Federal “Good Neighbor Plan” for the 2015 Ozone National Ambient Air Quality Standards

AGENCY:

Environmental Protection Agency (EPA).

ACTION:

Final rule.

SUMMARY:

This action finalizes Federal Implementation Plan (FIP) requirements to address 23 states' obligations to eliminate significant contribution to nonattainment, or interference with maintenance, of the 2015 ozone National Ambient Air Quality Standards (NAAQS) in other states. The U.S. Environmental Protection Agency (EPA) is taking this action under the “good neighbor” or “interstate transport” provision of the Clean Air Act (CAA or Act). The Agency is defining the amount of ozone-precursor emissions (specifically, nitrogen oxides) that constitute significant contribution to nonattainment and interference with maintenance from these 23 states. With respect to fossil fuel-fired power plants in 22 states, this action will prohibit those emissions by implementing an allowance-based trading program beginning in the 2023 ozone season. With respect to certain other industrial stationary sources in 20 states, this action will prohibit those emissions through emissions limitations and associated requirements beginning in the 2026 ozone season. These industrial source types are: reciprocating internal combustion engines in Pipeline Transportation of Natural Gas; kilns in Cement and Cement Product Manufacturing; reheat furnaces in Iron and Steel Mills and Ferroalloy Manufacturing; furnaces in Glass and Glass Product Manufacturing; boilers in Iron and Steel Mills and Ferroalloy Manufacturing, Metal Ore Mining, Basic Chemical Manufacturing, Petroleum and Coal Products Manufacturing, and Pulp, Paper, and Paperboard Mills; and combustors and incinerators in Solid Waste Combustors and Incinerators.

DATES:

This final rule is effective on August 4, 2023.

ADDRESSES:

The EPA has established a docket for this rulemaking under Docket ID No. EPA–HQ–OAR–2021–0668. All documents in the docket are listed in the https://www.regulations.gov index. Although listed in the index, some information is not publicly available, e.g., Confidential Business Information or other information whose disclosure is restricted by statute. Certain other material, such as copyrighted material, will be publicly available only in hard copy. Publicly available docket materials are available either electronically at https://www.regulations.gov or in hard copy at the U.S. Environmental Protection Agency, EPA Docket Center, William Jefferson Clinton West Building, Room 3334, 1301 Constitution Ave. NW, Washington, DC. The Public Reading Room is open from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding legal holidays. The telephone number for the Public Reading Room is (202) 566–1744, and the telephone number for the Office of Air and Radiation Docket is (202) 566–1742.

FOR FURTHER INFORMATION CONTACT:

Ms. Elizabeth Selbst, Air Quality Policy Division, Office of Air Quality Planning and Standards (C539–01), Environmental Protection Agency, 109 TW Alexander Drive, Research Triangle Park, NC 27711; telephone number: (312) 886–4746; email address: selbst.elizabeth@epa.gov.

SUPPLEMENTARY INFORMATION:

Preamble Glossary of Terms and Abbreviations

The following are abbreviations of terms used in the preamble.

2016v1 2016 Version 1 Emissions Modeling Platform

2016v2 2016 Version 2 Emissions Modeling Platform

4-Step Framework 4-Step Interstate Transport Framework

ABC Associated Builders and Contractors

ACS American Community Survey

ACT Alternative Control Techniques

AEO Annual Energy Outlook

AQAT Air Quality Assessment Tool

AQS Air Quality System

BACT Best Available Control Technology

BART Best Available Retrofit Technology

BOF Basic Oxygen Furnace

BPT Benefit Per Ton

C1C2 Category 1 and Category 2

C3 Category 3

CAA or Act Clean Air Act

CAIR Clean Air Interstate Rule

CBI Confidential Business Information

CCR Coal Combustion Residual

CDC Centers for Disease Control and Prevention

CDX Central Data Exchange

CEDRI Compliance and Emissions Data Reporting Interface

CEMS Continuous Emissions Monitoring Systems

CES Clean Energy Standards

CFB Circulating Fluidized Bed Units

CHP Combined Heat and Power

CMDB Control Measures Database

CMV Commercial Marine Vehicle

CoST Control Strategy Tool

CPT Cost Per Ton

CRA Congressional Review Act

CSAPR Cross-State Air Pollution Rule

DAHS Data Acquisition and Handling System

DOE Department of Energy

EAF Electric Arc Furnace

EGU Electric Generating Unit

EIA U.S. Energy Information Agency

EIS Emissions Inventory System

EISA Energy Independence and Security Act

ELG Effluent Limitation Guidelines

E.O. Executive Order

EPA or the Agency United States Environmental Protection Agency

ERT Electronic Reporting Tool

FERC Federal Energy Regulatory Commission

FFS Findings of Failure to Submit

FIP Federal Implementation Plan

GIS Geographic Information System

g/hp-hr grams per horsepower per hour

HDGHG Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles

HEDD High Electricity Demand Days

ICI Industrial, Commercial, and Institutional

I/M Inspection and Maintenance

IPM Integrated Planning Model

IRA Inflation Reduction Act

LAER Lowest Achievable Emission Rate

LDC Local Distribution Company

LME Low Mass Emissions

LNB Low-NO_X Burners

MATS Mercury and Air Toxics Standards

MCM Menu of Control Measures

MDA8 Maximum Daily Average 8-Hour

MJO Multi-Jurisdictional Organization

MOU Memorandum of Understanding

MOVES Motor Vehicle Emissions Simulator

MSAT2 Mobile Source Air Toxics Rule

MWC Municipal Waste Combustor

NAAQS National Ambient Air Quality Standards

NACAA National Association of Clean Air Agencies

NAICS North American Industry Classification System

NEEDS National Electric Energy Data System

NEI National Emissions Inventory

NERC North American Electric Reliability Corporation

NESHAP National Emissions Standards for Hazardous Air Pollutants

NMB Normalized Mean Bias

NME Normalized Mean Error

No SISNOSE No Significant Economic Impact on a Substantial Number of Small Entities

Non-EGU Non-Electric Generating Unit

NODA Notice of Data Availability

NO_X  Nitrogen Oxides

NREL National Renewable Energy Lab

NSCR Non-Selective Catalytic Reduction

NSPS New Source Performance Standard

NSR New Source Review

NTTAA National Technology Transfer and Advancement Act

OFA Over-Fire Air

OMB United States Office of Management and Budget

OSAT/APCA Ozone Source Apportionment Technology/Anthropogenic Precursor Culpability Analysis

OTC Ozone Transport Commission

OTR Ozone Transport Region

OTSA Oklahoma Tribal Statistical Area

PDF Portable Document Format

PEMS Predictive Emissions Monitoring Systems

PM_2.5  Fine Particulate Matter

ppb parts per billion

ppm parts per million

ppmv parts per million by volume

ppmvd parts per million by volume, dry

PRA Paperwork Reduction Act

PSD Prevention of Significant Deterioration

PTE Potential to Emit

RACT Reasonably Available Control Technology

RATA Relative Accuracy Test Audit

RCF Relative Contribution Factor

RFA Regulatory Flexibility Act

RICE Reciprocating Internal Combustion Engines

ROP Rate of Progress

RPS Renewable Portfolio Standards

RRF Relative Response Factor

RTC Response to Comments

RTO Regional Transmission Organization

SAFETEA Safe, Accountable, Flexible, Efficient, Transportation Equity Act

SCC Source Classification Code

SCR Selective Catalytic Reduction

SIL Significant Impact Level

SIP State Implementation Plan

SMOKE Sparse Matrix Operator Kernel Emissions

SNCR Selective Non-Catalytic Reduction

SO₂  Sulfur Dioxide

tpd ton per day

TAS Treatment as State

TSD Technical Support Document

UMRA Unfunded Mandates Reform Act

VMT Vehicle Miles Traveled

VOCs Volatile Organic Compounds

WRAP Western Regional Air Partnership

WRF Weather Research and Forecasting

Table of Contents

I. Executive Summary

A. Purpose of the Regulatory Action

1. Emissions Limitations for EGUs Established by the Final Rule

2. Emissions Limitations for Industrial Stationary Point Sources Established by the Final Rule

B. Summary of the Regulatory Framework of the Rule

C. Costs and Benefits

II. General Information

A. Does this action apply to me?

B. What action is the Agency taking?

C. What is the Agency's legal authority for taking this action?

D. What actions has the EPA previously issued to address regional ozone transport?

III. Air Quality Issues Addressed and Overall Rule Approach

A. The Interstate Ozone Transport Air Quality Challenge

1. Nature of Ozone and the Ozone NAAQS

2. Ozone Transport

3. Health and Environmental Effects

B. Final Rule Approach

1. The 4-Step Interstate Transport Framework

a. Step 1 Approach

b. Step 2 Approach

c. Step 3 Approach

d. Step 4 Approach

2. FIP Authority for Each State Covered by the Rule

C. Other CAA Authorities for This Action

1. Withdrawal of Proposed Error Correction for Delaware

2. Application of Rule in Indian Country and Necessary or Appropriate Finding

a. Indian Country Subject to Tribal Jurisdiction

b. Indian Country Subject to State Implementation Planning Authority

D. Severability

IV. Analyzing Downwind Air Quality Problems and Contributions From Upwind States

A. Selection of Analytic Years for Evaluating Ozone Transport Contributions to Downwind Air Quality Problems

B. Overview of Air Quality Modeling Platform

C. Emissions Inventories

1. Foundation Emissions Inventory Data Sets

2. Development of Emissions Inventories for EGUs

a. EGU Emissions Inventories Supporting This Rule

b. Impact of the Inflation Reduction Act on EGU Emissions

3. Development of Emissions Inventories for Stationary Industrial Point Sources

4. Development of Emissions Inventories for Onroad Mobile Sources

5. Development of Emissions Inventories for Commercial Marine Vessels

6. Development of Emissions Inventories for Other Nonroad Mobile Sources

7. Development of Emissions Inventories for Nonpoint Sources

D. Air Quality Modeling To Identify Nonattainment and Maintenance Receptors

E. Methodology for Projecting Future Year Ozone Design Values

F. Pollutant Transport From Upwind States

1. Air Quality Modeling To Quantify Upwind State Ozone Contributions

2. Application of Ozone Contribution Screening Threshold

a. States That Contribute Below the Screening Threshold

b. States That Contribute Above the Screening Threshold

G. Treatment of Certain Monitoring Sites in California and Implications for Oregon's Good Neighbor Obligations for the 2015 Ozone NAAQS

V. Quantifying Upwind-State NO_X Emissions Reduction Potential To Reduce Interstate Ozone Transport for the 2015 Ozone NAAQS

A. The Multi-Factor Test for Determining Significant Contribution

B. Identifying Control Stringency Levels

1. EGU NO_X Mitigation Strategies

a. Optimizing Existing SCRs

b. Installing State-of-the-Art NO_X Combustion Controls

c. Optimizing Already Operating SNCRs or Turning on Idled Existing SNCRs

d. Installing New SNCRs

e. Installing New SCRs

f. Generation Shifting

g. Other EGU Mitigation Measures

2. Non-EGU or Stationary Industrial Source NO_X Mitigation Strategies

3. Other Stationary Sources NO_X Mitigation Strategies

a. Municipal Solid Waste Units

b. Electric Generating Units Less Than or Equal to 25 MW

c. Cogeneration Units

4. Mobile Source NO_X Mitigation Strategies

C. Control Stringencies Represented by Cost Threshold ($ per ton) and Corresponding Emissions Reductions

1. EGU Emissions Reduction Potential by Cost Threshold

2. Non-EGU or Industrial Source Emissions Reduction Potential

D. Assessing Cost, EGU and Industrial Source NO_X Reductions, and Air Quality

1. EGU Assessment

2. Stationary Industrial Sources Assessment

3. Combined EGU and Non-EGU Assessment

4. Over-Control Analysis

VI. Implementation of Emissions Reductions

A. NO_X Reduction Implementation Schedule

1. 2023–2025: EGU NO_X Reductions Beginning in 2023

2. 2026 and Later Years: EGU and Stationary Industrial Source NO_X Reductions Beginning in 2026

a. EGU Schedule for 2026 and Later Years

b. Non-EGU or Industrial Source Schedule for 2026 and Later Years

B. Regulatory Requirements for EGUs

1. Trading Program Background and Overview of Revisions

a. Current CSAPR Trading Program Design Elements and Identified Concerns

b. Enhancements To Maintain Selected Control Stringency Over Time

i. Revised Emissions Budget-Setting Process

ii. Allowance Bank Recalibration

c. Enhancements To Improve Emissions Performance at Individual Units

i. Unit-Specific Backstop Daily Emissions Rates

ii. Unit-Specific Emissions Limitations Contingent on Assurance Level Exceedances

d. Responses to General Comments on the Revisions to the Group 3 Trading Program

2. Expansion of Geographic Scope

3. Applicability and Tentative Identification of Newly Affected Units

4. State Emissions Budgets

a. Methodology for Determining Preset State Emissions Budgets for the 2023 through 2029 Control Periods

b. Methodology for Determining Dynamic State Emissions Budgets for Control Periods in 2026 Onwards

c. Final Preset State Emissions Budgets

5. Variability Limits and Assurance Levels

6. Annual Recalibration of Allowance Bank

7. Unit-Specific Backstop Daily Emissions Rates

8. Unit-Specific Emissions Limitations Contingent on Assurance Level Exceedances

9. Unit-Level Allowance Allocation and Recordation Procedures

a. Set-Asides of Portions of State Emissions Budgets

b. Allocations to Existing Units, Including Units That Cease Operation

c. Allocations From Portions of State Emissions Budgets Set Aside for New Units

d. Incorrectly Allocated Allowances

10. Monitoring and Reporting Requirements

a. Monitor Certification Deadlines

b. Additional Recordkeeping and Reporting Requirements

11. Designated Representative Requirements

12. Transitional Provisions

a. Prorating Emissions Budgets, Assurance Levels, and Unit-Level Allowance Allocations in the Event of an Effective Date After May 1, 2023

b. Creation of Additional Group 3 Allowance Bank for 2023 Control Period

c. Recall of Group 2 Allowances for Control Periods After 2022

13. Conforming Revisions to Regulations for Other CSAPR Trading Programs

C. Regulatory Requirements for Stationary Industrial Sources

1. Pipeline Transportation of Natural Gas

2. Cement and Concrete Product Manufacturing

3. Iron and Steel Mills and Ferroalloy Manufacturing

4. Glass and Glass Product Manufacturing

5. Boilers at Basic Chemical Manufacturing, Petroleum and Coal Products Manufacturing, Pulp, Paper, and Paperboard Mills, Iron and Steel and Ferroalloys Manufacturing, and Metal Ore Mining Facilities

a. Coal-fired Industrial Boilers

b. Oil-fired Industrial Boilers

c. Natural gas-fired Industrial Boilers

6. Municipal Waste Combustors

D. Submitting a SIP

1. SIP Option To Modify Allocations for 2024 under EGU Trading Program

2. SIP Option To Modify Allocations for 2025 and Beyond Under EGU Trading Program

3. SIP Option To Replace the Federal EGU Trading Program With an Integrated State EGU Trading Program

4. SIP Revisions That Do Not Use the New Trading Program

5. SIP Revision Requirements for Non-EGU or Industrial Source Control Requirements

E. Title V Permitting

1. Title V Permitting Considerations for EGUs

2. Title V Permitting Considerations for Industrial Stationary Sources

F. Relationship to Other Emissions Trading and Ozone Transport Programs

1. NO_X SIP Call

2. Acid Rain Program

3. Other CSAPR Trading Programs

VII. Environmental Justice Analytical Considerations and Stakeholder Outreach and Engagement

A. Introduction

B. Analytical Considerations

C. Outreach and Engagement

VIII. Costs, Benefits, and Other Impacts of the Final Rule

IX. Summary of Changes to the Regulatory Text for the Federal Implementation Plans and Trading Programs for EGUs

A. Amendments to FIP Provisions in 40 CFR Part 52

B. Amendments to Group 3 Trading Program and Related Regulations

C. Transitional Provisions

D. Clarifications and Conforming Revisions

X. Statutory and Executive Order Reviews

A. Executive Order 12866: Regulatory Planning and Review and Executive Order 13563: Improving Regulation and Regulatory Review

B. Paperwork Reduction Act (PRA)

1. Information Collection Request for EGUs

2. Information Collection Request for Non-EGUs

C. Regulatory Flexibility Act (RFA)

D. Unfunded Mandates Reform Act (UMRA)

E. Executive Order 13132: Federalism

F. Executive Order 13175: Consultation and Coordination With Indian Tribal Governments

G. Executive Order 13045: Protection of Children From Environmental Health Risks and Safety Risks

H. Executive Order 13211: Actions Concerning Regulations That Significantly Affect Energy Supply, Distribution or Use

I. National Technology Transfer and Advancement Act (NTTAA)

J. Executive Order 12898: Federal Actions To Address Environmental Justice in Minority Populations and Low-Income Populations

K. Congressional Review Act

L. Determinations Under CAA Section 307(b)(1) and (d)

I. Executive Summary

This final rule resolves the interstate transport obligations of 23 states under CAA section 110(a)(2)(D)(i)(I), referred to as the “good neighbor provision” or the “interstate transport provision” of the Act, for the 2015 ozone NAAQS. On October 1, 2015, the EPA revised the primary and secondary 8-hour standards for ozone to 70 parts per billion (ppb). States were required to submit to EPA ozone infrastructure State Implementation Plan (SIP) revisions to fulfill interstate transport obligations for the 2015 ozone NAAQS by October 1, 2018. The EPA proposed the subject rule to address outstanding interstate ozone transport obligations for the 2015 ozone NAAQS in the Federal Register on April 6, 2022 (87 FR 20036).

The EPA is making a finding that interstate transport of ozone precursor emissions from 23 upwind states (Alabama, Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin) is significantly contributing to nonattainment or interfering with maintenance of the 2015 ozone NAAQS in downwind states, based on projected ozone precursor emissions in the 2023 ozone season. The EPA is issuing FIP requirements to eliminate interstate transport of ozone precursor emissions from these 23 states that significantly contributes to nonattainment or interferes with maintenance of the NAAQS in downwind states. The EPA is not finalizing its proposed error correction for Delaware's ozone transport SIP, and we are deferring final action at this time on the proposed FIPs for Tennessee and Wyoming pending further review of the updated air quality and contribution modeling and analysis developed for this final action. As discussed in section III of this document, the EPA's updated analysis of 2023 suggests that the states of Arizona, Iowa, Kansas, and New Mexico may be significantly contributing to one or more nonattainment or maintenance receptors. The EPA is not making any final determinations with respect to these states in this action but intends to address these states, along with Tennessee and Wyoming, in a subsequent action or actions.

The EPA is finalizing FIP requirements for 21 states for which the Agency has, in a separate action, disapproved (or partially disapproved) ozone transport SIP revisions that were submitted for the 2015 ozone NAAQS: Alabama, Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Texas, Utah, West Virginia, and Wisconsin. See88 FR 9336. In this final rule, the EPA is issuing FIPs for two states—Pennsylvania and Virginia—for which the EPA issued Findings of Failure to Submit for 2015 ozone NAAQS transport SIPs. See84 FR 66612 (December 5, 2019). Under CAA section 301(d)(4), the EPA is extending FIP requirements to apply in Indian country located within the upwind geography of the final rule, including Indian reservation lands and other areas of Indian country over which the EPA or a tribe has demonstrated that a tribe has jurisdiction.

This final rule defines ozone season nitrogen oxides (NO_X) emissions performance obligations for Electric Generating Unit (EGU) sources and fulfills those obligations by implementing an allowance-based ozone season trading program beginning in 2023. This rule also establishes emissions limitations beginning in 2026 for certain other industrial stationary sources (referred to generally as “non-Electric Generating Units” (non-EGUs)). Taken together, these regulatory requirements will fully eliminate the amount of emissions that constitute the covered states' significant contribution to nonattainment and interference with maintenance in downwind states for purposes of the 2015 ozone NAAQS.

This final rule implements the necessary emissions reductions as follows. Under the FIP requirements, EGUs in 22 states (Alabama, Arkansas, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin) are required to participate in a revised version of the Cross-State Air Pollution Rule (CSAPR) NO_X Ozone Season Group 3 Trading Program that was previously established in the Revised CSAPR Update. In addition to reflecting emissions reductions based on the Agency's determination of the necessary control stringency in this rule, the revised trading program includes several enhancements to the program's design to better ensure achievement of the selected control stringency on all days of the ozone season and over time. For 12 states already required to participate in the CSAPR NO_X Ozone Season Group 3 Trading Program (Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, New Jersey, New York, Ohio, Pennsylvania, Virginia, and West Virginia) under the Revised CSAPR Update (with respect to the 2008 ozone NAAQS), the FIPs are amended by the revisions to the Group 3 trading program regulations. For seven states currently covered by the CSAPR NO_X Ozone Season Group 2 Trading Program under SIPs or FIPs, the EPA is issuing new FIPs for two states (Alabama and Missouri) and amending existing FIPs for five states (Arkansas, Mississippi, Oklahoma, Texas, and Wisconsin) to transition EGU sources in these states from the Group 2 program to the revised Group 3 trading program, beginning with the 2023 ozone season. The EPA is issuing new FIPs for three states not currently covered by any CSAPR NO_X ozone season trading program: Minnesota, Nevada, and Utah.

This rulemaking requires emissions reductions in the selected control stringency to be achieved as expeditiously as practicable and, to the extent possible, by the next applicable nonattainment dates for downwind areas for the 2015 ozone NAAQS. Thus, initial emissions reductions from EGUs will be required beginning in the 2023 ozone season and prior to the August 3, 2024, attainment date for areas classified as Moderate nonattainment for the 2015 ozone NAAQS.

The remaining emissions reduction obligations will be phased in as soon as possible thereafter. Substantial additional reductions from potential new post-combustion control installations at EGUs as well as from installation of new pollution controls at non-EGUs, also referred to in this action as industrial sources, will phase in beginning in the 2026 ozone season, associated with the August 3, 2027, attainment date for areas classified as Serious nonattainment for the 2015 ozone NAAQS. The EPA had proposed to require all emissions reductions to eliminate significant contribution to be in place by the 2026 ozone season. While we continue to view 2026 as the appropriate analytic year for purposes of applying the 4-step interstate transport framework, as discussed in section V.D.4 and VI.A.2 of this document, the final rule will allow individual facilities limited additional time to fully implement the required emissions reductions where the owner or operator demonstrates to the EPA's satisfaction that more rapid compliance is not possible. For EGUs, the emissions trading program budget stringency associated with retrofit of post-combustion controls will be phased in over two ozone seasons (2026–2027). For industrial sources, this final rule provides a process for individual facilities to seek a one year extension, with the possibility of up to two additional years, based on a specific showing of necessity.

The EGU emissions reductions are based on the feasibility of control installation for EGUs in 19 states that remain linked to downwind nonattainment and maintenance receptors in 2026. These 19 states are: Arkansas, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, and West Virginia. The emissions reductions required for EGUs in these states are based primarily on the potential retrofit of additional post-combustion controls for NO_X on most coal-fired EGUs and a portion of oil/gas-fired EGUs that are currently lacking such controls.

The EPA is finalizing, with some modifications from proposal in response to comments, certain additional features in the allowance-based trading program approach for EGUs, including dynamic adjustments of the emissions budgets and recalibration of the allowance bank over time as well as backstop daily emissions rate limits for large coal-fired units. The purpose of these enhancements is to better ensure that the emissions control stringency the EPA found necessary to eliminate significant contribution at Step 3 of the 4-step interstate transport framework is maintained over time in Step 4 implementation and is durable to changes in the power sector. These enhancements ensure the elimination of significant contribution is maintained both in terms of geographical distribution (by limiting the degree to which individual sources can avoid making emissions reductions) and in terms of temporal distribution (by better ensuring emissions reductions are maintained throughout each ozone season, year over year). As we further discuss in section V.D of this document, these changes do not alter the stringency of the emissions trading program over time. Rather, they ensure that the trading program (as the method of implementation at Step 4) remains aligned with the determinations made at Step 3. These enhancements are further discussed in section VI.B of this document.

The EPA is making a finding that NO_X emissions from certain non-EGU sources are significantly contributing to nonattainment or interfering with maintenance of the 2015 ozone NAAQS and that cost-effective controls for NO_X emissions reductions are available in certain industrial source categories that would result in meaningful air quality improvements in downwind receptors. The EPA is establishing emissions limitations beginning in 2026 for non-EGU sources located within 20 states: Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, and West Virginia. The final rule establishes NO_X emissions limitations during the ozone season for the following unit types for sources in non-EGU industries: reciprocating internal combustion engines in Pipeline Transportation of Natural Gas; kilns in Cement and Cement Product Manufacturing; reheat furnaces in Iron and Steel Mills and Ferroalloy Manufacturing; furnaces in Glass and Glass Product Manufacturing; boilers in Iron and Steel Mills and Ferroalloy Manufacturing, Metal Ore Mining, Basic Chemical Manufacturing, Petroleum and Coal Products Manufacturing, and Pulp, Paper, and Paperboard Mills; and combustors and incinerators in Solid Waste Combustors and Incinerators.

A. Purpose of the Regulatory Action

The purpose of this rulemaking is to protect public health and the environment by reducing interstate transport of certain air pollutants that significantly contribute to nonattainment, or interfere with maintenance, of the 2015 ozone NAAQS in downwind states. Ground-level ozone has detrimental effects on human health as well as vegetation and ecosystems. Acute and chronic exposure to ozone in humans is associated with premature mortality and certain morbidity effects, such as asthma exacerbation. Ozone exposure can also negatively impact ecosystems by limiting tree growth, causing foliar injury, and changing ecosystem community composition. Section III of this document provides additional evidence of the harmful effects of ozone exposure on human health and the environment. Studies have established that ozone air pollution can be transported over hundreds of miles, with elevated ground-level ozone concentrations occurring in rural and metropolitan areas. Assessments of ozone control approaches have concluded that control strategies targeting reduction of NO_X emissions are an effective method to reduce regional-scale ozone transport.

CAA section 110(a)(2)(D)(i)(I) requires states to prohibit emissions that will contribute significantly to nonattainment or interfere with maintenance in any other state with respect to any primary or secondary NAAQS. Within 3 years of the EPA promulgating a new or revised NAAQS, all states are required to provide SIP submittals, often referred to as “infrastructure SIPs,” addressing certain requirements, including the good neighbor provision. See CAA section 110(a)(1) and (2). The EPA must either approve or disapprove such submittals or make a finding that a state has failed to submit a complete SIP revision. As with any other type of SIP under the Act, when the EPA disapproves an interstate transport SIP or finds that a state failed to submit an interstate transport SIP, the CAA requires the EPA to issue a FIP to directly implement the measures necessary to eliminate significant contribution under the good neighbor provision. See generally CAA section 110(k) and 110(c). As such, in this rule, the EPA is finalizing requirements to fully address good neighbor obligations for the covered states for the 2015 ozone NAAQS under its authority to promulgate FIPs under CAA section 110(c). By eliminating significant contribution from these upwind states, this rule will make substantial and meaningful improvements in air quality by reducing ozone levels at the identified downwind receptors as well as many other areas of the country. At any time after the effective date of this rule, states may submit a Good Neighbor SIP to replace the FIP requirements contained in this rule, subject to EPA approval under CAA section 110(a).

The EPA conducted air quality modeling for the 2023 and 2026 analytic years to identify (1) the downwind areas identified as “receptors” (which are associated with monitoring sites) that are expected to have trouble attaining or maintaining the 2015 ozone NAAQS in the future and (2) the contribution of ozone transport from upwind states to the downwind air quality problems. We use the term “downwind” to describe those states or areas where a receptor is located, and we use the term “upwind” to describe states whose emissions are linked to one or more receptors. States may be both downwind and upwind depending on the receptor or linkage in question. Section IV of this document provides a full description of the results of the EPA's updated air quality modeling and relevant analyses for the rulemaking, including a discussion of how updates to the modeling and air quality analysis following the proposed rule have resulted in some modest changes in the overall geography of the final rule. Based on the EPA's air quality analysis, the 23 upwind states covered in this action are linked above the 1 percent of the NAAQS threshold to downwind air quality problems in downwind states. The EPA intends to expeditiously review the updated air quality modeling and related analyses to address potential good neighbor requirements of six additional states—Arizona, Iowa, Kansas, New Mexico, Tennessee, and Wyoming—in a subsequent action. The EPA had previously approved 2015 ozone transport SIPs submitted by Oregon and Delaware, but in the proposed FIP action the EPA found these states potentially to be linked in the modeling supporting our proposal. We proposed to issue an error correction for our prior approval of Delaware's 2015 ozone transport SIP; however, in this final rule, the EPA is withdrawing the proposed error correction and the proposed FIP for Delaware, because our updated modeling for this final rule confirms that Delaware is not linked above the 1 percent of NAAQS threshold ( see section III.C.1 of this document for additional information). The EPA is deferring finalizing a finding at this time for Oregon ( see section IV.G of this document for additional information).

1. Emissions Limitations for EGUs Established by the Final Rule

In this rule, the EPA is issuing FIP requirements that apply the provisions of the CSAPR NO_X Ozone Season Group 3 Trading Program as revised in the rule to EGU sources within the borders of the following 22 states: Alabama, Arkansas, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin. Implementation of the revised trading program provisions begins in the 2023 ozone season.

The EPA is expanding the CSAPR NO_X Ozone Season Group 3 Trading Program beginning in the 2023 ozone season. Specifically, the FIPs require power plants within the borders of the 22 states listed in the previous paragraph to participate in an expanded and revised version of the CSAPR NO_X Ozone Season Group 3 Trading Program created by the Revised CSAPR Update. Affected EGUs within the borders of the following 12 states currently participating in the Group 3 Trading Program under existing FIPs remain in the program, with revised provisions beginning in the 2023 ozone season, under this rule: Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, New Jersey, New York, Ohio, Pennsylvania, Virginia, and West Virginia. The FIPs also require affected EGUs within the borders of the following seven states currently covered by the CSAPR NO_X Ozone Season Group 2 Trading Program (the “Group 2 trading program”) under existing FIPs or existing SIPs to transition from the Group 2 program to the revised Group 3 trading program beginning with the 2023 control period: Alabama, Arkansas, Mississippi, Missouri, Oklahoma, Texas, and Wisconsin. Finally, the EPA is issuing new FIPs for EGUs within the borders of three states not currently covered by any existing CSAPR trading program for seasonal NO_X emissions: Minnesota, Nevada, and Utah. Sources in these states will enter the Group 3 trading program in the 2023 control period following the effective date of the final rule. Refer to section VI.B of this document for details on EGU regulatory requirements.

2. Emissions Limitations for Industrial Stationary Point Sources Established by the Final Rule

The EPA is issuing FIP requirements that include new NO_X emissions limitations for industrial or non-EGU sources in 20 states, with sources expected to demonstrate compliance no later than 2026. The EPA is requiring emissions reductions from non-EGU sources to address interstate transport obligations for the 2015 ozone NAAQS for the following 20 states: Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia and West Virginia.

The EPA is establishing emissions limitations for the following unit types in non-EGU industries: reciprocating internal combustion engines in Pipeline Transportation of Natural Gas; kilns in Cement and Cement Product Manufacturing; reheat furnaces in Iron and Steel Mills and Ferroalloy Manufacturing; furnaces in Glass and Glass Product Manufacturing; boilers in Iron and Steel Mills and Ferroalloy Manufacturing, Metal Ore Mining, Basic Chemical Manufacturing, Petroleum and Coal Products Manufacturing, and Pulp, Paper, and Paperboard Mills; and combustors and incinerators in Solid Waste Combustors and Incinerators. Refer to Table II.A–1 for a list of North American Industry Classification System (NAICS) codes for each entity included for regulation under this rule.

B. Summary of the Regulatory Framework of the Rule

The EPA is applying the 4-step interstate transport framework developed and used in CSAPR, the CSAPR Update, the Revised CSAPR Update, and other previous ozone transport rules under the authority provided in CAA section 110(a)(2)(D)(i)(I). The 4-step interstate transport framework provides a stepwise method for the EPA to define and implement good neighbor obligations for the 2015 ozone NAAQS. The four steps are as follows: (Step 1) identifying downwind receptors that are expected to have problems attaining or maintaining the NAAQS; (Step 2) determining which upwind states contribute to these identified problems in amounts sufficient to “link” them to the downwind air quality problems ( i.e., in this rule as in prior transport rules beginning with CSAPR in 2011, above a contribution threshold of 1 percent of the NAAQS); (Step 3) for states linked to downwind air quality problems, identifying upwind emissions that significantly contribute to downwind nonattainment or interfere with downwind maintenance of the NAAQS through a multifactor analysis; and (Step 4) for states that are found to have emissions that significantly contribute to nonattainment or interfere with maintenance of the NAAQS in downwind areas, implementing the necessary emissions reductions through enforceable measures. The remainder of this section provides a general overview of the EPA's application of the 4-step framework as it applies to the provisions of the rule; additional details regarding the EPA's approach are found in section III of this document.

To apply the first step of the 4-step framework to the 2015 ozone NAAQS, the EPA performed air quality modeling to project ozone concentrations at air quality monitoring sites in 2023 and 2026. The EPA evaluated projected ozone concentrations for the 2023 analytic year at individual monitoring sites and considered current ozone monitoring data at these sites to identify receptors that are anticipated to have problems attaining or maintaining the 2015 ozone NAAQS. This analysis of projected ozone concentrations was then repeated for 2026.

To apply the second step of the framework, the EPA used air quality modeling to quantify the contributions from upwind states to ozone concentrations in 2023 and 2026 at downwind receptors. Once quantified, the EPA then evaluated these contributions relative to a screening threshold of 1 percent of the NAAQS ( i.e., 0.70 ppb). States with contributions that equaled or exceeded 1 percent of the NAAQS were identified as warranting further analysis at Step 3 of the 4-step framework to determine if the upwind state significantly contributes to nonattainment or interference with maintenance in a downwind state. States with contributions below 1 percent of the NAAQS were considered not to significantly contribute to nonattainment or interfere with maintenance of the NAAQS in downwind states.

Based on the EPA's most recent air quality modeling and contribution analysis using 2023 as the analytic year, the EPA finds that the following 23 states have contributions that equal or exceed 1 percent of the 2015 ozone NAAQS, and, thereby, warrant further analysis of significant contribution to nonattainment or interference with maintenance of the NAAQS: Alabama, Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin.

There are locations in California to which Oregon contributes greater than 1 percent of the NAAQS; the EPA proposed that downwind areas represented by these monitoring sites in California should not be considered interstate ozone transport receptors at Step 1. However, the EPA is deferring finalizing a finding at this time for Oregon ( see section IV.G of this document for additional information).

Based on the air quality analysis presented in section IV of this document, the EPA finds that, with the exception of Alabama, Minnesota, and Wisconsin, the states found linked in 2023 will continue to contribute above the 1 percent of the NAAQS threshold to at least one receptor whose nonattainment and maintenance concerns persist through the 2026 ozone season. As a result, the EPA's evaluation of significantly contributing emissions at Step 3 for Alabama, Minnesota, and Wisconsin is limited to emissions reductions achievable by the 2023 and 2024 ozone seasons.

At the third step of the 4-step framework, the EPA applied a multifactor test that incorporates cost, availability of emissions reductions, and air quality impacts at the downwind receptors to determine the amount of ozone precursor emissions from the linked upwind states that “significantly” contribute to downwind nonattainment or maintenance receptors. The EPA is applying the multifactor test described in section V.A of this document to both EGU and industrial sources. The EPA assessed the potential emissions reductions in 2023 and 2026, as well as in intervening and later years to determine the emissions reductions required to eliminate significant contribution in 2023 and future years where downwind areas are projected to have potential problems attaining or maintaining the 2015 ozone NAAQS.

For EGU sources, the EPA evaluated the following set of widely-available NO_X emissions control technologies: (1) fully operating existing selective catalytic reduction (SCR) controls, including both optimizing NO_X removal by existing operational SCRs and turning on and optimizing existing idled SCRs; (2) installing state-of-the-art NO_X combustion controls; (3) fully operating existing selective non-catalytic reduction (SNCR) controls, including both optimizing NO_X removal by existing operational SNCRs and turning on and optimizing existing idled SNCRs; (4) installing new SNCRs; (5) installing new SCRs; and (6) generation shifting. For the reasons explained in section V of this document and supported by the “Technical Support Document (TSD) for the Final Federal Good Neighbor Plan for the 2015 Ozone National Ambient Air Quality Standard, Docket ID No. EPA–HQ–OAR–2021–0668, EGU NO_X Mitigation Strategies Final Rule TSD” (Mar. 2023), hereinafter referred to as the EGU NO_X Mitigation Strategies Final Rule TSD, included in the docket for this action, the EPA determines that for the regional, multi-state scale of this rulemaking, only fully operating and optimizing existing SCRs and existing SNCRs (EGU NO_X emissions controls options 1 and 3 in the list earlier) are possible for the 2023 ozone season. The EPA determined that state-of-the-art NO_X combustion controls at EGUs (emissions control option 2 in the list above) are available by the beginning of the 2024 ozone season. See section V.B.1 of this document for a full discussion of EPA's analysis of NO_X emissions mitigation strategies for EGU sources.

The EPA is requiring control stringency levels that offer the most incremental NO_X emissions reduction potential from EGUs—among the uniform mitigation measures assessed for the covered region—and the most corresponding downwind ozone air quality improvements to the extent feasible in each year analyzed. The EPA is making a finding that the required controls provide cost-effective reductions of NO_X emissions that will provide substantial improvements in downwind ozone air quality to address interstate transport obligations for the 2015 ozone NAAQS in a timely manner. These controls represent greater stringency in upwind EGU controls than in the EPA's most recent ozone transport rulemakings, such as the CSAPR Update and the Revised CSAPR Update. However, programs to address interstate ozone transport based on the retrofit of post-combustion controls are by no means unprecedented. In prior ozone transport rulemakings such as the NO_X SIP Call and the Clean Air Interstate Rule (CAIR), the EPA established EGU budgets premised on the widespread availability of retrofitting EGUs with post-combustion emissions controls such as SCR. While these programs successfully drove many EGUs to retrofit post-combustion controls, other EGUs throughout the present geography of linked upwind states continue to operate without such controls and continue to emit at relatively high rates more than 20 years after similar units reduced these emissions under prior interstate ozone transport rulemakings.

Furthermore, the CSAPR Update provided only a partial remedy for eliminating significant contribution for the 2008 ozone NAAQS, as needed to obtain available reductions by the 2017 ozone season. In that rule, the EPA made no determination regarding the appropriateness of more stringent EGU NO_X controls that would be required for a full remedy for interstate transport for the 2008 ozone NAAQS. Following the remand of the CSAPR Update in Wisconsin v. EPA, 938 F.3d 303 (D.C. Cir. 2019) ( Wisconsin), the EPA again declined to require the retrofit of new post-combustion controls on EGUs in the Revised CSAPR Update, but that determination was based on a specific timing consideration: downwind air quality problems under the 2008 ozone NAAQS were projected to resolve before post-combustion control retrofits could be accomplished on a fleetwide, regional scale. See86 FR 23054, 23110 (April 30, 2021).

In this rulemaking, the EPA is addressing good neighbor obligations for the more protective 2015 ozone NAAQS, and the Agency observes ongoing and persistent contribution from upwind states to ozone nonattainment and maintenance receptors in downwind states under that NAAQS. As further discussed in section V of this document, the nature of this contribution warrants a greater degree of control stringency than the EPA determined to be necessary to eliminate significant contribution of ozone transport in prior CSAPR rulemakings. In this rule, the EPA is requiring emissions performance levels for EGU NO_X control strategies commensurate with those determined to be necessary in the NO_X SIP Call and CAIR.

Based on the Step 3 analysis described in section V of this document, the EPA finds that emissions reductions commensurate with the full operation of all existing post-combustion controls (both SCRs and SNCRs) and state-of-the-art combustion control upgrades constitute the Agency's selected control stringency for EGUs within the borders of 22 states linked to downwind nonattainment or maintenance in 2023 (Alabama, Arkansas, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin). For 19 of those states that are also linked in 2026 (Arkansas, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, and West Virginia), the EPA is determining that the selected EGU control stringency also includes emissions reductions commensurate with the retrofit of SCR at coal-fired units of 100 MW or greater capacity (excepting circulating fluidized bed units (CFB)), new SNCR on coal-fired units of less than 100 MW capacity and on CFBs of any capacity size, and SCR on oil/gas steam units greater than 100 MW that have historically emitted at least 150 tons of NO_X per ozone season.

To identify appropriate control strategies for non-EGU sources to achieve NO_X emissions reductions that would result in meaningful air quality improvements in downwind areas, for the proposed FIP, the EPA evaluated air quality modeling information, annual emissions, and information about potential controls to determine which industries, beyond the power sector, could have the greatest impact in providing ozone air quality improvements in affected downwind states. Once the EPA identified the industries, the EPA used its Control Strategy Tool to identify potential emissions units and control measures and to estimate emissions reductions and compliance costs associated with application of non-EGU emissions control measures. The technical memorandum Screening Assessment of Potential Emissions Reductions, Air Quality Impacts, and Costs from Non-EGU Emissions Units for 2026 lays out the analytical framework and data used to prepare proxy estimates for 2026 of potentially affected non-EGU facilities and emissions units, emissions reductions, and costs. This information helped shape the proposal and final rule. To further evaluate the industries and emissions unit types identified by the screening assessment and to establish the applicability criteria and proposed emissions limits, the EPA reviewed Reasonably Available Control Technology (RACT) rules, New Source Performance Standards (NSPS) rules, National Emissions Standards for Hazardous Air Pollutants (NESHAP) rules, existing technical studies, rules in approved SIPs, consent decrees, and permit limits. That evaluation is detailed in the “Technical Support Document (TSD) for the Proposed Rule, Docket ID No. EPA–HQ–OAR–2021–0668, Non-EGU Sectors TSD” (Dec. 2021), hereinafter referred to as the Proposed Non-EGU Sectors TSD, prepared for the proposed FIP.

In this final rule, the EPA is retaining the industries and many of the emissions unit types included in the proposal in its findings of significant contribution at Step 3, as discussed in section V of this document. As discussed in the memorandum for the final rule, titled “Summary of Final Rule Applicability Criteria and Emissions Limits for Non-EGU Emissions Units, Assumed Control Technologies for Meeting the Final Emissions Limits, and Estimated Emissions Units, Emissions Reductions, and Costs,” the EPA uses the 2019 emissions inventory, the list of emissions units estimated to be captured by the applicability criteria, the assumed control technologies that would meet the emissions limits, and information on control efficiencies and default cost/ton values from the Control Measures Database, to estimate NO_X emissions reductions and costs for the year 2026. In this final rule, the EPA made changes to the applicability criteria and emissions limits following consideration of comments on the proposal and reassessed the overall non-EGU emissions reduction strategy based on the factors at Step 3 to render a judgment as to whether the level of emissions control that would be achievable from these units meets the criteria for “significant contribution.” In the final rule, we affirm our proposed determinations of which industries and emissions units are potentially impactful and warrant further analysis at Step 3, and we find that the available emissions reductions are cost-effective and make meaningful improvements at the identified downwind receptors. For a detailed discussion of the changes, between the proposal and this final rule, in emissions unit types included and in emissions limits, see section VI.C. of this document.

The EPA performed air quality analysis using the Ozone Air Quality Assessment Tool (AQAT) to evaluate the air quality improvements anticipated to result from the implementation of the selected EGU and non-EGU emissions reduction strategies. See section V.D of this document. We also used AQAT to determine whether the emissions reductions for both EGUs and non-EGUs potentially create an “over-control” scenario. As in prior transport rules following the holdings in EME Homer City, overcontrol would be established if the record indicated that, for any given state, there is a less stringent emissions control approach for that state, by which (1) the expected ozone improvements would be sufficient to resolve all of the downwind receptor(s) to which that state is linked; or (2) the expected ozone improvements would reduce the upwind state's ozone contributions below the screening threshold ( i.e., 1 percent of the NAAQS or 0.70 ppb) to all of linked receptors. The EPA's over-control analysis, discussed in section V.D.4 of this document, shows that the control stringencies for EGU and non-EGU sources in this final rule do not over-control upwind states' emissions either with respect to the downwind air quality problems to which they are linked or with respect to the 1 percent of the NAAQS contribution threshold, such that over-control would trigger re-evaluation at Step 3 for any linked upwind state.

Based on the multi-factor test applied to both EGU and non-EGU sources and our subsequent assessment of over-control, the EPA finds that the selected EGU and non-EGU control stringencies constitute the elimination of significant contribution and interference with maintenance, without over-controlling emissions, from the 23 upwind states subject to EGU and non-EGU emissions reductions requirements under the rule. For additional details about the multi-factor test and the over-control analysis, see the document titled “Technical Support Document (TSD) for the Final Federal Good Neighbor Plan for the 2015 Ozone National Ambient Air Quality Standard, Docket ID No. EPA–HQ–OAR–2021–0668, Ozone Transport Policy Analysis Proposed Rule TSD” (Mar. 2023), hereinafter referred to as Ozone Transport Policy Analysis Final Rule TSD, included in the docket for this rulemaking.

In this fourth step of the 4-step framework, the EPA is including enforceable measures in the promulgated FIPs to achieve the required emissions reductions in each of the 23 states. Specifically, the FIPs require covered power plants within the borders of 22 states (Alabama, Arkansas, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin) to participate in the CSAPR NO_X Ozone Season Group 3 Trading Program created by the Revised CSAPR Update. Affected EGUs within the borders of the following 12 states currently participating in the Group 3 Trading Program will remain in the program, with revised provisions beginning in the 2023 ozone season, under this rule: Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, New Jersey, New York, Ohio, Pennsylvania, Virginia, and West Virginia. Affected EGUs within the borders of the following seven states currently covered by the CSAPR NO_X Ozone Season Group 2 Trading Program (the “Group 2 trading program”)—Alabama, Arkansas, Mississippi, Missouri, Oklahoma, Texas, and Wisconsin—will transition from the Group 2 program to the revised Group 3 trading program beginning with the 2023 control period, and affected EGUs within the borders of three states not currently covered by any CSAPR trading program for seasonal NO_X emissions—Minnesota, Nevada, and Utah—will enter the Group 3 trading program in the 2023 control period following the effective date of the final rule. In addition, the EPA is revising other aspects of the Group 3 trading program to better ensure that this method of implementation at Step 4 provides a durable remedy for the elimination of the amount of emissions deemed to constitute significant contribution at Step 3 of the interstate transport framework. These enhancements, summarized later in this section, are designed to operate together to maintain that degree of control stringency over time, thus improving emissions performance at individual units and offering a necessary measure of assurance that NO_X pollution controls will be operated throughout each ozone season, as described in section VI.B of this document. This rulemaking does not revise the budget stringency and geography of the existing CSAPR NO_X Ozone Season Group 1 trading program. Aside from the seven states moving from the Group 2 trading program to the Group 3 trading program under the final rule, this rule otherwise leaves unchanged the budget stringency of the existing CSAPR NO_X Ozone Season Group 2 trading program.

The EPA is establishing preset ozone season NO_X emissions budgets for each ozone season from 2023 through 2029, using generally the same Group 3 trading program budget-setting methodology used in the Revised CSAPR Update, as explained in section VI.B of this document and as shown in Table I.B–1. The preset budgets for the 2026 through 2029 ozone seasons incorporate EGU emissions reductions to eliminate significant contribution and also take into account a substantial number of known retirements over that period to ensure the elimination of significant contribution is maintained as intended by this rule. These budgets serve as floors and may be supplanted by a budget that the EPA calculates for that control period using more recent information (a “dynamic budget”) if that dynamic budget yields a higher level of allowable emissions—still consistent with the Step 3 level of emissions control stringency—than the preset budget. As reflected in Table I.B–1, and accounting for both the stringency of the rule and known fleet change, the 2026 preset budget is 23 percent lower than the 2025 preset budget; the 2027 preset budget is 20 percent lower than the 2026 preset budget; the 2028 preset budget is 4 percent lower than the 2027 preset budget; and the 2029 preset budget is 8 percent lower than the 2028 preset budget.

While it is possible that additional EGUs may seek to retire in this 2026–2029 period than are currently scheduled and captured in the preset emissions budgets, it is also possible that EGUs with currently scheduled retirements may adjust their retirement timing to accommodate the timing of replacement generation and/or transmission upgrades necessitated by their retirement. While the EPA designed this final rule to provide preset budgets through 2029 to incorporate known retirement-related emissions reductions to ensure the elimination of significant contribution as identified at Step 3 is maintained over time, the use of these floors also provides generators and grid operators enhanced certainty regarding the minimum amount of allowable NO_X emissions for reliability planning through the 2020s. By providing the opportunity for dynamic budgets to subsequently calibrate budgets to any unforeseen increases in fleet demand, it also ensures this rule will not interfere with ongoing retirement scheduling or adjustments and thus is robust to future uncertainty during a transition period.

The EPA also believes the likelihood and magnitude of a scenario in which a state's preset emissions budgets during this period would authorize more emissions than the corresponding dynamic budget is low. As described elsewhere, dynamic budgets are incorporated to best calibrate the rule's stringency to future unknown changes to the fleet. The circumstances in which a dynamic budget would produce a level of allowable emissions less than preset budgets is most pronounced for future periods in which there is a high degree of unknown retirements (increasing the risk that budgets are not appropriately calibrated to the reduced fossil fuel heat input post retirement). However, the 2026–2029 period presents a case where retirement planning has been announced with greater lead time than normal due to a combination of utility 2030 decarbonization commitments, and Effluent Limitation Guideline (ELG) and Coal Combustion Residual (CCR) alternative compliance pathways available to units planning to cease combustion of coal by December 31, 2028. For each of these existing rules, facilities that are planning to retire have already conveyed that intention to EPA in order to take advantage of the alternative compliance pathways available to such facilities. Therefore, the likelihood of unknown retirements—leading to lower dynamic budgets—is much lower than typical for this time horizon. This makes EPA's balanced use of preset emissions budgets or dynamic budgets if they exceed preset levels a reasonable mechanism to accommodate planning and fleet transition dynamics during this period. The need and reasoning for the limited-period preset budget floor is further discussed in section VI.B.4.

For control periods in 2030 and thereafter, the emissions budgets will be the amounts calculated for each state and noticed to the public roughly one year before the control period, using the dynamic budget-setting methodology. In this manner, the stringency of the program will be secured and sustained in the dynamic budgets of this program, regardless of whatever EGU transition activities ultimately occur in this 2026–2029 transition period.

The budget-setting methodology that the EPA will use to determine dynamic budgets for each control period starting with 2026 is an extension of the methodology used to determine the preset budgets and will be used routinely to determine emissions budgets for each future control period in the year before that control period, with each emissions budget reflecting the latest available information on the composition and utilization of the EGU fleet at the time that emissions budget is determined. The stringency of the dynamic emissions budgets will simply reflect the stringency of the emissions control strategies selected in the rulemaking more consistently over time and ensure that the annual updates would eliminate emissions determined to be unlawful under the good neighbor provision. As already noted, for the control periods in which both preset budgets and dynamic budgets are determined for a state ( i.e., 2026 through 2029), the state's dynamic budget will apply only if it is higher than the state's preset budget. See section VI.B of this document for additional discussion of the EPA's method for adjusting emissions budgets to ensure elimination of significant contribution from EGU sources in the linked upwind states.

In conjunction with the levels of the emissions budgets, the carryover of unused allowances for use in future control periods as banked allowances affects the ability of a trading program to maintain the rule's selected control stringency and related EGU effective emissions rate performance level as the EGU fleet evolves over time. Unrestricted banking of allowances allows what might otherwise be temporary surpluses of allowances in some individual control periods to accumulate into a long-term allowance surplus that reduces allowance prices and weakens the trading program's incentives to control emissions. To prevent this outcome, the EPA is also revising the Group 3 trading program by adding provisions that establish a routine recalibration process for banked allowances using a target percentage of 21 percent for the 2024–2029 control periods and 10.5 percent for control periods in 2030 and later years.

As an enhancement to the structure of the trading program originally promulgated in the Revised CSAPR Update, the EPA is also establishing backstop daily emissions rates for coal steam EGUs greater than or equal to 100 MW in covered states. Starting with the 2024 control period, a 3-for-1 allowance surrender ratio (instead of the usual 1-for-1 surrender ratio) will apply to emissions during the ozone season from any large coal-fired EGU with existing SCR controls exceeding by more than 50 tons a daily average NO_X emissions rate of 0.14 lb/mmBtu. The daily average emissions rate provisions will apply to large coal-fired EGUs without existing SCR controls starting with the second control period in which newly installed SCR controls are operational at the unit, but not later than the 2030 control period.

The backstop daily emissions rates work in tandem with the ozone season emissions budgets to ensure the elimination of significant contribution as determined at Step 3 is maintained over time and more consistently throughout each ozone season. They will offer downwind receptor areas a necessary measure of assurance that they will be protected on a daily basis during the ozone season by more continuous and consistent operation of installed pollution controls. The EPA's experience with the CSAPR trading programs has revealed instances where EGUs have reduced their SCRs' performance on a given day, or across the entire ozone seasons in some cases, including high ozone days. In addition to maintaining a mass-based seasonal requirement, this rule will achieve a much more consistent level of emissions control in line with our Step 3 determination of significant contribution while maintaining compliance flexibility consistent with that determination. These trading program improvements will promote consistent emissions control performance across the power sector in the linked upwind states, which protects communities living in downwind ozone nonattainment areas from exceedances of the NAAQS that might otherwise occur.

The EPA is including enforceable emissions control requirements that will apply during the ozone season (annually from May to September) for nine non-EGU industries in the promulgated FIPs to achieve the required emissions reductions in 20 states with remaining interstate transport obligations for the 2015 ozone NAAQS in 2026: Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, and West Virginia. These requirements would apply to all existing emissions units and to any future emissions units constructed in the covered states that meet the relevant applicability criteria. Thus, the emissions limitations for non-EGU sources and associated compliance requirements would apply in all 20 states listed in this paragraph, even if some of these states do not currently have any existing emissions units meeting the applicability criteria for the identified industries.

Based on our evaluation of the time required to install controls at the types of non-EGU sources covered by this rule, the EPA has identified the 2026 ozone season as a reasonable compliance date for industrial sources. The EPA is therefore finalizing control requirements for non-EGU sources that take effect in 2026. However, in recognition of comments and additional information indicating that not all facilities may be capable of meeting the control requirements by that time, the final rule provides a process by which the EPA may grant compliance extensions of up to 1 year, which if approved by the EPA, would require compliance no later than the 2027 ozone season, followed by an additional possible extension of up to 2 more years, where specific criteria are met. For sources located in the 20 states listed in the previous paragraph, the EPA is finalizing the NO_X emissions limits listed in Table I.B–2 for reciprocating internal combustion engines in Pipeline Transportation of Natural Gas; the NO_X emissions limits listed in Table I.B–3 for kilns in Cement and Cement Product Manufacturing; the NO_X emissions limits listed in Table I.B–4 for reheat furnaces in Iron and Steel Mills and Ferroalloy Manufacturing; the NO_X emissions limits listed in Table I.B–5 for furnaces in Glass and Glass Product Manufacturing; the NO_X emissions limits listed in Table I.B–6 for boilers in Iron and Steel Mills and Ferroalloy Manufacturing, Metal Ore Mining, Basic Chemical Manufacturing, Petroleum and Coal Products Manufacturing, and Pulp, Paper, and Paperboard Mills; and the NO_X emissions limits listed in Table I.B–7 for combustors and incinerators in Solid Waste Combustors or Incinerators.

Based on evaluation of comments received, the EPA is not, at this time, finalizing the source cap limit as proposed at 87 FR 20046 ( see section VII.C.2 of the April 6, 2022, Proposal).

Section VI.C of this document provides an overview of the applicability criteria, compliance assurance requirements, and the EPA's rationale for establishing these emissions limits and control requirements for each of the non-EGU industries covered by the rule.

The remainder of this preamble is organized as follows: section II of this document outlines general applicability criteria and describes the EPA's legal authority for this rule and the relationship of the rule to previous interstate ozone transport rulemakings. Section III of this document describes the human health and environmental challenges posed by interstate transport contributions to ozone air quality problems, as well as the EPA's overall approach for addressing interstate transport for the 2015 ozone NAAQS in this rule. Section IV of this document describes the Agency's analyses of air quality data to inform this rulemaking, including descriptions of the air quality modeling platform and emissions inventories used in the rule, as well as the EPA's methods for identifying downwind air quality problems and upwind states' ozone transport contributions to downwind states. Section V of this document describes the EPA's approach to quantifying upwind states' obligations in the form of EGU NO_X control stringencies and non-EGU emissions limits. Section VI of this document describes key elements of the implementation schedule for EGU and non-EGU emissions reductions requirements, including details regarding the revised aspects of the CSAPR NO_X Group 3 trading program and compliance deadlines, as well as regulatory requirements and compliance deadlines for non-EGU sources. Section VII of this document discusses the environmental justice analysis of the rule, as well as outreach and engagement efforts. Section VIII of this document describes the expected costs, benefits, and other impacts of this rule. Section IX of this document provides a summary of changes to the existing regulatory text applicable to the EGUs covered by this rule; and section X of this document discusses the statutory and executive orders affecting this rulemaking.

C. Costs and Benefits

A summary of the key results of the cost-benefit analysis that was prepared for this final rule is presented in Table I.C–1. Table I.C–1 presents estimates of the present values (PV) and equivalent annualized values (EAV), calculated using discount rates of 3 and 7 percent as recommended by OMB's Circular A–4, of the health and climate benefits, compliance costs, and net benefits of the final rule, in 2016 dollars, discounted to 2023. The estimated monetized net benefits are the estimated monetized benefits minus the estimated monetized costs of the final rule. These results present an incomplete overview of the effects of the rule because important categories of benefits—including benefits from reducing other types of air pollutants, and water pollution—were not monetized and are therefore not reflected in the cost-benefit tables. We anticipate that taking non-monetized effects into account would show the rule to be more net beneficial than this table reflects.

As shown in Table I.C–1, the PV of the monetized health benefits, associated with reductions in ozone and PM _2.5 concentrations, of this final rule, discounted at a 3-percent discount rate, is estimated to be about $200 billion ($200,000 million), with an EAV of about $13 billion ($13,000 million). At a 7-percent discount rate, the PV of the monetized health benefits is estimated to be $130 billion ($130,000 million), with an EAV of about $12 billion ($12,000 million). The PV of the monetized climate benefits, associated with reductions in GHG emissions, of this final rule, discounted at a 3-percent discount rate, is estimated to be about $15 billion ($15,000 million), with an EAV of about $970 million. The PV of the monetized compliance costs, discounted at a 3-percent rate, is estimated to be about $14 billion ($14,000 million), with an EAV of about $910 million. At a 7-percent discount rate, the PV of the compliance costs is estimated to be about $9.4 billion ($9,400 million), with an EAV of about $770 million.

II. General Information

A. Does this action apply to me?

This rule affects EGU and non-EGU sources, and regulates the groups identified in Table II.A–1.

This table is not intended to be exhaustive, but rather provides a guide for readers regarding entities likely to be regulated by this rule. This table lists the types of entities that the EPA is now aware could potentially be regulated by this rule. Other types of entities not listed in the table could also be regulated. To determine whether your EGU entity is regulated by this rule, you should carefully examine the applicability criteria found in 40 CFR 97.1004, which are unchanged in this rule. If you have questions regarding the applicability of this rule to a particular entity, consult the person listed in the FOR FURTHER INFORMATION CONTACT section.

B. What action is the Agency taking?

The EPA evaluated whether interstate ozone transport emissions from upwind states are significantly contributing to nonattainment, or interfering with maintenance, of the 2015 ozone NAAQS in any downwind state using the same 4-step interstate transport framework that was developed in previous ozone transport rulemakings. The EPA finds that emissions reductions are required from EGU and non-EGU sources in a total of 23 upwind states to eliminate significant contribution to downwind air quality problems for the 2015 ozone standard under the interstate transport provision of the CAA. The EPA will ensure that these NO_X emissions reductions are achieved by issuing FIP requirements for 23 states: Alabama, Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin.

The EPA is revising the existing CSAPR Group 3 Trading Program to include additional states beginning in the 2023 ozone season. EGUs in three states not currently covered by any CSAPR trading program for seasonal NO_X emissions—Minnesota, Nevada, and Utah—will be added to the CSAPR Group 3 Trading Program under this rule. EGUs in twelve states currently participating in the Group 3 Trading Program will remain in the program under this rule: Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, New Jersey, New York, Ohio, Pennsylvania, Virginia, and West Virginia. EGUs in seven states (Alabama, Arkansas, Mississippi, Missouri, Oklahoma, Texas, and Wisconsin) will transition from the CSAPR Group 2 Trading Program to the CSAPR Group 3 Trading Program under this rule beginning in the 2023 ozone season. The EPA is establishing control stringency levels reflecting installation of state-of-the-art combustion controls on certain covered EGU sources in emissions budgets beginning in the 2024 ozone season. The EPA is establishing control stringency levels reflecting installation of new SCR or SNCR controls on certain covered EGU sources in emissions budgets beginning in the 2026 ozone season.

As a complement to the ozone season emissions budgets, the EPA is also establishing a backstop daily emissions rate of 0.14 lb/mmBtu for coal-fired steam units greater than or equal to 100 MW in covered states. The backstop emissions rate will first apply in 2024 for coal-fired steam sources with existing SCRs, and in the second control period in which a new SCR operates, but not later than 2030, for those currently without SCRs.

This rule establishes emissions limitations for non-EGU sources in 20 states: Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, and West Virginia. In these states, the EPA is establishing control requirements for the following unit types in non-EGU industries: reciprocating internal combustion engines in Pipeline Transportation of Natural Gas; kilns in Cement and Cement Product Manufacturing; reheat furnaces in Iron and Steel Mills and Ferroalloy Manufacturing; furnaces in Glass and Glass Product Manufacturing; boilers in Iron and Steel Mills and Ferroalloy Manufacturing, Metal Ore Mining, Basic Chemical Manufacturing, Petroleum and Coal Products Manufacturing, and Pulp, Paper, and Paperboard Mills; and combustors and incinerators in Solid Waste Combustors and Incinerators. See Table II.A–1 in this document for a list of NAICS codes for each entity included for regulation in this rule.

This rule reduces the transport of ozone precursor emissions to downwind areas, which is protective of human health and the environment because acute and chronic exposure to ozone are both associated with negative health impacts. Ozone exposure is also associated with negative effects on ecosystems. Additional information on the air quality issues addressed by this rule are included in section III of this document.

C. What is the Agency's legal authority for taking this action?

The statutory authority for this rule is provided by the CAA as amended (42 U.S.C. 7401 et seq.). Specifically, sections 110 and 301 of the CAA provide the primary statutory underpinnings for this rule. The most relevant portions of CAA section 110 are subsections 110(a)(1), 110(a)(2) (including 110(a)(2)(D)(i)(I)) and 110(c)(1)).

CAA section 110(a)(1) provides that states must make SIP submissions “within 3 years (or such shorter period as the Administrator may prescribe) after the promulgation of a national primary ambient air quality standard (or any revision thereof),” and that these SIP submissions are to provide for the “implementation, maintenance, and enforcement” of such NAAQS. The statute directly imposes on states the duty to make these SIP submissions, and the requirement to make the submissions is not conditioned upon the EPA taking any action other than promulgating a new or revised NAAQS.

The EPA has historically referred to SIP submissions made for the purpose of satisfying the applicable requirements of CAA sections 110(a)(1) and 110(a)(2) as “infrastructure SIP” or “iSIP” submissions. CAA section 110(a)(1) addresses the timing and general requirements for iSIP submissions, and CAA section 110(a)(2) provides more details concerning the required content of these submissions. It includes a list of specific elements that “[e]ach such plan” must address.

CAA section 110(c)(1) requires the Administrator to promulgate a FIP at any time within 2 years after the Administrator: (1) finds that a state has failed to make a required SIP submission; (2) finds a SIP submission to be incomplete pursuant to CAA section 110(k)(1)(C); or (3) disapproves a SIP submission. This obligation applies unless the state corrects the deficiency through a SIP revision that the Administrator approves before the FIP is promulgated.

CAA section 110(a)(2)(D)(i)(I), also known as the “good neighbor” provision, provides the primary basis for this rule. It requires that each state SIP include provisions sufficient to “prohibit[ ], consistent with the provisions of this subchapter, any source or other type of emissions activity within the State from emitting any air pollutant in amounts which will—(I) contribute significantly to nonattainment in, or interfere with maintenance by, any other State with respect to any [NAAQS].” The EPA often refers to the emissions reduction requirements under this provision as “good neighbor obligations” and submissions addressing these requirements as “good neighbor SIPs.”

Once the EPA promulgates a NAAQS, the EPA must designate areas as being in “attainment” or “nonattainment” of the NAAQS, or “unclassifiable.” CAA section 107(d). For ozone, nonattainment is further split into five classifications based on the severity of the violation—Marginal, Moderate, Serious, Severe, or Extreme. Higher classifications provide states with progressively more time to attain while imposing progressively more stringent control requirements. See CAA sections 181, 182. In general, states with nonattainment areas classified as Moderate or higher must submit plans to the EPA to bring these areas into attainment according to the statutory schedule. CAA section 182. If an area fails to attain the NAAQS by the attainment date associated with its classification, it is “bumped up” to the next classification. CAA section 181(b).

Section 301(a)(1) of the CAA gives the Administrator the general authority to prescribe such regulations as are necessary to carry out functions under the Act. Pursuant to this section, the EPA has authority to clarify the applicability of CAA requirements and undertake other rulemaking action as necessary to implement CAA requirements. CAA section 301 affords the Agency any additional authority that may be needed to make certain other changes to its regulations under 40 CFR parts 52, 75, 78, and 97, to effectuate the purposes of the Act. Such changes are discussed in section IX of this document.

Tribes are not required to submit state implementation plans. However, as explained in the EPA's regulations outlining Tribal Clean Air Act authority, the EPA is authorized to promulgate FIPs for Indian country as necessary or appropriate to protect air quality if a tribe does not submit, and obtain the EPA's approval of, an implementation plan. See40 CFR 49.11(a); see also CAA section 301(d)(4). In the proposed rule, the EPA proposed an “appropriate or necessary” finding under CAA section 301(d) and proposed tribal FIP(s) as necessary to implement the relevant requirements. The EPA is finalizing these determinations, as further discussed in section III.C.2 of this document.

D. What actions has the EPA previously issued to address regional ozone transport?

The EPA has issued several previous rules interpreting and clarifying the requirements of CAA section 110(a)(2)(D)(i)(I) with respect to the regional transport of ozone. These rules, and the associated court decisions addressing these rules, summarized here, provide important direction regarding the requirements of CAA section 110(a)(2)(D)(i)(I).

The “NO_X SIP Call,” promulgated in 1998, addressed the good neighbor provision for the 1979 1-hour ozone NAAQS. The rule required 22 states and the District of Columbia to amend their SIPs to reduce NO_X emissions that contribute to ozone nonattainment in downwind states. The EPA set ozone season NO_X budgets for each state, and the states were given the option to participate in a regional allowance trading program, known as the NO_X Budget Trading Program. The D.C. Circuit largely upheld the NO_X SIP Call in Michigan v. EPA, 213 F.3d 663 (D.C. Cir. 2000), cert. denied, 532 U.S. 904 (2001).

The EPA's next rule addressing the good neighbor provision, CAIR, was promulgated in 2005 and addressed both the 1997 fine particulate matter (PM_2.5 ) NAAQS and 1997 ozone NAAQS. CAIR required SIP revisions in 28 states and the District of Columbia to reduce emissions of sulfur dioxide (SO₂) or NO_X —important precursors of regionally transported PM_2.5 (SO₂ and annual NO_X) and ozone (summer-time NO_X). As in the NO_X SIP Call, states were given the option to participate in regional trading programs to achieve the reductions. When the EPA promulgated the final CAIR in 2005, the EPA also issued findings that states nationwide had failed to submit SIPs to address the requirements of CAA section 110(a)(2)(D)(i) with respect to the 1997 PM_2.5 and 1997 ozone NAAQS. On March 15, 2006, the EPA promulgated FIPs to implement the emissions reductions required by CAIR. CAIR was remanded to EPA by the D.C. Circuit in North Carolina v. EPA, 531 F.3d 896 (D.C. Cir.), modified on reh'g, 550 F.3d 1176 (D.C. Cir. 2008). For more information on the legal issues underlying CAIR and the D.C. Circuit's holding in North Carolina, refer to the preamble of the CSAPR rule.

In 2011, the EPA promulgated CSAPR to address the issues raised by the remand of CAIR. CSAPR addressed the two NAAQS at issue in CAIR and additionally addressed the good neighbor provision for the 2006 PM_2.5 NAAQS. CSAPR required 28 states to reduce SO₂ emissions, annual NO_X emissions, or ozone season NO_X emissions that significantly contribute to other states' nonattainment or interfere with other states' abilities to maintain these air quality standards. To align implementation with the applicable attainment deadlines, the EPA promulgated FIPs for each of the 28 states covered by CSAPR. The FIPs require EGUs in the covered states to participate in regional trading programs to achieve the necessary emissions reductions. Each state can submit a good neighbor SIP at any time that, if approved by EPA, would replace the CSAPR FIP for that state.

CSAPR was the subject of an adverse decision by the D.C. Circuit in August 2012. However, this decision was reversed in April 2014 by the Supreme Court, which largely upheld the rule, including the EPA's approach to addressing interstate transport in CSAPR. EPA v. EME Homer City Generation, L.P., 572 U.S. 489 (2014) ( EME Homer City I). The rule was remanded to the D.C. Circuit to consider claims not addressed by the Supreme Court. Id. In July 2015 the D.C. Circuit generally affirmed the EPA's interpretation of various statutory provisions and the EPA's technical decisions. EME Homer City Generation, L.P. v. EPA, 795 F.3d 118 (2015) ( EME Homer City II). However, the court remanded the rule without vacatur for reconsideration of the EPA's emissions budgets for certain states, which the court found may have over-controlled those states' emissions with respect to the downwind air quality problems to which the states were linked. Id. at 129–30, 138. For more information on the legal issues associated with CSAPR and the Supreme Court's and D.C. Circuit's decisions in the EME Homer City litigation, refer to the preamble of the CSAPR Update.

In 2016, the EPA promulgated the CSAPR Update to address interstate transport of ozone pollution with respect to the 2008 ozone NAAQS. The final rule updated the CSAPR ozone season NO_X emissions budgets for 22 states to achieve cost-effective and immediately feasible NO_X emissions reductions from EGUs within those states. The EPA aligned the analysis and implementation of the CSAPR Update with the 2017 ozone season to assist downwind states with timely attainment of the 2008 ozone NAAQS. The CSAPR Update implemented the budgets through FIPs requiring sources to participate in a revised CSAPR NO_X ozone season trading program beginning with the 2017 ozone season. As under CSAPR, each state could submit a good neighbor SIP at any time that, if approved by the EPA, would replace the CSAPR Update FIP for that state. The final CSAPR Update also addressed the remand by the D.C. Circuit of certain states' CSAPR phase 2 ozone season NO_X emissions budgets in EME Homer City II.

In December 2018, the EPA promulgated the CSAPR “Close-Out,” which determined that no further enforceable reductions in emissions of NO_X were required with respect to the 2008 ozone NAAQS for 20 of the 22 eastern states covered by the CSAPR Update.

The CSAPR Update and the CSAPR Close-Out were both subject to legal challenges in the D.C. Circuit. Wisconsin v. EPA, 938 F.3d 303 (D.C. Cir. 2019) ( Wisconsin); New York v. EPA, 781 Fed. App'x 4 (D.C. Cir. 2019) ( New York). In September 2019, the D.C. Circuit upheld the CSAPR Update in virtually all respects but remanded the rule because it was partial in nature and did not fully eliminate upwind states' significant contribution to nonattainment or interference with maintenance of the 2008 ozone NAAQS by “the relevant downwind attainment deadlines” in the CAA. Wisconsin, 938 F.3d at 313–15. In October 2019, the D.C. Circuit vacated the CSAPR Close-Out on the same grounds that it remanded the CSAPR Update in Wisconsin, specifically because the Close-Out rule did not address good neighbor obligations by “the next applicable attainment date” of downwind states. New York, 781 Fed. App'x at 7.

In response to the Wisconsin remand of the CSAPR Update and the New York vacatur of the CSAPR Close-Out, the EPA promulgated the Revised CSAPR Update on April 30, 2021. The Revised CSAPR Update found that the CSAPR Update was a full remedy for nine of the covered states. For the 12 remaining states, the EPA found that their projected 2021 ozone season NO_X emissions would significantly contribute to downwind states' nonattainment or maintenance problems. The EPA issued new or amended FIPs for these 12 states and required implementation of revised emissions budgets for EGUs beginning with the 2021 ozone season. Based on the EPA's assessment of remaining air quality issues and additional emissions control strategies for EGUs and emissions sources in other industry sectors (non-EGUs), the EPA determined that the NO_X emissions reductions achieved by the Revised CSAPR Update fully eliminated these states' significant contributions to downwind air quality problems for the 2008 ozone NAAQS. As under the CSAPR and the CSAPR Update, each state can submit a good neighbor SIP at any time that, if approved by the EPA, would replace the Revised CSAPR Update FIP for that state.

On March 3, 2023, the D.C. Circuit Court of Appeals denied the Midwest Ozone Group's (MOG) petition for review of the Revised CSAPR Update. MOG v. EPA, No. 21–1146 (D.C. Cir. March 3, 2023). The court noted that it has “exhaustively” addressed the interstate transport framework before, citing relevant cases, and “incorporate them herein by reference.” Slip Op. 1 n.1. In response to MOG's arguments, the court upheld the Agency's air quality analysis. Id. at 10–11. The court noted that in light of the statutory timing framework and court-ordered schedule the EPA was under, the Agency's methodological choices were reasonable and provided “an appropriately reliable projection of air quality conditions and contributions in 2021.” Id. at 11–12.

III. Air Quality Issues Addressed and Overall Rule Approach

A. The Interstate Ozone Transport Air Quality Challenge

1. Nature of Ozone and the Ozone NAAQS

Ground-level ozone is not emitted directly into the air but is created by chemical reactions between NO_X and volatile organic compounds (VOCs) in the presence of sunlight. Emissions from electric utilities and industrial facilities, motor vehicles, gasoline vapors, and chemical solvents are some of the major sources of NO_X and VOCs.

Because ground-level ozone formation increases with temperature and sunlight, ozone levels are generally higher during the summer months. Increased temperature also increases emissions of volatile man-made and biogenic organics and can also indirectly increase NO_X emissions ( e.g., increased electricity generation for air conditioning).

On October 1, 2015, the EPA strengthened the primary and secondary ozone standards to 70 ppb as an 8-hour level. Specifically, the standards require that the 3-year average of the fourth highest 24-hour maximum 8-hour average ozone concentration may not exceed 70 ppb as a truncated value ( i.e., digits to right of decimal removed). In general, areas that exceed the ozone standard are designated as nonattainment areas, pursuant to the designations process under CAA section 107(d), and are subject to heightened planning requirements depending on the severity of their nonattainment classification, see CAA sections 181, 182.

In the process of setting the 2015 ozone NAAQS, the EPA noted that the conditions conducive to the formation of ozone ( i.e., seasonally-dependent factors such as ambient temperature, strength of solar insolation, and length of day) differ by location, and that the Agency believes it is important that ozone monitors operate during all periods when there is a reasonable possibility of ambient levels approaching the level of the NAAQS. At that time, the EPA stated that ambient ozone concentrations in many areas could approach or exceed the level of the NAAQS, more frequently and during more months of the year compared with the historical ozone season monitoring lengths. Consequently, the EPA extended the ozone monitoring season for many locations. See80 FR 65416 for more details.

Furthermore, the EPA stated that in addition to being affected by changing emissions, future ozone concentrations may also be affected by climate change. Modeling studies in the EPA's Interim Assessment (U.S. EPA, 2009a) that are cited in support of the 2009 Greenhouse Gas Endangerment Finding under CAA section 202(a) (74 FR 66496, Dec. 15, 2009) as well as a recent assessment of potential climate change impacts (Fann et al., 2015) project that climate change may lead to future increases in summer ozone concentrations across the contiguous U.S. (80 FR 65300). The U.S. Global Change Research Program's Impacts of Climate Change on Human Health in the United States: A Scientific Assessment and Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment, Volume II reinforced these findings. The increase in ozone results from changes in local weather conditions, including temperature and atmospheric circulation patterns, as well as changes in ozone precursor emissions that are influenced by meteorology (Nolte et al., 2018). While the projected impact may not be uniform, climate change has the potential to increase average summertime ozone relative to a future without climate change. Climate change has the potential to offset some of the improvements in ozone air quality, and therefore some of the improvements in public health, that are expected from reductions in emissions of ozone precursors (80 FR 65300). The EPA responds to comments received on the impacts of climate change on ozone formation in section 11 of the Response to Comments ( RTC) document.

2. Ozone Transport

Studies have established that ozone formation, atmospheric residence, and transport occur on a regional scale ( i.e., thousands of kilometers) over much of the U.S. While substantial progress has been made in reducing ozone in many areas, the interstate transport of ozone precursor emissions remains an important contributor to peak ozone concentrations and high-ozone days during the summer ozone season.

The EPA has previously concluded in the NO_X SIP Call, CAIR, CSAPR, the CSAPR Update, and the Revised CSAPR Update that a regional NO_X control strategy would be effective in reducing regional-scale transport of ozone precursor emissions. NO_X emissions can be transported downwind as NO_X or as ozone after transformation in the atmosphere. In any given location, ozone pollution levels are impacted by a combination of background ozone concentration, local emissions, and emissions from upwind sources resulting from ozone transport, in conjunction with variable meteorological conditions. Downwind states' ability to meet health-based air quality standards such as the NAAQS is challenged by the transport of ozone pollution across state borders. For example, ozone assessments conducted for the October 2015 Regulatory Impact Analysis of the Final Revisions to the National Ambient Air Quality Standards for Ground-Level Ozone continue to show the importance of NO_X emissions for ozone transport. This analysis is included in the docket for this rulemaking.

Further, studies have found that EGU NO_X emissions reductions can be effective in reducing individual 8-hour peak ozone concentrations and in reducing 8-hour peak ozone concentrations averaged across the ozone season. For example, a study of the EGU NO_X reductions achieved under the NO_X Budget Trading Program ( i.e., the NO_X SIP Call) shows that regulating NO_X emissions in that program was highly effective in reducing ozone concentrations during the ozone season.

Previous regional ozone transport efforts, including the NO_X SIP Call, CAIR, CSAPR, the CSAPR Update, and the Revised CSAPR Update, required ozone season NO_X reductions from EGU sources to address interstate transport of ozone. Together with NO_X, the EPA has also identified VOCs as a precursor in forming ground-level ozone. Ozone formation chemistry can be “NO_X -limited,” where ozone production is primarily determined by the amount of NO_X emissions or “VOC-limited,” where ozone production is primarily determined by the amount of VOC emissions. The EPA and others have long regarded NO_X to be the more significant ozone precursor in the context of interstate ozone transport.

The EPA has determined that the regulation of VOCs as an ozone precursor is not necessary to eliminate significant contribution of ozone transport to downwind areas in this rule. As described in section V.A of this document, the EPA examined the results of the contribution modeling performed for this rule to identify the portion of the ozone contribution attributable to anthropogenic NO_X emissions versus VOC emissions from each linked upwind state to each downwind receptor. Our analysis of the ozone contribution from upwind states subject to regulation demonstrates that regional ozone concentrations affecting the vast majority of the downwind areas of air quality concern are NO_X -limited, rather than VOC-limited. Therefore, the rule's strategy for reducing regional-scale transport of ozone targets NO_X emissions from stationary sources to achieve the most effective reductions of ozone transport over the geography of the affected downwind areas. The potential impacts of NO_X mitigation strategies from other sources are discussed in section V.B of this document.

In section V of this document, the EPA describes the multi-factor test that is used to determine NO_X emissions reductions that are cost-effective and reduce interstate transport of ground-level ozone. Our analysis indicates that the EGU and non-EGU control requirements included in this rule will provide meaningful improvements in air quality at the downwind receptors. Based on the implementation schedule established in section VI.A of this document, the EPA finds that the regulatory requirements included in the rule are as expeditious as practicable and are aligned with the attainment schedule of downwind areas.

3. Health and Environmental Effects

Exposure to ambient ozone causes a variety of negative effects on human health, vegetation, and ecosystems. In humans, acute and chronic exposure to ozone is associated with premature mortality and certain morbidity effects, such as asthma exacerbation. In ecosystems, ozone exposure causes visible foliar injury, decreases plant growth, and affects ecosystem community composition. See EPA's October 2015 Regulatory Impact Analysis of the Final Revisions to the National Ambient Air Quality Standards for Ground-Level Ozone in the docket for this rulemaking for more information on the human health and ecosystem effects associated with ambient ozone exposure.

Commenters on prior ozone transport rules have asserted that VOC emissions harm underserved and overburdened communities experiencing disproportionate environmental health burdens and facing other environmental injustices. The EPA acknowledges that VOCs can contain toxic chemicals that are detrimental to public health. The EPA conducted a demographic analysis as part of the regulatory impact analysis for the 2015 revisions to the primary and secondary ozone NAAQS. This analysis, which is included in the docket for this rulemaking, found greater representation of minority populations in areas with poor air quality relative to the revised ozone standard than in the U.S. as a whole. The EPA concluded that populations in these areas would be expected to benefit from implementation of future air pollution control actions from state and local air agencies in implementing the strengthened standard. This rule is an example of air pollution control actions implemented by the Federal Government in support of the more protective 2015 ozone NAAQS, and populations living in downwind ozone nonattainment and maintenance areas are expected to benefit from improved air quality that will result from reducing ozone transport. Further discussion of the environmental justice analysis of this rule is located in section VII of this document and in the accompanying regulatory impact analysis, titled “Regulatory Impact Analysis for Final Federal Good Neighbor Plan Addressing Regional Ozone Transport for the 2015 Ozone National Ambient Air Quality Standard” [EPA–452/D–22–001], which is available in the docket for this rulemaking.

The Agency regulates exposure to toxic pollutant concentrations and ambient exposure to criteria pollutants other than ozone through other sections of the Act, such as the regulation of hazardous air pollutants under CAA section 112 or the process for revising and implementing the NAAQS under CAA sections 107–110. The purpose of the subject rulemaking is to protect public health and the environment by eliminating significant contribution from 23 states to nonattainment or maintenance of the 2015 ozone NAAQS to meet the requirements of the CAA's interstate transport provision. In this rule, the EPA continues to observe that requiring NO_X emissions reductions from stationary sources is an effective strategy for reducing regional ozone transport in the U.S.

The EPA responds to other comments received on the health and environmental impacts of ozone exposure in section 11 of the RTC document.

B. Final Rule Approach

1. The 4-Step Interstate Transport Framework

The EPA first developed a multi-step process to address the requirements of the good neighbor provision in the 1998 NO_X SIP Call and the 2005 CAIR. The Agency built upon this framework and further refined the methodology for addressing interstate transport obligations in subsequent rules such as CSAPR in 2011, the CSAPR Update in 2016, and the Revised CSAPR Update in 2021. In CSAPR, the EPA first articulated a “4-step framework” within which to assess interstate transport obligations for ozone. In this rule to address interstate transport obligations for the 2015 ozone NAAQS, the EPA is again utilizing the 4-step interstate transport framework. These steps are: (1) identifying downwind receptors that are expected to have problems attaining the NAAQS (nonattainment receptors) or maintaining the NAAQS (maintenance receptors); (2) determining which upwind states are “linked” to these identified downwind receptors based on a numerical contribution threshold; (3) for states linked to downwind air quality problems, identifying upwind emissions on a statewide basis that significantly contribute to downwind nonattainment or interfere with downwind maintenance of the NAAQS, considering cost- and air quality-based factors; and (4) for upwind states that are found to have emissions that significantly contribute to nonattainment or interfere with maintenance of the NAAQS in any downwind state, implementing the necessary emissions reductions through enforceable measures.

Comment: The EPA received comments supporting the Agency's use of the 4-step interstate transport framework as a permissible method for assigning the required amount of emissions reductions necessary to eliminate upwind states' significant contribution. Commenters also noted that the 4-step interstate transport framework was reviewed by the Supreme Court in EPA vs. EME Homer City Generation, 572 U.S. 489 (2014), and upheld. However, other commenters took exception to the overall approach of this proposed action. These commenters alleged that the EPA is ignoring the “flexibility” in addressing good neighbor obligations that it had purportedly suggested to states would be permissible in memoranda that the EPA issued in 2018. Commenters also raised concerns that the air quality modeling (2016v2) the EPA used to propose to disapprove SIP submittals and as the basis for the proposed FIP was not available to states at the time they made their submissions and that the changes in results at Steps 1 and 2 from prior rounds of modeling rendered the new modeling unreliable. Commenters also raised a number of arguments that the EPA should allow states an additional opportunity to submit SIPs before promulgating a FIP, advocated that the EPA should issue a “SIP call” under CAA section 110(k)(5), asked for the EPA to issue new or more specific guidance, or otherwise suggested that the EPA should defer acting to promulgate a FIP at this time.

Response: As an initial matter, comments regarding the EPA's basis for disapproving SIPs are beyond the scope of this action. To the extent these comments relate to the legal basis for the EPA to promulgate a FIP, the EPA disagrees that it is acting in a manner contrary to the memoranda it released in 2018 related to good neighbor obligations for the 2015 ozone NAAQS. Arguments that the EPA must or should allow states to re-submit SIP submissions based on the most recent modeling information before the EPA promulgates a FIP ignore the plain language of the statute and relevant caselaw. CAA section 110(c) authorizes the EPA to promulgate a FIP “at any time within 2 years” of a SIP disapproval. No provision of the Act requires the EPA to give states an additional opportunity to prepare a new SIP submittal once the EPA has proposed a FIP or proposed disapproval of a SIP submittal. Comments regarding the timing of the EPA's actions and calls for the EPA to allow time for states to resubmit SIPs are further addressed in RTC sections 1.1 and 2.4.

With regard to the need for the EPA to develop and issue guidance in addressing good neighbor obligations, in EPA v. EME Homer City Generation, L.P., the Supreme Court held that “nothing in the statute places the EPA under an obligation to provide specific metrics to States before they undertake to fulfill their good neighbor obligations.” While we have taken a different approach in some prior rulemakings by providing states with an opportunity to submit a SIP after we quantified the states' budgets ( e.g., the NO_X SIP Call and CAIR ), the CAA does not require such an approach.

2018 Memoranda. As commenters point out, the EPA issued three “memoranda” in 2018 to provide some assistance to states in developing these SIP submittals. Each memorandum made clear that the EPA's action on SIP submissions would be through a separate notice-and-comment rulemaking process and that SIP submissions seeking to rely on or take advantage of any so-called “flexibilities” in these memoranda would be carefully reviewed against the relevant legal requirements and technical information available to the EPA at the time it would take such rulemaking action. Further, certain aspects of discussions in those memoranda were specifically identified as not constituting agency guidance (especially Attachment A to the March 2018 memorandum, which comprised an unvetted list of external stakeholders' ideas). And, although outside the scope of this action, as the EPA has explained in disapproving states' SIP submittals, those submittals did not meet the terms of the August 2018 or October 2018 memoranda addressing contribution thresholds and maintenance receptors, respectively.

Commenters mistakenly view Attachment A to the March 2018 memorandum as constituting agency guidance. This memorandum was primarily issued to share modeling results for 2023 that represented the best information available to the Agency as of March 2018, while Attachment A then listed certain ideas from certain stakeholders that the EPA said could be further discussed among states and stakeholders. The EPA disagrees with commenters' characterization of the EPA's stance regarding these so-called “flexibilities” listed (without analysis) in Attachment A. The March 2018 memorandum provided, “While the information in this memorandum and the associated air quality analysis data could be used to inform the development of these SIPs, the information is not a final determination regarding states' obligations under the good neighbor provision.” The EPA again affirms that the concepts listed in Attachment A to the March 2018 memorandum require unique consideration, and these ideas do not constitute agency guidance with respect to transport obligations for the 2015 ozone NAAQS. Attachment A to the March 2018 memorandum identified a “Preliminary List of Potential Flexibilities” that could potentially inform SIP development. However, the EPA made clear in both the March 2018 memorandum and in Attachment A that the list of ideas was not endorsed by the Agency but rather “comments provided in various forums” on which the EPA sought “feedback from interested stakeholders.” Further, Attachment A stated, “EPA is not at this time making any determination that the ideas discussed below are consistent with the requirements of the CAA, nor are we specifically recommending that states use these approaches.” Attachment A to the March 2018 memorandum, therefore, does not constitute agency guidance, but was intended to generate further discussion around potential approaches to addressing ozone transport among interested stakeholders. The EPA emphasized in these memoranda that such alternative approaches must be technically justified and appropriate in light of the facts and circumstances of each particular state's submittal. To the extent states sought to develop or rely on one or more of these ideas in support of their SIP submissions, the EPA reviewed their technical and legal justifications for doing so.

Regarding the October 2018 memorandum, that document recognized that states may be able to demonstrate in their SIPs that conditions exist that would justify treating a monitoring site as not being a maintenance receptor despite results from our modeling methodology identifying it as such a receptor. The EPA explained that this demonstration could be appropriate under two circumstances: (1) the site currently has “clean data” indicating attainment of the 2015 ozone NAAQS based on measured air quality concentrations, or (2) the state believes there is a technical reason to justify using a design value from the baseline period that is lower than the maximum design value based on monitored data during the same baseline period. To justify such an approach, the EPA anticipated that any such showing would be based on an analytical demonstration that (1) meteorological conditions in the area of the monitoring site were conducive to ozone formation during the period of clean data or during the alternative base period design value used for projections; (2) ozone concentrations have been trending downward at the site since 2011 (and ozone precursor emissions of NO_X and VOC have also decreased); and (3) emissions are expected to continue to decline in the upwind and downwind states out to the attainment date of the receptor. Although this is beyond the scope of this action, the EPA explained in its final SIP disapproval action that no state successfully demonstrated that one of these alternative approaches is justified. In this action, our analysis of the air quality data and projections in section IV of this document indicate that trends in historic measured data do not necessarily support adopting a less stringent approach for identifying maintenance receptors for purposes of the 2015 ozone NAAQS. In fact, as explained in section III.B.1.a and IV.D of this document, the EPA has found in its analysis for this final rule that, in general, recent measured data from regulatory ambient air quality ozone monitoring sites suggest that a number of receptors with elevated ozone levels will persist in 2023 even though our traditional methodology at Step 1 did not identify these monitoring sites as receptors in 2023. Thus, the EPA is not acting inconsistently with that memorandum—the factual conditions that would need to exist for the suggested approaches of that memorandum to be applicable have not been demonstrated as being applicable or appropriate based on the relevant data.

Regarding the August 2018 memorandum, as discussed in section IV.F.2 of this document, for purposes of Step 2 of our ozone transport evaluation framework, we are applying a 1 percent of NAAQS threshold rather than a 1 ppb threshold, as this memorandum had suggested might be appropriate for states to apply as an alternative. The EPA is finalizing its proposed approach of consistently using a 1 percent of the NAAQS contribution threshold at Step 2 to evaluate whether states are linked to downwind nonattainment and maintenance concerns for purposes of this FIP.

The approach of this FIP ensures both national consistency across all states and consistency and continuity with our prior interstate transport actions for other NAAQS. Further, in this action the EPA is promulgating FIPs under the authority of CAA section 110(c). In doing so, the EPA has exercised its discretion to determine how to define and apply good neighbor obligations in place of the discretion states otherwise would exercise (subject to the EPA's approval as compliant with the Act). In general, the EPA is applying the 4-step interstate transport framework it devised over the course of its prior good neighbor rulemakings, including applying a consistent definition of nonattainment and maintenance-only receptors, and applying the 1 percent of NAAQS threshold at Step 2. The basis for these decisions is further explained in sections IV.F.1 and IV.F.2 of the document. These policy judgments reflect consistency with relevant good neighbor case law and past agency practice implementing the good neighbor provision as reflected in the original CSAPR, CSAPR Update, Revised CSAPR Update, and related rulemakings. Nationwide consistency in approach is particularly important in the context of interstate ozone transport, which is a regional-scale pollution problem involving the collective emissions of many smaller contributors. Effective policy solutions to the problem of interstate ozone transport dating back to the NO_X SIP Call (63 FR 57356 (October 27, 1998)) have necessitated the application of a uniform framework of policy judgments, and the EPA's framework applied here has been upheld as ensuring an “efficient and equitable” approach. See EME Homer City Generation, LP v. EPA, 572 U.S. 489, 519 (2014).

Updated modeling. The EPA had originally provided 2023 modeling results in its March 2018 memorandum, which used a 2011-based platform. Many states used this modeling in providing good neighbor SIP submittals for the 2015 ozone NAAQS. While our action on the SIP submittals is not within scope of this action, commenters claim the use of new modeling or other information not available to states at the time they made their submittals renders this action promulgating a FIP unlawful. Notwithstanding whether that is an accurate characterization of the EPA's basis for disapproving the SIPs, we note that the court in Wisconsin rejected this precise argument against the CSAPR Update FIPs as a collateral attack on the SIP disapprovals. 938 F.3d at 336 (“That is the hallmark of an improper collateral attack. The true gravamen of the claim lies in the agency's failure to timely act upon the States' SIP submissions and, relatedly, its reliance on data compiled after the SIP action deadline. Both go directly to the legitimacy of the SIP denials.”).

Nonetheless, we offer the following explanation of the evolution of the EPA's understanding of projected air quality conditions and contributions in 2023 resulting from the iterative nature of our modeling efforts. These modeling efforts are further addressed in section IV of this document. We acknowledge that to evaluate transport SIPs and support our proposed FIP the EPA reassessed receptors at Step 1 and states' contribution levels at Step 2 through additional modeling (2016v2) before proposing this action and have reassessed again to inform the final action (2016v3). At proposal, we relied on CAMx Version 7.10 and the 2016v2 emissions platform to make updated determinations regarding which receptors would likely exist in 2023 and which states are projected to contribute above the contribution threshold to those receptors. As explained in the preamble of the EPA's proposed FIP and further detailed in the “Air Quality Modeling Technical Support Document for the Federal Implementation Plan Addressing Regional Ozone Transport for the 2015 Ozone National Ambient Air Quality Standards Proposed Rulemaking” (Dec. 2021), hereinafter referred to as Air Quality Modeling Proposed Rule TSD, and the “Technical Support Document (TSD): Preparation of Emissions Inventories for the 2016v2 North American Emissions Modeling Platform” (Dec. 2021), hereinafter referred to as the 2016v2 Emissions Inventory TSD, both available in the docket for this action (docket ID no. EPA–HQ–OAR–2021–0668), this modeling built off of previous modeling iterations used to support the EPA's action on interstate transport obligations. The EPA periodically refines its modeling to ensure the results are as indicative as possible of air quality in future years. This includes making any necessary adjustments to our modeling platform and updating our emissions inventories to reflect current information, including information submitted during public comments on proposed actions.

For this final rule, the EPA has evaluated a raft of technical information and critiques of its 2016v2 modeling provided by commenters on this action (as well as comments on the SIP actions) and has responded to those comments and incorporated updates into the version of the modeling used to support this final rule (2016v3). As explained in section IV.B of the document, in response to additional information provided by stakeholders following a solicitation of feedback during the release of the 2016v2 emissions inventory and during the comment periods on the proposed SIP actions, the EPA has reviewed and revised its 2016v2 modeling platform and input since the platform was made available for comment. The new modeling platform 2016v3 was developed from this input, and the modeling results using platform 2016v3 are available with this action. See section IV of this document for further discussion. Thus, the EPA's final rule is based on a comprehensive record of data and technical evaluation, including the updated modeling information used at proposal (2016v2), the comments received on that modeling, and the latest modeling used in this final rule (2016v3).

The changes in projected outcomes at Steps 1 and 2 are a product of these changes; these updates between the data released in 2018 to now are an outgrowth of this iterative process, including updating the platform from a 2011 to a 2016 base year, updates to the emissions inventory information and other updates. It is reasonable for the Agency to improve its understanding of a situation before taking final action, and the Agency uses the best information available to it in taking this action.

Further, these modeling updates have not uniformly resulted in new linkages—the 2016v2 modeling, for instance, corroborated the proposed approval of Montana and supported approval of Colorado's SIP in October of 2022. Although some commenters indicate that our modeling iterations have provided differing outcomes and are therefore unreliable, this is not what the overall record indicates. Rather, in general, although the specifics of states' linkages may have changed to some extent, our modeling on the whole has provided consistent outcomes regarding which states are linked to downwind air quality problems. For example, the EPA's modeling shows that most states that were linked to one or more receptors using the 2011-based platform ( i.e., the March 2018 data release) are also linked to one or more receptors using the newer 2016-based platform. Because the new platform uses different meteorology ( i.e., 2016 instead of 2011), it is not unexpected that an upwind state would be linked to different receptors using 2011 versus 2016 meteorology. In addition, although a state may be linked to a different set of receptors, those receptors are within the same areas that have historically had a persistent air quality problem. Only three upwind states included in the FIP went from being unlinked to being linked in 2023 between the 2011-based modeling provided in the March 2018 memorandum and the 2016v3-based modeling—Alabama, Minnesota, and Nevada.

Additionally, we disagree with commenters who claim that the 2016v2 modeling results were sprung upon the states with the publication of the proposed SIP disapprovals. In fact, states had prior access to a series of data and modeling releases beginning as early as the publication of the 2016v1 modeling with the proposed Revised CSAPR Update in October 2020. States could have reviewed and used this technical information to understand and track how the EPA's modeling updates were affecting the list of potential receptors and linkages for the 2015 ozone NAAQS in the 2023 analytic year.

The 2016-based meteorology and boundary conditions used in the modeling have been available through the 2016v1 platform, which was used for the Revised CSAPR Update (proposed, 85 FR 68964; October 30, 2020). The updated emissions inventory files used in the current modeling were publicly released September 21, 2021, for stakeholder feedback, and have been available on our website since that time. The CAMx modeling software that the EPA used has likewise been publicly available for over a year before this final rule was proposed on April 6, 2022. CAMx version 7.10 was released by the model developer, Ramboll, in December 2020. On January 19, 2022, we released on our website and notified a wide range of stakeholders of the availability of both the modeling results for 2023 and 2026 (including contribution data) along with many key underlying input files.

By providing the 2016 meteorology and boundary conditions (used in the 2016v1 version) in fall of 2020, and by releasing updated emissions inventory information used in 2016v2 in September of 2021, we gave states and other interested parties multiple opportunities prior to proposal of this rule on April 6, 2022, to consider how our modeling updates could affect their status for purposes of evaluating potential linkages for the 2015 ozone NAAQS. In this final rule, we have updated our modeling to 2016v3, incorporating and reflecting the feedback and additional information we received through the multiple public comment opportunities the EPA made available on the 2016v2 modeling.

The EPA's development of and reliance on newer modeling is reasonable and is simply another iteration of the EPA's longstanding scientific and technical work to improve our understanding of air quality issues and causes going back many decades.

Comment: Commenters asserted that the EPA lacks authority under the good neighbor provision to do more than establish state-wide emissions budgets, which states may then implement through their own choice of emissions controls. The commenters claim that the EPA lacks authority to directly regulate emissions sources under the good neighbor provision, and they cite to case law that they view as establishing a “federalism bar” to direct Federal regulation. Commenters assert that the term “amounts” as used in the good neighbor provision prevents the agency from establishing emissions limits at individual sources, such as the non-EGU industrial units that the EPA proposed to regulate or implementing “enhancements” in its mass-based emissions trading approach for EGUs as it had proposed. Commenters claim these aspects of the rule are an unlawful or arbitrary and capricious departure from the EPA's prior transport rulemakings, which they claim only set mass-based emissions budgets as the means to eliminate “significant contribution.”

Response: To the extent these comments challenge the EPA's disapproval of states' 2015 ozone NAAQS good neighbor SIP submissions, they are out of scope of this action, which promulgates a FIP under the authority of CAA section 110(c)(1). To the extent commenters assert that the EPA does not have the authority to directly implement source-specific emissions control requirements or other emissions control measures, means, or techniques, including emissions trading programs, in the exercise of that FIP authority, the EPA disagrees. While the courts have long recognized that the states have wide discretion in the design of SIPs to attain and maintain the NAAQS, see, e.g., Union Electric Co v. EPA, 427 U.S. 246 (1976), when the EPA promulgates a FIP to cure a defective SIP, the Act, including the definition of a FIP in section 302(y), provides for the EPA to directly implement the Act's requirements. The EPA is granted authority to choose among a broad range of “emission limitations or other control measures, means, or techniques (including economic incentives, such as marketable permits or auctions of emissions allowances) . . . .” CAA section 302(y); see also CAA section 110(a)(2) (empowering states to implement an identical set of emissions control mechanisms).

The courts have also recognized that the EPA has broad authority to cure a defective SIP, that the EPA may exercise its own, independent regulatory authority in implementing a FIP in accordance with the CAA, and that the EPA in effect steps into the shoes of a state when it promulgates a FIP. See, e.g., Central Ariz. Water Conservation Dist. v. EPA, 990 F.2d 1531 (9th Cir. 1993); South Terminal Corp. v. EPA, 504 F.2d 646 (1st Cir. 1974). Accord Virginia v. EPA, 108 F.3d 1397, 1406–07 (D.C. Cir. 1997) (“The Federal Plan `provides an additional incentive for state compliance because it rescinds state authority to make the many sensitive and policy choices that a pollution control regime demands.'”) (quoting Natural Resources Defense Council v. Browner, 57 F.3d 1122, 1124 (D.C. Cir. 1995)). Cf. District of Columbia v. Train, 521 F.2d 971 (D.C. Cir. 1975), vacated sub nom. EPA v. Brown, 431 U.S. 99 (1977) (“[W]here cooperation [from states] is not forthcoming, we believe that the recourse contemplated by the commerce clause is direct federal regulation of the offending activity . . . .”).

These same principles apply where the EPA must promulgate a FIP to address good neighbor requirements under CAA section 110(a)(2)(D)(i)(I). The EPA has promulgated a series of FIPs in the past to address the relevant requirements for prior ozone and PM NAAQS. See, e.g., CAIR FIP, 71 FR 25328 (April 28, 2006); CSAPR, 76 FR 48208 (August 8, 2011); the CSAPR Update, 81 FR 74504 (October 26, 2016); and the Revised CSAPR Update, 86 FR 23054 (April 30, 2021). Courts have upheld the EPA's exercise of this authority. See EME Homer City Generation v. EPA, 572 U.S. 489 (2014); Wisconsin v. EPA, 938 F.3d 303 (D.C. Cir. 2019). Indeed, in EME Homer City, the U.S. Supreme Court held that the EPA is not obligated to provide guidance to states before acting on their good neighbor submissions or give states a second chance at correcting the deficiencies before promulgating a FIP, and the EPA may promulgate a FIP at any time after finalizing its disapproval of SIP submissions. 572 U.S. at 508–11.

The cases cited by commenters, which they refer to as establishing the Train - Virginia federalism bar, were not reviewing the exercise of the EPA's authority in promulgating a FIP under CAA section 110(c)(1) but rather were describing the scope of the EPA's authority in acting on SIP submissions under CAA section 110(k)(3) or in issuing a “SIP call” under section 110(k)(5). In those latter contexts, the courts have held that the EPA may not dictate the specific control measures states must implement to meet the Act's requirements. See Virginia, 108 F.3d at 1409–10. In Michigan, the D.C. Circuit upheld the EPA's exercise of CAA section 110(k)(5) authority in issuing the “NO_X SIP Call,” because, “EPA does not tell the states how to achieve SIP compliance. Rather, EPA looks to section 110(a)(2)(D) and merely provides the levels to be achieved by state-determined compliance mechanisms. . . . However, EPA made clear that states do not have to adopt the control scheme that EPA assumed for budget-setting purposes.” Michigan v. EPA, 213 F.3d 663, 687–88 (D.C. Cir. 2000).

Commenters' position that the EPA must provide similar flexibility to the states in this action ( i.e., only provide a general emissions reduction target and leave to states how to meet that target) is a non sequitur. The EPA is implementing a FIP in this action and must directly implement the necessary emissions controls. The EPA is not empowered to require states to implement FIP mandates. Such an approach would conflict with constitutional anti-commandeering principles, is not provided for in the Act, and would only constitute a partial implementation of FIP obligations in contravention of the holding in Wisconsin v. EPA, 938 F.3d at 313–20.

Commenters' attempt to contrast the implementation of source-specific emissions limitations at industrial sources with the establishment of a specific mass-based budget (as the EPA has set for power plants in prior good neighbor FIPs) is unavailing. CAA section 110(c)(1) and 302(y) authorize the EPA in promulgating a FIP to establish “enforceable emission limitations” in addition to other types of control measures like mass-based trading programs. Further, in this action, the EPA has developed an emissions control strategy that prohibits the “amount” of pollution that significantly contributes to nonattainment and/or interferes with maintenance. We determine that amount, as we have in prior transport actions, at Step 3 of the analysis, by applying a multifactor analysis that includes considering cost and downwind air quality effects. See section V.A of this document. With the implementation of the selected controls (at Step 4) through both an emissions trading program for power plants and source-specific emissions limitations for industrial sources, those “amounts” that had been emitted prior to imposition of the controls will be eliminated.

The Act does not mandate that the EPA must set a specific mass-based budget for each state to eliminate significant contribution based on the use of the term “amounts” in CAA section 110(a)(2)(D)(i). As the Supreme Court recognized, the statute “ requires States to eliminate those `amounts' of pollution that `contribute significantly to nonattainment' in downwind States,” and it delegates to states or EPA acting in their stead discretion to determine how to apportion responsibility among those upwind states. 572 U.S. at 514 (emphasis added). The statute does not define the term “amount” in the way commenters suggest (or in any other way), and neither the Agency nor any court has reached that conclusion. The Supreme Court itself has recognized that the language of the good neighbor provision is amenable to different types of metrics for quantification of “significant contribution.” See EME Homer City Generation, L.P., 572 U.S. at 514 (“How is EPA to divide responsibility among the . . . States? Should the Agency allocate reductions proportionally . . ., on a per capita basis, on the basis of the cost of abatement, or by some other metric? . . . The Good Neighbor Provision does not answer that question for EPA.”); see also Michigan v. EPA, 213 F.3d 663, 677 D.C. Cir. 2000) (“Nothing in the text of . . . the statute spells out a criterion for classifying `emissions activity' as `significant.' ”); id. at 677 (“Must EPA simply pick some flat `amount' of contribution . . . ?”). When the State of Delaware petitioned the Agency under CAA section 126(b) to establish daily emissions rates for EGUs to remedy what it saw as continuing violations of the good neighbor provision for the 2008 ozone NAAQS, neither the EPA nor the reviewing court questioned whether the Agency had the statutory authority to do so. The EPA's decision not to was upheld on record grounds. See Maryland v. EPA, 958 F.3d 1185, 1207 D.C. Cir. 2020) (“In other words, Delaware's concern makes sense but has not been observed in practice.”).

The term “amounts” can be interpreted to refer to any number of metrics, and in fact the CAA uses the term in several contexts where it is clear Congress did not intend the term to refer to a fixed, mass-based quantity of emissions. For example, in the definition of “lowest achievable emission rate” (LAER) in CAA section 171, the Act provides that the application of LAER shall not permit a proposed new or modified source to emit any pollutant in excess of “the amount allowable under applicable new source standards of performance [NSPS].” NSPS may be, and usually are, set as emissions standards or limitations that are rate- or concentration-based. See, e.g.,40 CFR part 60, subpart KKKK, table I (establishing concentration-based and rate-based emissions limits for stationary combustion turbines). Congress has elsewhere used the term “amount” in the CAA to refer to concentration-based standards. For example, in CAA section 163(b), Congress provided that maximum allowable increases in concentrations of certain pollutants “shall not exceed the following amounts,” with a list of allowable increases provided that are expressed in micrograms per cubic meter. As a third example, in the 1990 CAA Amendments, Congress provided that ozone nonattainment areas classified as Serious must provide a reasonable further progress demonstration of reductions in VOC emissions “equal to the following amount,” which is then described as a percentage reduction from baseline emissions. CAA section 182(c)(2)(B). These examples illustrate that the word “amounts” is amenable to a variety of meanings depending on what is being measured or quantified. It would therefore be highly unlikely that Congress could have intended that “amount” as used in the good neighbor provision must signify only a fixed mass budget of emissions for each state expressed as total tons per ozone season.

Such an approach would, in fact, fail to address an important aspect of the problem of interstate transport. As explained in sections III.B.1.d, V.D.4, and VI.B.1, the EPA in this rule seeks to better address the need for emissions reductions on each day of the ozone season, reflecting the daily, but unpredictably recurring, nature of the air pollution problem, short-term health impacts, and the form of the 2015 ozone NAAQS, wherein nonattainment for downwind areas (and thus heightened regulatory requirements) could be based on ozone exceedances on just a few days of the year. The expression of the “amount” of pollution that should be eliminated to address upwind states' “significant contribution” to that type of air pollution problem may appropriately take into account those aspects of the problem, and the EPA may appropriately conclude, as we do here, that a single, fixed, emissions budget covering an entire ozone season is not sufficient to the task at hand.

In this action, the EPA reasonably applies the good neighbor provision, including the term “amount,” through the 4-step interstate transport framework. Under this approach, the EPA here, as it has in prior transport rulemakings for regional pollutants like ozone, identifies a uniform level of emissions reduction that the covered sources in the linked upwind states can achieve that cost-effectively delivers improvement in air quality at downwind receptors on a regional scale. The “amount” of pollution that is identified for elimination at Step 3 of the framework is therefore that amount of emissions that is in excess of the emissions control strategies the EPA has deemed cost-effective. Contrary to commenters' views, in prior transport rules utilizing emissions trading, the mass budgets through which the elimination of significant contribution was effectuated did not constitute the “amounts” to be eliminated but rather the residual emissions remaining following the elimination of significant contribution through the control stringency selected based on our multifactor assessment at Step 3. Nor did the EPA consider a mass-based budget to be the sole expression, even indirectly, of what constituted “significant contribution.” See, e.g., CSAPR, 76 FR 48256–57 (discussing the evaluation of the control strategies that would eliminate significant contribution for the 1997 ozone NAAQS, including combustion controls, and explaining, “[I]t would be inappropriate for a state linked to downwind nonattainment or maintenance areas to stop operating existing pollution control equipment (which would increase their emissions and contribution).”).

In other actions the EPA has taken to implement good neighbor obligations, the EPA has required or allowed for reliance on source-specific emissions limitations rather than defining significant contribution as a mass-based budget. For example, the EPA imposed unit-specific emissions limitations in granting a CAA section 126(b) petition from the State of New Jersey in 2011. Final Response to Petition From New Jersey Regarding SO₂ Emissions From the Portland Generating Station, 76 FR 69052, 69063–64 (Nov. 7, 2011) (discussing the analytical basis for the establishment of emissions limits at specific units). This action was upheld by the Third Circuit in Genon Rema LLC v. EPA, 722 F.3d 513, 526 (3d. Cir. 2013).

Even where the EPA has provided for implementation of good neighbor requirements through mass-based budgets, it has recognized that other approaches may be acceptable as providing an equivalent degree of emissions reduction to eliminate significant contribution. See, e.g., NO_X SIP Call, 63 FR 57378–79 (discussing approvability of rate-based emissions limit approaches for implementing NO_X SIP Call and providing, “the 2007 overall budget is an important accounting tool. However, the State is not required to demonstrate that it has limited its total NO_X emissions to the budget amounts. Thus, the overall budget amount is not an independently enforceable requirement.”); CAIR, 70 FR 25261–62 (discussing ways states could implement CAIR obligations, including through emission-rate limitations, so long as adequately demonstrated to achieve comparable reductions to CAIR's emissions budgets).

Finally, as it has in its prior transport FIP actions, the EPA has in this action provided guidance for states on methods by which they could replace this FIP with SIPs, and in so doing, continues to recognize substantial state flexibility in achieving an equivalent degree of emissions reduction that would successfully eliminate significant contribution for the 2015 ozone NAAQS. See section VI.D of this document. While the EPA has exercised the responsibility it has under CAA section 110(c)(1) to step into the shoes of the covered states and directly implement good neighbor requirements through a particular set of regulatory mechanisms in this action, we anticipate that states may identify alternative, equivalent mechanisms that we would be bound to evaluate and approve if satisfactory, should states seek to replace this FIP with a SIP.

For these reasons, the EPA disagrees with the contention that it is constrained by the good neighbor provision to define upwind state obligations solely by reference to a fixed, mass budget. We find it reasonable in this action to again determine the amount of “significant contribution” at Step 3 by reference to uniform levels of cost-effective emissions controls that can be applied across the upwind sources. And, we find it appropriate to implement those emissions reductions at Step 4 through mechanisms that go beyond fixed, mass-based, ozone-season long budgets.

The EPA's authority for its industrial source control strategies is further discussed in sections II.C. and III.B.1.c of this document. The relationship of the control strategy to the assessment of overcontrol is discussed in section V.D.4 of this document. The relationship of our FIP authority to state authorities and SIP calls under CAA section 110(k)(5) is further discussed in RTC sections 1 and 2.

a. Step 1 Approach

As proposed, the EPA applies the same basic method of the CSAPR Update and the Revised CSAPR Update for identifying nonattainment and maintenance receptors. However, we received comments arguing that the outcome of applying our methodology to identify receptors in 2023 appears overly optimistic in light of current measured data from the network of ambient air quality monitors across the country. These commenters suggest that the EPA give greater weight to current measured data as part of the method for identifying projected receptors. As discussed further in section IV.D of this document, the EPA has modified its approach for identifying receptors for this final rule in response to these comments.

This concern is more evident given that the 2023 ozone season is just a few months away, and the most recent measured ozone values in many areas strongly suggest that these areas will not likely see the substantial reduction in ozone levels that the 2016v2 and 2016v3 modeling continue to project.

It would not be reasonable to ignore recent measured ozone levels in many areas that are clearly not fully consistent with certain concentrations in the Step 1 analysis for 2023. Therefore, the EPA has developed an additional maintenance-only receptor category, which includes what we refer to as “violating monitor” receptors, based on current ozone concentrations measured by regulatory ambient air quality monitoring sites. We acknowledge that the traditional modeling plus monitoring methodology we used at proposal and in prior ozone transport rules would otherwise have identified such sites as being in attainment in 2023. Despite the implications of the current measured data suggesting there will be a nonattainment problem at these sites in 2023, we cannot definitively establish that such sites will be in nonattainment in 2023 in light of our modeling projections. In the face of this uncertainty, we regard our ability to consider such sites as receptors for purposes of good neighbor analysis under CAA section 110(a)(2)(D)(i)(I) to be a function of the requirement to prohibit emissions that interfere with maintenance of the NAAQS; even if our transport modeling projects that an area may reach attainment in 2023, we have other information indicating that there is an identified risk that attainment will not in fact be achieved in 2023. The EPA's analysis of these additional receptors further is explained in section IV.D of this document.

However, because we did not identify this basis for receptor-identification at proposal, in this final action we are only using this receptor category on a confirmatory basis. That is, for states that we find linked based on our traditional modeling-based methodology in 2023, we find in this final analysis that the linkage at Step 2 is strengthened and confirmed if that state is also linked to one or more “violating monitor” receptors. If a state is only linked to a violating-monitor receptor in this final analysis, we are deferring promulgating a final FIP (and we have also deferred taking final action on that state's SIP submittal). This is the case for the State of Tennessee. Among the states that previously had their transport SIPs fully approved for the 2015 ozone NAAQS, the EPA has also identified a linkage to violating-monitor receptors for the State of Kansas. The EPA intends to further review its air quality modeling results and recent measured ozone levels, and we intend to address these states' good neighbor obligations as expeditiously as practicable in a future action.

b. Step 2 Approach

The EPA applies the same approach for identifying which states are contributing to downwind nonattainment and maintenance receptors as it has applied in the three prior CSAPR rulemakings. CSAPR, the CSAPR Update, and the Revised CSAPR Update used a screening threshold of 1 percent of the NAAQS to identify upwind states that were “linked” to downwind air pollution problems. States with contributions greater than or equal to the threshold for at least one downwind nonattainment or maintenance receptor identified in Step 1 were identified in these rules as needing further evaluation of their good neighbor obligations to downwind states at Step 3. The EPA evaluated each state's contribution based on the average relative downwind impact calculated over multiple days. States whose air quality impacts to all downwind receptors were below this threshold did not require further evaluation for measures to address transport. In other words, the EPA determined that these states did not contribute to downwind air quality problems and therefore had no emissions reduction obligations under the good neighbor provision. The EPA applies a relatively low contribution screening threshold because many downwind ozone nonattainment and maintenance receptors receive transport contributions from multiple upwind states. While the proportion of contribution from a single upwind state may be relatively small, the effect of collective contribution resulting from multiple upwind states may substantially contribute to nonattainment of or interference with maintenance of the NAAQS in downwind areas. The preambles to the proposed and final CSAPR rules discuss the use of the 1 percent threshold for CSAPR. See75 FR 45237 (August 2, 2010); 76 FR 48238 (August 8, 2011). The same metric is discussed in the CSAPR Update, see81 FR 74538, and in the Revised CSAPR Update, see86 FR 23054. In this final rule, the EPA has updated the air quality modeling data used for determining contributions at Step 2 of the 4-step interstate transport framework using the 2016v3 modeling platform. The EPA continues to find that this threshold is appropriate to apply for the 2015 ozone NAAQS. This rule's application of the Step 2 approach is comprehensively described in section IV of this document.

Many commenters challenged the use of a 1 percent of NAAQS threshold or otherwise raised issues with the EPA's Step 2 methodology. These comments are addressed in section IV.F of this document and in the RTC document.

c. Step 3 Approach

The EPA continues to apply the same approach as the prior three CSAPR rulemakings for evaluating “significant contribution” at Step 3. For states that are linked at Step 2 to downwind air quality problems, CSAPR, the CSAPR Update, and the Revised CSAPR Update evaluated NO_X reduction potential, cost, and downwind air quality improvements available at various mitigation technology breakpoints (represented by cost thresholds) in the multi-factor test. In CSAPR, the CSAPR Update, and the Revised CSAPR Update, the EPA selected the technology breakpoint (represented by a cost threshold) that, in general, maximized cost-effectiveness— i.e., that achieved a reasonable balance of incremental NO_X reduction potential and corresponding downwind ozone air quality improvements, relative to the other emissions budget levels evaluated. See, e.g.,81 FR 74550. The EPA determined the level of emissions reductions associated with that level of control stringency to constitute significant contribution to nonattainment or interfere with maintenance of a NAAQS downwind. See, e.g.,86 FR 23116. This approach was upheld by the U.S. Supreme Court in EPA v. EME Homer City.

In this action, the EPA applies this approach to identify EGU and non-EGU NO_X control stringencies necessary to address significant contribution for the 2015 ozone NAAQS. The EPA applies a multifactor assessment using cost-thresholds, total emissions reduction potential, and downwind air quality effects as key factors in determining a reasonable balance of NO_X controls in light of the downwind air quality problems. The EPA's evaluation of available NO_X mitigation strategies for EGUs focuses on the same core set of measures as prior transport rules, and the EPA finalizes a control stringency for EGUs from these measures that is commensurate with the nature of the ongoing ozone nonattainment and maintenance problems observed for the 2015 ozone NAAQS. Similarly, in this action, the EPA includes other industrial sources (non-EGUs) in its Step 3 analysis and finalizes emissions limitations for certain non-EGU sources as needed to eliminate significant contribution and interference with maintenance. The available reductions and cost-levels for the non-EGU stringency is commensurate with the control strategy for EGUs.

In CSAPR, the CSAPR Update, and the Revised CSAPR Update, the EPA focused its Step 3 analysis on EGUs. In the Revised CSAPR Update, in response to the Wisconsin decision's finding that the EPA had not adequately evaluated potential non-EGU reductions, see 938 F.3d at 318, the EPA determined that the available NO_X emissions reductions from non-EGU sources, for purposes of addressing good neighbor obligations for the 2008 ozone NAAQS, at a comparable cost threshold to the required EGU emissions reductions (for which the EPA used an adjusted representative cost of $1,800 per ton), and based on the timing of when such measures could be implemented, did not provide a sufficiently meaningful and timely air quality improvement at the downwind receptors before those receptors were projected to resolve. See86 FR 23110. On that basis, the EPA made a finding that emissions reductions from non-EGU sources were not required to eliminate significant contribution to downwind air quality problems under the interstate transport provision for the 2008 ozone NAAQS. In this rule, the EPA's “significant contribution” analysis at Step 3 of the 4-step framework includes a comprehensive evaluation of major stationary source non-EGU industries in the linked upwind states. The EPA finds that emissions from certain non-EGU sources in the upwind states significantly contribute to downwind air quality problems for the 2015 ozone NAAQS, and that cost-effective emissions reductions from these sources are required to eliminate significant contribution under the interstate transport provision. Therefore, this rule requires emissions reductions from non-EGU sources in upwind states to fulfill interstate transport obligations for the 2015 ozone NAAQS. This analysis is described fully in section V of this document.

In this rule, the EPA also continues to apply its approach for assessing and avoiding “over-control.” In EME Homer City, the Supreme Court held that “EPA cannot require a State to reduce its output of pollution by more than is necessary to achieve attainment in every downwind State or at odds with the one-percent threshold the Agency has set.” 572 U.S. at 521. The Court acknowledged that “instances of `over-control' in particular downwind locations may be incidental to reductions necessary to ensure attainment elsewhere.” Id. at 492.

Because individual upwind States often `contribute significantly' to nonattainment in multiple downwind locations, the emissions reductions required to bring one linked downwind State into attainment may well be large enough to push other linked downwind States over the attainment line. As the Good Neighbor Provision seeks attainment in every downwind State, however, exceeding attainment in one State cannot rank as `over-control' unless unnecessary to achieving attainment in any downwind State. Only reductions unnecessary to downwind attainment anywhere fall outside the Agency's statutory authority.

Id. at 522 (footnotes omitted).

The Court further explained that “while EPA has a statutory duty to avoid over-control, the Agency also has a statutory obligation to avoid `under-control,' i.e., to maximize achievement of attainment downwind.” Id. at 523. Therefore, in the CSAPR Update and Revised CSAPR Update, the EPA evaluated possible over-control by considering whether an upwind state is linked solely to downwind air quality problems that can be resolved at a lower cost threshold, or if upwind states would reduce their emissions at a lower cost threshold to the extent that they would no longer meet or exceed the 1 percent air quality contribution threshold. See, e.g.,81 FR 74551–52. See also Wisconsin, 938 F.3d at 325 (over-control must be proven through a “ `particularized, as-applied challenge' ”) (quoting EME Homer City Generation, 572 U.S. at 523–24). The EPA continues to apply this framework for assessing over-control in this rule, and, as discussed in section V.D.4 of this document, does not find any over-control at the final control stringency selected.

This evaluation of cost, NO_X reductions, and air quality improvements, including consideration of whether there is proven over-control, results in the EPA's determination of the appropriate level of upwind control stringency that would result in elimination of emissions that significantly contribute to nonattainment or interfere with maintenance of the NAAQS in downwind areas.

Comment: Commenters alleged that the EPA lacks authority to regulate EGUs under the good neighbor provision of the CAA, or at least in the manner proposed, because in their view, this regulation would intrude into areas of regulation that are reserved to other Federal agencies or are beyond the EPA's expertise. They focused in particular on the EGU trading program enhancements, which they alleged would threaten electric grid reliability, and asserted that EPA lacks authority or expertise to dictate the mix of electricity generation in the country.

Response: The EPA disagrees that the regulation of EGUs in this action is unlawful or unsupported. The Agency has consistently and successfully regulated EGUs' ozone season NO_X emissions under the good neighbor provision for over 25 years, beginning with the 1997 NO_X SIP Call. This action does not intrude on other Federal agencies' authorities and responsibilities with respect to managing the electric power grid and ensuring reliable electricity. While other agencies such as the Federal Energy Regulatory Commission (FERC) have primary responsibility for ensuring reliability of the bulk electric system, the EPA has ensured that its final rule here will not create electric reliability concerns. See section VI.B.1.d of this document. Thus, to the extent commenters are raising a record-based issue that the EPA through this action has created a reliability concern, we disagree. The EPA engaged in a series of stakeholder meetings with Reliability Coordinators who commented on the proposed rule, including several Regional Transmission Organizations (RTOs) as well as non-RTO entities throughout the rulemaking process.

To the extent commenters maintain that—despite this record of collaboration and sensitivity to the need to ensure reliability in the implementation of its mandates, including in this rule—the EPA nonetheless fundamentally lacks authority to regulate the electric-power sector in any way that “impact[s] national electricity and energy markets,” the EPA disagrees. The EPA has successfully regulated interstate ozone-precursor emissions from the power sector since the NO_X SIP Call and the establishment of the NO_X Budget Trading Program. See generally Michigan v. EPA, 213 F.3d 663 (D.C. Cir. 2000); Appalachian Power Co. v. EPA, 249 F.3d 1032 (D.C. Cir. 2001). In fact, each of the EPA's interstate ozone transport rulemakings has focused on the regulation of ozone-precursor emissions from the power sector (all but the NO_X SIP Call exclusively), because substantial, cost-effective reductions in ozone-precursor emissions have been and continue to be available from fossil-fuel fired EGUs. See, e.g.,63 FR 57399–400 (NO_X SIP Call); 70 FR 25165 and 71 FR 25343 (CAIR and CAIR FIP); 76 FR 48210–11 (CSAPR); 81 FR 74507 (CSAPR Update); 86 FR 23061 (Revised CSAPR Update).

This rule, like all prior EPA ozone-transport rulemakings, regulates only one aspect of the operation of fossil-fuel fired EGUs, that is, the emissions of NO_X as an ozone-precursor pollutant during the ozone season. This rule limits EGU NO_X emissions that interfere with downwind states' ability to attain and maintain the 2015 ozone NAAQS. The rule does not regulate any other aspect of energy generation, distribution, or sale. For these reasons, the rule does not intrude on FERC's power under the Federal Power Act, 16 U.S.C. 791a, et seq. And, as in prior transport rules, the EPA implements this regulation through a proven, flexible mass-based emissions trading program that integrates well with, and in no way intrudes upon, the management of the power sector under other state and Federal authorities. This rule will not alter the procedures system operators employ to dispatch resources or force changes to FERC-jurisdictional electricity markets, nor have commenters offered any explanation in this regard themselves.

The actual compliance requirement that the EGUs must meet in the allowance trading system finalized here—just as in all prior interstate transport trading programs—is simply to hold sufficient allowances to cover emissions during a given control period, not to undertake any specific compliance strategy. The owner or operator of an EGU has flexibility in determining how it will meet this requirement, whether through the add-on emissions controls that the EPA has selected in our Step 3 analysis, or through some other method or methods of compliance. The costs of meeting this allowance-holding requirement—just like the cost associated with meeting any other regulatory requirements—could possibly then be factored into what that unit bids in the wholesale electricity market (or in regulated jurisdictions, would factor into utility regulators' determinations of what can be cost-recovered).

Those costs could, in turn, result in a reduction in electricity generation from higher-emitting sources and an increase in electricity generation from lower-emitting or zero-emitting generators, but that kind of generation shifting (not mandated but occurring as an economic choice by the regulated sources) is consistent, and in no way interferes with, the existing security-constrained economic dispatch protocols of the modern electrical grid. Further, this type of “impact” on electricity markets—merely incidental, not mandated or even intended—is of the same type that results from any other kind of regulation, environmental or otherwise. Indeed, the U.S. Supreme Court recognizes that regulatory actions that may have some “effect,” or impact, in electricity markets do not on that basis alone intrude into authorities reserved to electricity rate-setting regulators by the Federal Power Act. See FERC v. Electric Power Supply Ass'n, 577 U.S. 260, 282–84 (2016) (distinguishing between actions that have an effect on retail rates and actual intrusion into retail rate-setting itself); see also Hughes v. Talen, 578 U.S. 150, 166 (2016). The Supreme Court again recognized this distinction between “incidental” effects caused by lawfully issued environmental regulations and attempts to mandate a particular energy mix in West Virginia v. EPA. See 142 S. Ct. 2587, 2613 n.4 (2022) (“[T]here is an obvious difference between (1) issuing a rule that may end up causing an incidental loss of coal's market share, and (2) simply announcing what the market share of coal, natural gas, wind, and solar must be . . . .”).

This rule is squarely in the former camp; as the most stringent component of its emissions controls strategy for EGUs, the EPA has determined that to eliminate significant contribution to harmful levels of ozone in other states, certain fossil-fuel fired EGUs in “linked” upwind states that do not already have selective catalytic reduction (SCR) post-combustion control technology, should install it (or achieve emissions reductions commensurate with that technology). SCR is a well-established at-the-source NO_X control technology already in use by EGUs representing roughly 60 percent of the existing coal-fired generating capacity in the United States. This technology can be installed and operated to reduce NO_X emissions without forcing the retirement or reduced utilization of any EGU. However, if market conditions are such that an EGU faced with this mandate (again, as expressed through an emissions trading budget) finds it more economic to comply with the mandate through the purchase of allowances, installation of other types of pollution control, reduced utilization, and/or retirement, rather than installing SCR technology, that is a choice that the EGU owner/operator can freely make under this rule. Security constrained economic dispatch is thereby maintained and is in no way interfered with.

The EPA recognizes that cost to operate generators is one of the major factors that system operators utilize to determine “merit” order in dispatching resources. However, this rule does not intrude in any way into that process. To the extent that compliance with environmental regulations is a kind of cost that may need to be factored into generators' bids, this rule is no different than many other such requirements EGUs are already subject to. Further, as in prior transport rules, this rule applies a uniform control stringency to EGUs within the covered upwind states. EGUs that may have enjoyed a competitive advantage in the past through not bearing the costs of installing and running state-of-the-art emissions control technology now must bear that cost just as their competitors with that technology already are. Cf. EME Homer City, 572 U.S. 489, 519 (CSAPR is “[e]quitable because, by imposing uniform cost thresholds on regulated States, EPA's rule subjects to stricter regulation those States that have done relatively less in the past to control their pollution. Upwind States that have not yet implemented pollution controls of the same stringency as their neighbors will be stopped from free riding on their neighbors' efforts to reduce pollution. They will have to bring down their emissions by installing devices of the kind in which neighboring States have already invested.”).

Finally, we note that this final rule does not include “generation shifting” as a component of the budget-setting process, even in the limited way that it had been used in prior transport rules like CSAPR and the CSAPR Update, i.e., to ensure the budget provided adequate incentive to ensure implementation of the selected emission-control strategy. See section V.B.1.f of this document. Further comments regarding legal authority for “generation shifting,” relationship to state authorities, and expertise associated with grid reliability are addressed in section 1.3 of the RTC. We further discuss our consideration of grid reliability concerns and adjustments in the approach to the EGU emissions trading program from proposal in section VI.B.1.d of this document.

Comment: Commenters generally challenged the EPA's authority to establish emissions control requirements for non-EGU industrial sources in this action, or argued that such controls are unnecessary or unsupported, or run contrary to the EPA's prior actions under the good neighbor provision.

Response: The states and the EPA have authority under CAA section 110(a)(2)(D)(i)(I) to prohibit emissions from “any source or other type of emissions activity” that are found to significantly contribute to nonattainment or interfere with maintenance of the NAAQS in downwind states. This language is not limited only to power plant emissions, nor is it limited only to “major” sources or “stationary” sources. Thus, as a legal matter, the emissions control requirements for certain large “non-EGU” industrial sources in this action are grounded in unambiguous statutory authority, in particular the statute's use of the broad term “any source.” Whereas the Act elsewhere includes definitions of “major stationary source,” “small source,” and “stationary source,” see, e.g., CAA section 302(j), (x), and (z), no such qualifying terms are used with respect to the term “any source” at CAA section 110(a)(2)(D)(i). Rather, the scope of authority in this provision expands to encompass “other type of emissions activity” in addition to “any source.” The EPA has previously included non-EGU industrial sources in findings quantifying states' obligations under the good neighbor provision, in the 1998 NO_X SIP Call, see63 FR 57365. ⁹² See also Michigan v. EPA, 213 F.3d 663, 690–93 (upholding the inclusion of certain non-EGU boilers in the NO_X SIP Call). The EPA's determinations in prior transport rules not to regulate sources beyond the power sector were grounded in considerations not related to the Agency's statutory authority. For example, in the original CSAPR rulemaking, the EPA determined that the analytical effort needed to regulate non-EGU industrial sources would substantially delay the implementation of emissions reductions from the power sector. See, e.g.,76 FR 48247–48 (“[D]eveloping the additional information needed to consider NO_X emissions from non-EGU source categories to fully quantify upwind state responsibility with respect to the 1997 ozone NAAQS would substantially delay promulgation of the Transport Rule. . . . [W]e do not believe that effort should delay the emissions reductions and large health benefits this final rule will deliver[.]”). The EPA acknowledged that by not addressing non-EGUs, it may not have promulgated a complete remedy to good neighbor obligations in CSAPR, id. at 48248. Nonetheless, the EPA went on to explain that there were limited emissions reductions available from non-EGUs at the cost thresholds the EPA determined would deliver substantial reductions from power plants. See id. at 48249 (the EPA's “preliminary assessment in the rule proposal suggested that there likely would be very large emissions reductions available from EGUs before costs reach the point for which non-EGU sources have available reductions . . . . EPA revisited these non-EGU reduction cost levels in this final rulemaking and verified that there are little or no reductions available from non-EGUs at costs lower than the thresholds that EPA has chosen . . . .”). The EPA noted in CSAPR that states retained the authority to regulate non-EGUs as a method of addressing their good neighbor obligations. Id. at 48320. The EPA also noted in CSAPR that “potentially substantial” non-EGU emissions reductions could be available in future rulemakings applying a higher cost threshold. See id. at 48256.

Similarly, in the CSAPR Update, which addressed good neighbor obligations for the 2008 ozone NAAQS, the EPA found that regulation of non-EGUs was not warranted as the analysis required could delay the expeditious implementation of power plant reductions. The EPA found that the availability and cost-effectiveness of non-EGU reductions was uncertain and further analysis could delay implementation of the EGU strategy beyond 2017. The EPA acknowledged that it was not promulgating a complete remedy for good neighbor obligations for the 2008 ozone NAAQS and indicated its intention to further review emissions-reduction opportunities from non-EGU and EGU sources. 81 FR 74521–22.

In Wisconsin, the court held that the EPA's deferral of a complete good neighbor remedy by 2017, on the basis, among other things, of uncertainty regarding non-EGU emissions reductions and the need for further regulatory analysis, was unlawful. 938 F.3d at 318–19. The court noted that “ `the statutes and common sense demand regulatory action to prevent harm, even if the regulator is less than certain.' ” Id. at 319 (quoting Ethyl Corp. v. EPA, 541 F.2d 1, 24–25 (D.C. Cir. 1976)), and that agencies can only avoid meeting their statutory obligations where “scientific uncertainty is so profound that it precludes EPA from making a reasoned judgment.” Id. (citing Massachusetts v. EPA, 549 U.S. 497, 534 (2007)). Further, the court rejected the EPA's argument that it would have delayed its rulemaking if the EPA needed to complete a non-EGU analysis in a timely manner, holding that “administrative infeasibility” is not sufficient to “justify . . . noncompliance with the statute.” Id. Rather, the Agency would need to “meet the `heavy burden to demonstrate the existence of an impossibility.' ” Id. (quoting Sierra Club v. EPA, 719 F.2d 436, 462 (D.C. Cir. 1983)).

Following the remand of the CSAPR Update in Wisconsin, in the Revised CSAPR Update, the EPA conducted an analysis of non-EGUs to ensure it had implemented a complete remedy to eliminate significant contribution for the covered states for the 2008 ozone NAAQS. While acknowledging uncertainty in the datasets for non-EGUs, the EPA concluded: “[U]sing the best information currently available to the Agency, . . . the EPA is concluding that there are relatively fewer emissions reductions available at a cost threshold comparable to the cost threshold selected for EGUs. In the EPA's reasoned judgment, the Agency concludes such reductions are estimated to have a much smaller effect on any downwind receptor in the year by which the EPA finds such controls could be installed.” 86 FR 23059. Therefore, the EPA determined control of non-EGU emissions was not required to eliminate significant contribution for the 2008 ozone NAAQS.

The circumstances that led the EPA to defer or decline regulation of non-EGU sources in CSAPR, the CSAPR Update, and the Revised CSAPR Update, are not present here, and the EPA's determination in this action that prohibiting certain emissions from certain non-EGU sources is necessary to eliminate significant contribution for the 2015 ozone NAAQS is a logical extension of the analyses and evolution of regulatory policy development spanning its prior good neighbor rules, now applied to implement this more protective NAAQS. As the EPA explained at proposal, unlike in CSAPR and the Revised CSAPR Update, in this action the EPA finds that available reductions and cost-levels for the non-EGU stringency are commensurate with the control strategy for EGUs. Following consideration of comments and after some adjustments in the non-EGU analysis and control strategy, in this final rule, the EPA continues to find this to be the case. See sections V.C and V.D of this document.

In particular, the EPA continues to find that cost-effective emissions reductions are available for non-EGUs at a representative cost-threshold that is lower than the cost-threshold the EPA is applying for EGUs. See section V.C. of this document. These emissions control strategies are generally comparable to the emissions reduction requirements that similar sources in downwind states are already required to meet. See section V.B.2 of this document. The EPA finds that the implementation of these emissions control strategies at non-EGUs, in conjunction with the strategies for EGU, will make a cost-effective and meaningful improvement in air quality through reducing ozone levels at the identified downwind receptors, and, therefore, the EPA has determined that these strategies will eliminate the amount of upwind emissions needed to address significant contribution under the good neighbor provision. The EPA's action here is focused on the most impactful industries and emissions units as determined by our evaluation of the power sector and the non-EGU screening assessment prepared for the proposal; indeed, of the 41 industries, as identified by North American Industry Classification System codes, we analyzed, only nine industries met the criteria for further evaluation of significant contribution. See section V.B.2 of this document. Further, the EPA finds that these strategies do not result in “overcontrol.” See section V.D.4 of this document. As such, the EPA maintains that its final determinations regarding non-EGUs and its inclusion of non-EGU emissions sources within this final rule are statutorily authorized and lawful.

The EPA disagrees that it should defer regulation of industrial sources to the NSPS program under CAA section 111(b). CAA section 111(b) does not expressly provide for the elimination of “significant contribution” as is required under CAA section 110(a)(2)(D)(i)(I). In particular, commenter's statement that NSPS rulemakings under section 111(b) will appropriately address the emissions that we find must be eliminated in this action is not correct. Standards under section 111(b) apply only to new and modified sources, not existing sources. This action, however, finds that reductions in ongoing emissions from existing sources are needed to eliminate significant contribution. An NSPS standard for new and modified sources would not address such emissions from existing sources. To the extent that covered sources in this action also may be covered by an older NSPS, these sources nonetheless continue to have emissions that the EPA finds significantly contribute and can be eliminated through further emissions control as determined in this action. We further disagree with commenter's separate suggestion that the EPA use section 111(b) and (d) to regulate both new and existing sources of ozone season NO_X, which is premised on the incorrect notion that the EPA's action here is an attempt to regulate entire source categories nationwide, rather than to eliminate significant contribution pursuant to CAA section 110(a)(2)(D)(i)(I). This action applies only to the extent a state is “linked” to downwind receptors, and therefore this action only regulates covered non-EGU industrial sources in 20 states. Further, this comment ignores that the regulation of criteria pollutant emissions from existing sources under CAA section 111(d) is limited by the criteria pollutant exclusion in CAA section 111(d)(1)(A)(i).

The EPA agrees with the commenters who assert that the EPA's authority to regulate non-EGUs under the good neighbor provision is well-grounded in administrative precedent and case law. Our previous discussion briefly recites several of the most salient aspects of that history. We also agree that the statutory language is not limited only to those sources that emit above 100 tons per year. The EPA's Step 3 and Step 4 analyses in this regard, which establish certain thresholds based on historical actual emissions, potential to emit and/or metrics for unit design capacity, reflect a reasoned judgment by the Agency regarding which emissions can be cost-effectively eliminated to address significant contribution, under the facts and circumstances of this action. That these thresholds are designed to exclude certain smaller or lower-emitting units does not reflect a determination that the EPA lacks legal authority to regulate such sources under different facts and circumstances.

The EPA identified two industry tiers of potential non-EGU emissions reductions in its non-EGU screening assessment at proposal, based on screening metrics intended to capture different kinds of impacts that non-EGU sources may have on identified receptors. The EPA agrees that it is only authorized to prohibit emissions under the good neighbor provision that significantly contribute to nonattainment or interfere with maintenance in downwind states, and we determined that these industries did so. The EPA sought comment on whether additional non-EGU industries significantly contributed to nonattainment or interfered with maintenance in downwind states. The EPA did not receive comments identifying other industrial stationary sources that are more impactful that should be regulated instead of those the EPA identified. We believed at proposal and confirm here in our final rule that the methodology used in the screening assessment comported with the factors that we consider at Step 3. Further, the EPA's 4-step interstate transport framework, including the Step 3 analysis and an overcontrol assessment, ensure that the emissions reductions achieved at each source covered by this rule are in fact justified as part of an overall, complete remedy to eliminate significant contribution for the covered states for the 2015 ozone NAAQS. The EPA has decided to finalize emissions limitations for all of the non-EGU industries, with some modifications from proposal reflecting public input, as discussed in section VI.C of this document. The Agency's authority to establish unit- and/or source-specific emissions limitations in exercising our FIP authority is further discussed in section III.B.1 of this document.

Comment: Commenters raise additional issues with the overall approach of the rule at Step 3 to address significant contribution through our evaluation of EGU and non-EGU strategies through parallel but separate analyses. They stated that the EPA failed to establish that the identified non-EGU emissions reductions are needed to eliminate significant contribution. Commenters stated that the identified non-EGU emissions reductions are not impactful of air quality at receptors or that they are much less cost-effective than the EGU emissions reductions. Commenters stated that the EPA grouped all non-EGU emissions reductions together in making a cost-effectiveness determination that is only an average and ignores significant variation in costs associated with controls on different types of non-EGU emissions units. They also stated the EPA did not assess multiple control technologies in the way that it did for EGUs, and they argued there is great variation in the profile of non-EGU industries and emissions unit types in the different upwind states or that individual emissions units do not contribute to an out-of-state air quality problem at all. Commenters argued that certain non-EGU controls were not feasible, or that the EPA had applied a different standard for “feasibility” for non-EGUs than it did for EGUs. Commenters stated that the EPA should have provided a mass-based trading option for non-EGUs just as it had for EGUs. By contrast, other commenters supported the regulation of non-EGUs in this action as necessary to ensure a complete remedy to good neighbor obligations, since the statute is not limited to regulating power plants. Some commenters further stated that EGUs should not face any further emissions reduction obligation because all cost-effective controls have already been identified through prior transport rules, and that any further regulation of EGUs would only lead to the retirement of coal plants, which they believe is the EPA's true objective. Finally, some commenters argued that the EPA had not ensured that it only regulated up to the minimum needed for downwind areas to come into attainment.

Response: Issues related to the specific technical bases for the Agency's determinations of what emissions constitute “significant contribution” at Step 3 of the 4-step framework are addressed in section V of this document. Here, we evaluate commenters' more general assertions that this action addresses non-EGU or EGU emissions in an inconsistent way. First, the EPA agrees with commenters that the task of evaluating significant contribution from the non-EGU industries is complex compared to EGUs in light of the much greater diversity in industries and emissions unit types. This, however, is not a valid basis to avoid emissions control requirements on such sources if needed to eliminate significant contribution. In this respect, the EPA's analysis in this final rule is that the 4-step framework, as upheld by the Supreme Court in EME Homer City, can be adequately applied even to this more complex set of sources in a way that parallels the analysis previously conducted only for EGUs. This analysis relies on evaluation of uniform levels of control stringency across all upwind states to find a level of emissions control that is cost-effective and collectively delivers meaningful downwind air quality improvement. For non-EGUs, the EPA identified the most impactful industries and emissions unit types and evaluated emissions control strategies for these units that have been demonstrated or applied across many similar facilities and emissions units. The EPA has evaluated whether these strategies are cost-effective on a cost-per-ton basis, and in particular has compared these strategies to those selected for EGUs. This analysis is set forth in sections V and VI of this document and associated technical support documents.

Commenter's statement that the establishment of a uniform level of control for each group of industrial units across the linked upwind states fails to assess with greater precision or define a state-specific proportion of emissions reduction that is needed for each downwind receptor is effectively an attempt to relitigate EME Homer City. The Court in that case rejected that the EPA must define significant contribution by reference to a specific quantum of reductions that each state must achieve that is proportional to its impact at a downwind receptor. The Court agreed with the EPA's concerns as to why that approach would be problematically complicated or even impossible to apply in light of the complex set of linkages among states for a regional pollutant like ozone. See 572 U.S. at 515–17. The Court found that the use of uniform cost thresholds to allocate responsibility for good neighbor obligations to be efficient and equitable, in that it requires those sources that have done less to reduce their emissions to come up to a minimum level of performance to what other sources are already achieving. Id. at 519. The EPA's analysis in this action in section V of this document establishes that this continues to be an appropriate means of delivering meaningful air quality improvement to downwind receptors, taking into consideration the complexities of interstate pollution transport.

Not every upwind state has the same mix of non-EGU industries and emissions unit types, and it is also the case that the costs for installation of the selected level of control technology will vary from facility to facility based on site-specific considerations. This is also true for the set of EGU sources regulated here and in previous CSAPR rulemakings. These real-world complexities do not obviate the broader policy and technical judgements that the EPA makes at Step 3 regarding what level of emissions control performance can be achieved on a region-wide basis to resolve significant contribution for a regional-scale pollutant like ozone. The EPA's design of cost thresholds derives from the identification of discrete types of NO_X emissions control strategies. The EPA then identifies a representative cost-effectiveness on a per ton basis for that technology. In the Step 3 analysis, it is not the cost per ton value itself that is inherently meaningful, but rather how that cost-effectiveness value relates to other control stringencies, how many emissions reductions may be obtained, and how air quality is ultimately impacted. The selected level of control stringency reflects a point at which further emissions mitigation strategies become excessively costly on a per-ton basis while also delivering far fewer additional emissions reductions and air quality benefits. This is often referred to as a “knee in the curve” analysis. There are always inherent uncertainties in identifying a representative cost per ton value for any particular control stringency, but this in itself does not upset the EPA's ability to render an overall policy judgment based on the Step 3 factors as to a set of emissions control strategies that together eliminate significant contribution. See86 FR 23054, 23073 (responding to similar comments on the Revised CSAPR Update).

We note that the EPA has made a number of adjustments to the non-EGU emissions limits identified at Step 4 to accommodate legitimate concerns regarding the ability of certain non-EGU facilities to meet the emissions control requirements that the EPA had proposed. The Agency's determinations regarding feasibility and installation timing for pollution controls are comparable and not inconsistent between EGUs and non-EGUs. The EPA is not establishing a trading program for non-EGUs because the Agency does not have adequate baseline emissions data and information on monitoring currently at many of these emissions units to develop emissions budgets that could reliably implement the Step 3 determinations made in this action. However, for most of the non-EGU industries, the EPA is not mandating a specific control technology and is instead establishing numeric emissions limits that are uniform across the region and that allow sources to choose how to comply. The EPA's analysis, including review of RACT determinations, consent decrees, and permitting actions, shows that these emissions limits and control requirements are achievable by existing units in the non-EGU industries covered by this final rule. This rule will therefore bring all of these impactful industries and unit types across the region of linked upwind states up to this standard of performance, and thus will result collectively in a relatively substantial decrease in ozone-season NO_X emissions, with associated reductions in ozone levels projected to result at the downwind receptors. This is further discussed in section V.D.

Some commenters alleged that the EPA's EGU control strategy goes beyond the cost-effectiveness determinations of prior transport rules, and they believe that the EPA's true objective is to force the retirement of coal plants. First, we note that the EGU emissions control strategy is premised entirely on at-the- source emissions control technologies that are widely available and in use across the EGU fleet. It is not the EPA's intention in this rule to force the retirement of any EGU or non-EGU facilities or emissions units but to identify and eliminate significant contribution under CAA section 110(a)(2)(D)(i)(I) based on cost-effective and proven control technologies that are appropriate in relation to address the problem of interstate transport for the 2015 ozone NAAQS. Further, determinations of cost-effectiveness must be made in relation to the particular statutory provision and its purpose. The EPA recognized in CSAPR, for example, that additional emissions reductions beyond what were determined to be cost-effective in that action could be required to implement good neighbor obligations if a NAAQS were revised to a more protective level. See76 FR 48210. Here it is not surprising that a more stringent level of control could be found justified in implementing transport obligations for the more protective 2015 ozone NAAQS. Those reductions are projected to deliver meaningful air quality improvement to downwind receptors, as discussed in section V.D of this document. Those air quality benefits continue to compare favorably to the air quality benefits that will be delivered through the combined non-EGU emissions limits, which apply to nine non-EGU industries (see section V.C of this document). We find that the implementation of both the EGU and non-EGU strategies identified in section V of this document together represent the appropriate level of emissions control stringency to eliminate significant contribution under CAA section 110(a)(2)(D)(i)(I).

Finally, the EPA also analyzed for overcontrol and does not identify any. Some commenters misstate the purpose of this rule as bringing downwind receptors into attainment. In line with the statutory directive in CAA section 110(a)(2)(D)(i)(I), this rule eliminates “significant contribution” from upwind states; while the rule has substantial air quality benefits for downwind receptors, in many cases we project that a nonattainment or maintenance problem will continue to persist through 2023 and 2026 despite the emissions reductions achieved by this rule. Commenters alleging overcontrol have not met the requirement that overcontrol be established by particularized evidence through as-applied challenges. The Supreme Court has recognized that the EPA also has an obligation to avoid under-control and must have some leeway in fulfilling the good neighbor mandate of the Act given uncertainty in making forward projections of air quality and the efficacy or impact of emissions control determinations. See EME Homer City, 572 U.S. at 523. This is further addressed in section V.D.4 of this document.

d. Step 4 Approach

The EPA is finalizing an approach similar to its prior transport rulemakings to implement the necessary emissions reductions through permanent and enforceable measures. The EPA is requiring EGU sources to participate in an emissions trading program and is making additional enhancements to the trading regime to maintain the selected control stringency over time and improve emissions performance at individual units, offering a necessary measure of assurance that emissions controls will be operated throughout the ozone season. For non-EGUs, the EPA is finalizing permanent and enforceable emissions rate limits and work practice standards, and associated compliance requirements, for several types of NO_X -emitting combustion units across several industrial sectors. The measures for both EGUs and non-EGUs are required throughout the May 1-September 30 ozone season of each year. The EGU program will begin with the 2023 ozone season, and the non-EGU implementation schedule is targeted to the 2026 ozone season. Refer to section VI.A of this document for details on the implementation schedule.

Based on the EPA's experience in implementing prior transport rulemakings, the Agency is making several enhancements to its trading-program approach for implementing good neighbor requirements for EGUs. In CSAPR, the CSAPR Update, and the Revised CSAPR Update, the EPA established interstate trading programs for EGUs to implement the necessary emissions reductions. In each of these rules, EGUs in each covered state are assigned an emissions budget in each control period for their collective emissions. Emissions allowances are allocated to units covered by the trading program, and the covered units then surrender allowances after the close of the control period, usually in an amount equal to their ozone season EGU NO_X emissions. While these programs have been effective in achieving overall reductions in emissions, experience has shown that these programs may not fully reflect in perpetuity the degree of emissions stringency determined necessary to eliminate significant contribution in Step 3 and may not adequately ensure the control of emissions throughout all days of the ozone season. At the same time, the EPA continues to find that an interstate-trading program approach delivers substantial benefits at Step 4 in terms of affording an appropriate degree of compliance flexibility, certainty in emissions outcomes, data and performance transparency, and cost-effective achievement of a high degree of aggregate emissions reductions. As such, the EPA is retaining an interstate trading program approach while making several enhancements to that approach.

Thus, in this rulemaking, the EPA is including dynamic budget-setting procedures in the regulations that will allow state emissions budgets for control periods in 2026 and later years to reflect more current data on the composition and utilization of the EGU fleet ( e.g., the 2026 budgets will reflect recent data through 2024 data, the 2027 budgets will reflect data through 2025, etc.). These enhancements will enable the trading program to better maintain over time the selected control stringency that was determined to be necessary to address states' good neighbor obligations with respect to the 2015 ozone NAAQS. In prior programs, where state emissions budgets were static across years rather than calibrated to yearly fleet changes, the EPA has observed instances of units idling their emissions controls in the latter years of the program. To provide greater certainty regarding the minimum quantities of allowances that will be available for compliance for the control periods in 2026 through 2029, the EPA is also establishing preset state emissions budgets for these control periods, and a dynamic state emissions budget determined for one of these control periods will apply only if it is higher than the state's preset budget for the control period.

In the trading programs established for ozone season NO_X emissions under CSAPR, the CSAPR Update, and the Revised CSAPR Update, the EPA included assurance provisions to limit state emissions to levels below 121 percent of the state's budget by requiring additional allowance surrenders in the instance that emissions in the state exceed this level. This limit on the degree to which a state's emissions can exceed its budget is designed to allow for a certain level of year-to-year variability in power sector emissions to account for fluctuations in demand and EGU operations and is responsive to previous court decisions (see discussion in section VI.B.5 of this document). In this action, the EPA is maintaining the existing assurance provisions that limit state emissions to levels below a percentage of the state's budget by requiring additional allowance surrenders in any instance where emissions in the state exceed the specified level, but with adjustments that allow the level to exceed 121 percent of a state's budget in a given control period if necessary to account for actual operational conditions in that control period. In addition, the EPA is also making several additional enhancements to the EGU trading program in this action, including routine recalibrations of the total amount of banked allowances, unit-specific backstop daily emissions rates for certain units, and unit-specific secondary emissions limitations for certain units that contribute to exceedances of the assurance levels, to ensure EGU emissions control operation and associated air quality improvements. Implementation of the EGU emissions reductions using a CSAPR NO_X trading program is further described in section VI.B of this document.

In this rule, the EPA is also establishing emissions limitations for the non-EGU industry sources listed in Table II.A–1. The EPA has the authority to require emissions limitations from stationary sources, as well as from other sources and emissions activities, under CAA section 110(a)(2)(D)(i)(I). The EPA finds that requiring NO_X emissions reductions through emissions rate limits and control technology requirements for certain non-EGU industrial sources that the EPA found at Step 3 to be relatively impactful on downwind air quality is an effective strategy for reducing regional ozone transport. Therefore, the EPA is establishing NO_X emissions limitations and associated compliance requirements for non-EGU sources to ensure the elimination of significant contribution of ozone precursor emissions required under the interstate transport provision for the 2015 ozone NAAQS.

Finally, the EPA finds that the control measures determined to be required for the identified EGU and non-EGU sources apply to both existing units and any new, modified, or reconstructed units meeting the applicability criteria established in this final rule. This is consistent with the EPA's transport actions dating back to the NO_X SIP Call and the NO_X Budget Trading Program. In all CSAPR EGU trading programs, for instance, new EGUs are subject to the program, and the EPA has established provisions for the allocation of allowances to such units through “new unit set asides.” See, e.g.,86 FR 23126. In the NO_X SIP Call, the EPA required that states cover new and existing units in the relevant source sectors through an enforceable cap or other emissions limitation. See40 CFR 51.121(f). The EPA's approach of including new units in the NO_X Budget Trading Program promulgated under the EPA's CAA section 126 authority was upheld by the D.C. Circuit in Appalachian Power v. EPA, 249 F.3d 1032 (2001). As the court noted, the EPA explained in its action:

Once EPA has determined that the emissions from the existing sources in an upwind State already make a significant contribution to one or more petitioning downwind States, any additional emissions from a new source in that upwind State would also constitute a portion of that significant contribution, unless the emissions from that new source are limited to the level of highly effective controls.

Id. at 1058 (quoting EPA 1999 RTC at 39). The court affirmed this approach: “Indeed, it would be irrational to enable the EPA to make findings that a group of sources in an upwind state contribute to downwind nonattainment, but then preclude the EPA from regulating new sources that contribute to that same pollution.” Id. at 1057–58. The EPA is implementing the same court-affirmed approach in this action because this reasoning is equally applicable to addressing interstate transport obligations under CAA section 110(a)(2)(D)(i)(I) for the 2015 ozone NAAQS.

Comment: Commenters took issue with aspects of the EPA's proposed Step 4 approach. Commenters argued the EPA could not set unit- or source-specific emissions limits or other control requirements, for EGUs or non-EGUs. Commenters argued that various aspects of the non-EGU emissions control strategy would not be feasible for their facilities or were otherwise flawed. Many industrial-source and EGU commenters argued that the EPA had not provided sufficient time for sources to come into compliance. Commenters also challenged the EGU trading program “enhancements” as unnecessary or beyond the EPA's authority. In this regard, commenters argued that these changes deviated from the EPA's prior approach, were unnecessary overcontrol, constituted a command-and-control approach, could not be supported on the basis of environmental justice benefits, or were otherwise unlawful for other reasons. These commenters argue that the EPA's Step 4 dynamic budget approach for EGU regulation purportedly re-defines each state's “significant contribution” annually and independent of any impact (or lack thereof) on air quality. They further argue that under this dynamic budgeting approach, even if a state eliminates the “amount” the EPA has identified as the state's significant contribution by respecting a given control period's emissions budget, sources within that state are expected to continue to make further reductions by operating their controls in a particular manner in subsequent control periods under potentially lower emissions budgets, which these commenters argue is inconsistent with case law on prior CSAPR rules.

Response: Many of these comments regarding Step 4 issues are addressed elsewhere in this document or in the RTC document. The EPA's authority to establish unit- or source-specific emissions rates is addressed in section IV.B.1 of this document. Responses to comments and adjustments in the timing requirements of the final rule compared to proposal are discussed in VI.A. Responses to comments and adjustments in emissions control requirements for non-EGUs in the final rule compared to proposal are in section VI.C of this document.

Responses to comments on the EGU trading program enhancements and adjustments in the final rule are contained in section VI.B of this document. However, here, in light of the changes in the emissions trading program for EGUs that we are finalizing in this action as compared to prior EGU emissions trading programs promulgated to address good neighbor obligations under other NAAQS, we set forth responses to comments specific to this topic.

The EPA finds that these comments confuse Step 3 emissions reduction stringency determinations with Step 4 implementation program details. In this rulemaking's Step 3 analysis, the EPA is measuring emissions reduction potential from improving effective emissions rates across groups of EGUs adopting applicable pollution control measures and selecting a uniform control level whose effective emissions rates deliver an acceptable outcome under the multifactor test (including a finding of no overcontrol at the selected control stringency level). The “amounts” defined as significant contribution to nonattainment and interference with maintenance are emissions that occur at effective emissions rates above the control stringency level selected at Step 3. That is, if a state's affected EGUs fail to reduce their effective emissions rates in line with the widely available and cost-effective control measures identified, they have therefore failed to eliminate their significant contribution to nonattainment and interference with maintenance of this NAAQS.

In this rule, the EPA is finalizing several “enhancements” to its existing Group 3 emissions trading program for ozone season NO_X, for reasons explained in section VI.B.1 of this document. In general, these changes will ensure that the emissions control program promulgated for EGUs at Step 4 of the EPA's 4-step interstate transport framework is in alignment with the emissions control stringency determinations the EPA made at Step 3. These enhancements reflect lessons learned through the EPA's experience with prior trading programs implemented under the good neighbor provision and ensure that the implementation of the elimination of significant contribution through an emissions trading program remains durable through a period of power sector transition. None of commenters' arguments against the EPA's authority to implement these enhancements are persuasive.

First, the EPA is not mandating that any EGU must install SCR technology. All but one of the enhancements to the trading program continue to be implemented through allowance-holding requirements under the mass-based emissions budget and trading system, including the backstop rate. (The secondary emissions limitation, which is not implemented through allowance-holding requirements under the mass-based emissions budget and trading system, and which is discussed in section VI.B.1.c.ii of this document, merely establishes a stronger deterrent for a type of conduct that was already strongly discouraged under the pre-existing trading program regulations). Nonetheless, the EPA does have the authority to impose unit-specific emissions limits under the exercise of its FIP authority, and it has done so in this action for non-EGU industrial sources. This authority is distinct from the EPA's title I permitting authority as discussed by certain commenters, and the scope of that permitting authority is not relevant to this action.

The quantification of emissions budgets in an allowance-based emissions trading program is one of multiple potential Step 4 implementation program design choices that states and the EPA have authority to select in securing the emissions reductions deemed necessary under Step 3. See CAA section 110(a)(2)(A). The EPA and the states routinely determine control stringency on an emissions rate basis in line with demonstrated pollution control opportunities, and both the EPA and the states have implementation program design discretion to determine what compliance requirements, whether expressed on a rate, mass, concentration, or percentage basis, will assure an emissions performance that reflects the control stringency required. Dynamic budgets in the Step 4 implementation of this rule are simply to ensure the trading program continues to incentivize the implementation of the EGU control strategies we find are necessary to eliminate significant contribution at Step 3. The key distinction between dynamic budget approaches and preset budget approaches is not one in stringency or authority, but rather in timing and data resources for determining the suitable mass-based limits that are as well-matched as possible to expected emissions of the affected EGUs achieving the emissions rate-based control stringency deemed necessary under Step 3 to eliminate significant contribution to nonattainment and interference with maintenance of the NAAQS.

The EPA does not agree that the administrative mechanisms by which it will implement “dynamic budgeting” conflict with CAA section 307(d) or the Administrative Procedure Act. The EPA is promulgating a complete FIP in this action, and the codified language of that FIP will not need to be modified as budgets are adjusted. This is because the FIP establishes the formula by which the budgets will be calculated each year (with preset budgets functioning as a floor from 2026 through 2029). This is no different than how the EPA has implemented other calculations such as updating allocations using a rolling set of data in its prior CSAPR trading programs. See, e.g.,87 FR 10786. We view these actions as fundamentally ministerial in nature in that no exercise of Agency discretion is required. This process will rely on notices of availability of the relevant data in the Federal Register , coupled with an opportunity for the public to correct any errors they may identify in the data before the EPA sets each updated budget. See section VI.B.4 for more detail on how the EPA intends to implement dynamic budgeting. As in prior transport rules, this rule provides the opportunity for administrative appeal should an interested party identify some flaw in the EPA's updated data. See 40 CFR 78.1(b)(19)(i) (2023). That process is coupled with the availability of judicial review should the party remain dissatisfied with the EPA's resolution of complaints. See40 CFR 78.1(a)(2) (requiring administrative adjudication as a prerequisite for judicial review). This administrative process has worked well throughout the history of implementing good neighbor trading programs under Part 97, and no such disputes have necessitated judicial resolution.

Further, because the dynamic budgets simply implement the stringency level reflective of the emissions control performance the EPA has determined at Step 3 for the covered EGUs, the EPA does not agree that any “potential variables” that are unforeseeable now could upset the basis for the formula the EPA is establishing in this action. The EPA has adjusted the role of dynamic budgeting in this final rule as compared to the proposal. See sections VI.B.1 and VI.B.4 of the preamble. In particular, the EPA is applying an approach to budget setting through 2029 that will use the greater of either a preset budget based on information known to the Agency at the time of this action, or the dynamic budget to be calculated based upon future data yet to be reported. Thus, through 2029 the imposition of a dynamic budget would only increase rather than diminish the emissions allowed for that control period compared to the preset budgets established in this action. In addition, the EPA will determine each state's dynamic budget based on a rolling 3-year average of the state's heat input, thus smoothing out trends to account for interannual variability in demand and heat input and provide greater certainty and predictability as the budget updates from year to year.

Moreover, the EPA does not agree that the EPA is constrained by the statute to only implement good neighbor obligations through fixed, unchanging, mass-based emissions budgets. See section III.B.1 of this document. The EPA finds good reason based on its experience with trading programs using fixed budgets why this approach does not necessarily ensure the elimination of significant contribution in perpetuity. The EPA has already once adjusted its historical approach to better account for known, upcoming changes in the EGU fleet to ensure mass-based emissions budgets adequately incentivize the control strategy determined at Step 3. This adjustment was introduced in the Revised CSAPR Update. See 82 FR 23121–22. The EPA now believes it is appropriate to ensure in a more comprehensive manner, and in perpetuity, that the mass-based emissions budget incentivize continuing implementation of the Step 3 control strategies to ensure significant contribution is eliminated in all upwind states and remains so. The dynamic budget-setting process preserves these incentives over time by calculating the state emissions budgets for each future control period so as to reflect the Step 3 control stringency finalized in this rule as applied to the most current information regarding the composition of the power sector in the control period. This is fully analogous in material respect to an approach to implementation at Step 4 that relies on application of unit-specific emissions rates that apply in perpetuity. The availability of unit-specific emissions rates as a means to eliminate significant contribution is discussed in further detail in section III.B.1 of this document. The EPA also explained this in the proposal. See87 FR 20095–96. The EPA does not agree that either dynamic budgeting or the backstop rate results in overcontrol. See section V.D.4 of this document.

The EPA is enhancing the trading program to help reconcile the approach of using mass-based budgets to achieve the elimination of significant contribution with the Wisconsin directive to provide a complete remedy under the good neighbor provision. This approach also better accords with ensuring measures to attain and maintain the NAAQS are permanent and enforceable. The dynamic budget approach recognizes that the uncertainty around future fleet conditions increases the further into the future one looks (and the EPA must look further under the “full remedy” directive). To preserve its ability to successfully implement its identified Step 3 stringency, the EPA is designing the implementation of this rule's emissions control program to benefit from the future availability of better data from the regulated sources to inform its application of its stringency measures identified in this rule.

The EPA does not agree with commenters who suggest that these enhancements are undertaken for the purpose of a non-statutory “environmental justice” objective. As explained in section VI.B of this document, certain enhancements to the trading program ensure that each EGU is adequately incentivized to continuously operate its emissions controls once those controls are installed. One commenter contends that the backstop emissions rate is not authorized based on environmental justice considerations, since it is not necessary and is overcontrol with respect to the EPA's statutory authority to address good neighbor obligations. But the EPA disagrees with the premise that these enhancements are unrelated to the statutory obligation to eliminate significant contribution. Taking measures to ensure that each upwind source covered by an emissions trading program to eliminate significant contribution is operating its installed pollution controls on a more continuous and consistent basis throughout the ozone season is entirely appropriate in light of the daily nature of the ozone problem, the impacts to public health and the environment from ozone that can occur through short-term exposure ( e.g., over a course of hours), the fact that the 2015 ozone NAAQS is expressed as an 8-hour average, and that only a small number of days in excess of the ozone NAAQS are necessary to place a downwind area in nonattainment, resulting in continuing and/or increased regulatory burden on the downwind jurisdiction. See section III.A of this document.

Further, the D.C. Circuit has held that the EPA must ensure that its good neighbor program has eliminated each state's sources from continuing to significantly contribute to nonattainment or interfere with maintenance in downwind states. See North Carolina, 531 F.3d at 921. The commenters neglect to acknowledge the scenario that has frequently borne out in prior programs, in which future fleet changes that were not known at the time of initial setting of state emissions budgets produce unexpected “hot air” in the budget that, if unaccounted for, other units can exploit to forgo identified cost-effective mitigation measures deemed necessary to eliminate significant contribution to nonattainment and interference with maintenance of the NAAQS.

The EPA's experience is that fixed mass-based budgets that are determined based only on the profile of the power sector at the time the rule is promulgated, and without any additional requirement for pollution controls operation, can become quickly obsolete if the composition of the group of affected EGUs changes notably over time. As some sources retire, other sources relax their operation of NO_X controls in response to a growing surplus of allowances, even though the EPA had concluded that ongoing operation of those controls is necessary to meet the statutory good neighbor requirements. For instance, under the CSAPR Update, in the 2018–2020 period, the fixed budget approach enabled large, frequently run units with existing SCR controls to not optimize those controls even though the EPA's assessment (as reflected in the CSAPR Update) was that the optimization of those controls was necessary to eliminate significant contribution. This deterioration in emission rate at SCR-controlled coal plants was widely observed across the CSAPR Update geography as the program advanced into later years and allowance price deteriorated. Whereas coal sources with SCR performed, on average, at a 0.086 lb/mmBtu rate in 2017, that same set of sources saw their environmental performance worsen to a 0.099 lb/mmBtu rate in 2020. A Congressional Research Service Report on EPA prior CSAPR trading programs indicated low prices observed in later years “could lead to some decisions not to run some pollution controls at maximum output. This would, in turn, lead to higher emissions”.

In the case of individual units, this deterioration in performance can be quite pronounced and can occur as quickly as the second or third control period, as in the case of Miami Fort Unit 7 in Ohio in 2019, discussed in section V.B of this document. The absence of a sufficient incentive under the trading program to implement the identified control strategy at Step 3 can even result in collective emissions that exceed state-wide assurance levels. The EPA established these levels beginning with CSAPR, above which enhanced allowance-surrender requirements are triggered, in an effort to ensure sources in each state are held to eliminate their own significant contribution, which the D.C. Circuit has held is legally required, see North Carolina, 531 F.3d 896, 906–08 (D.C. Cir. 2008). In four instances over the course of the 2019, 2020, and 2021 control periods under the CSAPR Update, sources in Mississippi and Missouri collectively exceeded their state-wide assurance levels in part due to deterioration in emissions performance that can be attributed to a glut of allowances within the CSAPR Update. See section VI.B.8 of the preamble.

Thus, while this trading program structure may achieve some environmental benefit through fixed emissions budgets for initial control periods, over time those fixed budgets cease to have their intended effect, and remaining operating facilities can, and have, increased emissions or even discontinued the operation of their emissions controls. This, in turn, can lead to the continuation (or re-emergence) of significant contribution in terms of a recurrence of excessive emissions that had been slated for permanent elimination under the EPA's determinations at Step 3. Although the EPA has always intended for its trading programs to provide flexibility, the Agency did not expect and has certainly never endorsed the use of that flexibility to stop the operation of controls that have already been installed. See, e.g.,76 FR 48256–57 (“[I]t would be inappropriate for a state linked to downwind nonattainment or maintenance areas to stop operating existing pollution control equipment (which would increase their emissions and contribution).”). Despite the EPA's expectations in CSAPR, the historical data establishes a real risk of “under-control” if the existing trading framework is not improved upon. See EME Homer City, 572 U.S. at 523 (“[T]he Agency also has a statutory obligation to avoid `under-control,' i.e., to maximize achievement of attainment downwind.”).

This result is also inconsistent with the statutory mandate to “prohibit” significant contribution and interference with maintenance of the NAAQS in downwind states, as evidenced most clearly in CAA section 126, which makes it unlawful for a source “to operate more than three months after [a finding that the source emits or would emit in violation of the good neighbor provision] has been made with respect to it.” 42 U.S.C. 7426(c)(2) (emphasis added). See also North Carolina, 531 F.3d at 906–08 (each state must be held to the elimination of its own significant contribution). The purpose of the Agency's interstate trading programs under the good neighbor provision is to afford sources some flexibility in achieving region-wide emissions reductions; however, there is no justification that can be sustained within that framework for sources in certain areas within that region, or during periods of high ozone when good emissions performance is most essential, to emit at levels well in excess of the EPA's Step 3 determinations of significant contribution. Significant contribution, according to the statute, must be “prohibited.” CAA section 110(a)(2)(D)(i).

Thus, these trading program enhancements are within the EPA's authority under CAA section 110(a)(2)(D)(i)(I) to eliminate interstate ozone pollution that significantly contributes to nonattainment or interferes with maintenance in downwind states. These enhancements ensure the elimination of significant contribution across all upwind states and throughout each ozone season. We observe in the Ozone Transport Policy Analysis Final Rule TSD, section E, that the trading program enhancements may also benefit underserved and overburdened communities downwind of EGUs in the covered geography of the final rule. See section VI.B of this document. This does not detract from the statutorily-authorized basis for these changes, and the EPA finds nothing impermissible in acknowledging the reality of these potential benefits for underserved and overburdened communities.

The EPA appreciates a commenter's concern that our actions be legally defensible. The EPA acknowledges that the changes to the trading program structure for implementing good neighbor obligations discussed here constitute a change in the policy underlying its prior transport-rule trading programs for EGUs. However, the EPA is confident that these changes are in compliance with the holdings in judicial decisions reviewing prior transport rules. The fact that the EPA is making changes does not somehow render these enhancements legally impermissible or even subject to a heightened standard of review. See FCC v. Fox Television Stations, 556 U.S. 502, 514 (2009) (“We find no basis in the Administrative Procedure Act or in our opinions for a requirement that all agency change be subjected to more searching review.”). We have explained previously and elsewhere in the record that there are “good reasons” for the “new policy.” See id. at 515. And, we are of course fully aware that we have changed our position. See id. at 514–15. Specifically, we have gone from previously treating fixed, mass-based budgets as sufficient to eliminate significant contribution, to an approach for purposes of the 2015 ozone NAAQS reflecting a more nuanced understanding of how an emissions trading program that does not properly anticipate future fleet conditions at Step 4 may fail to achieve the elimination of emissions that should be prohibited based on our findings at Step 3. Further, we find there to be no “serious reliance interests” that have been or even could have been “engendered” by any prior policy on these issues, see id. at 515–16. The EPA is implementing these enhancements for the first time with respect to a new obligation—good neighbor requirements for the 2015 ozone NAAQS. No party reasonably could have invested substantial resources to-date to comply with an obligation that was heretofore undefined; and no commenter has supplied any information to the contrary.

2. FIP Authority for Each State Covered by the Rule

On October 26, 2015, the EPA promulgated a revision to the 2015 8-hour ozone NAAQS, lowering the level of both the primary and secondary standards to 0.070 parts per million (ppm). These revisions of the NAAQS, in turn, established a 3-year deadline for states to provide SIP submissions addressing infrastructure requirements under CAA sections 110(a)(1) and CAA 110(a)(2), including the good neighbor provision, by October 1, 2018. If the EPA makes a determination that a state failed to submit a SIP, or if EPA disapproves a SIP submission, then the EPA is obligated under CAA section 110(c) to promulgate a FIP for that state within 2 years. For a more detailed discussion of CAA section 110 authority and timelines, refer to section III.C of this document.

The EPA is finalizing this FIP action now to address 23 states' good neighbor obligations for the 2015 ozone NAAQS. For each state for which the EPA is finalizing this FIP, the EPA either issued final findings of failure to submit or has issued a final disapproval of that state's SIP submission.

Several commenters asserted that the sequence of the EPA's actions, and in particular, the timing of its proposed FIP (which was signed on February 28, 2022, and published on April 6, 2022) in relation to the timing of its proposed SIP disapprovals (most of which were published on February 22, 2022, four of which were published on May 24, 2022, and one of which was published on October 25, 2022), was either unlawful or unreasonable in light of the sequence of steps required under CAA section 110(k) and (c).

These commenters are incorrect. As an initial matter, concerns about the timing or substance of the EPA's actions on the SIP submittals are beyond the scope of this action. Nor are the timing or contents of merely proposed actions to be considered final agency actions or subject to judicial review. See In re Murray Energy, 788 F.3d 330 (D.C. Cir. 2015). With these principles in mind, the timing of this final action is lawful under the Act. First, the EPA is not required to wait to propose a FIP until after the Agency proposes or finalizes a SIP disapproval or makes a finding of failure to submit. CAA section 110(c) authorizes the EPA to promulgate a FIP “at any time within 2 years” of a SIP disapproval or making a finding of failure to submit. The Supreme Court recognized in EME Homer City that the EPA is not obligated to first define a state's good neighbor obligations or give the state an additional opportunity to submit an approvable SIP before promulgating a FIP: “EPA is not obliged to wait two years or postpone its action even a single day: The Act empowers the Agency to promulgate a FIP `at any time' within the two-year limit.” Thus, the EPA may promulgate a FIP contemporaneously with or immediately following predicate final SIP disapproval (or finding no SIP was submitted). To accomplish this, the EPA must necessarily be able to propose a FIP prior to taking final action to disapprove a SIP or make a finding of failure to submit.

Second, and more importantly, the EPA has established predicate authority to promulgate FIPs for all of the covered states through its action with respect to the relevant SIP submittals. A brief history of these actions follows:

On February 22, 2022, the EPA proposed to disapprove 19 good neighbor SIP submissions (Alabama, Arkansas, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, New Jersey, New York, Ohio, Oklahoma, Tennessee, Texas, West Virginia, Wisconsin). Alabama subsequently withdrew its SIP submission and re-submitted a SIP submission on June 22, 2022. The EPA proposed to disapprove that SIP submittal on October 25, 2022. The EPA proposed to disapprove good neighbor SIP submissions for four additional states, California, Nevada, Utah, and Wyoming, on May 24, 2022.

Subsequently, on January 31, 2023, the EPA Administrator signed a single disapproval action for all of the above states, with the exception of Tennessee and Wyoming. This action established the EPA's authority to promulgate FIPs for the disapproved states. (As explained in section IV.F of this document, the Agency is deferring action at this time for Tennessee and Wyoming with respect to its proposed FIP actions for those states. As discussed in section IV.F of this document, the EPA's most recent modeling and air quality analysis indicates that several states may be linked to downwind receptors for which we had not previously proposed disapproval or FIP action. The EPA anticipates addressing remaining interstate transport obligations for the 2015 ozone NAAQS for these in a subsequent rulemaking.)

Additionally, the EPA has taken action that has triggered the EPA's obligation under CAA section 110(c) to promulgate FIPs addressing the good neighbor provision for several downwind states. On December 5, 2019, the EPA published a rule finding that seven states (Maine, New Mexico, Pennsylvania, Rhode Island, South Dakota, Utah, and Virginia) failed to submit or otherwise make complete submissions that address the requirements of CAA section 110(a)(2)(D)(i)(I) for the 2015 ozone NAAQS. This finding triggered a 2-year deadline for the EPA to issue FIPs to address the good neighbor provision for these states by January 6, 2022. As the EPA has subsequently received and taken final action to approve good neighbor SIPs from Maine, Rhode Island, and South Dakota, the EPA currently has authority under the December 5, 2019, findings of failure to submit to issue FIPs for New Mexico, Pennsylvania, Utah, and Virginia. In this final rule, the EPA is issuing FIP requirements for Pennsylvania, Utah, and Virginia.

Further information on the procedural history establishing the EPA's authority for this final rule is provided in a document in the docket.

While the EPA's previous actions are sufficient to establish that the EPA's promulgation of this FIP action at this time is lawful, the timing of this action is all the more reasonable in light of the need for the EPA to address good neighbor obligations consistent with the rest of title I of the CAA. In particular, the D.C. Circuit in Wisconsin held that states and the EPA are obligated to fully address good neighbor obligations for ozone “as expeditiously as practical” and in no event later than the next relevant downwind attainment dates found in CAA section 181(a). In Maryland v. EPA, the D.C. Circuit made clear that Wisconsin' s and North Carolina' s holdings are fully applicable to the Marginal area attainment date for the 2015 ozone NAAQS, which fell on August 3, 2021. As discussed in section VI.A of this document, by finalizing this action now, the EPA is able to implement initial required emissions reductions to eliminate significant contribution by the 2023 ozone season, which is the last full ozone season before the next attainment date, the Moderate area attainment date of August 3, 2024. The Wisconsin court emphasized that the EPA has the authority under CAA section 110 to structure and time its actions in a manner such that the Agency can ensure necessary reductions are achieved in alignment with the downwind attainment schedule, and that is precisely what the EPA is doing here. The EPA provides further response to the comments on this issue in section 1 of the RTC document.

C. Other CAA Authorities for This Action

1. Withdrawal of Proposed Error Correction for Delaware

The EPA proposed at 87 FR 20036 to make an error correction under CAA section 110(k)(6) of its May 1, 2020, approval at 85 FR 25307 of the interstate transport elements for Delaware's October 11, 2018, and December 26, 2019, ozone infrastructure SIP submissions as satisfying the requirements of CAA section 110(a)(2)(D)(i)(I) for the 2015 ozone NAAQS. The EPA proposed to determine that the basis for the prior SIP approval was invalidated by the Agency's more recent technical evaluation of air quality modeling performed in support of the proposed rule, and that Delaware had unresolved interstate transport obligations for the 2015 ozone NAAQS. The EPA also proposed to issue a FIP for Delaware given these unresolved interstate transport obligations. However, based on the updated air quality modeling described in section IV.F. of this document and the technical assessment that informs this final rule, the EPA finds that Delaware is not projected to be linked to any downwind receptor above the 1 percent of the NAAQS threshold in 2023. Thus, based on the record before the Agency now, the original approval of Delaware's SIP submission was not in error, and the EPA is withdrawing its proposed error correction and proposed FIP for Delaware.

2. Application of Rule in Indian Country and Necessary or Appropriate Finding

The EPA is finalizing its determination that this rule will be applicable in all areas of Indian country (as defined at 18 U.S.C. 1151) within the covered geography of the final rule, as defined in this section. Certain areas of Indian country within the geography of the rule are or may be subject to state implementation planning authority. Other areas of Indian country within that geography are subject to tribal planning authority, although none of the relevant tribes have as yet sought eligibility to administer a tribal plan to implement the good neighbor provision. As described later, the EPA is including all areas of Indian country within the covered geography, notwithstanding whether those areas are currently subject to a state's implementation planning authority or the potential planning authority of a tribe.

a. Indian Country Subject to Tribal Jurisdiction

With respect to areas of Indian country not currently subject to a state's implementation planning authority— i.e., Indian reservation lands (with the partial exception of reservation lands located in the State of Oklahoma, as described further in this section) and other areas of Indian country over which the EPA or a tribe has demonstrated that a tribe has jurisdiction—the EPA here makes a “necessary or appropriate” finding that direct Federal implementation of the rule's requirements is warranted under CAA section 301(d)(4) and 40 CFR 49.11(a) (the areas of Indian country subject to this finding will be referred to as the CAA section 301(d) FIP areas). Indian Tribes may, but are not required to, submit tribal plans to implement CAA requirements, including the good neighbor provision. Section 301(d) of the CAA and 40 CFR part 49 authorize the Administrator to treat an Indian Tribe in the same manner as a state ( i.e., TAS) for purposes of developing and implementing a tribal plan implementing good neighbor obligations. See40 CFR 49.3; see also “Indian Tribes: Air Quality Planning and Management,” hereafter “Tribal Authority Rule” (63 FR 7254, February 12, 1998). The EPA is authorized to directly implement the good neighbor provision in the 301(d) FIP areas when it finds, consistent with the authority of CAA section 301—which the EPA has exercised in 40 CFR 49.11—that it is necessary or appropriate to do so.

The EPA hereby finds that it is both necessary and appropriate to regulate all new and existing EGU and industrial sources meeting the applicability criteria set forth in this rule in all of the 301(d) FIP areas that are located within the geographic scope of coverage of the rule. For purposes of this finding, the geographic scope of coverage of the rule means the areas of the United States encompassed within the borders of the states the EPA has determined to be linked at Steps 1 and 2 of the 4-step interstate transport framework. For EGU applicability criteria, see section VI.B of this document; for industrial-source applicability criteria, see section VI.C of this document. To EPA's knowledge, only one existing EGU or industrial source is located within the CAA section 301(d) FIP areas: the Bonanza Power Plant, an EGU source, located on the Uintah and Ouray Reservation, geographically located within the borders of Utah.

This finding is consistent with the EPA's prior good neighbor rules. In prior rulemakings under the good neighbor provision, the EPA has included all areas of Indian country within the geographic scope of those FIPs, such that any new or existing sources meeting the rules' applicability criteria would be subject to the rule irrespective of whether subject to state or tribal underlying CAA planning authority. In CSAPR, the CSAPR Update, and the Revised CSAPR Update, the scope of the emissions trading programs established for EGUs extended to cover all areas of Indian country located within the geographic boundaries of the covered states. In these rules, at the time of their promulgation, no existing units were located in the covered areas of Indian country; under the general applicability criteria of the trading programs, however, any new sources locating in such areas would become subject to the programs. Thus, the EPA established a separate allowance allocation that would be available for any new units locating in any of the relevant areas of Indian country. See, e.g.,76 FR 48293 (describing the CSAPR methodology of allowance allocation under the “Indian country new unit set-aside” provisions); see also id. at 48217 (explaining the EPA's source of authority for directly regulating in relevant areas of Indian country as necessary or appropriate). Further, in any action in which the EPA subsequently approved a state's SIP submittal to partially or wholly replace the provisions of a CSAPR FIP, the EPA has clearly delineated that it will continue to administer the Indian country new unit set aside for sources in any areas of Indian country geographically located within a state's borders and not subject to that state's CAA planning authority, and the state may not exercise jurisdiction over any such sources. See, e.g.,82 FR 46674, 46677 (October 6, 2017) (approving Alabama's SIP submission establishing a state CSAPR trading program for ozone season NO_X, but providing, “The SIP is not approved to apply on any Indian reservation land or in any other area where EPA or an Indian tribe has demonstrated that a tribe has jurisdiction.”).

In this rule, the EPA is taking an approach similar to the prior CSAPR rulemakings with respect to regulating sources in the CAA section 301(d) FIP areas. The EPA believes this approach is necessary and appropriate for several reasons. First, the purpose of this rule is to address the interstate transport of ozone on a national scale, and the technical record establishes that the nonattainment and maintenance receptors located throughout the country are impacted by sources of ozone pollution on a broad geographic scale. The upwind regions associated with each receptor typically span at least two, and often far more, states. Within the broad upwind region covered by this rule, the EPA is applying—consistent with the methodology of allocating upwind responsibility in prior transport rules going back to the NO_X SIP Call—a uniform level of control stringency (as determined separately for linkages existing in 2023, and linkages persisting in 2026). ( See section V of this document for a discussion of EPA's determination of control stringency for this rule.) Within this approach, consistency in rule requirements across all jurisdictions is vital in ensuring the remedy for ozone transport is, in the words of the Supreme Court, “efficient and equitable,” 572 U.S. 489, 519. In particular, as the Supreme Court found in EME Homer City Generation, allocating responsibility through uniform levels of control across the entire upwind geography is “equitable” because, by imposing uniform cost thresholds on regulated States, the EPA's rule subjects to stricter regulation those States that have done relatively less in the past to control their pollution. Upwind States that have not yet implemented pollution controls of the same stringency as their neighbors will be stopped from free riding on their neighbors' efforts to reduce pollution. They will have to reduce their emissions by installing devices of the kind in which neighboring States have already invested. Id.

In the context of addressing regional-scale ozone transport in this rule, the importance of a uniform level of stringency that extends to and includes the CAA section 301(d) FIP areas geographically located within the boundaries of the linked upwind states carries significant force. Failure to include all such areas within the scope of the rule creates a significant risk that these areas may be targeted for the siting of facilities emitting ozone-precursor pollutants, to avoid the regulatory costs that would be imposed under this rule in the surrounding areas of state jurisdiction. Electricity generation or the production of other goods and commodities may become more cost-competitive at any EGU or industrial sources not subject to the rule but located in a geography where the same types of sources are subject to the rule. For instance, the affected EGU source located on the Uintah and Ouray Reservation of the Ute Tribe is in an area that is interconnected with the western electricity grid and is owned and operated by an entity that generates and provides electricity to customers in several states. It is both necessary and appropriate, in the EPA's view, to avoid creating, via this rule, a structure of incentives that may cause generation or production—and the associated NO_X emissions—to shift into the CAA section 301(d) FIP areas to escape regulation needed to eliminate interstate transport under the good neighbor provision.

The EPA finds it is appropriate to directly implement the rule's requirements in the CAA section 301(d) FIP areas in this action rather than at a later date. Tribes have the opportunity to seek treatment as a state (TAS) and to undertake tribal implementation plans under the CAA. To date, the one tribe which could develop and seek approval of a tribal implementation plan to address good neighbor obligations with respect to an existing EGU in the CAA section 301(d) FIP areas for the 2015 ozone NAAQS (or for any other NAAQS), the Ute Indian Tribe of the Uintah and Ouray Reservation, has not expressed an intent to do so. Nor has the EPA heard such intentions from any other tribe, and it would not be reasonable to expect tribes to undertake that planning effort, particularly when no existing sources are currently located on their lands. Further, the EPA is mindful that under court precedent, the EPA and states bear an obligation to fully implement any required emissions reductions to eliminate significant contribution under the good neighbor provision as expeditiously as practicable and in alignment with downwind areas' attainment schedule under the Act. As discussed in section VI.A of this document, the EPA is implementing certain required emissions reductions by the 2023 ozone season, the last full ozone season before the 2024 Moderate area attainment date, and other key additional required emissions reductions by the 2026 ozone season, the last full ozone season before the 2027 Serious area attainment date. Absent the application of this FIP in the CAA section 301(d) FIP areas, NO_X emissions from any existing or new EGU or non-EGU sources located in, or locating in, the CAA section 301(d) FIP areas within the covered geography of the rule would remain unregulated for purposes of CAA section 110(a)(2)(D)(i)(I) for the 2015 ozone NAAQS and could continue or potentially increase. This would be inconsistent with the EPA's overall goal of aligning good neighbor obligations with the downwind areas' attainment schedule and to achieve emissions reductions as expeditiously as practicable.

Further, the EPA recognizes that Indian country, including the CAA section 301(d) FIP areas, is often home to communities with environmental justice concerns, and these communities may bear a disproportionate level of pollution burden as compared with other areas of the United States. The EPA's Fiscal Year 2022–2026 Strategic Plan includes an objective to promote environmental justice at the Federal, Tribal, state, and local levels and states: “Integration of environmental justice principles into all EPA activities with Tribal governments and in Indian country is designed to be flexible enough to accommodate EPA's Tribal program activities and goals, while at the same time meeting the Agency's environmental justice goals.” As described in section X.F of this document, the EPA offered Tribal consultation to 574 Tribes in April of 2022 and received no requests for Tribal consultation after publication of the proposed rulemaking. By including all areas of Indian country within the covered geography of the rule, the EPA is advancing environmental justice, lowering pollution burdens in such areas, and preventing the potential for “pollution havens” to form in such areas as a result of facilities seeking to locate there to avoid the requirements that would otherwise apply outside of such areas under this rule.

Therefore, to ensure timely alignment of all needed emissions reductions within the timetables of this rule, to ensure equitable distribution of the upwind pollution reduction obligation across all upwind jurisdictions, to avoid perverse economic incentives to locate sources of ozone-precursor pollution in the CAA section 301(d) FIP areas, and to deliver greater environmental justice to tribal communities in line with Executive Order 13985: Advancing Racial Equity and Support for Underserved Communities Through the Federal Government, the EPA finds it both necessary and appropriate that all existing and new EGU and industrial sources that are located in the CAA section 301(d) FIP areas within the geographic boundaries of the covered states, and which would be subject to this rule if located within areas subject to state CAA planning authority, should be included in this rule. The EPA issues this finding under CAA section 301(d)(4) of the Act and 40 CFR 49.11. Further, to avoid “unreasonable delay” in promulgating this FIP, as required under section 49.11, the EPA makes this finding now, to align emissions reduction obligations for any covered new or existing sources in the CAA section 301(d) FIP areas with the larger schedule of reductions under this rule. Because all other covered EGU and non-EGU sources within the geography of this rule would be subject to emissions reductions of uniform stringency beginning in the 2023 ozone season, and as necessary to fully and expeditiously address good neighbor obligations for the 2015 ozone NAAQS, there is little benefit to be had by not including the CAA section 301(d) FIP areas in this rule now and a potentially significant downside to not doing so.

The Agency recognizes that Tribal governments may still choose to seek TAS to develop a Tribal plan with respect to the obligations under this rule, and this determination does not preclude the tribes from taking such actions. Although the formal tribal consultation process associated with this action has concluded, the EPA is willing and available to engage with any tribe as this rule is implemented.

b. Indian Country Subject to State Implementation Planning Authority

Following the U.S. Supreme Court decision in McGirt v. Oklahoma, 140 S. Ct. 2452 (2020), the Governor of the State of Oklahoma requested approval under section 10211(a) of the Safe, Accountable, Flexible, Efficient Transportation Equity Act of 2005: A Legacy for Users, Public Law 109–59, 119 Stat. 1144, 1937 (August 10, 2005) (“SAFETEA”), to administer in certain areas of Indian country (as defined at 18 U.S.C. 1151) the State's environmental regulatory programs that were previously approved by the EPA for areas outside of Indian country. The State's request excluded certain areas of Indian country further described later. In addition, the State only sought approval to the extent that such approval is necessary for the State to administer a program in light of Oklahoma Dept. of Environmental Quality v. EPA, 740 F.3d 185 (D.C. Cir. 2014).

On October 1, 2020, the EPA approved Oklahoma's SAFETEA request to administer all the State's EPA-approved environmental regulatory programs, including the Oklahoma SIP, in the requested areas of Indian country. As requested by Oklahoma, the EPA's approval under SAFETEA does not include Indian country lands, including rights-of-way running through the same, that: (1) qualify as Indian allotments, the Indian titles to which have not been extinguished, under 18 U.S.C. 1151(c); (2) are held in trust by the United States on behalf of an individual Indian or Tribe; or (3) are owned in fee by a Tribe, if the Tribe (a) acquired that fee title to such land, or an area that included such land, in accordance with a treaty with the United States to which such Tribe was a party, and (b) never allotted the land to a member or citizen of the Tribe (collectively “excluded Indian country lands”).

The EPA's approval under SAFETEA expressly provided that to the extent EPA's prior approvals of Oklahoma's environmental programs excluded Indian country, any such exclusions are superseded for the geographic areas of Indian country covered by the EPA's approval of Oklahoma's SAFETEA request. The approval also provided that future revisions or amendments to Oklahoma's approved environmental regulatory programs would extend to the covered areas of Indian country (without any further need for additional requests under SAFETEA).

In a Federal Register document published on February 13, 2023 (88 FR 9336), the EPA disapproved the portion of an Oklahoma SIP submittal pertaining to the state's interstate transport obligations under CAA section 110(a)(2)(D)(i)(I) for the 2015 ozone NAAQS. Consistent with the D.C. Circuit's decision in ODEQ v. EPA and with the EPA's October 1, 2020 SAFETEA approval, the EPA has authority under CAA section 110(c) to promulgate a FIP as needed to address the disapproved aspects of Oklahoma's good neighbor SIP submittal. In accordance with the previous discussion, the EPA's FIP authority in this circumstance extends to all Indian country in Oklahoma, other than the excluded Indian country lands, as described previously. Because—per the State's request under SAFETEA—EPA's October 1, 2020 approval does not displace any SIP authority previously exercised by the State under the CAA as interpreted in ODEQ v. EPA, the EPA's FIP authority under CAA section 110(c) also applies to any Indian allotments or dependent Indian communities located outside of an Indian reservation over which there has been no demonstration of tribal authority. The EPA's FIP authority under CAA section 110(c) similarly applies to Indian allotments or dependent Indian communities located outside of an Indian reservation over which there has been no demonstration of tribal authority located in any other state within the geographic scope of this rule.

In light of the relevant legal authorities discussed above regarding the scope of the State of Oklahoma's regulatory jurisdiction under the CAA, the EPA has FIP authority under CAA section 110(c) with respect to all Indian country in Oklahoma other than excluded Indian country lands. To the extent any change occurs in the scope of Oklahoma's SIP authority in Indian country following finalization of this rule, and such change affects the exercise of FIP authority provided under section 110(c) of the Act, then, to the extent any such areas would fall more appropriately within the CAA section 301(d) FIP areas as described in section III.C.2.a of this document, the EPA's necessary or appropriate finding as set forth above with respect to all other CAA section 301(d) FIP areas within the geographic scope of coverage of the rule would apply.

D. Severability

The EPA regards this action as a complete remedy, which will as expeditiously as practicable implement good neighbor obligations for the 2015 ozone NAAQS for the covered states, consistent with the requirements of the Act. See North Carolina v. EPA, 531 F.3d 896, 911–12 (D.C. Cir. 2008); Wisconsin v. EPA, 938 F.3d 303, 313–20 (D.C. Cir. 2019); Maryland v. EPA, 958 F.3d 1185, 1204 (D.C. Cir. 2020); New York v. EPA, 964 F.3d 1214, 1226 (D.C. Cir. 2020); New York v. EPA, 781 Fed. App'x 4, 7–8 (D.C. Cir. 2019) (all holding that the EPA must address good neighbor obligations as expeditiously as practicable and by no later than the next applicable attainment date). Yet should a court find any discrete aspect of this document to be invalid, the Agency believes that the remaining aspects of this rule can and should continue to be implemented to the extent possible. In particular, this action promulgates a FIP for each covered state (and, pursuant to CAA section 301(d), for each area of tribal jurisdiction within the geographic boundaries of those states). Should any jurisdiction-specific aspect of the final rule be found invalid, the EPA views this rule as severable along those state and/or tribal jurisdictional lines, such that the rule can continue to be implemented as to any remaining jurisdictions. This action promulgates discrete emissions control requirements for the power sector and for each of seven other industries. Should any industry-specific aspect of the final rule be found invalid, the EPA views this rule as severable as between the different industries and different types of emissions control requirements. This is not intended to be an exhaustive list of the ways in which the rule may be severable. In the event any part of it is found invalid, our intention is that the remaining portions should continue to be implemented consistent with any judicial ruling.

The EPA's conclusion that this rule is severable also reflects the important public health and environmental benefits of this rulemaking in eliminating significant contribution and to ensure to the greatest extent possible the ability of both upwind states and downwind states and other relevant stakeholders to be able to rely on this final rule in their planning. Cf. Wisconsin, 938 F.3d at 336–37 (“As a general rule, we do not vacate regulations when doing so would risk significant harm to the public health or the environment.”); North Carolina v. EPA, 550 F.3d 1176, 1178 (D.C. Cir. 2008) (noting the need to preserve public health benefits); EME Homer City v. EPA, 795 F.3d 118, 132 (D.C. Cir. 2015) (noting the need to avoid disruption to emissions trading market that had developed).

IV. Analyzing Downwind Air Quality Problems and Contributions From Upwind States

A. Selection of Analytic Years for Evaluating Ozone Transport Contributions to Downwind Air Quality Problems

In this section, the EPA describes its process for selecting analytic years for air quality modeling and analyses performed to identify nonattainment and maintenance receptors and identify upwind state linkages. For this final rule, the EPA evaluated air quality to identify receptors at Step 1 for two analytic years: 2023 and 2026. The EPA evaluated interstate contributions to these receptors from individual upwind states at Step 2 for these two analytic years. In selecting these years, the EPA views 2023 and 2026 to constitute years by which key emissions reductions from EGUs and non-EGUS can be implemented “as expeditiously as practicable.” In addition, these years are the last full ozone seasons before the Moderate and Serious area attainment dates for the 2015 ozone NAAQS (ozone seasons run each year from May 1–September 30). To demonstrate attainment by these deadlines, downwind states would be required to rely on design values calculated using ozone data from 2021 through 2023 and 2024 through 2026, respectively. By focusing its analysis, and, potentially, achieving emissions reductions by, the last full ozone seasons before the attainment dates ( i.e., in 2023 or 2026), this final rule can assist the downwind areas with demonstrating attainment or receiving extensions of attainment dates under CAA section 181(a)(5). (The EPA explains in detail in sections V and VI of this document its determinations regarding which emissions reduction strategies can be implemented by 2023, and which emissions reduction strategies require additional time beyond that ozone season, or the 2026 ozone season.)

It would not be logical for the EPA to analyze any earlier year than 2023. The EPA continues to interpret the good neighbor provision as forward-looking, based on Congress's use of the future-tense “will” in CAA section 110(a)(2)(D)(i), an interpretation upheld in Wisconsin, 938 F.3d at 322. It would be “anomalous,” id., for the EPA to impose good neighbor obligations in 2023 and future years based solely on finding that “significant contribution” had existed at some time in the past. Id.

Applying this framework in the proposal, the EPA recognized that the 2021 Marginal area attainment date had already passed. Further, based on the timing of the proposal, it was not possible to finalize this rulemaking before the 2022 ozone season had also passed. Thus, the EPA has selected 2023 as the first appropriate future analytic year for this final rule because it reflects implementation of good neighbor obligations as expeditiously as practicable and coincides with the August 3, 2024, Moderate area attainment date established for the 2015 ozone NAAQS.

The EPA conducted additional analysis for 2026 to ensure a complete Step 3 analysis for future ozone transport contributions to downwind areas. As noted above, 2023 and 2026 coincide with the last full ozone seasons before future attainment dates for the 2015 ozone NAAQS. In addition, 2026 coincides with the ozone season by which key additional emissions reductions from EGUs and non-EGUs become available. Thus, the EPA analyzed additional years beyond 2023 to determine whether any additional emissions reductions that are impossible to obtain by the 2024 attainment date could still be necessary to fully address significant contribution. In all cases, implementation of necessary emissions reductions is as expeditiously as practicable, with all possible emissions reductions implemented by the next applicable attainment date.

The timing framework and selection of analytic years set forth above comports with the D.C. Circuit's direction in Wisconsin that implementing good neighbor obligations beyond the dates established for attainment may be justified on a proper showing of impossibility or necessity. See 938 F.3d at 320.

Comment: A commenter claims that the EPA has not followed the holdings of Wisconsin v. EPA, 938 F.3d 303 (D.C. Cir. 2019), North Carolina v. EPA, 550 F.3d 1176 (D.C. Cir. 2008), and Maryland v. EPA, 958 F. 3d 1185 (D.C. Cir. 2020) in the selection of analytic years, in that commenter interprets those decisions as holding that the EPA must “harmonize” the exact timing of upwind emissions reductions with when downwind states implement their required reductions. Commenter also points to the EPA's proposed action on New York's Good Neighbor SIP submission specifically to argue that the EPA is treating upwind and downwind states dissimilarly. Commenter also cites CAA sections 172, 177, and 179 to argue the EPA did not properly align upwind and downwind obligations. Several commenters believe the EPA should defer implementing good neighbor requirements until downwind receptor areas have first implemented their own emissions control strategies.

Response: The EPA maintains that 2023 is an appropriate analytic year and comports with the relevant caselaw. Section VI.A further discusses the compliance schedule for emissions reductions under this rule. Commenter misreads the North Carolina, Wisconsin, and Maryland decisions as calling for good neighbor analysis and emissions controls to be aligned with the timing of the implementation of nonattainment controls by downwind states. However, the D.C. Circuit has held that the statutory attainment dates are the relevant downwind deadlines the EPA must align with in implementing the good neighbor provision. In Wisconsin, the court held, “In sum, under our decision in North Carolina, the Good Neighbor Provision calls for elimination of upwind States' significant contributions on par with the relevant downwind attainment deadlines.” Wisconsin, 938 F.3d. at 321 (emphasis added).

After that decision, the EPA interpreted Wisconsin as limited to the attainment dates for Moderate or higher classifications under CAA section 181 on the basis that Marginal nonattainment areas have reduced planning requirements and other considerations. See, e.g.,85 FR 29882, 29888–89 (May 19, 2020) (proposed approval of South Dakota's 2015 ozone NAAQS good neighbor SIP). However, on May 19, 2020, the D.C. Circuit in Maryland v. EPA, 958 F.3d 1185 (D.C. Cir. 2020), applying the Wisconsin decision, rejected that argument and held that the EPA must assess air quality at the next downwind attainment date, including Marginal area attainment dates under CAA section 181, in evaluating the basis for the EPA's denial of a petition under CAA section 126(b). 958 F.3d at 1203–04. After Maryland, the EPA acknowledged that the Marginal attainment date is the first attainment date to consider in evaluating good neighbor obligations. See, e.g.,85 FR 67653, 67654 (Oct. 26, 2020) (final approval of South Dakota's 2015 ozone NAAQS good neighbor SIP).

The D.C. Circuit again had occasion to revisit the Agency's interpretation of North Carolina, Wisconsin, and Maryland, in a challenge to the Revised CSAPR Update brought by the Midwest Ozone Group (MOG). The court declined to entertain similar arguments to those presented by commenters here and instead in a footnote explained that it had “exhaustively summarized the regulatory framework governing EPA's conduct” and that it “[drew] on those decisions and incorporate them herein by reference,” citing, among other cases, Maryland, 958 F.3d 1185, and New York, 781 F. App'x 4. MOG v. EPA, No. 21–1146 (D.C. Cir. March 3, 2023), Slip Op. at 3 n.1.

The relevance of CAA sections 172, 177, and 179 to the selection of the analytic year in this action is not clear. Commenter cites these provisions to conclude that the EPA did not appropriately consider downwind attainment deadlines and the timing of upwind good neighbor obligations. These provisions are found in subpart I, and while they may have continuing relevance or applicability to aspects of ozone nonattainment planning requirements, the nonattainment dates for the 2015 ozone NAAQS flow from subpart 2 of title I of the CAA, and specifically CAA section 181(a). Applying that statutory schedule to the designations for the 2015 ozone NAAQS, the EPA has promulgated the applicable attainment dates in its regulations at 40 CFR 51.1303. The effective date of the initial designations for the 2015 ozone NAAQS was August 3, 2018 (83 FR 25776, June 4, 2018, effective August 3, 2018). Thus, the first deadline for attainment planning under the 2015 ozone NAAQS was the Marginal attainment date of August 3, 2021, and the second deadline for attainment planning is the Moderate attainment date of August 3, 2024. If a Marginal area fails to attain by the attainment date it is reclassified, or “bumped up,” to Moderate. Indeed, the EPA has just completed a rulemaking action reclassifying many areas of the country from Marginal to Moderate nonattainment, including all of the areas where downwind receptors have been identified in our 2023 modeling as well as many other areas of the country. 87 FR 60897, 60899 (Oct. 7, 2022).

Other than under the narrow circumstances of CAA section 181(a)(5) (discussed further in this section), the EPA is not permitted under the CAA to extend the attainment dates for areas under a given classification. That is, no matter when or if the EPA finalizes a determination that an area failed to attain by its attainment date and reclassifies that area, the attainment date remains fixed, based on the number of years from the area's initial designation. See, e.g., CAA section 182(i) (authorizing the EPA to adjust any applicable deadlines for newly reclassified areas “other than attainment dates”). As the D.C. Circuit has repeatedly made clear, the statutory attainment schedule of the downwind nonattainment areas under subpart 2 is rigorously enforced and is not subject to change based on policy considerations of the EPA or the states.

[T]he attainment deadlines, the Supreme Court has said, are “the heart” of the Act. Train v. Nat. Res. Def. Council, 421 U.S. 60, 66, 95 S.Ct. 1470, 43 L.Ed.2d 731 (1975); see Sierra Club v. EPA, 294 F.3d 155, 161 (D.C. Cir. 2002) (“the attainment deadlines are central to the regulatory scheme”) (alteration and internal quotation marks omitted). The Act's central object is the “attain[ment] [of] air quality of specified standards [within] a specified period of time.” Train, 421 U.S. at 64–65, 95 S.Ct. 1470.

Wisconsin, 938 F.3d at 316. See also Natural Resources Defense Council v. EPA, 777 F.3d 456, 466–68 (D.C. Cir. 2014) (holding the EPA cannot adjust the section 181 attainment schedule to run from any other date than from the date of designation); id. at 468 (“EPA identifies no statutory provision giving it free-form discretion to set Subpart 2 compliance deadlines based on its own policy assessment concerning the number of ozone seasons within which a nonattainment area should be expected to achieve compliance.”) (citing and quoting Whitman v. American Trucking Ass'ns, 531 U.S. 457, 484, (2001) (“The principal distinction between Subpart 1 and Subpart 2 is that the latter eliminates regulatory discretion that the former allowed.”). Furthermore, as the court in NRDC noted, “[T]he `attainment deadlines . . . leave no room for claims of technological or economic infeasibility.' ” 777 F.3d at 488 (quoting Sierra Club, 294 F.3d at 161) (internal quotation marks and brackets omitted).

With the exception of the Uinta Basin, which is not an identified receptor in this action, no Marginal nonattainment area met the conditions of CAA section 181(a)(5) to obtain a one-year extension of the Moderate area attainment date. 87 FR 60899. Thus, all Marginal areas (other than Uinta) that failed to attain have been reclassified to Moderate. Id. (And the New York City Metropolitan nonattainment area was initially classified as Moderate (see following text for further details).) Even if the EPA had extended the attainment date for any of the downwind areas, it is not clear that it would necessarily follow that the EPA must correspondingly extend or delay the implementation of good neighbor obligations. While the Wisconsin court recognized extensions under CAA section 181(a)(5) as a possible source of timing flexibility in implementing the good neighbor provision, 938 F.3d at 320, the EPA and the states are still obligated to implement good neighbor reductions as expeditiously as practicable and are also obligated under the good neighbor provision to address “interference with maintenance.” Areas that have obtained an extension under CAA section 181(a)(5) or which are not designated as in nonattainment could still be identified as struggling to maintain the NAAQS, and the EPA is obligated under the good neighbor provision to eliminate upwind emissions interfering with the ability to maintain the NAAQS, as well. North Carolina, 531 F.3d at 908–11. Thus, while an extension under CAA section 181(a)(5) may be a source of flexibility for the EPA to consider in the timing of implementation of good neighbor obligations, as Wisconsin recognized, it is not the case that the EPA must delay or defer good neighbor obligations for that reason, and neither the D.C. Circuit nor any other court has so held.

Commenter is therefore incorrect to the extent that they argue the selection of 2023 as an analytic year for upwind obligations results in the misalignment of downwind and upwind state obligations. To the contrary, both downwind and upwind state obligations are driven by the statutory attainment date of August 3, 2024 for Moderate areas, and the last year that air quality data may impact whether nonattainment areas are found to have attained by the attainment date is 2023. That is why, in the recent final rulemaking determinations that certain Marginal areas failed to attain by the attainment date, bumping those areas up to Moderate, and giving them SIP submission deadlines, reasonably available control measures (RACM), and reasonably RACT implementation deadlines, the EPA set the attainment SIP submission deadlines for the bumped up Moderate areas to be January 1, 2023. See 87 FR 60897, 60900 (Oct. 7, 2022). The implementation deadline for RACM and RACT is also January 1, 2023. Id. This was in large part driven by the EPA's ozone implementation regulations, 40 CFR 51.1312(a)(3)(i), which previously established a RACT implementation deadline for initially classified Moderate as no later than January 1, 2023, and the modeling and attainment demonstration requirements in 40 CFR 51.1308(d), which require a state to provide for implementation of all control measures needed for attainment no later than the beginning of the attainment year ozone season ( i.e., 2023). Given this regulatory history, the EPA can hardly be accused of letting states with nonattainment areas for the 2015 ozone NAAQS avoid or delay their mandatory CAA obligations.

Commenter's proposal that the EPA align good neighbor obligations with the actual implementation of measures in downwind areas is untethered from the statute, as discussed above. It is also unworkable in practice. It would necessitate coordinating the activities of multiple states and EPA regional and headquarters offices to an impossible degree and effectively could preclude the implementation of good neighbor obligations altogether. Commenter does not explain how the EPA or upwind states should coordinate upwind emissions control obligations for states linked to multiple downwind receptors whose states may be implementing their requirements on different timetables. Less drastic mechanisms than subjecting people living in downwind receptor areas to continuing high levels of air pollution caused in part by upwind-state pollution are available if the actual implementation of mandatory CAA requirements in the downwind areas is delayed: CAA section 304(a)(2) provides for judicial recourse where there is an alleged failure by the Agency to perform a nondiscretionary duty; that recourse is for the Agency to be placed on a court-ordered deadline to address the relevant obligations. See Oklahoma v. U.S. EPA, 723 F.3d 1201, 1223–24 (10th Cir. 2013); Montana Sulphur and Chemical Co. v. U.S. EPA, 666 F.3d 1174, 1190–91 (9th Cir. 2012). Commenter focuses on the EPA's evaluation of New York's Good Neighbor SIP submission to argue the EPA is treating upwind and downwind states dissimilarly. The argument conflates New York's role as both a downwind and an upwind state. In evaluating the Good Neighbor SIP submission that New York submitted, the EPA identified as a basis for disapproval that none of the state emissions control programs New York cited included implementation timeframes to achieve the reductions, let alone ensure they were achieved by 2023. 87 FR 9484, 9494 (Feb. 22, 2022). The EPA conducted the same inquiry into other states' claims regarding their existing or proposed state laws or other emissions reductions claimed in their SIP submissions. See, e.g.,87 FR 9472–73 (evaluating claims regarding emissions reductions anticipated under Maryland's state law); 87 FR 9854 (evaluating claims regarding emissions reductions anticipated under Illinois' state law). Consistent with its treatment of the other upwind states included in this action, the EPA in a separate action disapproved New York's good neighbor SIP submission for the 2015 ozone NAAQS because its arguments did not demonstrate that it had fully prohibited emissions significantly contributing to out of state nonattainment or maintenance problems.

Commenter attempts to contrast this evaluation with what it believes is the EPA's permissive attitude toward delays by downwind states, specifically claiming that “certain nonattainment areas have delayed implementation of nonattainment controls until 2025 and beyond.” This apparently references New York's simple cycle and regenerative combustion turbines (SCCT) controls, which commenter cited elsewhere in its comments. New York's SCCT controls were not included by New York in its good neighbor SIP submission, nor was the prior approval of the SCCT controls reexamined by the EPA or reopened for consideration by the Agency in this action. Although not part of this rulemaking, the EPA notes that the SCCT controls were approved by the EPA as a SIP strengthening measure and not to satisfy any specific planning requirements for the 2015 ozone NAAQS under CAA section 182. 86 FR 43956, 43958 (Aug. 11, 2021). The SCCT controls submitted to the EPA were already a state rule, and the only effect under the CAA of the EPA approving them into New York's SIP was to make them federally enforceable. 86 FR 43956, 43959 (Aug. 11, 2021). In other words, approval of the SCCT controls did not relieve New York of its nonattainment planning obligations for the 2015 ozone NAAQS.

The EPA notes that the New York-Northern New Jersey-Long Island, NY-NJ-CT nonattainment area was initially designated as Moderate nonattainment. 83 FR 25776 (June 4, 2018). Pursuant to this designation, New York was required to submit a RACT SIP submission and an attainment demonstration no later than 24 months and 36 months, respectively, after the effective date of the Moderate designation. CAA section 182; 40 CFR 51.1308(a), 51.1312(a)(2). New York submitted a RACT SIP for the 2015 ozone standards on January 29, 2021, and the EPA is currently evaluating that submission. New York has not yet submitted its attainment demonstration, which was due August 3, 2021. Further, the New York-Northern New Jersey-Long Island, NY-NJ-CT nonattainment area remains subject to the Moderate nonattainment area date of August 3, 2024. If it fails to attain the 2015 ozone NAAQS by August 3, 2024, it will be reclassified to Serious nonattainment, resulting in additional requirements on the New York nonattainment area.

In any case, regardless of the status of New York's and the EPA's efforts in relation to the New York-Northern New Jersey-Long Island, NY-NJ-CT nonattainment area (which are outside the scope of this action), the EPA's evaluation of 2023 as the relevant analytic year in assessing New York's and other states' good neighbor obligations is consistent with the statutory framework and court decisions calling on the agency to align these obligations with the downwind areas' statutory attainment schedule. The EPA further responds to these comments in the RTC document in the docket.

The remainder of this section includes information on (1) the air quality modeling platform used in support of the final rule with a focus on the base year and future year base case emissions inventories, (2) the method for projecting design values in 2023 and 2026, and (3) the approach for calculating ozone contributions from upwind states. The Agency also provides the design values for nonattainment and maintenance receptors and the largest predicted downwind contributions in 2023 and 2026 from each state. The 2016 base period and 2023 and 2026 projected design values and contributions for all ozone monitoring sites are provided in the docket for this rule. The “Air Quality Modeling Technical Support Document for the Federal Good Neighbor Plan for the 2015 Ozone National Ambient Air Quality Standards Final Rulemaking” (Mar. 2023), hereinafter referred to as the Air Quality Modeling Final Rule TSD, in the docket for this final rule contains more detailed information on the air quality modeling aspects of this rule.

B. Overview of Air Quality Modeling Platform

The EPA used version 3 of the 2016-based modeling platform ( i.e., 2016v3) for the air quality modeling for this final rule. This modeling platform includes 2016 base year emissions from anthropogenic and natural sources and anthropogenic emissions projections for 2023 and 2026. The emissions data contained in this platform represent an update to the 2016 version 2 inventories used for the proposal modeling.

The air quality modeling for this final rule was performed for a modeling region ( i.e., modeling domain) that covers the contiguous 48 states using a horizontal resolution of 12 x 12 km. The EPA used the CAMx version 7.10 for air quality modeling which is the same model that EPA used for the proposed rule air quality modeling. Additional information on the 2016-based air quality modeling platform can be found in the Air Quality Modeling Final Rule TSD.

Comment: Commenters noted that the 2016 base year summer maximum daily average 8-hour (MDA8) ozone predictions from the proposal modeling were biased low compared to the corresponding measured concentrations in certain locations. In this regard, commenters said that model performance statistics for a number of monitoring sites, particularly those in portions of the West and in the area around Lake Michigan, were outside the range of published performance criteria for normalized mean bias (NMB) and normalized mean error (NME) of less than ±15 percent and less than 25 percent, respectively (Emory, et al., 2017). The commenters said EPA must investigate the factors contributing to low bias and make necessary corrections to improve model performance in the final rule modeling. Some commenters said that EPA should include NO_X emissions from lightning strikes and assess the treatment of other background sources of ozone to improve model performance for the final rule. Additional information on the comments on model performance can be found in the RTC document for this final rule.

Response: In response to these comments EPA examined the temporal and spatial characteristics of model under prediction to investigate the possible causes of under prediction of MDA8 ozone concentrations in different regions of the U.S. in the proposal modeling. EPA's analysis indicates that the under prediction was most extensive during May and June with less bias during July and August in most regions of the U.S. For example, in the Upper Midwest region model under prediction was larger in May and June compared to July through September. Specifically, in the proposal modeling, the normalized mean bias for days with measured concentrations ≥60 ppb improved from a 21.4 percent under prediction for May and June to a 12.6 percent under prediction in the period July through September. As described in the Air Quality Modeling Final Rule TSD, the seasonal pattern in bias in the Upper Midwest region improves somewhat gradually with time from the middle of May to the latter part of June. In view of the seasonal pattern in bias in the Upper Midwest and in other regions of the U.S., EPA focused its investigation of model performance on model inputs that, by their nature, have the largest temporal variation within the ozone season. These inputs include emissions from biogenic sources and lightning NO_X, and contributions from transport of international anthropogenic emissions and natural sources into the U.S. Both biogenic and lightning NO_X emissions in the U.S. dramatically increase from spring to summer. In contrast, ozone transported into the U.S. from international anthropogenic and natural sources peaks during the period March through June, with lower contributions during July through September. To investigate the impacts of these sources, EPA conducted sensitivity model runs which focused on the effects on model performance of adding NO_X emissions from lightning strikes, updating biogenic emissions, and using an alternative approach for quantifying transport of ozone and precursor pollutants into the U.S. from international anthropogenic and natural sources. The development of lightning NO_X emissions and the updates to biogenic emissions, are described in section IV.C of this document. In the proposal modeling the amount of transport from international anthropogenic and natural sources was based on a simulation of the hemispheric version of the Community Multi-scale Air Quality Model (H–CMAQ) for 2016. The outputs from this hemispheric modeling were then used to provide boundary conditions for national scale air quality modeling at proposal. Overall, H–CMAQ tends to under-predict daytime ozone concentrations at rural and remote monitoring sites across the U.S. during the spring of 2016 whereas the predictions from the GEOS-Chem global model were generally less biased. During the summer of 2016 both models showed varying degrees of over prediction with GEOS-Chem showing somewhat greater over-prediction, compared to H–CMAQ. In view of those results, EPA examined the impacts of using GEOS-Chem as an alternative to H–CMAQ for providing boundary conditions for the final rule modeling.

For the lightning NO_X, biogenics, and GEOS-Chem sensitivity runs, EPA reran the proposal modeling using each of these inputs, individually. Results from these sensitivity runs indicate that each of the three updates provides an improvement in model performance. However, by far the greatest improvement in model performance is attributable to the use of GEOS-Chem. In view of these results EPA has included lightning NO_X emissions, updated biogenic emissions, and international transport from GEOS-Chem in the final rule air quality modeling. Details on the results of the individual sensitivity runs can be found in the Air Quality Modeling Final Rule TSD. For the air quality modeling supporting this final action, model performance based on days in 2016 with measured MDA8 ozone ≥60 ppb is considerably improved ( i.e., less bias and error) compared to the proposal modeling in nearly all regions of the U.S. For example, in the Upper Midwest, which includes monitoring sites along Lake Michigan, the normalized mean bias improved from a 19 percent under prediction to a 6.9 percent under prediction and in the Southwest region, which includes monitoring sites in Denver and Salt Lake City, normalized mean bias improved from a 13.6 percent under prediction to a 4.8 percent under prediction. In all regions, the normalized mean bias and normalized mean error statistics for high ozone days based on the final rule modeling are within the range of performance criteria benchmarks ( i.e., < ±15 percent for normalized mean bias and <25 percent for normalized mean error). Additional information on model performance is provided in the Air Quality Modeling Final Rule TSD. In summary, EPA included emissions of lightning NO_X, as requested by commenters, and investigated and addressed concerns about model performance for the final rule modeling.

C. Emissions Inventories

The EPA developed emissions inventories to support air quality modeling for this final rule, including emissions estimates for EGUs, non-EGU point sources ( i.e., stationary point sources), stationary nonpoint sources, onroad mobile sources, nonroad mobile sources, other mobile sources, wildfires, prescribed fires, and biogenic emissions that are not the direct result of human activities. The EPA's air quality modeling relies on this comprehensive set of emissions inventories because emissions from multiple source categories are needed to model ambient air quality and to facilitate comparison of model outputs with ambient measurements.

Prior to air quality modeling, the emissions inventories were processed into a format that is appropriate for the air quality model to use. To prepare the emissions inventories for air quality modeling, the EPA processed the emissions inventories using the Sparse Matrix Operator Kernel Emissions (SMOKE) Modeling System version 4.9 to produce the gridded, hourly, speciated, model-ready emissions for input to the air quality model. Additional information on the development of the emissions inventories and on data sets used during the emissions modeling process are provided in the document titled, “Technical Support Document (TSD): Preparation of Emissions Inventories for the 2016v3 North American Emissions Modeling Platform” (Jan. 2023), hereafter known as the 2016v3 Emissions Modeling TSD. This TSD is available in the docket for this rule.

1. Foundation Emissions Inventory Data Sets

The 2016v3 emissions platform is comprised of data from various sources including data developed using models, methods, and source datasets that became available in calendar years 2020 through 2022, in addition to data retained from the Inventory Collaborative 2016 version 1 (2016v1) Emissions Modeling Platform, released in October 2019. The 2016v1 platform was developed through a national collaborative effort between the EPA and state and local agencies along with MJOs. The 2016v2 platform used to support the proposed action included updated data from the 2017 NEI along with updates to models and methods as compared to 2016v1. The 2016v3 platform includes updates to the 2016v2 platform implemented in response to comments along with other updates to the 2016v2 platform such as corrections and the incorporation of updated data sources that became available prior to the 2016v3 inventories being developed. Several commenters noted that the 2016v2 platform did not include NO_X emissions that resulted from lightning strikes. To address this, lightning NO_X emissions were computed and included in the 2016v3 platform.

For this final rule, the EPA developed emissions inventories for the base year of 2016 and the projected years of 2023 and 2026. The 2023 and 2026 inventories represent changes in activity data and of predicted emissions reductions from on-the-books actions, planned emissions control installations, and promulgated Federal measures that affect anthropogenic emissions. The 2016 emissions inventories for the U.S. primarily include data derived from the 2017 National Emissions Inventory (2017 NEI) and data specific to the year of 2016. The following sections provide an overview of the construct of the 2016v3 emissions and projections. The fire emissions were unchanged between the 2016v2 and 2016v3 emissions platforms. For the 2016v3 platform, the biogenic emissions were updated to use the latest available versions of the Biogenic Emissions Inventory System and associated land use data to help address comments related to a degradation in model performance in the 2016v2 platform as compared to the 2016v1 platform. Details on the construction of the inventories are available in the 2016v3 Emissions Modeling TSD. Details on how the EPA responded to comments related to emissions inventories are available in the RTC document for this rule.

2. Development of Emissions Inventories for EGUs

a. EGU Emissions Inventories Supporting This Final Rule

Development of emissions inventories for annual NO_X and SO₂ emissions for EGUs in the 2016 base year inventory are based primarily on data from continuous emissions monitoring systems (CEMS) and other monitoring systems allowed for use by qualifying units under 40 CFR part 75, with other EGU pollutants estimated using emissions factors and annual heat input data reported to the EPA. For EGUs not reporting under Part 75, the EPA used data submitted to the NEI by the state, local, and tribal agencies. The Air Emissions Reporting Rule (80 FR 8787; February 19, 2015), requires that Type A point sources large enough to meet or exceed specific thresholds for emissions be reported to the EPA every year, while the smaller Type B point sources must only be reported to EPA every 3 years. Emissions data for EGUs that did not have data submitted to the NEI specific to the year 2016 were filled in with data from the 2017 NEI. For more information on the details of how the 2016 EGU emissions were developed and prepared for air quality modeling, see the 2016v3 Emissions Modeling TSD.

The EPA projected 2023 and 2026 baseline EGU emissions using the version 6—Updated Summer 2021 Reference Case of the Integrated Planning Model (IPM). IPM, developed by ICF Consulting, is a state-of-the-art, peer-reviewed, multi-regional, dynamic, deterministic linear programming model of the contiguous U.S. electric power sector. It provides forecasts of least cost capacity expansion, electricity dispatch, and emissions control strategies while meeting energy demand and environmental, transmission, dispatch, and reliability constraints. The EPA has used IPM for over two decades, including all prior implemented CSAPR rulemakings, to better understand power sector behavior under future business- as-usual conditions and to evaluate the economic and emissions impacts of prospective environmental policies. The model is designed to reflect electricity markets as accurately as possible. The EPA uses the best available information from utilities, industry experts, gas and coal market experts, financial institutions, and government statistics as the basis for the detailed power sector modeling in IPM. The model documentation provides additional information on the assumptions discussed here as well as all other model assumptions and inputs. The EPA relied on the same model platform at final as it did at proposal, but made substantial updates to reflect public comments on near-term fossil fuel market price volatility and updated fleet information reflecting Summer 2022 U.S. Energy Information Agency (EIA) 860 data, unit-level comments, and additional updates to the National Electric Energy Data System (NEEDS) inventory.

The IPM version 6—Updated Summer 2021 Reference Case incorporated recent updates through the Summer of 2022 to account for updated Federal and state environmental regulations (including Renewable Portfolio Standards (RPS), Clean Energy Standards (CES) and other state mandates), fleet changes (committed EGU retirements and new builds), electricity demand, technology cost and performance assumptions from recent data (for renewables adopting from National Renewable Energy Lab (NREL's) Annual Technology Baseline 2020 and for fossil sources from EIA's Annual Energy Outlook (AEO) 2020. Natural gas and coal price projections reflect data developed in Fall 2020 but updated in summer of 2022 to capture near-term price volatility and current market conditions. The inventory of EGUs provided as an input to the model was the NEEDS fall 2022 version and is available on EPA's website. This version of NEEDS reflects announced retirements and under-construction new builds known as of early summer 2022. This projected base case accounts for the effects of the finalized Mercury and Air Toxics Standards rule, CSAPR, the CSAPR Update, the Revised CSAPR Update, NSR enforcement settlements, the final ELG Rule, CCR Rule, and other on-the-books Federal and state rules (including renewable energy tax credit extensions from the Consolidated Appropriations Act of 2021) through early 2021 impacting SO₂, NO_X, directly emitted particulate matter, CO₂, and power plant operations. It also includes final actions the EPA has taken to implement the Regional Haze Rule and best available retrofit technology (BART) requirements. Documentation of IPM version 6 and NEEDS, along with updates, is in Docket ID No. EPA–HQ–OAR–2021–0668 and available online at https://www.epa.gov/airmarkets/power-sector-modeling . IPM has projected output years for 2023 and 2025. IPM year 2025 outputs were adjusted for known retirements to be reflective of year 2026, and IPM year 2030 outputs were used for the year 2032 as is specified by the mapping of IPM output years to specific years.

Additional 2023 through 2026 EGU emissions baseline levels were developed through engineering analytics as an alternative approach that did not involve IPM. The EPA developed this inventory for use in Step 3 of this final rule, where it determines emissions reduction potential and corresponding state-level emissions budgets. IPM includes optimization and perfect foresight in solving for least cost dispatch. Given that this final rule will likely become effective immediately prior to the start of the 2023 ozone season, the EPA adopted a similar approach to the CSAPR Update and the Revised CSAPR Update where it utilized historical data and an engineering analytics approach in Step 3 to avoid overstating optimization and dispatch decisions in state-emissions budget quantification that may not be possible in a short time frame. The EPA does this by starting with unit-level reported data and only making adjustments to reflect known baseline changes such as planned retirements and new builds (for the base case scenarios) and also identified mitigation strategies for determining state emissions budgets. In both the CSAPR Update and in this rule at Step 3, the EPA complemented that projected IPM EGU outlook with an historical ( e.g., engineering analytics) perspective based on historical data that only factors in known changes to the fleet. This 2023 engineering analytics data set is described in more detail in the Ozone Transport Policy Analysis Final Rule TSD and corresponding Appendix A: State Emissions Budgets Calculations and Underlying Data. The Engineering Analysis used in Step 3 is also discussed further in section VII.B of this document.

Both IPM and the Engineering Analytics tools are valuable for estimating future EGU emissions and examining the cone of uncertainty around any future sector-level inventory estimate. A key difference between the two tools is that IPM reflects both announced and projected changes in fleet operation, whereas the Engineering Analytics tool only reflects announced changes. By not including projected regional changes that are anticipated in response to market forces and fleet trends, the Engineering Analysis deliberately creates future estimates of the power sector where state estimates are limited to known changes. Throughout all of the CSAPR rules to date, and prior interstate transport actions, the EPA has used IPM at Steps 1 and 2 as it is best suited for projecting emissions in an airshed, at projecting emissions for time horizons more than a few years out (for which changes would not yet be announced and thus projecting changes is critical), and for scenarios where the assumed change in emissions is not being codified into a state emissions reduction requirement. Using IPM at Steps 1 and 2 helps the EPA avoid overstating the current analytic year receptor values (Step 1) and future year linkages (Step 2) by reflecting reductions anticipated to occur within the airshed in the relevant timeframe.

Engineering analytics has been a useful tool for Step 3 state-level emissions reduction estimates in CSAPR rulemaking, because at that step the EPA is dealing with more geographic granularity (state-level as opposed to regional air shed), more near-term (as opposed to medium-term) assessments, and scenarios where reduction estimates are codified into regulatory requirements. Using the Engineering Analytics tool at this step ensures that the EPA is not codifying into the base case, and consequently into state emissions budgets, changes in the power sector that are merely modeled to occur rather than announced by real-world actors.

Finally, both in the Revised CSAPR Update and in this rule, the EPA was able to use the Air Quality Assessment Tool to determine that regardless of which EGU inventory is used, the 2023 geography of the program is not impacted. In other words, regardless of whether a stakeholder takes a more comprehensive view of the EGU future (IPM) or one limited to current data and known changes (Engineering Analysis), the states that are linked to receptors at Steps 1 and 2 would be the same. This finding is consistent with the observation that EGUs are now less than 10 percent of the total ozone-season NO_X inventory and the degree of near-term difference between the IPM and Engineering Analytic regional projections is relatively small on the regional level. The EPA continues to believe that IPM is best suited for Step 1 and Step 2, and engineering analytics is best suited for Step 3 efforts in this rulemaking. The Ozone Transport Policy Analysis Final Rule TSD contains data on 2023 and 2026 AQ impacts of each dataset.

Comment: Some commenters express concern that using IPM for Step 1 and Step 2 captures generation shifting across state lines, which exceeds the EPA's authority. Moreover, the commenters suggest that the resulting proposed baseline EGU inventory may understate emissions levels as it projects economic retirements that are not yet announced or firm. Other commenters more generally allege that the EPA is using different modeling tools at different steps in its analysis, and this introduces confusion or uncertainty into the basis for the EPA's regulatory conclusions.

Response: The EPA believes the first aspect of this comment, in regards to its focus on generation shifting, is misguided in several ways. For Step 1 and Step 2, the EPA models no incremental generation shifting attributable to the implementation of an emissions control policy at Step 3. Rather, any generation patterns are merely a reflection of the model's projection of how regional load requirements will be met with the generation sources serving that region in the baseline. The EPA is not modeling any additional generation shifting, but merely capturing the expected generation dispatch under anticipated baseline market conditions. Electricity generated in one state regularly is transmitted across state boundaries and is used to serve load in other states; IPM is not incentivizing or requiring any additional generation transfer across state lines in this scenario but is merely projecting the pattern of this behavior in the future. Moreover, as noted previously, the EPA affirms its geographic findings at Step 2 (states contributing over 1 percent of the NAAQS to a downwind receptor) using historical data (engineering analysis) in a sensitivity analysis. These historical data reflect the actual generation patterns observed to meet regional load. Therefore, any suggestion by the commenter that the EPA's projected view of baseline grid dispatch is unreasonable, is mooted by the fact that the use of historical reported generation patterns produces the same result. Additionally, at the time of the proposal's analysis, the 2023 ozone season was still nearly two years away. Therefore, it was appropriate for EPA's modeling to project economic retirements as those retirements—which are regularly occurring—are often not firm or announced two years in advance. However, for this final rule, the 2023 analytic year was close enough to the period in which EPA was conducting its analysis that such retirements would likely be announced. Therefore, the EPA was able to incorporate those announced and firm retirements to occur in the 2023 year. Further, in recognition of this very near timeframe, we deactivated IPM's ability to project additional economic retirements for the 2023 year (reflecting the notion that any retirements occurring by 2023 would be known at this point). This adjustment further accommodates the commenters' concern that the baseline overstates generation shifting (driven by retirements) in the near term, and consequently understates emissions levels. Finally, with respect to comments that the EPA is using different modeling tools at different steps in the framework, we previously explained why these techniques are appropriate for the purposes at each step of the analysis, and they are not incompatible nor do they produce results so different as to call into question their reliability or the bases for our regulatory determinations (EPA notes that the nationwide projected ozone season total NO_X emissions vary by less than 1 percent in the 2023 analytic year). Nonetheless, we also observe that the effect of using engineering analytics to inform analysis at Steps 1 and 2 would tend to produce higher assumed emissions from EGUs in the baseline than IPM would project in 2026 and beyond and therefore only strengthen and further affirm the Step 1 and Step 2 geographic findings. EPA's use of different tools to project EGU scenarios is not inconsistent, but rather it is carefully explained as a deliberate measure taken to preserve—not introduce—consistency across each of the Steps in the 4-step framework. By using IPM at Step 1 and 2, EPA is selecting the more conservative approach for identifying the degree of nonattainment and geography of states contributing above 1 percent. By using Engineering Analytics at Step 3, EPA is selecting the more conservative value to codify into state-level budgets.

b. Impact of the Inflation Reduction Act on EGU Emissions

The EGU modeling used to construct the EGU emissions inventories used to inform the modeling projections for 2023 and 2026 was conducted prior to the passage of the Inflation Reduction Act (IRA), Public Law 117–169. The EPA did not have time to incorporate updated EGU projections reflecting the passage of the IRA into the primary air quality modeling for this final rule. However, the EPA was able to perform a sensitivity analysis reflecting the IRA in its EGU NO_X emissions inventories. The results from this scenario were run through AQAT and demonstrated that the status of states identified as linked at the 1 percent of NAAQS contribution threshold (based on the modeling and air quality analysis described in this section) would not change regardless of which inventory (with or without IRA) is used. This sensitivity analysis is presented in the Regulatory Impact Analysis accompanying this rule, and that discussion provides additional detail on the emissions consequences of including the IRA in a baseline EGU inventory. The air quality impact of including the IRA in EPA's emissions inventories and in its Step 3 scenarios is discussed in Appendix K of the Ozone Transport Policy Analysis Final Rule TSD.

The results of this analysis are not surprising and accord with what is generally understood to be the overall effect of the IRA over the short to long term. While the IRA is anticipated to have a potentially dramatic effect on reducing both GHG and conventional pollutant emissions from the power sector, it is likely to have a more substantial impact later in the forecast period ( i.e., beyond the attainment deadlines by which the emissions reductions under this final rule must occur). This timing reflects a realistic assessment of utilities', regulators', and transmission authorities' planning requirements associated with the addition of substantial new renewable and storage capacity to the grid, as well as the time needed to integrate that capacity and retire existing capacity. Additionally, the IRA incentives span a longer time period (for example, certain tax incentives for clean energy sources are available until the later of 2032 or the year in which power sector emissions are 75 percent below 2022 levels) and therefore there is no IRA-related deadline to build cleaner generation by 2026. Recent analysis by the Congressional Budget Office supports the finding that the majority of power sector EGU emissions reductions expected from the IRA occur well after the 2023 and 2026 analytic years relevant to the attainment dates and this rulemaking. While the report focuses on CO₂ rather than NO_X, the drivers of the emissions reductions (primarily increased zero-emitting generation) would generally have a downward impact on both pollutants.

We note that important uncertainties remain at this time in the implementation of the IRA that further counsel against over-assuming short-term emissions reductions for purposes of this rule. The legislation provides economic incentives for shifting to cleaner forms of power generation but does not mandate emissions reductions through an enforceable regulatory program. The strength of those incentives will vary to some extent depending on other key market factors (such as the cost of natural gas or renewable energy technologies). Further, some incentives, such as tax credits for carbon capture and storage, could lead EGUs to remain in operation longer, which could in turn result in greater NO_X emissions, if those emissions are not also well controlled.

Nonetheless, while we find that the passage of the IRA does not affect the geography of the rule in terms of which states we identify as linked, the Agency is confident that the incentives toward clean technology provided in the IRA will, in the longer run beyond the 2015 ozone NAAQS attainment deadlines, facilitate ongoing EGU compliance with the emissions reduction requirements of this rule and will reduce costs borne by EGUs and their customers as the U.S. power sector transitions. As discussed in greater detail in section VI.B of this document, we have made several adjustments in the final rule to provide greater flexibility to EGU owners and operators to integrate this rule's requirements with and facilitate the accelerating transition to an overall cleaner electricity-generating sector, which the IRA represents. Despite the uncertainties inherent in the implementation of the IRA at this time, the EPA also has performed a sensitivity analysis on the final rule to confirm that our finding of no overcontrol is robust to a future with the IRA in effect.

3. Development of Emissions Inventories for Stationary Industrial Point Sources

Non-EGU point source emissions are mostly consistent with those in the proposal modeling except where they were updated in response to comments. Several commenters mentioned that point source emissions carried forward from 2014 NEI were not the best estimates of 2017 emissions. Thus, emissions sources in 2016v2 that had been projected from the 2014 NEI in the proposal were replaced with emissions based on the 2017 NEI. Point source emissions submitted to the 2016 NEI or to the 2016v1 platform development process specifically for the year 2016 were retained in 2016v3. Other 2016 non-EGU updates in 2016v3 include a few sources being moved to the EGU inventory, the addition of some control efficiency information for the year 2016, the replacement of most emissions projected from 2014 NEI with data from 2017 NEI, and the inclusion of point source data for solvent processes that had not been included in the 2016v2 non-EGU inventory.

The 2023 and 2026 non-EGU point source emissions were grown from 2016 to those years using factors based on the AEO 2022 and reflect emissions reductions due to known national and local rules, control programs, plant closures, consent decrees, and settlements that could be computed as reductions to specific units by July 2022.

Aircraft emissions and ground support equipment at airports are represented as point sources and are based on adjustments to emissions in the January 2021 version of the 2017 NEI. The EPA developed and applied factors to adjust the 2017 airport emissions to 2016, 2023 and 2026 based on activity growth projected by the Federal Aviation Administration Terminal Area Forecast 2021 data, the latest available version at the time the factors were developed. By basing the factors on the latest available Terminal Area Forecast that was released following the most significant pandemic impacts on the aviation sector, the reduction and rebound impacts of the pandemic on aircraft and ground support equipment were reflected in the 2023 and 2026 airport emissions.

Emissions at rail yards were represented as point sources. The 2016 rail yard emissions are largely consistent with the 2017 NEI rail yard emissions. The 2016 and 2023 rail yard emissions were developed through the 2016v1 Inventory Collaborative process, with the 2026 emissions interpolated between the 2023 and 2028 emissions from 2016v1 rail yard emissions were interpolated from the 2016 and 2023 emissions. Class I rail yard emissions were projected based on the AEO freight rail energy use growth rate projections for 2023, and 2026 with the fleet mix assumed to be constant throughout the period.

The EPA made multiple updates to point source oil and gas emissions in response to comments. For the final rule, the point source oil and gas emissions for 2016 were based on the 2016v2 point inventory except that most 2014 NEI-based emissions were replaced with 2017 NEI emissions. Additionally, in response to comments, state-provided emissions equivalent to those in the 2016v1 platform were used for Colorado, and some New Mexico emissions were replaced with data backcast from 2020 to 2016. To develop inventories for 2023 and 2026 for the final rule, the year 2016 oil and gas point source inventories were first projected to 2021 values based on actual historical production data, then those 2021 emissions were projected to 2023 and 2026 using regional projection factors based on AEO 2022 projections. This was an update from the proposal approach that used actual data only through the year 2019, because 2021 data were not yet available. NO_X and VOC reductions resulting from co-benefits of NSPS for Stationary Reciprocating Internal Combustion Engines (RICE) are reflected, along with Natural Gas Turbine and Process Heater NSPS NO_X controls and Oil and Gas NSPS VOC controls. In some cases, year 2019 point source inventory data were used instead of the projected future year emissions except for the Western Regional Air Partnership (WRAP) states of Colorado, New Mexico, Montana, Wyoming, Utah, North Dakota, and South Dakota. The WRAP future year inventory was used in these WRAP states in all future years except in New Mexico where the WRAP base year emissions were projected using the EIA historical and AEO forecasted production data. Estimated impacts from the New Mexico Administrative code 20.2.50 were also included.

4. Development of Emissions Inventories for Onroad Mobile Sources

Onroad mobile sources include exhaust, evaporative, and brake and tire wear emissions from vehicles that drive on roads, parked vehicles, and vehicle refueling. Emissions from vehicles using regular gasoline, high ethanol gasoline, diesel fuel, and electric vehicles were represented, along with buses that used compressed natural gas. The EPA developed the onroad mobile source emissions for states other than California using the EPA's Motor Vehicle Emissions Simulator (MOVES). MOVES3 was released in November 2020 and has been followed by some minor releases that improved the usage of the model but that do not have substantive impacts on the emissions estimates. For the proposal, MOVES3 was run using inputs provided by state and local agencies through the 2017 NEI where available, in combination with nationally available data sets to develop a complete inventory. Onroad emissions were developed based on emissions factors output from MOVES3 runs for the year 2016, coupled with activity data ( e.g., vehicle miles traveled and vehicle populations) representing the year 2016. The 2016 activity data were provided by some state and local agencies through the 2016v1 process, and the remaining activity data were derived from those used to develop the 2017 NEI. The onroad emissions were computed within SMOKE by multiplying emissions factors developed using MOVES with the appropriate activity data. Prior to computing the final rule emissions, updates to some onroad inputs were made in response to comments and to implement corrections. Onroad mobile source emissions for California were consistent with the updated emissions data provided by the state for the final rule.

The 2023 and 2026 onroad emissions reflect projected changes to fuel properties and usage, along with the impact of the rules included in MOVES3 for each of those years. MOVES emissions factors for the years 2023 and 2026 were used. A comprehensive list of control programs included for onroad mobile sources is available in the 2016v3 Emissions Modeling TSD. Year 2023 and 2026 activity data for onroad mobile sources were provided by some state and local agencies, and otherwise were projected to 2023 and 2026 by first projecting the 2016 activity to year 2019 based on county level vehicle miles traveled (VMT) from the Federal Highway Administration. Because VMT for onroad mobile sources were substantially impacted by the pandemic and took about two years to rebound to pre-pandemic levels, in the 2016v3 platform no growth in VMT was implemented from 2019 to. The estimated 2021 VMT were then grown from 2021 to 2023 and 2026 using AEO 2022-based factors. Recent updates to inspection and maintenance programs in North Carolina and Tennessee were reflected in the MOVES inputs for the final rule modeling. The 2023 and 2026 onroad mobile emissions were computed within SMOKE by multiplying the respective emissions factors developed using MOVES with the year-specific activity data. Prior to computing the final rule emissions for 2023, the EPA made updates to some onroad inputs in response to comments and to implement corrections.

5. Development of Emissions Inventories for Commercial Marine Vessels

The commercial marine vessel (CMV) emissions in the 2016 base case emissions inventory for this rule were based on those in the 2017 NEI. Factors were applied to adjust the 2017 NEI emissions backward to represent emissions for the year 2016. The CMV emissions reflect reductions associated with the Emissions Control Area proposal to the International Maritime Organization control strategy (EPA–420–F–10–041, August 2010); reductions of NO_X, VOC, and CO emissions for new category 3 (C3) engines that went into effect in 2011; and fuel sulfur limits that went into effect prior to 2016. The cumulative impacts of these rules through 2023 and 2026 were incorporated into the projected emissions for CMV sources. The CMV emissions were split into emissions inventories from the larger C3 engines, and those from the smaller category 1 and 2 (C1C2) engines. CMV emissions in California are based on emissions provided by the state. The CMV emissions are consistent with the emissions for the 2016v1 platform updated CMV emissions released by February 2020 although they include projected emissions for the years of 2023 and 2026 instead of 2023 and 2028. In addition, in response to comments, the EPA implemented an improved process for spatial allocating CMV emissions along state and county boundaries.

6. Development of Emissions Inventories for Other Nonroad Mobile Sources

The EPA developed nonroad mobile source emissions inventories (other than CMV, locomotive, and aircraft emissions) for 2016, 2023, and 2026 from monthly, county, and process level emissions output from MOVES3. Types of nonroad equipment include recreational vehicles, pleasure craft, and construction, agricultural, mining, and lawn and garden equipment. State-submitted emissions data for nonroad sources were used for California. The nonroad emissions for the final rule were unchanged from those at the proposal. The nonroad mobile emissions control programs include reductions to locomotives, diesel engines, and recreational marine engines, along with standards for fuel sulfur content and evaporative emissions. A comprehensive list of control programs included for mobile sources is available in the 2016v3 Emissions Modeling TSD.

Line haul locomotives are also considered a type of nonroad mobile source but the emissions inventories for locomotives were not developed using MOVES3. Year 2016 locomotive emissions were developed through the 2016v1 collaborative process and the year 2016 emissions are mostly consistent with those in the 2017 NEI. More information on the development of the Class I, Class II and III, and commuter rail line haul locomotive emissions is available in the 2016v3 Emissions Modeling TSD. The projected locomotive emissions for 2023 and 2026 were developed by applying factors to the 2016 emissions using activity data based on AEO freight rail energy use growth rate projections along with emissions rates adjusted to account for recent historical trends. The emission factors used for NO_X, PM10 and VOC for line haul locomotives in the analytic years were derived from trend lines based on historic line-haul emission factors from the period of 2007 through 2017 and extrapolated to 2023 and 2026.

7. Development of Emissions Inventories for Nonpoint Sources

For stationary nonpoint sources, some emissions in the 2016 base case emissions inventory come directly from the 2017 NEI, others were adjusted from the 2017 NEI to represent 2016 levels, and the remaining emissions including those from oil and gas, fertilizer, and solvents were computed specifically to represent 2016. Stationary nonpoint sources include evaporative sources, consumer products, fuel combustion that is not captured by point sources, agricultural livestock, agricultural fertilizer, residential wood combustion, fugitive dust, and oil and gas sources. The emissions sources derived from the 2017 NEI include agricultural livestock, fugitive dust, residential wood combustion, waste disposal (including composting), bulk gasoline terminals, and miscellaneous non-industrial sources such as cremation, hospitals, lamp breakage, and automotive repair shops. A recent method to compute solvent VOC emissions was used.

Where comments were provided about projected control measures or changes in nonpoint source emissions, those inputs were first reviewed by the EPA. Those found to be based on reasonable data for affected emissions sources were incorporated into the projected inventories for 2023 and 2026 to the extent possible. Where possible, projection factors based on the AEO used data from AEO 2022, the most recent AEO at the time available at the time the inventories were developed. Federal regulations that impact the nonpoint sources were reflected in the inventories. Adjustments for state fuel sulfur content rules for fuel oil in the Northeast were included along with solvent controls applicable within the ozone transport region. Details are available in the 2016v3 Emissions Modeling TSD.

Nonpoint oil and gas emissions inventories for many states were developed based on outputs from the 2017 NEI version of the EPA Oil and Gas Tool using activity data for year 2016. Production-related emissions data from the 2017 NEI were used for Oklahoma, 2016v1 emissions were used for Colorado and for Texas production-related sources to response to comments. Data for production-related nonpoint oil and gas emissions in the states of Colorado, Montana, New Mexico, North Dakota, South Dakota, Utah, and Wyoming were obtained from the WRAP baseline inventory. A California Air Resources Board-provided inventory was used for 2016 oil and gas emissions in California. Nonpoint oil and gas inventories for 2023 and 2026 were developed by first projecting the 2016 oil and gas inventories to 2021 values based on actual production data. Next, those 2021 emissions were projected to 2023 and 2026 using regional projection factors by product type based on AEO 2022 projections. A 2017–2019 average inventory was used for oil and natural gas exploration emissions in 2023 and 2026 except for California and in the WRAP states in which data from the WRAP future year inventory were used. NO_X and VOC reductions that are co-benefits to the NSPS for RICE are reflected, along with Natural Gas Turbines and Process Heaters NSPS NO_X controls and NSPS Oil and Gas VOC controls. The WRAP future year inventory was used for oil and natural gas production sources in 2023 and 2026 except in New Mexico where the WRAP Base year emissions were projected using the EIA historical and AEO forecasted production data. Estimated impacts from the New Mexico Administrative Code 20.2.50 were included.

D. Air Quality Modeling To Identify Nonattainment and Maintenance Receptors

In this section, the Agency describes the air quality modeling and analyses performed in Step 1 to identify locations where the Agency expects there to be nonattainment or maintenance receptors for the 2015 ozone NAAQS in the 2023 and 2026 analytic years. Where the EPA's analysis shows that an area or site does not fall under the definition of a nonattainment or maintenance receptor in these analytic years, that site is excluded from further analysis under this rule.

In the proposed rule, the EPA applied the same approach used in the CSAPR Update and the Revised CSAPR Update to identify nonattainment and maintenance receptors for the 2008 ozone NAAQS. See86 FR 23078–79. The EPA's approach gives independent effect to both the “contribute significantly to nonattainment” and the “interfere with maintenance” prongs of section 110(a)(2)(D)(i)(I), consistent with the D.C. Circuit's direction in North Carolina. Further, in its decision on the remand of the CSAPR from the Supreme Court in the EME Homer City case, the D.C. Circuit confirmed that EPA's approach to identifying maintenance receptors in the CSAPR comported with the court's prior instruction to give independent meaning to the “interfere with maintenance” prong in the good neighbor provision. EME Homer City II, 795 F.3d at 136.

In the CSAPR Update and the Revised CSAPR Update, the EPA identified nonattainment receptors as those monitoring sites that are projected to have average design values that exceed the NAAQS and that are also measuring nonattainment based on the most recent monitored design values. This approach is consistent with prior transport rulemakings, such as the NO_X SIP Call and CAIR, where the EPA defined nonattainment receptors as those areas that both currently monitor nonattainment and that the EPA projects will be in nonattainment in the future compliance year.

The Agency explained in the NO_X SIP Call and CAIR and then reaffirmed in the CSAPR Update that the EPA has the most confidence in our projections of nonattainment for those monitoring sites that also measure nonattainment for the most recent period of available ambient data. The EPA separately identified maintenance receptors as those monitoring sites that would have difficulty maintaining the relevant NAAQS in a scenario that accounts for historical variability in air quality at that site. The variability in air quality was determined by evaluating the “maximum” future design value at each monitoring site based on a projection of the maximum measured design value over the relevant period. The EPA interprets the projected maximum future design value to be a potential future air quality outcome consistent with the meteorology that yielded maximum measured concentrations in the ambient data set analyzed for that receptor ( i.e., ozone conducive meteorology). The EPA also recognizes that previously experienced meteorological conditions ( e.g., dominant wind direction, temperatures, and air mass patterns) promoting ozone formation that led to maximum concentrations in the measured data may reoccur in the future. The maximum design value gives a reasonable projection of future air quality at the receptor under a scenario in which such conditions do, in fact, reoccur. The projected maximum design value is used to identify upwind emissions that, under those circumstances, could interfere with the downwind area's ability to maintain the NAAQS.

Therefore, applying this methodology in this rule, the EPA assessed the magnitude of the projected maximum design values for 2023 and 2026 at each monitoring site in relation to the 2015 ozone NAAQS and, where such a value exceeds the NAAQS, the EPA determined that receptor to be a “maintenance” receptor for purposes of defining interference with maintenance, consistent with the method used in CSAPR and upheld by the D.C. Circuit in EME Homer City II. That is, monitoring sites with a maximum design value that exceeds the NAAQS are projected to have maintenance problems in the future analytic years.

Recognizing that nonattainment receptors are also, by definition, maintenance receptors, the EPA often uses the term “maintenance-only” to refer to receptors that are not also nonattainment receptors. Consistent with the concepts for maintenance receptors, as described previously, the EPA identifies “maintenance-only” receptors as those monitoring sites that have projected average design values above the level of the applicable NAAQS, but that are not currently measuring nonattainment based on the most recent official design values. In addition, those monitoring sites with projected average design values below the NAAQS, but with projected maximum design values above the NAAQS are also identified as “maintenance only” receptors, even if they are currently measuring nonattainment based on the most recent official design values.

Comment: The EPA received comments claiming that the projected design values for 2023 were biased low compared to recent measured data. Commenters noted that a number of monitoring sites that are projected to be below the NAAQS in 2023 based on the EPA's modeling for the proposed action are currently measuring nonattainment based on data from 2020 and 2021. One commenter requested that the EPA determine whether its past modeling tends to overestimate or underestimated actual observed design values. If EPA finds that the agency's model tends to underestimate future year design values, the commenter requests that EPA re-run its ozone modeling, incorporating parameters that account for this tendency.

Response: In response to comments, the EPA compared the projected 2023 design values based on the proposal modeling to recent trends in measured data. As a result of this analysis, the EPA agrees that current data indicate that there are monitoring sites at risk of continued nonattainment in 2023 even though the model projected average and maximum design values at these sites are below the NAAQS ( i.e., sites that are not modeling-based receptors). It would not be reasonable to ignore recent measured ozone levels in many areas that are clearly not fully consistent with certain concentrations in the Step 1 analysis for 2023. Therefore, the EPA has also developed an additional maintenance-only receptor category, which includes what we refer to as “violating monitor” receptors, based on current ozone concentrations measured by regulatory ambient air quality monitoring sites.

Specifically, the EPA has identified monitoring sites with measured 2021 and preliminary 2022 design values and 4th high maximum daily 8-hour average (MDA8) ozone in both 2021 and 2022 (preliminary data) that exceed the NAAQS, although projected to be in attainment in 2023, as having the greatest risk of continuing to have a problem attaining the standard in 2023. These criteria sufficiently consider measured air quality data so as to avoid including monitoring sites that have measured nonattainment data in recent years but could reasonably be anticipated to not have a nonattainment or maintenance problem in 2023, in line with our modeling results. Our methodology is intended only to identify those sites that have sufficiently poor ozone levels that there is clearly a reasonable expectation that an ozone nonattainment or maintenance problem will persist in the 2023 ozone season. Moreover, 2023 is so near in time that recent measured ozone levels can be used to reasonably project whether an air quality problem is likely to persist. We view this approach to identifying additional receptors in 2023 as the best means of responding to the comments on this issue in this action, while also identifying all transport receptors.

For purposes of this action, we treat these violating monitors as an additional type of maintenance-only receptor. Because our modeling did not identify these sites as receptors, we do not believe it is sufficiently certain that these sites will be in nonattainment such that they should be considered nonattainment receptors. Rather, our authority for treating these sites as receptors in 2023 flows from the responsibility in CAA section 110(a)(2)(i)(I) to prohibit emissions that interfere with maintenance of the NAAQS. See, e.g., North Carolina, 531 F.3d at 910–11 (failing to give effect to the interfere with maintenance clause “provides no protection for downwind areas that, despite EPA's predictions, still find themselves struggling to meet NAAQS due to upwind interference . . . .”) (emphasis added). Recognizing that no modeling can perfectly forecast the future, and “a degree of imprecision is inevitable in tackling the problem of interstate air pollution,” this approach in the Agency's judgement best balances the need to avoid both “under-control” and “overcontrol,” EME Homer City, 572 U.S. at 523.

We acknowledge that the traditional modeling plus monitoring methodology we used at proposal and in prior ozone transport rules would otherwise have identified such sites as being in attainment in 2023. Despite the implications of the current measured data suggesting there will be a nonattainment problem at these sites in 2023, we cannot definitively establish that such sites will be in nonattainment in 2023 in light of our modeling projections. In the face of this uncertainty, we regard our ability to consider such sites as receptors for purposes of good neighbor analysis under CAA section 110(a)(2)(D)(i)(I) to be a function of the requirement to prohibit emissions that interfere with maintenance of the NAAQS; even if an area may be technically in attainment, we have reliable information indicating that there is an identified risk that attainment will not in fact be achieved.

However, because we did not identify this basis for receptor-identification at proposal, in this final action we are only using this receptor category on a confirmatory basis. That is, for states that we find linked based on our traditional modeling-based methodology in 2023, we find in this final analysis that the linkage at Step 2 is strengthened and confirmed if that state is also linked to one or more “violating monitor” receptors. If a state is only linked to a violating-monitor receptor in this final analysis, we are deferring taking final action on that state's SIP submittal. This is the case for the State of Tennessee. Among the states that previously had their transport SIPs fully approved for the 2015 ozone NAAQS, the EPA has also identified a linkage to violating-monitor receptors for the State of Kansas. The EPA intends to further review its air quality modeling results and recent measured ozone levels, and we intend to address these states' good neighbor obligations as expeditiously as practicable in a future action.

E. Methodology for Projecting Future Year Ozone Design Values

Consistent with the EPA's modeling guidance, the 2016 base year and future year air quality modeling results were used in a relative sense to project design values for 2023 and 2026. That is, the ratios of future year model predictions to base year model predictions are used to adjust ambient ozone design values up or down depending on the relative (percent) change in model predictions for each location. The modeling guidance recommends using measured ozone concentrations for the 5-year period centered on the base year as the air quality data starting point for future year projections. This average design value is used to dampen the effects of inter-annual variability in meteorology on ozone concentrations and to provide a reasonable projection of future air quality at the receptor under average conditions. In addition, the Agency calculated maximum design values from within the 5-year base period to represent conditions when meteorology is more favorable than average for ozone formation. Because the base year for the air quality modeling used in this final rule is 2016, measured data for 2014–2018 ( i.e., design values for 2016, 2017, and 2018) were used to project average and maximum design values in 2023 and 2026.

The ozone predictions from the 2016 and future year air quality model simulations were used to project 2016–2018 average and maximum ozone design values to 2023 and 2026 using an approach similar to the approach in EPA's guidance for attainment demonstration modeling. This guidance recommends using model predictions from the 3 x 3 array of grid cells surrounding the location of the monitoring site to calculate a Relative Response Factor (RRF) for that site. However, the guidance also notes that an alternative array of grid cells may be used in certain situations where local topographic or geographical feature ( e.g., a large water body or a significant elevation change) may influence model response.

The 2016–2018 base period average and maximum design values were multiplied by the RRF to project each of these design values to each of the three future years. In this manner, the projected design values are grounded in monitored data, and not the absolute model-predicted future year concentrations. Following the approach in the CSAPR Update and the Revised CSAPR Update, the EPA also projected future year design values based on a modified version of the “3 × 3” approach for those monitoring sites located in coastal areas. In this alternative approach, the EPA eliminated from the RRF calculations the modeling data in those grid cells that are dominated by water ( i.e., more than 50 percent of the area in the grid cell is water) and that do not contain a monitoring site ( i.e., if a grid cell is more than 50 percent water but contains an air quality monitor, that cell would remain in the calculation). The choice of more than 50 percent of the grid cell area as water as the criteria for identifying overwater grid cells is based on the treatment of land use in the Weather Research and Forecasting model (WRF). Specifically, in the WRF meteorological model those grid cells that are greater than 50 percent overwater are treated as being 100 percent overwater. In such cases the meteorological conditions in the entire grid cell reflect the vertical mixing and winds over water, even if part of the grid cell also happens to be over land with land-based emissions, as can often be the case for coastal areas. Overlaying land-based emissions with overwater meteorology may be representative of conditions at coastal monitors during times of on-shore flow associated with synoptic conditions or sea-breeze or lake-breeze wind flows. But there may be other times, particularly with off-shore wind flow, when vertical mixing of land-based emissions may be too limited due to the presence of overwater meteorology. Thus, for our modeling the EPA projected average and maximum design values at individual monitoring sites based on both the “3 × 3” approach as well as the alternative approach that eliminates overwater cells in the RRF calculation for near-coastal areas ( i.e., “no water” approach). The projected 2023 and 2026 design values using both the “3 × 3” and “no-water” approaches are provided in the docket for this final rule. For this final rule, the EPA is relying upon design values based on the “no water” approach for identifying nonattainment and maintenance receptors.

Consistent with the truncation and rounding procedures for the 8-hour ozone NAAQS, the projected design values are truncated to integers in units of ppb. Therefore, projected design values that are greater than or equal to 71 ppb are considered to be violating the 2015 ozone NAAQS. For those sites that are projected to be violating the NAAQS based on the average design values in the future analytic years, the Agency examined the measured design values for 2021, which are the most recent official measured design values at the time of this final rule. As noted earlier, the Agency is identifying nonattainment receptors in this rulemaking as those sites that are violating the NAAQS based on current measured air quality and also have projected average design values of 71 ppb or greater. Maintenance-only receptors include both (1) those sites with projected average design values above the NAAQS that are currently measuring clean data ( i.e., ozone design values below the level of the 2015 ozone NAAQS) and (2) those sites with projected average design values below the level of the NAAQS, but with projected maximum design values of 71 ppb or greater. In addition to the maintenance-only receptors, ozone nonattainment receptors are also maintenance receptors because the maximum design values for each of these sites is always greater than or equal to the average design value. The monitoring sites that the Agency projects to be nonattainment and maintenance receptors for the ozone NAAQS in the 2023 and 2026 base case are used for assessing the contribution of emissions in upwind states to downwind nonattainment and maintenance of the 2015 ozone NAAQS as part of this final rule.

Table IV.D–1 contains the 2016-centered base period average and maximum 8-hour ozone design values, the 2023 base case average and maximum design values and the measured 2021 design values for the sites that are projected to be nonattainment receptors in 2023. Table IV.D–2 contains this same information for monitoring sites that are projected to be maintenance-only receptors in 2023. The design values for all monitoring sites in the U.S. are provided in the docket for this rule. Additional details on the approach for projecting average and maximum design values are provided in the Air Quality Modeling Final Rule TSD.

In total, in the 2023 base case there are a total of 33 projected modeling-based receptors nationwide including 14 nonattainment receptors in 9 different counties and 19 maintenance-only receptors in 13 additional counties (Harris County, TX, has both nonattainment and maintenance-only receptors). Of the 14 nonattainment receptors in 2023, 7 remain nonattainment receptors, 5 are projected to become maintenance-only receptors and 2 are projected to be in attainment in 2026. Of the 19 maintenance-only receptors in 2023, 7 are projected to remain maintenance-only receptors and 12 are projected to be in attainment in 2026. The projected average and maximum design values in 2026 for all receptors are included in the Air Quality Modeling Final Rule TSD.

Comment: EPA received comments saying that the projected design values for 2023 were biased low compared to recent measured data. Commenters noted that a number of monitoring sites that are projected to be below the NAAQS in 2023 based on EPA's modeling for the proposed rule are currently measuring nonattainment. Because 2023 is only a year later than the most recent measured data some commenters said that EPA should give greater weight to measured data when identifying downwind receptors.

Response: Based on an analysis of model projections for 2023 and recent trends in measured data, the EPA agrees that current data indicate that there are monitoring sites at risk of continued nonattainment in 2023 even though the model projected average and maximum design values at these sites are below the NAAQS ( i.e., sites that are not modeling-based receptors). Specifically, the EPA believes that monitoring sites with measured design values and 4th high maximum daily 8-hour average (MDA8) ozone based on 2021 and preliminary 2022 data have the greatest risk of continuing to have a problem attaining the standard in 2023, even when the modeling projects these sites will attain. These criteria are sufficiently conservative that we avoid including monitoring sites that have measured nonattainment data in recent years but could reasonably be anticipated to not have a nonattainment or maintenance problem in 2023, in line with our modeling results. Our methodology is intended only to identify those sites that have sufficiently poor ozone levels that there is clearly a reasonable expectation that an ozone nonattainment or maintenance problem will persist in the 2023 ozone season. We do not apply this methodology for the 2026 analytic year, because that year is sufficiently farther in the future that we do not believe there would be a reasonable basis to supplement our modeling analysis with this “violating monitor” methodology. By comparison, 2023 is so near in time that recent measured ozone levels can be used reasonably to project whether an air quality problem is likely to persist. We view this approach to identifying additional receptors in 2023 as the best means of responding to the comments on this issue in this action. The monitoring sites that meet these criteria, along with the corresponding measured and modeled data, are provided in Table IV.D–3.

For purposes of this action, we will treat these sites as an additional type of maintenance-only receptor. Because our modeling did not identify these sites as receptors, we do not believe it is sufficiently certain that these sites will be in nonattainment that they should be considered nonattainment receptors for purposes of this final rule. Rather, our authority for treating these sites as receptors in 2023 flows from the responsibility in CAA section 110(a)(2)(i)(I) to prohibit emissions that interfere with maintenance of the NAAQS. See, e.g., North Carolina, 531 F.3d at 910–11 (failing to give effect to the interfere with maintenance clause “provides no protection for downwind areas that, despite EPA's predictions, still find themselves struggling to meet NAAQS due to upwind interference . . . .”) (emphasis added). Recognizing that no modeling can perfectly forecast the future, and “a degree of imprecision is inevitable in tackling the problem of interstate air pollution,” this approach in the Agency's judgement best balances the need to avoid both “under-control” and “overcontrol,” EME Homer City, 572 U.S. at 523.

In this action, we identify “violating monitor” maintenance-only receptors for purposes of more firmly establishing that the states we have otherwise identified as linked at Step 2 in our modeling-based methodology can indeed be reasonably anticipated to be linked to air quality problems in downwind states in 2023 for reasons that extend beyond that methodology. In this sense, this approach is “confirmatory” and does not alter the geography of the final rule compared to the application of the modeling-based receptor definitions used at proposal. Rather, it strengthens the analytical basis for our Step 2 findings by establishing that many upwind states covered in this action are also projected to contribute above 1 percent of the NAAQS to these types of receptors. For purposes of this final rule, we will not finalize FIPs for any states that this analysis indicates contribute greater than 1 percent of the NAAQS only to a “violating monitor” receptor. Our analysis suggests this would be the case for two states, Kansas and Tennessee (see section IV.F of this document). We are making no final decisions with respect to these states in this action and intend to address these states in a subsequent action.

F. Pollutant Transport From Upwind States

1. Air Quality Modeling To Quantify Upwind State Contributions

This section documents the procedures the EPA used to quantify the impact of emissions from specific upwind states on ozone design values in 2023 and 2026 for the identified downwind nonattainment and maintenance receptors. The EPA used CAMx photochemical source apportionment modeling to quantify the impact of emissions in specific upwind states on downwind nonattainment and maintenance receptors for 8-hour ozone. CAMx employs enhanced source apportionment techniques that track the formation and transport of ozone from specific emissions sources and calculates the contribution of sources and precursors to ozone for individual receptor locations. The benefit of the photochemical model source apportionment technique is that all modeled ozone at a given receptor location in the modeling domain is tracked back to specific sources of emissions and boundary conditions to fully characterize culpable sources.

The EPA performed nationwide, state-level ozone source apportionment modeling using the CAMx Ozone Source Apportionment Technology/Anthropogenic Precursor Culpability Analysis (OSAT/APCA) technique to quantify the contribution of 2023 and 2026 base case NO_X and VOC emissions from all sources in each state to the corresponding projected ozone design values in 2023 and 2026 at air quality monitoring sites. The CAMx OSAT/APCA model run was performed for the period May 1 through September 30 using the projected future base case emissions and 2016 meteorology for this time period. In the source apportionment modeling the Agency tracked ( i.e., tagged) the amount of ozone formed from anthropogenic emissions in each state individually as well as the contributions from other sources ( e.g., natural emissions).

In the state-by-state source apportionment model runs, the EPA tracked the ozone formed from each of the following tags:

• States—anthropogenic NO_X and VOC emissions from each state tracked individually (emissions from all anthropogenic sectors in a given state were combined);

• Biogenics—biogenic NO_X and VOC emissions domain-wide ( i.e., not by state);

• Boundary Concentrations—concentrations transported into the air quality modeling domain;
• Tribes—the emissions from those tribal lands for which the Agency has point source inventory data in the 2016v3 emissions modeling platform (EPA did not model the contributions from individual tribes);
• Canada and Mexico—anthropogenic emissions from sources in the portions of Canada and Mexico included in the modeling domain (the EPA did not model the contributions from Canada and Mexico separately);

• Fires—combined emissions from wild and prescribed fires domain-wide ( i.e., not by state); and

• Offshore—combined emissions from offshore marine vessels and offshore drilling platforms.

The contribution modeling provided contributions to ozone from anthropogenic NO_X and VOC emissions in each state, individually. The contributions to ozone from chemical reactions between biogenic NO_X and VOC emissions were modeled and assigned to the “biogenic” category. The contributions from wildfire and prescribed fire NO_X and VOC emissions were modeled and assigned to the “fires” category. That is, the contributions from the “biogenic” and “fires” categories are not assigned to individual states nor are they included in the state contributions.

For the Step 2 analysis, the EPA calculated a contribution metric that considers the average contribution on the 10 highest ozone concentration days ( i.e., top 10 days) in 2023. This average contribution metric is intended to provide a reasonable representation of the contribution from individual states to projected future year design values, based on modeled transport patterns and other meteorological conditions generally associated with modeled high ozone concentrations at the receptor. An average contribution metric constructed in this manner is beneficial since the magnitude of the contributions is directly related to the magnitude of the design value at each site.

The analytic steps for calculating the contribution metric for the 2023 analytic year are as follows:

(1) Calculate the 8-hour average contribution from each source tag to each monitoring site for the time period of the 8-hour daily maximum modeled concentrations in 2023;

(2) Average the contributions and average the concentrations for the top 10 modeled ozone concentration days in 2023;

(3) Divide the average contribution by the corresponding average concentration to obtain a Relative Contribution Factor (RCF) for each monitoring site;

(4) Multiply the 2023 average design values by the 2023 RCF at each site to produce the average contribution metric values in 2023.

This same approach was applied to calculate contribution metric values at individual monitoring sites for 2026.

The resulting contributions from each tag to each monitoring site in the U.S. for 2023 and 2026 can be found in the docket for this final rule. Additional details on the source apportionment modeling and the procedures for calculating contributions can be found in the Air Quality Modeling Final Rule TSD. The EPA's response to comments on the method for calculating the contribution metric can be found in the RTC document for this final rule.

The largest contribution from each state that is the subject of this rule to modeled 8-hour ozone nonattainment and maintenance receptors in downwind states in 2023 and 2026 are provided in Table IV.F–1 and Table IV.F–2, respectively. The largest contribution from each state to a “violating monitor” maintenance-only receptor is provided in Table IV.F–3.

2. Application of Contribution Screening Threshold

In Step 2 of the interstate transport framework, the EPA uses an air quality screening threshold to identify upwind states that contribute to downwind ozone concentrations in amounts sufficient to “link” them to these to downwind nonattainment and maintenance receptors. The contributions from each state to each downwind nonattainment or maintenance receptor that were used for the Step 2 evaluation can be found in the Air Quality Modeling Final Rule TSD.

The EPA applies an air quality screening threshold of 1 percent of the NAAQS, which has been used since the CSAPR rulemaking, including in the CSAPR Update, the Revised CSAPR Update, and numerous actions evaluating states' transport SIP submittals. The explanation for how this value was originally derived is available in the CSAPR rulemaking from 2011. See76 FR 48208, 48237–38. As originally explained there, the application of a relatively low threshold is intended to capture a relatively large percentage of the contribution from upwind states to downwind receptors in light of the regional-scale, collective contribution problem associated with both ozone and PM_2.5 NAAQS. Id. The Agency also explained that the use of a higher threshold in transport rules prior to CSAPR was based on single-day maximum contribution, whereas in CSAPR (and continuing in subsequent rules including this one), the Agency uses a more robust, average contribution metric over multiple days. Thus, it was not the case that 1 percent of NAAQS was substantially more stringent than that prior approach. Id. at 48238. In the 2016 CSAPR Update, the EPA reviewed the 1 percent threshold (as coupled with multi-day averaging) and determined it was appropriate to continue to apply this threshold. The EPA compared the 1 percent threshold to a 0.5 percent of NAAQS threshold and a 5 percent of NAAQS threshold. The EPA found that the lower threshold did not capture appreciably more upwind state contribution compared to the 1 percent threshold, while the 5 percent threshold allowed too much upwind state contribution to drop out from further analysis. The EPA continues to observe that nonattainment and maintenance receptors identified at Step 1 are impacted collectively by emissions from numerous upwind contributors. Therefore, application of a low, uniform screening threshold allows the EPA to identify upwind states that share a responsibility under the interstate transport provision to eliminate their significant contribution.

As we explained at proposal, the EPA recognizes that in 2018 it issued a memorandum indicating the potential for states to use a higher threshold at Step 2 in the development of their good neighbor SIP submissions where it could be technically justified. The August 2018 memorandum stated that “it may be reasonable and appropriate” for states to rely on an alternative 1 ppb threshold at Step 2. (The memorandum also indicated that any higher alternative threshold, such as 2 ppb, would likely not be appropriate.) The EPA nonetheless proposed to fulfill its role under CAA section 110(c) in promulgating FIPs to directly implement good neighbor requirements, and in this role, proposed retaining use of the 1 percent threshold for all states. We noted that in several documents proposing transport SIP disapprovals, see, e.g.,87 FR 9498 and 87 FR 9510 (Feb. 22, 2022), we explained that our experience since the issuance of the August 2018 memorandum regarding use of alternative thresholds led the Agency to believe it may not be appropriate to continue to attempt to recognize alternative contribution thresholds at Step 2, either in the context of SIPs or FIPs.

We went on to explain that the EPA's experience since 2018 is that allowing for alternative Step 2 thresholds may be impractical or otherwise inadvisable for a number of additional policy reasons. For a regional air pollutant such as ozone, consistency in requirements and expectations across all states is essential. Using multiple different thresholds at Step 2 with respect to the 2015 ozone NAAQS raises substantial policy consistency and practical implementation concerns. The application of different thresholds at Step 2 has the potential to result in inconsistent determination of good neighbor obligations. From the perspective of ensuring effective regional implementation of good neighbor obligations, the more important analysis is the evaluation of the emissions reductions needed, if any, to address a state's significant contribution after consideration of a multifactor analysis at Step 3, including a detailed evaluation that considers air quality factors and cost. We explained that while alternative thresholds for purposes of Step 2 may be “similar” in terms of capturing the relative amount of upwind contribution (as described in the August 2018 memorandum), nonetheless, use of alternative thresholds would allow certain states to avoid further evaluation of potential emissions controls while other states must proceed to a Step 3 analysis. This could create significant equity and consistency problems among states.

The EPA further proposed that, in promulgating FIPs to address these obligations on a nationwide scale, national ozone transport policy would not be well-served by applying a single, less stringent threshold at Step 2. The EPA recognized in the August 2018 memo that there was some similarity in the amount of total upwind contribution captured (on a nationwide basis) between 1 percent and 1 ppb. However, the EPA noted at proposal that while this may be true in some sense, that is hardly a compelling basis to move to a 1 ppb threshold. Indeed, the 1 ppb threshold has the disadvantage of losing a certain amount of total upwind contribution for further evaluation at Step 3. Considering the core statutory objective of ensuring elimination of all significant contribution to nonattainment or interference of the NAAQS in downwind states and the broad, regional nature of the collective contribution problem with respect to ozone, EPA could not identify a compelling policy imperative to move to a 1 ppb threshold.

In the proposal, we also found consistency with past interstate transport actions such as CSAPR, and the CSAPR Update and Revised CSAPR Update rulemakings (which used a Step 2 threshold of 1 percent of the NAAQS for two less protective ozone NAAQS) to be an important consideration. Continuing to use a 1 percent of NAAQS approach ensures that as the NAAQS are revised and made more stringent, an appropriate increase in stringency at Step 2 occurs, so as to ensure an appropriately larger amount of total upwind-state contribution is captured for purposes of fully addressing interstate transport for the more protective NAAQS.

The Agency also questioned whether it would be a good use of limited resources to attempt to further justify the use of alternative thresholds for certain states at Step 2 for purposes of the 2015 ozone NAAQS. Therefore, while EPA articulated the possibility of an alternative threshold in the August 2018 memorandum, the EPA concluded in the proposal that our experience and further evaluation since the issuance of that memo has revealed substantial programmatic and policy difficulties in attempting to implement this approach, and therefore we proposed to apply the 1 percent of NAAQS threshold.

Comment: Many commenters disagreed with our proposal to continue using a 1 percent of NAAQS threshold. They argued that the EPA was reversing course from its policy as articulated in the August 2018 memorandum and that the EPA was now bound to use a 1 ppb threshold rather than 1 percent of NAAQS, even in promulgating a FIP rather than evaluating SIPs. Commenters further argued that a 1 ppb threshold would be more consistent with the EPA's “significant impact level” (SIL) guidance related to implementing prevention of significant deterioration (PSD) permitting requirements. They argued that the 1 percent threshold was below precision limits of regulatory ozone monitors, and they argued it was within the “margin of error” of the EPA's modeling.

Response: The EPA is finalizing its proposed approach of consistently using a 1 percent of the NAAQS threshold at Step 2 in this action to determine which states contribute to identified nonattainment and maintenance receptors. This approach ensures both national consistency across all states and consistency and continuity with our prior interstate transport actions for other NAAQS. We do not agree that this approach is inconsistent with or a reversal in policy from the August 2018 memorandum, which only suggested that states in the development of their SIPs “may” be able to establish that 1 ppb could be an appropriate alternative threshold. The EPA has been consistent in that memorandum, and since that time, that final determinations on alternative thresholds would be made through rulemaking action, as the EPA is taking here.

The August 2018 memorandum made clear that the Agency had substantial doubts that any threshold greater than 1 ppb (such as 2 ppb) would be acceptable, and the Agency is affirming that a threshold higher than 1 ppb would not be justified under any circumstance for purposes of this action. No commenter credibly provided a basis for using a threshold even higher than 1 ppb, and so this issue is primarily limited to the difference between a 0.7 ppb threshold (the 1 percent of the NAAQS threshold discussed previously in this section) and a 1.0 ppb threshold. Therefore, before proceeding in responding to these comments, we note that this issue is only relevant to a small number of states whose contributions to any receptor are above 1 percent of the NAAQS but lower than 1 ppb. Under the 2016v3 modeling of 2023 being used in this final rule, the states in this rule with contributions that fall between 0.70 ppb and 1 ppb are Alabama, Kentucky, and Minnesota. Similarly, the EPA applies the 1 percent threshold in its 2026 modeling projections to determine if any states will not be linked to an ozone receptor by that year, and therefore should not be subject to the more stringent requirements that take effect in 2026. The states in this rule in that year with contribution between 0.70 ppb and 1 ppb are Kentucky, Nevada, and Oklahoma. For all other states covered in this action, at least one linkage exists in 2023 (and, as relevant, in 2026) that is greater than 1 ppb, and therefore the question of whether the EPA must recognize a 1 ppb threshold would not have a dispositive effect on the regulatory determination being made at Step 2.

The 1 percent of the NAAQS threshold is consistent with the Step 2 approach that the EPA applied in CSAPR for the 1997 ozone NAAQS and has subsequently been applied in the CSAPR Update and Revised CSAPR Update when evaluating determining interstate transport obligations for the 2008 ozone NAAQS. The EPA continues to find 1 percent of the ozone NAAQS to be an appropriate threshold. For ozone, as the EPA found in CAIR, CSAPR, and the CSAPR Update, a portion of the nonattainment and maintenance problems in the U.S. results from the combined impact of relatively small contributions from many upwind states, along with contributions from in-state sources and other sources. The EPA's analysis shows that the ozone transport problem being analyzed in this rule is still the result of the collective impacts of emissions from multiple upwind contributors. Therefore, application of a consistent contribution threshold is necessary to identify those upwind states that should have responsibility for addressing their contribution (to the extent found “significant” at Step 3) to the downwind nonattainment and maintenance problems to which they collectively contribute. Where a great number of geographically dispersed emissions sources contribute to a downwind air quality problem, which is the case for ozone, EPA believes that, in the context of CAA section 110(a)(2)(D)(i)(I), a state-level threshold of 1 percent of the NAAQS is a reasonably small enough value to identify only the greater-than-de minimis contributors yet is not so large that it unfairly focuses attention for further action only on the largest single or few upwind contributors. Continuing to use 1 percent of the NAAQS as the screening metric to evaluate collective contribution from many upwind states also allows the EPA (and states) to apply a consistent framework to evaluate interstate emissions transport under the interstate transport provision from one NAAQS to the next. See86 FR 23054, 23085; 81 FR 74504, 74518; 76 FR 48208, 48237–38.

Further, the EPA notes that the role of the Step 2 threshold is limited and just one step in the larger 4-Step Framework. It serves to screen in states for further evaluation of emissions control opportunities applying a multifactor analysis at Step 3. Thus, as the Supreme Court has recognized, the contribution threshold essentially functions to exclude states with “ de miminis ” impacts. EME Homer City, 572 U.S. 489, 500.

Comments related to the August 2018 memorandum argued that the EPA legally committed itself to approving SIP submissions from states with contributions below 1 ppb and so now the EPA must apply that threshold in this FIP action. (Comments regarding this issue as related to the EPA's action on SIPs is addressed in that rulemaking and is beyond the scope of this action.) This is not what the memorandum said. The memorandum merely provided an analysis regarding “the degree to which certain air quality threshold amounts capture the collective amount of upwind contribution from upwind states.” It interpreted “that information to make recommendations about what thresholds may be appropriate for use in” SIP submissions (emphasis added). Specifically, the August 2018 memorandum said, “Because the amount of upwind collective contribution capture with the 1 percent and the 1 ppb thresholds is generally comparable, overall, we believe it may be reasonable and appropriate for states to use a 1 ppb contribution threshold, as an alternative to a 1 percent threshold, at Step 2 of the 4-step framework in developing their SIP revisions addressing the good neighbor provision for the 2015 ozone NAAQS” (emphasis added). Thus, the text of the August 2018 memorandum in no way committed that the EPA would be using a 1 ppb threshold going forward either in its evaluation of SIPs or in promulgating a FIP. The August 2018 memorandum indicated that “[f]ollowing these recommendations does not ensure that EPA will approve a SIP revision in all instances where the recommendations are followed, as the guidance may not apply to the facts and circumstances underlying a particular SIP. Final decisions by the EPA to approve a particular SIP revision will only be made based on the requirements of the statute and will only be made following an air agency's final submission of the SIP revision to the EPA, and after appropriate notice and opportunity for public review and comment.” Further, the August 2018 memorandum said that “EPA and air agencies should consider whether the recommendations in this guidance are appropriate for each situation.” The memorandum said nothing regarding what threshold the EPA would apply if promulgating a FIP.

As explained in the SIP disapproval action and again here, the EPA finds it would not be sound policy to apply an alternative contribution threshold or thresholds to one or more states within the 4-step interstate transport framework for the 2015 ozone NAAQS. However, the EPA disagrees with commenters' claims that the agency has reversed course on applying the August 2018 memorandum, because the memorandum never adopted a view that the use of 1 ppb or other alternative thresholds would in fact be acceptable. Although the EPA said at proposal that the EPA may rescind the guidance in the future, we took comment on the subject and also stated, “EPA is not at this time rescinding the August 2018 memorandum.” The EPA is not formally rescinding the August 2018 memorandum in this action or at this time. However, it is not required that agencies must “rescind” a memorandum or guidance the moment it becomes outdated or called into question. The August 2018 memorandum was not issued through notice-and-comment rulemaking and is not binding on the Agency or other parties. While the willingness of the Agency as expressed in that memorandum to entertain the possibility of an alternative threshold of 1 ppb may be considered a kind of policy position, agencies may change their non-binding policies without going through notice and comment rulemaking. Catawba County v. EPA, 571 F.3d 20, 34 (D.C. Cir. 2009). In this case, we went through notice and comment rulemaking on this topic in the SIP-disapproval action (88 FR 9336) and here, even though the August 2018 memorandum was issued without such opportunity for public input. We further address the basis for the consistent use of a 1 percent of NAAQS threshold and summarize our conclusions under the FCC v. Fox factors below.

We continue to believe, as set forth in our proposed action, that national ozone transport policy is not well served by allowing for less protective thresholds than 1 percent of the NAAQS at Step 2. Furthermore, the EPA disagrees with commenters who suggest that national consistency is an inappropriate consideration in the context of interstate ozone transport. The Good Neighbor provision, CAA section 110(a)(2)(D)(i)(I), requires to a unique degree of concern for consistency, parity, and equity across state lines. For a regional air pollutant such as ozone, consistency in requirements and expectations across all states is essential. Based on the EPA's review of good neighbor SIP submissions to-date and after further consideration of the policy implications of attempting to recognize an alternative Step 2 threshold for certain states, the Agency concludes that the attempted use of different thresholds at Step 2 with respect to the 2015 8-hour ozone NAAQS raises substantial policy consistency and practical implementation concerns. The availability of different thresholds at Step 2 has the potential to result in inconsistent application of good neighbor obligations based solely on the strength of a state's SIP submission at Step 2 of the 4-step interstate transport framework. The steps of the analysis that lead up to evaluating emissions reductions opportunities to address states' significant contribution at Step 3 should be applied on a consistent basis. Where alternative thresholds for purposes of Step 2 may be “similar” in terms of capturing the relative amount of upwind contribution (as described in the August 2018 memorandum), nonetheless, use of an alternative threshold would allow certain states to avoid further evaluation of potential emissions controls while other states must proceed to a Step 3 analysis. This can create significant equity and consistency problems among states and could lead to ineffective or inefficient approaches to eliminating significant contribution.

One commenter suggested the EPA could address this potentially inequitable outcome by simply adopting a 1 ppb contribution threshold for all states. However, the August 2018 memorandum did not conclude that 1 ppb would be appropriate for all states and the EPA does not view that conclusion to be supported at present. The EPA recognized in the August 2018 memorandum that there was some similarity in the amount of total upwind contribution captured (on a nationwide basis) between 1 percent and 1 ppb. However, while this may be true in some sense, that is hardly a compelling basis to move to a 1 ppb threshold for every state. Indeed, the 1 ppb threshold has the disadvantage of losing a certain amount of total upwind contribution for further evaluation at Step 3 ( e.g., roughly 7 percent of total upwind state contribution was lost according to the modeling underlying the August 2018 memorandum; in the EPA's 2016v2 modeling, the amount lost is 5 percent; in the EPA's 2016v3 modeling used for final, the amount lost is also 5 percent). Further, this logic has no end point. A similar observation could be made with respect to any incremental change. For example, should the EPA next recognize a 1.2 ppb threshold because that would only cause some small additional loss in capture of upwind state contribution as compared to 1 ppb? If the only basis for moving to a 1 ppb threshold is that it captures a “similar” (but actually smaller) amount of upwind contribution, then there is no basis for moving to that threshold at all. Considering the core statutory objective of ensuring elimination of all significant contribution to nonattainment or interference with maintenance of the NAAQS in other states and the broad, regional nature of the collective contribution problem with respect to ozone, we continue to find no compelling policy reason to adopt a new threshold for all states of 1 ppb.

Nor have commenters explained why use of a 1 ppb threshold would be appropriate under the more protective 2015 ozone NAAQS when a 1 percent of the NAAQS contribution threshold has been used for less protective ozone NAAQS. To illustrate, a state contributing greater than 0.75 ppb but less than 1 ppb to a receptor under the 2008 ozone NAAQS was “linked” at Step 2, but if a 1 ppb threshold were used for the 2015 ozone NAAQS then that same state would not be “linked” to a receptor at Step 2 under a NAAQS that is set to be more protective of human health and the environment. Consistency with past interstate transport actions such as CSAPR, and the CSAPR Update and Revised CSAPR Update rulemakings (which all used the 1 percent of the NAAQS for less protective ozone NAAQS), is an important consideration. We affirm our view in CSAPR that continuing to use a 1 percent of NAAQS approach ensures that if the NAAQS are revised and made more stringent, an appropriate increase in stringency at Step 2 occurs, so as to ensure an appropriately larger amount of total upwind-state contribution is captured for purposes of fully addressing interstate transport. See76 FR 48208, 48237–38.

We note further that application of a 1 percent of NAAQS threshold has been the EPA's consistent approach in each of our notice-and-comment rulemakings beginning with CSAPR and continuing with the CSAPR Update, the Revised CSAPR Update, and numerous actions on ozone transport SIP submissions. In each case, the 1 percent of the NAAQS threshold was subject to rigorous vetting through public comment and the Agency's response to those comments, including through the use of analytical evaluations of alternative thresholds. See, e.g.,81 FR 74518–19. By contrast, the August 2018 memorandum was not issued through notice-and-comment rulemaking procedures, and the EPA was careful to caveat its utility and ultimate reliability for that reason.

The EPA disagrees with claims that the EPA is applying the August 2018 memorandum inconsistently based on the EPA's actions with regard to Arizona, Iowa, and Oregon. The EPA withdrew a previously proposed approval of Iowa's SIP submission that was premised on a 1 ppb contribution threshold, and re-proposed and finalized approval of that SIP based on a different rationale using a 1 percent of the NAAQS contribution threshold. 87 FR 9477 (Feb. 22, 2022); 87 FR 22463 (April 15, 2022). The EPA also disagrees with any claim that Oregon and Arizona were “allowed” to use a 1 ppb or higher threshold. The EPA approved Oregon's SIP submission for the 2015 ozone NAAQS on May 17, 2019, and both Oregon and the EPA relied on a 1 percent of the NAAQS contribution threshold. 84 FR 7854, 7856 (March 5, 2019) (proposal); 84 FR 22376 (May 17, 2019) (final). In the proposal for this action, the EPA explained it was not proposing to conduct an error correction for Oregon even though updated modeling indicated Oregon contributed above 1 percent of the NAAQS to monitors in California.

The EPA is deferring finalizing a finding at this time for Oregon ( see section IV.G of this document for additional information). In 2016, the EPA approved Arizona's SIP for the earlier 2008 ozone NAAQS based on a similar rationale with regard to certain monitors in California. 81 FR 15200 (March 22, 2016) (proposal); 81 FR 31513 (May 19, 2016) (final rule). We are deferring finalizing a finding at this time that such a rationale is appropriate with respect to the more protective 2015 ozone NAAQS. While Arizona and Oregon's interstate transport obligations for the 2015 ozone NAAQS remain pending (along with several other states), there is no inconsistency in the treatment of these states or any other state at Step 2.

Some commenters claim the EPA must use a 1 ppb threshold based on the identification of 1 ppb as a significance threshold in one step of the PSD permitting process. The EPA's SIL guidances, however, relate to a different provision of the Clean Air Act regarding implementation of the prevention of significant deterioration (PSD) permitting program. This program applies in areas that have been designated attainment of the NAAQS and is intended to ensure that such areas remain in attainment even if emissions were to increase as a result of new sources or major modifications to existing sources located in those areas. This purpose is different than the purpose of the good neighbor provision, which is to assist downwind areas (in some cases hundreds or thousands of miles away) in resolving ongoing nonattainment of the NAAQS or difficulty maintaining the NAAQS through eliminating the emissions from other states that are significantly contributing to those problems. In addition, as discussed in preceding paragraphs, the purpose of the Step 2 threshold within the EPA's interstate transport framework for ozone is to broadly sweep in all states contributing to identified receptors above a de minimis level in recognition of the collective-contribution problem associated with regional-scale ozone transport. The threshold used in the context of PSD SIL serves a different purpose, and so it does not follow that they should be made equivalent. Further, commenters incorrectly associate the EPA's Step 2 contribution threshold with the identification of “significant” emissions (which does not occur until Step 3), and so it is not the case that the EPA is interpreting the same term differently.

The EPA has previously explained this distinction between the good neighbor framework and PSD SILs. See70 FR 25162, 25190–25191 (May 12, 2005); 76 FR 48208, 48237 (Aug. 8, 2011). Importantly, the implication of the PSD SIL threshold is not that single-source contribution below this level indicates the absence of a contribution or that no emissions control requirements are warranted. Rather, the PSD SIL threshold addresses whether further, more comprehensive, multi-source review or analysis of air quality impacts are required of the source to support a demonstration that it meets the criteria for a permit. A source with estimated impacts below the PSD SIL may use this to demonstrate that it will not cause or contribute (as those terms are used within the PSD program) to a violation of an ambient air quality standard, but is still subject to meeting applicable control requirements, including best available control technology, designed to moderate the source's impact on air quality.

Moreover, other aspects of the technical methodology in the SILs guidance compared to the good neighbor framework make a direct comparison between these two values misleading. For instance, in PSD permit modeling using a single year of meteorology the maximum single-day 8-hour contribution is evaluated with respect to the SIL. The purpose of the contribution threshold at Step 2 of the 4-step good neighbor framework is to determine whether the average contribution from a collection of sources in a state is small enough not to warrant any additional control for the purpose of mitigating interstate transport, even if that control were highly cost effective. Using a 1 percent of the NAAQS threshold is more appropriate for evaluating multi-day average contributions from upwind states than a 1 ppb threshold applied for a single day, since that lower value of 1 percent of the NAAQS will capture variations in contribution. If EPA were to use a single day reflecting the maximum amount of contribution from an upwind state to determine whether a linkage exists at Step 2, commenters' arguments for use of the PSD SIL might have more force. This would in effect be a return to the pre-CSAPR contribution calculation methodology of using a single day, see76 FR 48238. However, that would likely cause more states to become linked, not less. And in any case, consistent with the method in our modeling guidance for projecting future attainment/nonattainment and as the EPA concluded in 2011 in CSAPR, the present good neighbor methodology of using multiple days provides a more robust approach to establishing that a linkage exists at the state level than relying on a single day of data.

A commenter also claimed the 1 percent of NAAQS threshold is inconsistent with the standards of precision for Federal reference monitors for ozone and the rounding requirements found in 40 CFR part 50, appendix U, Interpretation of the Primary and Secondary National Ambient Air Quality Standards for Ozone. Commenter claimed that the 1 percent contribution threshold of 0.7 ppb is lower than the manufacturer's reported precision of these reference monitors and that the requirements found in Appendix U truncates monitor values of 0.7 ppb to 0 ppb. However, the commenter is mistaken in applying criteria related to the precision of monitoring technology to the modeling methodology by which we project contributions when quantifying and evaluating interstate transport at Step 2. Indeed, contributions by source or state cannot be derived from the total ambient concentration of ozone at a monitor at all but must be apportioned through modeling. Under our longstanding methodology for doing so, the contribution values identified from upwind states are based on a robust assessment of the average impact of each upwind state's ozone-precursor emissions over a range of scenarios, as explained in the 2016v3 modeling's Air Quality Modeling Final Rule TSD, in the docket for this rule, Docket ID No. EPA–HQ–OAR–2021–0668. This analysis is in no way connected with or dependent on monitoring instruments' precision of measurement. See EME Homer City, 795 F.3d 118, 135–36 (“[A] model is meant to simplify reality in order to make it tractable.' ”) (quoting Chemical Manufacturers Association v. EPA, 28 F.3d 1259, 1264 (D.C. Cir. 1994).

To the extent that commenters argue that the EPA consider a less stringent threshold as a result of modeling uncertainty, the EPA disagrees with this notion. The EPA has successfully applied a 1 percent of NAAQS threshold to identify linked upwind states using modeling in three prior FIP rulemakings and numerous state-specific actions on good neighbor obligations. This continues to be a reasonable approach, and indeed courts have repeatedly declined to establish bright line criteria for model performance. In upholding the EPA's approach to evaluating interstate transport in CSAPR, the D.C. Circuit held that it would not “invalidate EPA's predictions solely because there might be discrepancies between those predictions and the real world. That possibility is inherent in the enterprise of prediction.” EME Homer City Generation, L.P. v. EPA, 795 F.3d 118, 135 (2015). “[T]he fact that a `model does not fit every application perfectly is no criticism; a model is meant to simplify reality in order to make it tractable.' ” Id. at 135–36 (quoting Chemical Manufacturers Association v. EPA, 28 F.3d 1259, 1264 (D.C. Cir. 1994). See also Sierra Club v. EPA, 939 F.3d 649, 686–87 (5th Cir. 2019) (upholding EPA's modeling in the face of complaints regarding an alleged “margin of error,” noting challengers face a “considerable burden” in overcoming a “presumption of regularity” afforded “the EPA's choice of analytical methodology”) (citing BCCA Appeal Grp. v. EPA, 355 F.3d 817, 832 (5th Cir. 2003)).

The Agency will continue to use the CAMx model to evaluate contributions from upwind states to downwind areas. The agency has used CAMx routinely in previous notice and comment transport rulemakings to evaluate contributions relative to the 1 percent threshold for both ozone and PM_2.5. In fact, in the original CSAPR, the EPA found that “[t]here was wide support from commenters for the use of CAMx as an appropriate, state-of-the science air quality tool for use in the [Cross-State Air Pollution] Rule. There were no comments that suggested that the EPA should use an alternative model for quantifying interstate transport.” 76 FR 48229 (August 8, 2011). In this action, the EPA has taken a number of steps based on comments and new information to ensure to the greatest extent the accuracy and reliability of its modeling projections at Step 1 and 2, as discussed elsewhere in this section.

The EPA disagrees with commenters that case law reviewing changes in agency positions such as FCC v. Fox TV Stations, Inc., 556 U.S. 502, 515 (2009), is applicable with respect to this issue. As explained above, under the terms of the August 2018 memorandum, the Agency did not conclude that the use of an alternative contribution threshold was justified for any states. But even if it were found that the Agency's position had changed between this rulemaking action and the August 2018 memorandum, the FCC v. Fox factors are met. We have explained above that there are good reasons for continuing to use a 1 percent of NAAQS threshold. We also are aware that we are not using a 1 ppb threshold despite acknowledging the potential for doing so in the August 2018 memorandum. We do not believe that any party has a serious reliance interest that would be sufficient to overcome the countervailing public interest that is served through the EPA's determination to maintain continuity with its longstanding, more protective 1 percent of NAAQS threshold in this action. Cf.88 FR 9373 (reviewing reliance in the context of the SIP-disapproval action).

The EPA therefore will continue its longstanding practice of applying the 1 percent of NAAQS threshold in this action.

a. States That Contribute Below the Screening Threshold

Based on the EPA's modeling and considering measured data at violating monitors, the contributions from each of the following states to nonattainment or maintenance-only receptors in the 2023 analytic year are below the 1 percent of the NAAQS threshold: Colorado, Connecticut, the District of Columbia, Delaware, Florida, Georgia, Idaho, Maine, Massachusetts, Montana, Nebraska, New Hampshire, North Carolina, North Dakota, Rhode Island, South Carolina, South Dakota, Vermont, and Washington. The EPA has already approved these states' 2015 ozone good neighbor SIP submittals. Because the contributions from these states to projected downwind air quality problems are below the screening threshold in the current modeling, these states are not within the scope of this final rule. Additionally, the EPA has made final determinations that two states outside the modeling domain for the air quality modeling analyzed in this final rulemaking—Hawaii and Alaska —do not significantly contribute to nonattainment or interfere with maintenance of the NAAQS in any other state.

With respect to Wyoming, our methodology when applied using the 2016v3 modeling suggests that whether the state is linked is uncertain and warrants further analysis. The EPA intends to expeditiously review its assessment with respect to Wyoming and take action addressing Wyoming's good neighbor obligations for the 2015 ozone NAAQS through a separate action.

b. States That Contribute at or Above the Screening Threshold

Based on the maximum downwind contributions in Table IV.F–1, the Step 2 analysis identifies that the following 21 states contribute at or above the 0.70 ppb threshold to downwind nonattainment receptors in 2023: Alabama, Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, and West Virginia. Based on the maximum downwind contributions in Table IV.F–1, the following 23 states contribute at or above the 0.70 ppb threshold to downwind modeling-based maintenance-only receptors in 2023: Arizona, Arkansas, California, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New Mexico, New York, Ohio, Oklahoma, Texas, Virginia, West Virginia, and Wisconsin. Based on the maximum downwind contribution in Table IV.F–3, the following additional states contribute at or above the 0.70 ppb threshold to downwind violating monitor maintenance-only receptors in 2023: Kansas and Tennessee. (However, the EPA is not taking final action based on this analytical result for these two states at this time.) The levels of contribution between each of these linked upwind states and downwind nonattainment receptors and maintenance-only receptors are provided in the Air Quality Modeling Final Rule TSD.

Among the linked states are several western states—California, Nevada, and Utah. While the EPA has not previously included action on linked western states in its prior CSAPR rulemakings, the EPA has consistently applied the 4-step framework in evaluating good neighbor obligations from these states. On a case-by-case basis, the EPA has found in some instances with respect to the 2008 ozone NAAQS that a unique consideration has warranted approval of a western state's good neighbor SIP submittal that might otherwise be found to contribute above 1 percent of the NAAQS without concluding that additional emissions reductions are required at Step 3 of the framework. The EPA has also explained in prior actions that its air quality modeling is reliable for assessing downwind air quality problems and ozone transport contributions from upwind states throughout the nationwide modeling domain. The EPA is deferring finalizing a finding at this time for Oregon ( see section IV.G of this document for additional information).

As explained in the following section, the EPA is not, in this action, altering its prior approval of Oregon's good neighbor SIP submission for the 2015 ozone NAAQS. For the remaining western states included in this rule, the EPA's modeling supports a conclusion that these states are linked above the contribution threshold to identified ozone transport receptors in downwind states, and therefore, consistent with the treatment of all other states within the modeling domain, the EPA proposes to proceed to evaluate these states for a determination of “significant contribution” at Step 3.

In conclusion, as described above, states with contributions that equal or exceed 1 percent of the NAAQS to either nonattainment or maintenance-only receptors are identified as “linked” at Step 2 of the good neighbor framework and warrant further analysis for significant contribution to nonattainment or interference with maintenance under Step 3. The EPA finds that for purposes of this final rule, the following 23 states are linked at Step 2 in 2023: Alabama, Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, West Virginia, and Wisconsin. In addition, the EPA finds that the following 20 States are linked at Step 2 in 2026: Arkansas, California, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Nevada, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, Texas, Utah, Virginia, and West Virginia. We note that our updated modeling for this final rule shows that two states, Minnesota and Wisconsin, that we found linked in 2026 at proposal are no longer projected to be linked in that year but are linked in 2023. As at proposal, Alabama is only projected to be linked in 2023, not 2026.

For six states, the EPA's analysis at this time indicates that a linkage may exist in 2023 for which the EPA had not proposed FIP requirements, or the updated analysis for this final rule suggests that linkages we had previously found in the proposed action are now uncertain and warrant further analysis. The EPA intends to expeditiously address these states in a separate action or actions: Arizona, Iowa, Kansas, New Mexico, Tennessee, and Wyoming.

G. Treatment of Certain Monitoring Sites in California and Implications for Oregon's Good Neighbor Obligations for the 2015 Ozone NAAQS

The EPA previously approved Oregon's September 25, 2018 transport SIP submittal for the 2015 ozone NAAQS on May 17, 2019 (84 FR 22376), because in an earlier round of modeling Oregon was not projected to contribute above 1 percent of the NAAQS to any downwind receptors. In the EPA's updated modeling used at proposal (2016v2) and again in the final modeling (2016v3), Oregon is modeled to contribute above the 1 percent of NAAQS threshold to several monitoring sites in California that would generally meet the EPA's definition of nonattainment or maintenance “receptors” at Step 1. At proposal, the EPA explained that our analysis of the nature of the air quality problem at these monitoring sites led us to propose a determination that these monitoring sites should not be treated as receptors for purposes of determining interstate transport obligations of upwind states under CAA section 110(a)(2)(D)(i)(I). We explained that we reached this conclusion at Step 1 of our 4-step framework.

The EPA previously made a similar assessment of the nature of certain other monitoring sites in California in approving Arizona's 2008 ozone NAAQS transport SIP submittal. There, the EPA noted that a “factor [. . .] relevant to determining the nature of a projected receptor's interstate transport problem is the magnitude of ozone attributable to transport from all upwind states collectively contributing to the air quality problem.” The EPA observed that only one upwind state (Arizona) was linked above 1 percent of the 2008 ozone NAAQS to the two relevant monitoring sites in California, and the cumulative ozone contribution from all upwind states to those sites was 2.5 percent and 4.4 percent of the total ozone, respectively. The EPA determined the size of those cumulative upwind contributions was “negligible, particularly when compared to the relatively large contributions from upwind states in the East or in certain other areas of the West.” In that action, the EPA concluded the two California sites to which Arizona was linked should not be treated as receptors for the purposes of determining Good Neighbor obligations for the 2008 ozone NAAQS.

Comment: Commenters criticized what they considered to be unfair treatment of Oregon, stating that the EPA is applying a higher contribution threshold than it applies to other states. Commenters argued that EPA has not established a specific threshold for why the level of upwind-state impact at these sites should not be considered meaningful. Commenters argued that our analysis ignored the fact that there are many monitoring sites in California to which Oregon contributes above 1 percent of the NAAQS. Commenters state that EPA has failed to explain why Oregon is not subject to this rulemaking, while other states contribute lower total downwind ozone contributions and fewer receptors. Commenters concluded that since Oregon is linked it should be subject to the same emissions control determinations at Step 3 and 4 as every other state, or otherwise apply the same “nature of the air quality problem” consideration to eliminate other receptors.

Response: The EPA acknowledges that several commenters opposed the proposed treatment of Oregon and the California monitoring sites to which it is linked in the proposed and final modeling. We also recognize that other commenters expressed confusion regarding the role of this proposed determination at Step 1 and how it relates to the longstanding 4-step interstate transport framework that the EPA is otherwise applying in this action. In recognition of these concerns and the need to give further thought to the appropriate treatment of both upwind states and downwind receptors in these circumstances, the EPA is deferring finalizing a finding at this time for Oregon. The current approval of the state's SIP submission will remain in place for the time being, pending further review. We make no final determination in this action regarding whether the California monitoring sites at issue should or should not be treated as receptors for purposes of addressing interstate transport for the 2015 ozone NAAQS.

V. Quantifying Upwind-State NO _X Emissions Reduction Potential To Reduce Interstate Ozone Transport for the 2015 Ozone NAAQS

A. The Multi-Factor Test for Determining Significant Contribution

This section describes the EPA's methodology at Step 3 of the 4-step framework for identifying upwind emissions that constitute “significant” contribution for the states subject to this final rule and focuses on the 23 states with FIP requirements identified in the previous sections. Following the existing framework as applied in the prior CSAPR rulemakings, the EPA's assessment of linked upwind state emissions is based primarily on analysis of several alternative levels of NO_X emissions control stringency applied uniformly across all of the linked states. The analysis includes assessment of non-EGU stationary sources in addition to EGU sources in the linked upwind states.

The EPA applies a multi-factor test—the same multi-factor test that was used in CSAPR, the CSAPR Update, and the Revised CSAPR Update —to evaluate increasing levels of uniform NO_X control stringency. The multi-factor test, which is central to EPA's Step 3 quantification of significant contribution, considers cost, available emissions reductions, downwind air quality impacts, and other factors to determine the appropriate level of uniform NO_X control stringency that would eliminate significant contribution to downwind nonattainment or maintenance receptors. The selection of a uniform level of NO_X emissions control stringency across all of the linked states, reflected as a representative cost per ton of emissions reduction (or a weighted average cost per ton in the case of EPA's non-EGU and EGU analysis for 2026 mitigation measures), also serves to apportion the reduction responsibility among collectively contributing upwind states. This approach to quantifying upwind state emission-reduction obligations using uniform cost was reviewed by the Supreme Court in EME Homer City Generation, which held that using such an approach to apportion emissions reduction responsibilities among upwind states that are collectively responsible for downwind air quality impacts “is an efficient and equitable solution to the allocation problem the Good Neighbor Provision requires the Agency to address.” 572 U.S. at 519.

There are four stages in developing the multi-factor test: (1) identify levels of uniform NO_X control stringency; (2) evaluate potential NO_X emissions reductions associated with each identified level of uniform control stringency; (3) assess air quality improvements at downwind receptors for each level of uniform control stringency; and (4) select a level of control stringency considering the identified cost, available NO_X emissions reductions, and downwind air quality impacts, while also ensuring that emissions reductions do not unnecessarily over-control relative to the contribution threshold or downwind air quality.

As mentioned in section III.A.2 of this document, commenters on the proposed rule and previous ozone transport rules have suggested that the EPA should regulate VOCs as an ozone precursor. For this final rule, the EPA examined the results of the contribution modeling performed for this rule to identify the portion of the ozone contribution attributable to anthropogenic NO_X emissions versus VOC emissions from each linked upwind state to each downwind receptor. Of the total upwind-downwind linkages in 2023, the contributions from NO_X emissions comprise 80 percent or more of the total anthropogenic contribution for nearly all of the linkages (121 out of 124 total). Across all receptors, the contribution from NO_X emissions ranges from 84 percent to 97 percent of the total anthropogenic contribution from upwind states. This review of the portion of the ozone contribution attributable to anthropogenic NO_X emissions versus VOC emissions from each linked upwind state leads the Agency to conclude that the vast majority of the downwind air quality areas addressed by the final rule under are primarily NO_X -limited, rather than VOC-limited. Therefore, the EPA continues to find that regulation of VOCs as an ozone precursor in upwind states is not necessary to eliminate significant contribution or interference with maintenance in downwind areas in this final rule. The remainder of this section focuses on EPA's strategy for reducing regional-scale transport of ozone by targeting NO_X emissions from stationary sources to achieve the most effective reductions of ozone transport over the geography of the affected downwind areas.

For both EGUs and non-EGUs, section V.B of this document describes the available NO_X emissions controls that the EPA evaluated for this final rule and their representative cost levels (in 2016$). Section V.C of this document discusses EPA's application of that information to assess emissions reduction potential of the identified control stringencies. Finally, section V.D of this document describes EPA's assessment of associated air quality impacts and EPA's subsequent identification of appropriate control stringencies considering the key relevant factors (cost, available emissions reductions, and downwind air quality impacts).

This multi-factor approach is consistent with EPA's approach in prior transport actions, such as CSAPR. In addition, as was evaluated in the CSAPR Update and Revised CSAPR Update, the EPA evaluated whether, based on particularized evidence, its selected control strategy would result in over-control for any upwind state by examining whether an upwind state is linked solely to downwind air quality problems that could have been resolved at a lesser threshold of control stringency and whether an upwind state could reduce its emissions below the 1 percent air quality contribution threshold at a lesser threshold of control stringency. This analysis is described in section V.D of this document.

Finally, while the EPA has evaluated potential emissions reductions from non-EGU sources in prior rules and found certain non-EGU emissions reductions should inform the budgets established in the NO_X SIP Call, this is the first action for which the EPA is finalizing non-EGU emissions reductions within the context of the specific, 4-step interstate transport framework established in CSAPR. The EPA applies its multi-factor test to non-EGUs and independently evaluates non-EGU industries in a consistent but parallel track to its Step 3 assessment for EGUs. This is consistent with the parallel assessment approach taken for EGUs and non-EGUs in the Revised CSAPR Update. Following the conclusions of the EGU and non-EGU multi-factor tests, the identified reductions for EGUs and non-EGUs are combined and collectively analyzed to assess their effects on downwind air quality and whether the rule achieves a full remedy to eliminate “significant contribution” while avoiding over-control.

To ensure that this rule implements a full remedy for the elimination of significant contribution from upwind states, the EPA has reviewed available information on all major industrial source sectors in the upwind states inclusive of commenter-provided data. This analysis leads the EPA to conclude that both EGUs and certain large sources in several specific industrial categories should be evaluated for emissions control opportunities. As discussed in the sections that follow, the EPA determines, for both EGUs and the selected non-EGU source categories, there are impactful emissions reduction opportunities available at reasonable cost-effectiveness thresholds. As in the Revised CSAPR Update, the EPA examines EGUs and non-EGUs in this section on consistent but distinct parallel tracks due to differences stemming from the unique characteristics of the power sector compared to other industrial source categories.

Since the NO_X SIP Call, EGUs have consistently been regulated under ozone transport rules. These units operate in a coordinated manner across a highly interconnected electrical grid. Their configuration and emissions control strategies are relatively homogenous, and their emissions levels and emissions control opportunities are generally very well understood due to longstanding monitoring and data-reporting requirements. Non-EGU sources, by contrast, are relatively heterogeneous, even within a single industrial category, and have far greater variation in existing emissions control requirements, emissions levels, and technologies to reduce emissions. In general, despite these differences, the information available for this rulemaking indicates that both EGUs and certain non-EGU categories have available cost-effective NO_X emissions reduction opportunities at relatively commensurate cost per ton levels, and these emissions reductions will make a meaningful improvement in air quality at the downwind receptors. Section V.B.2 of this document describes EPA's process for selecting specific non-EGU industries and emissions unit types included in this final rulemaking.

The EPA notes that its Step 3 analysis for this FIP does not assess additional emissions reduction opportunities from mobile sources. The EPA continues to believe that title II of the CAA provides the primary authority and process for reducing these emissions at the Federal level. EPA's various Federal mobile source programs, summarized in this section, have delivered and are projected to continue to deliver substantial nationwide reductions in both VOCs and NO_X emissions; these reductions from final rules are factored into the Agency's assessment of air quality and contributions at Steps 1 and 2. Further, states are generally preempted from regulating new vehicles and engines with certain exceptions, and therefore a question exists regarding EPA's authority to address such emissions through such means when regulating in place of the states under CAA section 110(c). See generally CAA section 209. See also86 FR 23099. As noted earlier, the EPA accounted for mobile source emissions reductions resulting from other federally enforceable regulatory programs in the development of emissions inventories used to support analysis for this final rulemaking, and the EPA does not evaluate any mobile source control measures in its Step 3 evaluation in this rule. For further discussion of EPA's existing and ongoing mobile source measures, see section V.B.4 of this document.

B. Identifying Control Stringency Levels

1. EGU NO_X Mitigation Strategies

In identifying levels of uniform control stringency for EGUs, the EPA assessed the same NO_X emissions controls that the Agency analyzed in the CSAPR Update and the Revised CSAPR Update, all of which are considered to be widely available in this sector: (1) fully operating existing SCR, including both optimizing NO_X removal by existing operational SCRs and turning on and optimizing existing idled SCRs; (2) installing state-of-the-art NO_X combustion controls; (3) fully operating existing SNCRs, including both optimizing NO_X removal by existing operational SNCRs and turning on and optimizing existing idled SNCRs; (4) installing new SNCRs; and (5) installing new SCRs. Finally, for each of these combustion and post combustion technologies identified, EPA evaluated whether emissions reduction potential from generation shifting at that representative dollar per ton level was appropriate at this Step. Shifting generation to lower NO_X emitting or zero-emitting EGUs may occur in response to economic factors. As the cost of emitting NO_X increases, it becomes increasingly cost-effective for units with lower NO_X rates to increase generation, while units with higher NO_X rates reduce generation. Because the cost of generation is unit-specific, this generation shifting occurs incrementally on a continuum. For the reasons explained in the following sections and supported by technical information provided in the EGU NO_X Mitigation Strategies Final Rule TSD included in the docket for this final rule, the EPA determined that for the regional, multi-state scale of this rulemaking, only EGU NO_X emissions controls 1 and 3 are possible for the 2023 ozone season (fully operating existing SCRs and SNCRs). The EPA finds that it is not possible to install state-of-the-art NO_X combustion controls by the 2023 ozone season on a regional scale; those controls are assumed to be available by the beginning of the 2024 ozone season. All cost values discussed in the rest of the section for EGUs are in 2016 dollars.

a. Optimizing Existing SCRs

Optimizing ( i.e., turning on idled or improving operation of partially operating) existing SCRs can substantially reduce EGU NO_X emissions quickly, using investments that have already been made in pollution control technologies. With the promulgation of the CSAPR Update and the Revised CSAPR Update, most operators in the covered states improved their SCR performance and have continued to maintain that level of improved operation. However, this optimized SCR performance was not universal and not always sustained. Between 2017 and 2020, as the CSAPR Update ozone-season NO_X allowance price declined, NO_X emissions rates at some SCR-controlled EGUs increased. For example, power sector data from 2019 revealed that, in some cases, operating units had SCR controls that had been idled or were operating partially, and therefore suggested that there remained emissions reduction potential through optimization. The EPA determined in the Revised CSAPR Update that optimizing SCRs was a readily available approach for EGUs to reduce NO_X emissions in the 12 states addressed by a FIP in that rulemaking. Noticeable improvements in emissions rates at units with SCRs during the 2021 and 2022 compliance period further affirm the ability of sources to quickly implement this mitigation strategy and to realize emissions reductions from doing so. This emissions reduction measure is currently available at EGUs across the broader geography affected in this final rulemaking (including in states not previously affected by the Revised CSAPR Update). The EPA thus determines that SCR optimization, of both idled and partially operating controls, is a viable mitigation strategy for the 2023 ozone season.

The EPA estimates a representative marginal cost of optimizing SCR controls to be approximately $1,600 per ton, consistent with its estimation in the Revised CSAPR Update for this technology. EPA's EGU NO_X Mitigation Strategies Final Rule TSD for this rule describes a range of cost estimates for this technology noting that the costs are frequently lower than—and for the majority of EGUs, significantly lower than—this representative marginal cost. While the costs of optimizing existing, operational SCRs include only variable costs, the cost of optimizing SCR units that are currently idled considers both variable and fixed costs of returning the control into service. Variable and fixed costs include labor, maintenance and repair, parasitic load, and ammonia or urea for use as a NO_X reduction reagent in SCR systems. Depending on a unit's control operating status, the representative cost at the 90th percentile unit (among the relevant fleet of coal units with SCR covered in this rulemaking) ranges between $900 and $1,700 per ton. The EPA performed an in-depth cost assessment for all coal-fired units with SCRs and found that for the subset of SCRs that are already partially operating, the cost of optimizing is often much lower than $1,600 per ton and is often under $900 per ton. The EPA anticipates the vast majority of realized cost for compliance with this strategy to be better reflected by the $900 per ton end of that range (reflecting the 90th percentile of EGUs optimizing SCRs that are already partially operating) because this circumstance is considerably more common than EGUs that have ceased operating their SCR. This cost distinction is reflected in the EPA's RIA cost estimates. When representing the cost of optimization here, the EPA uses the higher value to reflect both optimization of partially operating and idled controls. EPA's analysis of this emissions control is informed by the latest engineering modeling equations used in EPA's IPM platform. These cost and performance equations were recently updated in the summer of 2021 in preparation for this rule, and subsequently evaluated for the final rule in 2022 and determined to still be appropriate. The description and development of the equations are documented in EGU NO_X Mitigation Strategies Final Rule TSD and accompanying documents. They are also implemented in an interactive spreadsheet tool called the Retrofit Cost Analyzer and applied to all units in the fleet. These materials are available in the docket for this action.

The EPA is using the same methodology to identify SCR performance as it did in the Revised CSAPR Update. To estimate EGU NO_X reduction potential from optimizing, the EPA considers the difference between the non-optimized NO_X emissions rates and an achievable operating and optimized SCR NO_X emissions rate. To determine this rate, EPA evaluated nationwide coal-fired EGU NO_X ozone season emissions data from 2009 through 2019 and calculated an average NO_X ozone season emissions rate across the fleet of coal-fired EGUs with SCR for each of these eleven years. The EPA found it prudent to not consider the lowest or second-lowest ozone season NO_X emissions rates, which may reflect SCR systems that have all new components ( e.g., new layers of catalyst). Data from these systems are potentially not representative of ongoing achievable NO_X emissions rates considering broken-in components and routine maintenance schedules. Considering the emissions data over the full time period from 2009–2019 results in a third-best rate of 0.079 pounds NO_X per million British thermal units (lb/mmBtu). Therefore, consistent with the Revised CSAPR Update, where EPA identified 0.08 lb/mmBtu as a reasonable level of performance for units with optimized SCR, the EPA finalizes a rate of 0.08 lb/mmBtu as the optimized rate for this rule. The EPA notes that half of the SCR-controlled EGUs achieved a NO_X emissions rate of 0.064 lb/mmBtu or lower over their third-best entire ozone season. Moreover, for the SCR-controlled coal units that the EPA identified as having a 2021 emissions rate greater than 0.08 lb/mmBtu, the EPA verified that in prior years, the majority (more than 90 percent) of these same units had demonstrated and achieved a NO_X emissions rate of 0.08 lb/mmBtu or less on a seasonal or monthly basis. This further supports EPA's determination that 0.08 lb/mmBtu reflects a reasonable emissions rate for representing SCR optimization at coal steam units in identifying uniform control stringency. This emissions rate assumption of 0.08 lb/mmBtu reflects what those units would achieve on average when optimized, recognizing that individual units may achieve lower or higher rates based on unit-specific configuration and dispatch patterns. Units historically performing at, or better, than this rate of 0.08 lb/mmBtu are assumed to continue to operate at that prior performance level.

Given the magnitude and duration of the air quality problems addressed by this rulemaking, the EPA also applied the same methodology to identify a reasonable level of performance for optimizing existing SCRs at oil- and gas-fired steam units and simple cycle units (for which EPA determined that a 0.03 lb/mmBtu emissions rate reflected SCR optimization) as well as at combined-cycle units (for which the EPA determined that a 0.012 lb/mmBtu emissions rate reflected SCR optimization).

The EPA evaluated the feasibility of optimizing idled SCRs for the 2023 ozone season. Based on industry past practice, the EPA determined that idled controls can be restored to operation quickly ( i.e., in less than 2 months). This timeframe is informed by many electric utilities' previous long-standing practice of utilizing SCRs to reduce EGU NO_X emissions during the ozone season while putting the systems into protective lay-up during the non-ozone season months. For example, this was the long-standing practice of many EGUs that used SCR systems for compliance with the NO_X Budget Trading Program. It was quite typical for SCRs to be turned off following the end of the ozone season control period on September 30. These controls would then be put into protective lay-up for several months of non-use before being returned to operation by May 1 of the following ozone season. Therefore, the EPA believes that optimization of existing SCRs is possible for the portion of the 2023 ozone season covered under this final rule. The recent successful implementation of this strategy for the Revised CSAPR Update Rule, and corresponding fast improvement in SCR performance rates at units with optimization potential, provides further supporting evidence of the viability of this timeframe.

The vast majority of SCR-controlled units (nationwide and in the 23 linked states for which EPA is issuing a FIP for EGUs) are already partially operating these controls during the ozone season based on reported 2021 and 2022 emissions rates. Notably, the higher ozone season NO_X allowance price observed in 2022 resulted in more units operating their controls closer to their potential and bringing collective emissions from those 12 states closer to the 2023 emissions budgets for those states in this final rule, accordingly. Existing SCRs operating at partial capacity still provide functioning, maintained systems that may only require an increased chemical reagent feed rate ( i.e., ammonia or urea) up to their design potential and catalyst maintenance for mitigating NO_X emissions; such units may require increased frequency or quantity of deliveries, which can be accomplished within a few weeks. In many cases, EGUs with SCR have historically achieved more efficient NO_X removal rates than their current performance and can therefore simply revert to earlier operation and maintenance plans that achieved demonstrably better SCR performance.

In the 12 states subject to this control stringency in the Revised CSAPR Update, the EPA observed significant immediate-term improvements in SCR performance in the first ozone season following finalization of that rule, as evidenced in particular by the sharp drop in emissions rate at Miami Fort unit 7 ( see EGU NO_X Mitigation Strategies Final Rule TSD). For instance, in June of 2021—within months of the Revised CSAPR Rule being finalized—Miami Fort Unit 7 and Unit 8 (which had substantial SCR optimization potential) were able to reach levels of 0.07 lb/mmBtu of NO_X (a greater than 50 percent reduction from where they had operated the prior year during the same month). Such empirical data further illustrates the viability of this mitigation strategy for the 2023 control period in response to this rule.

Comment: EPA received comments supporting the 0.08 lb/mmBtu emissions rate as achievable and, according to some commenters, underestimate the control's potential. Some of these commenters went on to provide their own analysis demonstrating that the 0.08 lb/mmBtu was achievable not only on average for the non-optimized fleet, but also for these individual units and that the resulting state emissions budgets were likewise achievable. Some commenters suggested that the rate should be lower and premised on EPA using the first- or second-best year instead of the third best year of SCR performance. Some commenters observed that using the same methodology, but omitting SCR units that have since retired, could deliver an even lower SCR performance benchmark rate.

Response: The EPA notes that updating the inventory of coal-fired EGUs to reflect recent retirements and to include data reported since 2019 ( e.g., 2009–2021) would provide a lower value of 0.071 lb/mmBtu. However, EPA acknowledges that 2020 operational data included impacts from COVID–19 pandemic shutdowns (such as atypical electricity demand patterns) which complicate interpretations of typical EGU emissions performance. Additionally, EPA believes that in this context, a unit's retirement in 2020 or 2021 does not obviate the usefulness of its prior SCR operational data for assessing the emissions control performance of other existing SCRs across the fleet. Consequently, EPA is continuing to use the same value of the 0.08 lb/mmBtu emissions rate calculated from the 2009–2019 data set identified at the time of the final Revised CSAPR Update Rule in this rulemaking. EPA's analysis focuses on the third best ozone season average rate because EPA believes that the first- or second-best rate, consistent with its CSAPR Update final rule and in the Revised CSAPR Update, could give undue weight to the emissions control performance of new SCRs in their first year of service and their corresponding newer SCR components. It does not necessarily reflect achievable ongoing NO_X emissions rates at relatively older SCRs. The third-lowest season was selected because it represents a time when the unit was most likely consistently and efficiently operating its SCR in a manner representative of sustained future operation.

Comment: Other commenters suggested that EPA should apply a higher NO_X emissions rate than 0.08 lb/mmBtu to existing SCR at coal EGUs premised on considerations such as: a generally reduced average capacity factor for coal units in recent years, the age of the boiler, coal rank (bituminous or subbituminous), or other unit-specific considerations that commenters claim make the 0.08 lb/mmBtu rate unattainable for a specific unit.

Response: EPA did not find sufficient justification to apply a higher average emissions rate than 0.08 lb/mmBtu. EPA found that some commenters were misunderstanding or misconstruing both EPA's assumption and implementation mechanism as a unit-level requirement for every SCR-controlled unit instead of a reflection of a fleet-wide average based on a third-best rate. The commenters' observation—that 0.08 lb/mmBtu may be difficult for some units to achieve or may not be a preferred compliance strategy for a given unit given its dispatch levels—does not contradict EPA's assumption, but rather supports its methodology and assumptions. As EPA pointed out in the proposed rule, this fleet-level emissions rate assumption of 0.08 lb/mmBtu for non-optimized units reflects, on average, what those units would achieve when optimized. Some of these units may achieve rates that are lower than 0.08 lb/mmBtu, and some units may operate above that rate based on unit-specific configuration and dispatch patterns. In other words, EPA is using this assumption as the average performance of a unit that optimizes its SCR, recognizing that heterogeneity within the fleet will likely lead some units to overperform and others to underperform this rate. Moreover, a review of unit-specific historical data indicates that this is a reasonable assumption: not only has the group of units with SCR optimization potential demonstrated they can perform at or better than the 0.08 lb/mmBtu rate on average, over 90 percent of the individual units in this group have already met this rate on a seasonal and/or monthly basis based on their reported historical data.

Additionally, EPA's examination of units experiencing SCR performance deterioration included notable instances of poor NO_X control at increased capacity factors. As an example, Miami Fort Unit 7 had considerably more hours of operation at a 70 to 79 percent capacity factor in 2019 compared to previous years. However, Miami Fort Unit 7's ozone-season NO_X emissions rate substantially increased in 2019 compared to previous years. This SCR performance deterioration runs counter to the notion that an increase in emissions rates is purely driven by reduced capacity factor, as suggested by commenters. This substantial deterioration in the median emissions rate performance is observable even when comparing specific hours in 2019 to specific hours in prior years when the unit operated in the same 70 to 79 percent capacity factor range. In fact, in 2019 the unit experienced notable emissions rate increases from prior years across multiple capacity factor ranges as low as 40 percent to as high as 80 percent. This type of data indicates instances where the increase in emissions rate (and emissions) is not necessitated by load changes but is more likely due to the erosion of the existing incentive to optimize controls ( i.e., the ozone-season NO_X allowance price has fallen so low that unit operators find it more economic to surrender additional allowances instead of continuing to operate pollution controls at an optimized level).

EPA observed this pattern in other units identified in this rulemaking as having significant SCR optimization emissions reduction potential. In the accompanying Emissions Data TSD for the supplemental notice that EPA recently released in a proceeding to address a recommendation submitted to EPA by the Ozone Transport Commission under CAA section 184(c), EPA noted, “In their years with the lowest average ozone season NO_X emissions rates in this analysis, these EGUs had relatively low NO_X emissions rates at mid- and high-operating levels; moreover, there was little variability in NO_X emissions rates at these operating levels. However, during the 2019 ozone season, these EGUs had higher NO_X emissions rates and greater variability in NO_X emissions rates across operating levels than in the past, particularly at mid-operating levels.” That hourly data analysis, included in this docket, controls for operating level changes and still finds there to be instances across multiple SCR-controlled units where hourly emissions rates are increasing even when compared to the same load levels in previous years.

Some commenters have alleged that in recent years coal-fired EGUs have declined in capacity factor and that SCR performance declines at those lower operating levels. However, hourly data indicate that maintaining consistent SCR performance at lower capacity factors is possible. For example, the unit-level performance data in Figure 2 to section VI.B of this document show the emissions rate at a coal-fired EGU with existing SCR staying relatively low (consistent with our optimization assumption of 0.08 lb/mmBtu) and stable across a wide range of capacity factors.

Furthermore, most recent data from 2022 illustrates that cycling units do have the ability to adjust cycling patterns in a manner that enables them to maintain a lower emissions rate throughout the season while still achieving a load cycling pattern at the unit. For example, the SCR-controlled Conemaugh Unit 2 in Pennsylvania adjusted operating patterns in 2022 to have a slightly higher minimum load in most hours (maintaining a range of 550 MW–900 MW for most hours as opposed to 450 MW–900 MW observed in 2021). This change in minimum load, and corresponding minimum operating temperature, enabled the unit to maintain emissions rates in the 0.05 lb/mmBtu to 0.10 lb/mmBtu range for most of the 2022 season (as opposed to NO_X emissions rates that regularly exceeded 0.25 lb/mmBtu in the 2021 season). This 2022 improvement in SCR operation occurred during a period when allowance prices increased relative to prior years, creating an incentive for potential emissions reductions through SCR optimization.

Comment: EPA also received comment suggesting it should deviate from its approach in the CSAPR Update of using a nationwide data set of all SCR controlled coal units to establish a third best year, and instead limit the dataset to either just the covered states, or—in the case of some commenters—just to the baseline years of those units at which EPA is identifying optimization potential. They claim the current methodology may capture extremely efficient SCR performance years at the best performing units and that level of performance may not be available at all units with optimization potential. These commenters also disagree with the EPA finding that SCRs can consistently maintain a 0.08 lb/mmBtu rate over time.

Response: EPA reviewed the data and its methodology and evaluated it against its intention to identify a technology-specific representative emissions rate for SCR optimization. In doing so, EPA did not identify any need to make the suggested change. EPA is interested in the performance potential of a technology, and a larger dataset provides a superior indication of that potential as opposed to a smaller, state-limited dataset. Moreover, EPA's use of the third best year (as opposed to best) from its baseline period results in an average optimization level that is robust to the commenters' concern that EPA should not overstate the fleetwide representative optimization level. Prior experience with EPA's methodology and program has borne out empirical evidence of its reasonableness. In both the CSAPR Update and in Revised CSAPR Update rule, EPA appropriately relied on the largest dataset possible ( i.e., nationwide) to derive technology performance averages that it then applied respectively to the CSAPR Update 22-state region and the Revised CSAPR Update's 12-state region. EPA repeats that successful approach in this rule. Finally, as noted in the preceding paragraphs, in affirming the reasonableness of this approach, EPA examined the historical reported data (pre-2021) for the units in the states with SCR optimization potential and found the nationwide derived average appropriate and consistent with demonstrated capability and performance of units within those states. That is, the vast majority of units to which this resulting emissions rate assumption was being applied had demonstrated the ability to achieve this rate in some prior year for an extended monthly or seasonal basis. This information is discussed further in the EGU NO_X Mitigation Strategies Final Rule TSD in the docket.

Comment: Some commenters suggested the price of SCR optimization is higher than the $1,600 per ton figure proposed due to current market conditions for aqueous ammonia or other input prices.

Response: EPA provides a representative cost for this mitigation technology which is anticipated to reflect the cost, on average, throughout the compliance period for the rule. While there may be volatility in the market during that period where the price falls above or below the single representative threshold value, EPA's EGU NO_X Mitigation Strategies Final Rule TSD explains how the representative cost is derived and is inclusive of consultation and vetting by third party air pollution control consulting groups. Commenters did not demonstrate that observed 2021 elevated prices amid market volatility would continue into the future compliance periods discussed in this rule. Moreover, the selection of the mitigation technology is reflective of a variety of factors including reduction potential and air quality impact. A higher cost (commenter suggests up to $3,800 per ton) would not change EPA's determination that optimizing already existing SCRs is an appropriate mitigation strategy for Step 3 emissions reduction analysis in this rulemaking as it would remain one of the most widely available, widely practiced, and lowest cost mitigation measures with meaningful downwind air quality benefit. Appendix B of the EGU NO_X Mitigation Strategies Final Rule TSD further addresses commenters' concerns as it provides a variety of sensitivities showing cost per ton levels under a variety of different input assumptions (including higher material and reagent cost). It supports the continued inclusion of this technology in the rule even in the event that higher reagent costs extend into compliance years.

Comment: While many commenters supported the feasibility of 2023 ozone-season implementation by noting the “immediate availability” of SCR optimization, other commenters argued that the engineering, procurement, and other steps required for SCR optimization were not feasible given the anticipated limited window between rule finalization and the start of the 2023 ozone season.

Response: There is ample evidence of units restoring their optimal performance within a two-month timeframe. Not only do units reactivate SCR performance level at the start of an ozone-season when tighter emissions limits begin, but unit-level data also shows instances where sources have demonstrated the ability to quickly alter their emissions rate within an ozone-season and even within the same day in some cases. Moreover, this emissions control is familiar to sources and was analyzed and included in the Revised CSAPR Update emissions budgets finalized in 2021 and the CSAPR Update emissions budgets finalized in 2016. With this experience, and notice through the March 2022 proposed rule, as well as over two months from final rule to effective date, the viability of this emissions control for the 2023 ozone season is consistent with the 2-week to 2-month timeframe that EPA identified as reasonable in the CSAPR Update, Revised CSAPR Update, and in this rulemaking. Similar to prior rules, commenters provide some unit-level examples where it has taken longer. Also similar to those prior rules, EPA does not find those unit-level examples compelling in the context of its fleet average assumptions and in the implementation context of a trading program which provides compliance alternatives in the event a specific unit prefers more time to implement a given control measure. As noted in Wisconsin, “. . . all those anecdotes show is that installation can drag on when companies are unconstrained by the ticking clock of the law.” 938 F.3d at 330.

b. Installing State-of-the-Art NO_X Combustion Controls

The EPA estimates that the representative cost of installing state-of-the-art combustion controls is comparable to, if not notably less than, the estimated cost of optimizing existing SCR (represented by $1,600 per ton). State-of-the-art combustion controls such as low-NO_X burners (LNB) and over-fire air (OFA) can be installed or updated quickly and can substantially reduce EGU NO_X emissions. Nationwide, approximately 99 percent of coal-fired EGU capacity greater than 25 MW is equipped with some form of combustion control; however, the control configuration or corresponding emissions rates at a small portion of those units (including units in those states covered in this action) indicate they do not currently have state-of-the-art combustion control technology. For this rulemaking, the Agency re-evaluated its NO_X emissions rate assumptions for upgrading existing combustion controls to state-of-the-art combustion control. The EPA is maintaining its determination that NO_X emissions rates of 0.146 to 0.199 lb/mmBtu can be achieved on average depending on the unit's boiler configuration, and, once installed, reduce NO_X emissions at all times of EGU operation.

These assumptions are consistent with the Revised CSAPR Update. They are further discussed in the EGU NO_X Mitigation Strategies Final Rule TSD. In particular, the EPA is finalizing, as proposed, the application of the 0.199 lb/mmBtu emissions rate assumption for both boiler types (tangentially and wall fired). EPA's analysis calculated average emissions rates of 0.199 lb/mmBtu for combustion controls on dry bottom wall fired units and 0.146 lb/mmBtu for tangentially fired units. However, many of the likely impacted units burn bituminous coal, and the 0.146 lb/mmBtu nationwide average for tangentially-fired (inclusive of subbituminous units) appears to be below the demonstrated emissions rate of state-of-the-art combustion controls for bituminous coal units of this boiler type. Therefore, EPA's assignment of a 0.199 lb/mmBtu emissions rate for combustion controls at all affected unit types is robust to current and future coal choice at a unit.

The EPA has previously examined the feasibility of installing combustion controls and found that industry had demonstrated ability to install state-of- the-art LNB controls on a large unit (800 MW) in under six months when including the pre-installation phases (design, order placement, fabrication, and delivery). In prior rules, the EPA has documented its own assessment of combustion control timing installation as well as evaluated comments it received regarding installation of combustion controls from the Institute of Clean Air Companies. Those comments provided information on the equipment and typical installation time frame for new combustion controls, accounting for all steps. To date, EPA has found it generally takes between 6–8 months on a typical boiler—covering the time through bid evaluation through start-up of the technology. The deployment schedule is repeated here as:

• 4–8 weeks—bid evaluation and negotiation
• 4–6 weeks—engineering and completion of engineering drawings
• 2 weeks—drawing review and approval from user
• 10–12 weeks—fabrication of equipment and shipping to end user site
• 2–3 weeks—installation at end user site
• 1 week—commissioning and start-up of technology

Given the referenced timeframe of approximately 6 to 8 months to complete combustion control installation in the region, the EPA is finalizing that installation of state-of-the-art combustion controls is a readily available approach for EGUs to reduce NO_X emissions by the start of the 2024 ozone season. More details on these analyses can be found in the EGU NO _X Mitigation Strategies Final Rule TSD.

The cost of installing state-of-the-art combustion controls per ton of NO_X reduced is dependent on the combustion control type and unit type. The EPA estimates the cost per ton of state-of-the-art combustion controls to be $400 per ton to $1,200 per ton of NO_X removed using a representative capacity factor of 85 percent. This cost fits well within EPA's representative cost threshold observed for SCR optimization and combustion controls (of $1,600 per ton) which would accommodate combustion control upgrade even under scenarios where a lower capacity factor is assumed. 99 percent of units have some form of combustion controls, indicating the widespread cost-effectiveness of this control. See the EGU NO _X Mitigation Strategies Final Rule TSD for additional details.

At proposal EPA assumed that emissions reductions from combustion control upgrades at affected EGUs in states subject to the Revised CSAPR Update program could occur by 2023 given that those EGUs may have already begun pursuing such upgrades in response to that previous rule. However, EPA does not have data to confirm that presumption, and hence EPA is determining in this final rule that combustion control upgrades for all affected EGUs, regardless of whether they were previously subject to the Revised CSAPR Update program, should be considered available by the 2024 ozone season, consistent with the deployment schedule noted in this section.

Comment: Some commenters suggested that EPA, in its modeling for the proposed rule, overestimated the ability of combustion control technologies to achieve very low NO_X emissions rates. The commenters claim EPA's assumptions are derived from projected NO_X emissions rates based on ideal circumstances for NO_X emissions reductions, including combinations of fuel composition and unit design that are not typical and should not be extrapolated to the national inventory.

Response: EPA's emissions performance rate for state-of-the-art combustion controls is derived from historical data and takes both boiler type and coal choice into account. EPA reviewed historical data and identified the average emissions rates for units with this technology already in place. It segmented this analysis by boiler type (dry-bottom wall-fired boiler and tangentially-fired, and further segmented by coal rank to assess the average performance among these varying parameters. As explained in the EGU NO _X Mitigation Strategies Final Rule TSD, EPA chose an emissions rate for which it verified accommodated ( i.e., was greater than or equal to) the average performance rate identified above for each boiler configuration with state-of-the-art combustion controls and resulted in reductions consistent with the technology's assumed percent reduction potential when applied to this subset of units. It also assessed whether the rate had been demonstrated by both subbituminous and bituminous coal units with state-of-the-art combustion controls. EPA further assessed the percent reduction that achieving this rate would require from the specific segment of the fleet identified as having this mitigation measure available. Here too, EPA found that the effective percent reduction for the identified fleet (inclusive of their existing coal rank choice) is well within the historical performance range for this technology. Therefore, EPA is finalizing the combustion control upgrade performance assumption of 0.199 lb/mmBtu as appropriate representative average performance rate for this technology and robust to different boiler types and coal ranks.

c. Optimizing Already Operating SNCRs or Turning on Idled Existing SNCRs

Optimizing already operating SNCRs or turning on idled existing SNCRs can also reduce EGU NO_X emissions quickly, using investments in pollution control technologies that have already been made. Compared to no post-combustion controls on a unit, SNCRs can achieve a 25 percent reduction on average in EGU NO_X emissions (with sufficient reagent). They are less capital intensive but less efficient at NO_X removal than SCRs. These controls are in use to some degree across the U.S. power sector. In the 22 linked states with EGU reductions identified in this final rule, approximately 11 percent of coal-fired EGU capacity is equipped with SNCR. Recent power sector data suggest that, in some cases, SNCR controls have been operating less in 2021 relative to performance in prior years. For instance, EPA reviewed the last five years of performance data for all the units with SNCR optimization potential in its Engineering Analysis. It found that in 2021—the most recent year reviewed—that the weighted average ozone season emissions rate for these units was higher than the prior three years (indicating some deterioration in average performance). Moreover, a unit level review illustrated that 80 of the 107 units had performed better in a prior year by an average of 13 percent—indicating substantial optimization potential.

The EPA determined that optimizing already operating SNCRs or turning on idled SNCRs is an available approach for EGUs to reduce NO_X emissions, has similar implementation timing to restarting idled SCR controls (less than 2 months for a given unit), and therefore could be implemented in time for the 2023 ozone season. In this final rule, the EPA is determining that this emissions control measure is available beginning in the 2023 ozone season.

Using the Retrofit Cost Analyzer described in the EGU NO _X Mitigation Strategies Final TSD, the EPA estimates a representative cost of optimizing SNCR ranging from approximately $1,800 per ton (for partially operating SNCRs) to $3,900 per ton (for idled SNCRs). For existing SNCRs that have been idled, unit operators may need to restart payment of some fixed and variable operating costs including labor, maintenance and repair, parasitic load, and ammonia or urea. The EPA determined that the majority of units with existing SNCR optimization potential were already partially operating their controls. Therefore, the EPA finalizes a representative cost of $1,800 per ton for SNCR optimization as this value best reflects the circumstances of the majority of the affected EGUs with SNCR.

d. Installing New SNCRs

The EPA evaluated potential emissions reductions and associated costs from retrofitting EGUs with new SNCR post-combustion controls at steam units lacking such controls, which can achieve a 25 percent NO_X reduction on average. New SNCR technology provides owners with a relatively less capital-intensive option for reducing NO_X emissions compared to new SCR technology, albeit at the expense of higher operating costs on a per-ton basis and less total emissions reduction potential. SNCR is more widely observed on relatively smaller coal units given its low capital/variable cost ratio. The average capacity of a coal unit with SNCR is half the size of the average capacity of coal unit with SCR. Given these observations, the EPA identifies this technology as an emissions reduction measure for coal units less than 100 MW lacking post-combustion NO_X control technology. As described in the EGU NO _X Mitigation Strategies Final Rule TSD, the EPA estimated that $6,700 per ton reflects a representative SNCR retrofit cost level for these units.

For this rulemaking, EPA is not considering SNCR installation timing unto itself but is instead considering how long eligible EGUs may need to adopt either SNCR or SCR as a post-combustion control measure. SNCR installations generally have shorter project installation timeframes relative to other post-combustion controls. The time for engineering review, contract award, fabrication, delivery, and hookup is as little as 16 months including pre-contract award steps for an individual power plant installing controls on more than one boiler. However, SNCR retrofits have less pollution reduction potential than SCRs, and as explained further in the next section, the EPA is identifying the retrofit of new SCR rather than SNCR as a strategy for larger steam units due to this lower removal efficiency. This approach respects empirical evidence that larger coal-fired EGUs which installed post-combustion NO_X control technology have overwhelmingly chosen SCRs over SNCRs. Even for smaller units less than 100 MW identified as potential candidates for SNCR technology, the EPA does not want to preclude those units from pursuing SCR in lieu of SNCR.

Therefore, in this final rule the EPA defines the availability of emissions reductions from post-combustion control installation to be in 2026, the same period as the start of SCR-based reductions becoming available, to allow enough time for eligible EGUs to choose between SCR or SNCR. SNCR installation shares similar implementation steps with and also need to account for the same regional factors as SCR installations, which are described in the next section. While the EPA is determining that at least 16 months would be needed to complete all necessary steps of SNCR development and installation, an eligible EGU choosing new SCR instead would require installation timing of 36 to 48 months. EPA believes its finalized joint timing considerations for post-combustion control retrofits (SNCR and SCR) are justified given that post-combustion control retrofit decisions are subject to unit-specific economic and engineering factors and are sensitive to operator compliance strategy choices with respect to multiple regulatory requirements.

Comment: Some commenters argued that post-combustion control timing assumptions (SCR and SNCR) should be decoupled, which could result in the EPA using the 16-month time frame specific to SNCR installation to require emissions reductions related to new SNCR installations by the 2025 ozone season.

Response: The EPA does not agree that decoupling SCR and SNCR timing consideration is justified in the context of this final rule's emissions control program for EGUs. Approximately 1,000 tons of emissions reduction potential are estimated for the small coal EGUs deemed eligible for SNCR retrofit. The incentives provided through the implementation of this rule's trading program will encourage these EGUs to determine and adopt emissions reduction measures (including SNCR or SCR) as soon as possible to reduce their allowance holding compliance burden. By scheduling SNCR-related emissions reductions potential for the 2026 ozone season, the EPA preserves the opportunity for considerably superior emissions reduction potential from these EGUs should they select SCR retrofit instead, while still requiring post-combustion control emissions reduction potential ahead of the next attainment date.

Comment: Some commenters argued that the upper range of SNCR NO_X removal performance (40 percent) referenced by EPA is optimistic for many boilers.

Response: EPA evaluated both actual performance and engineering literature regarding SNCR retrofit technology and found both sources supported the range of reduction estimates cited by EPA. (Refer to the EGU NO _X Mitigation Strategies Final Rule TSD in the docket for this rulemaking for additional information.) Moreover, for purposes of calculating state budgets, EPA assumes 25 percent reduction from this technology—not 40 percent—which reflects a value well within the range of documented performance for this technology. Remaining comments on SNCR performance potential are addressed in the RTC Document and in the EGU NO _X Mitigation Strategies Final Rule TSD.

e. Installing New SCRs

Selective Catalytic Reduction (SCR) controls already exist on over 66 percent of the coal fleet in the linked states that are subject to a FIP in this rulemaking. Nearly every pulverized coal unit larger than 100 MW built in the last 30 years has installed this control, which is generally required for Best Available Control Technology (BACT) purposes. Other than circulating fluidized bed coal units which can achieve a comparably low emissions rate without this technology, the EPA identifies this emissions reduction measure for coal steam units greater than or equal to 100 MW. SCR is widely available for existing coal units of this size and can provide significant emissions reduction potential, with removal efficiencies of up to 90 percent. The EPA limited its consideration of SCR technology to steam units greater than or equal to 100 MW. The costs for retrofitting a plant smaller than 100 MW with SCR increase rapidly due to a lack of economies of scale.

The amount of time needed to retrofit an EGU with new SCR extends beyond the 2023 ozone season. Similar to the SNCR retrofits discussed in this section, the EPA evaluated potential emissions reductions and associated costs from this control technology, as well as the impacts and need for this emissions control strategy, at the earliest point in time when their installation could be achieved. EPA notes that it has previously determined in the context of ozone transport that regional scale implementation of SCRs at numerous EGUs is achievable in 36 months. See63 FR 57356, 57447–50 (October. 27, 1998). However, since that time, the EPA has found up to 36–48 months to be a more appropriate installation timeframe for regionwide actions when the EPA is evaluating multiple installations at multiple locations.

In the past, the EPA has found the amount of time to retrofit a single EGU with new SCR, depending on the regulatory program under which such control may be required, may vary between approximately 2 and 4 years depending on site-specific engineering considerations and on the number of installations being considered. This includes steps for engineering review, construction permit, operating permit, and control technology installation (including fabrication, pre hookup, control hookup, and testing). EPA's assessment of installation procedures suggests as little as 21 months may be needed for a single SCR at an individual plant and 36 months at a single plant with multiple boilers. EPA's assessment of units with SCR retrofit potential indicate the majority fall into this first classification, i.e., a single SCR at a power plant.

While EPA finds that 36 months is a possible time frame for SCR installation at individual units or plants, the total of nearly 31 GW of coal capacity with SCR retrofit potential and 19 GW of oil/gas steam capacity with SCR retrofit potential within the geographic footprint of the final rule is a scale of retrofit activity that is not demonstrated to have been achieved within a three-year span based on data from the past two decades. Given that some of the assumed SCR retrofit potential occurs at plants with multiple units identified with retrofit potential, and given the total volume of SCR retrofit capacity being implemented across the region, EPA is allowing in this final rule between 36 to 48 months, consistent with the regional time frame discussed for SCR retrofit in prior rules, for the full implementation of reductions commensurate with this volume of SCR retrofit capacity, as described further in section VI.A of this document.

The Agency examined the cost for retrofitting a coal unit with new SCR technology, which typically attains controlled NO_X rates of 0.05 lb/mmBtu or less. These updates are further discussed in the EGU NO_X Mitigation Strategies Final Rule TSD. Based on the characteristics of coal units of 100 MW or greater capacity that do not have post-combustion

NO_X control technology, the EPA estimated a weighted-average representative SCR cost of $11,000 per ton.

The 0.05 lb/mmBtu emissions rate performance assumption for new SCR retrofits is supported by historical data and third party independent review by pollution control engineering and consulting firms. The EPA first examined unit-level emissions rate data for coal-fired units that had a relatively recent SCR installation (within the last 10 years). The best performing 10 percent of these SCRs were demonstrating seasonal emissions rates of 0.036 lb/mmBtu during this time.

While the EPA identified the 0.05 lb/mmBtu performance assumption consistent with historical data, these performance levels are also informed and consistent with the Agency's IPM modeling assumptions used for more than a decade. These modeling assumptions are based on input from leading engineering and pollution control consulting entities. Most recently, these data assumptions were affirmed and updated in the summer of 2021 and included in the docket for this rulemaking. The EPA relies on a global firm providing engineering, construction management, and consulting services for power and energy with expertise in grid modernization, renewable energy, energy storage, nuclear power, and fossil fuels. Their familiarity with state-of-the art pollution controls at power plants derives from experience providing comprehensive project services—from consulting, design, and implementation to construction management, commissioning, and operations/maintenance. This review and update supported the 0.05 lb/mmBtu performance assumption as a representative emissions rate for new SCR across coal types.

The EPA performed an assessment for oil/gas steam units in which it evaluated the nationwide performance of those units with SCR technology. For these units, the EPA tabulated EGU NO_X ozone season emissions data from 2009 through 2021 and calculated an average NO_X ozone season emissions rate across the fleet of oil- and gas-fired EGUs with SCR for each of these years. The EPA identified the third lowest year which yielded an SCR performance rate of 0.03 lb/mmBtu as representative of performance for this retrofit technology applied to this type of EGU. Next, the EPA evaluated the emissions and operational characteristics for the existing oil/gas steam fleet lacking SCR technology. EPA's analysis indicated that the majority of reduction potential (approximately 76 percent) from these units occurred at units greater than or equal to 100 MW and that were emitting more than 150 tons per ozone season ( i.e., approximately 1 ton per day). Moreover, the cost of reductions for units falling below these criteria increased significantly on a dollar per ton basis. Therefore, the EPA identified the portion of the oil/gas steam fleet meeting these criteria ( i.e., greater than or equal to 100 MW and emitting more than 150 tons per ozone season) as representative of the SCR retrofit reduction potential. For this segment of the oil/gas steam units lacking post-combustion NO_X control technology, the EPA estimated a weighted-average representative SCR cost of $7,700 per ton.

Comment: Some commenters disagreed with EPA's proposed 36-month timeframe for SCR retrofit. These commenters noted that, while possible at the unit or plant level, the collective volume of SCR installation occurring in a limited region of the country would not be possible given the labor constraints, supply constraints, and simultaneous outages necessary to complete SCR retrofit projects on such a schedule. They noted that achieving such a timeframe against a backdrop of such challenging circumstances is unprecedented and that EPA's assumptions ignore that many of the remaining unretrofitted coal units reflect more site-specific challenges than those that were already retrofitted on a quicker timeframe.

Response: EPA reviewed the comments and is making several changes in this final rule to address some of the concerns identified by the commenters. In particular, EPA found that its own review of historical retrofit patterns as well as technical information submitted by commenters supported commenters' concerns regarding: (1) current and anticipated constraints in labor and supply markets, (2) the potential collective capacity levels of SCR retrofit within 36 months, and (3) possible site-specific complexities at the remaining units without an existing SCR. To address these concerns, EPA is phasing in its SCR installation requirement over a 48-month time frame in this final rule, instead of a 36-month time frame as proposed (see additional detail and discussion in section VI.A.2.a and the EGU NO_X Mitigation Strategies Final Rule TSD). EPA will require half of the reductions associated with SCR installation in 2026 and the other half in 2027. Additionally, EPA is moving the daily backstop rate for these units with identified SCR reduction potential from 2027 to no later than 2030, which defers the increased allowance surrender ratio for emissions above the backstop rate at any outlier units unable to complete the retrofit during that time frame. These adjustments continue to incentivize reductions in NO_X emissions by the attainment date that are consistent with cost-effective SCR controls, but provide more flexibility (both from timing and technology perspective) in how they are procured.

Some commenters requested more than 48 months to install SCR controls based on the collective total volume of SCR retrofit volume identified and past projects that took five or more years. EPA disagrees with these comments and finds that they ignored key aspects of the proposed rule. First, the final rule does not directly require implementation of SCR; rather, it requires reductions commensurate with SCR installations based on a rigorous assessment of SCR retrofit potential. Implementing the reductions through a trading program means that sources in many cases, as suggested by the Regulatory Impact Analysis (RIA), will find alternative, and more economic means, of reducing emissions—including reduced generation and retirements that are already planned based on the age of the unit, decarbonization goals, or compliance with other Federal/state/local regulation compliance dates. Moreover, the additional new generation incentives provided by the Inflation Reduction Act (enacted after the proposed rule) will further increase the pace of new generation replacing some of the older generating capacity identified as having retrofit potential. In short, although EPA identified the total SCR retrofit capacity potential for today's existing fleet and does not premise any reduction requirements of incremental retirements, the announced and planned futures for these units indicates that many will likely retire instead of installing SCR. For the capacity identified at Step 3 which lacks SCR, the planned or projected retirement in place of a retrofit moots the SCR timing for these units. Moreover, it also reduces the demand for associated labor and materials which, in turn, frees up resources for any units proceeding with a SCR retrofit. Therefore, comments which cite labor and supply chain challenges for accommodating the entire fleet capacity identified as having SCR retrofit potential significantly overstate the supply-side challenge—as it ignores the fact that much of this capacity has explicit or expected operation plans that will result in compliance without a retrofit.

Even for sources choosing a SCR retrofit compliance pathway, many of these comments ignore the timing flexibilities of the trading program, which (particularly with the changes to the backstop daily emissions rate in this final rule) allow sources to temporarily comply through means other than SCR retrofit if they experience any site-specific retrofit limitations that increase their time frame. Also, historical examples of SCR retrofit projects that exceeded 48 months in duration do not necessarily demonstrate that such projects are impossible in less than 48 months, but rather that they can extend beyond the timeframe if no requirements or incentives are in place for a faster installation. Some also cite site-specific conditions that resulted an outlier cases of project timing that would not be representative of the conditions expected at future retrofit projects.

Comment: Some stakeholders suggested that EPA's cost estimates of $11,000 per ton are premised on a 15-year book life of the equipment and are therefore too optimistic for units that plan to retire in well under 15 years.

Response: EPA analysis of SCR retrofit cost reflects a representative value for the technology based on a weighted average cost. The underlying data and the discussion in the EGU NO_X Mitigation Strategies Final TSD illustrates that these costs can vary significantly at the unit level based on factors such as the length of time a pollution control technology would be in operation, the capacity factor of the unit ( i.e., how much does it operate), its size or potential to emit, and its baseline emissions rate. The EPA has not in prior transport rulemakings used such factors as justification to excuse any source that is significantly contributing to nonattainment or interfering with maintenance in another state from eliminating that significant contribution as expeditiously as practicable. Unlike under other statutory provisions that may require retrofit of emissions controls on existing sources, such as under CAA section 111(d) or CAA section 169A, there is no remaining useful life factor expressly identified as a justification to relax the requirements of CAA section 110(a)(2)(D)(i)(I). EPA continues to believe that where an emissions control strategy has been identified at Step 3 that is cost-effective on a regional scale and provides meaningful downwind air quality improvement, and is thus appropriately identified as necessary to eliminate significant contribution under the good neighbor provision, it would not be appropriate to allow emissions to continue in excess of those achievable emissions reductions beyond the timeframe for expeditious implementation of reductions as provided under the larger title I structure of the Act for attaining and maintaining the NAAQS. The court in Wisconsin recognized that where such emissions have been identified, they should be eliminated as expeditiously as practicable, and in line with the attainment schedule for downwind areas, which, for the 2015 ozone NAAQS, is provided in CAA section 181. 938 F.3d at 313–20.

Further, EPA observes that more than one-third of the identified SCR retrofit potential (in terms of generating capacity) has no planned retirement date within 15 years, and therefore the cost of pollution control technology on such units would likely be lower, holding all other parameters equal, on a dollar per ton basis by virtue of the length of time the pollution control equipment may be in operation. Nor does EPA agree that units that would retire in less than 15 years should automatically be considered to face an unreasonably higher cost burden. Based on data analyzed in the EGU NO_X Mitigation Strategies Final Rule TSD, we find that the cost per ton associated with SCR retrofit technology does not begin to increase significantly above the $11,000/ton benchmark unless units have dramatically lower operating capacity or retire in less than 5 years' time—as illustrated in Figure 1 to section V.B.1.e of this document.

Finally, EPA's identification of this mitigation strategy is not meant to be limited only to units that experience a retrofit cost that is less than the representative cost threshold. First, that threshold represents an average, meaning that EPA's analysis already recognizes that some units on a facility-specific basis may face costs higher than that threshold. Further, EPA identifies this technology as widely available, implemented in practice already at many existing EGUs, and now standard for any coal-fired unit coming online in the past 25 years. More than 66 percent of the current large coal fleet already has such controls in place. Even if the cost were higher for some units for the reasons provided by commenters—and there were no less costly means provided to them to achieve the same level of emissions reduction (which the trading program allows for)—that would not necessarily obviate EPA's basis for finding that an emissions-reduction requirement commensurate with this standard pollution control practice for this unit type is warranted. The implementation of emissions reductions through a trading program, and its corresponding compliance flexibilities, make the use of a single representative cost all the more appropriate in this assessment. Therefore, upon reviewing all of the data including the information supplied by commenters, and even accounting for certain units' announced plans to retire earlier than an assumed 15-year book life for SCR retrofit technology, EPA finds its representative cost for this technology to be appropriate and reasonable for purposes of analysis under CAA section 110(a)(2)(D)(i)(I) and maintains this cost estimate in the final rule.

However, in recognition of the unique circumstances related to the transition of the power sector away from coal-fired and other high-NO_X emitting fuels and generating technologies, which is anticipated to accelerate in the late 2020s and into the 2030s, EPA has adjusted the final rule to avoid imposing a capital-intensive control technology retrofit obligation which could have overall net-negative environmental consequences ( e.g., by extending the life of a higher-emitting EGU or necessitating the allocation of material and personnel that could be used for more advanced clean-technology innovations). For units that plan to retire by 2030, the final rule—by extending the daily backstop rate to 2030—allows these units to continue to operate, so long as they comply with the mass-based emissions trading program requirements. Therefore, a unit experiencing a higher dollar per ton retrofit cost due to retirement plans has the flexibility to install less capital intensive controls such as SNCR, procure less costly allowances through either banking or purchase, or they may also reduce their allowance holding requirement through reduced utilization consistent with their phasing out towards a planned retirement date. This flexibility that EPA has included in the final rule is discussed in further detail in section VI.B of this document.

Comment: Some commenters suggested that the 0.05 lb/mmBtu emissions rate assumed for new SCRs at large coal units is not achievable at all coal units with retrofit potential and that EPA should raise this performance assumption to a value of 0.08 lb/mmBtu consistent with that assumption for existing SCRs.

Response: First, EPA believes the commenter misunderstands its intention with the 0.05 lb/mmBtu SCR rate assumption. This is meant to reflect a representative assumption for emissions rate performance for new SCR installed on the currently unretrofitted coal fleet—in this respect, it represents an average, not a maximum. EPA recognizes that some units will likely perform better ( i.e., lower) than this rate and some will potentially perform worse ( i.e., higher) than this rate—but that 0.05 lb/mmBtu is a reasonable representation of new SCR retrofit potential on a fleet-wide basis and for identifying expected state and regional emissions reduction potential from this technology. It would be inappropriate for EPA to use the worst performing tier of new SCR retrofit for this representative value. Moreover, EPA's review of historical environmental performance for recently installed SCRs does not support any indication that 0.05 is not representative of the retrofit potential for the fleet. EPA found that three quarters of the SCR retrofit projects completed in the last 15 years have achieved a rate of 0.05 lb/mmBtu or better on a monthly or seasonal basis. Moreover, its review of the engineering literature and consultation with third party pollution control engineering consultancies suggests that vendors are often willing to guarantee 0.05 lb/mmBtu seasonal performance for new SCR retrofit projects. Current SCR catalyst suppliers provide NO_X emissions warranties based at the catalyst's end-of-life period, often after 16,000 to 24,000 hours of operations, with newer catalyst achieving similar or better NO_X removal rates. Standard commercial terms, made by the purchaser to the SCR Retrofit supplier, can specify a system capable of meeting the proposed NO_X emissions rate and define the catalyst operational life before replacement. Thus, achieving the proposed reduction rates is accomplished through the buyer specifying the SCR retrofit requirements and the supplier providing an optimized system design and installing sufficient catalyst for the targeted end-of-life NO_X emissions rate. The agency is confident that SCR retrofit suppliers will be able to warrant their offerings for the emissions rates proposed in the regulation and to provide sufficient operating life for the affected sector.

Comment: Some commenters suggest that the evaluation of pollution control installation cost at Step 3 should be segmented depending on unit characteristics, and by failing to do so understate the cost of retrofitting SCR controls. In particular, these commenters note that units with lower capacity factors, different coal ranks, with pre-existing controls—such as SNCR—face substantially higher dollar per ton reduced costs than those that do not have such controls in place and should not be identified as a cost-effective mitigation strategy.

Response: Consistent with prior CSAPR rulemakings, at Step 3 EPA evaluates a mitigation technology and its representative cost and performance for the fleet on average. This representative cost is inclusive and robust to the portion of the fleet that may face higher dollar per ton cost. Both the “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, EGU NO_X Mitigation Strategies Proposed Rule TSD” (Feb. 2022), hereinafter referred to as the EGU NO_X Mitigation Strategies Proposed Rule TSD, and the EGU NO_X Mitigation Strategies Final TSD discuss the SCR retrofit cost specific to the segment of the fleet that has a SNCR in place and notes that those unit-level higher retrofit cost estimates are factored into its determination of the fleet-wide representative number. Although EPA believes its representative cost are appropriate and underpinned by operating assumptions reflective of the fleet averages, it nevertheless examined how cost would vary based on some of the variables highlighted by commenter. The EPA derived its capacity factor assumption based on expected future operations of this fleet segment that are inclusive of units operating at a range of capacity factors. It also examined how cost would change assuming different coal rank, assuming different book life, and different reagent cost. These analyses are discussed and shown in Appendix B of the EGU NO_X Mitigations Strategies Final Rule TSD and demonstrate that even under different operating assumptions, the variation in cost does not reach a point that would reverse EPA's finding regarding the appropriateness of this technology as part of this final rule's control stringency. Moreover, as discussed in section V.D of this document, EPA identifies appropriate mitigation strategies based on multiple factors—not solely on cost, and there is no indication that an individual unit's higher retrofit cost would obviate the appropriateness of retrofitting this standard and best practice technology at the unit. Finally, in prior rules and in the proposal, EPA recognized that some units will have higher cost and some will have lower cost relative the fleetwide representative value provided. Implementing the region and state reduction requirements through a mass-based trading program provides a means of alternative lower cost compliance for those sources particularly concerned about the higher retrofit cost at their unit.

Comment: Some commenters suggested that EPA's proposed representative cost for SCR pollution control is likely too high and overstates the true cost of such control. They also noted it aligns with agency precedent. These commenters claim that EPA's cost recovery factor is higher than necessary (thus inflating the cost) as it reflects a weighting of utility-owned to merchant-owned plants that is representative of the fleet, but not the unretrofitted fleet with this retrofit potential identified in this rule. They also noted that EPA's assumed interest rate informing the cost estimate was higher than the prime rate in June of 2022.

Response: EPA agrees that its approach for identifying representative cost thresholds is aligned with prior rules and agrees that its approach is reasonable. As the commenter points out, prime rates and cost recovery factors may indeed be lower in recent data than those assumed by EPA for future years. However, given the volatility among these metrics, EPA believes its choices are appropriate to build cost estimates that are robust to future uncertainty, and if these cost input factors do materialize to be the lower values highlighted by commenter, then it will result in a lower cost assumed in this final rule, but would not otherwise alter any of the stringency identification or regulatory findings put forward in this final rule. EPA performed a cost sensitivity analysis in Appendix B of the EGU NO_X Mitigation Strategies Final Rule TSD which shows how cost for this technology would vary based on different assumed levels for this variable. This analysis shows that under lower interest rates such as those put forward by commenter, that technology cost would drop by approximately 15 percent relative to the representative values put forward in this rule.

f. Generation Shifting

At proposal, EPA considered intrastate emissions reduction potential from generation shifting across the representative dollar per ton levels estimated for the emissions controls considered in previous sections. As the cost of emitting NO_X increases, it becomes increasingly cost-effective for units with lower NO_X rates to increase generation, while units with higher NO_X rates reduce generation. Because the cost of generation is unit-specific, this generation shifting occurs incrementally on a continuum. Consequently, there is more generation shifting at higher cost NO_X -control levels.

The EPA recognizes that imposing a NO_X -control requirement on affected EGUs, like any environmental regulation, internalizes the cost of their pollution, which could result in generation shifting away from those sources toward other generators offering electricity at a lower pollution cost. If, in the context of a market-based allowance trading program form of implementation, the EPA imposes a preset emissions budget that is premised only on assumed installation, optimization, and continued operation of unit-specific pollution control technologies, with no accounting for the likely generation shift in the marketplace away from these higher-polluting sources, that preset emissions budget will contain more tons than would be emitted if the affected EGUs achieved the emissions performance level (on a rate basis) selected at step 3. Hence, EPA has previously quantified and required expected emissions reductions from generation shifting in prior transport rules to avoid undermining the program's incentive to install, optimize, and operate controls identified in the Agency's determinations regarding the requisite level of emissions control at Step 3. See, e.g.,81 FR 74544–45; 76 FR 48280.

As in these prior rules, at proposal, the EPA did not identify generation shifting as a primary mitigation strategy and stringency measure on its own, but included emissions reductions from this strategy as it would be projected to occur in response to the selected emissions control stringency levels (and corresponding allowance price signals in step 4 implementation). For this rule's proposal, the EPA only specified emissions reductions from generation shifting in its preset budget calculations for 2023 and 2024. Because this rule's dynamic budget methodology applies the selected control stringency's emissions rates to the most recently reported heat input at each affected EGU, dynamic budgeting effectively serves a similar purpose to our ex ante quantification of emissions reduction potential from generation shifting for preset budgets in prior transport rules, i.e., to adequately and continuously incentivize the implementation of the emissions control strategies selected at Step 3. Therefore, dynamic budgets under this rule's program moot the need to specify discrete emissions reduction potential from generation shifting for those control periods, as they automatically reflect whatever generation balance affected EGUs would determine in the marketplace inclusive of their response to the emissions performance levels imposed by this rule.

Comment: Commenters offered both support for and opposition against the inclusion of generation shifting at Step 3 analysis for EGUs. Those in support noted that inclusion of emissions reductions from generation-shifting is integral to the successful implementation of the pollution control measures identified in the selected control stringency at Step 3. Those opposed generally argued the EPA was overestimating reduction potential from generation shifting in light of recent volatility and high prices in the markets for lower emitting fuels such as natural gas. Commenters also noted the electrical grid in certain regions has constraints that would make generation shifting more difficult than the EPA assumed. Commenters also asserted that the EPA did not have the legal authority to require generation shifting.

Response: The EPA disagrees with these comments regarding our legal authority but notes this issue is not relevant for purposes of this final action. The EPA continues to believe it has authority under CAA section 110(a)(2)(D)(i)(I) to consider and require emissions reductions from generation shifting if the EPA were to find that strategy was necessary to eliminate significant contribution. However, based on circumstances currently facing affected EGUs, as well as the inherent strength of the dynamic budget methodology to automatically reflect the market-determined balance of generation across sources responding to this rule, the EPA is not specifying emissions reduction potential from generation shifting as a part of the Step 3 analysis, nor to require any emissions reductions from generation shifting in preset budgets formulated under Step 4 for any control period, for this final rule.

Currently observable market conditions ( e.g., fuel prices) present unusual uncertainty with respect to key economic drivers of generation shifting. The availability of emissions reductions through generation shifting, and the magnitude of those emissions, is dependent on the availability and cost of substitute generation. The primary driver of near-term generation shifting-based emissions reductions has been shifting to lower-emitting natural gas generation. Recent volatility and high prices in the natural gas market have increased the uncertainty and reduced the potential of this emissions control strategy at any given cost threshold in the near term. For example, Henry Hub natural gas prices went from under $3.00/mmBtu during most of the last decade to an average of nearly $8.00/mmBtu for the most recent (2022) ozone season before declining sharply at the start of 2023. The current volatility in natural gas prices reduces the availability of emissions reductions from generation shifting and make its identification and quantification too uncertain for incorporation into Step 3 emissions reduction estimates for this rulemaking.

The Step 4 dynamic budget-setting process of this rule obviates the need to specify and require discrete emissions reductions from generation shifting under Step 3. As discussed in section VI of this document, the EPA in this final rule will implement a budget-setting approach that relies on two components: first, we have calculated “preset” budgets that reflect the best information currently available about fleet change over the period 2023 through 2029. Second, beginning in 2026, dynamic state emissions budgets will be calculated that will reflect the balance of generation across sources reported to EPA by EGU operators. Between 2026 and 2029, the actual budget that will be implemented will reflect the greater of either the preset budget or the dynamic budget calculation; from 2030 onwards, the budgets will be set only through the dynamic budget calculation. This overall approach is well suited for a period of significant power sector transition driven by a variety of economic, policy, and regulatory forces and allows for the balance of generation in this period to adjust in response to these forces while nonetheless ensuring that the budgets will continuously incentivize the emissions control stringency identified at Step 3. See section VI.B.4 of this document for further discussion on the interaction of preset and dynamic budgets during the 2026–2029 time period. With these approaches, and on the present record before the Agency, we conclude that the estimation and incorporation of specified emissions reductions from generation shifting at Step 3 is not necessary to eliminate significant contribution from EGUs for the 2015 ozone NAAQS through this rule's program implementation.

In previous CSAPR rulemakings, the EPA included generation shifting in the budget setting process to capture those reductions that would occur through shifting generation as an economic response to the control stringency determined based on the selected NO_X control strategies. See, e.g.,81 FR 74544–45. “Because we have identified discrete cost thresholds resulting from the full implementation of particular types of emissions controls, it is reasonable to simultaneously quantify the reduction potential from generation shifting strategy at each cost level. Including these reductions is important, ensuring that other cost-effective reductions ( e.g., fully operating controls) can be expected to occur.” EGU NO_X Mitigation Strategies Final Rule TSD (EPA–HQ–OAR–2015–0500–0554), at 11–12.

Commenters on this rule and prior transport rules have observed that using preset budgets to factor in generation shifting is flawed in that it results in EPA incorporating specific quantities of emissions reductions from discrete levels of generation shifting that are projected to occur but may in fact ultimately transpire differently in the marketplace. Commenters on this rule claim that other variables, such as constraints in transmission capacity or changes in fuel prices, can drive such differences in projected versus realized generation shifting, and these concerns are particularly exacerbated in a time of significant uncertainty around energy supplies and markets together with new laws passed by Congress ( e.g., the Infrastructure Investment and Jobs Act and the Inflation Reduction Act) driving the current transformation of the power sector. By refraining in this rule from specifying discrete emissions reductions from generation shifting in preset budgets and instead relying on a dynamic budgeting approach to reflect market-driven generation patterns, EPA ensures that its budgets remain sufficiently stringent over the long term to continually incentivize the emissions control stringency it determined to be cost-effective and therefore appropriate to eliminate significant contribution at Step 3. Thus, dynamic budgeting addresses the same concern that animated our use of generation shifting in the CSAPR rulemakings, but in doing so uses a market-following approach that will accommodate, over the long term, unforeseen drops or increases in heat input levels.

g. Other EGU Mitigation Measures

The EPA requested comment on whether other EGU ozone-season NO_X Mitigation technologies should be required to eliminate significant contribution. For instance, the EGU NO_X Mitigation Strategies Proposed and Final Rule TSDs discussed certain mitigation technologies that have been applied to “peaking” units (small, low-capacity factor gas combustion turbines often only operating during periods of peak demand).

Comment: Some commenters emphasized that simple cycle combustion turbines play a significant role in downwind contribution, and they highlight that states such as New York have imposed emissions limits on these sources acknowledging their impact on downwind nonattainment. These commenters suggest that EPA pursue and expedite the implementation of these or similar mitigation measures.

Response: As explained in greater detail in the EGU NO_X Mitigation Strategies Final TSD, both the configuration and operation of this segment of the EGU fleet reflects significant variability among units and across time. In other words, one unit may have a capacity factor in a given year that is one hundred times greater than a similar unit in that same year, or even than its own capacity factor from a preceding year. This type of variability and heterogeneity make it unlikely that there is a single cost-effective control strategy across this fleet segment, and commenters did not provide evidence to the contrary. EPA's analysis discussed in the EGU NO_X Mitigation Strategies Final Rule TSD highlights that there are 32 units emitting more than 10 tons per year on average for the 2019–2021 ozone seasons and lacking combustion controls or more advanced controls (totaling approximately 1,000 tons of ozone season NO_X emissions in 2021). EPA analysis estimates a representative cost of $22,000 per ton for dry low NO_X burners or ultra-low NO_X burners at these simple cycle combustion turbines, and over $100,000 per ton for SCR retrofit at some combustion turbines. Therefore, EPA does not identify any such uniform mitigation measure at Step 3 when estimating reduction potential.

Nonetheless, the EPA recognizes that these simple cycle combustion turbines may have cost-effective emissions-reduction opportunities. These units are included in the emissions trading program and therefore, as in prior transport rules, the program continues to subject them to an allowance holding requirement under this rule which will likely incentivize any available cost-effective NO_X reductions from these EGUs. For instance, emissions rates from these units in New York were considerably lower in 2022, when they faced a high allowance price, versus 2021, when the allowance price was much lower. Therefore, we find that the appropriate treatment of these units in this final rule is to continue to include them in the emissions trading program to incentivize cost-effective emissions reductions, but EPA does not find the magnitude or consistency of cost-effective mitigation potential to establish a specific increment of emissions reduction through a specific Step 3 emissions control determination. Moreover, while EPA's program will incentivize any available cost-effective reductions within this cadre of units (and such behavior is captured in its final program evaluation and modeling the RIA), it does not obviate the need for the other EGU cost-effective reductions elsewhere as suggested by some commenters.

2. Non-EGU or Stationary Industrial Source NO_X Mitigation Strategies

In the early stages of preparing the proposed FIP, the EPA evaluated air quality modeling information, annual emissions, and information about potential controls to determine which industries, beyond the power sector, could have the greatest impact on downwind receptors' air quality and therefore the greatest impact in providing ozone air quality improvements in affected downwind states through reducing those emissions. Specifically, the EPA conducted a screening assessment focused on individual emissions units with >100 tpy of actual NO_X emissions in 23 upwind states. Once the industries were identified, the EPA used its Control Strategy Tool to identify potential emissions units and control measures and to estimate emissions reductions and compliance costs associated with application of non-EGU emissions control measures. The technical memorandum “Screening Assessment of Potential Emissions Reductions, Air Quality Impacts, and Costs from Non-EGU Emissions Units for 2026” (“Non-EGU Screening Assessment” or “screening assessment”) lays out the analytical framework and data used to prepare proxy estimates for 2026 of potentially affected non-EGU facilities and emissions units, emissions reductions, and costs.

This screening assessment was not intended to identify the specific emissions units subject to the proposed emissions limits for non-EGU sources but was intended to inform the development of the proposed rule by identifying proxies for (1) non-EGU emissions units that potentially had the most impact in terms of the magnitude of emissions and potential for emissions reductions, (2) potential controls for and emissions reductions from these emissions units, and (3) control costs from the potential controls on these emissions units. This information helped shape the proposed rule.

To further evaluate the industries and emissions unit types identified by the screening assessment and to establish the applicability criteria and proposed emissions limits, the EPA reviewed RACT rules, NSPS rules, NESHAP rules, existing technical studies, rules in approved SIP submittals, consent decrees, and permit limits. That evaluation is detailed in the Proposed Non-EGU Sectors TSD prepared for the proposed FIP.

In this final rule, for purposes of this part of the Step 3 analysis, the EPA is retaining emissions control requirements for these industries and many of the emissions unit types included in the proposal. However, based on comments that credibly indicated in certain cases that emissions reduction opportunities are either not available for certain unit types or are at costs that are far greater than the EPA estimated at proposal, the EPA has changed the final rule to either remove or adjust the applicability criteria for such units. For a detailed discussion of the changes between the proposed FIP and this final rule, in emissions unit types included and in emissions limits, see section VI.C of this document. Tables I.B–2 through I.B–7 in section I.B of this document identify the emissions units and applicable emissions limitations, and Table II.A–1 in section II.A of this document identifies the industries included in the final rule.

For the final rule, to determine NO_X emissions reduction potential for the non-EGU industries and emissions unit types, with the exception of Solid Waste Combustors and Incinerators, we used a 2019 inventory prepared from the emissions inventory system (EIS) to estimate a list of emissions units captured by the applicability criteria for the final rule. For Solid Waste Combustors and Incinerators, the EPA estimated the list of covered units using the 2019 inventory, as well as the NEEDS-v6-summer-2021-reference-case workbook. Based on the review of RACT, NSPS, NESHAP rules, as well as SIPs, consent decrees, and permits, we also assumed certain control technologies could meet the final emissions limits. We did not run the Control Strategy Tool to estimate emissions reductions and costs and instead programmed the assessment using R. Using the list of emissions units estimated to be captured by the final rule applicability criteria, the assumed control technologies that would meet the emissions limits, and information on control efficiencies and default cost/ton values from the control measures database (CMDB), the EPA estimated NO_X emissions reductions and costs for the year 2026. We estimated emissions reductions using the actual emissions from the 2019 emissions inventory. In the assessment, we matched emissions units by Source Classification Code (SCC) from the inventory to the applicable control technologies in the CMDB. We modified SCC codes as necessary to match control technologies to inventory records.

The EPA recognized both at proposal and in the final rule that the cost per ton of emissions controls could vary by industry and by facility. The $7,500 marginal cost/ton threshold reflected in the Non-EGU Screening Assessment functioned as a relative, representative cost/ton level. Similar to the role of cost-effectiveness thresholds the EPA uses at Step 3 to evaluate EGU emissions control opportunities, this threshold is not intended to represent the maximum cost any facility may need to expend but is rather intended to be a representative figure for evaluating technologies to allow for a relative comparison between different levels of control stringency. The value was used to identify potentially cost-effective controls for further evaluation.

In the final rule, partly in recognition of the many comments indicating widely varying cost-per-ton values across industries and facilities, the EPA has updated its analysis of costs for the covered non-EGU industries. This data is summarized in the Technical Memorandum “Summary of Final Rule Applicability Criteria and Emissions Limits for Non-EGU Emissions Units, Assumed Control Technologies for Meeting the Final Emissions Limits, and Estimated Emissions Units, Emissions Reductions, and Costs,” available in the docket. We further respond to comments on the screening assessment in section 2.2 of the response to comments document.

3. Other Stationary Sources NO_X Mitigation Strategies

As part of its analysis for this final rule, the EPA also reviewed whether NO_X mitigation strategies for any other stationary sources may be appropriate. In this section, the EPA discusses three classes of units that have historically been excluded from our interstate air transport programs: (1) solid waste incineration units, (2) electric generating units less than or equal to 25 MW, and (3) cogeneration units. EPA's initial assessment did not lead it to propose inclusion of the units in these categories. However, EPA requested comment on whether any particular units within this category may offer cost-effective reduction potential.

Based on our request for comment, comments received, and our further evaluation, the EPA is including emissions limits and associated control requirements for the ozone season for solid waste incinerator units in this final rule, in line with the requirements we laid out for comment at proposal. Our analysis in this final rule confirms that these units have emissions reductions of a magnitude, degree of beneficial impact, and cost-effectiveness that is on par with the units in other industrial sectors included in this final rule.

For electric generating units less than 25 MW and cogeneration units previously exempted from EGU emissions budgets established through ozone interstate transport rules, the EPA has determined that these units should not be treated as EGUs in this final rule.

The EPA provides a summary of these three segments, their emissions control opportunities, and potential air quality benefits in the following sections. Additional considerations are further discussed in the EGU NO_X Mitigation Strategies Final TSD and in the RTC Document.

a. Municipal Solid Waste Units

At proposal, the EPA solicited comments on whether NO_X emissions reductions should be sought from municipal waste combustors (MWCs) to address interstate ozone transport, specifically on potential emissions limits, control technologies, and control costs. The EPA requested comment on emissions limits of 105 ppmvd on a 30-day rolling average and a 110 ppmvd on a 24-hour block average based on determinations made in the June 2021 Ozone Transport Commission (OTC) Municipal Waste Combustor Workgroup Report (OTC MWC Report). See87 FR 20085–20086. The OTC MWC Report found that MWCs in the Ozone Transport Region (OTR) are a significant source of NO_X emissions and that significant annual NO_X reductions could be achieved from MWCs in the OTR using several different technologies, or combination of technologies at a reasonable cost. The OTC MWC report is included in the docket for this action.

Comment: The EPA received multiple comments supporting the inclusion of emissions limits for MWCs in the final rule. Commenters noted that MWCs are significant sources of NO_X that contribute to ozone problems in the states covered by the proposal. Multiple commenters referenced the OTC MWC report to contend that NO_X emissions from MWCs could be significantly reduced at a reasonable cost. Some commenters reasoned that sources closer to downwind monitors, including MWCs, should be regulated as a more targeted approach and a means to prevent overcontrol of upwind sources. Commenters also noted that the OTC recently signed a memorandum of understanding (MOU) requesting that OTC member states develop cost effective solutions and select the strategy or combination of strategies, as necessary and appropriate, that provides both the maximum certainty and flexibility for that state and its MWCs. Additionally, multiple commenters noted that MWCs are often located in economically marginalized communities or communities of color. Lastly, one commenter stated that MWCs were arbitrarily excluded from the non-EGU screening assessment prepared for the proposal.

Response: As described in section VI.B.2 of the notice of proposed rulemaking, the EPA assessed emissions reduction potential from non-EGUs by preparing a screening assessment to identify those industries that could have the greatest air quality impact at downwind receptors. While the EPA did not prepare an updated non-EGU screening assessment in preparation for this final rule, the Agency did evaluate MWCs using the criteria developed in the screening assessment for proposal and determined that MWCs should be included in this rulemaking. A discussion of this analysis for MWCs is available in the Municipal Waste Combustor Supplement to February 28, 2022 Screening Assessment of Potential Emissions Reductions, Air Quality Impacts, and Costs from Non-EGU Emissions Units for 2026, which is available in the docket for this rule.

Considering EPA's conclusion that MWCs should be included in this final rule if EPA applied the same criteria developed in the screening assessment for proposal, the findings from the OTC MWC report and recent MOU, the fact that many state RACT NO_X rules apply to MWCs, and information received during public comment, the EPA finds that MWCs should be included in this final rule. Thus, the EPA is finalizing NO_X emissions limits and compliance assurance requirements for large MWCs as defined in the regulatory text at § 52.46 and as described in this section.

Comment: Some commenters did not support the inclusion of emissions limits for MWCs in the final rule. Some commenters suggested that the inclusion of NO_X limits in a FIP is not necessary to continue to reduce NO_X emissions from MWCs or to address interstate transport problems. Some commenters noted that many of the MWCs in the states covered by the proposal are already subject to RACT-based NO_X emissions limits that are below the current Federal NSPS NO_X emissions limits for MWCs under 40 CFR part 60, subparts Cb and Eb. One commenter noted that MWCs do not always account for a large percentage of statewide NO_X emissions. Others suggested that voluntary industry actions are also driving downward trends of NO_X emissions for some MWCs. Some commenters also asserted that regulation could interfere with state waste reduction policies and associated environmental considerations.

Response: Regarding the comments that some MWCs are already subject to RACT NO_X emissions limits, the EPA acknowledges that some states included in this rulemaking have promulgated RACT NO_X emissions limits that apply to certain MWCs, including some that are lower than current MWC NSPS NO_X emissions limits. The EPA does not consider a source to be exempt from this rulemaking just because the source may be subject to other regulatory requirements. As noted, the Agency did evaluate MWCs using the criteria developed in the screening assessment for proposal and has concluded that MWCs should be included in this rulemaking. In considering the emissions limits that are being finalized in this rulemaking, the EPA reviewed existing state RACT rules as described in section VI.C.6 of this document and the “Technical Support Document (TSD) for the Final Rule, Docket ID No. EPA–HQ–OAR–2021–0668, Non-EGU Sectors TSD” (Mar. 2023), hereinafter referred to as Final Non-EGU Sectors TSD. We note that sources already subject to RACT NO_X emissions limits that are equal to or more stringent than the limits finalized in this rulemaking will have the option to streamline regulatory requirements through the Title V permitting process.

Regarding the statement that regulation could interfere with state waste reduction policies and associated environmental considerations, the EPA acknowledges that MWCs serve an important role in municipal solid waste management programs, and that many function as cogeneration facilities that produce electrical power for the power grid. The EPA also analyzed control costs and determined that the required NO_X emissions limits for MWCs can be achieved at a reasonable cost, as described in section VI.C.6 of this document, the Final Non-EGU Sectors TSD, and the OTC MWC Report. Although the EPA does not expect these regulations to disrupt the ability of the industry to provide municipal solid waste and electric services, to the extent a facility is unable to comply with the standards due to technical impossibility or extreme economic hardship, the final rule includes provisions for facility operators to apply for a case-by-case alternative emissions limit. See section VI.C of this document and 40 CFR 52.40(d). In addition, for MWC facilities that are unable to comply with the standard by the 2026 ozone season, the final rule includes provisions for requesting limited extensions of time to comply. See section VI.C and 40 CFR 52.40(c).

b. Electric Generating Units Less Than or Equal to 25 MW

The EPA has historically not included control requirements for emissions for electric generating units less than or equal to 25 MW of generation for three primary reasons: low potential reductions, relatively high cost per ton of reduction, and high monitoring and other compliance burdens. In the January 11, 1993, Acid Rain permitting rule, the EPA provided for a conditional exemption from the emissions reduction, emitting, and emissions monitoring requirements of the Acid Rain Program for new units having a nameplate capacity of 25 MWe or less that burn fuels with a sulfur content no greater than 0.05 percent by weight, because of the de minimis nature of their potential SO_2, CO₂ and NO_X emissions. See63 FR 57484. The NO_X SIP Call identified these as Small Point Sources. For the purposes of that rulemaking, the EPA considered electricity generating boilers and turbines serving a generator 25 MWe or less, to be small point sources. The EPA noted that the collective emissions from small sources were relatively small and the administrative burden to the states and regulated entities of controlling such sources was likely to be considerable. As a result, the rule did not assume reductions from those sources in state emissions budgets requirements (63 FR 57402). Similar size thresholds have been incorporated in subsequent transport programs such as CAIR and CSAPR. As these sources were not identified as having cost-effective reductions and so were not included in those programs, they were also exempted from certain reporting requirements and the data for these sources is, therefore, not of the same caliber as that of covered larger sources.

EPA's preliminary survey of current data, compared to this initial justification, does not appear to offer a compelling reason to depart from this past practice by requiring emissions reductions from these small EGU sources as part of this rule. For instance, as explained in the EGU NO_X Mitigation Strategies Final Rule TSD, EPA has evaluated the costs of SCR retrofits at small EGUs using its Retrofit Cost Analyzer and found that such controls become markedly less cost-effective at lower levels of generating capacity. This analysis concluded that, after controlling for all other unit characteristics, the dollar per ton cost for a SCR retrofit increases by about a factor of 2.5 when moving from a 500 MW to a 10 MW unit, and a factor of 8 when moving to a 1 MW unit. Moreover, the EPA estimates that under 6 percent of nationwide EGU emissions come from units that are less than 25 MW and not covered by current applicability criteria due to this size exemption threshold. Therefore, the EPA is not finalizing any emissions reductions for these units.

Comment: EPA received comment supporting the continued application of the 25 MW threshold.

Response: Consistent with prior rules, the proposal, and stakeholder comment, EPA is continuing to apply its 25 MW applicability threshold for EGUs in this rulemaking. EPA did not find compelling comment to reverse its determination that (1) these sources offer low potential reductions, (2) have relatively high cost per ton, and (3) have high monitoring and other compliance burdens.

c. Cogeneration Units

Consistent with prior transport rules, fossil fuel-fired boilers and combustion turbines that produce both electricity and useful thermal energy (generally referred to as “cogeneration units”) and that meet the applicability criteria to be included in the CSAPR NO_X Ozone Season Group 3 Trading Program would be subject to the emissions reduction requirements established in this rulemaking for EGUs. However, those applicability criteria—which the EPA is not altering in this rulemaking ( see section VI.B.3 of this document)—exempt some cogeneration units from coverage as EGUs under the trading program. The EPA is finalizing that fossil fuel-fired boilers and combustion turbines that produce both electricity and useful thermal energy and that do not meet the applicability criteria to be included in the CSAPR NO_X Ozone Season Group 3 Trading Program as EGUs would not be subject to the Group 3 emissions trading program. However, to the extent a cogeneration unit meets the applicability criteria for industrial non-EGU boilers covered by this rule, that unit will be subject to the relevant requirements and is not exempted by virtue of being a cogeneration unit.

According to information contained in the EPA's Combined Heat and Power Partnership's document “Catalog of CHP Technologies”, there are 4,226 CHP installations in the U.S. providing 83,317 MWe of electrical capacity. Over 99 percent of the installations are powered by 5 equipment types, those being reciprocating engines (52 percent), boilers/steam turbines (17 percent), gas turbines (16 percent), microturbines (8 percent), and fuel cells (4 percent). The majority of the electrical capacity is provided by gas turbine CHP systems (64 percent) and boiler/steam turbine CHP systems (32 percent). The various CHP technologies described herewith are available in a large range of sizes, from as small as 1 kilowatt reciprocating engine systems to as large as 300 megawatt gas turbine powered systems.

NO_X emissions from rich burn reciprocating engine, gas turbine, and microturbine systems are low, ranging from 0.013 to 0.05 lb/mmBtu. NO_X emissions from lean burn reciprocating engine systems and gas-powered steam turbines systems range from 0.1 to 0.2 lb/mmBtu. The highest NO_X emitting CHP units are solid fuel-fired boiler/steam turbine systems which emit NO_X at rates ranging from 0.2 to 1.2 lb/mmBtu.

Under the final rule (consistent with prior CSAPR rulemakings), certain cogeneration units would be exempt from coverage under the CSAPR NO_X Ozone Season Group 3 Trading Program as EGUs. Specifically, the trading program regulations include an exemption for a unit that qualifies as a cogeneration unit throughout the later of 2005 or the first 12 months during which the unit first produces electricity and continues to qualify through each calendar year ending after the later of 2005 or that 12-month period and that meets the limitation on electricity sales to the grid. To meet the trading program's definition of “cogeneration unit” under the regulations, a unit ( i.e., a fossil-fuel-fired boiler or combustion turbine) must be a topping-cycle or bottoming-cycle type that operates as part of a “cogeneration system.” A cogeneration system is defined as an integrated group of equipment at a source (including a boiler, or combustion turbine, and a generator) designed to produce useful thermal energy for industrial, commercial, heating, or cooling purposes and electricity through the sequential use of energy. A topping-cycle unit is a unit where the sequential use of energy results in production of useful power first and then, through use of reject heat from such production, in production of useful thermal energy. A bottoming-cycle unit is a unit where the sequential use of energy results in production of useful thermal energy first, and then, through use of reject heat from such production, in production of useful power. To qualify as a cogeneration unit, a unit also must meet certain efficiency and operating standards in 2005 and each year thereafter. The electricity sales limitation under the exemption is applied in the same way whether a unit serves only one generator or serves more than one generator. In both cases, the total amount of electricity produced annually by a unit and sold to the grid cannot exceed the greater of one-third of the unit's potential electric output capacity or 219,000 MWh. This is consistent with the approach taken in the Acid Rain Program (40 CFR 72.7(b)(4)), where the cogeneration-unit exemption originated.

The EPA requested comment on requiring fossil fuel-fired boilers in the non-EGU industries identified in section VI.C of this document that serve electricity generators and that qualify for an exemption from inclusion in the CSAPR NO_X Ozone Season Group 3 Trading Program as EGUs to instead meet the same emissions standards, if any, that would apply under this rulemaking to fossil fuel-fired boilers at facilities in the same non-EGU industries that do not serve electricity generators.

Comment: Some stakeholders support the continued exclusion of qualifying cogenerators from the EGU program, but suggested they be regulated as non-EGUs if they don't fit the EGU applicability criteria.

Response: The EPA agrees that there is no basis within the four-step framework to exempt cogeneration units that fall under the applicability criteria of the final rule for non-EGU boilers simply because they are cogeneration units. While cogeneration units do have environmental benefits as noted at proposal, some cogeneration unit-types, particularly boilers, are estimated to have NO_X emissions that would otherwise meet this rule's criteria at Step 3 for constituting “significant contribution.” These units can meet the emissions limits that are otherwise finalized for these unit types, and the EPA does not find a basis to exclude them simply because they may have other environmentally-beneficial attributes.

These emissions limits are set forth in section VI.C.5 of this document. Therefore, the final requirements for non-EGUs do not exempt cogeneration units and any cogeneration emissions units meeting the applicability criteria for non-EGUs will be subject to the final emissions limits for the appropriate non-EGU emissions unit. Based on EPA's review of available data, across all of the non-EGU industries covered by this rule, there are four cogeneration boilers (two in Pulp and Papermill and two in Basic Chemical Manufacturing) that would meet the final rule's applicability criteria for non-EGU units and are included in the analysis of non-EGU emissions reduction potential in section V.C.2 of this document.

4. Mobile Source NO_X Mitigation Strategies

Under a variety of CAA programs, the EPA has established Federal emissions and fuel quality standards that reduce emissions from cars, trucks, buses, nonroad engines and equipment, locomotives, marine vessels, and aircraft ( i.e., “mobile sources”). Because states are generally preempted from regulating new vehicles and engines with certain exceptions ( see generally CAA section 209), mobile source emissions are primarily controlled through EPA's Federal programs. The EPA has been regulating mobile source emissions since it was established as a Federal agency in 1970, and all mobile source sectors are currently subject to NO_X emissions standards. The EPA factors these standards and associated emissions reductions into its baseline air quality assessment in good neighbor rulemaking, including in this final rule. These data are factored into EPA's analysis at Steps 1 and 2 of the 4-step framework. As a result of this long history, NO_X emissions from onroad and nonroad mobile sources have substantially decreased (73 percent and 57 percent since 2002, for onroad and nonroad, respectively) and are predicted to continue to decrease into the future as newer vehicles and engines that are subject to the most recent, stringent standards replace older vehicles and engines.

For example, in 2014, the EPA promulgated new, more stringent emissions and fuel standards for light-duty passenger cars and trucks. The fuel standards took effect in 2017, and the vehicle standards phase in between 2017 and 2025. Other EPA actions that are continuing to reduce NO_X emissions include the Heavy-Duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control Requirements (66 FR 5002; January 18, 2001); the Clean Air Nonroad Diesel Rule (69 FR 38957; June 29, 2004); the Locomotive and Marine Rule (73 FR 25098; May 6, 2008); the Marine Spark-Ignition and Small Spark-Ignition Engine Rule (73 FR 59034; October 8, 2008); the New Marine Compression-Ignition Engines at or Above 30 Liters per Cylinder Rule (75 FR 22895; April 30, 2010); and the Aircraft and Aircraft Engine Emissions Standards (77 FR 36342; June 18, 2012).

Most recently, EPA finalized more stringent emissions standards for NO_X and other pollution from heavy-duty trucks (Control of Air Pollution from New Motor Vehicles: Heavy-Duty Engine and Vehicle Standards, 88 FR 4296, January 24, 2023). These standards will take effect beginning with model year 2027. Heavy-duty vehicles are the largest contributor to mobile source emissions of NO_X and will be one of the largest mobile source contributors to ozone in 2025. Reducing heavy-duty vehicle emissions nationally will improve air quality where the trucks are operating as well as downwind. The EPA's existing regulatory program for mobile sources will continue to reduce NO_X emissions into the future.

Comment: The EPA received comments on ozone-precursor emissions from mobile sources, including cars, trucks, trains, ships, and planes. Commenters broadly encouraged the EPA to require emissions reductions from mobile sources in this rule. Commenters stated that the transportation sector plays a significant role in NO_X pollution and ozone formation and urged the EPA to finalize emissions reductions for the transportation sector that will enable attainment of the 2015 ozone NAAQS. Some commenters noted that high proportions of NO_X emissions in various upwind states are attributable to the transportation sector, and stated that EPA should have targeted emissions reductions from mobile sources first before requiring more stringent emissions controls from stationary sources in the same upwind states.

Response: The EPA agrees with commenters that a variety of sources, including mobile sources in the transportation sector, produce NO_X emissions that contribute to ozone air quality problems across the U.S. This rule, as with prior interstate transport actions, does not ignore those emissions, and it credits those on-the-books measures of states and the Federal Government within the four-step framework by including emissions and emissions reductions from these sources in the emissions inventory for air quality modeling, which informs Steps 1 and 2 of this analysis. Thus, this rule accurately represents emissions from mobile sources that are used to evaluate the contribution of states to ozone air quality problems in other states. See section IV.C of this document.

The EPA notes that its Step 3 analysis for this FIP does not assess additional emissions reductions opportunities from mobile sources. The EPA continues to believe that title II of the CAA provides the primary authority and process for reducing these emissions at the Federal level. EPA's various Federal mobile source programs, summarized above in this section, have delivered and are projected to continue to deliver substantial nationwide reductions in both VOCs and NO_X emissions; these reductions from final rules are factored into the Agency's assessment of air quality and contributions at Steps 1 and 2. Further, states are generally preempted from regulating new vehicles and engines with certain exceptions, and therefore a question exists regarding the EPA's authority to address such emissions through such means when regulating in place of the states under CAA section 110(c). See generally CAA section 209. See also86 FR 23099. In any case, the existence of mobile source emissions noted by commenters does not lead to the conclusion that the EPA must require mobile source reductions in this rule or that the EPA has not properly identified “source[s] or other type[s] of emissions activity” in upwind states that “significantly contribute” for purposes of the Good Neighbor Provision. The EPA is committed to continuing the effective implementation and enforcement of current mobile source standards and continuing its efforts on new standards. The EPA will continue to work with state and local air agencies to incorporate emissions reductions from the transportation sector into required ozone attainment planning elements.

C. Control Stringencies Represented by Cost Threshold ($ per ton) and Corresponding Emissions Reductions

1. EGU Emissions Reduction Potential by Cost Threshold

For EGUs, as discussed in section V.A of this document, the multi-factor test considers increasing levels of uniform control stringency in combination with considering total NO_X reduction potential and corresponding air quality improvements. The EPA evaluated EGU NO_X emissions controls that are widely available (described previously in section V.B.1 of this document), that were assessed in previous rules to address ozone transport, and that have been incorporated into state planning requirements to address ozone nonattainment.

The EPA evaluated the EGU sources within the State of California and found there were no covered coal steam sources greater than 100 MW that would have emissions reduction potential according to EPA's assumed EGU SCR retrofit mitigation technologies. The EGUs in the state are sufficiently well-controlled resulting in the lowest fossil-fuel emissions rate and highest share of renewable generation among the 23 states examined at Step 3. EPA's Step 3 analysis, including analysis of the emissions reduction factors from EGU sources in the state, therefore resulted in no additional emissions reductions required to eliminate significant contribution from any EGU sources in California.

The following tables summarize the emissions reduction potentials (in ozone season tons) from these emissions controls across the affected jurisdictions. Table V.C.1–1 focuses on near-term emissions controls while Table V.C.1–2 includes emissions controls with extended implementation timeframes.

2. Non-EGU or Industrial Source Emissions Reduction Potential

As described in the memorandum titled “Summary of Final Rule Applicability Criteria and Emissions Limits for Non-EGU Emissions Units, Assumed Control Technologies for Meeting the Final Emissions Limits, and Estimated Emissions Units, Emissions Reductions, and Costs,” the EPA uses the 2019 emissions inventory, the list of emissions units estimated to be captured by the applicability criteria, the assumed control technologies that would meet the emissions limits, and information on control efficiencies and default cost/ton values from the CMDB, to estimate NO_X emissions reductions and costs for the year 2026. The estimates using the 2019 inventory and information from the CMDB identify proxies for emissions units, as well as emissions reductions, and costs associated with the assumed control technologies that would meet the final emissions limits. Emissions units subject to the final rule emissions limits may differ from those estimated in this assessment, and the estimated emissions reductions from and costs to meet the final rule emissions limits may also differ from those estimated in this assessment. The costs do not include monitoring, recordkeeping, reporting, or testing costs.

Table V.C.2–1 summarizes the industries, estimated emissions unit types, assumed control technologies, estimated annual costs (2016$), and estimated ozone season emissions reductions in 2026, and Table V.C.2–2 summarizes the estimated reductions by state.

In Table V.C.2–3 by industry and emissions unit type, the EPA provides a summary of the control technologies applied and their average costs across all of the non-EGU emissions units. The average cost per ton values range from $939 to $14,595 per ton. Note that the average cost per ton values are in 2016 dollars and reflect simple averages and not a percentile or other representative cost values from a distribution of cost estimates.

Refer to the memorandum titled “Summary of Final Rule Applicability Criteria and Emissions Limits for Non-EGU Emissions Units, Assumed Control Technologies for Meeting the Final Emissions Limits, and Estimated Emissions Units, Emissions Reductions, and Costs” for additional estimates—including by industry and by state. These estimates are proxy estimates, and the EPA also did not prepare detailed engineering analyses for the industries, facilities, and individual emissions units identified for the final rule. Emissions units subject to the final rule emissions limits may differ from those estimated in this assessment, and the estimated emissions reductions from and costs to meet the final rule emissions limits may also differ from those estimated in this assessment.

Comment: Regarding the marginal cost threshold of $7,500/ton used to assess potential emissions reductions in the non-EGU screening assessment prepared for proposal, commenters raised a range of questions, including (1) why the EPA used a marginal cost threshold that is much higher than the $2,000/ton threshold used in the 2021 Revised CSAPR Update Rule, (2) why the EPA used a “one size fits all” approach for addressing the estimated cost and actual emissions reductions achievable, particularly for existing sources of NO_X emissions, (3) why the EPA set a $7,500/ton marginal cost threshold for all non-EGUs, despite acknowledging the heterogeneity of industry, emissions unit types and control options and failing to consider the actual costs associated with achieving the proposed reductions at different types of emissions units in order to artificially inflate the marginal cost threshold and to justify otherwise cost-prohibitive NO_X control technologies. Commenters also stated that controls for their industry are not cost-effective using the EPA's presumptive value of $7,500/ton and that the value may not be technically feasible to apply to existing sources that would have to retrofit controls.

Response: The EPA notes that the primary purpose of the Screening Assessment of Potential Emissions Reductions, Air Quality Impacts, and Costs from Non-EGU Emissions Units for 2026 (non-EGU screening assessment) was to identify potentially impactful industries and emissions unit types for further evaluation. In the non-EGU screening assessment memorandum we presented an analytical framework to further analyze potential emissions reductions and costs and included proxy estimates for 2026.

As noted in section V.D. of this document, at proposal the EPA found that based on data available at that time and for the purposes of the non-EGU screening assessment, it appeared that a $7,500 marginal cost-per-ton threshold could be used as a proxy to identify cost-effective emissions control opportunities. Also, the $7,500 marginal cost-per-ton threshold is higher than the cost-per-ton value used in the Revised Cross-State Air Pollution Rule Update because that rulemaking assessed significant contribution for the less protective 2008 ozone NAAQS, and it is reasonable when assessing significant contribution associated with the more protective 2015 ozone NAAQS, that a potentially more costly universe of emissions controls and related potential reductions should be included in the analysis. Similar to the role of cost-effectiveness thresholds the EPA uses at Step 3 to evaluate EGU emissions control opportunities, this threshold is not intended to represent the maximum cost any facility may need to expend but is rather intended to be a representative figure for evaluating technologies to allow for a relative comparison between different levels of control stringency. The EPA's potential cost threshold for non-EGU controls at proposal was intended to serve a similar representative purpose. Based on the EPA's updated analysis for this final rule, the EPA recognizes that the $7,500/ton threshold does not reflect the full range of cost-effectiveness values that are likely present across the many different types of non-EGU industries and emissions units assessed.

While the potentially impactful industries (identified in Step 1 of the analytical framework presented in the non-EGU screening assessment) were directly used, the proxy estimates for emissions unit types, emissions reductions, and costs from the non-EGU screening assessment were not directly used to establish applicability thresholds and emissions limits in the proposal. To further evaluate the impactful industries and emissions unit types and establish the proposed emissions limits, the EPA reviewed RACT rules, NSPS rules, NESHAP rules, existing technical studies ( e.g., Ozone Transport Commission, Technical Information Oil and Gas Sector Significant Stationary Sources of NO_X Emissions, October 17, 2012), rules in approved SIP submittals, consent decrees, and permit limits.

D. Assessing Cost, EGU and Non-EGU NO _X Reductions, and Air Quality

To determine the emissions that are significantly contributing to nonattainment or interfering with maintenance, the EPA applied the multi-factor test to EGUs and non-EGUs separately, considering for each the relationship of cost, available emissions reductions, and downwind air quality impacts. Specifically, for each sector, the EPA finalizes a determination regarding the appropriate level of uniform NO_X control stringency that would collectively eliminate significant contribution to downwind nonattainment and maintenance receptors. Based on the air quality results presented in this section, we find that the emissions control strategies that were identified and evaluated in sections V.B and V.C of this document and found to be both cost-effective and feasible, deliver meaningful air quality benefits through projected reductions in ozone levels across the linked downwind nonattainment and maintenance receptors in the relevant analytic years 2023 and 2026. Further, EPA finds the emissions control strategies in upwind states that would deliver these benefits to be widely available and in use at many other similar EGU and non-EGU facilities throughout the country, particularly in those areas that have historically or now continue to struggle to attain and maintain the 2015 ozone NAAQS. Applying these emissions control strategies on a uniform basis across all linked upwind states continues to constitute an efficient and equitable solution to the problem of allocating upwind-state responsibility for the elimination of significant contribution. This approach continues to effectively address the “thorny” causation problem of interstate pollution transport for regional-scale pollutants like ozone that transport over large distances and are affected by the vagaries of meteorology. EME Homer City, 572 U.S. at 514–16. It requires the most impactful sources in each state that has been found to contribute to ozone problems in other states to come up to minimum standards of environmental performance based on demonstrated NO_X pollution-control technology. Id. at 519. When the effects of these emissions reductions are assessed collectively across the hundreds of EGU and non-EGU industrial sources that are subject to this rule, the cumulative improvements in ozone levels at downwind receptors, while they may vary to some extent, are both measurable and meaningful and will assist downwind areas in attaining and maintaining the 2015 ozone NAAQS.

In addition to the findings of cost-effectiveness, feasibility and widespread availability that support EPA's identification of the appropriate level of emissions-control stringency at Step 3 discussed in sections V.B and V.C, the findings regarding air quality improvement in this section—as in prior transport rules—are a central component of our Step 3 analytic findings as to the definition of “significant contribution.” EPA's assessment of air quality improvement for all of the emissions control strategies included shows continued air quality improvement with each additional control strategy measure. Within the group of selected control strategies for EGUs and non-EGUs no clear “knee-in-the-curve” is evident; i.e., there is no point at which there is a noticeable decline in the rate of air quality improvement up through the control stringency level selected. However, if EPA were to go beyond the selected control stringency through inclusion of additional EGU or non-EGU NO_X mitigation technologies for the covered sources and unit-types that are, at least on the record of this action, not widely available, uncertain or untested, and/or far more costly, a “knee-in-the-curve” does materialize, where the incremental air quality benefit per dollar spent per ton on mitigation measures plateaus even as costs increase dramatically. In the Revised CSAPR Update, EPA explained that a knee in the curve “is not on its own a justification for not requiring reductions beyond that point,” 86 FR 23107, but does indicate that it is a useful indicator for informing potential stopping points. The observation that no “knee-in-the-curve” materializes at the stringency levels up through that selected by EPA supports EPA's identified control stringency.

Further, as the Supreme Court has explained, “while EPA has a statutory duty to avoid over-control, the Agency also has a statutory obligation to avoid `under-control,' i.e., to maximize achievement of attainment downwind.” 572 U.S. at 523. While the ultimate purpose of the good neighbor provision is to eliminate significant contribution and not necessarily to resolve downwind areas' nonattainment and maintenance problems, we have evaluated the expected attainment status at each identified receptor as we examine the air quality effects of the different emissions control strategies identified. As discussed further in this section, the EPA notes that multiple receptors shift into projected attainment status or shift from projected nonattainment to maintenance status up through the stringency level ultimately selected by EPA. (And all receptors show improvement in air quality even if their status does not change.) These analytic findings at Step 3 cement EPA's identification of the selected EGU and non-EGU mitigation measures as the appropriate control stringency to fulfill its statutory obligation to eliminate significant contribution for the 2015 ozone NAAQS for the covered states. The EPA also evaluated whether the final rule resulted in possible over-control scenarios by evaluating if an upwind state is linked solely to downwind air quality problems that could have been resolved at a lower cost threshold, or if an upwind state could have reduced its emissions below the 1 percent of NAAQS air quality contribution threshold at a lower cost threshold. The Agency finds no overcontrol from this rule. See section V.D.4 of this document.

1. EGU Assessment

For EGUs, the EPA examined the emissions reduction potential associated with each EGU emissions control technology (presented in section V.C.1 of this document) and its impact on the air quality at downwind receptors. Specifically, EPA identified and assessed the projected average air quality improvements relative to the base case and whether these improvements are sufficient to shift the status of receptors from projected nonattainment to maintenance or from maintenance to attainment. Combining these air quality factors, costs, and emissions reductions, the EPA identified a control stringency for EGUs that results in substantial air quality improvement from emissions controls that are available in the timeframe for which air quality problems at downwind receptors persist. For all affected jurisdictions, this control stringency reflects, at a minimum, the optimization of existing post-combustion controls and installation of state-of-the-art NO_X combustion controls, which are widely available at a representative cost of $1,800 per ton. EPA's evaluation also shows that the effective emissions rate performance across affected EGUs consistent with realization of these mitigation measures does not over-control upwind states' emissions relative to either the downwind air quality problems to which they are linked at Step 1 or the 1 percent contribution threshold that triggers further evaluation at Step 3 of the 4-step framework for the 2015 ozone NAAQS.

Similarly, the EPA also identified installation of new SCR post-combustion controls at coal steam sources greater than or equal to 100 MW and for a more limited portion of the oil/gas steam fleet that had higher levels of emissions as components of the required control stringency. These SCR retrofits are widely available starting in the 2026 ozone season at $11,000 and $7,700 per ton respectively. For all but 3 of the affected states (Alabama, Minnesota, and Wisconsin, which are no longer linked in 2026 at Steps 1 and 2 in EPA's base case air quality modeling for this final rule), EPA's evaluation shows that the effective emissions rate performance across EGUs consistent with the full realization of these mitigation measures does not over-control upwind states' emissions in 2026 relative to either the downwind air quality problems to which they are linked at Step 1 or the 1 percent contribution threshold that triggers further evaluation at Step 3 of the 4-step framework for the 2015 ozone NAAQS ( see the Ozone Transport Policy Analysis Final Rule TSD for details).

To assess downwind air quality impacts for the nonattainment and maintenance receptors identified in section IV.D of this document, the EPA evaluated the air quality change at that receptor expected from the progressively more stringent upwind EGU control stringencies that were available for that time period in upwind states linked to that receptor. This assessment provides the downwind ozone improvements for consideration and provides air quality data that is used to evaluate potential over-control situations.

To assess the air quality impacts of the various control stringencies at downwind receptors for the purposes of Step 3, the EPA evaluated changes resulting from the emissions reductions associated with the identified emissions controls in each of the upwind states, as well as assumed corresponding reductions of similar stringency in the downwind state containing the receptor to which they are linked. By applying these emissions reductions to the state containing the receptor, the EPA assumes that the downwind state will implement (if it has not already) an emissions control stringency for its sources that is comparable to the upwind control stringency identified here. Consequently, the EPA is accounting for the downwind state's “fair share” of the responsibility for resolving a nonattainment or maintenance problem as a part of the over-control evaluation.

For this assessment, the EPA used an ozone air quality assessment tool (ozone AQAT) to estimate downwind changes in ozone concentrations related to upwind changes in emissions levels. The EPA focused its assessment on the years 2023 and 2026 as they pertain to the last years for which ozone season emissions data can be used for purposes of determining attainment for the Moderate (2024) and Serious (2027) attainment dates. For each EGU emissions control technology, the EPA first evaluated the magnitude of the change in ozone concentrations at the nonattainment and maintenance receptors for each relevant year ( i.e., 2023 and 2026). Next, the EPA evaluated whether the estimated change in concentration would resolve the receptor's nonattainment or maintenance concern by lowering the average or maximum design values, respectively, below 71 ppb. For a complete set of estimates, see the Ozone Transport Policy Analysis Final Rule TSD or the ozone AQAT Excel file.

For 2023, the EPA evaluated potential air quality improvements at the downwind receptors outside of California associated with available EGU emissions control technologies in that timeframe. The EPA determined for the purposes of Step 3 that the average air quality improvement at the receptors relative to the engineering analytics base case was 0.06 ppb for emissions reductions commensurate with optimization of existing SCRs/SNCRs and combustion control upgrades. The EPA determined for the purposes of Step 3 that no receptors switch from maintenance to attainment or from nonattainment to maintenance with these mitigation strategies in place. Table V.D.1–1 summarizes the results of EPA's Step 3 evaluation of air quality improvements at these receptors using AQAT.

For 2026, the EPA determined that the average air quality improvement at these receptors relative to the engineering analytics base case was 0.47 ppb for emissions reductions commensurate with optimization of existing SCRs/SNCRs, combustion control upgrades, and new post-combustion control (SCR and SNCR) retrofits at eligible units are assumed to be implemented. The EPA determined for the purposes of Step 3 that in 2026, all but one of the receptors are expected to remain nonattainment or maintenance across these control stringencies, with one receptor in Larimer County, Colorado (Monitor 080690011), switching from maintenance to attainment and two receptors (one in Fairfield County, Connecticut (Monitor 90013007), and one in Galveston, Texas (Monitor ID 481671034)) switching from nonattainment to maintenance with these mitigation strategies in place. Table V.D.1–2 summarizes the results of EPA's Step 3 evaluation of air quality improvements at the receptors included in the AQAT analysis. For more information about how this assessment was performed and the results of the analysis for each receptor, refer to the Ozone Transport Policy Analysis Final Rule TSD and to the Ozone AQAT included in the docket for this rule.

Figures 1 and 2 to section V.D.1 of this document, included in Appendix I of the Ozone Transport Policy Analysis Final Rule TSD available in the docket for this rulemaking, illustrate the air quality improvement relative to the estimated representative cost associated with the previously identified emissions control technologies. The graphs show improving air quality at the downwind receptors as emissions reductions commensurate with the identified control technologies are assumed to be implemented. Figure 1 to section V.D.1 of this document reflects emissions reductions commensurate with optimization of existing SNCRs and SCRs. Figure 2 to section V.D.1 of this document reflects emissions reductions commensurate with installation of new post combustion controls (mainly SCRs) layered on top of the emissions reduction potential from the technologies represented in Figure 1 to section V.D.1 of this document. The graphic, and underlying AQAT receptor-by-receptor analysis demonstrates that air quality continues to improve at downwind receptors as EPA examines increasingly stringent EGU NO_X control technologies. While all major technology breakpoints identified in sections V.B and V.C of this document show continued air quality improvements at problematic receptors and at cost and technology levels that are commensurate with mitigation strategies that are proven to be widely available and implemented, EPA's quantification and application of those breakpoints reflect certain exclusions to: (1) preserve this consistency with widely observed mitigation measures in states, and (2) remove any retrofit assumptions at marginal units that would have much higher dollar per ton representative cost and little or no air quality benefit. For instance, the EPA does not define the SCR retrofit breakpoint ($11,000 per ton) to include retrofit application at steam units less than 100 MW or at oil/gas steam units emitting at less than 150 tons per ozone season. The emissions reductions from these potential categories of measures are small and do not constitute additional “breakpoints” in EPA's estimation. They would entail much higher dollar per ton costs, going beyond what is widely observed in the fleet. This careful calibration of technology breakpoints through exclusion of measures that are clearly not cost-effective in terms of air quality benefit allows for the identification of an EGU uniform control stringency that is an appropriate reflection of those readily available and widely implemented emissions reduction strategies that will have meaningful downwind air quality impact.

Moreover, these technologies (and representative cost) are demonstrated ozone pollution mitigation strategies that are widely practiced across the EGU fleet and are of comparable stringency to emissions reduction measures that many downwind states have already instituted. The coal SCR retrofit measures driving the majority of the emissions reductions in this action not only reflect industry best practice, but they also reflect prevailing practice among EGUs. More than 66 percent of the existing coal capacity already has this technology in place. For nearly 25 years, all new coal-fired EGUs that commenced construction have had SCR (or equivalent emissions rates). The 1997 proposed amendments to subpart Da revised the NO_X standard based on the use of SCR. The NO_X SIP Call (promulgated in 1998) established emissions reduction requirements premised on extensive SCR installation (142 units) and incentivized well over 40 GWs of SCR retrofit in the ensuing years. Similarly, the Clean Air Interstate Rule established emissions reductions requirements in 2006 that assumed SCR would be installed on another 58 units (15 GW) in the ensuing years among just 10 states, and an even greater volume of capacity chose SCR retrofit measures in the wake of finalizing that action.

Basing emissions reduction requirements for EGUs on SCR retrofits is also consistent with regulatory approaches adopted by states, which—particularly in downwind areas more impacted by ozone transport contribution from upwind state emissions—have already adopted SCR-based standards as part of stringent NO_X control programs. Regulatory programs that impose stringent RACT requirements on all major power plants and Lowest Achievable Emission Rate (LAER) standards on all new major sources of NO_X have resulted in remaining coal-fired generating resources in states along the Northeast Corridor such as Connecticut, Delaware, New Jersey, New York, and Massachusetts all being retrofitted with SCR. The Maryland Code of Regulations requires coal-fired sources to operate existing SCR controls or install SCR controls by specified dates. Programs like North Carolina's Clean Smokestacks Act and Colorado's Clean Air, Clean Jobs Act have also required or prompted SCR retrofits on units. Unit-level BART requirements for the first Regional Haze planning period also determined SCR retrofits (and corresponding emissions rates) were cost-effective controls for a variety of sources in the U.S.

As shown in Figure 1 to section V.D.1 of this document, the majority of EGU emissions reduction potential and associated air quality improvements estimated for 2023 occurs from optimization of existing SCRs, with some additional reductions from installation of state-of-the-art combustion controls at the same representative cost threshold. At the slightly higher representative cost threshold of $1,800 per ton, there is some additional air quality improvement from optimization of existing SNCRs. These measures taken together represent the control stringency at which near-term incremental EGU NO_X reduction potential and corresponding downwind ozone air quality improvements are maximized. This evaluation shows that EGU NO_X reductions for each of the near-term emissions control technologies are available at reasonable cost and that these reductions provide meaningful improvements in downwind ozone concentrations at the identified nonattainment and maintenance receptors. Figure 1 to section V.D.1 of this document highlights (1) the continuous connection between identified emissions reduction potential and downwind air quality improvement across the range of near-term mitigation measures assessed, and (2) the cost-effective availability of these reductions and corresponding air quality improvements.

Additional considerations that are unique to EGUs provide additional support for EPA's determination to include SCR and SNCR optimization as part of the identified near-term control stringency, including:

• these controls are already installed and available for operation on these units;
• they are on average already partially operating, but not necessarily optimized;
• the reductions are available in the near-term (during ozone seasons when the problematic receptors are projected to persist), including by the 2023 ozone season aligned with the Moderate area attainment date; and

• these sources are already covered under the existing CSAPR NO_X Ozone Season Group 2 or Group 3 Trading Programs or the Acid Rain Program and thus have the monitoring, reporting, recordkeeping, and all other necessary elements of compliance with the trading program already in place.

The majority of EGU emissions reduction potential and associated air quality improvements estimated to start in 2026 occur from retrofitting uncontrolled steam sources with post-combustion controls. At the representative cost threshold of $11,000 per ton, there are significant additional air quality improvements from emissions reductions commensurate with installation of new SCRs and SNCRs. These measures taken together with the near-term emissions reduction measures described previously represent the level of control stringency in 2026 at which incremental EGU NO_X reduction potential and corresponding downwind ozone air quality improvements are maximized. This evaluation shows that EGU NO_X reductions for each of the emissions control technologies are available at reasonable cost and that these reductions can provide improvements in downwind ozone concentrations at the identified nonattainment and maintenance receptors.

The EPA finds that the control stringency that reflects optimization of existing SCRs and SNCRs, installation of state-of-the-art combustion controls, and the retrofitting of new post combustion controls at the coal and oil/gas steam capacity described previously is projected to result in nearly 73,000 tons of NO_X reduction (approximately 40 percent of the 2026 baseline level) for the 19 linked states in 2026 subject to a FIP for EGUs, which will deliver notable air quality improvements across all transport-impacted receptors and assist in fully resolving one downwind air quality receptor for the 2015 ozone NAAQS. Figure 2 to section V.D.1 of this document demonstrates the continuous connection between identified emissions reduction potential and downwind air quality improvement across the range of mitigation measures assessed in 2026. At no point do the additional emissions mitigation measures examined here fail to produce corresponding downwind air quality improvements.

The EPA is determining that the appropriate EGU control stringency is commensurate with the full operation of all existing post-combustion controls (both SCRs and SNCRs) and state-of-the-art combustion control upgrades for those states linked to downwind nonattainment or maintenance receptors in 2023. For those states also linked in 2026, the EPA is determining that the appropriate EGU control stringency also includes emissions reductions commensurate with the retrofit of SCR at coal steam units of 100 MW or greater capacity (excepting circulating fluidized bed units), new SNCR on coal steam units of less than 100 MW capacity and circulating fluidized bed units, and SCR on oil/gas steam units greater than 100 MW that have historically emitted at least 150 tons of NO_X per ozone season.

As noted previously in section V.B of this document and in the EGU NO_X Mitigation Strategies Final Rule TSD, the EPA considered other methods of identifying mitigation measures ( e.g., SCRs on smaller units, combustion control upgrades on combustion turbines, SCRs on combined cycle and simple cycle combustion turbines). The emissions reductions from these potential categories of measures do not constitute additional “technology breakpoints” in EPA's estimation, but rather reflect a different tier of assessment where further mitigation measures are based on inclusion of smaller and/or different generator-type units (rather than different pollution control technologies). Emissions reductions from these measures are relatively small and would entail much higher dollar per ton costs, going beyond what is widely observed in the fleet. Although these additional measures are not included in EPA's technology breakpoint analysis discussed in this section, the EPA did analyze the cost, potential reductions, and air quality impact of these additional measures to affirm that they do not merit inclusion in the final stringency for this action. That analysis shows the potential emissions reductions and air quality improvements from these additional measures occur beyond a notable “knee-in-the-curve” breakpoint. In other words, there are very little additional emissions reductions and air quality improvement at problematic receptors, and the cost associated with these measures increases substantially on a dollar per ton basis. The graphic capturing this effect (located in Appendix I of the Ozone Transport Policy Analysis Final Rule TSD) illustrates the significant decline in cost-effectiveness of reductions if these measures had been included in EPA's final stringency.

2. Non-EGU Assessment

Using a 2019 emissions inventory, the list of emissions units estimated to be captured by the applicability criteria, the assumed control technologies that would meet the emissions limits, and information on control efficiencies and default cost/ton values from the control measures database, the EPA estimated NO_X emissions reductions and costs for the year 2026. Given the EPA's conclusion that the 2026 ozone season is the earliest date by which the required controls can be installed across the identified non-EGU industries, the EPA assessed the effects of these controls in 2026 under its multi-factor test. In the assessment, we matched emissions units by Source Classification Code (SCC) from the inventory to the applicable control technologies in the CMDB. We modified SCC codes as necessary to match control technologies to inventory records. For additional details about the steps taken to estimate emissions units, emissions reductions, and costs, see the memorandum titled “Summary of Final Rule Applicability Criteria and Emissions Limits for Non-EGU Emissions Units, Assumed Control Technologies for Meeting the Final Emissions Limits, and Estimated Emissions Units, Emissions Reductions, and Costs” available in the docket. The estimates using the 2019 inventory and information from the CMDB identify proxies for emissions units, as well as emissions reductions, and costs associated with the assumed control technologies that would meet the final emissions limits. Emissions units subject to the final rule emissions limits may differ from those estimated in this assessment, and the estimated emissions reductions from, and costs to meet, the final rule emissions limits may also differ from those estimated in this assessment. The costs do not include monitoring, recordkeeping, reporting, or testing costs.

After reviewing public comments and updating some of the data used to provide an accurate assessment of the likely potential emissions reductions that could be achieved from the identified emissions units in the industries analyzed for proposal, the EPA finds that in general, these emissions reductions (with some modifications from proposal) are necessary to eliminate significant contribution at Step 3. The EPA's use of the analytical framework presented in the non-EGU screening assessment to identify potentially impactful industries and emissions unit types in the proposal remains valid. The EPA's criteria were intended to identify industries and emissions unit types that on a broad scale impact multiple receptors to varying degrees. The EPA focused its non-EGU screening assessment on (1) emissions and potential emissions reductions from these industries and emissions units and (2) the potential impact that emissions reductions from those industries and emissions units could deliver to the receptors.

While commenters criticized the analytical framework in the non-EGU screening assessment for assuming potentially unachievable emissions reductions at Step 3, or for not corresponding to a precise list of emissions units that would be covered at Step 4, these comments did not offer an alternative methodology for the Step 3 analysis to identify those industries and emissions units that potentially have the greatest impact and therefore should be scrutinized more closely for emissions reduction opportunities. Further, contrary to some commenters' assertions, the EPA's assessment did not result in an unbounded scope of regulation of industrial sources. Of the approximately 40 industries defined by North American Industry Classification System codes the EPA analyzed, only seven industries were identified as having emissions and potential emissions reduction opportunities that met the EPA's air quality criteria for further assessment.

At proposal, the EPA found that based on data available at that time and for the purposes of the screening assessment, it appeared that a $7,500 marginal cost-per-ton threshold could be used as a proxy to identify cost-effective emissions control opportunities. Similar to the role of cost-effectiveness thresholds the EPA uses at Step 3 to evaluate EGU emissions control opportunities, this threshold is not intended to represent the maximum cost any facility may need to expend but is rather intended to be a representative figure for evaluating technologies to allow for a relative comparison between different levels of control stringency. For example, in the EGU analysis, the $11,000/ton average cost threshold for an SCR retrofit represents a range of SCR retrofit costs for units for which the 90th percentile cost-per-ton is roughly $21,000. See section V.B.a of this document. The EPA's potential cost threshold for non-EGU controls at proposal was intended to serve a similar representative purpose. We respond briefly to comments regarding the use of the $7,500/ton threshold in section V.C of this document. Comments regarding the screening assessment are further addressed in section 2.2 of the response to comments document in the docket.

Based on the EPA's updated analysis for this final rule, the EPA recognizes that the $7,500/ton threshold does not reflect the full range of cost-effectiveness values that are likely present across the many different types of non-EGU industries and emissions units assessed. However, the EPA nonetheless finds that, with some adjustments from proposal, the overall mix of emissions controls it identified at proposal is appropriate to eliminate significant contribution to nonattainment or interference with maintenance in downwind areas. In the final analysis, we find that the average cost-per-ton of emissions reductions across all non-EGU industries in this rule generally ranges from approximately $939/ton to $14,595/ton, with an overall average of approximately $5,339/ton. See memorandum titled “Summary of Final Rule Applicability Criteria and Emissions Limits for Non-EGU Emissions Units, Assumed Control Technologies for Meeting the Final Emissions Limits, and Estimated Emissions Units, Emissions Reductions, and Costs,” available in the docket.

Nonetheless, overall the EPA finds that the range of cost-effectiveness values for non-EGU industries and emissions units compares favorably with the values used to evaluate EGUs. As discussed in the preceding paragraphs, the representative cost for EGUs to retrofit SCR is $11,000/ton. This reflects a range of cost estimates, with $20,900/ton reflecting the 90th percentile of units (see section V.B.a of this document). The higher end of the estimated average cost range for certain non-EGU industrial emissions units is also in that range. While specific emissions units may have higher costs associated with installing pollution control technologies than other similar unit types, this does not in itself undermine the Agency's conclusion that a level of emissions control associated with a specific emissions limit or control technology is appropriate to require across the linked upwind state region, in light of the overall emissions reductions and air quality benefits at downwind receptors that those controls are projected to deliver.

We note that the non-EGU control cost estimates in this final rule were based on historical actual emissions. This can affect the presentation of cost-per-ton values at the unit level, and it would not be appropriate to abandon uniform control stringency among like units in the covered industries across or within upwind states based on such cost differentials.

The EPA finds it appropriate to require a uniform level of emissions control across similar emissions unit types to, among other things, prevent two potential outcomes related to shifting production, either between units within the same facility or between units at different facilities. First, if some units were exempted from control requirements because of historically low actual emissions, there is a risk that source owners or operators may shift production to these specific units, increasing their utilization and resulting in emissions increases from these units. Second, if some owners or operators were able to avoid the control requirements of the final rule on this basis, they could gain a competitive advantage vis-à-vis other facilities within their respective industries. Production could shift from units at another facility subject to the control requirements to the units that avoided control requirements (and thus avoid costs the regulated facility should bear), potentially resulting in emissions increases. The effect of such an approach in such circumstances would be mere emissions shifting rather than the elimination of significant contribution. Finally, as we have explained in prior transport actions, the cost-effectiveness figure is not the only factor that the agency considers at Step 3, see86 FR 23073, and if used in isolation to make a policy decision without considering other information, could produce a result that is inconsistent with the objective of ensuring significant contribution is eliminated.

In addition to our evaluation of cost-effectiveness on a cost per ton basis, the EPA's determination at Step 3 for non-EGUs is also informed by the overall level of emissions reductions that will be achieved across the region and the effect those reductions are projected to have on air quality at the downwind receptors (discussed more later in this section). We are also influenced by the fact that these emissions control strategies for non-EGUs are generally well demonstrated to be feasible on many existing units, as established through our review of consent decrees, permits, RACT determinations, and other data sources. These levels of emissions control have in many cases already been required by states with downwind nonattainment areas for the 2015 ozone NAAQS.

The EPA determined that, for 2026, the incremental average air quality improvement at receptors relative to the EGU case when SCR post-combustion controls were installed was 0.19 ppb when non-EGU controls were applied, based on the Step 3 analysis. The total average air quality improvement was 0.66 ppb when the non-EGU improvement was added to the EGU improvement, meaning that the non-EGU increment accounts for about 29 percent of this average air quality improvement. In general, the air quality results from non-EGU emissions reductions yield additional important downwind benefits to the air quality benefits of the EGU strategy. For example, the total ppb improvement summed over all of the receptors from EGUs was 7.04 ppb and the non-EGU increment adds another 2.82 ppb of improvement bringing the total to 9.87 (when accounting for rounding). Non-EGUs account for 29 percent of this total air quality improvement as well. Further, these figures should not be considered in isolation; EPA is not comparing EGU strategy effects and non-EGU effects to make a selection between two different approaches. Rather, both the selected EGU and non-EGU emissions reduction strategies at the cost-effectiveness values identified in section V.B and V.C of this document present a comprehensive solution to eliminating significant contribution for the covered states. The combined effect of the EGU and non-EGU strategies is further presented in the following section.

For more information about how this assessment was performed and the results of the analysis for each receptor, refer to the Ozone Transport Policy Analysis Final Rule TSD and to the Ozone AQAT included in the docket for this rule.

3. Combined EGU and Non-EGU Assessment

The EPA used the Ozone AQAT to evaluate the combined impact of these selected stringency levels for both EGUs and non-EGUs on all receptors remaining in the 2026 air quality modeling base case to inform the air quality effects of the rule and to conduct our over-control analysis. EPA's evaluation demonstrated air quality improvement at the remaining nonattainment or maintenance receptors outside of California ( see section IV.D of this document for receptor details). The EPA estimated that the average air quality improvement at these receptors relative to the engineering analytics base case was 0.66 ppb for emissions reductions commensurate with optimization of existing SCRs/SNCRs, combustion control upgrades, application of new post-combustion control (SCR and SNCR) retrofits at eligible units, and all estimated emissions reductions from the non-EGU industries. Table V.D.3–1 summarizes the results of EPA's Step 3 evaluation of air quality improvements at these receptors using AQAT. In summary, the collective application of these mitigation measures and emissions reductions are projected to deliver meaningful downwind air quality improvements.

4. Over-Control Analysis

The EPA applied its over-control test to this same set of aggregated EGU and non-EGU data described in the previous section. The EPA performed air quality analysis using the Ozone AQAT to determine whether the emissions reductions for both EGUs and non-EGUs potentially create an “over-control” scenario. As in prior transport rules following the holdings in EME Homer City, overcontrol would be established if the record indicated that, for any given state, there is an identified, less stringent emissions control approach for that state, by which (1) the expected ozone improvements would be sufficient to resolve all of the downwind receptor(s) to which that state is linked; or (2) the expected ozone improvements would reduce the upwind state's ozone contributions below the screening threshold ( i.e., 1 percent of the NAAQS or 0.70 ppb) to all receptors. In EME Homer City, the Supreme Court held that the EPA cannot “require[] an upwind State to reduce emissions by more than the amount necessary to achieve attainment in every downwind State to which it is linked.” 572 U.S. at 521. On remand from the Supreme Court, the D.C. Circuit held that this means that the EPA might overstep its authority “when those downwind locations would achieve attainment even if less stringent emissions limits were imposed on the upwind States linked to those locations.” EME Homer City II, 795 F.3d at 127. The D.C. Circuit qualified this statement by noting that this “does not mean that every such upwind state would then be entitled to less stringent emissions limits. Some of those upwind States may still be subject to the more stringent emissions limits so as not to cause other downwind locations to which those States are linked to fall into nonattainment.” Id. at 14–15. Further, as the Supreme Court explained, “while EPA has a statutory duty to avoid over-control, the Agency also has a statutory obligation to avoid `under-control,' i.e., to maximize achievement of attainment downwind.” 572 U.S. at 523. The Court noted that “a degree of imprecision is inevitable in tackling the problem of interstate air pollution” and that incidental over-control may be unavoidable. Id. “Required to balance the possibilities of under-control and over-control, EPA must have leeway in fulfilling its statutory mandate.” Id.

Consistent with these instructions from the Supreme Court and the D.C. Circuit, using the Ozone AQAT, the EPA first evaluated whether reductions resulting from the selected control stringencies for EGUs in 2023 and 2026 combined with the emissions reductions selected for non-EGUs in 2026 can be anticipated to resolve any downwind nonattainment or maintenance problems (see the Ozone Transport Policy Analysis Final Rule TSD for details on the construction and application of AQAT).

Similar to our approach in the CSAPR Update and the Revised CSAPR Update, our primary overcontrol assessment examines the receptor changes from the emissions reductions of the upwind states found linked to a receptor. Consistent with prior Rules, EPA also assumed that downwind states that are not upwind states in this rule implement reductions commensurate with the rule's requirements (this treatment applies specifically to Colorado and Connecticut). This configuration effectively presents an equitable representation of the effects of the rule in that linked upwind states do not shift their responsibility to other upwind states linked to different receptors. It also effectively resolves any interdependence and “which state goes first?” questions. Furthermore, the downwind states in which a receptor is located are held to a “fair share” of emissions reductions— i.e., the same level of emissions control stringency that the upwind states must implement.

The EPA also repeated this analysis using an alternative configuration, as described in the Ozone Transport Policy Analysis Final Rule TSD. In this configuration, we looked at the combined effect of the entire program across all linked upwind states on each receptor and did not assume that a downwind state that is not also an upwind state makes any additional emissions reductions beyond the baseline in the relevant year. This configuration effectively isolates how the rule as a whole, and just the rule, will affect air quality and linkages. While the first configuration described is, in the Agency's view, the more appropriate way to evaluate overcontrol, taken together the configurations provide a more robust basis on which to rest our conclusions regarding overcontrol. In any case, as further illustrated in the Ozone Transport Policy Analysis Final Rule TSD, our analysis under both configurations establishes that there is no overcontrol and so there is no need to reconcile any difference in results between them.

We also looked at the ordering of increments of emissions reduction and have found that it does not matter whether we assume EGU emissions controls would be applied first, followed by non-EGU controls, or vice-versa. For 2023, the question is moot as there are only EGU reductions to examine. For 2026, the analysis showed there would be no overcontrol either way. In 2026, the EPA's overcontrol analysis (as presented here) examined all EGU reductions first and layered in non-EGU reductions in the last step of the overcontrol check. However, the EPA also examined an alternative ordering scenario where the non-EGU reductions were assessed prior to the EGU reductions associated with installation of new SCR post-combustion controls (see the Ozone Transport Policy Analysis Final Rule TSD for details). This ordering did not impact the results of the overcontrol test. The specific results of these analyses are presented in the TSD.

The control stringency selected for 2023 (a representative cost threshold of $1,800 per ton for EGUs) includes emissions reductions commensurate with optimization of existing SCRs and SNCRs and installation of state-of-the-art combustion controls, is not estimated to change the status of any receptors. Thus, the nonattainment or maintenance receptors that the states are linked to remain unresolved. Nor do any states' contribution levels drop below the 1 percent of NAAQS threshold. Thus, the EPA determined that none of the 23 linked states have all of their linkages resolved at the final EGU level of control stringency in 2023, and hence, the EPA finds no over-control in the final level of stringency.

Based on the air quality baseline modeling for 2026, all receptors to which Alabama, Minnesota, and Wisconsin are linked in 2023 are projected to be in attainment in 2026. Therefore, no additional stringency is finalized for EGUs or non-EGUs in those states beyond the 2023 level of stringency. For the remaining 20 states, the selected control stringency beginning in 2026 includes additional EGU controls and the non-EGU emissions reductions.

The EPA assesses air quality impacts and overcontrol in the year 2026 in this final rule, even though the rule accommodates the potential need for individual facilities (both EGU and non-EGU) to have some additional time to come into compliance. The EPA views this additional time to be a reflection of need (based on demonstrated impossibility) that is justified at Step 4 of the interstate transport framework rather than at Step 3. As explained in section VI.A of this document, with respect to EGUs, the EPA extends the full implementation of the SCR retrofit-based reductions across 2026 and 2027 to accommodate any unit-level scheduling challenges. However, we find that many sources can meet a three-year installation time and the trading program features and the allowance price will incentivize these reductions to occur as soon as possible. Similarly, with respect to non-EGU industrial sources, the final rule provides limited circumstances for individual facilities to seek and to be granted extensions of time to install required pollution controls and achieve the emissions rates established in this rule based on a showing of necessity. Those circumstances where an extension may be warranted for any specific facility are unknown at this time and will be evaluated through a source-specific application process, where the need for extension can be established with source-specific evidence. See section VI.C of this document. Further, 2026 is the critical analytic year associated with the last full ozone season before the 2027 Serious area attainment date and is the year by which significant contribution must be eliminated if at all possible. Therefore, for purposes of this analysis, the collective state and regional representation of these reductions are fully assumed in 2026. The potential ability of both EGU and non-EGU sources to have some amount of additional time beyond 2026 to comply with requirements that we have determined at Step 3 are necessary to eliminate significant contribution does not necessitate evaluating a later year than 2026 for overcontrol. The stringency of the control program does not alter in any year beyond 2026. By fully reflecting all Step 3 emissions reductions in its overcontrol test for 2026, EPA ensures that it is not understating the emissions impact and benefit when performing the test.

The EPA used the Ozone AQAT to evaluate the impact of this selected stringency level (as well as other potential stringency levels) on all receptors remaining in the 2026 air quality modeling base case. This assessment shows that the selected control stringency level is estimated to change the status of three receptors to attainment or maintenance in 2026. Brazoria County, Texas (Monitor ID 480391004); and Galveston County, Texas (Monitor ID 481671034), are estimated to come into attainment. We observe that one of the Fairfield, Connecticut, receptors (Monitor ID 090013007) is estimated to go from nonattainment to maintenance (when EGU emissions reductions with SCR are applied, prior to the application of the non-EGU emissions reductions). This receptor is expected to remain in maintenance even after the application of the non-EGU emissions reductions. Based on these data, EPA finds that all linked states except Arkansas, Mississippi, and Oklahoma are projected to continue to be linked to nonattainment or maintenance receptors after implementation of all identified Step 3 reductions, and hence, the EPA finds no over-control in its determination of that level of stringency for those states. Arkansas, Mississippi, and Oklahoma are linked to at least one of the two Texas receptors that are projected to come into attainment with the full implementation of the control strategy at Step 3. However, these two Texas receptors are expected to remain as maintenance-only receptors prior to the final increment of reductions assessed (the addition of the non-EGU reductions), so EPA concludes that imposition of the incremental non-EGU level is appropriate to avoid under-control as to these states and does not constitute overcontrol.

Next, the EPA evaluated the potential for over-control with respect to the 1 percent of the NAAQS threshold applied in this final rulemaking at Step 3 of the good neighbor framework, assessed for the selected control stringencies for each state for each period that downwind nonattainment and maintenance problems persist ( i.e., 2023 and 2026). Specifically, the EPA evaluated whether the selected control stringencies would reduce upwind emissions to a level where the contribution from any of the 23 linked states in 2023 or 20 linked states in 2026 would be below the 1 percent threshold. The EPA finds that for the mitigation measures assumed in 2023 and in 2026, all states that contributed greater than or equal to the 1 percent threshold in the base case are projected to continue to contribute greater than or equal to 1 percent of the NAAQS to at least one remaining downwind nonattainment or maintenance receptor for as long as that receptor remained in nonattainment or maintenance. EPA notes that in 2026, for Oklahoma, when the incremental level of stringency associated with the non-EGU control strategy is applied, Oklahoma's contribution to Galveston County Texas is expected to drop below the 1 percent threshold (at the same time that the receptor has its maintenance problems resolved). EPA concludes that this does not constitute overcontrol because both the receptor and the contribution are estimated to remain above the maintenance level and linkage threshold at the prior level of stringency and, thus, since otherwise justified at Step 3, the full stringency for 2026 is appropriate to avoid under-control. For more information about this assessment, refer to the Ozone Transport Policy Analysis Final Rule TSD and the Ozone AQAT.

Therefore, EPA finds that all of the selected EGU and non-EGU NO_X reduction strategies selected in EPA's Step 3 analysis can be applied to all states linked in 2026 to eliminate significant contribution to nonattainment and interference with maintenance of the 2015 ozone NAAQS without introducing an overcontrol problem based on the present record. The Supreme Court has directed the EPA to avoid both over-control and under-control in addressing good neighbor obligations. In addition, the D.C. Circuit has reinforced that over-control must be established based on particularized, record evidence on an as-applied basis.

The determination that the stringency of this action does not constitute overcontrol for any linked state is further reinforced by EPA's observation in section III.A of this document regarding the nature of the ozone problem. Ozone levels are known to vary, at times dramatically, from year to year. Future ozone concentrations and the formation of ground level ozone may also be impacted by factors in future years that the EPA cannot fully account for at present. For example, changes to meteorological conditions could affect future ozone levels. Climate change could also contribute to higher than anticipated ozone levels in future years through wildfires and heat waves, which can contribute directly and indirectly to higher levels of ozone. Any modeling projection can be characterized as having some uncertainty, and that is not a sufficient reason to ignore modeling results. However, in the context of the overcontrol test, the question is whether it is clear according to particularized evidence that there is no need for the emissions reductions in question. See EME Homer City, 572 U.S. at 523 (“[A] degree of imprecision is inevitable in tackling the problem of interstate air pollution. Slight changes in wind patterns or energy consumption, for example, may vary downwind air quality in ways EPA might not have anticipated.”). Under this standard, the degree of attainment that is projected to occur under the rule in relation to the Texas receptors discussed above is not so large or certain to occur that it would be appropriate to attempt to devise a less stringent emissions control strategy for the relevant linked states as a result, particularly in light of the fact that at the penultimate stringency level the receptors are not resolved.

It is also possible that ozone-precursor emissions from certain sources may decline beyond what we currently project in this rule. For example, the IRA may result in reductions in fossil-fuel fired generation, which should in turn result in lower NO_X emissions during the ozone season. We have assessed this scenario to ensure our overcontrol conclusions are robust even if the IRA has those effects. As discussed in the Regulatory Impact Analysis, the EPA conducted additional modeling of the final policy scenario (inclusive of economically efficient methods of compliance available within the Step 4 implementation programs) using its IPM tool. The EPA observes that the differences in estimated costs and emissions reductions in the IRA sensitivity (presented in Appendix 4A of the RIA) suggests that there would also be differences in estimated health and climate benefits under that scenario, although the Agency did not have time under this rulemaking schedule to quantify those differences. The EPA also used AQAT to conduct an additional EGU modeling sensitivity reflecting the IRA. Both the IPM sensitivity and the corresponding AQAT assessment of the IRA scenarios demonstrated no overcontrol as every state linkage to a downwind problematic receptor persisted in the penultimate level of stringency when EPA performed its Step 3 evaluation—even when the impacts of the IRA are incorporated. This further affirmed EPA's conclusion of no overcontrol concerns at the stringency level of the final rule. This overcontrol sensitivity is further discussed in the Ozone Transport Policy Analysis Final Rule TSD, Appendix K.

In light of the mandate of the CAA to protect the public health and environment through the elimination of significant contribution under the Good Neighbor Provision for the 2015 ozone NAAQS, nothing in the present record establishes on an as-applied, particularized basis that this rule will result in an unnecessary degree of control of upwind-state emissions.

Comment: Many commenters alleged that the rule overcontrols emissions by more than necessary to eliminate significant contribution for the 2015 ozone NAAQS, on the basis that the emissions reductions are unnecessary or are unnecessarily stringent.

Response: As discussed earlier in this section, EPA has analyzed whether this rule “overcontrols” emissions and has found based on a robust, multi-faceted analysis, that it does not. In particular, EPA has not identified a lesser-stringency emissions control strategy for any state that would either fully resolve the air quality problems at a downwind receptor location or resolve that upwind state's linkage to a level below the 1 percent of NAAQS contribution threshold. No commenter has provided a particularized, as-applied analysis demonstrating that EPA's emissions control strategy will actually result in any overcontrol of emissions in the manner the EPA or courts have understood that term, and overcontrol allegations must be proven through particularized, as-applied challenges. See EME Homer City, 795 F.3d at 127; see also Wisconsin, 938 F.3d at 325 (“[T]he way to contest instances of over-control is not through generalized claims that EPA's methodology would lead to over-control, but rather through a `particularized, as-applied challenge.' ” Accordingly, as we did when presented with similar arguments in EME Homer III, we reject Industry Petitioners' arguments because they do no more than speculate that aspects of `EPA's methodology could lead to over-control of upwind States.' ”) (cleaned up) (citing EME Homer City, 795 F.3d at 136–137).

Comment: For 2 of the 20 states linked in 2026, Arkansas and Mississippi, the last downwind receptor to which these two states are linked ( i.e., Brazoria County, Texas) was estimated to achieve attainment and maintenance after full application of EGU reductions and Tier 1 non-EGU reductions at proposal. Commenters noted that this suggested application of the estimated non-EGU, and/or some EGU, emissions reductions constituted over-control for these states.

Response: EPA notes that at proposal, this downwind receptor only resolved by a small margin after the application of all EGU and Tier 1 non-EGU emissions reductions. As explained earlier in this section, the final rule air quality modeling shows that the receptors to which these states are linked do not resolve upon full implementation of the identified EGU reductions by themselves, and only reach attainment by a small degree following the additional reductions from the non-EGU control strategy. If the EPA were to select the control stringency of this penultimate step, both upwind-state contribution and downwind-state air quality receptors would persist while the cost-effective emissions reductions that were identified to eliminate significant contribution remain available but un-implemented. This would constitute under-control. Consequently, as described, the EPA views the control stringency required of these states in this final rule as not constituting over-control and appropriate to eliminate significant contribution to nonattainment and interference with maintenance of this NAAQS in line with our Step 3 determinations for all other states. See the Ozone Transport Policy Analysis Final Rule TSD section C.3 for discussion and analysis regarding overcontrol for states solely linked to one or both of these receptors.

Comment: Commenters raised a variety of arguments that the enhancements to the EGU trading program in this action will result in overcontrol of power plant emissions. They alleged that dynamic budgeting would cause the budget to continually decrease even after significant contribution is eliminated. They similarly argue that annual emissions bank recalibration and the emissions backstop emissions rate have not been shown to be justified to eliminate significant contribution.

Response: This final rule's determination regarding the appropriate level of control stringency for EGUs finds that the amounts of NO_X emissions reduction achieved through these strategies at EGUs are appropriate and cost-justified under the Step 3 multifactor analysis. These determinations are associated with particular emissions control technologies and strategies as detailed in sections V.B.1 and V.C.1 above. It is the implementation of those strategies at the covered EGU sources and the air quality effects of those strategies (coupled with non-EGUs) in the relevant analytic year of 2026 on which we base our determination of significant contribution at Step 3. This includes the evaluation of whether there is overcontrol, which is also conducted for the 2026 analytic year as explained above. As explained below, we disagree that the enhancements to the trading program at Step 4 implicate the need for further overcontrol analysis. These enhancements operate together to ensure the trading program continues to maintain the Step 3 emissions control stringency over time. These enhancements reflect lessons learned through EPA's experience with prior trading programs implemented under the good neighbor provision. None of commenters' arguments that these enhancements result in overcontrol are persuasive.

Commenters contend that these enhancements to the trading program go beyond a mass-based budget approach as applied in CSAPR. Because these improvements in the program result in a continuing incentive for each covered EGU source to maintain the pollution control performance the EPA found appropriate to eliminate significant contribution at Step 3, commenters believe these enhancements must necessarily result in prohibited overcontrol. These arguments appear to be premised on the assumption that overall emissions may later decline to such a point that there is no longer a linkage between a particular state and any downwind receptors for reasons other than the requirements of this rule.

As an initial matter, no commenter has provided an empirical analysis demonstrating that the control stringency identified at Step 3 to eliminate significant contribution would actually result in any overcontrol. The case law is clear that over-control allegations must be proven through particularized, as-applied challenges. See prior response to comments. More importantly here, the Group 3 trading program enhancements do not impose increased stringency in years after 2030 and do not force emissions to continually be reduced to ever lower levels. They are only designed to incentivize the implementation of the Step 3 emissions control stringency that eliminates significant contribution. The circumstances that could potentially cause a receptor or linkage to resolve at some point in the future after 2026 are not circumstances that are within the power of this rule to control. Nor would those circumstances present a justification as to why upwind sources should no longer be obligated to eliminate their own significant contribution. Wisconsin, 938 F.3d at 324–25 (rejecting overcontrol arguments premised on attributing air quality problems to other emissions).

Further, the EPA is not constrained by the statute to only implement good neighbor obligations through fixed, unchanging, mass-based emissions budgets. See section III.B.1 of this document. The EPA has defined the “amount” of emissions that must be prohibited to eliminate significant contribution in this action based on a series of determinations of which emissions control strategies, for certain identified EGU and non-EGU sources, are appropriate applying the Step 3 multifactor analysis. Notably, the non-EGU industrial source emissions reductions in this action are not being achieved at Step 4 through mass-based emissions trading, nor are they required to be by any provision of the CAA. See section III.B.1.

As explained in sections III.B.1.d and VI.B.1 of this document, the EPA finds good reason based on its experience with trading programs that using fixed, mass-based, ozone-season wide budgets does not necessarily ensure the elimination of significant contribution over the entire region of linked states or throughout each ozone season. Even in the original CSAPR rulemaking, which promulgated only fixed, mass-based budgets, such outcomes were never the EPA's intention to allow. See, e.g.,76 FR 48256–57 (“[I]t would be inappropriate for a state linked to downwind nonattainment or maintenance areas to stop operating existing pollution control equipment (which would increase their emissions and contribution).”). Despite the EPA's expectations in CSAPR, the experience of the Agency since that time establishes a real risk of “under-control” if the existing trading framework is not enhanced. See EME Homer City, 572 U.S. at 523 (“[T]he Agency also has a statutory obligation to avoid `under-control,' i.e., to maximize achievement of attainment downwind.”).

Further, the EPA has already once adjusted its historical approach to better account for known, upcoming changes in the EGU fleet to ensure mass-based emissions budgets adequately incentivize the control strategy determined at Step 3. This adjustment was introduced in the Revised CSAPR Update. See82 FR 23121–22. The EPA now believes it is appropriate to ensure in a more comprehensive manner, and in perpetuity, that a mass-based emissions-trading framework incentivizes continuing implementation of the Step 3 control strategies to ensure significant contribution is eliminated in all upwind states and remains so. This is fully analogous in material respect to an approach to implementation at Step 4 that relies on application of unit-specific emissions limitations, which under the Act would typically apply in perpetuity and may only be modified through a future SIP- or FIP-revision rulemaking process. See CAA section 110(i) prohibiting modifications to implementation plan requirements except by enumerated processes. The availability of unit-specific emissions rates as a means to eliminate significant contribution is discussed in further detail in section III.B.1 of this document. The EPA also explained this in the proposal. See87 FR 20095–96.

Further, these enhancements are directly related to assisting downwind areas specifically with the goal of attaining and maintaining the 2015 8-hour ozone NAAQS. In this respect, they are not “unnecessary” or “unrelated” to carrying out the mandates of CAA section 110(a)(2)(D)(i)(I). Taking measures to ensure that each upwind source covered by an emissions trading program is adequately incentivized to eliminate excessive emissions (as found at Step 3) throughout the entirety of each ozone season is entirely appropriate in light of the nature of the ozone problem. Ozone exceedances recur on varying days throughout the summertime ozone season, and it is not possible to predict in advance which specific days will have high ozone. Further, impacts to public health and the environment from ozone can occur through short-term exposure ( e.g., over a course of hours, i.e., on a daily basis). The 2015 ozone NAAQS is expressed as an 8-hour average, and only a small number of days in excess of the ozone NAAQS can cause a downwind area to be in nonattainment. Thus, even a small number of exceedances can result in continuing and/or increased regulatory burdens on the downwind jurisdiction. Taking these considerations into account, it is evident that a fixed, mass-based emissions program that does not adequately incentivize emissions reductions commensurate with our Step 3 determinations on each day of every ozone season going forward does not provide a sufficient guarantee that the emissions that significantly contribute on those particular days and at particular receptor locations when ozone levels are at risk of exceeding the NAAQS have been eliminated. See section V.B.1.a and VI.B of this document for more discussion of data observations regarding SCR optimization.

These enhancements are also consistent with the general policies and principles EPA has long applied in implementing the NAAQS through the SIP/FIP framework of section 110. Emissions control measures relied on to meet CAA requirements must be permanent and enforceable and included in the implementation plan itself. See, e.g., Montana Sulfur & Chem. Co. v. EPA, 666 F.3d 1174, 1196 (9th Cir. 2012); 40 CFR 51.112(a). In the General Preamble laying out EPA's plans for implementing the 1990 CAA Amendments, the EPA identified a core “principle” that control strategies should be “accountable.” “This means, for example, that source-specific limits should be permanent and must reflect the assumptions used in the SIP demonstrations.” 57 FR 13498, 13568 (April 16, 1992). EPA went on, “The principles of quantification, enforceability, replicability, and accountability apply to all SIPs and control strategies, including those involving emissions trading, marketable permits and allowances.” Id. EPA also explained that its “emissions trading policy provides that only trades producing reductions that are surplus, enforceable, permanent, and quantifiable can get credit and be banked or used in an emissions trade.” Id. These principles follow from the language of the Act, including CAA section 110(a)(2), 107(d)(3)(E)(iii), 110(i), and 110(l). These provisions and principles further underscore the importance of ensuring that the emissions reductions the EPA has found necessary to eliminate significant contribution are in fact implemented on a consistent and permanent basis even within the context of an emissions trading program.

The EPA disagrees that the budget adjustments that would occur over time under this final rule (for example, the annual dynamic-budget adjustment) must be reassessed each time they occur through notice and comment rulemaking under CAA section 307(d). This would serve no purpose. The formulas that the EPA will apply to adjust the budgets and allowance bank are set in this final rule and are intended to maintain, not increase (or decrease), program stringency. While the EPA intends to provide an opportunity for stakeholders to review and propose corrections to its data as it implements the established budget formulas, no larger reassessment of the emissions control program is needed on an ongoing basis, because, again, that program is simply calibrated to ensure that emissions reductions commensurate with the determination of “significance” in Step 3 continue to be obtained over the long term. As described earlier, these trading program provisions are analogous to, or mimic, the effect of unit-specific emissions limitations that apply in perpetuity.

Commenters also confuse the “amount” of emissions that must be eliminated under CAA section 110(a)(2)(D)(i)(I) as being synonymous with a fixed, mass-based budget that reflects the residual emissions allowed following the elimination of significant contribution. However, EPA views the “amount” to be eliminated as those emissions that are in excess of the cost-effective emissions control strategies identified in Step 3. This is further explained in section III.B.1 of this document.

Thus, this rule is in compliance with the overcontrol principles that the D.C. Circuit applied on remand in EME Homer City to find certain instances of overcontrol in CSAPR's emissions control strategies. The D.C. Circuit found that EPA had imposed more stringent emissions-control strategies for certain states than were necessary to resolve all of those states' linkages. 795 F.3d at 128–30. Specifically, for sulfur dioxide, the court found certain receptors would reach attainment if all linked upwind states had implemented “cost controls” at $100/ton or $400/ton, rather than EPA's selected stringency level of $500/ton. Similarly, for ozone season NO_X, the court found that receptors were projected to attain the NAAQS at stringencies below $500/ton. The court's focus was on the stringency of the emissions control obligations as determined through the application of cost thresholds at Step 3 of the analysis. The court did not hold that EPA may only use fixed, mass-based budgets to implement those reductions. The court did not hold that EPA must permit individual polluting sources to be allowed to increase their emissions at some point in the future. The court did not hold that EPA's good neighbor FIPs must, effectively, contain termination clauses, such that they cease to ensure the implementation of the control stringency determined as necessary at Step 3, the moment a downwind receptor reaches attainment. Indeed, such a rule would contravene the statute's clear, forward-looking directive that EPA must also eliminate upwind emissions that interfere with maintenance of the NAAQS; see North Carolina, 531 F.3d at 908–911; Wisconsin, 938 F.3d at 325–26.

The EME Homer City court on remand in fact rejected various arguments that other aspects of EPA's emissions control strategy in CSAPR resulted in overcontrol, holding that EPA had properly given effect to the interfere with maintenance prong, and noting that petitioners failed to make out proven, as-applied demonstrations of overcontrol:

At bottom, each of those claims is an argument that EPA's methodology could lead to over-control of upwind States that are found to interfere with maintenance at a downwind location. That could prove to be correct in certain locations. But the Supreme Court made clear in EME Homer that the way to contest instances of over-control is not through generalized claims that EPA's methodology would lead to over-control, but rather through a “particularized, as-applied challenge.” EME Homer, 134 S. Ct. at 1609, slip op. at 31. And petitioners do not point to any actual such instances of over-control at downwind locations.

795 F.3d at 137. The court went on to observe, “EPA may only limit emissions `by just enough to permit an already-attaining State to maintain satisfactory air quality.' If States have been forced to reduce emissions beyond that point, affected parties will have meritorious as-applied challenges.” Id. (quoting 572 U.S. at 521–22). But this too was not a holding that EPA may not ensure effective and permanent implementation of an emissions control stringency that EPA has found warranted under CAA section 110(a)(2)(D)(i)(I). Such an approach is available through the more conventional CAA practice of setting unit-specific emissions limitations that would apply on a permanent and enforceable basis. See CAA sections 110(a)(2) and 302(y) (providing for SIPs and FIPs to include “enforceable emissions limitations” in addition to economic incentive measures like trading programs). This is in fact how EPA intends to ensure significant contribution is eliminated from non-EGU industrial sources for which a mass-based trading regime is, at least at the present time, unworkable (see section VI.C of this document). And EPA has provided for the elimination of significant contribution through source-specific emissions limitations in prior transport actions as well, so this position is not novel. See section III.B of this document.

Nonetheless, EPA recognizes that under the Act, both FIPs and SIPs may be revised, and states may replace FIPs with SIPs if EPA approves them. Any such revision must be evaluated to ensure no applicable CAA requirements are interfered with. See, e.g., Indiana v. EPA, 796 F.3d 803 (7th Cir. 2015). For example, states may be able to demonstrate in the future that through some other permanent and enforceable methods of emissions reduction that they have adopted into their SIP, they will be able to achieve a similar emissions control stringency with different emissions reduction requirements imposed on different sources as compared to the FIPs finalized in this action. See section VI.D of this document.

Therefore, commenters' contentions that EPA's trading program enhancements result in prohibited overcontrol are not proven through as-applied, particularized challenges, and they are premised on an incorrect understanding of the CAA and the relevant case law. The Agency rejects the contention that it must somehow provide in the present FIP action for a relaxation in the stringency of the Step 4 implementation program and thus allow for the recurrence of pollution that we have found here, in this action, significantly contributes to downwind ozone nonattainment and maintenance problems.

VI. Implementation of Emissions Reductions

A. NO _X Reduction Implementation Schedule

This action will ensure that emissions reductions necessary to eliminate significant contribution will be achieved “as expeditiously as practicable” and no later than the downwind attainment dates except where compliance by those dates is not possible. See CAA section 181(a); Wisconsin, 938 F.3d at 318–20. The timing of this action will provide for all possible emissions reductions to go into effect beginning in the 2023 ozone season for the covered states, which is aligned with the next upcoming attainment date of August 3, 2024, for areas classified as Moderate nonattainment under the 2015 ozone standard. Additional emissions reductions that the EPA finds not possible to implement by that attainment date will take effect as expeditiously as practicable. Emissions reductions commensurate with SCR mitigation measures for EGUs will start in 2026 and be fully implemented by 2027. Emissions reductions through the mitigation measures for industrial sources will generally go into effect in 2026; however, as explained in section VI.C of this document, we have provided for case-by-case extensions of up to one year based on a demonstration of necessity (with the potential for up to an additional two years based on a further demonstration). The full suite of emissions reductions is generally anticipated to take effect by the 2027 ozone season, which is aligned with the August 3, 2027, attainment date for areas classified as Serious nonattainment under the 2015 ozone NAAQS. This rule constitutes a full remedy for interstate transport for the 2015 ozone NAAQS for the states covered; the EPA does not anticipate further rulemaking to address good neighbor obligations under this NAAQS will be required for these states with the finalization of this rule.

EPA's determinations regarding the timing of this rule are informed by and in compliance with several recent court decisions. The D.C. Circuit has reiterated several times that, under the terms of the Good Neighbor Provision, upwind states must eliminate their significant contributions to downwind areas “consistent with the provisions of [title I of the Act],” including those provisions setting attainment deadlines for downwind areas. In North Carolina, the D.C. Circuit found the 2015 compliance deadline that the EPA had established in CAIR unlawful in light of the downwind nonattainment areas' 2010 deadline for attaining the 1997 NAAQS for ozone and PM_2.5. Similarly, in Wisconsin, the Court found the CSAPR Update unlawful to the extent it allowed upwind states to continue their significant contributions to downwind air quality problems beyond the downwind states' statutory deadlines for attaining the 2008 ozone NAAQS. In Maryland, the Court found the EPA's selection of a 2023 analysis year in evaluating state petitions submitted under CAA section 126 unlawful in light of the downwind Marginal nonattainment areas' 2021 deadline for attaining the 2015 ozone NAAQS. The Court noted in Wisconsin that the statutory command—that compliance with the Good Neighbor Provision must be achieved in a manner “consistent with” title I of the CAA—may be read to allow for some deviation from the mandate to eliminate prohibited transport by downwind attainment deadlines, “under particular circumstances and upon a sufficient showing of necessity,” but concluded that “[a]ny such deviation would need to be rooted in Title I's framework” and would need to “provide a sufficient level of protection to downwind States.”

1. 2023–2025: EGU NO_X Reductions Beginning in 2023

The near-term EGU control stringencies and corresponding reductions in this rulemaking cover the 2023, 2024, and 2025 ozone seasons. This is the period in which some reductions will be available, but the portion of full remedy reductions related to post combustion control installation identified in sections V.B through V.D of this document are not yet available. The EGU NO_X mitigation strategies available during these initial 3 years are the optimization of existing post-combustion controls (SCRs and SNCRs) and combustion control upgrades. As described in sections V.B through V.D of this document and in accompanying TSDs, these mitigation measures can be implemented in under two months in the case of existing control optimization and in 6 months in the case of combustion control upgrades. These timing assumptions account for planning, procurement, and any physical or structural modification necessary. The EPA provides significant historical data, including the implementation of the most recent Revised CSAPR Update, as well as engineering studies and input factor analysis documenting the feasibility of these timing assumptions. However, these timing assumptions are representative of fleet averages, and the EPA has noted that some units will likely overperform their installation timing assumptions, while others may have unit configuration or operational considerations that result in their underperforming these timing assumptions. As in prior interstate transport rules, the EPA is implementing these EGU reductions through a trading program approach. The trading program's option to buy additional allowances provides flexibility in the program for outlier sources that may need more time than what is representative of the fleet average to implement these mitigation strategies while providing an economic incentive to outperform rate and timing assumptions for those sources that can do so. In effect, this trading program implementation operationalizes the mitigation measures as state-wide assumptions for the EGU fleet rather than unit-specific assumptions.

However, starting in 2024, as described in section VI.B.7 of this document, unit-specific backstop daily emissions rates are applied to coal units with existing SCR at a level consistent with operating that control. The EPA believes that implementing these emissions reductions through state emissions budgets starting in 2023 while imposing the unit-specific backstop emissions rates in 2024 achieves the necessary environmental performance as soon as possible while accommodating any heterogeneity in unit-level implementation schedules regarding daily operation of optimized SCRs.

Additionally, as in prior rules, the EPA assumes combustion control upgrade implementation may take up to 6 months. In the Revised CSAPR Update, covering 12 of the 22 states for which emissions reduction requirements for EGUs are established under this action, the EPA finalized the rule in March of 2021 and thus did not require these combustion control-based emissions reductions in ozone-season state emissions budgets until 2022 (year two of that program). The EPA is applying the same timing assumption regarding combustion control upgrades for this rulemaking. Given the same relationship here between the date of final action and the year one ozone season, the EPA is not assuming the implementation of any additional combustion control upgrades in state emissions budgets until year two ( i.e., the 2024 ozone season). Any identified combustion control upgrade emissions reductions are reflected beginning in the 2024 ozone-season budgets for all covered states. For the 12 states covered under the Revised CSAPR Update, any identified emissions reduction potential from combustion control upgrade is included and reflected in those state budgets beginning in 2024—which means EGUs in those states have even more time than the 14 months between finalization of this rule and the 2024 ozone season if they started any planning or installation earlier in response to the Revised CSAPR Update.

2. 2026 and Later Years: EGU and Stationary Industrial Source NO_X Reductions Beginning in 2026

The EPA finds that it is not possible to implement all necessary emissions controls across all of the affected EGU and non-EGU sources by the August 3, 2024, Moderate area attainment date. In accordance with the good neighbor provision and the downwind attainment schedule under CAA section 181 for the 2015 ozone NAAQS, the EPA is aligning its analysis and implementation of the emissions reductions addressing significant contribution from EGU and non-EGU sources that require relatively longer lead time at a sectoral scale with the 2026 ozone season. The 2026 ozone season is the last full ozone season that precedes the August 3, 2027, Serious area attainment date for the 2015 ozone NAAQS. The EPA proposed to require compliance with all of the remaining EGU and non-EGU control requirements beginning in the 2026 ozone season. The EPA continues to find 2026 to be the relevant analytic year for purposes of its Step 3 analysis, including its analysis of overcontrol, as discussed in section V.D.4 of this document. However, many commenters argued that full implementation of the EGU and industrial source control strategies is not feasible for every source by the 2026 ozone season. The EPA addresses these technical comments specifically in sections V.B and VI.C of this document. The EPA also commissioned a study to develop a better understanding of the time needed for installation of emissions controls for the industrial sector units covered in this rule, which is included in the docket and discussed in section VI.A.2.b of this document. While the EPA does not agree with all of the commenters' assertions regarding the time they claim is needed for control installation, in other respects the concerns raised were sufficient to justify some adjustments to the compliance schedule for the final rule. We have provided for the emissions reductions commensurate with assumed EGU post-combustion emissions control retrofits to be phased in over the 2026 and 2027 ozone season emissions budgets, and we have provided a process in the final regulations for individual non-EGU industrial sources to seek limited compliance extensions extending no later than 2029 based on a case-by-case demonstration of necessity. This compliance schedule delivers substantial emissions reductions in the 2026 and 2027 ozone seasons and before the 2027 Serious area attainment date, and it only allows compliance extensions beyond that attainment date based on a rigorous, source-specific demonstration of need for the additional time.

The timing of this final rule provides three to four years for EGU and non-EGU sources to install whatever controls they deem suitable to comply with required emissions reductions by the start of the 2026 and 2027 ozone seasons. In addition, the publication of the proposal provided roughly an additional year of notice to these source owners and operators that they should begin engineering and financial planning (steps that can be taken prior to any capital investment) to be prepared to meet this implementation timetable.

The EPA views this timeframe for retrofitting post-combustion NO_X emissions controls and other non-EGU controls to be reasonable and achievable. A 3-year period for installation of control technologies is consistent with the statutory timeframe for implementation of the controls required to address interstate pollution under section 110(a)(2)(D) and 126 of the Act, the statutory timeframes for implementation of RACT in ozone nonattainment areas classified as Moderate or above, and other statutory provisions that establish control requirements for existing stationary sources of pollution.

For example, section 126 of the CAA authorizes a downwind state or tribe to petition the EPA for a finding that emissions from “any major source or group of stationary sources” in an upwind state contribute significantly to nonattainment in, or interfere with maintenance by, the downwind state. If the EPA makes a finding that a major source or a group of stationary sources emits or would emit pollutants in violation of the relevant prohibition in CAA section 110(a)(2)(D), the source(s) must shut down within three months from the finding unless the EPA directly regulates the source(s) by establishing emissions limitations and a compliance schedule extending no later than three years from the date of the finding, to eliminate the prohibited interstate transport of pollutants as expeditiously as practicable. Thus, in the provision that allows for direct Federal regulation of sources violating the good neighbor provision, Congress established three years as the maximum amount of time available from a final rule to when emissions reductions need to be achieved at the relevant source or group of sources. Because this action is not taken under CAA section 126(c), the mandatory timeframe for implementation of emissions controls under that provision is not directly applicable, but it is informative.

In response to arguments from sources that more time than has been provided in the final rule is necessary, this provision strongly indicates that allowing time beyond a three-year period must be based on a substantial showing of impossibility. Our analysis based on comments and considering additional information is that the additional time we have provided in the final rule is both justified and sufficient in light of the statutory objective of expeditious compliance.

Additionally, for ozone nonattainment areas classified as Moderate or higher, the CAA requires states to implement RACT requirements less than three years after the statutory deadline for submitting these measures to the EPA. Specifically, for these areas, CAA sections 182(b)(2) and 182(f) require that states implement RACT for existing VOC and NO_X sources as expeditiously as practicable but no later than May 31, 1995, approximately 30 months after the November 15, 1992, deadline for submitting RACT SIP revisions. For purposes of the 2015 ozone NAAQS, the EPA has interpreted these provisions to require implementation of RACT SIP revisions as expeditiously as practicable but no later than January 1 of the fifth year after the effective date of designation, which is less than three years after the deadline for submitting RACT SIP revisions. For areas initially designated nonattainment with a Moderate or higher classification effective August 3, 2018 (83 FR 25776), that implementation deadline falls on January 1, 2023, approximately 29 months after the August 3, 2020 submission deadline. Moderate ozone nonattainment areas must also implement all reasonably available control measures (including RACT) needed for expeditious attainment within three years after the statutory deadline for states to submit these measures to the EPA as part of a Moderate area attainment demonstration. Nonattainment areas for the 2015 ozone NAAQS that were reclassified to Moderate nonattainment in October 2022 face this same regulatory schedule, meaning that their sources are required to implement RACT controls in 2023. With the exception of the Uinta Basin, which is not an identified receptor in this action, no Marginal nonattainment area met the conditions of CAA section 181(a)(5) to obtain a one-year extension of the Moderate area attainment date. 87 FR 60899 (Oct. 7, 2022). Thus, all Marginal areas (other than Uinta) that failed to attain have been reclassified to Moderate. Id. In the October 2022 final rulemaking EPA made determinations that certain Marginal areas failed to attain by the attainment date, reclassified those areas to Moderate, and established SIP submission deadlines and RACM and RACT implementation deadlines. EPA set the attainment SIP submission deadlines for the bumped up Moderate areas to be January 1, 2023. See87 FR 60897, 60900. The implementation deadline for RACM and RACT is also January 1, 2023. Id.

The EPA notes that the types and sizes of the EGU and non-EGU sources that the EPA includes in this rule, as well as the types of emissions control technologies on which the EPA bases the emissions limitations that would take effect for the 2026 and 2027 ozone seasons, generally are consistent with the scope and stringency of RACT requirements for existing major sources of NO_X in downwind Moderate nonattainment areas and some upwind areas, which many states have already implemented in their SIPs. Thus, the timing Congress allotted for sources in downwind states to come into compliance with RACT requirements bears directly on the amount of time that should be allotted here and indicates, as does CAA section 126, that three years is an outer limit on the time that should be given sources to come into compliance where possible. In light of the January 1, 2023, deadline for implementation of RACT in Moderate nonattainment areas, the EPA finds that a May 1, 2026 deadline for full implementation of the emissions control requirements in this final rule would generally provide adequate time for any individual source to install the necessary controls, barring the circumstances of necessity discussed further in this section.

Finally, with respect to emissions standards for hazardous air pollutants, section 112(i)(3) of the CAA requires the EPA to establish compliance dates for each category or subcategory of existing sources subject to an emissions standard that “provide for compliance as expeditiously as practicable, but in no event later than 3 years after the effective date of such standard,” with limited exceptions. CAA section 112(i)(3)(B) authorizes the EPA to grant an extension of up to 1 additional year for an existing source to comply with emissions standards “if such additional period is necessary for the installation of controls,” and sections 112(i)(4) through (7) provide for limited compliance extensions where other conditions are met. Here again, where Congress was concerned with addressing emissions of pollutants that impact public health, a 3-year time period was allotted as the time needed for existing sources to come into compliance where possible. As discussed further in section VI.A.2.b of this document, the process for obtaining a compliance extension for industrial sources in this rule is generally modeled on 40 CFR 63.6(i)(3), which implements the extension provision for existing sources under CAA section 112(i)(3)(B).

All of these statutory timeframes for implementation of new control requirements on existing stationary sources indicate that Congress considered 3 years to be not only a sufficient amount of time but an upper bound of time allowable (barring instances of impossibility) for existing stationary sources to install or begin the installation of pollution controls as necessary for expeditious attainment, to eliminate prohibited interstate transport of pollutants, and to protect public health.

Further, the EPA notes that, given the number of years that have passed since EPA's promulgation of the 2015 ozone NAAQS and related nonattainment area designations in 2018, and in light of the Maryland court's holding that good neighbor obligations for the 2015 ozone NAAQS should have been implemented by the Marginal area attainment date in 2021, the implementation of good neighbor obligations for these NAAQS is already delayed, and the sources subject to NO_X emissions control in this rule have continued to operate for several years without the controls necessary to eliminate their significant contribution to ongoing and persistent ozone nonattainment and maintenance problems in other states. Under these circumstances, we find it reasonable to require compliance with the control requirements for all non-EGUs and the EGU reductions related to post-combustion control retrofit identified in section V.B.1.b of this document beginning in the 2026 ozone season (with full implementation by the 2027 ozone season for EGUs, and the availability of source-specific extensions based on a demonstration of necessity for non-EGUs).

As the D.C. Circuit noted in Wisconsin, the good neighbor provision requires upwind states to “eliminate their substantial contributions to downwind nonattainment in concert with the attainment deadlines” in the downwind states, even where those attainment deadlines occur before EPA's statutory deadline under CAA section 110(c) to promulgate a FIP. Referencing the Supreme Court's description of the attainment deadlines as “the heart” of the CAA, the Wisconsin court noted that some deviation from the mandate to eliminate prohibited transport by downwind attainment deadlines may be allowed only “under particular circumstances and upon a sufficient showing of necessity.”

For the reasons provided in the following sub-sections, the EPA finds that installation of certain EGU controls and all non-EGU controls is not possible by the Moderate area attainment date for the 2015 ozone NAAQS ( i.e., August 3, 2024), and, for certain sources, may not be possible by the 2026 ozone season or even the August 3, 2027, Serious area attainment date. While the EPA's technical analysis demonstrates that for any individual source, control installation could be accomplished by the start of the 2026 ozone season, in light of the scope of this rule coupled with current information on the present economic capacity of sources, control-installation vendors, and associated markets for labor and material, it is the EPA's judgment that a three-year timeframe is not possible for all sources subject to this rule collectively to come into compliance. Therefore, additional time beyond 2026 will be allowed for certain facilities in recognition of these constraints on the processes needed for installation of controls across all of the covered sources.

a. EGU Schedule for 2026 and Later Years

As discussed in sections V.B through V.D of this document, significant emissions reduction potential exists and is included in EPA's quantification of significant contribution based on the potential to install post-combustion controls (SCR and SNCRs) at EGUs. However, as discussed in detail in those sections, the assumption for installation of this technology on a region-wide scale is 36–48 months in this final rule. This amount of time allows for all necessary procurement, permitting, and installation milestones across multiple units in the covered region. Therefore, the EPA finds that these emissions reductions are not available any earlier than the 2026 compliance period. Starting in 2026, state emissions budgets will reflect full implementation of assumed SNCR mitigation measures and implementation of half the emissions reduction potential identified for assumed SCR mitigation measures. For each year in 2027 and beyond, state emissions budgets include all of the emissions reductions commensurate with these post-combustion control technologies identified for covered units in Step 3. The EPA notes that similar compliance schedules and post-combustion control retrofit installations have been realized successfully in prior programs allowing similar timeframes. Subsequent to the NO_X SIP Call and the parallel Finding of Significant Contribution and Rulemaking on Section 126 Petitions (which became effective December 28, 1998, and February 17, 2000, respectively ), nearly 19 GW of SCR retrofit came online in 2002 and another 42 GW of SCR retrofit came online for steam boilers in 2003, illustrating that a considerable volume of SCR retrofit capacity is possible within a 36-month period.

Comment: Some commenters disagreed with EPA's proposed 36-month timeframe for SCR retrofit. These commenters noted that, while possible at the unit or plant level, the collective volume of assumed SCR installation would not be possible given the labor constraints, supply constraints, and simultaneous outages necessary to complete SCR retrofit projects on such a schedule. They noted that many of the remaining coal units lacking SCR pose more site-specific installation challenges than those that were already retrofitted on a quicker timeframe.

Response: EPA is making several changes in this final rule to address these concerns. First, EPA is phasing in emissions reductions commensurate with assumed SCR installations consistent with a 36-to-48-month time frame in this final rule, instead of a 36-month time frame as proposed. EPA is implementing half of this emissions reduction potential in 2026 ozone-season NO_X budgets for states containing these EGUs and the other half of this emissions reduction potential in 2027 ozone-season NO_X budgets for those states. This phase-in approach to implementing SCR retrofit reduction potential over a three to four year period is in response to comments, including those from third-party full-service engineering firms. These commenters highlighted that while the proposed 36-month time frame is viable at the plant level, it would be “very unlikely” that the collective volume of SCR capacity could be installed in a three-year time frame based on a variety of factors. First, the commenters identified constraints on labor needed to retrofit 32 GW of capacity, highlighting that the Bureau of Labor and Statistics projects that there will be a decline in boilermaker employment over the decade and that the Associated Builders and Contractors (ABC) identifies the need for 650,000 additional skilled craft professionals on top of the normal hiring pace to meet the economy-wide demand created by infrastructure investment and other clean energy projects ( e.g., carbon capture and storage). They highlighted the decline in companies serving this type of large-scale retrofit project as the lack of new coal units and the retirement of coal units has curtailed activity in this area over the past five years. They also identified supply bottlenecks for key SCR components that would slow the ability to implement a large volume of SCR within 3 years, affecting electrical conduits, transformers, piping, structural and plate steel, and wire (with temporary price increases ranging from 30 percent to 200 percent). Finally, commenters note that site-specific conditions can make retrofits for individual units a lengthier process than historical averages ( e.g., under prior rules more accommodating sites retrofitted first) and that four years may be necessary for some projects, accordingly. EPA found the technical justification submitted in comment consistent with its prior assessments that a range of 39–48 months is appropriate for SCR-retrofit timing within regional-scale programs. Therefore, EPA is adjusting the timeframe to still incentivize these reductions by the attainment date while accommodating the potential for some SCR retrofits to require between 36–48 months for installation.

Some commenters requested more than 48 months for SCR installation based on past projects that took five or more years. EPA disagrees with these commenters for two reasons. First, while EPA is identifying SCR retrofit potential to define significant contribution at Step 3, the rule only requires emissions reductions commensurate with that technology, implemented through a trading program, meaning that operators of EGUs eligible for SCR retrofit may pursue a variety of strategies for reducing emissions. Such compliance flexibility will accommodate extreme or unique circumstances in which a desired SCR retrofit is not achieved by the 2027 ozone season, although EPA finds such a circumstance exceedingly unlikely. Second, the historical examples that exceeded 48 months do not necessarily demonstrate that such projects are impossible to execute in less than 48 months, but rather that they can extend beyond that timeframe if no requirements or incentives are in place for a faster installation. As the D.C. Circuit has recognized, historical data on the amount of time sources have taken to install pollution controls do not in themselves establish the minimum amount of time in which those controls could be installed if sources are subject to a legal mandate to do so. See Wisconsin, 938 F.3d at 330 (“[A]ll those anecdotes show is that installation can drag on when companies are unconstrained by the ticking clock of the law.”).

b. Non-EGU or Industrial Source Schedule for 2026 and Later Years

The EPA proposed to require that all emissions reductions associated with the requirements for non-EGU industrial sources go into effect by the start of the 2026 ozone season, but also requested comment on its control-installation timing estimates for non-EGUs and requested comment on the possibility of providing for limited compliance extensions based on a showing of necessity. See87 FR 20104–05.

Comment: The EPA received numerous comments regarding the inability of various non-EGU industries to install controls to comply with the emissions limits by 2026. Specifically, commenters raised concerns regarding the ability to meet these deadlines due to the ongoing geopolitical instability triggered by the war in Ukraine, COVID–19 pandemic-driven disruptions, and supply chain delays and shortages. Commenters also claimed that the EPA's three-year installation timeframe for non-EGUs does not account for the time needed to obtain necessary permits. Commenters stated that even where controls are feasible for a source, some sources would need to shut down due to their inability to install controls by 2026 and requested that the EPA provide additional time for sources to come into compliance. Commenters from multiple non-EGU industries stated that the proposed applicability criteria will require controls to be installed on thousands of non-EGU emissions units. Because of the number of emissions units, commenters raised concerns with permitting delays and the unavailability of skilled labor and necessary components. Commenters suggested various timelines for control installation timing ranging from one additional year to seven years. Other commenters asserted that the data supported the conclusion that all non-EGU sources, or at least some non-EGU sources, could install controls by 2026 or earlier, and that EPA has a legal obligation to impose good neighbor requirements as expeditiously as practicable by such sources, including earlier than 2026 if possible.

Response: After reviewing the information received during the public comment period and the additional information presented in the Non-EGU Control Installation Timing Report, the EPA has concluded that the majority of non-EGUs can install and operate the required controls by the 2026 ozone season. For the non-EGU control requirements on which the EPA has based its Step 3 findings as described in section V of this document, the emissions limits will generally go into effect starting with the 2026 ozone season (except where an individual source qualifies for a limited extension of time to comply based on a specific demonstration of necessity, as described in this section). The EPA finds that meeting the emissions limitations of this final rule through installation of necessary controls by an ozone season before 2026 is not expected to be possible for the industrial sources covered by this final rule.

The EPA recognizes that labor shortages, supply shortages, or other circumstances beyond the control of source owner/operators may, in some cases, render compliance by 2026 impossible for a particular industrial source. Therefore, the final rule contains provisions allowing source owner/operators to request limited compliance extensions based on a case-by-case demonstration of necessity. Under these provisions, the owner or operator of a source may initially apply for an extension of up to one year to comply with the applicable emissions control requirements, which if approved by the EPA, would require compliance no later than the 2027 ozone season. The EPA may grant an additional case-based extension of up to two additional years for full compliance, where specific criteria are met.

The EPA initiated a study to examine the time necessary to install the potential controls identified in the final rule's cost analysis for all of the non-EGU industries subject to the final rule, including SNCR, low NO_X burners, layered combustion, NSCR, SCR, fluid gas recirculation, and SNCR/advanced selective noncatalytic reduction (ASNCR). The resulting report, which we refer to as the “Non-EGU Control Installation Timing Report,” identified a range of estimated installation times with minimum estimated installation times ranging from 6–27 months without any supply chain delays and 6–40 months with potential supply chain delays depending on the industry. The Non-EGU Control Installation Timing Report also identified maximum estimated installation times ranging from 12–28 months without any supply chain delays and 12–72 months with potential supply chain delays depending on the industry. As indicated in the Non-EGU Control Installation Timing Report, the installation of layered combustion and NSCR control technology, in particular, could take between 9 and 72 months depending on supply chain delays. The report also indicated that permitting processes may take 6 to 12 months but noted that these processes typically can proceed concurrent with other steps of the installation process.

We find that the potential time needed for permitting processes is generally unlikely to significantly affect installation timeframes of at least three years given that a source that has three or more years to comply is expected, in most cases, to have adequate time to apply for and secure the necessary permits during that time. Permitting processes may, however, impact shorter installation times ranging from 12–28 months. Given the 12–28 month estimate for minimum and maximum installation times without supply chain delays and permitting timeframes typically ranging from 6–12 months, the EPA finds that the controls for non-EGU sources needed to comply with this final rule are generally not expected to be installed significantly before the 2026 ozone season.

Generally, the Non-EGU Control Installation Timing Report indicated that all non-EGU unit types subject to the final rule could install controls within 28 months if there are no supply chain delays. Thus, the Non-EGU Control Installation Timing Report confirms that for any individual facility, meeting the emissions limitations of this final rule through installation of controls can be completed by the start of the 2026 ozone season. It is only when the number of units in the U.S. potentially affected by the rule is taken into account, coupled with broader considerations of economic capacity including current information on supply-chain delays, that the potential need for additional time beyond 2026 becomes a possibility. Under ideal economic conditions ( i.e., no supply-chain delays or other constraints), affected units are estimated to be capable to install both combustion and post-combustion controls before the 2026 ozone season. Many commenters, however, provided information on installation timing estimates based on current supply chain delays and labor constraints. These commenters generally stated that installation of the necessary controls for some units would take longer than three years if supply chain delays similar to those that have occurred over the past few years continue. The Non-EGU Control Installation Timing Report reflected this information, together with additional information gathered from pollution control vendors, to develop ranges of estimates of possible installation times given current ( i.e., 2022) labor market conditions and material supplies. The Non-EGU Control Installation Timing Report also discussed how the installation and optimization of post-combustion controls over a similar timeframe at both EGUs and non-EGUs subject to this final rule would, considered cumulatively, potentially affect the installation timing needs of the covered non-EGU sources.

Based on information provided by commenters and vendors, the Non-EGU Control Installation Timing Report indicated that if current supply chain delays continue, control installations could take as long as 61 months for most non-EGU industries and possibly as long as 64–112 months in difficult cases. Notably, however, the conclusions in the Non-EGU Control Installation Timing Report reflect three key assumptions that could result in the relatively lengthy timing estimates at the outer end of this range: (1) the current state of supply chain delays and disruptions would continue without any increase in labor supply, materials, or reduction in fabrication timing; (2) the labor and materials markets would not adjust in response to this rule in the timeframe needed to meet the increased demand for control installations; and (3) the Report was unable to account for some of the flexibilities built into the final rule that will allow owners and operators to install controls on the most cost-effective units with shorter installation times.

As presented in the Non-EGU Control Installation Timing Report, supply chain delays and disruptions have generally been lessening since they peaked in 2020 during the COVID–19 pandemic, and many economic indicators have showed some improvement towards pre-pandemic levels, including freight transportation, inventory to sales ratios, interstate miles traveled, U.S. goods imports, and supply chain indices. If these economic indicators continue to improve and the availability of fabricators and materials continues to trend upward, the control timing estimates identified in the Non-EGU Control Installation Timing Report could prove to be overstated for some industries and control technologies. In addition, the Non-EGU Control Installation Timing Report did not account for the labor and supply market adjustments that would be anticipated to occur to meet increased demand for control technologies and related materials and labor over the next several years in response to the rule. Cf. Wisconsin, 938 F.3d at 330 (“[A]ll those anecdotes [of elongated control installation times] show is that installation can drag on when companies are unconstrained by the ticking clock of the law.”). For example, some of the longer installation timeframes identified in the Non-EGU Control Installation Timing Report are based on assumed limits on the current availability of skilled labor needed to install combustion controls and post combustion controls. If the market adjusts in response to increasing demand for this type of skilled labor in the timeframe needed for compliance ( e.g., there is an increase in boilermaker and engine controls labor), the installation timing estimates in the Non-EGU Control Installation Timing Report again could be overstated.

The Non-EGU Control Installation Timing Report also did not account for flexibilities provided in this final rule that will enable owners and operators of certain affected units to identify the most cost-effective and efficient means for installing any necessary controls. For example, one concern highlighted by commenters was the amount of time necessary to install controls on engines that have been in operation for 50 or more years. The requirements that we are finalizing for engines in the Pipeline Transportation of Natural Gas industry include an exemption for emergency engines and provisions allowing source owner/operators to request the EPA approval of facility-wide emissions averaging plans, both of which enable owners and operators of affected units to take costs, installation timing needs, and other considerations into account in deciding which engines to control.

In response to industry concern about the number and size of units captured by the proposed applicability criteria, the EPA has made several changes to the applicability criteria in the final rule to focus the control requirements on impactful non-EGU units. As explained further in section VI.C of this document, the EPA is establishing exemptions for low-use boilers and engines where it would not be cost-effective to require controls at this time. Finally, as discussed in section VI.C.3 of this document, the EPA is not finalizing the proposed requirements for most emissions unit types in the Iron and Steel Mills and Ferroalloy Manufacturing industry given the EPA does not currently have a sufficient technical basis for finalizing those proposed requirements. These changes reduce the number of non-EGU units that will actually need to install controls and should reduce the strain on the labor and supply chain and permitting processes. For example, for engines, the EPA estimates that the facility-wide emissions averaging provision would, in many cases, allow facilities to install controls on only one-third of their engines, on average (see section VI.C.1 of this document for further discussion).

Taking all of these considerations into account, the EPA finds that the outer range of timing estimates presented in the Non-EGU Control Installation Timing Report generally reflects a conservative set of installation timing estimates and that the factors described previously could result in installation timeframes that fall toward the shorter end of the ranges of time that factor in supply-chain delays or could obviate those supply-chain delay issues entirely.

Based on all of these considerations, the EPA has concluded that three years is generally an adequate amount of time for the non-EGU sources covered by this final rule to install the controls in the 20 states that remain linked in 2026. The EPA also recognizes, however, that some sources may not be able to install controls by the 2026 ozone season despite making good faith efforts to do so, due to the aforementioned supply chain delays or other circumstances entirely beyond the owner or operator's control. Therefore, the final FIPs require compliance with the emissions control requirements for non-EGUs by the beginning of the 2026 ozone season, with limited exceptions based on a showing of necessity for individual sources that meet specific criteria. Where an individual owner or operator submits a satisfactory demonstration that an extension of time to comply is necessary, due to circumstances entirely beyond the owner or operator's control and despite all good faith efforts to install the necessary controls by May 1, 2026, the EPA may determine that installation by 2026 is not possible and thereby grant an extension of up to one year for that source to fully implement the required controls. If, after the EPA has granted a request for an initial compliance extension, the source remains unable to comply by the extended compliance date due to circumstances entirely beyond the owner or operator's control and despite all good faith efforts to install the necessary controls by the extended compliance date, the owner or operator may request and the EPA may grant a second extension of up to two additional years for full compliance, where specific criteria are met. This application process is generally in accordance with the concept on which the Agency requested comment in the proposal, see87 FR 20104–05, and is modeled on a similar process provided for industrial sources subject to CAA section 112 NESHAPs, found at 40 CFR 63.6(i)(3).

The EPA intends to grant a request for an initial compliance extension only where a source demonstrates that it has taken all steps possible to install the necessary controls by the applicable compliance date and still cannot comply by the 2026 ozone season, due to circumstances entirely beyond its control. Any request for a compliance extension must be received by the EPA at least 180 days before the May 1, 2026, compliance date. The request must include all information obtained from control technology vendors demonstrating that the necessary controls cannot be installed by the applicable compliance date, any permit(s) secured for the installation of controls or information from the permitting authority on the timeline for issuance of such permit(s) if the source has not yet obtained the required permit(s); and any contracts entered into by the source for the installation of the control technology or an explanation as to why no contract is necessary. The EPA may also consider documentation of a source owner's/operator's plans to shut down a source by the 2027 ozone season in determining whether a source is eligible for a compliance extension. The owner or operator of an affected unit remains subject to the May 1, 2026 compliance date unless and until the Administrator grants a compliance extension.

The EPA intends to grant a request for a second compliance extension beyond 2027 only where a source owner/operator submits updated documentation showing that it is not possible to install and operate controls by the 2027 ozone season, despite all good faith efforts to comply and due to circumstances entirely beyond its control. The request must be received by the EPA at least 180 days before the extended compliance date and must include, at minimum, the same types of information as that required for the initial extension request. The owner or operator of an affected unit remains subject to the initial extended compliance date unless and until the Administrator grants a second compliance extension. A denial will be effective on the date of denial.

As discussed earlier in section VI.A, in Wisconsin the court held that some deviation from the CAA's mandate to eliminate prohibited transport by downwind attainment deadlines may be allowed only “under particular circumstances and upon a sufficient showing of necessity.” This standard is met when, in the EPA's judgment, compliance by the attainment date amounts to an impossibility. The EPA cannot allow a covered industrial source to avoid timely compliance with the emissions control requirements established in this final rule unless the source owner/operator can demonstrate that compliance by the 2026 ozone season is not possible due to circumstances entirely beyond their control. The criteria that must be met to qualify for limited extensions of time to comply are designed to meet this statutory mandate. The EPA anticipates that the majority of the industrial sources covered by this final rule will not qualify for a compliance extension.

B. Regulatory Requirements for EGUs

To implement the required emissions reductions from EGUs, the EPA is revising the existing CSAPR NO_X Ozone Season Group 3 Trading Program (the “Group 3 trading program”) established in the Revised CSAPR Update both to expand the program's geographic scope and to enhance the program's ability to ensure favorable environmental outcomes. The EPA is using a trading program for EGUs because of the inherently greater flexibility that a trading program can provide relative to more prescriptive, “command-and-control” forms of regulation of sufficient stringency to achieve the necessary emissions reductions. In the electric power sector, EGUs' extensive interconnectedness and coordination create the ability to shift both electricity production and emissions among units, providing a closely related ability to achieve emissions reductions in part by shifting electricity production from higher-emitting units to lower-emitting or non-emitting units. Thus, while the Step 3 control-stringency determination for EGUs to eliminate significant contribution is based on strategies that do not require generation shifting or reduced utilization of EGUs, the sector's unusual flexibility with respect to how emissions reductions can be achieved makes the flexibility of a trading program particularly useful as a means of lowering the overall costs of obtaining such reductions. In addition, it is essential for the electric power sector to retain short-term operational flexibility sufficient to allow electricity to be produced at all times in the quantities needed to meet demand simultaneously, and the flexibility of a trading program can be helpful in supporting this aspect of the industry as well.

To ensure emissions reductions necessary to eliminate significant contribution are maintained, in this rulemaking, the EPA is making certain enhancements to the current provisions of the Group 3 trading program addressing emissions-control performance by some kinds of individual units that will necessarily reduce the flexibility of the program to some extent for those units. In analyzing significant contribution at Step 3, once a linkage has been established between an upwind state and a downwind receptor, we identify an appropriate set of emissions control strategies, considering cost and other factors, that would eliminate significant contribution from the upwind state without leading to undercontrol or overcontrol at the downwind linked receptors. At Step 4, for EGUs, we develop emissions budgets based on consistent application of the identified strategies to the sources. This level of emission control at each source identified in Step 3 is what the EPA deems to eliminate significant contribution, while the design of emission budgets that successfully implement that level of emission control is determined at Step 4. See section III.B and V.

The trading program enhancements discussed in this section are designed to ensure that sources actually achieve that level of emission control and thereby eliminate significant contribution on a permanent basis at Step 4. The enhancements ensure that the emissions budgets for EGUs continue to secure the level of emission control identified at Step 3 at the sources active in the trading program on a more consistent basis throughout each ozone season than prior transport trading programs (including those that did not provide complete remedies for interstate pollution transport) have required. An alternative form of implementation at Step 4 would be to implement source-specific emissions limitations ( e.g., rate-based standards expressed as mass per unit of heat input) reflecting the control strategies identified at Step 3. This is a very common form of implementation for many other CAA requirements and is indeed the manner of implementation selected in this very rulemaking for other affected industrial sources. See sections III.B, V.D.4, and VI.C. But doing so would require loss of the flexibilities inherent in a trading program, inclusive of these enhancements, that facilitate orderly and timely achievement of the required emission reductions in the power sector.

Prior to this rule, the Group 3 trading program has applied to EGUs meeting the program's applicability criteria within the borders of twelve states: Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, New Jersey, New York, Ohio, Pennsylvania, Virginia, and West Virginia. Affected EGUs in these twelve states will continue to participate in the Group 3 trading program as revised in this rulemaking, with some revised provisions taking effect in the 2023 control period and other revised provisions taking effect later as discussed elsewhere in this document. The EPA is expanding the Group 3 trading program's geographic scope to include all of the additional states for which EGU emissions reduction requirements are being established in this rulemaking. Affected EGUs within the borders of seven states currently covered by the CSAPR NO_X Ozone Season Group 2 Trading Program (the “Group 2 trading program”)—Alabama, Arkansas, Mississippi, Missouri, Oklahoma, Texas, and Wisconsin—will transition from the Group 2 trading program to the revised Group 3 trading program at the beginning of the 2023 control period, and affected EGUs within the borders of the three states not currently covered by any CSAPR trading program for seasonal NO_X emissions—Minnesota, Nevada, and Utah—will enter the Group 3 trading program in the 2023 control period on the effective date of this rule. As discussed in section VI.B.12.a of this document, because the effective date of the rule will likely be sometime during the 2023 ozone season, special transitional provisions have been developed to allow for efficient administration of the rule's EGU requirements through the Group 3 trading program while not imposing any new substantive obligations on parties prior to the rule's effective date, similar to the transitional provisions implemented under the Revised CSAPR Update.

As is the case for the states already in the Group 3 trading program, for each state added to the program, the set of affected EGUs will include new units as well as existing units and will also include units located in Indian country within the state's borders. Sections VI.B.2 and VI.B.3 of this rule provide additional discussion of the geographic expansion of the Group 3 trading program and the units in the expanded geography that will become subject to the program under the program's existing applicability provisions.

In addition to expanding the Group 3 trading program's geographic scope, the EPA is modifying the program's regulations prospectively to include certain enhancements to improve environmental outcomes. Two of the proposed enhancements will adjust the overall quantities of allowances available for compliance in the trading program in each control period so as to maintain the rule's selected control stringency and related EGU effective emissions rate performance level as the EGU fleet evolves. First, instead of establishing emissions budgets for all future years under the program at the time of the rulemaking, which cannot reflect future changes in the EGU fleet unknown at the time of the rulemaking, the EPA is revising the trading program regulations to include a dynamic budgeting procedure. Under this procedure, the EPA will calculate emissions budgets for control periods in 2026 and later years based on more current information about the composition and utilization of the EGU fleet, specifically data available from the 2024 ozone season and following ( e.g., for 2026, data from periods through 2024; for 2027, data from periods through 2025; etc.). Through the 2029 control period, the dynamically determined budgets will apply only if they are higher than preset budgets established in the rule. (Associated revisions to the program's variability limits and unit-level allowance allocation procedures will coordinate these provisions with the revised budget-setting procedures.) Second, starting with the 2024 control period, the EPA will annually recalibrate the quantity of accumulated banked allowances under the program to prevent the quantity of allowances carried over from each control period to the next from exceeding the target bank level, which would be revised to represent a preset percentage of the sum of the state emissions budgets for each control period. The preset percentage will be 21 percent for control periods through 2029 and 10.5 percent for control periods in 2030 and later years. Together, these enhancements will protect the intended stringency of the trading program against potential erosion caused by EGU fleet turnover and will better sustain over time the incentives created by the trading program to achieve the degree of emissions control for EGUs that the EPA has determined is necessary to address states' good neighbor obligations.

Two further enhancements to the Group 3 trading program establish provisions designed to promote more consistent emissions control by individual EGUs within the context of the trading program. First, starting with the 2024 control period for coal-fired EGUs with existing SCR controls and the earlier of the 2030 control period or the control period after which an SCR is installed for other large coal-fired EGUs, a daily NO_X emissions rate of 0.14 lb/mmBtu will apply as a backstop to the seasonal emissions budgets (which are based on an assumed seasonal average emissions rate of 0.08 lb/mmBtu for EGUs with existing SCR controls). Each ton of emissions exceeding a unit's backstop daily emissions rate, after the first 50 such tons, in a given control period will incur a 3-for-1 allowance surrender ratio instead of the usual 1-for-1 allowance surrender ratio. Second, also starting with the 2024 control period, the trading program's existing assurance provisions, which require extra allowance surrenders from sources that are found responsible for contributing to an exceedance of the relevant state's “assurance level” ( i.e., typically 121 percent of the state's emissions budget), will be strengthened by the addition of another backstop requirement. Specifically, for any unit equipped with post-combustion controls that is found responsible for contributing to an exceedance of the state's assurance level, the revised regulations will prohibit the unit's seasonal emissions from exceeding by more than 50 tons the emissions that would have resulted if the unit had achieved a seasonal average emissions rate equal to the higher of 0.10 lb/mmBtu or 125 percent of the unit's lowest previous seasonal average emissions rate under any CSAPR seasonal NO_X trading program.

These two enhancements are designed to ensure that all individual units with SCR controls have strong incentives to continuously operate and optimize their controls, and also to ensure that all units with post-combustion controls have strong incentives to optimize their emissions performance when a state's assurance level might otherwise be exceeded. These enhancements are generally designed to ensure consistency with the EPA's determination regarding the emissions control stringency needed from EGUs to eliminate significant contribution under the Step 3 multifactor analysis as discussed in section V of this document. Further, these enhancements are designed to provide greater assurance that emissions controls will be operated on all days of the ozone season and therefore necessarily on the days that turn out to be most critical for downwind ozone levels. The EPA expects that promoting more consistently good emissions performance by individual EGUs will better ensure that each state's significant contribution is fully eliminated by this action, see North Carolina, 531 F.3d at 919–21. In addition to addressing the statutory requirements of eliminating significant contribution, the EPA anticipates that these enhancements will also deliver public health and environmental benefits to underserved and overburdened communities.

The revisions to the Group 3 trading program being finalized in this rule are very similar to the proposed revisions. The changes from proposal to the set of states covered are driven largely by updates to the air quality modeling performed for the final rule, as described in section IV of this document. The changes from proposal to the trading program enhancements are generally being made in response to comments on the proposal, as discussed in more detail in the remainder of section VI.B of this document.

1. Trading Program Background and Overview of Revisions

a. Current CSAPR Trading Program Design Elements and Identified Concerns

The use of allowance trading programs to achieve required emissions reductions from the electric power sector has a long history, rooted in the Clean Air Act Amendments of 1990. In Title IV of those amendments, Congress specified the design elements for a 48-state allowance trading program to reduce SO₂ emissions and the resulting acid precipitation. Building on the success of that first allowance trading program as a tool for addressing multi-state air pollution issues, since 1998 EPA has promulgated and implemented multiple allowance trading programs for SO₂ or NO_X emissions to address the requirements of the CAA's good neighbor provision with respect to successively more protective NAAQS for fine particulate matter and ozone. Most of these trading programs have applied either exclusively or primarily to EGUs.

The EPA currently administers six CSAPR trading programs for EGUs (promulgated in CSAPR, the CSAPR Update, and the Revised CSAPR Update) that differ in the pollutants, geographic regions, and time periods covered and in the levels of stringency, but that otherwise have been nearly identical in their core design elements and their regulatory text. The principal common design elements currently reflected in all of the programs are as follows:

• An “emissions budget” is established for each state for each control period, representing the EPA's quantification of the emissions that would remain under certain projected conditions after elimination of the emissions prohibited by the good neighbor provision under those projected conditions. For each control period of program operation, a quantity of newly issued “allowances” equal to the amount of each state's emissions budget is allocated among the state's sources. (States have options to replace the EPA's default allocations or to institute an auction process.) Total emissions in a given control period from all sources in the program are effectively capped at a level no higher than the total quantity of allowances available for use in the control period, consisting of the sum of all states' emissions budgets for the control period plus any unused allowances carried over from previous control periods as “banked” allowances.

• “Assurance provisions” in each program establish an “assurance level” for each state for each control period, defined as the sum of the state's emissions budget plus a specified “variability limit.” The purpose of the assurance provisions is to limit the total emissions from each state's sources in each control period to an amount close to the state's emissions budget for the control period, consistent with the good neighbor provision's mandate that required emissions reductions must be achieved within the state, while allowing some flexibility beyond the emissions budget to accommodate year-to-year operational variability. In the event a state's assurance level is exceeded, responsibility for the exceedance is apportioned among the state's sources through a procedure that accounts for the sources' shares of the state's total emissions for the control period as well as the sources' shares of the state's assurance level for the control period.
• At the program's compliance deadlines after each control period, sources are required to hold for surrender specified quantities of allowances. The minimum quantities of allowances that must be surrendered are based on the sources' reported emissions for the control period at a 1-for-1 ratio of allowances to tons of emissions (or 2-for-1 in instances of late compliance). In addition, two more allowances must be surrendered for each ton of emissions exceeding a state's assurance level for a control period, yielding an overall 3-for-1 surrender ratio for those emissions (or 4-for-1 in instances of late compliance). Failure to timely surrender all required allowances is potentially subject to penalties under the CAA's enforcement provisions.

• To continuously incentivize sources to reduce their emissions even when they already hold sufficient allowances to cover their expected emissions for a control period, and to promote compliance cost minimization, operational flexibility, and allowance market liquidity, the programs allow trading of allowances—both among sources in the program and with non-source entities—and also let allowances that are unused in one control period be carried over for use in future control periods as banked allowances. Although the CSAPR programs do not limit trading of allowances, and prior to this rule have not limited banking of allowances within a given trading program, the 3-for-1 surrender ratio imposed by the assurance provisions on any emissions exceeding a state's assurance level disincentivizes sources from relying on either in-state banked allowances or net out-of-state purchased allowances to emit over the assurance level.

• Finally, other common design elements ensure program integrity, source accountability, and administrative transparency. Most notably, each unit must monitor and report emissions and operational data in accordance with the provisions of 40 CFR part 75; all allowance allocations or auction results, transfers, and deductions must be properly recorded in the EPA's Allowance Management System; each source must have a designated representative who is authorized to represent all of the source's owners and operators and is responsible for certifying the accuracy of the source's reports to the EPA and overseeing the source's Allowance Management System account; and comprehensive data on emissions and allowances are made publicly available.

The EPA continues to believe that the historical CSAPR trading program structure established by the common design elements just described has important positive attributes, particularly with respect to the exceptional degree of compliance flexibility it can provide to a sector such as the electric power sector where such flexibility is especially useful and valuable. However, the EPA also shares many stakeholders' concerns about whether the historical structure, without enhancements, is capable of adequately addressing states' good neighbor obligations with respect to the 2015 ozone NAAQS in light of the rapidly evolving EGU fleet and the protectiveness and short-term form of the ozone standard. One set of concerns relates to the historically observed tendency under the trading programs for the supply of allowances to grow over time while the demand for allowances falls, reducing allowance prices and eroding the consequent incentives for sources to effectively control their emissions. A second, overlapping set of concerns relates to the general absence of source- or unit-specific emissions reduction requirements, allowing some individual sources to idle or run less optimally existing emissions controls even when a linkage between the sources' state and a receptor persists. For example, certain units in Ohio and Pennsylvania have been found to have operated their controls below target emissions performance levels used for budget setting under the CSAPR Update in the 2019–2021 period, even though the Revised CSAPR Update found that these states remained linked through at least 2021 to receptors for the 2008 ozone NAAQS, and the CSAPR Update itself was only a partial remedy. See86 FR 23071, 23083. While this unit-level behavior may have been permissible under the prior program, emissions from these individual sources can contribute to increased pollution concentrations downwind on the particular days that matter for downwind exceedances of the relevant air quality standard. This indicates that the prior program design was not effectively ensuring the elimination of significant contribution.

The EPA has analyzed hourly emissions data reported in prior cap-and-trade programs and identified instances of sources that did not operate SCR controls for substantial portions of recent ozone seasons. In an effort to ensure emissions control on critically important highest ozone days, guard against non-operation of emissions controls under a more protective NAAQS, and provide assurance of elimination of significant contribution to downwind areas, while also maintaining appropriate compliance and operational flexibility for EGUs, the EPA in this rule is implementing a suite of enhancements to the trading program. These will help to ensure reductions occur on the highest ozone days commensurate with our Step 3 determinations, in addition to maintaining a mass-based seasonal requirement. To meet the statutory mandate to eliminate significant contribution and interference with maintenance on the critically important days, this combination of provisions will strongly incentivize sources to plan to run controls all season, including on the highest ozone days, while giving reasonable flexibility for occasional operational needs.

In this rulemaking, the EPA is revising the Group 3 trading program to include enhancements designed to address both sets of concerns described previously. The principles guiding the various revisions and the relationships of the revisions to one another are discussed in sections VI.B.1.b and VI.B.1.c of this document. The individual revisions are discussed in more detail in sections VI.B.4 through VI.B.9 of this document.

b. Enhancements To Maintain Selected Control Stringency Over Time

The first set of concerns noted about the current CSAPR trading program structure relates to the programs' ability to maintain the rule's selected control stringency and related EGU effective emissions performance level as the EGU fleet evolves over time. Under the historical structure of the CSAPR trading programs, the effectiveness of the programs at maintaining the rule's selected control stringency depends entirely on how allowance prices over time compare to the costs of sources' various emissions reduction opportunities, which in turn depends on the relationship between the supply for allowances and the demand for allowances. In considering possible ways to address concerns about the ability to enhance the historical trading program structure to better sustain incentives to control emissions over time, the EPA has focused on the trading program design elements that determine the supply of allowances, specifically the approach for setting state emissions budgets and the rules concerning the carryover of unused allowances for use in future control periods as banked allowances.

i. Revised Emissions Budget-Setting Process

In each of the previous rulemakings establishing CSAPR trading programs, the EPA has evaluated the emissions that could be eliminated through implementation of certain types of emissions control strategies available at various cost thresholds to achieve certain rates of emissions per unit of heat input ( i.e., the amount of fuel consumed) and the effects of the resulting emissions reductions on downwind air quality. After determining the emissions control strategies and associated emissions reductions that should be required under the good neighbor provision by considering these factors in a multifactor test at Step 3, the EPA has then for purposes of Step 4 implementation program design projected the amounts of emissions that would remain after the assumed implementation of the selected emissions control strategies at various points in the future and has established the projected remaining amounts of emissions as the state emissions budgets in trading programs.

Projecting the amounts of emissions remaining after implementation of selected emissions controls necessarily requires projections not only for sources' future emissions rates but also for other factors that influence total emissions, notably the composition of the future EGU fleet ( i.e., the capacity amounts of different types of sources with different emissions rates) and their future utilization levels ( i.e., their heat input). To the extent conditions unfold in practice that differ from the projections made at the time of a rulemaking for these other factors, over time the emissions budgets may not reflect the intended stringency of the emissions control strategies identified in the rulemaking as consistent with addressing states' good neighbor obligations. Further, projecting EGU fleet composition and utilization beyond the relatively near-term analytic years of 2023 and 2026 given particular attention in this rulemaking has become increasingly challenging in light of the anticipated continued evolution of the electric power sector toward more efficient and cleaner sources of generation, including as driven by incentives provided by the Infrastructure Investment and Jobs Act as well as the Inflation Reduction Act.

A consequence of using a trading program approach with preset emissions budgets that do not keep pace with the trends in EGU fleet composition and heat input is that the preset emissions budgets maintain the supply of allowances at levels that increasingly exceed the emissions that would occur even without implementation of the emissions control strategies used as the basis for determining the emissions budgets, causing decreases in allowance prices and hence the incentives to implement the control strategies. As an example, although the emissions budgets in the CSAPR Update established in 2016 reflected implementation of the emissions control strategy of operating and optimizing existing SCR controls, within four years the EPA found that EGU retirements and changes in utilization not anticipated in EPA's previous budget-setting computations had made it economically attractive for at least some sources to idle or reduce the effectiveness of their existing controls (relying on purchased allowances instead). While the EPA has provided analysis indicating that, on average, sources operate their controls more effectively on high electric demand days, it has also identified cases where units fail to optimize their controls on these days. Downwind states have suggested this type of reduced pollution control performance has occurred on the day and preceding day of an ozone exceedance. While the EPA had previously provided analysis focusing on the year of initial program implementation, when allowance prices were high ( i.e., 2017 for the CSAPR Update), to demonstrate that on average, sources operate their controls more effectively on high electric demand days, even in that case it had identified situations where particular units failed to optimize their controls on these days. In later years, when allowance prices had fallen, more sources, including some identified by commenters, had idled or reduced the effectiveness of their controls. Such an outcome undermined the ongoing achievement of emissions rate performance consistent with the control strategies identified in the CSAPR Update to eliminate significant contribution to nonattainment and interference with maintenance, despite the fact that the mass-based budgets were being met.

In the Revised CSAPR Update, the EPA took steps to better address the rapid evolution of the EGU fleet, specifically by setting updated emissions budgets for individual future years though 2024 that reflect future EGU fleet changes known with reasonable certainty at the time of the rulemaking. Some commenters in that rulemaking requested that the EPA also update the year-by-year emissions budgets to reflect future fleet changes that might become known after the time of the rulemaking, but the EPA declined to do so, in part because no methodology for making future emissions budget adjustments in response to post-rulemaking data had been included in the proposal for the rulemaking.

Based on information available as of December 2022, it appears that the emissions budgets set for the first two control periods covered by the Revised CSAPR Update generally succeeded at creating incentives to operate emissions controls under the Group 3 trading program for those control periods. However, the EPA recognizes that the lack of emissions budget adjustments after 2024 in conjunction with industry trends toward more efficient and cleaner resources will likely lead to a surplus of allowances after the adjustments end. This prospect for the existing Group 3 trading program should be avoided by the changes being made in this rulemaking. In this rulemaking, besides establishing new preset emissions budgets for the 2023 through 2029 control periods, the EPA is also extending the Group 3 trading program budget-setting methodology used in the Revised CSAPR Update to routinely calculate dynamic emissions budgets for each future control period from 2026 on, to be published in the year before that control period, with each dynamic emissions budget generally reflecting the latest available information on the composition and utilization of the EGU fleet at the time that dynamic emissions budget is determined. For the control periods in 2026 through 2029, each state's final emissions budget will be the preset budget determined for the state in this rulemaking except in instances when the dynamic budget determined for the state (and published approximately one year before the control period using the dynamic budget-setting methodology) is higher. For control periods in 2030 and thereafter, the emissions budgets will be the amounts determined for each state in the year before the control period using the dynamic budget-setting methodology.

The current budget-setting methodology established in the Revised CSAPR Update and the revisions being made to that methodology are discussed in detail in section VI.B.4 of this document and the Ozone Transport Policy Analysis Final Rule TSD. To summarize here, the methodology used to determine the preset budgets largely follows the Revised CSAPR Update's emissions budget-setting methodology, which included three primary steps: (1) establishment of a baseline inventory of EGUs adjusted for known retirements and new units, with heat input and emissions rate data for each EGU in the inventory based on recent historical data; (2) adjustment of the baseline data to reflect assumed emissions rate changes resulting from known new controls, known gas conversions, and implementation of the emissions control strategies used to determine states' good neighbor obligations; and (3) application of an increment or decrement to reflect the effect on emissions from projected generation shifting among the units in a state at the emissions reduction cost associated with the selected emissions control strategies. In this rulemaking, the EPA has determined the preset state emissions budgets for the control periods from 2023 through 2029 by using the Revised CSAPR Update's budget-setting methodology, except that the step of that methodology intended to reflect the effects of generation shifting has been eliminated.

The dynamic budget-setting methodology used to determine dynamic state emissions budgets in the year before each control period starting with the 2026 control period is set forth in the revised Group 3 trading program regulations at 40 CFR 97.1010(a). This methodology modifies the Revised CSAPR Update's budget-setting methodology in two ways. First, the baseline EGU inventory and heat input data, but not the emissions rate data, will be updated for each control period using the most recent available reported data in combination with reported data from the four immediately preceding years. For example, in early 2025, using the final data reported for 2020 through 2024, the EPA will update the baseline inventory and heat input data used to determine dynamic state emissions budgets for the 2026 control period. Second, the EPA will not apply an increment or decrement to any state emissions budget for projected generation shifting associated with implementation of the selected control strategies, because any such shifting should already be reflected in the reported heat input data used to update the baseline.

The EPA believes that the revisions to the emissions budget-setting process will substantially improve the ability of the emissions budgets to keep pace with changes in the composition and utilization of the EGU fleet. The dynamic budget-setting methodology will account for the electric power sector's overall trends toward more efficient and cleaner resources, both of which tend to decrease total heat input at affected EGUs, and through 2029 the preset budgets established in the rule will also account for these factors to the extent known. The dynamic budget-setting methodology will also account for other factors that could lead to increased heat input in some states, such as generation shifting from other states or increases in electricity demand caused by rising electrification. The dynamic budget-setting procedure is specified in this final rule's trading program regulations and the computations, which are straightforward, can be performed in a spreadsheet to deliver reliable results. The EPA will provide public notice of the preliminary calculations and the data used by March 1 of the year preceding the control period and will provide an opportunity for submission of any objections to the data and preliminary calculations before finalizing the dynamic budgets for each control period by May 1 of the year before the control period to which those dynamic budgets apply. Thus, for example, sources and other stakeholders will have certainty by May 1, 2025, of the dynamic emissions budgets that will be calculated for the 2026 control period that starts May 1, 2026. Moreover, as of the issuance of this final rule, stakeholders will know the state-level preset emissions budgets for the 2026–2029 control periods, which serve as floors that will only be supplanted by dynamic budgets calculated for those control periods if such a dynamic budget yields a higher amount of tons than the corresponding preset budget established in this action.

It bears emphasis that the annually updated information used in the dynamic budget-setting computations will concern only the composition and utilization of the EGU fleet and not the emissions rate data also used in those computations. The dynamically determined emissions budget computations for all years will reflect only the specific emissions control strategies used to determine states' good neighbor obligations as determined in this rulemaking, along with fixed historical emissions rates for units that are not assumed to implement additional control strategies, thereby ensuring that the annual updates will eliminate emissions as determined to be required under the good neighbor provision. The stringency of the emissions budgets will simply reflect the stringency of the emissions control strategies determined in the Step 3 multifactor analysis and will do so more consistently over time than the EPA's previous approach of computing emissions budgets for all future control periods at the time of the rulemaking.

The rule's revisions relating to state emissions budgets and the budget-setting process generally follow the proposal except for two changes we are making in response to comments, specifically: we will use historical data from multiple years rather than a single year in the dynamic budget-setting process, and we are establishing preset emissions budgets for the 2026–2029 control periods such that the dynamic budgets for those control periods will only be imposed where they exceed the corresponding preset budgets finalized in this rule. The rationale for these changes is discussed later in this section as part of the responses to the relevant comments. Details of the final budget-setting methodology and responses to additional comments are discussed further in section VI.B.4 of this document.

The final rule's provisions relating to the determination of state-level variability limits and assurance levels and unit-level allowance allocations are coordinated with the budget-setting methodology. These provisions generally follow the proposal except that the change to the methodology for determining variability limits is implemented starting with the 2023 control period instead of the 2025 control period and the final methodology for determining unit-level allocations of allowances to coal-fired units considers the controlled emissions rate assumptions applicable to the same units in the budget-setting process. Details of these provisions, including the rationales for the changes from proposal, are discussed in sections VI.B.5 and VI.B.9, respectively.

ii. Allowance Bank Recalibration

Besides the levels of the emissions budgets, the second design element of the trading program structure that affects the supply of allowances in each control period, and that consequently also affects the ability of a trading program to maintain the rule's selected control stringency as the EGU fleet evolves over time, is the set of rules concerning the carryover of unused allowances for use in future control periods as banked allowances. As noted previously, trading and banking of allowances in the CSAPR trading programs can serve a variety of purposes: continuously incentivizing sources to reduce their emissions even when they already hold sufficient allowances to cover their expected emissions for a control period, facilitating compliance cost minimization, accommodating necessary operational flexibility, and promoting allowance market liquidity. All of these purposes are advanced by rules that allow sources to trade allowances freely (both with other sources and with non-source entities such as brokers). All of these purposes are also advanced by rules that allow unused allowances to be carried over for possible use in future control periods, thereby preserving a value for the unused allowances. However, while the EPA considers it generally advantageous to place as few restrictions on the trading of allowances as possible, unrestricted banking of allowances has a potentially significant disadvantage offsetting its advantages, namely that it allows what might otherwise be temporary surpluses of allowances in some individual control periods to accumulate into a long-term allowance surplus that reduces allowance prices and weakens the trading program's incentives to control emissions. With weakened incentives, some operators would be more likely to choose not to continuously operate and optimize their emissions controls, imperiling the ongoing achievement of emissions rate performance consistent with the control strategies defined as eliminating significant contribution to nonattainment and interference with maintenance.

As discussed in detail in section VI.B.6 of this rule, the EPA is revising the Group 3 trading program by adding provisions that establish a routine recalibration process for banked allowances that will be carried out in August 2024 and each subsequent August, after the compliance deadline for the control period in the previous year. In each recalibration, the EPA will reset the total quantity of banked allowances for the Group 3 trading program (“Group 3 allowances”) held in all Allowance Management System accounts to a level computed as a target percentage of the sum of the state emissions budgets for the current control period. The target percentage will be 21 percent for the 2024–2029 control periods and 10.5 percent for control periods in 2030 and later years. The recalibration procedure entails identifying the ratio of the target bank amount to the total quantity of banked allowances held in all accounts before the recalibration and then, if the ratio is less than 1.0, multiplying the quantity of banked allowances held in each account by the ratio to identify the appropriate recalibrated amount for the account (rounded to the nearest allowance), and deducting any allowances in the account exceeding the recalibrated amount.

As noted previously, recalibration of the bank for each control period will be carried out in August of that control period. This timing will accommodate the process of deducting allowances for compliance for the previous control period, which cannot be completed before sources' June 1 compliance deadline for the previous control period, and will then provide approximately two additional months for sources to engage in any desired allowance transactions before recalibration occurs. However, data that can be used to estimate the bank recalibration ratio for each control period will be available shortly after the end of the previous control period, and the EPA will use these data to make information on the estimated bank recalibration ratio for each control period publicly available no later than March 1 of the year of that control period, thereby facilitating the ability of affected EGUs to anticipate their ultimate holdings of recalibrated banked allowances to inform their compliance planning for that control season. Affected EGUs will also have several months following the completed bank recalibration in August to transact allowances with other parties as needed before the allowance transfer deadline of June 1 of the following year.

The EPA believes this revision to the Group 3 trading program's banking provisions establishing an annual bank recalibration process will complement the revisions to the budget-setting process by preventing any surplus of allowances created in one control period from diminishing the intended stringency and resulting emissions reductions of the emissions budgets for subsequent control periods.

The calibration procedure will not erase the value of unused allowances for the holder, because the larger the quantity of banked allowances that is held in a given account before each recalibration, the larger the quantity of banked allowances that will be left in the account after the recalibration for possible sale or use in meeting future compliance requirements. Because the banked allowances will always have value, the opportunity to bank allowances will continue to advance the purposes served by otherwise unrestricted banking as described previously. Opportunities to bank unused allowances can serve all these same purposes whether a banked allowance is of partial value (if the bank needs recalibrating to its target level) or is of full value compared to a newly issued allowance for the next control period.

The final rule's provisions relating to bank recalibration generally follow the proposal except that, in response to comments, the target percentage used to determine the recalibrated bank levels for the 2024–2029 control periods is being set at 21 percent instead of 10.5 percent. The rationale for this change is discussed later in this section as part of the responses to the relevant comments. Details of the bank recalibration provisions are discussed further in section VI.B.6 of this rule.

c. Enhancements To Improve Emissions Performance at Individual Units

The second set of concerns about the structure of the current CSAPR trading programs relates to the general absence of source- or unit-specific emissions reduction requirements. Without such requirements, the programs affect individual sources' emissions performance only to the extent that the incentives created by allowance prices are high enough relative to the costs of the sources' various emissions control opportunities. In circumstances where the incentives to control emissions are insufficient, some individual sources even idle existing emissions controls. Emissions from these individual sources can contribute to increased pollution concentrations downwind on the particular days that matter for downwind exceedances of the relevant air quality standard.

This EPA intends that the trading program enhancements described in section VI.B.1.b of this rule will improve the Group 3 trading program's ability to sustain emissions control incentives over time such that needed emissions performance will be achieved by all participating units without the need for additional requirements to be imposed at the level of individual units. However, because obtaining needed emissions performance at individual units is also important to the elimination of significant contribution in keeping with the EPA's Step 3 determinations, the EPA is supplementing the previously discussed enhancements with two other new sets of provisions that will apply to certain individual units within the larger context of the Group 3 trading program. The allowance price will continue to be the most important driver of good environmental performance for most units, but the proposed unit-level requirements will be important supplemental drivers of performance and will offer additional assurance that significant contribution is eliminated on a daily basis during the ozone season by more continuous operation of existing pollution controls.

i. Unit-Specific Backstop Daily Emissions Rates

The first of the trading program enhancements intended to improve emissions performance at the level of individual units is the addition of backstop daily NO_X emissions rate provisions that will apply to large coal-fired EGUs, defined for this purpose as units serving electricity generators with nameplate capacities equal to or greater than 100 MW and combusting any coal during the control period in question. Starting with the 2024 control period, a 3-for-1 allowance surrender ratio (instead of the usual 1-for-1 surrender ratio) will apply to emissions during the ozone season from any large coal-fired EGU with existing SCR controls exceeding by more than 50 tons a daily average NO_X emissions rate of 0.14 lb/mmBtu. The additional allowance surrender requirement will be integrated into the trading program as a new component in the calculation of each unit's primary emissions limitation, such that the additional allowances will have to be surrendered by the same compliance deadline of June 1 after each control period. The amount of additional allowances to be surrendered will be determined by computing, for each day of the control period, any excess of the unit's reported emissions (in pounds) over the emissions that would have resulted from combusting that day's actual heat input at an average daily emissions rate of 0.14 lb/mmBtu, summing the daily amounts, converting from pounds to tons, computing the amount of any excess over 50 tons, and multiplying by two. Starting with the second control period in which newly installed SCR controls are operational, but not later than the 2030 control period, the 3-for-1 surrender ratio will apply in the same way to all large coal-fired EGUs except circulating fluidized bed units, consistent with EPA's determination that a control stringency reflecting installation and operation of SCR controls on all such large coal-fired EGUs is appropriate to address states' good neighbor obligations with respect to the 2015 ozone NAAQS.

In prior rules addressing interstate transport of air pollution, stakeholders have noted that while seasonal cap-and-trade programs are effective at lowering ozone and ozone-forming precursors across the ozone season, attainment of the standard is measured on key days and therefore it is necessary to ensure that the rule requires emissions reductions not just seasonally, but also on those key days. They have noted that while the trading programs established under the NO_X SIP Call, CAIR, and CSAPR have all been successful in ensuring seasonal reductions, states must remain below daily peak levels, not just seasonal levels, to reach attainment. These downwind stakeholder communities have suggested that operating pollution controls on the highest ozone days (and immediately preceding days) during the ozone season is of critical importance. The EPA has analyzed hourly emissions data reported in prior cap-and-trade programs and has identified instances of sources that did not operate SCR controls for substantial portions of recent ozone seasons. These instances are discussed in section V.B.1.a of this document and in the EGU NO_X Mitigation Strategies Final Rule TSD in the docket. While the EPA has in prior ozone transport actions not found sufficient evidence of emissions control idling or non-optimization to take the step of building in enhancements to the trading program to ensure unit-level control operation, our review of subsequent-year data for prior programs suggests that the non-optimization behavior increases in the latter years of a program. Applied to this context ( e.g., a rule providing a full remedy to interstate transport for the more protective 2015 ozone NAAQS and an extended period of expected persistence of receptors), this data suggests this deterioration in performance could become prevalent and problematic in future years if not addressed. Rather than allow for the potential of continued deterioration in the environmental performance of our trading programs, the EPA finds the evidence of declining SCR performance in later years of trading programs sufficient to justify prophylactic measures in this rule to ensure the emissions control strategy selected at Step 3 is indeed implemented at Step 4. Thus, particularly in the context of the more protective 2015 ozone NAAQS combined with the full remedy nature of this action and the extended timeframe for which upwind contribution to downwind nonattainment is projected to persist, the EPA agrees with these stakeholders that the set of measures promulgated in this rulemaking to implement the control stringency levels found necessary to address states' good neighbor obligations should include measures designed to more effectively ensure that individual units operate their emissions controls routinely throughout the ozone season, thereby also ensuring that the controls are planned to be in operation on the particular days that turn out to be most critical for ozone formation and for attainment of the NAAQS. Routine operation of emissions controls will also provide relief to overburdened communities downwind of any units that might otherwise have chosen not to operate their controls. In the Ozone Transport Policy Analysis Final Rule TSD, the EPA conducted a screening analysis that found nearly all of the EGUs included in this analysis are located within a 24-hour transport distance of many areas with potential EJ concerns. Thus, the EPA is adopting backstop daily rate limits at the individual unit level because it is appropriate and justified in the context of eliminating significant contribution under CAA section 110(a)(2)(D)(i)(I). While the former justification is sufficient to finalize this enhancement to the trading program, we also anticipate that this measure will deliver public health and environmental benefits to overburdened communities (as well as the rest of the population).

We considered whether, as some commenters suggested, it would be appropriate to simply implement unit-specific daily emissions limitation at all of the large, coal-fired EGUs, and forego an emissions trading approach altogether. While this is within the EPA's statutory authority, see CAA section 110(a)(2)(A) and 302(y), and merits careful consideration, we are declining to do so in this action but intend to closely monitor EGU emissions performance in response to the trading program finalized here. The purpose of establishing a backstop daily NO_X emissions rate and implementing it through additional allowance surrender requirements instead of as an enforceable emissions limitation is to incentivize improved emissions performance at the individual unit level while continuing to preserve, to the extent possible, the advantages that the flexibility of a trading program brings to the electric power sector. As discussed in section VI.B.7 of this document, under the EPA's historical trading programs without the enhancements made in this rulemaking, some individual coal-fired units with SCR controls have chosen to operate the controls at lower removal efficiencies than in past ozone seasons or even to idle the controls for entire ozone seasons. In addition, some SCR-equipped units have chosen to routinely cycle their emissions controls off at lower load levels, such as while operating overnight, instead of operating the controls, upgrading the units to enable the controls to be operated under those conditions, or not operating the units under those conditions. Collectively, this non-optimization of existing controls has a detrimental impact on problematic receptors. Table V.D.1–1 shows the expected air quality benefit from control optimization (totaling nearly 1.6 ppb change across all receptors).

The EPA has identified sources of interstate ozone pollution such as the New Madrid and Conemaugh plants (in Missouri and Pennsylvania, respectively) whose SCR controls were not operating for substantial portions of recent ozone seasons. The data included in Appendix G of the Ozone Transport Policy Analysis Final Rule TSD, available in the docket for this rulemaking, demonstrate that these units have operated their SCRs better and more consistently during years with higher NO_X allowance prices. Downwind stakeholders have noted that some of the higher emissions rates (specifically in the case of Conemaugh Unit 2 in 2019) have occurred on the day of and the preceding day of an ozone exceedance in bordering states.

The EPA believes that the design of the daily emissions rate provisions will be effective in addressing these types of high-emitting behavior by significantly raising the cost of planned operator decisions that substantially compromise environmental performance. At the same time, the provision will not unduly penalize an occasional unplanned exceedance, because the amount of additional allowances that would have to be surrendered to address a single day's exceedance would be much smaller than the amount that would have to be surrendered to address planned poor performance sustained over longer time periods. Moreover, the EPA believes that the inclusion of a 50-ton threshold before the increased surrender requirements would apply is sufficient to address virtually all instances where a unit's emissions would exceed the 0.14 lb/mmBtu daily rate because of unavoidable startup or shutdown conditions during which SCR equipment cannot be operated, thereby ensuring that the provision will not penalize units for emissions that are beyond their reasonable control.

The EPA is applying the daily emissions rate provisions to large coal-fired EGUs, and not to other types of units, for reasons that are consistent with EPA's determinations regarding the appropriate control stringency for EGUs to address states' good neighbor obligations with respect to the 2015 ozone NAAQS. Installation and operation of SCR controls is well-established as a common practice for the best control of NO_X emissions from coal-fired EGUs, as evidenced by the fact that the technology is already installed on more than 60 percent of the sector's total coal-fired capacity and installed on nearly 100 percent of the coal fired boilers in the top quartile of emissions rate performance. In the context of addressing good neighbor obligations with respect to the 2015 ozone NAAQS, the EPA is determining that a control stringency reflecting universal installation and operation of SCR technology at large coal-fired EGUs (other than circulating fluidized bed units) is appropriate at Step 3. Finally, where SCR controls are installed on such units, optimized operation of those controls is an extremely cost-effective method of achieving NO_X emissions reductions. The EPA believes these considerations support establishment of the daily emissions rate provisions on a universal basis for large coal-fired EGUs, with near-term application of the provisions for units that already have the controls installed and deferred application for other units, as discussed later.

With regard to gas-fired steam EGUs, SCR controls are nowhere near as prevalent, and while the EPA is including some SCR controls at gas-fired steam units in the selected control stringency at Step 3, the EPA is not including universal SCR controls at gas-fired steam units. Because the EPA is not determining that universal installation and operation of SCR controls at gas-fired steam EGUs is part of the selected control stringency, in order not to constrain the power sector's flexibility to choose which particular gas-fired steam EGUs are the preferred candidates for achieving the required emissions reductions, the EPA is not applying the daily emissions rate provisions to large gas-fired steam EGUs. Focusing the backstop daily emissions rates on coal-fired units is also consistent with stakeholder input which has emphasized the need for short-term rate limits at coal units given their relatively higher emissions rates.

The EPA developed the level of the daily average NO_X emissions rate—0.14 lb/mmBtu—through analysis of historical data, as described in section VI.B.7 of this document. A rate of 0.14 lb/mmBtu represents the daily average NO_X emissions rate that has been demonstrated to be achievable on approximately 95 percent of days covering more than 99 percent of total ozone-season NO_X emissions by coal-fired units with SCR controls that are achieving a seasonal NO_X average emissions rate of 0.08 lb/mmBtu (or less), which is the seasonal NO_X emissions rate that the EPA has determined is indicative of optimized SCR performance by units with existing SCR controls.

As noted previously, the daily average emissions rate provisions will apply beginning in the 2024 control period for large coal-fired units with installed SCR controls, one control period later than optimization of those controls will be reflected in the state emissions budgets under this rule. For these units, not applying the daily average rate provisions until 2024 serves three purposes. First, it provides all the units with a preparatory interval to focus attention on improving not only the average performance of their SCR controls but also the day-to-day consistency of performance before they will be held to increased allowance-surrender consequences for exceeding the daily rate. Second, it provides the subset of units that exhaust to common stacks with other units that currently lack SCR controls an opportunity to exercise the option to install and certify any additional monitoring systems needed to monitor the individual units' NO_X emissions rates separately; otherwise, the daily emissions rate provisions will apply to the SCR-equipped units based on the combined NO_X emissions rates measured in the common stacks. Third, it provides all units sufficient time to update the data handling software in their existing monitoring systems as needed to compute and report the additional hourly and daily data values needed for implementation of the provisions.

With respect to the units without existing SCR controls, the daily average emissions rate provisions will apply starting with the second control period in which newly installed SCR controls are operational at the unit, but not later than the 2030 control period. This implementation timing represents a change from the proposal, under which the daily average emissions rate provisions would have applied to units without existing SCR starting in the 2027 control period. Commenters noted that for many units without SCR, replacement of the unit within a few years, and shifting of some generation to cleaner units in the interim, would be a more economic compliance strategy than installation of new SCR controls. The commenters further noted that implementation of the daily average emissions rate for these units starting in 2027 would strongly disadvantage such an alternative strategy if the capacity replacement and any associated transmission improvements could not be implemented by 2027. In light of these comments, the EPA has determined that as long as the emissions budgets determined in this rule to eliminate significant contribution are still being implemented as expeditiously as practicable—which in this instance the EPA has determined requires phasing in the required emissions reductions by 2027—it is reasonable to defer implementation of the daily average emissions rate provisions to 2030 for units without SCR to allow temporarily greater flexibility to pursue compliance strategies other than installation of new controls. This lag is permissible consistent with the obligation to eliminate significant contribution for reasons that are further discussed in response to comments in section VI.B.1.d of this document. However, for any units that choose a compliance strategy of installing new SCR controls before 2030, the daily average emissions rate provisions would apply in the second control period of operation. Specification of the second control period rather than the first control period provides the unit operators with an opportunity to gain operational experience with the new equipment before the units will be held to increased allowance-surrender consequences for exceeding the daily rate.

The unit-specific daily emissions rate provisions are being finalized as proposed except for two changes noted in the previous summary: the exclusion from extra allowance surrender requirements of a unit's first 50 tons of emissions in a control period exceeding the backstop daily rate, and the revision of the starting date for implementation of the requirement for units without existing SCR controls to 2030 or the second control period of SCR operation, if earlier. The rationale for these changes is further discussed in the responses to comments later in this section. Additional details of the unit-specific daily emissions rate provisions are discussed in section VI.B.7 of this document.

ii. Unit-Specific Emissions Limitations Contingent on Assurance Level Exceedances

The second of the trading program enhancements intended to improve emissions performance at the level of individual units is the addition of unit-specific secondary emissions limitations for units with post-combustion controls starting with the 2024 control period. The secondary emissions limitations will be determined on a unit-specific basis according to each unit's individual performance but will apply to a given unit only under the circumstance where a state's assurance level for a control period has been exceeded, the unit is included in a group of units to which responsibility for the exceedance has been apportioned under the program's assurance provisions, and the unit operated during at least 10 percent of the hours in the control period. Where these conditions for application of a secondary emissions limitation to a given unit for a given control period are met, the unit's secondary emissions limitation consists of a prohibition on NO_X emissions during the control period that exceed by more than 50 tons the NO_X emissions that would have resulted if the unit had achieved an average emissions rate for the control period equal to the higher of 0.10 lb/mmBtu or 125 percent of the unit's lowest average emissions rate for any previous control period under any CSAPR seasonal NO_X trading program during which the unit operated for at least 10 percent of the hours.

The secondary emissions limitation is in addition to, not in lieu of, the primary emissions limitation applicable to each source, which continues to take the form of a requirement to surrender a quantity of allowances based on the source's emissions, and also in addition to the existing assurance provisions, which similarly continue to take the form of a requirement for the owners and operators of some sources to surrender additional allowances when a state's assurance level is exceeded. In contrast to these other requirements, the unit-specific secondary emissions limitation takes the form of a prohibition on emissions over a specified level, such that any emissions by a unit exceeding its secondary emissions limitation would be subject to potential administrative or judicial action and subject to penalties and other forms of relief under the CAA's enforcement authorities. The reason for establishing this form of limitation is that experience under the existing CSAPR trading programs has shown that, in some circumstances, the existing assurance provisions have been insufficient to prevent exceedances of a state's assurance level for a control period even when the likelihood of an exceedance has been foreseeable and the exceedance could have been readily avoided if certain units had operated with emissions rates closer to the lower emissions rates achieved in past control periods. The assurance levels exist to ensure that emissions from each state that contribute significantly to nonattainment or interfere with maintenance of a NAAQS in another state are prohibited. North Carolina v. EPA, 531 F.3d 896, 906–08 (D.C. Cir. 2008). The EPA's programs to eliminate significant contribution must therefore achieve this prohibition, and the evidence of foreseeable and avoidable exceedances of the assurance levels demonstrates that EPA's existing approach has not been sufficient to accomplish this.

The purpose of including assurance levels higher than the state emissions budgets in the CSAPR trading programs is to provide flexibility to accommodate operational variability attributable to factors that are largely outside of an individual owner's or operator's control, not to allow owners and operators to plan to emit at emissions rates that could be anticipated to cause a state's total emissions to exceed the state's emissions budget or assurance level. Conduct leading to a foreseeable, readily avoidable exceedance of a state's assurance level cannot be reconciled with the statutory mandate of the CAA's good neighbor provision that emissions “within the state” significantly contributing to nonattainment or interfering with maintenance of a NAAQS in another state must be prohibited. Because the current CSAPR regulations do not expressly prohibit such conduct and have proven insufficient to deter it in some circumstances, the EPA is correcting the regulatory deficiency in the Group 3 trading program by adding secondary emissions limitations that cannot be complied with through the use of allowances.

The EPA notes that although the purpose of the secondary emissions limitations is to strengthen the assurance provisions, which apply on a statewide, seasonal basis, the unit-specific structure of the new limitations will strengthen the incentives for individual units with post-combustion controls to maintain their emissions performance at levels consistent with their previously demonstrated capabilities. The new limitations will strengthen the incentives to operate and optimize the controls continuously, which can be expected to reduce some individual units' emissions rates throughout the ozone season, including on the days that turn out to be most critical for downwind ozone levels. Better emissions performance on average across the ozone season by individual units likely will also help address impacts of pollution on overburdened communities downwind from some such units. See Ozone Transport Policy Analysis Final Rule TSD, Section E.

The unit-specific secondary emissions limitations are being finalized as proposed except that the limitations will apply only to units with post-combustion controls. The rationale for this change, and additional details regarding the provisions, are discussed in section VI.B.8 of this document.

d. Responses to General Comments on the Revisions to the Group 3 Trading Program

This section summarizes and provides the EPA's responses to overarching comments received on the EPA's proposal to implement the emissions reductions required from EGUs under this rule through expansion and enhancement of the Group 3 trading program originally established in the Revised CSAPR Update, particularly comments on electric system reliability. Responses to comments about individual aspects of the enhanced trading program are addressed in the respective subsections of this section in which those aspects are discussed. Responses to comments concerning alleged overcontrol and the EPA's legal authority are in sections V.D. and III. Comments not addressed in this document are addressed in the separate RTC document available in the docket for this action.

Comment: Some commenters, including EGU owners, states, and several RTOs, expressed concern that the requirements for EGUs as formulated in the proposal could lead to a degradation in the reliability of the electric system. As background, some of these commenters noted that the power sector is currently undergoing rapid change, with older and less economic fossil-fuel-fired steam generating units retiring while the majority of the new capacity being added consists of wind and solar capacity. They noted that fossil-fuel-fired generating capacity provides reliability benefits not necessarily provided by other types of generating capacity, including not only the ability to generate electricity in the absence of wind or sunlight, but also inertia, ramping capability, voltage support, and frequency response. Commenters stated that past EGU retirements and the pace of change in the generating capacity mix have already been stressing the electric system in some regions, and that the forecasted risk of events where the electric system would be unable to fully meet load is rising.

For purposes of their comments, these commenters generally assumed that the rule would lead to additional retirements of fossil-fuel-fired generating capacity beyond the retirements that EGU owners have already planned and announced. Some of the commenters also suggested that remaining fossil-fuel-fired generators would be unwilling to operate when needed because allowances might be unavailable for purchase or too costly. In the context of an already-stressed electric system, the commenters predicted that these assumed consequences of the rule would threaten resource adequacy and result in degraded electric reliability. To support their assumptions concerning additional retirements, some of the commenters pointed to projections of incremental generating capacity retirements included in the results of modeling performed by the EPA to analyze the costs and benefits of the proposed rule. Some commenters indicated that they expected EGU owners to be interested in retiring and replacing uncontrolled units as of the date of implementation of the backstop daily rate requirement on uncontrolled units, and expressed concern that the proposal to implement that requirement as of the 2027 control period did not allow sufficient time for planning and implementation of all the necessary generation and transmission investments to make this a viable compliance strategy; for these commenters, 2027 and the immediately following years were the period of greatest concern. Some commenters appear simply to have assumed that owners of units not already equipped with SCR controls would choose to retire the units as of the ozone season in which the units would otherwise become subject to the backstop daily emissions rate provisions, regardless of whether replacement investments had been completed.

Some of the commenters raising concerns about electric system reliability suggested potential modifications to the proposed rule that the commenters believed could help address their concerns. The suggestions included various mechanisms for suspending some or all of the trading program's requirements for certain EGUs at times when an RTO or other entity responsible for overseeing a region of the interconnected electrical grid determines that generation from those EGUs is needed and the EGUs might not otherwise agree to operate. Other suggestions focused on ways of providing EGUs with greater confidence that allowances would be available to cover their incremental emissions during particular events. A number of commenters used the term “reliability safety valve,” in some cases with reference to the types of suggestions just mentioned and in other cases without details. Some commenters pointed to the “safety valve” provision included in the Group 2 trading program regulations under the Revised CSAPR Update. Another commenter pointed to provisions for a “reliability safety valve” included in the Clean Power Plan (80 FR 64662, Oct. 23, 2015).

In addition to offering critiques and recommendations concerning the proposed rule's contents, some commenters claimed that the EPA had failed to conduct sufficient analysis of the potential implications of the proposed rule on electrical system reliability. These commenters called on the EPA to consult with RTOs and other entities with responsibilities relating to electric system reliability and to perform additional analysis. Some commenters advocated for renewed consultations and analysis before each planned adjustment to emissions budgets under the dynamic budget-setting process. Commenters cited the consultation processes followed during implementation of other EPA rules, such as the Mercury and Air Toxics Standards (MATS) (77 FR 9304, Feb. 16, 2012).

Response: The EPA disagrees with the comments asserting that this rule would threaten resource adequacy or otherwise degrade electric system reliability. The emissions reduction requirements for EGUs under this rule are being implemented through the mechanism of an allowance trading program. Under the trading program, no EGU is required to cease operation. The core trading program requirements for a participating EGU are to monitor and report the unit's NO_X emissions for each ozone season period and to surrender a quantity of allowances after the end of the ozone season based on the reported emissions. To address states' obligations under the good neighbor provision, some units of course will have to take some type of action to reduce emissions, the actions taken to reduce emissions will generally have costs, and some EGU owners will conclude that, all else being equal, retiring a particular EGU and replacing it with cleaner generating capacity is likely to be a more economic option from the perspective of the unit's customers and/or owners than making substantial investments in new emissions controls at the unit. However, the EPA also understands that before implementing such a retirement decision, the unit's owner will follow the processes put in place by the relevant RTO, balancing authority, or state regulator to protect electric system reliability. These processes typically include analysis of the potential impacts of the proposed EGU retirement on electrical system reliability, identification of options for mitigating any identified adverse impacts, and, in some cases, temporary provision of additional revenues to support the EGU's continued operation until longer-term mitigation measures can be put in place. No commenter stated that this rule would somehow authorize any EGU owner to unilaterally retire a unit without following these processes, yet some comments nevertheless assume that is how multiple EGU owners would proceed, in violation of their obligations to RTOs, balancing authorities, or state regulators relating to the provision of reliable electric service. Assumptions of this nature are simply not reasonable. Like many commenters, the EPA does expect that retirement will be viewed as a more economic compliance strategy for some EGUs than installing new controls, but the Agency also expects that any resulting unit retirements will be carried out through an orderly process in which RTOs, balancing authorities, and state regulators use their powers to ensure that electric system reliability is protected. The trading program inherently provides ample flexibility to allow such an orderly transition to take place. In addition, as discussed later in this section, the EPA has adopted several changes in the final rule to increase flexibility specifically for the early years of the trading program for which commenters have indicated the greatest concerns about electric system reliability.

As an initial matter, the EPA notes two fundamental aspects of this rulemaking which together provide a strong foundation for the Agency's conclusion that the emissions reductions required from EGUs can be achieved with no adverse impacts on electric system reliability. First, there is ample evidence indicating that the required emissions reductions are feasible. As discussed in section V of this document, the magnitude and timing of the EGU emissions reductions required by this action reflect application of technologies that are already in widespread use, on schedules that are supported by industry experience. Second, the required emissions reductions are being implemented through the mechanism of a trading program. The enhanced trading program under this rule, like the trading programs established by the EPA under prior rules, provides EGU owners with opportunities to substitute emissions reductions from sources where achieving reductions is cheaper and easier for emissions reductions from other sources where achieving reductions is more costly or difficult. In general, an EGU owner has options to operate the emissions controls identified by the EPA for that type of unit (including installation or upgrade of controls where necessary), operate other types of emissions controls, or adapt the unit's levels of operation to produce less generation if the unit is a higher-emitting EGU or more generation if the unit is a lower-emitting EGU. The backstop daily emissions rate provisions in this rule reduce the degree of available flexibility relative to the degree of flexibility in the Agency's previous trading programs under CAIR and CSAPR but by no means eliminate it. Moreover, even the backstop rate provisions are structured as requirements to surrender additional allowances rather than as hard limits, providing a further element of flexibility No EGU is required to retire or is prohibited from operating at any time under this rule. EGUs only need to surrender of the appropriate quantities of allowances after the end of the control period.

Further, in the large number of comments submitted in this rulemaking that assert concerns over electric system reliability, no commenter has cited a single instance where implementation of an EPA trading program has actually caused an adverse reliability impact. Indeed, similar claims made in the context of the EPA's prior trading program rulemakings have shown a considerable gap between rhetoric and reality. For example, in the litigation over the industry's multiple motions to stay implementation of CSAPR, claims were made that allowing the rule to go into effect would compromise reliability. Yet in the 2012 ozone season starting just over 4 months after the rule was stayed, EGUs covered by CSAPR collectively emitted below the overall program budgets that the rule would have imposed in that year if the rule had been allowed to take effect, with most individual states emitting below their respective state budgets despite CSAPR not being in effect. Similarly, in the litigation over the 2015 Clean Power Plan, assertions that the rule would threaten electric system reliability were made by some utilities or their representatives, yet even though the Supreme Court stayed the rule in 2016, the industry achieved the rule's emissions reduction targets without the rule ever going into effect. See West Virginia v. EPA, 142 S. Ct. 2587, 2638 (2022) (Kagan, J., dissenting) (“[T]he industry didn't fall short of the [Clean Power] Plan's goal; rather, the industry exceeded that target, all on its own. . . . At the time of the repeal . . . `there [was] likely to be no difference between a world where the [Clean Power Plan was] implemented and one where it [was] not.' ”) (quoting 84 FR 32561). The claims that these rules would have had adverse reliability impacts were proved to be groundless.

Notwithstanding the long experience confirming the ability of the EPA's trading programs to obtain emissions reductions from EGUs without impairing the sector's ability to provide reliable electric service, the Agency of course does not rely here solely on its experience, but has carefully reviewed the comments on this topic for any information that might indicate the appropriateness of modifications to the enhanced trading program as proposed. In recognition of the important role that RTOs play in ensuring electric system reliability, and consistent with the requests of some commenters, the EPA has engaged in outreach to the RTOs that commented on the proposal to better understand their comments specifically and the reliability-related comments of other commenters more generally. Through these meetings, the central reliability-related concern was identified as one of timing. In order for retirement to be a viable compliance strategy for a unit that cannot be entirely spared until replacement investments in generation or transmission are completed, it must be possible for the unit to operate at critical times for a transition period. Like other stakeholders, the RTOs perceived implementation of the backstop daily emissions rate provisions on uncontrolled units as materially strengthening incentives for such units to either install controls or retire. The RTOs were concerned that the option for a coal-fired unit without SCR controls to maintain limited operation while surrendering allowances at a 3-for-1 ratio for all emissions exceeding the backstop daily rate was one that EGU owners would be reluctant to pursue. Accordingly, the RTOs expected considerable interest from EGU owners in retiring and replacing uncontrolled units as of the date of implementation of the backstop daily rate requirement on uncontrolled units, and they were concerned that the proposal to implement that requirement as of the 2027 control period did not allow sufficient time for planning and implementation of all the necessary generation and transmission investments to make this a viable compliance strategy. The RTOs described their concerns as greatest through approximately the 2029 control period.

The RTOs also described a concern about potentially illiquid allowance markets. They believed it was possible that some EGUs might claim an inability to operate at particular times when needed unless they had confidence that they would be able obtain additional allowances. The RTOs were particularly concerned that introduction of dynamic budgeting as proposed would create uncertainty for some EGUs regarding the quantities of allowances they would have available for use, particularly given the potentially large year-to-year swings if budgets were based on historical data from a single year. Some of the RTOs suggested potential solutions for these issues, principally in the form of auctions or RTO-administered allocations of allowances from pools of supplemental allowances, with access to the supplemental allowances triggered by certain indications of temporary stress on the electric system.

In the final rule, the EPA is adopting several changes from the proposal to help address the reliability-related concerns that were identified in comments and brought into greater focus by the consultations with the RTOs. The first change adopted in response to these comments is that application of the backstop daily NO_X emissions rate to units without existing SCR controls is being deferred until the 2030 control period, or the second control period in which a unit operates new SCR controls, if earlier. The purpose of this change is to address the concerns that application of the backstop daily NO_X emissions rate to EGUs without existing SCR starting in 2027 would provide insufficient time for planning and investments needed to facilitate unit retirement as a compliance pathway, which some commenters noted they prefer or have already planned. In particular, where an EGU owner would prefer to retire and replace an uncontrolled EGU rather than to install new controls, and in recognition that reliability-related needs may require some degree of operation from such units in the period before the investments needed to replace the unit can be completed, deferral of the backstop daily emissions rate provisions ensures that the necessary generation can be provided without being made subject to a 3-for-1 allowance surrender ratio that might render that compliance strategy uneconomic compared to the faster but less environmentally beneficial compliance strategy of installing new controls. The EPA has considered the statutory mandate that states' good neighbor obligations— including this action's requirement for large coal-fired EGUs to make emissions reductions commensurate with good SCR operation—be addressed as expeditiously as practicable. The EPA has also considered the fact that in this rule, the backstop daily emissions rate serves as a supplement to the broader requirement for emissions reductions commensurate with application of several control technologies at several types of EGUs, encompassing the extent of emissions reductions that would be incentivized by the backstop emissions rate requirement. The EPA views the backstop daily emissions rate as part of the solution to eliminating significant contribution in that it strongly incentivizes emissions-control operation throughout each day of the ozone season. See sections III.B.1.d, VI.B.1.b, VI.B.1.c.i. For that reason, in general we are finalizing the daily backstop emissions rate for units that have SCR installed or that install it in the future. It is only as an exception to that general rule that we defer the backstop daily emissions rate given the transition period and reliability concerns identified by commenters. The EPA finds that in this circumstance, as long as state emissions budgets continue to reflect the required degree of emissions reductions, deferral of the backstop rate requirement for uncontrolled units for a transition period can be justified on the basis of the greater long-term environmental benefits obtained through facilitating the replacement of these affected EGUs with cleaner sources of generation. Beginning in the 2030 ozone season, all coal-fired EGUs identified for SCR retrofit potential in this action will be subject to the backstop daily emissions rate. Any such units that remain in operation in that year can and should meet the backstop daily emissions rate or be subject to the heightened allowance surrender ratio.

The second change from the proposal adopted in response to the reliability-related comments is that the target percentage of the states' emissions budgets used to recalibrate the target bank level will be set at the proposed 10.5 percent starting in the 2030 control period, and for the control periods from 2024 through 2029, a target percentage of 21 percent will be used instead. The adoption of the higher target percentage for use through the 2029 control period is intended to promote greater allowance market liquidity during a period of relatively rapid fleet transition about which commenters expressed more focused reliability-related needs. As discussed later in this section, the EPA expects the introduction of the bank recalibration process in 2024 generally to boost market liquidity (by discouraging allowance hoarding) and also considers the target percentage of 10.5 percent set forth in the proposal well supported. Nevertheless, the Agency agrees with suggestions by commenters that, at least in the early years of the enhanced trading program, a larger bank would provide further liquidity and would give program participants greater confidence that allowances would be available for purchase when needed. Greater confidence by sources would help address RTOs' concern about the possibility that some sources could be reluctant to operate if they were unsure of their ability to procure allowances to cover their emissions. In finding that this modification from proposal is appropriate, the EPA has considered the fact that use of a higher target percentage will not result in the creation of any additional allowances in any control period, because under the recalibration provisions, when the total quantity of allowances banked from the previous control period is less than the bank target level, the consequence is not that additional allowances are created to raise the bank to the target level, but simply that no bank adjustment is carried out. We also note that while including an annual bank recalibration of any percentage is an enhancement in the trading program from prior trading programs under the good neighbor provision established in the CAIR, CSAPR, CSAPR Update, and Revised CSAPR Update rulemakings, it is not unprecedented; the trading program established under the NO_X SIP Call included “progressive flow control” provisions that were designed differently from the bank recalibration provisions in this rule but had the same purpose and general effect.

The third change from the proposal adopted in response to the reliability-related comments is that the EPA is determining preset state emissions budgets not only for the control periods in 2023 and 2024 as proposed, but also for the control periods in 2025 through 2029. Finalizing preset state emissions budgets through 2029 will establish predictable amounts for the minimum quantities of allowances available during the period when commenters have expressed concern that the reliability-related need for such predictability is greatest. Moreover, the EPA will also determine state emissions budgets using the final dynamic budget-setting methodology for the control periods in 2026 through 2029, and for each state and control period, the dynamic budget to be published in the future will only supplant the preset budget finalized in this rule for a control period in which that dynamic budget is higher than the corresponding preset budget. The reason for using dynamic budgets when they are higher than the corresponding preset budgets is that the EPA recognizes that evolution of the EGU fleet will not follow the exact path projected at the time of the rulemaking, and that by not accounting for certain events, the preset methodology could result in issuance of smaller quantities of allowances than the EPA would find consistent with the quantities of emissions from a well-controlled EGU fleet using the dynamic budget-setting methodology. Events that could cause preset budgets to underpredict a state's well-controlled emissions, which are more likely in years farther in the future from the time of the rulemaking, include deferral of a large EGU's previously planned retirement date or increases in electricity demand that outpace the general trend of lower-emitting or non-emitting generation replacing higher-emitting generation. After considering the commenters' interest in greater predictability during the early years of the amended trading program as well as the need to protect against instances where the preset budgets could underpredict a state's well-controlled emissions in years farther from the year of the rulemaking, the EPA finds that the combination of these factors justifies the approach of using the higher of the two budgets for the control periods from 2026 through 2029.

In addition to the changes made in response to reliability-related comments, several other changes to the proposal being adopted primarily for other reasons will also help address the factors identified as reliability-related concerns. Most notably, the EPA is adopting changes to the dynamic budget computation procedure to incorporate multiple years of heat input data, which will reduce year-to-year variability in the budgets determined under that procedure and should to some extent reduce uncertainty about the quantities of allowances available for use in instances where a dynamic budget is being used instead of preset budget. In addition, the adoption of a 50-ton threshold before application of the 3-for-1 surrender ratio to emissions exceeding the backstop daily NO_X emissions rate should ensure that no unit incurs the higher surrender ratio solely because of unavoidable emissions during startup and should help address concerns that some units might be reluctant to operate because of the associated emissions- related costs. Also, the 2026–2027 phase-in of emissions reductions commensurate with installation of new SCR controls will increase the quantities of allowances available in the 2026 state emissions budgets for most states in the trading program.

To summarize: in light of the strong record supporting the feasibility of the emissions reductions required from EGUs; the use of a trading program as the mechanism for achieving those emissions reductions, with multiple options for achieving compliance and no requirements to cease operation of any individual EGU at any time; the established processes of RTOs, other balancing authorities, and state regulators for managing any EGU retirement requests that do occur in an orderly manner with evaluation of potential reliability impacts and implementation of mitigation measures where needed; the unbroken, decades-long historical success of the EPA's trading programs at achieving emissions reductions without any adverse reliability impacts; the views expressed by commenters that facilitating EGU retirement and replacement as a possible compliance strategy through 2029 would be particularly helpful; the changes made in the final rule for control periods through 2029 specifically to increase flexibility during this transitional period, including deferring application of the backstop daily emissions rate provisions for EGUs without existing SCR controls, increasing the target percentage used to determine the target allowance bank level for purposes of the bank recalibration provisions, and establishing preset state emissions budgets which serve as floors against potential dynamic budget imposition in those control periods; and the changes made in the final rule incorporating multiple years of heat input data into the dynamic budget-setting procedure, adding a 50-ton threshold before application of the 3-for-1 surrender ratio to emissions exceeding the backstop daily NO_X emissions rate, and phasing in emissions reductions requirements commensurate with new SCR installations through 2027; the EPA concludes that this action does not pose any material risk of adverse impact to electric system reliability.

The EPA has also considered the other suggestions offered by commenters for addressing reliability-related issues. With respect to suggestions that the rule should include provisions allowing some or all of the trading program's requirements to be suspended at times when an RTO or other entity with grid management responsibilities determines there is a reliability-related need, the EPA again observes that the rule's emissions reduction requirements are being implemented through a trading program mechanism which makes exceptions of this nature unnecessary. Trading programs inherently offer the flexibility to accommodate variability in the utilization of individual units. The “reliability safety valve” provisions in the Clean Power Plan, which one commenter cited as a precedent to support some form of temporary exemption under this rule, in fact was available only in situations where a state plan did not allow emissions trading and instead imposed unit-specific emissions constraints. See80 FR 64877–879. Even the 3-for-1 allowance surrender ratio under the backstop daily NO_X emissions rate provisions can be met through the surrender of additional allowances. The rule does not bar any EGU from operating at any time as long as all allowance surrender requirements are met.

With respect to suggestions that the EPA must undertake recurring modeling of the evolving electrical system and consult with RTOs before each planned adjustment to emissions budgets, which start from the premise that the rule poses risk to electric system reliability that must be continuously monitored, the EPA disagrees with the premise and therefore also disagrees with the suggestions. As discussed in section V of this document, the EPA has taken care to ensure that the emissions reduction requirements applicable to EGUs under this rule are feasible through application of the control technologies selected as the basis of the emissions reductions. The EPA has also performed modeling in this rulemaking to assess the benefits and costs of the rule when all required emissions reductions are achieved. That modeling, which incorporates a representation of electrical grid regions and interregional constraints on energy and capacity exchange, affirms the feasibility of the overall emissions reduction requirements and is illustrative of a control strategy where some units retire and are replaced instead of installing new controls. The EPA has also consulted with the RTOs (as well as other balancing authorities) in the course of this rulemaking to ensure that the EPA understood the concerns expressed in their comments such that we could address those comments in this final rule. The EPA does not agree that further modeling or ongoing consultations with RTOs are needed in advance of the recurring dynamic budget adjustments, which do not increase the stringency of the rule's emissions reduction requirements established in the final rule. The extensive consultation processes adopted by the Agency in conjunction with the MATS rulemaking are not a relevant precedent; the MATS rule, which was promulgated to address a different statutory mandate, was structured in the form of unit-specific emissions constraints, fundamentally different from the requirements of this rule. The EPA notes that other entities responsible for maintaining reliability and managing entry and exit of resources, including the North American Electric Reliability Corporation (NERC) and RTOs and other balancing authorities, already routinely assess resource adequacy and reliability inclusive of meeting all regulatory requirements, including environmental requirements.

While the EPA does not agree that such consultations are a necessary precondition for successful implementation of this rule, the Agency remains available to engage with any affected EGU or reliability authority requesting to meet and discuss the intersection of its power sector regulatory programs with electric reliability planning and operations. The EPA is also continuing its practice of meeting with the U.S. Department of Energy and the Federal Energy Regulatory Commission to maintain mutual awareness of how Federal actions and programs intersect with the industry's responsibility to maintain electric reliability.

The EPA is not adopting the suggestion to replicate the so-called “safety valve” mechanism created under the Revised CSAPR Update. That mechanism, cited by some commenters as potential precedent for an unspecified form of “reliability safety valve” in this action, gave owners of covered EGUs a one-time opportunity to voluntarily convert allowances banked under the Group 2 trading program to allowances useable in the Group 3 trading program at an 18-for-1 ratio for use in the trading program's initial control period in 2021. See82 FR 23137–138. EGU owners chose to use the voluntary mechanism to acquire a total of 382 allowances, representing only 0.36 percent of the sum of the state emissions budgets and only 0.26 percent of the total quantity of allowances available for compliance in that control period. For the 2023 control period, the bank of allowances carried over from the 2022 control period plus the incremental starting bank that will be created by conversion of additional allowances banked under the Group 2 trading program (see section VI.B.12.b of this document) will total over 30 percent of the full-season emissions budgets. Given the larger starting bank and this rule's bank recalibration provisions (which will be implemented starting with the 2024 control period, but which the EPA expects will increase allowance market liquidity starting with the 2023 control period), the Agency views establishment of a one-time voluntary conversion opportunity for the 2023 control period analogous to the Revised CSAPR Update's “safety valve” provision as unnecessary.

Finally, in the final rule the EPA is not adopting any of the other suggestions concerning additional mechanisms to make additional allowances available through auctions or RTO-administered allowance pools. For the reasons discussed throughout this section, the EPA concludes that the trading program as established in this action provides a flexible compliance mechanism that will allow the required emissions reductions to be achieved without the need for creation of additional allowances. However, the EPA also recognizes the potential for allowance market liquidity to be further increased through some form of auction mechanism. For instance, it may be appropriate to pair the introduction of an auction with a reduction in the bank recalibration percentage that begins earlier than 2030. Through a supplemental rulemaking, the Agency intends to propose and take comment on potential amendments to the Group 3 trading program that would add such an auction mechanism to the regulations and make other appropriate adjustments in the implementation framework at Step 4.

2. Expansion of Geographic Scope

In light of the findings at Steps 1, 2, and 3 of the 4-step interstate transport framework, the EPA is expanding the geographic scope of the existing CSAPR NO_X Ozone Season Group 3 Trading Program to encompass additional states (and Indian country within the borders of such states) with EGU emissions that significantly contribute for purposes of the 2015 ozone NAAQS. Specifically, the EPA is expanding the Group 3 trading program to include the following states and Indian country within the borders of the states: Alabama, Arkansas, Minnesota, Mississippi, Missouri, Nevada, Oklahoma, Texas, Utah, and Wisconsin. Any unit located in a newly added jurisdiction that meets the applicability criteria for the Group 3 trading program will become an affected unit under the program, as discussed in section VI.B.3 of this document.

CSAPR, the CSAPR Update, and the Revised CSAPR Update also applied to sources in Indian country, although, when those rules were issued, no existing EGUs within the regions covered by the rules were located on lands that the EPA understood at the time to be Indian country. In contrast, within the geographic scope of this rulemaking, the EPA is aware of areas of Indian country within the borders of both Utah and Oklahoma with existing EGUs that meet the program's applicability criteria. Issues related to state, tribal, and Federal CAA implementation planning authority with respect to sources in Indian country in general and in these areas in particular are discussed in section III.C.2 of this document. EPA's approach for determining a portion of each state's budget for each control period that will be set aside for allocation to any units in areas of Indian country within the state not subject to the state's CAA implementation planning authority is discussed in section VI.B.9 of this document.

Units within the borders of each newly added state will join the Group 3 trading program on one of two possible dates during the program's 2023 control period (that is, the period from May 1, 2023, through September 30, 2023). The reason that two entry dates are necessary is that, as discussed in section VI.B.12.a of this document, the effective date is expected to fall after May 1, 2023. In the case of states (and Indian country within the states' borders) whose sources do not currently participate in the CSAPR NO_X Ozone Season Group 2 trading program—Minnesota, Nevada, and Utah—the sources will begin participating in the Group 3 trading program on the rule's effective date. However, in the case of the states (and Indian country within the states' borders) whose sources do currently participate in the Group 2 trading program—Alabama, Arkansas, Mississippi, Missouri, Oklahoma, Texas, and Wisconsin—the sources will begin participating in the Group 3 trading program on May 1, 2023, regardless of the rule's effective date, subject to transitional provisions designed to ensure that the increased stringency of the Group 3 trading program as revised in this rulemaking will not substantively affect the sources' requirements prior to the rule's effective date. This approach provides a simpler transition for the sources historically covered by the Group 2 trading program than the alternative approach of being required to switch from the Group 2 trading program to the Group 3 trading program in the middle of a control period, and it is the same approach that was followed for sources that transitioned from the Group 2 trading program to the Group 3 trading program in 2021 under the Revised CSAPR Update. Section VI.B.12.a of this document contains further discussion of the rationale for this approach and the specific transitional provisions.

The EPA notes that under the rule, the expanded Group 3 trading program will include not only 19 states for which the EPA is determining that the required control stringency includes, among other measures, installation of new post-combustion controls, but also three states—Alabama, Minnesota, and Wisconsin—for which the EPA is determining that the required control stringency does not include such measures. In previous rulemakings, the EPA has chosen to combine states in a single multi-state trading program only where the selected control stringencies were comparable, to ensure that states did not effectively shift their emissions reduction requirements to other states with less stringent emissions reduction requirements by using net out-of-state purchased allowances. Although the assurance provisions in the CSAPR trading programs were designed to address the same general concern about excessive shifting of emissions reduction activities between states, EPA chose not to rely on the assurance provisions as sufficient to allow for interstate trading in situations where the states were assigned differing emissions control stringencies.

In this rulemaking, the EPA believes the previous concern about the possibility that certain states might not make the required emissions reductions is sufficiently addressed through the various enhancements to the design of the trading program, even where states have been assigned differing emissions control stringencies. First, the existing assurance provisions are being substantially strengthened through the addition of the unit-specific secondary emissions limitations discussed in sections VI.B.1.c.ii and VI.B.8. Second, by ensuring that individual units operate their emissions controls effectively, the unit-specific backstop daily emissions rate provisions discussed in sections VI.B.1.c.i and VI.B.7 will necessarily also ensure that required emissions reductions occur within the state. With these enhancements to the design of the trading program, the EPA does not believe it is necessary for sources in Alabama, Minnesota, and Wisconsin to be excluded from the revised Group 3 trading program simply because their emissions budgets reflect a different selected emissions control stringency than the other states in the program.

The EPA's legal and analytic bases for expansion of the Group 3 trading program to each of the additional covered states, as well as responses to the principal related comments, are discussed in sections III, IV, and V of this document, respectively, and responses to additional comments are contained in the RTC document. With respect to the proposed approach of including all states covered by the rule in a single trading program even where the assigned control stringencies differ, the only comments received by the EPA supported the approach, which is finalized as proposed.

3. Applicability and Tentative Identification of Newly Affected Units

The Group 3 trading program generally applies to any stationary, fossil-fuel-fired boiler or stationary, fossil fuel-fired combustion turbine located in a covered state (or Indian country within the borders of a covered state) and serving at any time on or after January 1, 2005, a generator with nameplate capacity exceeding 25 MW and producing electricity for sale, with exemptions for certain cogeneration units and certain solid waste incineration units. To qualify for an exemption as a cogeneration unit, an otherwise-affected unit generally (1) must be designed to produce electricity and useful thermal energy through the sequential use of energy, (2) must convert energy inputs to energy outputs with efficiency exceeding specified minimum levels, and (3) may not produce electricity for sale in amounts above specified thresholds. To qualify for an exemption as a solid waste incineration unit, an otherwise-affected unit generally (1) must meet the CAA section 129(g)(1) definition of a “solid waste incineration unit” and (2) may not consume fossil fuel in amounts above specified thresholds. The complete text of the Group 3 trading program's applicability provisions and the associated definitions can be found at 40 CFR 97.1004 and 97.1002, respectively. The applicability of this rule to MWCs and cogeneration units outside the Group 3 trading program is discussed in sections V.B.3.a and V.B.3.c of this document, respectively, and MWC applicability criteria are further discussed in section VI.C.6 of this document.

In this rulemaking, the EPA did not propose and is not finalizing any revisions to the existing applicability provisions for the Group 3 trading program. Thus, any unit that is located in a newly added state and that meets the existing applicability criteria for the Group 3 trading program will become an affected unit under the program. The fact that the applicability criteria for all of the CSAPR trading programs are identical therefore is sufficient to establish that any units that are currently required to participate in another CSAPR trading program in any of the additional states where such other programs currently are in effect—Alabama, Arkansas, Minnesota, Mississippi, Missouri, Oklahoma, Texas, and Wisconsin (including Indian country within the borders of such states)—will also become subject to the Group 3 trading program.

In the additional states where other CSAPR trading programs are not currently in effect—Nevada and Utah (including Indian country within the borders of such states)—units already subject to the Acid Rain Program under that program's applicability criteria ( see40 CFR 72.6) generally also meet the applicability criteria for the Group 3 trading program. Based on a preliminary screening analysis of the units in these states that currently report emissions and operating data to the EPA under the Acid Rain Program, the Agency believes that all such units are likely to meet the applicability criteria for the Group 3 trading program.

Because the applicability criteria for the Acid Rain Program and the Group 3 trading program are not identical, it is possible that some units could meet the applicability criteria for the Group 3 trading program even if they are not subject to the Acid Rain Program. Using data reported to the U.S. Energy Information Administration, in the proposal the EPA identified six sources in Nevada and Utah (and Indian country within the borders of the states) with a total of 15 units that appear to meet the general applicability criteria for the Group 3 trading program and that do not currently report NO_X emissions and operating data to the EPA under the Acid Rain Program. These units were listed in a table in the proposed rule, and the data from that table for these units are reproduced as Table VI.B.3–1 of this document. For each of these units, the table shows the estimated historical heat input and emissions data that the EPA proposed to use for the unit when determining state emissions budgets if the unit was ultimately treated as subject to the Group 3 trading program. The EPA requested comment on whether each listed unit would or would not meet all relevant criteria set forth in 40 CFR 97.1004 and the associated definitions in 97.1002 to qualify for an exemption from the trading program and whether the estimated historical heat input and emissions data identified for each unit were representative. With respect to the listed units within the borders of Nevada or Utah, the EPA received no comments asserting either that the units qualified for applicability exemptions or that the estimated data identified by the EPA were unrepresentative. For purposes of this rule, the EPA is therefore presuming that the units listed in Table VI.B.3–1 do not qualify for applicability exemptions and that the estimated data shown in the table for each unit are representative. However, the owners and operators of the sources retain the option to seek applicability determinations under the trading program regulations at 40 CFR 97.1004(c).

4. State Emissions Budgets

In this final rule, the EPA is using a combination of a “preset” budget calculation methodology and a “dynamic” budget calculation methodology to establish state emissions budgets for the Group 3 trading program. A “preset” budget is one for which the absolute amount expressed as tons per ozone season control period is established in this final rule. It uses the latest data currently available on EGU fleet composition at the time of this final action. A “dynamic” budget is one for which the formula and emissions-rate information is finalized in this rule, but updated EGU heat input and inventory information is used on a rolling basis to set the total tons per ozone season for each control period. Both methods of budget calculation are designed to set budgets reflective of the emissions control strategies and associated stringency levels (expressed as an emissions rate of pounds of NO_X per mmBtu) identified for relevant EGU types at Step 3—which we will refer to in this section as the “Step 3 emissions control stringency.” Preset budgets provide greater certainty for planning purposes and can be reliably established in the short-term based on known, upcoming changes in the EGU fleet. Due to build time for new units and planning and approval processes for plant retirements, these major fleet alterations are often known several years in advance. This information facilitates presetting budgets that appropriately calibrate the identified control stringency to the fleet. Dynamic budgets better assure that the budgets remain commensurate with the Step 3 emissions control stringency over the longer term, as currently unknown changes in the EGU fleet occur. In this final rule, in response to comments, we have adjusted the proposal to give a greater role for preset budgets through 2029, while dynamic budgeting will be phased in to provide greater certainty in the short term and allow for a transition period to an exclusively “dynamic” approach beginning in 2030.

For the control periods from 2023 through 2025, the preset budgets established in the rule will serve as the state emissions budgets for the control periods in those years, with no role for dynamic budgeting. For the control periods from 2026 through 2029, the EPA is determining preset emissions budgets for each control period in the rule and will also calculate and publish dynamic budgets for each state in the year before each control period using the dynamic budget-setting methodology finalized in this rule, applied to data available at the time of the calculations. For these four control periods, each state's preset budget serves as a floor and may be supplanted by the dynamic emissions budget EPA calculates for the state for that control period only if the dynamic budget is higher than the preset budget. For control periods in 2030 and thereafter, the state emissions budgets will be the dynamic budgets calculated and published in the year before each control period.

In the dynamic budget calculation methodology, it is the fleet composition (reflected by heat input patterns across the fleet in service, inclusive of EGU entry and exit) that is dynamic, while the emissions stringency finalized in this rule is constant, as reflected in emissions rates for various types of units. Multiplying the assumed emissions rate for each unit (as finalized in this rule) by the identified recent historical heat input for each unit and summing the results to the state level would provide a given year's state dynamic emissions budgets. Dynamic budgets are a product of the formula promulgated in this action applied to a rolling three-year average of reported heat input data at the state level and a rolling highest-three-of-five-year average of reported heat input data at the unit level. As such, the EPA is confident that dynamic budgets will more accurately reflect power sector composition, particularly in later years, and certainly from 2030 and beyond, than preset budgets could and will therefore better implement the Step 3 emissions control stringency over long time horizons.

Starting in 2025 (for the 2026 control period), the dynamic budgets, along with the underlying data and calculations will be publicly announced, and this will occur approximately one year before the relevant control period begins. These will be published in the Federal Register through notices of data availability (NODAs), similar to how other periodic actions that are ministerial in nature to implement the trading programs are currently handled. And as with such other actions, interested parties will have the opportunity to seek corrections or administrative adjudication under 40 CFR part 78 if they believe any data used in making these calculations, or the calculations themselves, are in error.

To illustrate how dynamic budgeting will work after the transition from preset budgets, the dynamic budgets for the 2030 ozone season control period will be identified by May 1, 2029, using the latest available average of three years of reported operational data at that time ( i.e., the average of 2026–2028 heat input data at the state level and 2024–2028 years of rolling data at the unit level) applied in a simple mathematical formula finalized in this rule, which multiplies this heat input data by the emissions rates quantified in this rule. Therefore, if a unit retires before the start of the 2028 ozone season but had not announced its upcoming retirement at the time of this rule's finalization, the dynamic budget approach ensures that the dynamic budgets for 2030 and subsequent control periods would represent the identified control stringency applied to a fleet reflecting that retirement.

The two examples discussed next illustrate the implementation of the dynamic budget during the 2026–2029 time period. During this period, the state emissions budget for each state for a given control period will be the preset state emissions budget unless the dynamic budget is higher. This approach accommodates scenarios where baseline fossil heat input may exceed levels anticipated by EPA in the preset budgets ( e.g., this could result from greater electric vehicle penetration rates). Table VI.B.4–1 illustrates this scenario. In the preset budget approach for 2028, the 2028 heat input is estimated based on the latest available heat input data at the time of rule proposal ( i.e., 2021; see the subsection on preset budget methodology later in this section), which cannot reflect a subsequent change in fleet heat input values (column 2) due to, e.g., increased utilization to meet increased electric load. However, the dynamic budget would use 2022–2026 heat input values at the unit level and 2024–2026 heat input values at the state level—as opposed to 2021 heat input values—as the latest representative values to inform the 2028 state emissions budget. Therefore, the heat input values in column 2 under the dynamic scenario reflect the change in fleet utilization levels, and when multiplied by the emissions rates reflecting the Step 3 emissions control stringency in this final rule, the corresponding emissions (18,700 tons) summed in column 4 constitute a state budget that more accurately reflects the Step 3 emissions control stringency applied to the fleet composition for that year, as opposed to the 17,000 tons identified in the preset budget approach. As illustrated in the example, the dynamic variable is the heat input variable, which changes over time. In this instance, the dynamic budget value of 18,700 tons would be implemented for 2028 instead of the preset value, and thus accommodate the unforeseen utilization changes in response to higher demand.

In the second table, Table VI.B.4–2, the dynamic budget is lower than the preset budget due to retirements that were not foreseen at the time the preset budgets were determined. In the preset budget approach for 2028, the 2028 heat input is still estimated based on the latest available heat input data at the time of rule proposal ( i.e., 2021), which cannot reflect a subsequent fleet change in heat input values due to an unanticipated retirement of one of the state's coal-fired units before the start of the 2028 ozone season. However, the dynamic budget again would use 2022–2026 heat input values at the unit level and 2024–2026 heat input values at the state level—as opposed to 2021 heat input values—as the latest representative values to inform the 2028 state emissions budget, which would reflect the decline in coal heat input and replacement with natural gas heat input (capturing the coal unit's retirement). Therefore, the heat input values under the dynamic budget scenario reflect the change in fleet composition, and when multiplied by the relevant emissions rates reflecting the Step 3 emissions control stringency identified in this final rule, the corresponding emissions (15,000 tons) constitute a state budget that reflects the identified control stringency applied to the fleet composition for that year as opposed to the 17,000 tons in summed in the first table. However, for the 2026–2029 period, in which the EPA implements an approach that utilizes the higher of the dynamic budget or preset budget, the budget implemented for 2028 in this scenario would be the 17,000 ton preset amount.

During the 2026–2029 transition period—during which substantial, publicly announced utility commitments exist for higher emitting units to exit the fleet—it is still possible that yet-to-be known, unit-specific retirements (such as illustrated in this second scenario) may result in dynamic budgets that are lower than the preset budgets finalized in this rule. However, during this transition period EPA believes that having the preset budgets serve as floors for the state emissions budgets is appropriate for two primary reasons identified by commenters. First, commenters repeatedly emphasized the need for certainty and flexibility to successfully carryout plans for significant fleet transition through the end of the decade. The 2026–2029 period is expected to have substantial fleet turnover. Current Form EIA–860 data, in which utilities report their retirement plans, identify 2028 as the year with the most planned coal capacity retirements during the 2023–2029 timeframe. Using preset budgets as state emissions budget floors provides states and utilities with information on minimum quantities of allowances that can be used for planning purposes. In turn, this fosters the operational flexibility needed while putting generation and transmission solutions into place to accommodate such elevated levels of retirements. Second, the latter part of the decade has a significant amount of unit-level firm retirements already planned and announced for purposes of compliance with other power sector regulations or fulfillment of utility commitments. These known retirements are already captured in the preset state budgets, with the result that the likelihood and magnitude of instances where a state's dynamic budget for a given control period would be lower than its preset budget for the control period is reduced in this 2026–2029 period relative to control periods further in the future for which retirement plans have not yet been announced. After 2029, the dynamic budgets from 2030 forward will fully capture all prior retirements and new builds when the fleet is entering this period where unit-specific data on such plans is less frequently available. For instance, through the remaining portion of the decade, the amount of coal steam retirements identified and reported through Form EIA–860 is nearly 7 GW each year. However, for the decade beginning in 2030—the amount of capacity currently reported with a planned retirement is less than 2 GW each year. This yet-to-be available data and relative lack of currently known firm retirement plans for 2030 and beyond make dynamic budget implementation for those years essential for state emissions budgets to maintain the Step 3 control stringency required under this rule.

In summary, for the control periods in 2023 through 2025, EPA is providing only preset budgets in this final rule because those control periods are in the immediate future and would not substantially benefit from the use of future reported data. For these years, the certainty around new builds and retirements is higher than ensuing years. For the ozone season control periods of 2026 through 2029, EPA is providing both preset budgets in this final rule and dynamic budgets via future ministerial actions. For those control periods from 2026 through 2029, the preset budgets finalized in this rule serve as floors, such that a given state's dynamic budget ultimately calculated and published for that control period will apply to that state's affected EGUs only if it is higher than the corresponding preset budget finalized in this rulemaking. This approach is in response to stakeholder comments requesting more advance notice regarding the total quantities of allowances available to accommodate compliance planning through the latter half of the decade, during a period of particularly high fleet transition expected with or without this rulemaking.

EPA's emissions budget methodology and formula for establishing Group 3 budgets are described in detail in the Ozone Transport Policy Analysis Final Rule TSD and summarized later in this section.

a. Methodology for Determining Preset State Emissions Budgets for the 2023 Through 2029 Control Periods

To compose preset state emissions budgets, the EPA is using the best available data at the time of developing this final rule regarding retirements and new builds. The EPA relies on a compilation of data from Form EIA–860 (where facilities report their future retirement plans), the PJM Retirement Tracker, utilities' integrated resource plans, notification of compliance plans with other EPA power sector regulatory requirements, and other information sources that EPA routinely canvasses to populate the data fields included in the Agency's NEEDS database. The EPA has updated this data on retirements and new builds using the latest information available from these sources at the time of final rule development as well as input provided by commenters.

For determining preset state emissions budgets, the EPA generally uses historical ozone season data from the 2021 ozone season, the most recent data available to EPA and to commenters responding to this rulemaking's proposal and providing a reasonable representation of near-term fleet conditions. This is similar to the approach taken in the CSAPR Update and the Revised CSAPR Update, where the EPA likewise began with data for the most recent ozone season at the time of proposal (2015 and 2019, respectively).

By using historical unit-level NO_X emissions rates, heat input, and emissions data in the first stage of determining preset emissions budgets, the EPA is grounding its budgets in the most recent representative historical operation for the covered units at the time EPA began its final rulemaking. This data set is a reasonable starting point for the budget-setting process as it reflects recent publicly available and quality assured data reported by affected facilities under 40 CFR part 75, largely using CEMS. The reporting requirements include quality control measures, verification measures, and instrumentation to best record and report the data. In addition, the designated representatives of EGU sources are required to attest to the accuracy and completeness of the data.

The first step in deriving the future year state emissions budget is to calibrate historical data to planned future fleet conditions. EPA does this by adjusting this historical baseline information to reflect the known changes ( e.g., when deriving the 2023 state emissions budget, EPA starts by adjusting 2021 unit-level data to reflect changes announced and planned to occur by 2023). The EPA adjusted the 2021 ozone-season data to reflect committed fleet changes expected to occur in the baseline. This includes announced and confirmed retirements, new builds, and retrofits that occur after 2021 but prior to 2023. For example, if a unit emitted in 2021, but retired prior to May 1, 2022, its 2021 emissions would not be included in the 2023 baseline estimate. For units that had no known changes, the EPA uses the actual emissions, heat input, and emissions rates reported for 2021 as the baseline starting point for calculating the 2023 state emissions budgets. Using this method, the EPA arrived at a baseline emission, heat input, and emissions rate estimate for each unit for a future year ( e.g., 2023).

The second step in deriving the preset state emissions budgets is for EPA to take the adjusted historical data from Step 1, and adjust the emissions rates and mass emissions to reflect the control stringencies identified as appropriate for EGUs of that type. For instance, if an SCR-equipped unit was not operating its SCR so as to achieve a seasonal average emissions rate of 0.08 lb/mmBtu or less in the historical baseline, the EPA lowered that unit's assumed emissions rate to 0.08 lb/mmBtu and calculated the impact on the unit's mass emissions. Note that the heat input is held constant for the unit in the process, reflecting the same level of unit operation compared to historical 2021 data. The improved emissions rate of 0.08 lb/mmBtu is applied to this constant heat input, reflecting control optimization. In this manner, the unit-level totals from Step 1 are adjusted to reflect the additional application of the assumed control technology at a given control stringency. This is illustrated in Table VI.B.4.a–1. Row 1 reflects the 2021 historical data for this SCR-controlled unit. Row 2 reflects no change (as there are no known changes such as planned retirement or coal-to-gas conversion). Row 3 reflects application of the Step 3 stringency ( i.e., a 0.08 lb/mmBtu emissions rate from SCR optimization). The resulting impact on emissions is a reduction from the historical 4,700 tons to an expected future level of 615 tons. A state's preset budget for a given control period is the sum of the amounts computed in this manner for each unit in the state for the control period.

For each control period from 2026 onward, the unit-specific emissions rates assumed for all affected states except Alabama, Minnesota, and Wisconsin will reflect the selected control stringency that incorporates post-combustion control retrofit opportunities for the relevant units identified in the state emissions budgets and calculations appendix to the Ozone Transport Policy Analysis Final Rule TSD. The emissions rates assigned to large coal-fired EGUs for 2026 state emissions budget computations only reflect 50 percent of the SCR retrofit emissions reduction potential at each of those units, to capture the phase-in approach EPA is taking for this control as described in section VI.A of this document. The EPA calculates these unit-level emissions rates in 2026 as the sum of the unit's baseline emissions rate and its controlled emissions rate divided by two ( i.e., 50 percent of the emissions reduction potential of that pollution control measure). The emissions rates assigned to these large coal-fired EGUs for 2027 state emissions budget computations reflect the full assumed SCR retrofit emissions potential at those units, by applying the controlled emissions rate only. For example, a coal steam unit greater than or equal to 100 MW currently lacking a SCR and emitting at 0.20 lb/mmBtu would be assumed to reduce its emissions rate to 0.125 lb/mmBtu rate in 2026 and 0.050 lb/mmBtu rate in 2027 for purposes of deriving its preset state emissions budgets in those years.

Comment: Some commenters suggested that EPA should not reflect planned retirements in its preset budgets. The suggestion stems from commenters' observation that those retirement decisions may yet change.

Response: The effectiveness of EPA's future year preset state emissions budgets depends on how well they are calibrated to the expected future fleet. Therefore, EPA believes it is important to incorporate expected new builds, retirements, and unit changes already slated to occur. Ignoring these factors would dilute, rather than strengthen, the ability of preset budgets to capture the most representative fleet of EGUs to which they will be applied. Omitting scheduled retirements and new builds from state emissions budgets would reflect units that power sector operators and planning authorities do not expect to exist, while failing to reflect units that are expected to exist.

EPA notes it is using the best available data at the time of the final rule. EPA relies on a compilation of data from Form EIA–860 where facilities report their future retirement plans. In addition, EPA is using data from regional transmission organizations who are cataloging, evaluating, and approving such retirement plans and data; data from notifications submitted directly to EPA by the utility themselves through comments; and retirement notifications submitted to permitting authorities. This information is highly reliable, real-world information that provides EPA with the high confidence that such retirements will in fact occur.

If a unit's future retirement does not occur on the currently scheduled date, EPA observes that such an unexpected departure from the currently available evidence would still not undermine the ability of affected EGUs to comply with their applicable state budgets. EPA's approach of using historical data and incorporation only of announced fleet changes in estimating its future engineering analytics baseline means that its future year baseline generation and retirement outlook for higher emitting sources is more likely to understate future retirements (rather than overstate as suggested by commenter), as EPA does not assume for the purpose of preset budget quantification any retirements beyond those that are already planned. In other words, in the 2023 through 2029 timeframe for which EPA is establishing preset state emissions budgets in this rulemaking, there are more likely to be additional future EGU retirements beyond those scheduled prior to the finalization of this rule than there are to be reversed or substantially delayed changes to already announced EGU retirement plans. For instance, subsequent to the EPA's finalization of the Revised CSAPR Update Rule budgets for 2023 (rule finalized in March 2021), the owners of Sammis Units 5–7 and Zimmer Unit 1 in Ohio (totaling nearly 3 GW of coal capacity) announced that the units would retire by 2023—nearly 5 years earlier than previously planned. These coal retirements were not captured in Ohio's 2023 or 2024 state emissions budgets established under the Revised CSAPR Update. Meanwhile, there have been no announcements of previously announced retirement plans being rescinded or delayed for other Ohio units. Similarly, the Joppa Power Plant in Illinois accelerated its retirement from 2025 to 2022 shortly after the Revised CSAPR Update Rule was signed.

We further observe that the commenters' concern is only materially meaningful for the 2023 through 2025 preset budget periods, where the currently known information is generally the most reliable. For the 2026–2029 control periods, if an anticipated fleet change such as an EGU retirement does not actually occur, the dynamic budget setting methodology would, all else being equal, generate a budget reflective of that unit's continued operation (as the budget would be based on the preceding years of historical data), and that dynamic budget will supplant the preset budget for that state (if it represents a total quantity of emissions higher than the preset budget).

Because the future is inherently uncertain, all analytic tools and information resources used in any estimation of future EGU emissions will yield some differences between the projected future and the realized future. Such potential differences may either increase or decrease future emissions in practice, and the unavoidable existence of such differences does not, on its own, render the EPA's inclusion of currently announced retirements an unreasonable feature of the methodology for determining future year preset emissions budgets. To the contrary, if the EPA failed to include these announced retirements, the rule would knowingly authorize amounts of additional, sustained pollution that are not currently expected to occur. If those retirements largely or entirely occur as currently scheduled, the overestimated state budgets would allow other EGUs to emit additional pollution in place of the emissions from the retired EGUs instead of maintaining or improving their emissions performance to eliminate significant contribution with nonattainment and interference with maintenance of the NAAQS.

Additionally, as noted elsewhere, EPA's use of a market-based program, a starting bank of converted allowances, and variability limits are all features that will readily accommodate whatever relatively limited differences in emissions may occur if a currently scheduled EGU retirement is ultimately postponed during the preset budget years of 2023 through 2025. Therefore, EPA's resulting preset state emissions budgets—inclusive of expected fleet turnover—are robust to the inherent uncertainty in future year baseline conditions for the period in which they are applied.

Comment: Some commenters suggested that EPA should use a multi-year baseline for all of its state budget derivations, including preset budgets, to control for outlier years that may not be representative of future years due to major weather events or other fleet disruptions (such as a large nuclear unit outage).

Response: For preset state emissions budget derivation, EPA is finalizing use of the same single-year historical baseline approach it used in the proposed rule. This approach is similar to the Revised CSAPR Update, where EPA also relied on a single-year historical baseline to inform its Step 3 approach. EPA's interest in a historical data set to inform this part of the analysis is to capture the most representative view of the power sector. For estimating preset state budgets, EPA finds that, particularly at the state level, more recent data is a better representation and basis for future year baselines rather than incorporating older data. Taking as an example preset budget estimation for the 2023 through 2025 ozone seasons, the EPA is able to compare its single-year base line to an alternative multi-year baseline ( e.g., a 3-year baseline encompassing 2020–2022) and determine that the single year baseline better reflects future fleet operation expectation than a multi-year baseline that incorporates units which have since retired as well as outlier patterns in load during pandemic-related shutdowns.

EPA recognizes that 2021 is the latest available historical data as of the preparation of this rulemaking, and therefore the most up-to-date picture of the fleet at the time EPA began its analysis. EPA then further evaluates the 2021 historical data at the state level to determine whether it was a representative starting point for estimating future year baseline levels and subsequently deriving the preset state emissions budgets. If the Agency finds any state-level anomalies, it makes necessary adjustments to the data. While unit-level variation may occur from year-to-year, those variations are often offset by substitute generation from other units within the state. Therefore, EPA conducts its first screening at the state level by identifying any states where 2021 heat input and 2021 emissions were the lowest year for heat input and emissions relative to the past several years (2018–2022, excluding 2020 due to shut downs and corresponding reduced utilization related to the pandemic onset). Then, for that limited number of states (AL, LA, MS, and TX) in which 2021 reflects the minimum fossil fuel heat input and minimum emissions over the baseline evaluation period, EPA—similar to prior rules—evaluated whether any unit-level anomalies in operation were driving this lower heat input at the state level. EPA examined unit-level 2021 outages to determine where an individual unit-level outage might yield a significant difference in state heat input, corresponding emissions baseline and resulting state emissions budgets. When applying this test to all of the units in the previously identified states (and even when applying to EGUs in all states for whom Federal implementation plans are finalized in this rulemaking), the EPA determined that the only unit with a 2021 outage that (1) decreased its output relative to preceding or subsequent years by 75 percent or more (signifying an outage), and (2) could potentially impact the state's emissions budget substantially as it constituted more than 5 percent of the state's heat input in a non-outage year was Daniel Unit 2 in Mississippi. EPA therefore adjusted this state's baseline heat input and NO_X emissions to reflect the operation of this unit based on its 2019 data—which was the second most recent year of data available at the time of proposal (excluding 2020 given atypical impacts from pandemic-related shutdowns) for which this unit operated. The EPA then applied the Step 3 mitigation strategies as appropriate to this unit ( i.e., combustion controls upgrade in 2024, SCR retrofit in 2026/2027) to derive this portion of Mississippi's budget. This test, and subsequent adjustment as necessary, enables EPA to utilize the latest, most representative data in a manner that is robust to any substantial state-level or region-level outlier events within that dataset and further validates EPA's comprehensive approach to using the most recent single year of data for preset budgets.

b. Methodology for Determining Dynamic State Emissions Budgets for Control Periods in 2026 onwards

In this final rule, the EPA is finalizing an approach of using multi-year baseline data for purposes of dynamic budget computation. The aforementioned testing of the representative nature of a single year of baseline data for purposes of preset budget setting is not possible in the dynamic budget process as that data will not be available until a later date. Further, the EPA generally agrees with commenters that use of a multi-year period will be more robust to any unrepresentative outlier years in fleet operation and thus better suited for purposes of dynamic budgets. The methodology for determining dynamic state emissions budgets for later control periods (2026 and beyond) relies on a nearly identical methodology for applying unit-level emissions rate assumptions as the preset budget methodology. But it uses more recent heat input data that will become available by that future time, employing a multi-year approach for identifying the heat input data so as to ensure representativeness.

For dynamic budgets, EPA uses more years of baseline data to control for any state-level and unit-level variation that may occur in a future single year that is not possible to identify at present. First, for each unit operating in the most recent ozone season for which data have been reported, EPA identifies the average of the three highest unit-level heat input values from the five ozone seasons ending with that ozone season to get a representative unit-level heat input. Ozone seasons for which a unit reported zero heat input are excluded from the averaging of the three highest heat input values for that unit. These representative unit-level heat input values established for each unit individually are then summed for all units in each state. Each unit's representative unit-level heat input is then divided into this state-level sum to get that unit's representative percent of the aggregated average heat input values for all affected EGUs in that state.

Next, EPA calculates a representative state-level heat input by taking the average state-level total heat input across affected EGUs from the most recent three ozone seasons for which data have been reported, to which the above-derived representative unit-level percentages of heat input are applied. The EPA uses a three-year baseline period for state-level heat input versus the five-year baseline period noted previously for unit-level heat input because there is less variation from year to year at the state level compared to the unit level. Multiplying the representative unit-level percentages of heat input by the representative state-level heat input yields a normalized unit-level heat input value for each affected EGU. This step assures that the total heat input being reflected in a dynamic state budget does not exceed the average total heat input reported by affected EGUs in that state from the three most recent years. Finally, each normalized unit-level heat input value is multiplied by the emissions rate reflecting the assumed unit-specific control stringency for each particular year (determined at Step 3) to get a unit-level emissions estimate. These unit-level emissions estimates are then summed to the state level to identify the dynamic budget for that year. This procedure to derive normalized unit-level heat input is captured in the following table:

The EPA will issue these dynamic budget quantifications approximately 1 year before the relevant control period. We view such actions as ministerial in nature in that no exercise of agency discretion is required. For instance, starting in early 2025, the EPA would take the most recent three years of state-level heat input data and the most recent five years of unit-level heat input data and calculate 2026 state emissions budgets using the methodology described previously. For 2026–2029, EPA is establishing the preset state emissions budgets finalized in this rulemaking and will only supplant those preset emissions budgets with the to-be-published dynamic emissions budgets if, for a given state and a given control period, that dynamic budget yields a higher level of emissions than the corresponding preset budget finalized in this rulemaking. For 2030 and beyond, the EPA solely uses the dynamic budget process.

By March 1 of 2025, and each year thereafter, the EPA will make publicly available through a NODA the preliminary state emissions budgets for the subsequent control period and will provide stakeholders with a 30-day opportunity to submit any objections to the updated data and computations. (This process will be similar to the releases of data and preliminary computations for allocations from new unit set-asides that is already used in existing CSAPR trading programs.) By May 1 of 2025, and each year thereafter, the EPA will publish the dynamic budgets for the ozone-season control period in the following calendar year. Through the 2029 ozone season control period, these budgets will only be imposed if the applicable dynamic state budget is higher than the corresponding preset state budget finalized in this rulemaking. Preliminary and final unit-level allowance allocations for the units in each state in each control period will be published on the same schedule as the dynamic budgets for the control period. For the control periods from 2026 through 2029, the allocations will reflect the higher of the preset or dynamic budget for each state, and after 2030, the allocations will reflect the dynamic budgets. Additional details, corresponding data and formulas, and examples for the dynamic budget are described in the Ozone Transport Policy Analysis Final Rule TSD.

Comment: Multiple commenters claimed that designing a dynamic budget process that relies on a single year of yet-to-be known heat input data may produce an unrepresentative view of fleet operations for the immediate ensuing years. Commenters pointed to the hypothetical of another pandemic-like year ( e.g., 2020) occurring in the future, noting that 2020 would have been a poor choice for estimating 2022 fleet operation and the same would likely hold true if a similar event occurred, for example, in 2025—that would consequently make that year a poor choice as a representative of 2027 baseline. They further pointed out that severe weather events and operating disruptions (a large nuclear plant outage) can similarly render a single year baseline a risky choice to inform future expectations.

Response: Insofar as the commenters are addressing the reference period for dynamic budget computation regarding years of data that have not yet occurred and therefore not currently available for evaluating their representative nature, EPA agrees and is incorporating a rolling 3-year baseline at the state level and a rolling 5-year baseline at the unit level for determining dynamic budgets in this final rule. These multi-year rolling baseline (or reference periods) will minimize any otherwise undue impact from individual years where fleet-level or unit-level heat input was uncharacteristically high or low. EPA determined that such an approach, while not needed for preset budgets, is necessary in the case of dynamic budgets because the baseline in that instance is occurring in a future year and therefore is not knowable and available to test for representativeness at the time of the final rule. To control for this type of uncertainty, the EPA finds it appropriate to use a multi-year baseline in this instance per commenter suggestion. While a multi-year baseline may have a slight drawback of using a slightly more dated past fleet performance (including emissions from higher emitting EGUs that may have subsequently reduced utilization by the target year for which the dynamic budget is being calculated) to estimate the expected future fleet performance at the emissions performance levels determined by the Step 3 result in this rulemaking, that drawback is worth the advantage of protecting against instances where atypical circumstances in the most recent single year may occur and not be representative of the subsequent year for which the dynamic budget is being estimated. This singular drawback of moving to a multi-year baseline is most pronounced in the early years of dynamic budgeting. Therefore, EPA is able to lessen the impact of this drawback of the multi-year baseline by extending the earliest start date of dynamic budgets from 2025 (as proposed) to 2026 in the final rule.

Comment: Commenters suggested that the dynamic budget procedure would not provide enough advance notice of state budget and unit level allocation for sources to adequately plan future year operation.

Response: EPA disagrees with the notion that the timing of the dynamic budget determination would occur too close to the control period to allow adequate operations planning for compliance. As described previously, the dynamic budget level would be provided approximately 1 year in advance of the start of the control period ( i.e., around May 1), and the allowance allocations would occur on July 1, approximately 10 months prior to the start of the compliance period. Not only is this an adequate amount of time as demonstrated by the successful implementation of past rules that have been finalized and implemented within several months of the beginning of the first affected compliance period ( e.g., Revised CSAPR Update), but EPA notes it is maintaining similar trading program flexibility and banking flexibilities of past programs which provide further opportunities for sources to procure allowances and plan for any future operating conditions. Finally, as noted previously, the EPA is providing preset budgets for the years 2023–2029, which serve as an effective floor on the state's ultimate emissions budget level for years 2026–2029, as states will receive the higher of the preset or dynamic budget for those years. This provision of certain preset state emissions budgets serving as a floor level for 2026–2029 should further assuage commenters' concerns regarding planning certainty about allowance allocations and state emissions budget levels during this period of power sector transition to cleaner energy sources.

Comment: Commenters raised concerns that there is a two-year lag in the dynamic budgets in that, for example, for the dynamic budget in the 2026 control period, the calculations will be based on heat input and inventory information reflective of data through 2024. Commenters contend that, if there is a much greater need for allowances for compliance due to unavoidable or unforeseen need for a higher amount of heat input than reflected in prior years' data, the budget for that control period will not reflect this need, and the allowances will only become available when the dynamic budget is calculated using that information ( i.e., 2025 data would be reflected starting in the 2027 dynamic budget). According to commenters, this lag could present a serious compliance challenge. Other commenters raised a concern in the opposite direction about the potential “slack” created by the lag time—meaning that as high-emitting units retire, their emissions and operation will still inform the state emissions budgets for additional years beyond their retirement due to the lag.

Response: The EPA recognizes there will be a data lag inherent in the computation of future year dynamic emissions budgets, because the dynamic budgets will reflect fleet composition and utilization data from recent previous control periods rather than the control periods for which the dynamic budgets are being calculated. This means that the resulting dynamic budgets will reflect a limited lag behind the actual pace of the EGU fleet's trends. However, on the whole, those trends are clearly toward more efficient and cleaner generating resources. Thus, the data lag on the whole will inure to the compliance benefit of EGUs by resulting in dynamic budgets that are generally calculated at levels likely to be somewhat higher than what a dynamic budget calculation reflecting real-time EGU operations would produce. The EPA believes this data lag is worthwhile to provide more compliance planning certainty and advance notice to affected EGUs of the dynamic budget applicable to an upcoming control period. Furthermore, this data lag in dynamic budget computation is comparable to the data lag of quantifying preset state budgets for 2023 through 2025 based upon 2021 data, and at no point in the long history of EPA's trading programs has such a data lag in state budget computation yielded any compliance problems for affected EGUs. Without dynamic budgeting, the data lag inherent in calculating preset budgets would grow unabated with the passage of time, as a fixed reference year of heat input levels would continually apply regardless of potentially higher heat input levels farther and farther into the future. By eliminating the increase in the length of the data lag, this new dynamic budgeting approach is a substantial improvement in performance of the program relative to previous approaches that were not capable of capturing changes over time in the fleet and its utilization beyond the scheduled changes known to the EPA at the time of establishing preset budgets.

The EPA disagrees that this lag will in fact pose compliance challenges for EGUs even if the unlikely scenario described by commenters were to occur. Several factors influence this. First, the change in methodology to preset budgets serving as a floor on budgets through 2029 means that the dynamic budget methodology can only produce an increase in the budget from this final rule through that year. Second, the adoption of a multi-year approach for identifying the heat input used to calculate the dynamic budgets will smooth the year-to-year budget changes and effectively eliminate the possibility of greatest concern, which was that a single year of unusually low heat input would be used to set the budget for a subsequent year that turned out to have unusually high heat input. While a year of unusually high heat input for a given state may still occur, the state's budgets for those years will never be based on heat input from an anomalously low year, but instead will always be based on an average of several years' heat input. Third, because the Group 3 trading program is an interstate program implemented over a wide geographic region, and it is unlikely that all regions of the country would uniformly experience a marked increase in fossil fuel heat input necessitating an additional supply of allowances, it is likely that allowances will be available for trade from one area of the country where there is less demand to another area where there is greater demand. Fourth, as explained in section VI.B.5 of this document, each state's assurance level will adjust to reflect actual heat input in that year. Specifically, the EPA will determine each state's variability limit for a given control period so that the percentage value used will 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. Thus, if in year 2030, for example, a state's actual heat input levels increase to a level that is not reflected in the dynamic budget calculation using earlier years of data, the assurance level (which absent the unusually high heat input would be 121 percent of the state's budget) will be calculated by the EPA following the 2030 ozone season, using that higher reported heat input. This will avoid imposing a three-for-one allowance surrender penalty on sources except where emissions exceed the assurance level even factoring in the increase in heat input in that year. Finally, as some commenters observed, the inherent data lag in dynamic budget quantification means that a state budget for the year 2030 will continue to reflect emissions from any EGU that retires before the 2030 control period but is still operating anytime during the 2026–2028 reference years from which the 2030 dynamic budget will be calculated. Given the likely ongoing trend of relatively high-emitting EGU retirements over time, this method for determining dynamic budgets should further assist the ability of remaining EGUs to obtain sufficient allowances to cover future heat input levels.

With respect to the comments expressing concern that dynamic budgets would create too much slack because of the lag in incorporating retirements, the EPA observes that dynamic budgets will yield a closer representation of Step 3 control stringency across the future fleet than preset budgets for years in which retirement plans are currently relatively unknown. Moreover, any risk that the lag would lead to an unacceptably large surplus of allowances is limited by EPA's finalization of the annual bank recalibration to 21 percent and 10.5 percent of the budget beginning in 2024 and 2030 respectively. The corresponding risk that a lag will lead sources to not operate emissions controls, due to a surplus of allowances, is also limited by the backstop daily emissions rates that start in 2024 (for sources with existing SCR controls) and no later than 2030 for other coal-fired sources.

Comment: Commenters allege that the dynamic budget methodology is effectively a “one-way ratchet” because, if EGUs pursue compliance strategies such as reduced utilization or generation shifting to comply with the rule rather than install or optimize pollution controls pursuant to the identified Step 3 emissions control strategies, the effect will be that the dynamic budget calculated in a future year will reflect that reduced heat input, but the applied emissions rate assumption will be the same. Thus, the approach according to commenters actually “punishes” sources for achievement of emissions reductions commensurate with EPA's Step 3 determinations through alternative compliance means, by producing a smaller budget in later years (less heat input multiplied by the same emissions rate). If the source again reduces utilization or shifts generation to comply with this budget, then budgets in later years will again ratchet down, and so on.

Response: First, the claims of dynamic budgeting being a one-way ratchet are incorrect. As pointed out at proposal, the dynamic budget process would allow for increased utilization to result in increased budgets. Moreover, this concern is entirely mooted for the period 2026 through 2029 with the shift to preset budgets serving as a floor; dynamic budgeting can only increase the budget used in any given year in this time period. Additionally, the use of a multi-year average heat input in the budget-setting calculations will, on the whole, modulate the dynamic budgets such that the budgets over time will only gradually change with changes in the operating profile of the EGU fleet.

For the control periods 2030 and later, this rule is premised on the expectation that all large coal-fired EGU sources identified for SCR-retrofit potential will, if they continue operating in 2030 or later, have installed the requisite post-combustion controls. Thus, the backstop daily emissions rate applies for all such sources beginning in the 2030 ozone season. In this latter period (post-2030), the EPA disagrees that the dynamic budget will punish fleet segments seeking to continue to pursue a strategy of reduced utilization. Rather, the dynamic budget will simply continue to reflect the Step 3 emissions control stringency. For instance, if there are two otherwise high-emitting sources in a state that can reduce emissions by operating SCR, this rule's control stringency finds it cost effective for both sources to operate their controls. If one source retires and is replaced by new lower-emitting generation, it is not a punishment to have the budgets adjust in a way that still incentivize remaining units to operate their controls. This is simply right-sizing the budget to an evolving fleet. It is a feature of the rule, not a flaw, and is designed to address observed instances in prior rules where market-driven reduced utilization resulted in non-binding ( i.e., overly slack) budgets and corresponding conditions where the incentive to operate a control dissipated over time. In the event that sources reduce utilization whether for compliance purposes or market-driven reasons, that also does not obviate the importance of continuing to incentivize the Step 3 emissions control stringency at identified sources.

c. Final Preset State Emissions Budgets

For affected EGUs in each covered state (and Indian country within the state's borders), this final rule establishes preset budgets for the control periods 2023 through 2029. For control periods 2026 through 2029, any of those preset budgets may be supplanted by the corresponding dynamic budget that will be tabulated at later date, if and only if that dynamic budget yields a higher amount. For 2030 and beyond, the dynamic budget formula promulgated in this rule will be applied to future year data to quantify state emissions budgets for those control periods. The procedures for allocating the allowances from each state budget among the units in each state (and Indian country within the state's borders) are described in section VI.B.9 of this document. The amounts of the final preset state emissions budgets for the 2023 through 2029 control periods are shown in Table VI.B.4.c–1.