02
0.02
0.02
0.02
0.02
Massachusetts
0.03
0.03
0.03
0.03
0.03
0.03
Michigan
-0.01
0.00
0.00
0.00
0.00
-0.05
Minnesota
-0.01
-0.01
-0.01
-0.01
-0.01
-0.03
Mississippi
0.12
0.12
0.10
0.12
0.10
0.02
Missouri
0.13
0.00
0.00
0.00
0.00
-0.05
Montana
0.00
0.00
0.00
0.00
0.00
-0.07
Nebraska
-0.03
-0.03
-0.04
-0.03
-0.04
-0.18
Nevada
0.06
0.06
0.06
0.06
0.06
-0.02
New Hampshire
0.06
0.05
0.05
0.05
0.05
0.05
New Jersey
0.01
0.01
0.01
0.01
0.01
0.01
New Mexico
0.00
0.00
0.00
0.00
0.00
-0.01
New York
-0.01
-0.01
-0.01
-0.01
-0.01
-0.02
North Carolina
0.08
0.02
0.02
0.02
0.02
-0.03
North Dakota
0.02
0.02
0.02
0.01
0.01
-0.12
Ohio
0.00
-0.02
-0.02
-0.02
-0.02
-0.02
Oklahoma
0.09
0.09
0.08
0.09
0.08
0.02
Oregon
0.00
-0.01
-0.01
-0.01
-0.01
-0.01
Pennsylvania
0.00
0.00
0.00
0.00
0.00
-0.01
Rhode Island
0.02
-0.01
-0.01
-0.01
-0.01
-0.01
South Carolina
0.01
-0.01
-0.01
-0.01
-0.01
-0.01
South Dakota
0.01
0.01
0.01
0.01
0.01
0.01
Tennessee
0.07
0.07
0.07
0.07
0.07
0.07
Texas
0.07
0.06
0.06
0.06
0.06
0.00
Utah
0.24
0.22
0.22
0.22
0.22
-0.12
Vermont
0.02
0.02
0.02
0.02
0.02
0.02
Virginia
0.08
0.07
0.07
0.07
0.06
0.06
Washington
0.05
0.05
0.05
0.05
0.05
0.02
West Virginia
0.04
0.07
0.06
0.07
0.05
-0.01
Wisconsin
0.08
0.08
0.08
0.08
0.08
0.06
Wyoming
0.16
0.13
0.11
0.13
0.11
-0.01
Tribal Data
0.44
0.42
0.42
0.42
0.42
0.08
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
45
-------
Table C-5. 2026 Fractional Difference in Emissions Relative to 2026 Air Quality Modeling
Base Case for Each State.
State
Engineering
Optimize
Optimize
Optimize
Optimize
New
non-EGU
non-EGU
Baseline
SCR
SCR + SOA
SNCR+SCR
SNCR+ SCR
SCR/SNCR +
Tier 1 +New
Tier 1 +Tier
CC
+ SOA CC
Optimize
SNCR+ SCR
+ SOA CC
SCR/SNCR +
Optimize
SNCR+SCR
+ SOA CC
2 +New
SCR/SNCR +
Optimize
SNCR+SCR
+ SOA CC
Alabama
0.02
0.01
0.01
0.01
0.01
0.00
0.00
0.00
Arizona
0.13
0.12
0.12
0.12
0.12
0.06
0.03
0.03
Arkansas
0.09
0.09
0.09
0.09
0.09
-0.04
-0.06
-0.08
California
-0.02
-0.03
-0.03
-0.03
-0.03
-0.03
-0.04
-0.04
Colorado
0.01
0.01
0.01
0.01
0.01
-0.04
-0.05
-0.05
Connecticut
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Delaware
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
District of Columbia
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Florida
0.06
0.04
0.04
0.04
0.04
0.02
0.02
0.02
Georgia
0.07
0.07
0.07
0.07
0.07
0.06
0.06
0.06
Idaho
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
Illinois
0.00
-0.01
-0.01
-0.01
-0.01
-0.02
-0.05
-0.05
Indiana
-0.01
-0.02
-0.02
-0.02
-0.02
-0.03
-0.07
-0.08
Iowa
0.04
0.04
0.04
0.04
0.04
-0.09
-0.09
-0.09
Kansas
0.05
0.04
0.04
0.04
0.04
0.01
0.01
0.01
Kentucky
0.04
-0.02
-0.03
-0.02
-0.03
-0.09
-0.13
-0.13
Louisiana
0.08
0.02
0.02
0.02
0.02
-0.03
-0.07
-0.10
Maine
0.07
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
0.03
0.03
Massachusetts
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
Michigan
-0.01
0.00
0.00
0.00
0.00
-0.04
-0.08
-0.08
Minnesota
0.03
0.03
0.03
0.03
0.03
0.00
-0.01
-0.01
Mississippi
0.10
0.09
0.08
0.09
0.08
0.00
-0.05
-0.05
Missouri
0.16
0.03
0.03
0.03
0.03
-0.02
-0.07
-0.07
Montana
0.00
0.00
0.00
0.00
0.00
-0.07
-0.07
-0.07
Nebraska
0.01
0.01
0.00
0.01
0.00
-0.13
-0.13
-0.13
Nevada
0.06
0.06
0.06
0.06
0.06
-0.01
-0.01
-0.01
New Hampshire
0.06
0.05
0.05
0.05
0.05
0.05
0.05
0.05
New Jersey
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
New Mexico
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
New York
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.02
-0.02
North Carolina
0.11
0.05
0.05
0.05
0.05
0.00
0.00
0.00
North Dakota
0.03
0.03
0.03
0.02
0.02
-0.11
-0.11
-0.11
Ohio
0.02
0.00
0.00
0.00
0.00
0.00
-0.03
-0.04
Oklahoma
0.10
0.09
0.09
0.09
0.09
0.02
-0.02
-0.02
Oregon
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Pennsylvania
-0.02
-0.02
-0.02
-0.02
-0.02
-0.03
-0.06
-0.06
Rhode Island
0.02
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
South Carolina
0.04
0.02
0.02
0.02
0.02
0.02
0.02
0.02
South Dakota
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Tennessee
0.07
0.07
0.07
0.07
0.07
0.07
0.07
0.07
Texas
0.09
0.08
0.08
0.08
0.08
0.03
0.01
0.01
Utah
0.16
0.15
0.15
0.15
0.15
-0.03
-0.05
-0.05
Vermont
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
Virginia
0.07
0.05
0.05
0.05
0.05
0.04
0.01
0.01
Washington
0.05
0.04
0.04
0.04
0.04
0.02
0.02
0.02
West Virginia
0.14
0.17
0.15
0.16
0.15
0.08
0.06
0.06
Wisconsin
0.04
0.04
0.04
0.04
0.04
0.04
0.02
-0.02
Wyoming
0.16
0.13
0.11
0.13
0.11
-0.02
-0.04
-0.04
Tribal Data
0.44
0.43
0.43
0.43
0.43
0.08
0.08
0.08
46
-------
3. Description of the analytic results.
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 levels. At each cost
threshold level, using AQAT, for each receptor, 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). EPA also examined each state's air quality contributions at each emission
level, assessing whether a state maintained at least one linkage (i.e., greater than or equal to 1%
(0.70 ppb) to a receptor that was estimated to remain in nonattainment and/or maintenance. EPA
examined the engineering base case, $l,600/ton, $l,800/ton, $11,000/ton and non-EGU Tier 1
and Tier 1+Tier 2 scenarios. Some of the EGU scenarios include emissions with and without
installation of state of the art combustion controls. EPA assessed changes in air quality for the
Tier 1 and Tier 2 non-EGU scenarios for 2026. In these cases, we included EGU emission
reductions at the $11,000/ton cost threshold level when SCRs were installed. 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.
For each year, the average and maximum design values (in ppb) estimated using the
assessment tool for each identified receptor for each cost threshold level have been rounded to
hundredths of a ppb and can be found in Tables C-6 through C-9. There are 29 receptors in 2023
and 22 receptors in 2026. Scenarios that are not viable have been grayed out in these tables.
In 2023, we observe that the Clark County Nevada, monitor 320030075, switches from
maintenance to attainment when existing SCRs are optimized. In other words, its maximum
design value drops below 71 ppb (Table C-7). All other monitors consistently have their average
and/or maximum design values at or above 71 ppb for all viable scenarios.
In 2026, of the 22 receptors, three receptors have their maximum design values drop
below 71 ppb. The maximum design value for monitor 80350004 in Douglas County Colorado
drops below 71 ppb when EGU emission reductions associated with new SCR 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 550590025 in Kenosha County Wisconsin have their maximum design values drop
below 71 ppb when EGU SCRs and non-EGU Tier 1 emission reductions are applied. See Table
C-9 for the values.
In the assessment of air quality using the calibrated assessment tool, we are able to
estimate the change in the air quality contributions of each upwind state to each receptor (see the
description of the state and receptor-specific contributions in section 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 AQAT. For 2023 and 2026, these
results are shown in Tables C-10 and C-l 1, respectively.
To see static air quality contributions and design value estimates for the receptors of
interest for each of the years for each of the cost levels, see the individual worksheets (labeled in
Appendix B). For interactive worksheets, refer to the "202X_scenario" worksheets after setting
47
-------
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-12. For a cost threshold run, cell II would be a value of 0 through
10 (note that 6, 7, and 8 are invalid), while cell 12 should be fixed with a value of 0. Also
included in Table C-12 is a list of the three scenarios used in the RIA. For these scenarios, cell
12 should be set as the same number as cell II. The numbers are 17, 18, or 19 for the proposed
rule, less stringent, or more stringent cases, respectively. Consequently, for each monitor, the
linked, home, and nonlinked states are simulated using the same emission value that represents
the base or policy case for that particular state.
For all linked states, in all years, across all cost threshold levels, we did not see any
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 at all cost threshold levels. 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 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. In some cases, for individual linkages, a state does drop below the
threshold. However, while in these limited cases, an individual linkage to a particular receptor is
resolved, we did not see any instances where all of the linkages across all of the remaining
receptors drop below the linkage threshold. For a few states in 2026, namely Tennessee,
Alabama, and Delaware the receptors to which they were linked in 2023 have their design values
fall below the NAAQS, resulting in these state's maximum remaining air quality contribution to
remaining receptors being below 1% of the NAAQS (Table C-l 1).
Lastly, as an alternative assessment, it was possible to estimate air quality concentrations
in the "control scenario" at each downwind receptor using the ozone AQAT. Here, we apply a
scenario where all states (regardless of whether they are linked to a particular receptor or to a
different receptor in the geography) have the same cost threshold applied as do the "linked"
states. And, for these cases, we kept the states containing the receptor (such as Colorado and
Connecticut) that are not linked to receptors in other states at base case emission levels (rather
than modulate them up to the same threshold as the linked upwind states). This allows us to
assess whether impacts from states that are not specifically linked to a receptor would result in
potential overcontrol. It also allows us to assess whether the assumption that a receptor state
makes "fair share" emission reductions generates any instances of apparent "overcontrol" that is
not actually certain to occur. In general, the differences are relatively small (though, for the
receptors in Colorado to which Wyoming is linked), this difference is larger and it affects
whether or not the receptor has its maximum design value drop below 71 ppb. The average and
maximum design values for 2026 are shown in Tables C-13 and C-14.
48
-------
Table C-6. 2023 Average Ozone DVs (ppb) for NOx Emissions Cost Threshold Levels
($/1
ton) Assessed Using t
ie Ozone AQAT for All Recep
tors.
site
state
county
Engineering
Analysis
Base
SCR
Optimize +
Generation
Shifting
SCR /
Optimize /
SOA Cf/+
Generation
Shifting
SCR
Optimize +
SNCR
Optimize +
Generation
Shifting
SCR /
Optimize /
SOA ( (/+
SNCR
Optimize +
Generation
Shifting
SCR /
Optimize Y
soaco4
SNCR
Optimize +
SCR/SNCR
Retrofit +
/feneration
/ Shifting
40278011
Arizona
Yuma
70.53
70.53
70.53
70.53
70.53
70.50
80350004
Colorado
Douglas
72.35
72.29
72.28
72.29
72.28
71.38
80590006
Colorado
Jefferson
73.23
73.19
73.19
73.19
73.19
72.30
80590011
Colorado
Jefferson
74.41
74.38
74.38
74.38
74.38
73.51
90010017
Connecticut
Fairfield
73.11
73.14
73.14
73.14
73.14
73.04
90013007
Connecticut
Fairfield
74.45
74.47
74.45
74.45
74.44
74.21
90019003
Connecticut
Fairfield
76.30
76.32
76.30
76.31
76.29
76.11
90099002
Connecticut
New Haven
72.11
72.11
72.08
72.09
72.07
71.85
170310001
Illinois
Cook
70.02
70.01
70.01
70.02
70.02
69.85
170310032
Illinois
Cook
70.14
70.15
70.15
70.16
70.15
70.03
170310076
Illinois
Cook
69.64
69.64
69.64
69.65
69.65
69.52
170314201
Illinois
Cook
70.19
70.18
70.18
70.18
70.18
70.05
170317002
Illinois
Cook
70.42
70.34
70.34
70.33
70.33
70.18
320030075
Nevada
Clark
70.09
70.06
70.06
70.06
70.06
69.93
420170012
Pennsylvania
Bucks
71.09
71.07
71.04
71.05
71.03
70.80
480391004
Texas
Brazoria
71.71
71.31
71.30
71.30
71.29
70.04
481210034
Texas
Denton
71.20
71.06
71.04
71.05
71.03
70.50
482010024
Texas
Harris
76.92
76.57
76.57
76.55
76.55
75.30
482010055
Texas
Harris
72.50
72.17
72.15
72.16
72.14
71.00
482011034
Texas
Harris
72.07
71.69
71.69
71.67
71.67
70.28
482011035
Texas
Harris
69.69
69.32
69.32
69.31
69.31
67.98
490110004
Utah
Davis
73.65
73.59
73.59
73.59
73.59
72.58
490353006
Utah
Salt Lake
74.35
74.29
74.29
74.29
74.29
73.27
490353013
Utah
Salt Lake
75.27
75.21
75.21
75.21
75.21
74.04
490570002
Utah
Weber
71.35
71.29
71.29
71.29
71.29
70.29
490571003
Utah
Weber
71.24
71.19
71.19
71.19
71.19
70.19
550590019
Wisconsin
Kenosha
73.17
73.07
73.07
73.07
73.07
72.90
550590025
Wisconsin
Kenosha
69.62
69.47
69.47
69.46
69.46
69.28
551010020
Wisconsin
Racine
71.70
71.61
71.61
71.61
71.61
71.44
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
49
-------
Table C-7. 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 +
Generation
Shifting
SCR /
Optimize/
SOA CO+
Generation
Shifting
SCR
Optimize +
SNCR
Optimize +
Generation
Shifting
SCR /
Optimize/
SOA Ct/+
SNCR
Optimize +
Generation
Sfhifting
SCR /
Optimize -j/
soaccA
SNCR
Optimize +
SCKMSNCR
Retrofit +
/(.enerat ion
/ Shifting
40278011
Arizona
Yuma
72.25
72.24
72.24
72.24
72.24
72.21
80350004
Colorado
Douglas
72.96
72.91
72.89
72.91
72.89
71.98
80590006
Colorado
Jefferson
73.84
73.80
73.80
73.80
73.80
72.91
80590011
Colorado
Jefferson
75.13
75.09
75.09
75.09
75.09
74.21
90010017
Connecticut
Fairfield
73.82
73.86
73.86
73.85
73.85
73.75
90013007
Connecticut
Fairfield
75.37
75.39
75.37
75.37
75.36
75.13
90019003
Connecticut
Fairfield
76.51
76.52
76.51
76.51
76.50
76.32
90099002
Connecticut
New Haven
74.16
74.15
74.13
74.14
74.12
73.89
170310001
Illinois
Cook
73.90
73.89
73.89
73.89
73.89
73.71
170310032
Illinois
Cook
72.78
72.79
72.79
72.80
72.79
72.67
170310076
Illinois
Cook
72.49
72.49
72.49
72.49
72.49
72.37
170314201
Illinois
Cook
73.75
73.74
73.74
73.74
73.74
73.60
170317002
Illinois
Cook
73.37
73.29
73.29
73.29
73.29
73.12
320030075
Nevada
Clark
71.01
70.98
70.98
70.98
70.98
70.84
420170012
Pennsylvania
Bucks
72.63
72.61
72.58
72.59
72.57
72.33
480391004
Texas
Brazoria
73.89
73.48
73.47
73.47
73.45
72.17
481210034
Texas
Denton
73.06
72.91
72.89
72.90
72.89
72.34
482010024
Texas
Harris
78.48
78.12
78.12
78.10
78.10
76.82
482010055
Texas
Harris
73.54
73.20
73.19
73.19
73.17
72.02
482011034
Texas
Harris
73.32
72.93
72.93
72.91
72.91
71.49
482011035
Texas
Harris
73.32
72.93
72.93
72.92
72.92
71.52
490110004
Utah
Davis
75.91
75.85
75.85
75.85
75.85
74.80
490353006
Utah
Salt Lake
75.99
75.93
75.93
75.93
75.93
74.89
490353013
Utah
Salt Lake
75.78
75.72
75.72
75.72
75.72
74.55
490570002
Utah
Weber
73.29
73.23
73.23
73.23
73.23
72.20
490571003
Utah
Weber
72.16
72.11
72.11
72.11
72.11
71.10
550590019
Wisconsin
Kenosha
74.09
73.99
73.99
73.99
73.99
73.81
550590025
Wisconsin
Kenosha
72.69
72.53
72.53
72.52
72.52
72.34
551010020
Wisconsin
Racine
73.64
73.55
73.55
73.55
73.55
73.37
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
50
-------
Table C-8. 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 +
Generation
Shifting
SCR
Optimize +
SOA CC +
Generation
Shifting
SCR
Optimize +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting +
non-EGU
Tier 1
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting +
non-EGU
Tier 1 +
Tier 2
40278011
Arizona
Yuma
70.11
70.10
70.10
70.10
70.10
70.09
70.06
70.06
80350004
Colorado
Douglas
70.94
70.89
70.88
70.89
70.88
70.23
70.07
70.07
80590006
Colorado
Jefferson
72.09
72.05
72.05
72.05
72.05
71.42
71.26
71.26
80590011
Colorado
Jefferson
72.97
72.94
72.94
72.94
72.94
72.32
72.16
72.16
90010017
Connecticut
Fairfield
71.60
71.62
71.62
71.62
71.62
71.52
71.36
71.35
90013007
Connecticut
Fairfield
73.09
73.08
73.07
73.07
73.05
72.84
72.55
72.54
90019003
Connecticut
Fairfield
74.83
74.83
74.81
74.82
74.80
74.63
74.41
74.40
90099002
Connecticut
New
Haven
70.77
70.75
70.73
70.74
70.72
70.51
70.23
70.22
170310001
Illinois
Cook
69.05
69.05
69.05
69.05
69.05
68.96
68.83
68.73
170310032
Illinois
Cook
69.37
69.38
69.37
69.39
69.38
69.32
69.27
69.20
170310076
Illinois
Cook
68.75
68.76
68.76
68.76
68.76
68.71
68.59
68.51
170314201
Illinois
Cook
69.10
69.09
69.09
69.09
69.09
69.02
68.89
68.83
170317002
Illinois
Cook
69.36
69.29
69.29
69.29
69.29
69.18
69.02
68.98
480391004
Texas
Brazoria
70.93
70.54
70.52
70.52
70.51
69.35
68.88
68.72
482010024
Texas
Harris
76.28
75.92
75.92
75.91
75.91
74.77
74.33
74.23
490110004
Utah
Davis
72.20
72.16
72.16
72.16
72.16
71.61
71.51
71.51
490353006
Utah
Salt Lake
73.00
72.96
72.96
72.96
72.96
72.40
72.30
72.30
490353013
Utah
Salt Lake
74.10
74.05
74.05
74.05
74.05
73.45
73.34
73.34
490570002
Utah
Weber
70.30
70.26
70.26
70.26
70.26
69.74
69.64
69.63
550590019
Wisconsin
Kenosha
72.01
71.92
71.92
71.92
71.92
71.80
71.62
71.57
550590025
Wisconsin
Kenosha
68.46
68.32
68.32
68.32
68.32
68.19
67.99
67.95
551010020
Wisconsin
Racine
70.52
70.44
70.44
70.44
70.44
70.33
70.17
70.12
51
-------
Table C-9. 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 +
Generation
Shifting
SCR
Optimize +
SOA CC +
Generation
Shifting
SCR
Optimize +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting +
non-EGU
Tier 1
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting +
non-EGU
Tier 1 +
Tier 2
40278011
Arizona
Yuma
71.81
71.80
71.80
71.80
71.80
71.79
71.76
71.76
80350004
Colorado
Douglas
71.55
71.49
71.48
71.49
71.48
70.83
70.67
70.67
80590006
Colorado
Jefferson
72.69
72.66
72.66
72.66
72.66
72.02
71.86
71.86
80590011
Colorado
Jefferson
73.68
73.65
73.65
73.65
73.65
73.02
72.86
72.86
90010017
Connecticut
Fairfield
72.30
72.32
72.32
72.32
72.32
72.22
72.05
72.04
90013007
Connecticut
Fairfield
73.99
73.99
73.97
73.97
73.96
73.74
73.45
73.43
90019003
Connecticut
Fairfield
75.03
75.03
75.01
75.02
75.00
74.83
74.61
74.59
90099002
Connecticut
New
Haven
72.78
72.76
72.74
72.75
72.73
72.51
72.23
72.21
170310001
Illinois
Cook
72.87
72.87
72.87
72.87
72.87
72.77
72.63
72.53
170310032
Illinois
Cook
71.98
71.99
71.99
72.00
71.99
71.93
71.87
71.80
170310076
Illinois
Cook
71.56
71.57
71.57
71.57
71.57
71.52
71.40
71.31
170314201
Illinois
Cook
72.61
72.60
72.60
72.60
72.60
72.53
72.39
72.32
170317002
Illinois
Cook
72.27
72.20
72.20
72.20
72.20
72.09
71.92
71.88
480391004
Texas
Brazoria
73.09
72.68
72.67
72.67
72.65
71.46
70.97
70.81
482010024
Texas
Harris
77.82
77.46
77.46
77.44
77.44
76.28
75.83
75.73
490110004
Utah
Davis
74.42
74.37
74.37
74.37
74.37
73.81
73.70
73.70
490353006
Utah
Salt Lake
74.61
74.57
74.57
74.57
74.57
74.00
73.90
73.90
490353013
Utah
Salt Lake
74.60
74.56
74.56
74.56
74.56
73.95
73.84
73.84
490570002
Utah
Weber
72.22
72.17
72.17
72.17
72.17
71.64
71.53
71.53
550590019
Wisconsin
Kenosha
72.91
72.83
72.83
72.83
72.83
72.70
72.52
72.47
550590025
Wisconsin
Kenosha
71.48
71.33
71.33
71.32
71.32
71.19
70.98
70.95
551010020
Wisconsin
Racine
72.42
72.35
72.35
72.35
72.35
72.24
72.07
72.02
52
-------
Table C-10. 2023 Maximum Air Quality Contribution (ppb) to a Remaining Receptor.
51
state
Engineering
Analysis Base
SCR Optimize
+ Generation
Shifting
SCR /
Optimize y
SOA wA
Generation
SCR Optimize
+ SNCR
Optimize +
Generation
SCR /
Optimize y
SOACW
SNCR
SCR /
Optimize y
soa cm
SNCR
Shifting
Shifting
Optimize +
Generation
Shifting
Optimize +
SCR/SNCR
Retrofit +
/feneration
/ Shifting
Alabama
0.91
0.90
0.90
0.90
0.90
0.90
Arizona
0.46
0.46
0.46
0.46
0.46
0.46
Arkansas
1.48
1.48
1.48
1.48
1.48
1.34
California
7.33
5.04
5.04
5.04
5.04
5.04
Colorado
0.21
0.21
0.21
0.21
0.21
0.21
Connecticut
0.21
0.21
0.21
0.21
0.21
0.21
Delaware
1.43
1.42
1.42
1.42
1.42
1.42
District of
0.08
0.08
0.08
0.08
0.08
0.08
Columbia
Florida
0.18
0.18
0.18
0.18
0.18
0.18
Georgia
0.18
0.18
0.18
0.18
0.18
0.18
Idaho
0.59
0.59
0.59
0.59
0.59
0.59
Illinois
18.55
18.56
18.56
18.56
18.56
18.57
Indiana
7.20
7.18
7.18
7.18
7.18
7.09
Iowa
0.64
0.64
0.64
0.64
0.64
0.64
Kansas
0.62
0.62
0.62
0.62
0.62
0.62
Kentucky
0.91
0.88
0.87
0.88
0.87
0.82
Louisiana
7.51
7.23
7.23
7.22
7.22
6.95
Maine
0.02
0.02
0.02
0.02
0.02
0.02
Maryland
2.44
2.44
2.44
2.44
2.44
2.44
Massachusetts
0.31
0.31
0.31
0.31
0.31
0.31
Michigan
1.67
1.68
1.68
1.68
1.68
1.64
Minnesota
0.97
0.97
0.97
0.97
0.97
0.96
Mississippi
1.22
1.22
1.21
1.22
1.21
1.16
Missouri
1.81
1.67
1.67
1.67
1.67
1.61
Montana
0.12
0.12
0.12
0.12
0.12
0.12
Nebraska
0.36
0.36
0.36
0.36
0.36
0.36
Nevada
0.94
0.94
0.94
0.94
0.94
0.88
New Hampshire
0.10
0.10
0.10
0.10
0.10
0.10
New Jersey
8.84
8.85
8.85
8.85
8.85
8.85
New Mexico
0.30
0.31
0.30
0.30
0.30
0.30
New York
16.78
16.79
16.79
16.79
16.79
16.77
North Carolina
0.65
0.65
0.65
0.65
0.65
0.65
North Dakota
0.38
0.38
0.38
0.38
0.38
0.38
Ohio
1.95
1.93
1.93
1.93
1.93
1.93
Oklahoma
1.26
1.26
1.25
1.26
1.25
1.20
Oregon
1.10
1.10
1.10
1.10
1.10
1.10
Pennsylvania
6.90
6.93
6.93
6.92
6.92
6.85
Rhode Island
0.05
0.05
0.05
0.05
0.05
0.05
South Carolina
0.20
0.20
0.20
0.20
0.20
0.20
South Dakota
0.10
0.10
0.10
0.10
0.10
0.10
Tennessee
0.97
0.97
0.97
0.97
0.97
0.97
Texas
1.89
1.88
1.88
1.88
1.88
1.81
Utah
1.59
1.63
1.58
1.58
1.58
1.26
Vermont
0.03
0.03
0.03
0.03
0.03
0.03
Virginia
1.86
1.84
1.84
1.84
1.84
1.83
Washington
0.35
0.35
0.35
0.35
0.35
0.35
West Virginia
1.50
1.53
1.52
1.52
1.51
1.44
Wisconsin
2.75
2.75
2.75
2.75
2.75
2.72
Wyoming
0.93
0.91
0.90
0.91
0.90
0.81
Tribal Data
0.08
0.03
0.08
0.08
0.08
0.08
Note: Scenarios that are not viable have had column heads struck through and associated data
has been grayed out and
Values greater than or equal to 0.70 ppb indicate the state remains linked to a remaining downwind receptor.
53
-------
Table C-ll. 2026 Maximum Air Quality Contribution (ppb) to a Remaining Receptor.
State
Engineering
Analysis Base
SCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
Generation
Shifting
SCR
Optimize +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
Generation
Shifting
SCR Optimize
+ SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting
SCR Optimize +
SOA CC +
SNCR Optimize
+ SCR/SNCR
Retrofit +
Generation
Shifting + non-
EGU Tier 1
SCR Optimize +
SOA CC + SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting + non-
EGU Tier 1 + 2
Alabama
0.49
0.49
0.49
0.49
0.49
0.49
0.17
0.17
Arizona
0.39
0.39
0.39
0.39
0.39
0.39
0.39
0.39
Arkansas
1.41
1.40
1.40
1.40
1.40
1.26
0.68
0.68
California
4.79
4.79
4.79
4.79
4.79
4.79
4.76
4.75
Colorado
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
Connecticut
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
Delaware
0.55
0.55
0.55
0.55
0.55
0.55
0.55
0.55
District of
Columbia
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
Florida
0.11
0.11
0.11
0.11
0.11
0.11
0.11
0.11
Georgia
0.17
0.17
0.17
0.17
0.17
0.17
0.16
0.16
Idaho
0.49
0.49
0.49
0.49
0.49
0.49
0.49
0.49
Illinois
18.15
18.15
18.15
18.16
18.16
18.17
17.83
17.83
Indiana
6.96
6.95
6.95
6.95
6.95
6.92
6.81
6.80
Iowa
0.59
0.59
0.59
0.59
0.59
0.59
0.59
0.59
Kansas
0.60
0.60
0.60
0.60
0.60
0.60
0.60
0.60
Kentucky
0.83
0.79
0.79
0.79
0.79
0.75
0.72
0.72
Louisiana
7.38
7.10
7.10
7.09
7.09
6.82
4.03
3.95
Maine
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
Maryland
1.26
1.26
1.26
1.25
1.25
1.25
1.25
1.25
Massachusetts
0.30
0.30
0.30
0.30
0.30
0.30
0.30
0.30
Michigan
1.58
1.59
1.59
1.59
1.59
1.55
1.52
1.52
Minnesota
0.93
0.93
0.93
0.93
0.93
0.91
0.91
0.91
Mississippi
0.97
0.97
0.95
0.97
0.95
0.90
0.40
0.40
Missouri
1.70
1.57
1.57
1.56
1.56
1.51
0.95
0.95
Montana
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
Nebraska
0.23
0.23
0.23
0.23
0.23
0.21
0.21
0.21
Nevada
0.86
0.86
0.86
0.86
0.86
0.80
0.80
0.80
New
Hampshire
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
New Jersey
8.53
8.54
8.54
8.54
8.54
8.54
8.54
8.54
New Mexico
0.29
0.29
0.29
0.29
0.29
0.29
0.29
0.29
New York
16.55
16.57
16.57
16.57
16.57
16.55
16.53
16.53
North Carolina
0.59
0.59
0.59
0.59
0.59
0.59
0.59
0.59
North Dakota
0.35
0.35
0.35
0.35
0.35
0.35
0.35
0.35
Ohio
1.86
1.83
1.83
1.83
1.83
1.84
1.80
1.79
Oklahoma
0.77
0.77
0.77
0.77
0.77
0.74
0.71
0.71
Oregon
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
Pennsylvania
6.75
6.76
6.76
6.75
6.75
6.68
6.54
6.53
Rhode Island
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
South Carolina
0.18
0.18
0.18
0.18
0.18
0.18
0.18
0.18
South Dakota
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
Tennessee
0.36
0.36
0.36
0.36
0.36
0.36
0.26
0.26
Texas
1.81
1.80
1.80
1.80
1.80
1.73
1.62
1.62
Utah
1.34
1.33
1.33
1.33
1.33
0.94
0.92
0.92
Vermont
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
Virginia
1.75
1.74
1.74
1.74
1.74
1.73
1.69
1.69
Washington
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
West Virginia
1.50
1.53
1.52
1.52
1.51
1.44
1.41
1.41
Wisconsin
2.51
2.51
2.51
2.51
2.51
2.50
2.47
2.41
Wyoming
0.91
0.89
0.88
0.89
0.88
0.52
0.52
0.52
Tribal Data
0.08
0.08
0.08
0.08
0.08
0.06
0.06
0.06
52
52 Values greater than or equal to 0.70 ppb indicate the state remains linked to a remaining downwind receptor.
54
-------
Table C-12. Description of the Various Scenarios Modeled in AQAT.
Scenario
Cost Threshold
Level
Description
0
$0
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs
1
$1,600
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs +SCR
optimize + Generation Shifting
2
$1,600
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs +SCR
optimize + SOA CC + Generation Shifting
3
$1,800
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs
+SCR/SNCR optimize + Generation Shifting
4
$1,800
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs
+SCR/SNCR optimize + SOA CC + Generation Shifting
5
$11,000
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs
+SCR/SNCR optimize + SOA CC + Generation Shifting + SCR Retrofit +
Generation Shifting
9
$11,000+ non-
EGU Tier 1
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs
+SCR/SNCR optimize + SOA CC + Generation Shifting + SCR Retrofit +
Generation Shifting + non-EGU Tier 1
10
11,000+ non-
EGU Tier 1 +
Tier 2
Baseline Engineering Analysis 202x OS NOx + engineering nonCEMs
+SCR/SNCR optimize + SOA CC + Generation Shifting + SCR Retrofit +
Generation Shifting + non-EGU Tier 1 + non-EGU Tier 2
17
RIA Proposed
Rule
EGU and non-EGU controls associated with the proposed rule in the RIA.
18
RIA Less
Stringent
EGU and non-EGU controls associated with the less stringent case in the RIA.
19
RIA More
Stringent
EGU and non-EGU controls associated with the more stringent case in the RIA.
55
-------
Table C-l;
i. 2026 Average Ozone DVs
ppb) for
Cach "Control Scenario" Assessed.
Site
State
County
Engineering
Analysis
Base
SCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
Generation
Shifting
SCR
Optimize +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting +
non-EGU
Tier 1
SCR Optimize
+ SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting + non-
EGU Tier 1 +
Tier 2
40278011
Arizona
Yuma
70.11
70.10
70.10
70.10
70.10
70.09
70.07
70.07
80350004
Colorado
Douglas
70.94
70.90
70.88
70.90
70.88
70.56
70.49
70.49
80590006
Colorado
Jefferson
72.09
72.05
72.04
72.05
72.04
71.78
71.72
71.72
80590011
Colorado
Jefferson
72.97
72.94
72.93
72.94
72.93
72.67
72.62
72.62
90010017
Connecticut
Fairfield
71.60
71.58
71.57
71.58
71.57
71.37
71.14
71.12
90013007
Connecticut
Fairfield
73.09
73.04
73.02
73.03
73.01
72.72
72.39
72.37
90019003
Connecticut
Fairfield
74.83
74.79
74.78
74.78
74.77
74.54
74.28
74.26
90099002
Connecticut
New
Haven
70.77
70.71
70.69
70.70
70.68
70.40
70.09
70.07
170310001
Illinois
Cook
69.05
69.01
69.00
69.01
69.00
68.83
68.65
68.54
170310032
Illinois
Cook
69.37
69.34
69.34
69.35
69.34
69.24
69.15
69.08
170310076
Illinois
Cook
68.75
68.73
68.72
68.73
68.73
68.59
68.42
68.33
170314201
Illinois
Cook
69.10
69.05
69.05
69.05
69.05
68.91
68.75
68.68
170317002
Illinois
Cook
69.36
69.28
69.27
69.28
69.27
69.09
68.88
68.84
480391004
Texas
Brazoria
70.93
70.51
70.49
70.49
70.48
69.29
68.79
68.63
482010024
Texas
Harris
76.28
75.90
75.89
75.89
75.87
74.58
74.09
73.97
490110004
Utah
Davis
72.20
72.15
72.15
72.15
72.15
71.60
71.50
71.50
490353006
Utah
Salt
Lake
73.00
72.95
72.95
72.95
72.95
72.40
72.30
72.30
490353013
Utah
Salt
Lake
74.10
74.05
74.05
74.05
74.05
73.40
73.30
73.30
490570002
Utah
Weber
70.30
70.26
70.25
70.26
70.25
69.69
69.58
69.58
550590019
Wisconsin
Kenosha
72.01
71.91
71.90
71.91
71.90
71.70
71.47
71.41
550590025
Wisconsin
Kenosha
68.46
68.30
68.29
68.29
68.28
68.06
67.81
67.76
551010020
Wisconsin
Racine
70.52
70.42
70.41
70.42
70.41
70.21
69.99
69.93
56
-------
Table C-14. 2026 Maximum Ozone DVs (ppb) for Each "Control Scenario" Assessed.
Site
State
County
Engineering
Analysis
Base
SCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
Generation
Shifting
SCR
Optimize +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting +
non-EGU
Tier 1
SCR
Optimize +
SOA CC +
SNCR
Optimize +
SCR/SNCR
Retrofit +
Generation
Shifting +
non-EGU
Tier 1 +
Tier 2
40278011
Arizona
Yuma
71.81
71.80
71.80
71.80
71.80
71.79
71.77
71.77
80350004
Colorado
Douglas
71.55
71.50
71.49
71.50
71.49
71.16
71.09
71.09
80590006
Colorado
Jefferson
72.69
72.66
72.65
72.66
72.65
72.38
72.32
72.32
80590011
Colorado
Jefferson
73.68
73.64
73.63
73.64
73.63
73.38
73.32
73.32
90010017
Connecticut
Fairfield
72.30
72.28
72.27
72.28
72.27
72.07
71.84
71.82
90013007
Connecticut
Fairfield
73.99
73.95
73.93
73.93
73.91
73.62
73.29
73.26
90019003
Connecticut
Fairfield
75.03
75.00
74.98
74.98
74.97
74.74
74.48
74.46
90099002
Connecticut
New
Haven
72.78
72.72
72.70
72.71
72.69
72.40
72.09
72.06
170310001
Illinois
Cook
72.87
72.83
72.82
72.83
72.82
72.63
72.44
72.34
170310032
Illinois
Cook
71.98
71.95
71.95
71.96
71.95
71.84
71.75
71.68
170310076
Illinois
Cook
71.56
71.54
71.53
71.54
71.54
71.39
71.22
71.13
170314201
Illinois
Cook
72.61
72.56
72.56
72.56
72.56
72.42
72.24
72.17
170317002
Illinois
Cook
72.27
72.19
72.18
72.19
72.18
71.99
71.77
71.73
480391004
Texas
Brazoria
73.09
72.65
72.64
72.63
72.62
71.40
70.88
70.72
482010024
Texas
Harris
77.82
77.44
77.43
77.42
77.41
76.09
75.58
75.47
490110004
Utah
Davis
74.42
74.37
74.37
74.37
74.37
73.80
73.69
73.69
490353006
Utah
Salt Lake
74.61
74.56
74.56
74.56
74.56
74.00
73.89
73.89
490353013
Utah
Salt Lake
74.60
74.55
74.55
74.55
74.55
73.90
73.80
73.80
490570002
Utah
Weber
72.22
72.17
72.17
72.17
72.17
71.58
71.48
71.48
550590019
Wisconsin
Kenosha
72.91
72.81
72.81
72.81
72.80
72.60
72.37
72.31
550590025
Wisconsin
Kenosha
71.48
71.30
71.30
71.30
71.29
71.06
70.80
70.75
551010020
Wisconsin
Racine
72.42
72.33
72.32
72.32
72.32
72.11
71.88
71.82
57
-------
4. Comparison between the air quality assessment tool estimates
As described earlier, AQAT was calibrated using modeled ozone data from a 2026 case
where EGUs and non-EGUs were reduced by 30%. We also had a second set of calibration
factors, based on the change from the 2026 base to the 2023 base (which could be used to
modulate to alternative years, though these were not pursued). Thus, it was possible to evaluate
the estimates from the tool for a comparable scenario using alternative calibration factors. The
average design values from AQAT as well as the differences for the 2026 scenario with EGU
SCR and non-EGU Tier 1 + Tier 2 are shown in Table C-15. The AQAT values and the
differences in the table have been rounded to a hundredth of a ppb. For this set of scenarios, the
differences are moderate, with a maximum value of 0.37 ppb. Since the calibration factor based
on the 30% EGU and non-EGU emission reduction was developed based on modulating the
sectors being regulated in this rulemaking, we conclude that these factors were the ones to use
within the Step 3 methodology.
The results of this comparison, which are relatively similar, demonstrate that, considering
the time and resource constraints faced by the EPA, the AQAT provides reasonable estimates of
air quality concentrations for each receptor, and can provide reasonable inputs for the multi-
factor assessment and overcontrol assessment.
58
-------
Table C-15. 2026 Average Ozone DVs (ppb) for the EGU SCR and non-EGU Tier 1 and
Tier 2 Scenarios Using Two Calibration Factors.
Site
State
County
EGU SCR
and non-
EGU Tier
1+Tier 2
(30% EGU
and non-
EGU
Calibration)
EGU SCR and
non-EGU Tier
1+Tier 2
(2023
Calibration)
Delta AQ
between
Calibration
Approaches
40278011
Arizona
Yuma
70.06
70.02
0.04
80350004
Colorado
Douglas
70.07
69.89
0.18
80590006
Colorado
Jefferson
71.26
71.10
0.17
80590011
Colorado
Jefferson
72.16
71.90
0.26
90010017
Connecticut
Fairfield
71.35
71.37
-0.02
90013007
Connecticut
Fairfield
72.54
72.77
-0.24
90019003
Connecticut
Fairfield
74.40
74.54
-0.15
90099002
Connecticut
New Haven
70.22
70.38
-0.16
170310001
Illinois
Cook
68.73
68.50
0.23
170310032
Illinois
Cook
69.20
68.97
0.22
170310076
Illinois
Cook
68.51
68.14
0.37
170314201
Illinois
Cook
68.83
68.65
0.18
170317002
Illinois
Cook
68.98
68.76
0.22
480391004
Texas
Brazoria
68.72
69.01
-0.29
482010024
Texas
Harris
74.23
74.21
0.03
490110004
Utah
Davis
71.51
71.44
0.07
490353006
Utah
Salt Lake
72.30
72.24
0.06
490353013
Utah
Salt Lake
73.34
73.37
-0.03
490570002
Utah
Weber
69.63
69.46
0.18
550590019
Wisconsin
Kenosha
71.57
71.35
0.22
550590025
Wisconsin
Kenosha
67.95
67.71
0.24
551010020
Wisconsin
Racine
70.12
69.90
0.22
59
-------
D. Selection of Short-term Rate Limits
For the reasons described in the preamble, EPA is proposing to complement the longer-
term mass-based trading program (premised on seasonal emission rate performance) with a
short-term emission rate limit for some units. EPA considered hourly, 24-hour, 7-day and 30-
day limits as appropriate short-term rate limits. 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 identified the daily
(e.g., 24-hr) limit as an appropriate time-period for the short-term rate limit.
As described in the preamble, in establishing the 24-hour emission limits, 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)
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.
Focusing on the 0.14 lb/MMBtu rate, EPA identified 164 units that had ozone season
rates at or below 0.08 lb/MMBtu. 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. Considering the number of tons emitted on days when the daily
emission rates exceeded 0.14. 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.
60
-------
A
1.2
E S
~ E 0.8
oo F
i— . •—
-------
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.53'54 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 a
conservative 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.
The EPA continues to consider that approach to be acceptable. However, as discussed in
the 2014 Guidance, after receiving numerous comments, and analyzing the impact of emissions
variability on air quality, the EPA expects that it may also be possible in specific cases for states
to develop control strategies that account for variability in 1-hour emissions rates through
emission limits with averaging times that are longer than 1 hour, using averaging times as long
as 30-days, but still provide for attainment of the 2010 SO2 NAAQS. The EPA would need to
consider specific submitted candidate emission limits along with other elements of a submitted
SIP attainment demonstration to conclude whether such a limit would be approvable. This view
is based on the EPA's general expectation that, if periods of hourly emissions above the critical
emission value are a rare occurrence at a source, particularly if the magnitude of the emissions is
not substantially higher than the critical emissions value, these periods would be unlikely to have
a significant impact on air quality, insofar as they would be very unlikely to occur repeatedly at
the times when the meteorology is conducive for high ambient concentrations of S02. The EPA
believes that making this option available to states could reflect an appropriate balance between
providing a strong assurance that the NAAQS will be attained and maintained, while still
acknowledging the necessary variability in source operations and the impairment to source
operations that would occur under what could be in some cases an unnecessarily restrictive
approach to constraining that variability. Nevertheless, 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
53 https://www.epa.gov/sites/default/files/2016-06/documents/20140423guidance_nonattainment_sip.pdf
54 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).
62
-------
averaging time would reflect a higher numeric emission rates. 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 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 limit 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
different time periods. EPA notes that concept could be applied to help identify daily (e.g., 24-
hour) limits that are comparably stringent to longer-term limits. In other words, because these
sources are only a portion of the problem causing NAAQS violations, and because EPA defines
the emissions that are significantly contributing inclusive of emissions that are eliminated by
installation and full operation of post-combustion control equipment at a portion of the EGU
fleet, and 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 limits 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 limits 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 limits that would be adjusted higher to
accommodate the variation in operation, demand for electricity, variation in fuel, and other
technical and engineering limitations.
We expect that the use of shorter-term averages may be necessary in cases where sources'
emission rates exhibit a high degree of variability with some time-periods with high emission
rates (i.e., units that have post-combustion control equipment and units that need new post-
combustion control equipment installed). Therefore, EPA is limiting its application of short-term
limits to coal-fired units with SCR retrofit potential or that are already equipped with SCR. In
such cases, as previously noted, the EPA believes this approach provides appropriate flexibility
while still requiring approximately the same control strategy as demonstrated with longer-term
emission rate averages (in particular, the averages used to enshrine emissions budgets).
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 limit 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,
63
-------
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 limit (i.e., 24-hour). Here, the EPA envisions that meeting both
the short-term rate limits 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 derived55. 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 limit, specifically the operation of SCR post-combustion controls.
Emission limits are often expressed either in terms of emission rates (e.g., pounds per
hour) or in terms of emission factors (e.g., lb/MMBtu heat input). The variability of values for
these two parameters will likely be different. Therefore, analyses of a longer-term average
emission rate that is comparably stringent to a shorter-term emission rate limit would need to be
designed to assess variability for the parameter for which an emission limit is being set. Since we
are focused here on ensuring installation and operation of control equipment, rather than
constraining the operation of the unit through a mass limitation, we focused on variability in
emissions rate (lb/MMBtu).
We acknowledge that supplemental limits on the frequency and/or magnitude of
occasions of elevated emissions can be a valuable element of a plan that ensures control
operation and protects against NAAQS violations may be useful in some instances. However,
because of the differences between 1-hr S02 and 8-hr Ozone (with the latter being created based
on the emissions of NOx and VOCs from hundreds or thousands of individual point sources, and
millions of individual mobile sources, rather than a handful of large point sources), we find that a
long-term emission rate assumption (expressed as a seasonal mass limit) coupled with a short-
term daily emission limit applied to individual units incentivizes best performance of controls
while also ensuring operation of the controls each day.
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
55 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.
64
-------
lb/MMBtu basis (Table D-l).56'57 We show the estimated limits 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 for coal-steam fleet-wide value) 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 0.081b/MMBtu seasonal rate), the daily limits are
unlikely to be binding if an SCR is present.
56 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.
57 For this assessment, we assume that the 30-day and seasonal rates would be at comparable levels. Clearly, a 30-
day rate would have a larger variability than a seasonal rate, 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.
65
-------
Table D-l. Ratios to convert between various time-averages, applied to a 0.08 lb/MMBtu
seasonal limit.
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
O/G Steam
SCR
0.68
0.83
1
0.10
66
-------
E. Preliminary Environmental Justice Screening Analysis
In addition to the considerations above, EPA also considered potential environmental
justice concerns.58 EPA's EJ Technical Guidance59 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 in relatively
close proximity to the facility.
This initial screening analysis examines whether air pollution emitted from each
individual facility might reach any communities with potential environmental justice concerns.
Such an impact would support further consideration of additional pollution limits imposed at that
facility to address existing disproportionate impacts. This screening-level analysis helped EPA
identify potential concerns at the start of proposed rule development, while subsequent analysis
presented in the RIA provide a robust evaluation of the distributional impacts of the requirements
proposed 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 at a facility level early in the process, and the environmental justice analyses presented
in the RIA estimate the ultimate impacts of the proposed rule.
Based on this screening analysis, nearly all of the EGUs included in this analysis are
located within a 24-hour transport distance of many areas with potential EJ concerns. While this
screen does not identify all potentially impacted downwind areas or quantify the downwind
impact of these sources (the aggregate impact of which is evaluated and discussed in the RIA), it
does demonstrate that the potential exists for these sources to affect areas facing pre-existing
disproportionate impacts. An overview of the methodology is described below.
Methodology
The screening assessment in this TSD is carried out in two parts. First, to estimate which
census block groups have some potential to be affected 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
proposed rule/0 A forward trajectory is a modeled parcel of air that moves forward (i.e.,
58 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, 2015a). 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, 2015a).
59 U.S. Environmental Protection Agency (EPA), 2015. Guidance on Considering Environmental Justice During the
Development of Regulatory Actions
60 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
67
-------
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
day—12:00 AM, 6:00 AM, 12:00 PM, and 6:00 PM (local standard time). For simplicity, in
order to facilitate an initial screening-level analysis, EPA limited trajectories to the period June 1
to August 31 for the years 2017 to 2019. 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
(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.61 For this analysis, we limit our evaluation to coordinates of those
trajectories that are within the continental United States and within 500 meters of ground level
for simplicity in this initial screen. While the 24-hour transport time and 500 meter elevation
used in this screening analysis identifies many of the near-source areas that are the most
frequently impacted, emissions can travel over larger distances and longer times and have
substantive air quality impacts downwind (i.e., those impacts are analyzed in the RIA).62
It is important to note that unlike the other models used to quantify downwind ozone
concentrations related to this proposed 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.63 We are using HYSPLIT trajectories in a qualitative way to examine
the spatial patterns of pollutant transport from EGUs.64 The model results simply simulate the
path that the wind would carry a modeled parcel of air from the stack(s) of each EGU.2
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.
Next, EPA screened those downwind areas to identify census block groups with potential
environmental justice concerns. The intent of this screen is to broadly identify areas potentially
experiencing pre-existing disproportionate impacts, and as such, it does not quantify ozone-
specific health risks. The screen was performed using data from EPA's EJSCREEN, an
environmental justice mapping and screening tool that includes 11 different environmental
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).
61 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 are uploaded into an Oracle database. 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.
62 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
63 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).
64 In general, pollutant concentrations are the result of transport, dispersion, and transformation. As noted, this
analysis does not consider photochemical transformations.
68
-------
indicators and 6 different demographic indicators.65 For this analysis, EPA evaluated the
available information at the census block group level for one environmental indicator, ozone,66
and one demographic indicator, percent low-income.67 Note that this screening analysis is
limited to a single environmental burden indicator (pre-existing ozone exposure), and does not
consider the exposure and vulnerability of communities to multiple environmental burdens and
their cumulative impacts. For further discussion of these indicators and the other indicators
currently available in the EJSCREEN tool, see the EJSCREEN Technical Documentation.
Using these indicators to represent environmental burden and vulnerability generally, the
EPA identified block groups for which these two indicators each exceeded the 80th percentile on
a national basis. 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.68 While
communities exceeding this threshold may be exposed to pollution and potentially vulnerable to
its impacts, it is important to note that EPA is not designating these areas as being "EJ
communities." In line with this, the results of this screen should not be interpreted to suggest the
absence of environmental justice concerns in areas that fail to meet this screening threshold.
Rather, populations residing in these downwind areas are identified as being amongst the 20% of
the US population with the highest values for each of the respective EJSCREEN indicators.
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 with potential EJ
concerns to identify the EGUs that have the potential to impact those areas. These are EGUs
whose HYSPLIT trajectories cross over some portion of census block groups that meet the
screening criteria above. EGUs with at least one block group exceeding the screening threshold
that intersect with the EGU's respective HYSPLIT trajectory are highlighted in the figure below.
When viewed comprehensively, the results are used to provide a reasonable approximation of
downwind areas potentially exposed to air pollutants from each facility within a 24-hour period
from emissions for the 2017-2019 time period. The map in Figure 1 shows the number of block
groups exceeding the screening threshold that are identified as being downwind from each EGU,
based on this particular analysis.
65 U.S. Environmental Protection Agency (EPA), 2019. EJSCREEN Technical Documentation.
66 Ozone summer seasonal average of daily maximum 8-hour concentration in air in parts per billion (2017)
67 The percent of a block group's population in households where the household income is less than or equal to twice
the federal "poverty level."
68 U.S. Environmental Protection Agency (EPA), 2019. EJSCREEN Technical Documentation.
69
-------
Figure E-l. Number of Block Groups Downwind from Each EGU that Exceed the Screening
Threshold
70
-------
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).69,70 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)..71 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.
69 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
70 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.
71 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/ZyP URL.cgi ?Dockey=Pl 00KETF. txt.
71
-------
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 2018 as part of the Clean Air Market
Division (CAMD) annual power sector programs progress report.72 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. 73 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 2017-2019, 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 11.5% (Figure F-l)74.
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 2019 (to be updated in 2022), 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.
The reductions from this rule are likely to provide further protection to natural forest
ecosystems by reducing the potential for ozone-related impacts.
72 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/2019 full report.pdf. and
https://www3.epa.gov/airmarkets/progress/reports/pdfs/2018_full_report.pdf
73 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
74 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.
72
-------
Figure F-l: Estimated Black Cherry, Yellow Poplar, Sugar Maple, Eastern White Pine, Virginia
Pine, Red Maple, and Quaking Aspen Biomass Loss due to Ozone Exposure for 2016-2018.
Biomass (% Loss)
> 1%
1 to 3%
3 to 6%
6 to 9%
>9% Max =11.6%
See the annual progress reports at https://www3.epa.gov/airmarkets/progress/reports/index.html
and https://www3.epa.gov/airmarkets/progress/reports/pdfs/2018_full_report.pdf
73
-------
Appendix A: State Emission Budget Calculations and Engineering Analytics
See Excel workbook titled "Proposed Rule State Emission Budget Calculations and Engineering
Analytics" on EPA's website and in the docket for this rulemaking
74
-------
Appendix B: Description of Excel Spreadsheet Data Files Used in the AQAT
75
-------
EPA placed the Ozone=AQAT=Proposal.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 (including
generation shifting).
• "RIAcases" contains state specific ozone-season NOx emission total EGU and non-
EGU emissions and emissions changes for the 2023 and 2026 proposed rule, as well as
less and more-stringent cases. The emission changes are relative to the 2023 and 2026
base cases modeled in CAMx.
• "NOxnonCEM" has a breakdown of the point EGU nonCEM emission inventory
component used in the air quality modeling.
• "non-EGU emiss" has the total anthropogenic emission reductions by state and Tier for
each of the non-EGU cases.
• "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 20261] emission level (column P on the "2026 OS NOx" worksheet) to
make a fractional change in emissions in columns AM through AR. For 2026, Non-EGU
emissions change and fractional change) are found in columns AS through AY.
Air quality modeling design values and contributions from CAMx
• "2023fj_All" contains average and maximum design values as well as state by state
contributions for the 2023fj base case modeled in CAMx.
• "2026fj_All" contains average and maximum design values as well as state by state
contributions for the 2026fj base case modeled in CAMx.
• "2026fj_30NOx" 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%.
• "2016_2023_2026_2032 DVs contains average and maximum design values for each
receptor for each year.
• 2026fj_receptor_list contains a list of the receptors whose average and/or maximum
design values are greater than or equal to 71 ppb in 2026.
Calibration factor creation and assessment
• "2026to2026w30=calib=(rec, stat)" 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%. The
calibration factors can be found in columns I through BF.
76
-------
• "2026to2023_calib_(rec, state)" includes the state-by-state and receptor-by-receptor
calculation of the calibration factors based on the 2026 base and 2023 base contributions,
and fractional change of 2023 emissions relative to 2026 emissions. 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 approach for each cost
threshold 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 AR through AY) for a control
scenario, 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
approach) and J (a control scenario approach where the geography remains fixed) and the
maximum contributions to remaining linkages in column AD. Design value estimates for
the proposed rule and less and more stringent alternatives for the RIA are shown in
columns BC, BD, and BE (note that the linked, home, and nonlinked states are assigned
the same emission value). The maximum contribution to remaining receptors is shown in
columns AN through AP. The alternative calibration factor simulation results are shown
in columns B J and BK. Each column contains average design values followed by
maximum design values, below.
• " 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 approach for each cost
threshold are shown starting in column L. Under this approach, the maximum
contribution to remaining receptors is shown in columns AD through AI. Furthermore, a
set of design value estimates are shown (columns AN through AS) for a control scenario,
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 approach) and J (a
control scenario approach where the geography remains fixed) and the maximum
contributions to remaining linkages in column AB. Each column contains average design
values followed by maximum design values, below. Design value estimates for the
proposed rule and less and more stringent alternatives for the RIA are shown in columns
BC, BD, and BE (note that the linked, home, and nonlinked states are assigned the same
emission value). The maximum contribution to remaining receptors is shown in columns
A J through AL.
• "2023_scenario"and "2026_scenario" 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 states are shown in rows 2 and 3.
• "2023_scenario_links" and "2026_scenario_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.
77
-------
"2026_control_fixed" and "2023_contol_fixed" 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_calib" 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
states are shown in rows 2 and 3. This uses the calibration factor based on the 2023 air
quality modeling, rather than the calibration factor based on the 2026 air quality
modeling with the 30% reduction from EGUs and non-EGUs.
"2026_scenario_eng_base" and "2023_scenario_eng_base" contain air quality
contributions and design value estimates for the two base cases using the engineering
analysis emission estimates for EGUs.
The individual scenario worksheets labeled:
o
"2023_scenario_base",
o
"2023_scenario_SCRopt",
o
"2023 scenario SCRoptwCC",
o
"2023_scenario_SNCRopt",
o
"2023_scenario_SNCRoptwCC",
o
"2023_scenario_newSCR",
o
"2026_scenario_base",
o
"2026_scenario_SCRopt",
o
"2026 scenario SCRoptwCC",
o
"2026_scenario_SNCRopt",
o
"2026_scenario_SNCRoptwCC",
o
"2026_scenario_newSCR",
o
"2026 scenario Tierl",
o
"2026_scenario_Tierland2",
o
"2023_control_fixed_base",
o
"2023_control_fixed_SCRopt",
o
"2023_control_fixed_SCRoptwCC",
o
"2023_control_fixed_SNCRopt",
o
"2023_control_fixed_SNCRoptwCC"
o
"2023_control_fixed_newSCR",
o
"2026_control_fixed_base",
o
"2026_control_fixed_SCRopt",
o
"2026_control_fixed_SCRoptwCC",
o
"2026_control_fixed_SNCRopt",
o
"2026_control_fixed_SNCRoptwCC"
o
"2026_control_fixed_newSCR",
o
"2026 control fixed Tierl",
o
"2026 control fixed Tierland2"
o
"2023_proposed_rule"
o
"2023_less_stringent"
78
-------
o "2023_more_stringent"
o "2026_proposed_rule"
o "2026_less_stringent"
o "2026_more_stringent"
contain static air quality contributions and design value estimates for all monitors for the
particular year and scenario.
79
-------
Appendix C: IPM Runs Used in Transport Rule Significant Contribution Analysis
80
-------
Table 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 BC1K
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 2021.
Illustrative Base Case with optimization
technology + LNB upgrade
EPA620_TR_2e
Model run used as the base case for the Illustrative Analysis of
cost threshold analyses. Based on the air quality modeling base
case, but with projected retirements and retrofits in 2023 limited.
Also assumes optimization of existing post-combustion controls
and upgrade of combustion controls if mode 3>1. Imposes state-
level generation constraints starting in 2023 for fossil-fuel fired
units greater than 25 MW that is equal to Air Quality Modeling
Base Case levels.
Illustrative Base Case with optimization
technology + LNB upgrade + SCR retrofit
EPA620_TR_4e
Imposes state-level generation constraints starting in 2023 for
fossil-fuel fired units greater than 25 MW that is equal to
Illustrative Base Case levels. Also assumes optimization of
existing post-combustion controls and upgrade of combustion
controls if mode 3
-------
Appendix D: Generation Shifting Analysis
Table Appendix D-l. Tons of EGU NOx Reduction Potential from Shifting Generation
Compared to Adjusted Historical Baseline Emissions.
State
2023
2023 Reductions
2023 Reductions
2026
2026 Reductions
2026 Reductions
Baseline
from generation
from generation
Baseline
from generation
from generation
(Tons)
Shifting at
$1,800/Ton
Shifting at
$1,800/Ton (%)
(Tons)
Shifting at
$1,800/Ton
Shifting at
$1,800/Ton (%)
Alabama
6,648
231
3%
6,701
0
0%
Arkansas
8,955
38
0%
8,728
108
1%
Delaware
423
-4
-1%
473
0
0%
Illinois
7,662
-127
-2%
7,763
350
5%
Indiana
12,351
326
3%
9,737
206
2%
Kentucky
13,900
1,213
9%
13,211
188
1%
Louisiana
9,987
-96
-1%
9,854
0
0%
Maryland
1,208
5
0%
1,208
11
1%
Michigan
10,737
-15
0%
9,129
56
1%
Minnesota
4,207
107
3%
4,197
48
1%
Mississippi
5,097
0
0%
5,077
-1
0%
Missouri
20,094
444
2%
18,610
127
1%
Nevada
2,346
0
0%
2,438
0
0%
New Jersey
915
11
1%
915
11
1%
New York
3,927
100
3%
3,927
100
3%
Ohio
10,295
765
7%
10,295
355
3%
Oklahoma
10,463
0
0%
10,283
40
0%
Pennsylvania
12,242
309
3%
11,738
409
3%
Tennessee
4,319
-25
-1%
4,064
0
0%
Texas
40,860
1,190
3%
39,186
1,423
4%
Utah
15,500
512
3%
9,679
-16
0%
Virginia
3,415
-24
-1%
3,243
30
1%
West Virginia
14,686
547
4%
14,686
429
3%
Wisconsin
5,933
-76
-1%
3,628
102
3%
Wyoming
10,191
958
9%
10,249
90
1%
82
-------
Appendix E: Feasibility Assessment for Engineering Analytics Baseline
Similar to the Revised CSAPR Update Final Action, EPA analyzed and confirmed that the
assumed 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 expected to
occur in years 2023 through 2026. 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 baseline heat input and generation from the states
covered in this action and compared it to historical trends for these same states (Scenario
1). This illustrated that the assumed heat input and generation from fleet turnover 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 2019 levels instead
of continuing to decline (Scenario 2).
• Finally, EPA identified the 2021 Energy Information Administration's Annual Energy
Outlook (EIA AEO) annual growth projections from 2020 through 2026 total electricity
demand levels (1.1%) from its reference case, and estimated an upperbound future year
scenario where covered fossil generation grew at levels matching this fleet-wide total
growth rate (Scenario 3).75
• EPA's assessment illustrates the amount of generation in its baseline, factoring in
retirements and new fossil units, is more than sufficient to accommodate all three
scenarios.76 For instance, generation from covered fossil sources in these states has
dropped at an average rate greater than 1.6% per year between 2018 and 2021 (877 TWh
to 833 Twh). However, EPA's assumed baseline generation from covered fossil sources
for the states reflects a rate of decline less than 2% per year. See Table Appendix E-2.
• 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
75 Department of Energy, Annual Energy Outlook 2020. Available at
https://www.eia. gov/outlooks/aeo/data/browser/#/?id=8-AE02020&cases=ref2020&sourcekey=0
76 Based on historical trends, modeling, and company statements, EPA expects levels similar to scenario 1 and
scenario 2 to be most likely.
83
-------
regulatory approvals pending for years 2023 and beyond (as this capacity is unlikely to
have yet started construction).77
• 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% for NGCC.
• Using these technology-specific capacity factors based on past performance and IPM
documentation, EPA anticipated over 40 TWh from new non-fossil generation already
under construction or being planned with regulatory approval received. This combined
with the baseline generation from existing units exceeds the expected generation load for
the states under all three scenarios.78
• 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 40 TWh), exceeds the generation
assuming no change (scenario 2) and the upper bound analysis for future covered fossil
generation that assumes 1.1% 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 (low emitting generation) becomes available
in the outer years, that constitutes additional generation that further exceeds EPA's
upperbound generation levels below - further bolstering the observation that no
replacement generation from existing units needs to be assumed to fill generation from
retiring units.
77 Department of Energy, EIA Form 860, Generator Form 3-1. 2020. Available at
https://www.eia. gov/electricity/data/eia860/
78 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.
84
-------
Table Appendix E-l: Heat Input Change Due to Fleet Turnover (Historical and Future)
Values for 2018-2021 reflect reported data, while 2023-2026 reflects assumed heat input.
Region
2018
2H|«>
2H2H
2021
2023
2U24
2025
2026
Alabama
388
352
327
322
327
327
327
338
Arkansas
220
203
160
193
202
202
202
197
California
254
219
265
301
226
226
226
226
Delaware
22
20
21
19
34
34
34
34
Illinois
397
332
283
334
245
242
242 ;
232
Indiana
479
404
371
41 1
328
287
287
289
Kentucky
354
316
270
303
250
250
250
262
Louisiana
312
317
281
280
334
334
334
334
Maryland
105
92
82
88
78
93
93
93
Michigan
349
326
283
308
317
317
317
321
Minnesota
144
132
108
129
1 10
1 10
no
1 12
Mississippi
218
21 1
224
190
199
199
199
199
Missouri
313
269
254
288
240
240
240
254
Nevada
108
98
100
103
103
103
103
104
New Jersey
151
146
1 19
120
142
142
142
142
New York
238
202
234
240
: 222
: 222
: 222
221
Ohio
405
402
395
400
352
352
352
368
Oklahoma
276
235
232
213
: 235
: 235
235
233
Oregon
58
63
50
56
55
55
55
56
Pennsylvania
487
509
535
565
524
524
524
518
Tennessee
184
190
165
180
169
169
169
156
Texas
1,530
1,501
1,355
1,403
1,434
1,418
1,418
1,382
Utah
143
133
133
164
130
130
130
107
Virginia
251
249
261
215
258
249
249
273
West Virginia
309
295
268
313
260
260
260
251
Wisconsin
222
192
195
221
194
185
185
137
Wyoming
186
164
163
176
152
152
152
164
Total
8,101
7,570
7,137
7,535
7,121
7,058
7,058
7,003
85
-------
Appendix E-2: Assumed Baseline OS Generation and Expected New Build Generation from
Covered Fossil Units (TWh)
2023
2024
2025
2026
Scenario 1 - Generation Levels (with
continued pace of 1.6% decline)
806
793
780
767
Scenario 2 - Generation Levels (no change
from 2021)
833
833
833
833
Scenario 3 - Generation Levels (1.1% growth
from covered fossil)
843
852
862
872
Assumed Baseline Fossil Generation with
Reported Fossil Retirement and Reported New
Build
815
810
810
806
New Build (Non-Fossil)
40
57
64
75
Total Baseline Generation
855
867
874
881
86
-------
Appendix F: State Emission Budgets and Variability Limits
State
2023
Emission
Budgets
(tons)
2023
Variability
Limit
(tons)
2024
Emission
Budgets
(tons)
2024
Variability
Limit
(tons)
2025
Illustrative
Emission
Budgets
(tons)
2025
Illustrative
Variability
Limit
(tons)
2026
Illustrative
Emission
Budgets
(tons)
2026
Illustrative
Variability
Limit
(tons)
Alabama
6,364
1,336
6,306
1,324
6,306
1,324
6,306
1,324
Arkansas
8,889
1,867
8,889
1,867
8,889
1,867
3,923
824
Delaware
384
81
434
91
434
91
434
91
Illinois
7,364
1,546
7,463
1,567
7,463
1,567
6,115
1,284
Indiana
11,151
2,342
9,391
1,972
8,714
1,830
7,791
1,636
Kentucky
11,640
2,444
11,640
2,444
11,134
2,338
7,573
1,590
Louisiana
9,312
1,956
9,312
1,956
9,179
1,928
3,752
788
Maryland
1,187
249
1,187
249
1,187
249
1,189
250
Michigan
10,718
2,251
10,718
2,251
10,759
2,259
6,114
1,284
Minnesota
3,921
823
3,921
823
3,910
821
2,536
533
Mississippi
5,024
1,055
4,400
924
4,400
924
1,914
402
Missouri
11,857
2,490
11,857
2,490
10,456
2,196
7,246
1,522
Nevada
2,280
479
2,372
498
2,372
498
1,211
254
New Jersey
799
168
799
168
799
168
799
168
New York
3,763
790
3,763
790
3,763
790
3,238
680
Ohio
8,369
1,757
8,369
1,757
8,369
1,757
8,586
1,803
Oklahoma
10,265
2,156
9,573
2,010
9,393
1,973
4,275
898
Pennsylvania
8,855
1,860
8,855
1,860
8,855
1,860
6,819
1,432
Tennessee
4,234
889
4,234
889
4,008
842
4,008
842
Texas
38,284
8,040
38,284
8,040
36,619
7,690
21,946
4,609
Utah
14,981
3,146
15,146
3,181
15,146
3,181
2,620
550
Virginia
3,090
649
2,814
591
2,948
619
2,567
539
West Virginia
12,478
2,620
12,478
2,620
12,478
2,620
10,597
2,225
Wisconsin
5,963
1,252
5,057
1,062
4,198
882
3,473
729
Wyoming
9,125
1,916
8,573
1,800
8,573
1,800
4,490
943
87
-------
Appendix G: Figures Related to Preamble Section VI and Section VII
Figure 1 to Section VI.D.l - EGU Ozone Season NOx Reduction Potential in 26 Linked
States and Corresponding Total Reductions in Downwind Ozone Concentration at
Nonattainment and Maintenance Receptors for Each Cost Threshold Level Evaluated
(2023)
0.12
Ozone improvement
— • — NOx reduction
. . , potential. . . .
$1,000 $1,500
Cost per Ton
35,000
30,000 ^
¦o
e?
88
-------
Figure 3 to Section VI.D.l: EGU Ozone Season NOx Reduction Potential in 23 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)79
_Q
Q.
Q.
+J
C
(D
E
(D
>
o
I—
Q.
E
3
a
(D
tLO
ro
<
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
E f\
jfe _ _ _ _
F
^ —
/
— Ozone Improvement
- •
— NOx Reduction Potential
120,000
100,000
80,000
60,000
20,000
$0
$20,000
$40,000
$60,000
$80,000
0
$100,000
c
o
o
¦a
(D
cc
tn
c
o
c
o
e rr
c
o
to
CU
(D
OO
o
19 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 o/g steam units <150 tons
per season, combustion control upgrade on combustion turbines, and SCRs on combustion
turbines > 100 MW respectively. These mitigation measures and costs are further discussed in
the EGU NOx Mitigation Strategies Proposed Rule TSD.
89
-------
Figure 1 to Section VILB.l.c.i: New Madrid Unit 2 Daily Emissions Rate (2017 and 2019)
_ New Madrid Unit 2 Ozone Season Daily NOx Rates
3
m
0 5-
F
E
0.4-
-Q
0)
0.3-
m
a.
0.2-
x
o
z
0.1 -
>,
m
0.0-
O
_>
ns-
Ol
F
E
0.4-
n
¦—•
0.3-
OJ
m
01
0.2-
X
()
0.1 -
>.
OS
Q
o.o-
May 2017
Jun 2017
Jul 2017
Aug 2017
Sep 2017
Oct 2017
v.
•
¦
•
m
/ •«
¦ • ¦
••
•
• • •
¦ • ®
m • +
• •
• • • J
<%>v *». S-W- •
•
••
•
¦
•
•
1 1
¦
May 2019
Jun2019
Jul 2019
Aug 2019
Sep 2019
Oct 2019
Date
Conemaugh Unit 2 Ozone Season Daily NOx Rates
Oct 2017
May 2019
Jun 2019
Jul 2019 Aug 2019
Date
Sep 2019
Oct 2019
90
------- |