^EDSrX
I Q \
Regulatory Impact Analysis
of the Supplemental Proposal for the Standards
of Performance for New, Reconstructed, and
Modified Sources and Emissions Guidelines for
Existing Sources: Oil and Natural Gas Sector
Climate Review
-------
-------
EPA-452/R-22-006
November 2022
Regulatory Impact Analysis of the Supplemental Proposal for the Standards of Performance for
New, Reconstructed, and Modified Sources and Emissions Guidelines for Existing Sources: Oil
and Natural Gas Sector Climate Review
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Research Triangle Park, NC
-------
CONTACT INFORMATION
This document has been prepared by staff from the Office of Air and Radiation, U.S.
Environmental Protection Agency. Questions related to this document should be addressed to the
Air Economics Group in the Office of Air Quality Planning and Standards (email:
OAQPSeconomics@epa.gov).
ACKNOWLEDGEMENTS
In addition to U.S. EPA staff from the Office of Air and Radiation, personnel from SC&A
contributed data and analysis to this document.
iv
-------
TABLE OF CONTENTS
Table of Contents v
List of Tables vii
List of Figures ix
1 Executive summary 1
1.1 Introduction 1
1.2 Legal and economic basis for this rulemaking 2
1.2.1 Statutory Requirements 2
1.2.2 Market Failure 4
1.3 Baseline and Regulatory Requirements 4
1.4 Summary of Key Results 10
1.5 Organization of RIA 14
2 Projected Compliance Costs and Emissions Reductions 16
2.1 Emissions Sources and Regulatory Requirements Analyzed in this RIA 16
2.1.1 Emissions Sources 16
2.1.2 Regulatory Requirements 22
2.2 Methodology 26
2.2.1 Activity Data Projections 28
2.2.2 Model Plant Compliance Cost and Emissions Reductions 43
2.2.3 State Programs 47
2.3 Emissions Reductions 48
2.4 Product Recovery 49
2.5 Compliance Costs 51
2.6 Comparison of Regulatory Alternatives 55
3 Benefits 60
3.1 Emissions Reductions 63
3.2 Methane Climate Effects and Valuation 64
3.3 Ozone-Related Impacts Due to VOC Emissions 79
3.3.1 Ozone Health Effects 80
3.3.2 Ozone Vegetation Effects 80
3.3.3 Ozone Climate Effects 80
3.4 Ozone-Related Impacts Due to Methane 81
3.5 PM2.5-Related Impacts Due to VOC Emissions 81
3.5.1 PM2 5 Health Effects 82
3.5.2 PM Welfare Effects 83
3.6 Hazardous Air Pollutants (HAP) Impacts 83
3.6.1 Benzene 85
3.6.2 Formaldehyde 85
3.6.3 Toluene 86
3.6.4 Carbonyl Sulfide 87
3.6.5 Ethylbenzene 87
3.6.6 Mixed Xylenes 88
3.6.7 n-Hexane 88
3.6.8 Other Air Toxics 89
3.7 Secondary Air Emissions Impacts 89
3.8 Total Benefits 92
4 Economic Impact and Distributional Analysis 96
4.1 Oil and Natural Gas Market Impact Analysis 96
v
-------
4.1.1 Crude Oil Market Model 97
4.1.2 Natural Gas Market Model 98
4.1.3 Assumptions, Data, and Parameters Used in the Oil and Natural Gas Market Models 99
4.1.4 Results 101
4.1.5 Caveats and Limitations of the Market Analysis 103
4.2 Environmental Justice Analyses 104
4.2.1 Analyzing EJ Impacts in This Supplemental Proposal 106
4.2.2 Climate Impacts 107
4.2.3 Ozone from Oil and Natural Gas VOC Emission Impacts 112
4.2.4 Air Toxics Impacts 118
4.2.5 Demographic Characteristics of Oil and Natural Gas Workers and Communities 125
4.2.6 Household Energy Expenditures 132
4.2.7 Summary 135
4.3 Initial Regulatory Flexibility Analysis 136
4.3.1 Reasons Why Action is Being Considered 136
4.3.2 Statement of Objectives and Legal Basis for Proposed Rules 137
4.3.3 Description and Estimate of Affected Small Entities 139
4.3.4 Compliance Cost Impact Estimates 141
4.3.5 Caveats and Limitations 145
4.3.6 Projected Reporting, Recordkeeping and Other Compliance Requirements 146
4.3.7 Related Federal Rules 147
4.3.8 Regulatory Flexibility Alternatives 148
4.4 Employment Impacts of Environmental Regulation 156
5 Comparison of Benefits and Costs 158
5.1 Comparison of Benefits and Costs 158
5.2 Uncertainties and Limitations 165
6 References 171
APPENDIX A Additional Information on Cost and Emissions Analysis 182
A. 1 Calculation of Equipment Bin Proportions and Average Equipment Factors for Well Sites
from 2016 ICR Data 182
A.2 Equipment Count Calibration at Well Sites 186
APPENDIX B Sensitivity Analysis of Monetized Climate Benefits 189
B. 1 Updated Estimates of the Social Cost of Methane 189
B .2 Results of the Climate Benefits Sensitivity Analysis 193
B.3 References 195
APPENDIX C Illustrative Screening Analysis of Monetized VOC-Related
Ozone Health Benefits 197
C. 1 Air Quality Modeling Simulations 197
C. 1.1 Ozone Model Performance 198
C. 1.2 Source Apportionment Modeling 201
C.2 Applying Modeling Outputs to Quantify a National VOC-Ozone Benefit Per-Ton Value 202
C.3 Uncertainties and Limitations of Air Quality Methodology 204
C.4 References 209
APPENDIX D Fugitive Emissions Abatement Simulation Toolkit (FEAST) Memo
212
vi
-------
LIST OF TABLES
Table 1-1 NSPS OOOOb Emissions Sources, Baseline Requirements, and Requirements under the Proposed
Option 6
Table 1-2 EG 0000c Emissions Sources, Baseline Requirements, and Requirements under the Proposed Option7
Table 1-3 Projected Emissions Reductions under the Proposed NSPS OOOOb and EG 0000c Option, 2023-
2035: 11
Table 1-4 Projected Benefits, Compliance Costs, and Emissions Reductions for the Proposed NSPS OOOOb
Option, 2023-2035 (million 2019$) 12
Table 1-5 Projected Benefits, Compliance Costs, and Emissions Reductions for the Proposed EG 0000c Option,
2023-2035 (million 2019$) 13
Table 1-6 Projected Benefits, Compliance Costs, and Emissions Reductions for the Proposed NSPS OOOOb and
EG 0000c Option, 2023-2035 (million 2019$) 14
Table 2-1 NSPS OOOOb Emissions Sources, Baseline Requirements, and Requirements under the Proposed
Option 22
Table 2-2 EG 0000c Emissions Sources, Baseline Requirements, and Requirements under the Proposed Option
24
Table 2-3 Assumed Retirement Rates and Annual New Site Counts by Site Type 30
Table 2-4 Distribution of Well Sites in Equipment Bins 37
Table 2-5 Projection of Incrementally Impacted Affected Facilities under the Proposed NSPS OOOOb and EG
0000c Option, 2023 to 2035 42
Table 2-6 Projected Emissions Reductions under the Proposed NSPS OOOOb and EG 0000c Option, 2023-
2035 49
Table 2-7 Projected Increase in Natural Gas Recovery under the Proposed NSPS OOOOb and EG 0000c
Option, 2023-2035 50
Table 2-8 Projected Compliance Costs under the Proposed NSPS OOOOb and EG 0000c Option, 2023-2035
(millions 2019$) 52
Table 2-9 Undiscounted Projected Compliance Costs under the Proposed NSPS OOOOb and EG 0000c Option,
2023-2035 (millions 2019$) 54
Table 2-10 Discounted Projected Costs under the Proposed NSPS OOOOb and EG 0000c Option, 2023-2035
(millions 2019$) 55
Table 2-11 Summary of Regulatory Alternatives 57
Table 2-12 Comparison of Regulatory Alternatives in 2023, 2026, and 2035 for the Proposed NSPS OOOOb and
EG 0000c (millions 2019$) 59
Table 3-1 Climate and Human Health Effects of the Projected Emissions Reductions from this Proposal 61
Table 3-2 Projected Annual Reductions of Methane, VOC, and HAP Emissions under the Proposed NSPS
OOOOb and EG 0000c Option, 2023-2035 64
Table 3-3 Interim Estimates of the Social Cost of CH4, 2023-2035 (in 2019$ per metric ton CH4) 72
Table 3-4 Undiscounted Monetized Climate Benefits under the NSPS OOOOb and EG 0000c Option, 2023-
2035 (millions, 2019$) 77
Table 3-5 Discounted Monetized Climate Benefits under the Proposed NSPS OOOOb and EG 0000c Option,
2023-2035 (millions, 2019$) 78
Table 3-6 Top Annual HAP Emissions as Reported in 2017 NEI for Oil and Natural Gas Sources 84
Table 3-7 Increases in Secondary Air Pollutant Emissions, Vapor Combustion at Storage Vessels (short tons per
year) 89
Table 3-8 Comparison of PV and EAV of the Projected Benefits for the Proposed NSPS OOOOb and EG 0000c
across Regulatory Options, 2023-2035 (millions of 2019$) 93
Table 3-9 Comparison of PV and EAV of the Projected Benefits for the Proposed NSPS OOOOb across
Regulatory Options, 2023-2035 (millions of 2019$) 94
Table 3-10 Comparison of PV and EAV of the Projected Benefits for the Proposed EG 0000c Across Regulatory
Options, 2023-2035 (millions of 2019$) 95
Table 4-1 Parameters Used in Market Analysis 100
Table 4-2 Baseline Crude Oil and Natural Gas Production and Prices Used in Market Analysis 100
Table 4-3 Projected Regulatory Costs for the Proposed NSPS OOOOb and EG 0000c Option Applied in the
Market Analysis (millions 2019$) 101
Vll
-------
Table 4-4 Estimated Crude Oil Production and Prices Changes under the Proposed NSPS OOOOb and EG
0000c Option 102
Table 4-5 Estimated Natural Gas Production and Prices Changes under the Proposed NSPS OOOOb and EG
0000c Option 102
Table 4-6 Components of the Criteria Pollutant Environmental Justice Assessment 113
Table 4-7 Cancer Risk and Demographic Population Estimates for 2017 NEI Nonpoint Emissions 123
Table 4-8 Demographic Characteristics of Oil and Natural Gas Workers and Communities 128
Table 4-9 Demographic Characteristics of Oil and Natural Gas Communities by Oil and Natural Gas Intensity. 130
Table 4-10 Hispanic Population by Oil and Natural Gas Intensity 131
Table 4-11 Energy Expenditures by Quintiles of Income before Taxes, 2019 134
Table 4-12 SBA Size Standards by NAICS Code 140
Table 4-13 Counts and Estimated Percentages of Small Entities 141
Table 4-14 Summary Statistics for Revenues of Potentially Affected Entities 142
Table 4-15 Distribution of Estimated Compliance Costs across Segment and Firm Size Classes (2019$) 144
Table 4-16 Compliance Cost-to-Sales Ratios for Small Entities 145
Table 5-1 Projected Emissions Reductions under the Proposed NSPS OOOOb and EG 0000c across Regulatory
Options, 2023-2035 158
Table 5-2 Projected Benefits, Compliance Costs, and Emissions Reductions across Regulatory Options under the
Proposed NSPS OOOOb, 2023-2035 (million 2019$) 159
Table 5-3 Projected Benefits, Compliance Costs, and Emissions Reductions across Regulatory Options under the
Proposed EG 0000c, 2023-2035 (million 2019$) 160
Table 5-4 Projected Benefits, Compliance Costs, and Emissions Reductions across Regulatory Options under the
Proposed NSPS OOOOb and EG 0000c, 2023-2035 (million 2019$) 161
Table 5-5 Projected Emissions Reductions, Climate Benefits, and Compliance Costs (millions 2019$) for
Incrementally Affected Sources under the Proposed NSPS OOOOb and EG 0000c Option, 2023 to
2035 164
Table A-l Well Site Equipment/Tank Category Proportions Estimated From the 2016 ICR 184
Table A-2 Per-Well Average Equipment/Tank Counts Estimated From the 2016 ICR 185
Table B-l Updated Estimates of the Social Cost of CH4, 2023-2035 (in 2019$ per metric ton CH4) 192
Table B-2 Undiscounted Monetized Climate Benefits Using Updated SC-CH4 Estimates under the NSPS OOOOb
and EG 0000c Option, 2023-2035 (millions, 2019S) 193
Table B-3 Discounted Monetized Climate Benefits Using Updated SC-CH4 Estimates under the Proposed NSPS
OOOOb and EG 0000c Option, 2023-2035 (millions, 2019S) 194
Table C-l Summary of 2017 CAMx MDA8 ozone model performance for all April-September days 200
Table C-2 Estimated Avoided Ozone-Related Premature Respiratory Mortality and Illnesses for the Proposed
NSPS OOOOb and EG 0000c Option in 2026: 207
Table C-3 Benefit Per Ton Estimates of Ozone-Attributable Premature Mortality and Illnesses for the Proposal in
2026 207
Table C-4 Estimated Discounted Economic Value of Ozone-Attributable Premature Mortality and Illnesses under
the Proposed NSPS OOOOb and EG 0000c Option, 2023-2035 (million 2019$)a d 207
Table C-5 Stream of Human Health Benefits under the Proposed NSPS OOOOb and EG 0000c Option, 2023-
2035: Monetized Benefits Quantified as Sum of Avoided Morbidity Health Effects and Avoided Long-
term Ozone Mortality (discounted at 3 percent to 2021; million 2019$)a b 208
Table C-6 Stream of Human Health Benefits under the Proposed NSPS OOOOb and EG 0000c Option, 2023-
2035: Monetized Benefits Quantified as Sum of Avoided Morbidity Health Effects and Avoided Long-
term Ozone Mortality (discounted at 7 percent to 2021; million 2019$)a b 209
-------
LIST OF FIGURES
Figure 2-1 Projections of Cumulative Site Counts by Site Type and Vintage 31
Figure 3-1 Frequency Distribution of SC-CH4 Estimates for 2030 74
Figure 4-1 Map of Baseline Ozone Concentrations from Oil and Natural Gas VOC Emissions in 2017 114
Figure 4-2 Average Ozone Concentrations from Oil and Natural Gas VOC Emissions by Population and
Corresponding 2017 Population Counts 115
Figure 4-3 Distributions of Ozone from Oil and Natural Gas VOC Emissions Concentrations by Race/Ethnicity 116
Figure 4-4 Distributions of Ozone from Oil and Natural Gas VOC Emissions Concentrations by Age Range 117
Figure 4-5 Distributions of Ozone from Oil and Natural Gas VOC Emissions Concentrations by Sex 117
Figure 4-6 National Map of Grid Cell Median Cancer Risks for 2017 Nonpoint Oil and Natural Gas NEI Emissions
124
Figure 4-7 Local-Scale Map of Grid Cell Median Cancer Risks for 2017 Nonpoint Oil and Natural Gas NEI
Emissions 125
Figure 4-8 National-level Employment in Oil and Natural Gas Production 126
Figure 4-9 Continental U.S. Map of PUMAs and Oil and Natural Gas Intensive Communities 129
Figure 4-10 Map of Oil and Natural Gas Intensive Communities of Environmental Justice Note 132
Figure C-1 Air Quality Modeling Domain 198
Figure C-2 Climate Regions Used to Summarize 2017 CAMx Model Performance for Ozone 200
Figure C-3 Map of 2017 CAMx MDA8 Normalized Mean Bias (%) for April-September at all U.S. monitoring
sites in the model domain 201
Figure C-4 Contributions of 2017 Oil and Natural Gas VOC Emissions across the Contiguous U.S. to the April-
September Average of MDA8 Ozone 202
IX
-------
1 EXECUTIVE SUMMARY
1.1 Introduction
This document presents the regulatory impact analyses (RIA) for the supplemental notice
of proposed rulemaking (hereafter, "supplemental proposal") titled "Standards of Performance
for New, Reconstructed, and Modified Sources and Emissions Guidelines for Existing Sources:
Oil and Natural Gas Sector Climate Review." The supplemental proposal builds on the proposed
rule with the same title published in November 2021 (hereafter, "November 2021 proposal"),1
providing additions, amendments, and clarification to the November 2021 proposal. This RIA for
the supplemental notice projects the potential impacts of these proposed actions cumulatively,
including provisions from the November 2021 proposal that have not been updated in the
supplemental proposal.
The November 2021 proposal included three distinct actions. First, it proposed to amend
existing crude oil and natural gas new source performance standards (NSPS) under the Clean Air
Act (CAA) section 111(b); second, it proposed new NSPS for the crude oil and natural gas
source category; and third, it proposed emissions guidelines (EG) under CAA section 111(d)
which will inform states on the development, submittal, and implementation of state plans to
establish performance standards for existing crude oil and natural gas sources. Both the
November 2021 and supplemental proposals respond to the President's Executive Order (EO)
13990, "Protecting Public Health and the Environment and Restoring Science to Tackle the
Climate Crisis".2
A wide range of stakeholders as well as state and tribal governments submitted public
comments on the November 2021 proposal, submitting over 470,000 public comments in total.
Many commenters representing diverse perspectives expressed general support for the proposal
and requested that the EPA further strengthen the proposed standards and make them more
comprehensive. Other commenters highlighted implementation or cost concerns related to
elements of the November 2021 proposal or provided specific data and information that the EPA
1 86 FR 63110
2 86 FR 7037
1
-------
was able to use to refine or revise several of the standards included in the November 2021
proposal.
The purpose of the supplemental proposed rulemaking is to strengthen, update, and
expand the proposed standards for certain emissions sources, including: (1) to reduce emissions
from the source category more comprehensively by adding proposed standards for certain
sources that were not addressed in the November 2021 proposal, revising the proposed
requirements for fugitive emissions monitoring and repair, and establishing a super-emitter
response program; (2) to encourage the deployment of innovative technologies and techniques
for detecting and reducing methane emissions by providing additional options for the use of
advanced monitoring; (3) to modify and refine certain elements of the proposed standards in
response to concerns and information submitted in public comments; and (4) to provide
additional information not included in the November 2021 proposal for public comment, such as
the proposed regulatory text for the new subparts and details of the timelines and other
requirements that apply to states as they develop state plans to implement the emission
guidelines.
1.2 Legal and economic basis for this rulemaking
In this section, we summarize the statutory requirements in the Clean Air Act that serve
as the legal basis for the proposed rule and the economic theory that supports environmental
regulation as a mechanism to enhance social welfare. The Clean Air Act requires the EPA to
prescribe regulations for new and existing sources. In turn, those regulations attempt to address
negative externalities created when private entities fail to internalize the social costs of air
pollution.
1.2.1 Statutory Requirements
Clean Air Act section 111, which Congress enacted as part of the 1970 Clean Air Act
Amendments, establishes mechanisms for controlling emissions of air pollutants from stationary
sources. This provision requires the EPA to promulgate a list of categories of stationary sources
that the Administrator, in his or her judgment, finds "causes, or contributes significantly to, air
pollution which may reasonably be anticipated to endanger public health or welfare." The EPA
has listed more than 60 stationary source categories under this provision. Once the EPA lists a
2
-------
source category, the EPA must, under CAA section 111(b)(1)(B), establish "standards of
performance" for emissions of air pollutants from new sources in the source categories. Under
section 111(b), EPA identifies the "best system of emission reduction" (BSER) that has been
adequately demonstrated to control emissions of a particular pollutant from a particular type of
source and sets a standard for new sources based on the application of that BSER. These
standards are known as new source performance standards (NSPS), and they are national
requirements that apply directly to the sources subject to them.
When the EPA establishes NSPS for sources in a source category under CAA section
111(b), the EPA is also required, under CAA section 111(d)(1), to prescribe regulations for states
to submit plans regulating existing sources in that source category for any air pollutant that, in
general, is not regulated under the CAA section 109 requirements for the NAAQS or regulated
under the CAA section 112 requirements for hazardous air pollutants (HAP). CAA section
111(d)'s mechanism for regulating existing sources differs from the one that CAA section 111(b)
provides for new sources because CAA section 111(d) contemplates states submitting plans that
establish "standards of performance" for the affected sources and contain other measures to
implement and enforce those standards.
"Standards of performance" are defined under CAA section 111(a)(1) as standards for
emissions that reflect the emission limitation achievable from the "best system of emission
reduction," considering costs and other factors, that "the Administrator determines has been
adequately demonstrated." Under section 111(d), EPA determines the BSER, but, unlike with
new sources under 111(b), the states are the entities that establish performance standards. CAA
section 111(d)(1) grants states the authority, in applying a standard of performance, to take into
account the source's remaining useful life and other factors.
Under CAA section 111(d), a state must submit its plan to the EPA for approval, and the
EPA must approve the state plan if it is "satisfactory." If a state does not submit a plan, or if the
EPA does not approve a state's plan, then the EPA must establish a plan for that state. Once a
state receives the EPA's approval of its plan, the provisions in the plan become federally
enforceable against the entity responsible for noncompliance, in the same manner as the
provisions of an approved State Implementation Plan (SIP) under the Act.
3
-------
1.2.2 Market Failure
Many regulations are promulgated to correct market failures, which otherwise lead to a
suboptimal allocation of resources within the free market. Air quality and pollution control
regulations address "negative externalities" whereby the market does not internalize the full
opportunity cost of production borne by society as public goods such as air quality are unpriced.
While recognizing that optimal social level of pollution may not be zero, methane and
VOC emissions impose costs on society, such as negative health and welfare impacts, that are
not reflected in the market price of the goods produced through the polluting process. For the
proposed regulatory action analyzed in this RIA, the goods produced are crude oil and natural
gas. If crude oil and natural gas producers pollute the atmosphere when extracting, processing,
and transporting products, the social costs will not be borne exclusively by the polluting firm but
rather by society as a whole. Thus, the producer is imposing a negative externality, or a social
cost of emissions, on society. The equilibrium market price of crude oil and natural gas may fail
to incorporate the full opportunity cost to society of these products. Consequently, absent a
regulation on emissions, producers will not internalize the social cost of emissions and social
costs will be higher as a result. The proposed regulation will work towards addressing this
market failure by causing affected producers to begin internalizing the negative externality
associated with methane and VOC emissions.
1.3 Baseline and Regulatory Requirements
The impacts of proposed regulatory actions are evaluated relative to a baseline that
represents the world without the proposed action. In this case, we present results for the proposed
NSPS OOOOb and EG 0000c, taking into account both the November 2021 and supplemental
proposals. In other words, this analysis reflects the totality of the two proposals compared to a
baseline without either regulatory action. As in the RIA for the November 2021 proposal, the
baseline for the supplemental proposal incorporates changes to regulatory requirements induced
by the Congressional Review Act (CRA) resolution that disapproved the 2020 Policy Rule.
Throughout this document, we focus the analysis on the proposed requirements that result in
quantifiable compliance cost or emissions changes compared to the baseline. We do not analyze
the regulatory impacts of all proposed requirements because we lack sufficient data, require
additional work to adapt existing data into a coherent analysis framework, or believe the
4
-------
provisions would not result in compliance cost or emissions impacts; see Section 2.1.2 for a
discussion of provisions for which impacts were not quantified.
Compared to the analysis presented in the RIA for the November 2021 proposal, this
analysis reflects changes in the proposed regulation for some sources; new methodologies to
estimate and project the universe of affected facilities and their emissions profiles, as well as the
cost and emissions impacts of applying control strategies; and updated assumptions based on
new information on existing and projected source counts, emissions factors and control costs,
natural gas prices, and state and local regulations that have been promulgated. The updated
baseline represents the EPA's most recent assessment of the current and future state of the
industry absent the proposed requirements.
Table 1-1 and Table 1-2 summarize the baseline and proposed standards of performance
for the sources with impacts quantified in this RIA.3 In Table 1-2, requirements in the baseline
differ depending on when sources were constructed relative to previous NSPS proposal dates.
We define pre- and post-KKK as having construction dates prior to and after January 20, 1984,
respectively. The dividing dates for pre- and post-0000 and pre- and post-OOOOa are August
23, 2011 and September 18, 2015, respectively. The abbreviations used in the table are OGI
(optical gas imaging), AVO (auditory, visual, and optical), scfh (standard cubic feet per hour),
and scfm (standard cubic feet per minute).
3 See the preamble for a more comprehensive description of the proposed standards.
5
-------
Table 1-1 NSPS OOOOb Emissions Sources, Baseline Requirements, and
Requirements under the Proposed Option
Standards of Performance
Source
In the Baseline
Under the Proposal
Fugitive Emissions/Equipment Leaks3
Well Sites
Wellhead only, single well site
No requirement
Quarterly AVO monitoring
Wellhead only, multiple well site
No requirement
Quarterly AVO monitoring +
Semiannual OGI
Single well site with a single price of
major equipment and no tank battery
Semiannual OGI
Quarterly AVO monitoring
Multiple well site with a single piece
of major equipment, or any site with
two or more pieces of major
equipment or one piece of major
equipment and a tank battery
Semiannual OGI
Bimonthly AVO monitoring +
Quarterly OGI
Gathering and Boosting Stations
Transmission and Storage Compressor
Stations
Quarterly OGI
Monthly AVO monitoring +
Quarterly OGI
Natural Gas Processing Plants
NSPS Subpart Wa
Bimonthly OGI
Pneumatic Pumps
Well Sites
95% control
Zero emissions
Gathering and Boosting Stations
No requirement
Pneumatic Controllers'"
Well Sites
Gathering and Boosting Stations
Natural gas bleed rate no greater
Transmission and Storage Compressor
Stations
than 6 scfh
Zero emissions0
Natural Gas Processing Plants
Zero emissions
Reciprocating Compressors
Gathering and Boosting Stations
Natural Gas Processing Plants
Transmission and Storage Compressor
Stations
Rod-packing changeout on fixed
schedule
Volumetric flow rate of 2 scfm
Centrifugal Compressors
Wet-seal
Gathering and Boosting Stations
No requirement
Natural Gas Processing Plants
Transmission and Storage
95% control
95% control
Compressor Stations
Dry-seal
Gathering and Boosting Stations
Natural Gas Processing Plants
Transmission and Storage
No requirement
Volumetric flow rate of 3 scfm
Compressor Stations
6
-------
Liquids Unloading
Well Sites
No requirement
Zero emissions or best
management practices'1
Storage Vessels
PTE > 6 tpy VOC
95% control, affected facility is
the tank
95% control, affected facility is
the tank battery
PTE < 6 tpy VOC
No requirement
No requirement
a Well sites and compressor stations on the Alaska North Slope are subject to Annual OGI monitoring only.
b Specifically, the affected source is natural gas-driven controllers that vent to the atmosphere.
0 The zero emissions rate standard does not apply to pneumatic controllers at sites in Alaska for which on site power
is not available. Instead natural gas-driven continuous bleed controllers at those sites are required to achieve bleed
rates at or below 6 scfh, while natural gas-driven intermittent bleed controllers are subject to OGI monitoring and
repair of emissions from controller malfunctions.
d The proposed regulation requires liquids unloading events to be zero-emitting unless technical infeasibilities exist,
in which case the regulation requires that best management practices be adopted.
Table 1-2 EG OOOOc Emissions Sources, Baseline Requirements, and Requirements
under the Proposed Option
Presumptive Standards of Performance
Source
In the Baseline
Under the Proposal
Fugitive Emissions/Equipment Leaks"
Well Sites
Wellhead only, single well site
Quarterly AVO monitoring
Wellhead only, multiple well site
No requirement
Quarterly AVO monitoring +
Semiannual OGI
Single well site with a single price of
major equipment and no tank battery
Pre-OOOOa: No requirement
Post-OOOOa: Semiannual OGI
Quarterly AVO monitoring
Multiple well site with a single piece
of major equipment, or any site with
two or more pieces of major
equipment or one piece of major
equipment and a tank battery
Pre-OOOOa: No requirement
Post-OOOOa: Semiannual OGI
Bimonthly AVO monitoring +
Quarterly OGI
Gathering and Boosting Stations
Transmission and Storage Compressor
Stations
Pre-OOOOa: No requirement
Post-OOOOa: Quarterly OGI
Monthly AVO monitoring +
Quarterly OGI
Natural Gas Processing Plants
Pre-KKK: No requirement
Post-KKK and Pre-OOOO:
NSPS Subpart W
Post-OOOO: NSPS Subpart Wa
Bimonthly OGI
Pneumatic Pumps
Well Sites
Pre-OOOOa: No requirement
Post-OOOOa: 95% control
Methane emission rate of zero
Gathering and Boosting Stations
No requirement
7
-------
Pneumatic Controllers'"
Well Sites
Gathering and Boosting Stations
Pre-OOOO: No requirement
Post-OOOO: Natural gas bleed
rate no greater than 6 scfh
Methane emission rate of zero0
Transmission and Storage Compressor
Stations
Pre-OOOOa: No requirement
Post-OOOOa: Natural gas bleed
rate no greater than 6 scfh
Natural Gas Processing Plants
Pre-OOOO: No requirement
Post-OOOO: Zero emissions
Methane emission rate of zero
Reciprocating Compressors
Gathering and Boosting Stations
Natural Gas Processing Plants
Pre-OOOO: No requirement
Post-OOOO: Rod-packing
changeout on fixed schedule
Volumetric flow rate of 2 scfm
Transmission and Storage Compressor
Stations
Pre-OOOOa: No requirement
Post-OOOOa: Rod-packing
changeout on fixed schedule
Centrifugal Compressors
Wet-seal
Gathering and Boosting Stations
No requirement
Natural Gas Processing Plants
Transmission and Storage
Compressor Stations
Pre-OOOO: No requirement
Post-OOOO: 95% control
Volumetric flow rate of 3 scfm
Dry-seal
Gathering and Boosting Stations
Natural Gas Processing Plants
Transmission and Storage
No requirement
Volumetric flow rate of 3 scfm
Compressor Stations
Liquids Unloading
Well Sites
No requirement
Zero emissions or best
management practices'1
Storage Vessels
PTE > 20 tpy CH4
Pre-OOOO: No requirement
Post-OOOO: 95% control,
95% control, affected facility is
the tank battery
PTE < 20 tpy CH4 and > 6 tpy VOC
affected facility is the tank
No requirement
PTE < 20 tpy CH4 and < 6 tpy VOC
No requirement
a Well sites and compressor stations on the Alaska North Slope are subject to Annual OGI monitoring only.
b Specifically, the affected source is natural gas-driven controllers that vent to the atmosphere.
0 The zero emissions rate standard does not apply to pneumatic controllers at sites in Alaska for which on site power
is not available. Instead natural gas-driven continuous bleed controllers at those sites are required to achieve bleed
rates at or below 6 scfh, while natural gas-driven intermittent bleed controllers are subject to OGI monitoring and
repair of emissions from controller malfunctions.
d The proposed regulation requires liquids unloading events to be zero-emitting unless technical infeasibilities exist,
in which case the regulation requires that best management practices be adopted.
The net benefits analysis summarized in this RIA reflects a nationwide engineering
analysis of compliance cost and emissions reductions, of which there are two main components:
activity data and information on control measures. The activity data represents estimates of the
counts of affected facilities over time, and the control measure information includes data on costs
8
-------
and control efficiencies for typical facilities. Both components are described briefly below, with
more detailed information provided in Section 2.2.
The first component is activity data for a set of representative or model plants for each
regulated facility.4 To project activity data for regulated facilities, we first project activity data
for oil and gas sites, which include well sites, natural gas processing plants, and compressor
stations (gathering and boosting, transmission, and storage). Projections include addition of
newly constructed sites and retirement of previously constructed sites, with magnitudes based on
a combination of analysis of several data sources and, where necessary, assumptions. Using
representative "per-site" factors based on EPA's Greenhouse Gas Inventory (GHGI), regulated
facilities are apportioned to sites across all industry segments.5 We assume the per-site factors
are fixed over time, so that the projected counts of regulated facilities change in proportion to the
projected counts of sites.
The regulated facility projections are combined with information on control options,
including capital costs, annual operations and maintenance costs, and control efficiencies.
Information on control options is derived from the analysis underpinning the BSER
determinations. Impacts are calculated by setting parameters on how and when affected facilities
are assumed to respond to a regulatory regime, multiplying activity data by model plant cost and
emissions estimates, differencing from the baseline scenario, and then summing to the desired
level of aggregation. In addition to emissions reductions, some control options result in natural
gas recovery, which can then be combusted in production or sold. Where applicable, we present
projected compliance costs with and without the projected revenues from product recovery.
For the analysis, we calculate the cost and emissions impacts of the proposed NSPS
OOOOb and EG 0000c from 2023 to 2035. The initial analysis year is 2023 as we assume the
proposed rule will be finalized early in that year. The NSPS OOOOb is assumed to take effect
4 Regulated facilities include well site fugitives, gathering and boosting station fugitives, transmission and storage
compressor station fugitives, natural gas processing plant equipment leaks, pneumatic pumps, pneumatic controllers,
reciprocating compressors, centrifugal compressors, liquids unloading, and storage vessels.
5 Industry segments include production, gathering and boosting, processing, transmission, and storage.
9
-------
immediately and impact sources constructed after publication of the November 2021 proposal.6
We assume the EG OOOOc will take longer to go into effect as states will need to develop
implementation plans in response to the rule and have them approved by the Agency. We assume
that this process will take three years, and so EG OOOOc impacts will begin in 2026. The final
analysis year is 2035, which allows us to present ten years of regulatory impacts after state plans
under the EG OOOOc are assumed to take effect.
1.4 Summary of Key Results
A summary of the key results is shown below. All dollar estimates are in 2019 dollars.
Also, all compliance costs, emissions changes, and benefits are estimated for the years 2023 to
2035 relative to a baseline without the proposed NSPS OOOOb and EG OOOOc.
Table 1-3 summarizes the emissions reductions associated with the proposed standards
over the 2023 to 2035 period for the NSPS OOOOb, the EG OOOOc, and the NSPS OOOOb
and EG OOOOc combined. The emissions reductions are estimated by multiplying the source-
level emissions reductions associated with each applicable control and facility type by the
number of affected sources of that facility type. We present methane emissions in both short tons
and CO2 equivalents (CO2 Eq.) using a global warming potential (GWP) of 25.7
6 As explained in the preamble to supplemental proposal, NSPS OOOOb would apply to all emissions sources
("affected facilities") identified in the proposed 40 CFR 60.5365b, except dry seal centrifugal compressors, that
commenced construction, reconstruction, or modification after November 15, 2021. NSPS OOOOb would apply to
dry seal centrifugal compressor affected facilities that commence construction, reconstruction, or modification after
publication of the supplemental proposal in the Federal Register.
7 Global warming potential is a measure that allows comparisons of the global warming impacts of different
greenhouse gases. Specifically, it is a measure of how much energy the emission of 1 ton of a gas will absorb over a
given period of time, relative to the emission of 1 ton of carbon dioxide (C02).
10
-------
Table 1-3 Projected Emissions Reductions under the Proposed NSPS OOOOb and EG
OOOOc Option, 2023-2035a b
Emissions Changes
Proposal
Methane
(million short tons)
VOC
(million short tons)
HAP
(million short tons)
Methane
(million metric tons
CO2 Eq. using
GWP=25)
NSPS OOOOb
8.1
2.9
0.11
180
EG OOOOc
28
6.8
0.28
620
Total
36
9.7
0.39
810
a Numbers rounded to two significant digits unless otherwise noted. Totals may not appear to add correctly due to
rounding. To convert from short tons to metric tons, multiply the short tons by 0.907. Alternatively, to convert
metric tons to short tons, multiply metric tons by 1.102.
b The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
Table 1-4, Table 1-5, and Table 1-6 present results for the proposal for the NSPS
OOOOb, EG OOOOc, and NSPS OOOOb and EG OOOOc combined, respectively. Each table
presents the present value (PV) and equivalent annual value (EAV), estimated using discount
rates of 3 and 7 percent, of the changes in quantified benefits, costs, and net benefits, as well as
the emissions reductions relative to the baseline. These values reflect an analytical time horizon
of 2023 to 2035, are discounted to 2021, and presented in 2019 dollars. We present the total
compliance costs, the value of product recovery generated by the capture of natural gas, and the
net compliance costs, which treats the value of product recovery as an offset to the compliance
costs.8 The table includes consideration of the non-monetized benefits associated with the
emissions reductions projected under this proposal.
8 Under this proposal, over 80 percent of revenue from the sale of captured natural gas is projected to be earned by
operators in the production and processing segments of the industry, where we assume that the operators own the
natural gas and will receive the financial benefit from the captured natural gas. The remainder of the captured
natural gas is captured within the transmission and storage segment, where operators do not typically own the
natural gas they transport; rather, they receive payment for the transportation service they provide. In the RIA, we
treat these revenues as an offset to projected compliance costs, while the revenues may also be considered as a
benefit of the regulatory action. However, regardless of whether the revenue from capture of natural gas is
considered a compliance cost offset or a benefit, the net benefits are equivalent.
11
-------
Table 1-4 Projected Benefits, Compliance Costs, and Emissions Reductions for the
Proposed NSPS OOOOb Option, 2023-2035 (million 2019$)
3 Percent Discount Rate
PV
EAV
PV
EAV
Climate Benefits3
$11,000
$1,000
$11,000
$1,000
3 Percent Discount Rate
7 Percent Discount Rate
PV
EAV
PV
EAV
Net Compliance Costs
$3,300
$360
$3,000
$360
Compliance Costs
$4,400
$460
$3,700
$440
Value of Product Recovery
$1,000
$99
$730
$88
Net Benefits
$7,600
$670
$7,900
$670
Climate and ozone health benefits from reducing 8.1 million short
tons of methane from 2023 to 2035
PM2.5 and ozone health benefits from reducing 2.9 million short
tons of VOC from 2023 to 203 5b
Non-Monetized Benefits HAP benefits from reducing 110 thousand short tons of HAP from
2023 to 2035
Visibility benefits
Reduced vegetation effects
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
a Climate benefits are based on reductions in methane emissions and are calculated using four different estimates of
the social cost of methane (SC-CH4) (model average at 2.5 percent, 3 percent, and 5 percent discount rates; 95th
percentile at 3 percent discount rate). For the presentational purposes of this table, we show the benefits associated
with the average SC-CH4 at a 3 percent discount rate, but the Agency does not have a single central SC-CH4 point
estimate. We emphasize the importance and value of considering the benefits calculated using all four SC-CH4
estimates; the present value (and equivalent annual value) of the additional benefit estimates ranges from $4.4
billion to $29 billion ($470 million to $2.7 billion) over 2023 to 2035 for the proposed option. Please see Table 3-5
and Table 3-8 for the full range of SC-CH4 estimates. As discussed in Section 3 of the RIA, a consideration of
climate benefits calculated using discount rates below 3 percent, including 2 percent and lower, are also warranted
when discounting intergenerational impacts. Appendix B presents the results of a sensitivity analysis using a set of
SC-CH4 estimates that incorporates recent research addressing recommendations of the National Academies of
Sciences, Engineering, and Medicine (2017). All net benefits are calculated using climate benefits discounted at 3
percent.
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C of the RIA.
12
-------
Table 1-5 Projected Benefits, Compliance Costs, and Emissions Reductions for the
Proposed EG OOOOc Option, 2023-2035 (million 2019$)
3 Percent Discount Rate
PV
EAV
PV
EAV
Climate Benefits3
$37,000
$3,500
$37,000
$3,500
3 Percent Discount Rate
7 Percent Discount Rate
PV
EAV
PV
EAV
Net Compliance Costs
$11,000
$990
$8,700
$1,000
Compliance Costs
$14,000
$1,300
$11,000
$1,300
Value of Product Recovery
$3,600
$340
$2,500
$300
Net Benefits
$26,000
$2,500
$28,000
$2,400
Climate and ozone health benefits from reducing 28 million short
tons of methane from 2023 to 2035
PM2.5 and ozone health benefits from reducing 6.8 million short tons
of VOC from 2023 to 2035bc
Non-Monetized Benefits HAP benefits from reducing 280 thousand short tons of HAP from
2023 to 2035
Visibility benefits
Reduced vegetation effects
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
a Climate benefits are based on reductions in methane emissions and are calculated using four different estimates of
the social cost of methane (SC-CH4) (model average at 2.5 percent, 3 percent, and 5 percent discount rates; 95th
percentile at 3 percent discount rate). For the presentational purposes of this table, we show the benefits associated
with the average SC-CH4 at a 3 percent discount rate, but the Agency does not have a single central SC-CH4 point
estimate. We emphasize the importance and value of considering the benefits calculated using all four SC-CH4
estimates; the present value (and equivalent annual value) of the additional benefit estimates ranges from $ 15 billion
to $98 billion ($1.6 billion to $9.3 billion) over 2023 to 2035 for the proposed option. Please see Table 3-5 and
Table 3-8 for the full range of SC-CH4 estimates. As discussed in Section 3 of the RIA, a consideration of climate
benefits calculated using discount rates below 3 percent, including 2 percent and lower, are also warranted when
discounting intergenerational impacts. Appendix B presents the results of a sensitivity analysis using a set of SC-
CH4 estimates that incorporates recent research addressing recommendations of the National Academies of
Sciences, Engineering, and Medicine (2017). All net benefits are calculated using climate benefits discounted at 3
percent.
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C of the RIA.
0 The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
13
-------
Table 1-6 Projected Benefits, Compliance Costs, and Emissions Reductions for the
Proposed NSPS OOOOb and EG OOOOc Option, 2023-2035 (million 2019$)
3 Percent Discount Rate
PV
EAV
PV
EAV
Climate Benefits3
$48,000
$4,500
$48,000
$4,500
3 Percent Discount Rate
7 Percent Discount Rate
PV
EAV
PV
EAV
Net Compliance Costs
$14,000
$1,400
$12,000
$1,400
Compliance Costs
$19,000
$1,800
$15,000
$1,800
Value of Product Recovery
$4,600
$440
$3,300
$390
Net Benefits
$34,000
$3,200
$36,000
$3,100
Climate and ozone health benefits from reducing 36 million short
tons of methane from 2023 to 2035
PM2.5 and ozone health benefits from reducing 9.7 million short tons
of VOC from 2023 to 2035bc
Non-Monetized Benefits HAP benefits from reducing 390 thousand short tons of HAP from
2023 to 2035
Visibility benefits
Reduced vegetation effects
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
a Climate benefits are based on reductions in methane emissions and are calculated using four different estimates of
the social cost of methane (SC-CH4) (model average at 2.5 percent, 3 percent, and 5 percent discount rates; 95th
percentile at 3 percent discount rate). For the presentational purposes of this table, we show the benefits associated
with the average SC-CH4 at a 3 percent discount rate, but the Agency does not have a single central SC-CH4 point
estimate. We emphasize the importance and value of considering the benefits calculated using all four SC-CH4
estimates; the present value (and equivalent annual value) of the additional benefit estimates ranges from $ 19 billion
to $130 billion ($2.1 billion to $12 billion) over 2023 to 2035 for the proposed option. Please see Table 3-5 and
Table 3-8 for the full range of SC-CH4 estimates. As discussed in Section 3 of the RIA, a consideration of climate
benefits calculated using discount rates below 3 percent, including 2 percent and lower, are also warranted when
discounting intergenerational impacts. Appendix B presents the results of a sensitivity analysis using a set of SC-
CH4 estimates that incorporates recent research addressing recommendations of the National Academies of
Sciences, Engineering, and Medicine (2017). All net benefits are calculated using climate benefits discounted at 3
percent.
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C of the RIA.
0 The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
1.5 Organization of RIA
Section 2 describes the projected compliance cost and emissions impacts from the
proposal, including the PV and EAV of the projected costs over the 2023 to 2035 period and the
associated EAV. Section 3 describes the projected climate benefits resulting from this proposal,
including the PV and EAV of the projected climate benefits over the 2023 to 2035 period.
Section 3 additionally considers the potential beneficial climate, health, and welfare impacts that
could not be quantified. Section 4 describes the economic impact and distributional analysis
associated with the proposed rule. The economic impact and distributional analysis section
14
-------
includes analysis of oil and natural gas market impacts, environmental justice, small entities, and
employment. Section 5 compares the projected benefits and compliance cost reductions of this
action, as well as a summary of the net benefits with consideration of non-monetized benefits.
Section 5 also highlights uncertainties and limitations of the analysis. The RIA includes three
appendices, which provide further detail on the projection of affected sources (Appendix A), a
sensitivity analysis of the monetized climate benefits using newly developed SC-CH4 estimates
(Appendix B), and a screening analysis of monetized ozone benefits from VOC reductions
(Appendix C).
15
-------
2 PROJECTED COMPLIANCE COSTS AND EMISSIONS REDUCTIONS
In this section, we present estimates of the projected engineering compliance costs and
emissions reductions associated with the proposed rule for the 2023 to 2035 period. These
estimates are generated by combining model plant-level cost and emissions reductions based on
the BSER analysis with activity data projections based on a combination of historical trends and
third-party projections. The methods and assumptions used to construct the activity data
projections are also documented in this section.
2.1 Emissions Sources and Regulatory Requirements Analyzed in this RIA
A series of emissions sources and controls were evaluated as part of the proposed NSPS
OOOOb and EG OOOOc review. Section 2.1.1 provides a basic description of emissions sources
and the controls evaluated for each source to facilitate the reader's understanding of the
economic analysis. Section 2.1.2 describes the regulatory choices within the proposed NSPS
OOOOb and EG OOOOc that are examined in this RIA. Additional technical detail on the
engineering and cost basis of the analysis is available within the preamble, the Technical Support
Document (TSD) for the supplemental proposal, hereafter referred to as the Supplemental TSD
(U.S. EPA, 2022),9 and the TSD for the November 2021 proposal, hereafter referred to as the
November 2021 TSD (U.S. EPA, 202le).
2.1.1 Emissions Sources
The section provides brief descriptions of the emissions sources subject to the
requirements in the proposed NSPS OOOOb and EG OOOOc. EPA presents more detailed
modeling, assumptions and other crucial information, and additional technical detail in the
preamble, the Supplemental TSD and accompanying FEAST memo,10 and the November 2021
TSD.
Fugitive Emissions:11 There are several potential sources of fugitive emissions
throughout the crude oil and natural gas production source category. Fugitive emissions occur
9 Available at https://www.regulations.gov/ under Docket No. EPA-HQ-OAR-2021-0317.
10 Memorandum. Modeling Fugitive Emissions from Production Sites Using FEAST. Prepared by RTI International
for Karen Marsh, SPPD/OAQPS/EPA. July 27, 2022. Docket No. EPA-HQ-OAR-2021-0317.
11 See Chapter 5 of the Supplemental TSD and the FEAST memo for more information.
16
-------
when connection points are not fitted properly or when seals and gaskets start to deteriorate.
Changes in pressure and mechanical stresses can also cause components or equipment to emit
fugitive emissions. Poor maintenance or operating practices, such as improperly reseated
pressure relief valves (PRVs) or worn gaskets on thief hatches on controlled storage vessels are
also potential causes of fugitive emissions. Additional sources of fugitive emissions include
agitator seals, connectors, pump diaphragms, flanges, instruments, meters, open-ended lines
(OELs), pressure relief devices such as PRVs, pump seals, valves or controlled liquid storage
tanks. These fugitive emissions do not include devices that vent as part of normal operations,
such as natural gas-driven pneumatic controllers or natural gas-driven pneumatic pumps, insofar
as the natural gas discharged from the device's vent is not considered a fugitive emissions (e.g.,
an intermittent pneumatic controller that is venting continuously).
Pneumatic Controllers:12 Pneumatic controllers are devices used to regulate a variety of
physical parameters, or process variables, using air or gas pressure to control the operation of
mechanical devices, such as valves. The valves, in turn, control process conditions such as
levels, temperatures and pressures. When a pneumatic controller identifies the need to alter a
process condition, it will open or close a control valve. In many situations across all segments of
the oil and natural gas industry, pneumatic controllers make use of the available high-pressure
natural gas to operate or control the valve. In these "gas-driven" pneumatic controllers, natural
gas may be released with every valve movement and/or continuously from the valve control.
Pneumatic controllers can be categorized based on the emissions pattern of the controller.
Some controllers are designed to have the supply-gas provide the required pressure to power the
end-device, and the excess amount of gas is emitted. The emissions of this excess gas are
referred to as "bleed," and this bleed occurs continuously. Controllers that operate in this manner
are referred to as "continuous bleed" pneumatic controllers. These controllers can be further
categorized based on the amount of bleed they are designed to have. Those that have a bleed rate
of less than or equal to 6 standard cubic feet per hour (scfh) are referred to as "low bleed," and
those with a bleed rate of greater than 6 scfh are referred to as "high bleed." Another type of
controller is designed to release gas only when the process parameter needs to be adjusted by
opening or closing the valve, and there is no vent or bleed of gas to the atmosphere when the
12 See Chapter 3 of the Supplemental TSD for more information.
17
-------
valve is stationary. These types of controllers are referred to as "intermittent vent" pneumatic
controllers. A third type of controller releases gas to a downstream pipeline instead of the
atmosphere. These "closed loop" types of controllers can be used in applications with very low
pressure.
Not all pneumatic controllers are natural gas-driven. At sites with electricity, electrically
powered pneumatic devices or pneumatic controllers using compressed air can be used. As these
devices are not driven by pressurized natural gas, they do not emit any natural gas to the
atmosphere. At sites without electricity provided through the grid or on-site electricity
generation, solar power can be used in some instances.
Pneumatic Pumps:13 Most pneumatic pumps fall into two main types: diaphragm pumps,
generally used for heat tracing and plunger/piston pumps, generally used for chemical and
methanol injection. The pneumatic pump may use natural gas or another gas to drive the pump.
These pumps can also be electrically powered. "Non-natural gas-driven" pneumatic pumps can
be mechanically operated or use sources of power other than pressurized natural gas, such as
compressed "instrument air." Because these devices are not natural gas-driven, they do not
directly release natural gas or methane emissions. However, these systems have other energy
impacts, with associated secondary impacts related to generation of the electrical power required
to drive the instrument air compressor system. Instrument air systems are feasible only at oil and
natural gas locations where the devices can be driven by compressed instrument air systems and
have electrical service sufficient and reliable enough to power an air control system.
Reciprocating Compressors:14 In a reciprocating compressor, natural gas enters the
suction manifold, and then flows into a compression cylinder where it is compressed by a piston
driven in a reciprocating motion by the crankshaft powered by an internal combustion engine.
Emissions occur when natural gas leaks around the piston rod when pressurized natural gas is in
the cylinder. The compressor rod packing system consists of a series of flexible rings that create
a seal around the piston rod to prevent gas from escaping between the rod and the inboard
cylinder head. However, over time, during operation of the compressor, the rings become worn,
13 See Chapter 4 of the Supplemental TSD for more information.
14 See Chapter 7 of the November 2021 TSD for more information.
18
-------
and the packaging system needs to be replaced to prevent excessive leaking from the
compression cylinder.
Centrifugal Compressors:15 Centrifugal compressors use a rotating disk or impeller to
increase the velocity of the natural gas where it is directed to a divergent duct section that
converts the velocity energy to pressure energy. These compressors are primarily used for
continuous, stationary transport of natural gas in the processing and transmission systems. Some
centrifugal compressors use wet (meaning oil) seals around the rotating shaft to prevent natural
gas from escaping where the compressor shaft exits the compressor casing. The wet seals use oil
which is circulated at high pressure to form a barrier against compressed natural gas leakage. The
circulated oil entrains and adsorbs some compressed natural gas that may be released to the
atmosphere during the seal oil recirculation process. Off gassing of entrained natural gas from
wet seal centrifugal compressors is not suitable for sale and is either released to the atmosphere,
flared, or routed back to a process. Some centrifugal compressors utilize dry seal systems. Dry
seal systems minimize leakage by using the opposing force created by hydrodynamic grooves
and springs.
Storage vessels:16 Storage vessels, or storage tanks, in the oil and natural gas sector are
used to hold a variety of liquids, including crude oil, condensates, and produced water. Many
facilities operate a group of storage vessels, sometimes in series but most often in parallel, used
to store the same oil or condensate streams. This group of tanks used to store a common fluid is
typically called a tank battery.
Underground crude oil contains many light hydrocarbon gases in solution. When oil is
brought to the surface and processed, many of the dissolved lighter hydrocarbons are removed
through a series of high-pressure and low-pressure separators. The oil (or condensate or water)
from the separator is then directed to a tank battery where it is stored before being shipped off-
site. Some light hydrocarbon gases remain dissolved in the oil, condensate, or water because the
separator operates at pressures above atmospheric pressure. These dissolved hydrocarbon gases
are released from the liquid as vapors, commonly referred to as flash gas, when stored at
atmospheric pressures in the tank batteries. Typically, the larger the operating pressure of the
15 See Chapter 2 of the Supplemental TSD for more information.
16 See Chapter 6 of the November 2021 TSD for more information.
19
-------
separator, the more flash emissions will occur in the storage stage. Temperature of the liquid
may also influence the amount of flash emissions. Lighter crude oils and condensate generally
flash more hydrocarbons than heavier crude oils.
In addition to flash gas losses, other hydrocarbons may be emitted from the storage
vessels due to working and breathing (or standing) losses. Working losses occur when vapors are
displaced due to the emptying and filling of tank batteries. When the liquid level in the tank is
lowered, ambient air is drawn into the tank's headspace. Some hydrocarbons from the liquid will
volatilize into the headspace to reach equilibrium with the new headspace gas. When the liquid
level in the tank is increased, it will expel the saturated headspace gas into the atmosphere.
Breathing losses are the release of gas associated with daily temperature fluctuations when the
liquid level remains unchanged. As temperatures drop (or atmospheric pressure increases), gas in
the headspace contracts, drawing in ambient air. Again, hydrocarbons volatilize into this new gas
due to equilibrium effects. As the temperature rises (or atmospheric pressure falls), the gas in the
tank's headspace expands, expelling a portion of the hydrocarbon-saturated gas. Working losses
increase relative to the "turnover rate" (throughput rate divided by the tank capacity) and are
typically much greater than breathing losses.
Liquids Unloading:17 In new natural gas wells, there is generally sufficient reservoir
pressure/gas velocity to facilitate the flow of water and hydrocarbon liquids through the well
head and to the separator to the surface along with produced gas. In mature gas wells, the
accumulation of liquids in the wellbore can occur when the bottom well pressure/gas velocity
approaches the average reservoir pressure (i.e., volumetric average fluid pressure within the
reservoir across the areal extent of the reservoir boundaries). This accumulation of liquids can
impede and sometimes halt gas production. When the accumulation of liquid results in the
slowing or cessation of gas production (i.e., liquids loading), removal of fluids (i.e., liquids
unloading) is required to maintain production. These gas wells therefore often need to remove or
"unload" the accumulated liquids so that gas production is not inhibited.
The choice of what liquids unloading technique to employ is based on a well-by-well and
reservoir-by-reservoir analysis. To address the complex science and engineering considerations
17 See Chapter 11 of the November 2021 TSD for more information.
20
-------
to cover well unloading requirements, many differing technologies, techniques, and practices
have been developed to address an individual well's characteristics of the well to manage liquids
and maintain production of the well. At the onset of liquids loading, techniques that rely on the
reservoir energy are typically used. Eventually a well's reservoir energy is not sufficient to
remove the liquids from the well and it is necessary to add energy to the well to continue
production. Owners and operators can choose from several techniques to remove the liquids,
including manual unloading, velocity tubing or velocity strings, beam or rod pumps, electric
submergence pumps, intermittent unloading, gas lift (e.g., use of a plunger lift), foam agents and
wellhead compression. Each of these methods/procedures removes accumulated liquids and
thereby maintains or restores gas production. Although the unloading method employed by an
owner or operator can itself be a method that mitigates/eliminates venting of emissions from a
liquids unloading event, dictating a particular method to meet a particular well's unloading needs
is a production engineering decision.
Equipment Leaks at Gas Plants:18 The primary sources of equipment leak emissions
from natural gas processing plants are pumps, valves, and connectors. The major cause of
equipment leak emissions from valves and connectors is a seal or gasket failure due to normal
wear or improper maintenance. For pumps, emissions are often a result of a seal failure. The
large number of valves, pumps, and connectors at natural gas processing plants means emissions
from these components can be significant.
Common classifications of equipment at natural gas processing facilities include
components in VOC service and in non-VOC service. "In VOC service" is defined as a
component containing or in contact with a process fluid that is at least 10 percent VOC by
weight or a component "in wet gas service," which is a component containing or in contact with
field gas before extraction. "In non-VOC service" is defined as a component in methane service
(at least 10 percent methane) that is not also in VOC service.
The most common technique to reduce emissions from equipment leaks is to implement a
leak detection and repair (LDAR) program. Implementing an LDAR program can potentially
reduce product losses, increase safety for workers and operators, decrease exposure for the
18 See Chapter 10 of the November 2021 TSD for more information.
21
-------
surrounding community, reduce emissions fees, and help facilities avoid enforcement actions.
The effectiveness of an LDAR program is based on the frequency of monitoring, leak definition,
frequency of leaks, percentage of leaks that are repaired, and the percentage of reoccurring leaks.
2.1.2 Regulatory Requirements
Table 2-1 and Table 2-2 summarize the baseline and proposed standards of performance
for the sources with impacts quantified in this RIA.19 In Table 2-2, requirements in the baseline
differ depending on when sources were constructed relative to previous NSPS proposal dates.
We define pre- and post-KKK as dates prior to and after January 20, 1984, respectively. The
dividing dates for pre- and post-0000 and pre- and post-OOOOa are August 23, 2011 and
September 18, 2015, respectively. The abbreviations used in the table are OGI (optical gas
imaging), AVO (auditory, visual, and optical), scfh (standard cubic feet per hour), and scfm
(standard cubic feet per minute).
Table 2-1 NSPS OOOOb Emissions Sources, Baseline Requirements, and
Requirements under the Proposed Option
Standards of Performance
Source
In the Baseline
Under the Proposal
Fugitive Emissions/Equipment Leaks3
Well Sites
Wellhead only, single well site
No requirement
Quarterly AVO monitoring
Wellhead only, multiple well site
No requirement
Quarterly AVO monitoring +
Semiannual OGI
Single well site with a single price of
major equipment and no tank battery
Semiannual OGI
Quarterly AVO monitoring
Multiple well site with a single piece
of major equipment, or any site with
two or more pieces of major
equipment or one piece of major
equipment and a tank battery
Semiannual OGI
Bimonthly AVO monitoring +
Quarterly OGI
Gathering and Boosting Stations
Transmission and Storage Compressor
Stations
Quarterly OGI
Monthly AVO monitoring +
Quarterly OGI
Natural Gas Processing Plants
NSPS Subpart Wa
Bimonthly OGI
Pneumatic Pumps
Well Sites
95% control
Gathering and Boosting Stations
No requirement
Zero emissions
19 See the preamble for a more comprehensive description of the proposed standards.
22
-------
Pneumatic Controllers'"
Well Sites
Gathering and Boosting Stations
Natural gas bleed rate no greater
Transmission and Storage Compressor
Stations
than 6 scfh
Zero emissions0
Natural Gas Processing Plants
Zero emissions
Reciprocating Compressors
Gathering and Boosting Stations
Natural Gas Processing Plants
Transmission and Storage Compressor
Stations
Rod-packing changeout on fixed
schedule
Volumetric flow rate of 2 scfm
Centrifugal Compressors
Wet-seal
Gathering and Boosting Stations
No requirement
Natural Gas Processing Plants
Transmission and Storage
95% control
95% control
Compressor Stations
Dry-seal
Gathering and Boosting Stations
Natural Gas Processing Plants
Transmission and Storage
No requirement
Volumetric flow rate of 3 scfm
Compressor Stations
Liquids Unloading
Well Sites
No requirement
Zero emissions or best
management practices'1
Storage Vessels
PTE > 6 tpy VOC
95% control, affected facility is
the tank
95% control, affected facility is
the tank battery
PTE < 6 tpy VOC
No requirement
No requirement
a Well sites and compressor stations on the Alaska North Slope are subject to Annual OGI monitoring only.
b Specifically, the affected source is natural gas-driven controllers that vent to the atmosphere.
0 The zero emissions rate standard does not apply to pneumatic controllers at sites in Alaska for which on site power
is not available. Instead natural gas-driven continuous bleed controllers at those sites are required to achieve bleed
rates at or below 6 scfh, while natural gas-driven intermittent bleed controllers are subject to OGI monitoring and
repair of emissions from controller malfunctions.
d The proposed regulation requires liquids unloading events to be zero-emitting unless technical infeasibilities exist,
in which case the regulation requires that best management practices be adopted.
23
-------
Table 2-2 EG OOOOc Emissions Sources, Baseline Requirements, and Requirements
under the Proposed Option
Presumptive Standards of Performance
Source
In the Baseline
Under the Proposal
Fugitive Emissions/Equipment
Leaksab
Well Sites
Wellhead only, single well site
Quarterly AVO monitoring
Wellhead only, multiple well site
No requirement
Quarterly AVO monitoring +
Semiannual OGI
Single well site with a single price of
major equipment and no tank battery
Pre-OOOOa: No requirement
Post-OOOOa: Semiannual OGI
Quarterly AVO monitoring
Multiple well site with a single piece
of major equipment, or any site with
two or more pieces of major
equipment or one piece of major
equipment and a tank battery
Pre-OOOOa: No requirement
Post-OOOOa: Semiannual OGI
Bimonthly AVO monitoring +
Quarterly OGI
Gathering and Boosting Stations
Transmission and Storage Compressor
Stations
Pre-OOOOa: No requirement
Post-OOOOa: Quarterly OGI
Monthly AVO monitoring +
Quarterly OGI
Natural Gas Processing Plants
Pre-KKK: No requirement
Post-KKK and Pre-OOOO:
NSPS Subpart W
Post-OOOO: NSPS Subpart Wa
Bimonthly OGI
Pneumatic Pumps
Well Sites
Pre-OOOOa: No requirement
Post-OOOOa: 95% control
Methane emission rate of zero
Gathering and Boosting Stations
No requirement
Pneumatic Controllers'"
Well Sites
Gathering and Boosting Stations
Pre-OOOO: No requirement
Post-OOOO: Natural gas bleed
rate no greater than 6 scfh
Methane emission rate of zero0
Transmission and Storage Compressor
Stations
Pre-OOOOa: No requirement
Post-OOOOa: Natural gas bleed
rate no greater than 6 scfh
Natural Gas Processing Plants
Pre-OOOO: No requirement
Post-OOOO: Zero emissions
Methane emission rate of zero
Reciprocating Compressors
Gathering and Boosting Stations
Natural Gas Processing Plants
Pre-OOOO: No requirement
Post-OOOO: Rod-packing
changeout on fixed schedule
Volumetric flow rate of 2 scfm
Transmission and Storage Compressor
Stations
Pre-OOOOa: No requirement
Post-OOOOa: Rod-packing
changeout on fixed schedule
24
-------
Centrifugal Compressors
Wet-seal
Gathering and Boosting Stations
No requirement
Natural Gas Processing Plants
Transmission and Storage
Compressor Stations
Pre-OOOO: No requirement
Post-OOOO: 95% control
Volumetric flow rate of 3 scfm
Dry-seal
Gathering and Boosting Stations
Natural Gas Processing Plants
Transmission and Storage
No requirement
Volumetric flow rate of 3 scfm
Compressor Stations
Liquids Unloading
Well Sites
No requirement
Zero emissions or best
management practices'1
Storage Vessels
PTE > 20 tpy CH4
Pre-OOOO: No requirement
Post-OOOO: 95% control,
95% control, affected facility is
the tank battery
PTE < 20 tpy CH4 and > 6 tpy VOC
affected facility is the tank
PTE < 20 tpy CH4 and < 6 tpy VOC
No requirement
No requirement
a Well sites and compressor stations on the Alaska North Slope are subject to Annual OGI monitoring only.
b Specifically, the affected source is natural gas-driven controllers that vent to the atmosphere.
0 The zero emissions rate standard does not apply to pneumatic controllers at sites in Alaska for which on site power
is not available. Instead natural gas-driven continuous bleed controllers at those sites are required to achieve bleed
rates at or below 6 scfh, while natural gas-driven intermittent bleed controllers are subject to OGI monitoring and
repair of emissions from controller malfunctions.
d The proposed regulation requires liquids unloading events to be zero-emitting unless technical infeasibilities exist,
in which case the regulation requires that best management practices be adopted.
There are proposed requirements that we do not attempt to quantify regulatory impacts
for in this RIA. We do not attempt to quantify the impacts of the super-emitter response program
due to the unpredictable nature of super-emission events, resulting in a lack of specific data on
their frequency, intensity, and cost to mitigate. We note that our estimates may undercount the
emissions reductions achieved by this rule, as well as costs, because our analysis does not fully
account for cost-effective opportunities to prevent or quickly correct super-emitter emissions
events. Though we are not currently able to quantify the emissions reductions likely to result
from preventing or more quickly mitigating super-emitter emissions events, we note that the
information presented in Appendix D includes model simulations suggesting that covering large
emitters could "significantly impact[] the expected emissions from the fugitive emission
25
-------
program."20 Other requirements that we do not attempt to quantify regulatory impacts for in this
RIA include emissions control requirements for associated gas from oil wells, and storage vessel
control requirements at centralized production facilities (CPFs) and in the gathering and boosting
segment. While we expect the impacts from the associated gas and storage vessels at CPFs
provisions to be small relative to the overall impacts of the proposal, we do not expect them to be
insignificant, and quantifying their impacts is a priority for the final rule. We also do not account
for instances in which all or some sources in Alaska are subject to different requirements than
those in the rest of the country, both in the baseline due to previous rulemakings and in the
proposal; see Section 5.2 for additional discussion.
2.2 Methodology
The compliance cost and emissions reductions analysis summarized in this RIA reflects a
nationwide engineering analysis of which there are two main components: activity data and
information on control measures. The activity data represents estimates of the counts of affected
facilities over time, and the control measure information includes data on costs and control
efficiencies for typical facilities.
The first component is activity data for a set of representative (or model) plants for each
regulated facility.21 To project activity data for regulated facilities, we first project activity data
for oil and gas sites, which include well sites, natural gas processing plants, and compressor
stations (gathering and boosting, transmission, and storage). Projections include addition of
newly constructed sites and retirement of previously constructed sites, with magnitudes based on
a combination of analysis of several data sources and, where necessary, sensible assumptions.
Using representative "per-site" factors generated from EPA's Greenhouse Gas Inventory
(GHGI), regulated facilities are apportioned to sites across all industry segments.22 We assume
20 As stated, some of the model simulations in Appendix D suggest that large-emitters could significantly impact the
estimated emissions reductions; however, those simulations are not directly related to the definition of "super-
emitter" included in this proposal, thus the emissions and emission reductions cannot be used to directly assess the
emissions or emission reductions related to the proposed super-emitter program. The model simulations relied on
information of large emissions from a single basin (Permian), and available data suggest that the frequency of these
events may vary significantly across different production basins, which could lead to significant uncertainty if the
emission reductions were applied nationwide.
21 Regulated facilities include well site fugitives, gathering and boosting station fugitives, transmission and storage
compressor station fugitives, natural gas processing plant equipment leaks, pneumatic pumps, pneumatic controllers,
reciprocating compressors, centrifugal compressors, liquids unloading, and storage vessels.
22 Industry segments include production, gathering and boosting, processing, transmission, and storage.
26
-------
the per-site factors are fixed over time, so that the projected counts of regulated facilities change
in proportion to the projected counts of sites. The regulated facility projections are combined
with information on control options, including capital costs, annual operations and maintenance
costs, and control efficiencies. Information on control options is derived from the analysis
underpinning the BSER determinations. Impacts are calculated by setting parameters on how and
when affected facilities are assumed to respond to a regulatory regime, multiplying activity data
by model plant cost and emissions estimates, differencing from the baseline scenario, and then
summing to the desired level of aggregation. In addition to emissions reductions, some control
options result in natural gas recovery, which can then be combusted in production or sold. Where
applicable, we present projected compliance costs with and without the projected revenues from
product recovery.
For the analysis, we calculate the cost and emissions impacts of the proposed NSPS
OOOOb and EG 0000c from 2023 to 2035. The initial analysis year is 2023 as we assume the
proposed rule will be finalized early in that year. The NSPS OOOOb will take effect
immediately and impact sources constructed after publication of the proposed rule. We assume
the EG 0000c will take longer to go into effect as states will need to develop implementation
plans in response to the rule and have them approved by the Agency. We assume that this
process will take three years, and so EG 0000c impacts will begin in 2026. The final analysis
year is 2035, which allows us to provide ten years of impacts after the EG 0000c is assumed to
take effect.
While it would be desirable to analyze impacts beyond 2035, limited information
available to model long-term changes in practices and equipment use in the oil and natural gas
industry make the choice of a longer time horizon infeasible. In a dynamic industry like oil and
natural gas, technological progress is likely to change control methods to a greater extent over a
longer time horizon, creating more uncertainty about impacts of the NSPS OOOOb and the EG
0000c. For example, the current analysis does not include potential fugitive emissions controls
employing remote sensing technologies currently under development.
27
-------
2.2.1 Activity Data Projections
To construct the activity data projections used in this analysis, we rely on historical data
from the Greenhouse Gas Inventory (GHGI),23 industry data collected by EPA through an
information collection request (ICR) distributed in 2016 (hereafter, "2016 ICR"), information
from the private firm Enverus that provides energy sector data and analytical services,24 and
projections from the U.S. Energy Information Administration's (EIA) Annual Energy Outlook
(AEO).25 Our projections follow a two-step procedure. First, we construct projected counts of oil
and natural gas "sites," such as well sites, compressor stations, and processing plants, that
contain or are themselves facilities affected by the regulations. Second, using per-well factors,
we build upon the site projections to estimate the counts of these "affected facilities." The details
of these calculations are described by site/regulated facility type below.
In addition to sites and affected facilities, there is a third category of activity data that we
track. When comparing a new regulatory regime, such as the proposed rule, to the baseline
scenario, a subset of affected facilities is assumed to take action to comply with regulatory
requirements: we refer to these facilities as "incrementally impacted facilities." In Section
2.2.1.3 below, we provide a table of incrementally impacted facility counts for the proposed rule
relative to the baseline.
2.2.1.1 Projected Oil and Natural Gas Sites
There are three types of "sites" in our analysis of projected facilities: well sites,
compressor stations, and natural gas processing plants. Compressor stations are further
subdivided into sites located in different segments of the natural gas sector, that is, the gathering
and boosting, transmission, and storage segments. For each site type, we generate annual
projections of cumulative and new counts for four different "vintage" bins: the first vintage (VI)
represents sites constructed prior to NSPS OOOO, the second vintage (V2) represents sites
constructed after NSPS 0000 but prior to NSPS 0000a, the third vintage (V3) represents sites
constructed after NSPS 0000a but prior to NSPS 0000b, and the fourth vintage (V4)
23 See Methodology Annexes 3.5 and 3.6 at https://www.epa.gov/ghgemissions/natural-gas-and-petroleum-systems-
ghg-inventory-additional-information-1990-2019-ghg. Activity data is presented in Tables 3.5-5 and 3.6-7,
respectively.
24 Enverus: https://www.enverus.com/.
25 EIA AEO: https://www.eia.gov/outlooks/aeo/.
28
-------
represents sites constructed after NSPS OOOOb. Within V3, sites are further subdivided into
sites constructed after NSPS 0000a through the base year (2019) and 3 additional vintages
representing sites assumed to be constructed in each year from 2020 through 2022, while V4 is
subdivided into separate vintages for each year from 2023, when the NSPS OOOOb is assumed
to take effect, through 2035. In the case of well sites only, VI is further subdivided into sites
constructed prior to 2000 and sites constructed from 2000 on.
There are two countervailing forces that impact the overall trajectory of our estimated
sites beyond the base year: the rate at which new sites are constructed and the rate at which sites
retire (or cease operation). In our analysis, counts of newly constructed sites are based on either
analysis of historical trends from the GHGI (compressor stations), GHGI and the Department of
Homeland Security's Homeland Infrastructure Foundation-Level Data (HIFLD) (compressor
stations), or projections from AEO (well sites). Estimates of retirement rates are based on
analysis of Enverus data (well sites) and assumptions underlying analysis submitted in response
to the 2018 NSPS OOOOa Policy Reconsideration proposal (processing plants and compressor
stations);26 along with new site counts, those rates are summarized in Table 2-3. To avoid sites
having implausibly short lifespans in the analysis, we assume site retirements only apply to well
sites that are at least five years old and processing plants and compressor stations in VI (pre-
OOOO).
26 See page 4 of Appendix D of Docket ID No. EPA-HQ-OAR-2017-0757-0002.
29
-------
Table 2-3 Assumed Retirement Rates and Annual New Site Counts by Site Type
Annual Retirement Rate as a
Type of Site New Site Counts in Each Year Percentage of Existing Stock
Well Sites 14,000 - 28,000
Greater than 15 barrels of oil (j(y
equivalent (boe) per day 0
3-15 boe per day
Oil - 1.6%
Gas - 1%
Less than 3 boe per day
Oil - 6.7%
Gas - 4.4%
Compressor Stations
Gathering and Boosting 439 1%
Transmission 102 1%
Storage 2 1%
Natural Gas Processing Plants 7 1%
Our projections of the cumulative counts of sites for each vintage are illustrated in Figure
2-1. While the projected total counts of wells are relatively stable over the analysis horizon, the
projected total counts of well sites decline significantly, as older, smaller sites are displaced by
newer, larger sites. The total counts of natural gas processing plants and storage compressor
stations change slightly over time, due to very few assumed annual additions and retirements.
For gathering and boosting and transmission compressor stations, the total number of sites
increase significantly over the analysis horizon. Below, we describe how those trajectories are
generated for each site type.
30
-------
Wells
750,000-
500,000-
t/> 250,000-
-4—"
c
3
o
° o-
Q) '
U)
CD
>
+-»
TO
3
E
Well Sites
600,000-
600,000-
400,000-
200,000-
NG Proc. Plants
'0 2025 2030 2035
2020 2025 2030 2035 2020 2025 2030 2035
Stor. Stations
O
10,000
5,000
100-
Exists
Projected
2020 2025 2030 2035
2020 2025 2030 2035
Year
2020 2025 2030 2035
Figure 2-1 Projections of Cumulative Site Counts by Site Type and Vintage
(a) Well Sites
The dataset used to characterize the base year (2019) population of oil and natural gas
well sites is developed from data provided by Enverus, a private firm focused on the energy
industry that provides data and analytical services. The dataset includes two types of entities:
wells and leases. Whether a well is represented as its own entity or as part of a lease depends on
the state in which the well is located, as reporting requirements differ across state agencies. The
columns in the dataset include entity identifiers, well site identifiers (for wells), locations,
completion and initial production dates, well counts (for leases), and natural gas and liquids
production levels. We restricted the dataset to onshore wells with positive production values in
2019. The base year is chosen as 2019 as we assume that it is the most recent year with
comprehensive data coverage due to reporting lags.
Using the base year dataset, we perform a series of steps to convert from well- and lease-
level data to site-level data. First, we aggregate the well-level data into site-level data using the
Enverus-provided well site identifiers when available. Each data point includes information on
site location, date (based on the most recent well completion), count of oil wells, count of gas
wells, and liquids and gas production levels. Wells are assigned as oil or gas based on gas-to-oil
31
-------
ratios (GOR).27 For wells without site identifiers and leases, we track the same information, but
at the well or lease level. For each entity (site, well, or lease), we project production forward
through the end of the analysis horizon using simple decline rate assumptions based on analysis
of historical production data.28
Using the projections, we aggregate entities into representative groups for each year
(2019-2035). For well sites, each group characterized by a unique combination of state, vintage
(based on the bins described in the previous section), site type (oil or natural gas), well count bin
(single well or multi-well), base year production rate bin, and current year production rate bin.29
Each group includes total counts of sites, oil and gas wells, and oil and gas production. Likewise,
the well and lease entities for which we don't have site identifiers are aggregated analogously,
but without well count bins. To fill that gap, we apportion the well counts and production levels
of the well/lease entity groups into single and multi-well bins based on national-level proportions
derived from the subset of data with well site identifiers, stratifying over site types, vintage bins,
and initial production rate bins. To complete the imputation, we calculate the number of sites
within each group by dividing well counts by national average estimates of the number of wells
per site. The two sets of groups are then combined to form one cohesive dataset with projections
of production for a collection of representative well site groups.30
In addition to the projection of the base year dataset, we also implement a series of steps
to construct projections for well sites assumed to be constructed in years beyond 2019. First, we
implement the well site grouping procedure just described, but restricted to sites with completion
dates in a recent vintage (2018). Our operating assumption is that future sites will be distributed
27 If GOR > 100,000 mcf per bbl, then the well is designated as a gas well, otherwise, it is designated as an oil well.
28 Decline rates are estimated using well-level Enverus production data from 2010-2020. For each well and
production year, decline percentages for oil/condensate and gas production are calculated as the production level in
the next year less the production in the current year, divided by current production. We then calculate median
decline rates for four production rate bins, resulting in the following decline rate assumptions:
Production Rate Bin (barrels of oil equivalents/day, or BOE/d)
Production Type Greater than 100 15-100 3-15 Less than 3
Oil/Condensate 35% 18% 11% 10%
Gas 26% 13% 9% 7%
29 Sites are grouped the four production rate bins, based on the average BOE/d per well at the site, described in
footnote 27.
30 The dataset, along with the analysis code used to estimate impacts, can be found in the docket.
32
-------
similarly across locations, site types, well count bins, and production rate bins as sites recently
completed. We then use the representative grouping to distribute AEO2022 nationwide-
projections of new wells drilled from 2020 to 2035 based on the relative proportions of wells in
each group, with production for each vintage projected through 2035.
The last step in the well site projection procedure is to apply the retirement percentages
presented in Table 2-3. The retirement percentages differ by production bin and site type, and are
otherwise uniformly applied across groups regardless of other characteristics, such as location.31
Only low production sites (less than 15 BOE/d/well) are assumed to retire, and the bulk of
retirements come from sites with very low production (less than 3 BOE/d/well).
(b) Compressor Stations
We project compressor stations for three segments (gathering and boosting, transmission,
and storage) using data from GHGI; the approach for all three segments is analogous.32 The first
step is to estimate the number of stations in the base year, 2019. We assume that the number of
stations in 2011 are all VI stations (pre-OOOO). To get the counts of VI stations in subsequent
years, including the base year, we apply the relevant annual retirement rates to the 2011 station
counts. The number of V2 stations (post-OOOO, pre-OOOOa) in 2019 is estimated by
subtracting the estimated number of VI stations in 2015 from the total station counts from 2015.
The number of V3 stations (post-OOOOa) in 2019 is estimated by subtracting the estimated
number of VI and V2 stations in 2019 from the total number of stations.
To project the number of new stations constructed in the years after the base year, we
calculate a historical average number of new stations per year over a recent period (as presented
in Table 2-3), and apply it uniformly across all years. Specifically, we divide the calculated
number of V3 stations in 2019 and divide it by four, as the first V3 stations are assumed to be
constructed in 2016. This yields an estimate of the average number of V3 stations added per year
31 Retirement percentages are estimated using well-level Enverus production data from 2010-2020. For a subset of
those years (2012-2018), we identify wells that previously had production, but have no recorded oil or gas
production records for 2 consecutive years, as retired. Retirement percentages are then calculated by dividing the
count of retired wells in each year by the total count of producing wells from the previous year. The retirement rate
percentage assumptions result from averaging the estimated retirement rates over all years.
32 Station counts are extracted from the following rows: Yard Piping (gathering and boosting) and Station +
Compressor Fugitive Emissions (transmission and storage).
33
-------
through the base year, and we assume new stations are added at that same rate beyond the base
year. New stations assumed to be constructed in 2020 and 2021 are assigned to V3, while all
estimated new stations beyond 2021 are assigned to V4.
The final step is to project station counts for those existing in the base year and combine
those projections with the new construction projections. This results in a set of projections in
which VI station counts decline over the analysis horizon due to retirements and V2 station
counts are uniform over the analysis horizon. V3 station counts are also uniform over the
analysis horizon, but they are split across three vintages (2016-2019, 2020, and 2021), with the
latter two equal to the average number of new stations described above. Finally, V4 station
counts are equal to the new station estimates in all vintage/year combinations from 2022-2035.
(c) Natural Gas Processing Plants
To construct base year activity data counts for natural gas processing plants, we leverage
data from both the GHGI and HIFLD.33 The estimates of the counts of VI and V2 plants are
generated using the same process as for compressor stations: the 2011 count of plants are
assigned to VI, and the V2 count of plants in 2015 is estimated to be the 2015 count from the
GHGI minus the estimated count of VI plants in 2015 after the annual retirement rates are
applied. We use the HIFLD as a source of 2020 plant counts since plant counts have been fixed
in the GHGI in recent years due to lack of data, and the latest update date for HIFLD is from
October 2020. Estimates for the count of V3 plants in 2020 are then calculated using the 2020
total plant estimate and subtracting VI (after applying retirements) and V2 plant counts. The
estimated number of new plants in each year beyond the base year is then calculated by dividing
the number of V3 plants in 2020 by the number of years (5) assumed to have passed since the
first NSPS OOOOa-affected facilities were constructed. That estimate is used to calculate the
number of V3 plants in the base year by subtracting it from the 2020 count, as well as to populate
the counts of plants for all vintages and years beyond the base year.
33 The dataset of processing plants is downloaded from https://hifld-
geoplatform.opendata.arcgis.com/datasets/geoplatform: :natural-gas-processing-plants/explore. We filter out plants
located in Canada and Mexico.
34
-------
2.2.1.2 Affected Facilities
In most cases, estimates of projected affected facility counts are generated by assuming
fixed proportional relationships with the site counts. This means that as site counts are projected
to expand (construction of new sources) or contract (retirement of existing sources), the counts
of affected facilities expand and contract as well such that the ratio of facilities to sites remains
constant. Details for each affected facility type are provided below.
(a) Fugitives and Leaks
The proposed rule features LDAR requirements across all segments. Well site
requirements are the most nuanced and depend on the equipment present at the site, which we
characterize through a series of data processing steps. Compressor station requirements are
uniform across segments, and we rely on a single representative plant in each segment to
estimate the impacts of those requirements. Requirements at natural gas processing plants
distinguish the collection of VOC service components and the collection of non-VOC service
components. Our impacts analysis for processing plants differentiates between two model plant
types representing "large" and "small" facilities.
The proposed rule features different monitoring frequency requirements for well sites
depending on the equipment present at a site. The Enverus data does not provide information on
site equipment, so we assign well site groups to equipment categories in fixed proportions based
on analysis of data from EPA's 2016 Information Collection Request (ICR) for the Oil and
Natural Gas Industry (hereafter, 2016 ICR).34 The data captured a survey of major equipment
(separators, compressors, and dehydrators) and storage tanks at more than 100,000 well sites
across the U.S., which we use to separately estimate the proportions of sites in six equipment
categories for all combinations of oil and gas sites and single and multi-well sites.35 We also
34 See https://www.epa.gov/controlling-air-pollution-oil-and-natural-gas-industry/background-information-request-
oil-and for more information on the ICR. The ICR was withdrawn in 2017, but not before significant amounts of
data were collected. The data used for this analysis were obtained from the file
"OilandGasSpreadsheetUnredacted.xlsx" found at
https://foiaonline.gov/foiaonline/action/public/submissionDetails?trackingNumber=EPA-HQ-2017-
003014&type=Request.
35 The equipment categories are: (1) no equipment or tanks; (2) no equipment with storage tanks; (3) one piece of
major equipment without tanks; (4) one piece of major equipment with tanks; (5) more than one piece of major
equipment without tanks; and (6) more than one piece of major equipment with tanks. Since we estimate proportions
for all combinations of site type (oil, gas) and well count bin (single, multi), there are 24 possibilities in total.
35
-------
calculate, for each combination of equipment category, site type, and well count bin average
equipment and tank counts per oil and gas well. See Section A. 1 in the appendix for a detailed
discussion of how the ICR data is processed to construct equipment bin proportions and average
equipment counts.
The equipment category proportions and average counts generated from the ICR are
applied to the well site projections and scaled to match per-well equipment factors from the
GHGI from the base year. The calibration entails a series of steps, beginning with the imputation
of equipment (process heaters and heater-treaters) not surveyed in the 2016 ICR. Using site-level
survey data provided by the American Petroleum Institute (API),36 we estimate the proportion of
sites that have exactly one piece of major equipment captured in the ICR (separators,
compressors, and dehydrators) that also have heaters or heater-treaters and use this to adjust the
equipment category proportions estimated using the ICR data. The API survey data is also used
to estimate the average number of heaters and heater-treaters per gas and oil well for oil/gas and
single/multi-well sites. These estimates, along with the ICR average equipment estimates, are
applied to the well site group dataset for the base year, and then the per-well equipment counts
are scaled uniformly across equipment categories and site types such that the aggregate per-well
equipment counts across all well sites match the GHGI in 2019. As part of this process, we also
calculate the number of headers per oil well (only at sites with major equipment) and
meters/piping per gas well (at all sites) such that the per-well counts of that equipment also
matches the GHGI in 2019. More details on how the API survey data is used and the calibration
steps is available in Section A.2 in the appendix.
The equipment category proportions are illustrated in Table 2-4 for well sites in the base
year and newly constructed sites in subsequent vintages. The table reflects the distribution of
sites after making the adjustments for heaters and heater-treaters and applying the stratified
proportions to the well site group dataset. Importantly, we assume that equipment is assigned to
well sites based on either base year production levels (for sites in the base year dataset) or first
36 See Attachment 4 (Microsoft Excel workbook) of Docket ID No. EPA-HQ-OAR-2017-0757-0002, EPA Analysis
of Well Site Fugitive Emissions Monitoring Data Provided by API. The dataset contains survey data on 2,183 gas
well sites and 1,742 oil well sites.
36
-------
year production levels (for projected new construction beyond the base year) and does not
change as production declines.
Fugitive emissions monitoring requirements differ across the equipment bins captured in
the table. In the analysis of the proposed option, single well wellhead-only sites and sites with
only one major piece of equipment and no tank battery are assumed to perform quarterly AVO
inspections. Wellhead-only sites with multiple wells are assumed to perform quarterly AVO and
semiannual OGI monitoring. Sites with two or more major pieces of equipment, one piece of
major equipment and a tank battery, or multi-wellhead sites with one piece of major equipment
or a tank battery are assumed to perform bimonthly AVO and quarterly OGI monitoring. To
calculate impacts for the fugitive monitoring requirements at well sites, we allocate the total
number of well sites to the bins defined by counts of major equipment and tank batteries.
Table 2-4 Distribution of Well Sites in Equipment Bins
Proportions in the Proportions in the
Site Bin base year (2019) projected years
Natural Gas
Single wellhead
Wellhead only 34% 33%
One piece of major equipment, including tank batteries 44% 43%
More than one piece of major equipment, including tank batteries 16% 15%
Multi-wellhead
Wellhead only 0.13% 0.16%
One piece of major equipment, including tank batteries 5.0% 7.8%
More than one piece of major equipment, including tank batteries 0.49% 0.58%
Oil
Single wellhead
Wellhead only 51% 45%
One piece of major equipment, including tank batteries 31% 30%
More than one piece of major equipment, including tank batteries 8.6% 7.3%
Multi-wellhead
Wellhead only 0.6% 0.8%
One piece of major equipment, including tank batteries 7.2% 15%
More than one piece of major equipment, including tank batteries 1.3% 1.7%
Affected facility counts for compressor station fugitives are equal to the compressor
station counts detailed in the previous section. As such, compressor station fugitives affected
facility counts are binned according to segment, vintage, and year.
37
-------
There are two affected facility types associated with natural gas processing plant leaks:
the collection of VOC service components and the collection of non-VOC service components.
In each case, the number of affected facilities is equal to the number of processing plants, and so
the total number of affected facilities is twice the number of processing plants. For the purposes
of calculating impacts associated with LDAR at processing plants, we assume that 80 percent of
plants are "large" and 20 percent are "small".37
(b) Pneumatic Controllers
Pneumatic controllers are represented in the GHGI for all segments. For well sites, we
estimate the number of controllers at sites based on equipment counts, scaling the estimates such
that the aggregate controller-per-site counts match the GHGI in 2019. For compressor stations,
controller counts are directly based on per-station counts from the GHGI in 2019. For processing
plants, we assume that all controllers are already powered by compressed air and therefore do not
estimate any impacts from the proposal for that segment and affected facility.
To estimate controller counts at well sites, we proceed in three steps. First, we multiply,
for each well site group, equipment counts per oil and gas well by controller-per-equipment
factors presented in the Supporting Information of Zavala-Araiza et al. (2017).38 Second, we
uniformly scale the resulting controller counts per well across all sites in the base year such that
they match, in aggregate, the GHGI controller per-well counts in 2019 for both oil and gas wells.
Finally, we allocate the controllers across to three types (low-bleed, high-bleed, and intermittent
bleed) such that each type matches the corresponding GHGI controller per-well counts in 2019,
assuming that no high-bleed controllers exist at post-0000 sites in any state and at pre-0000
sites in California, Colorado, or Utah.
The estimation of controller counts at compressor stations is similar to the last step for
well sites. In that case, we assume that high-bleed controllers are only allocated to pre-0000
gathering and boosting stations and pre-OOOOa stations for transmission and storage. In
37 See page 6 of Chapter 10 of the November 2021 TSD.
38 Using data from Allen et al. (2015), the authors estimate 0.42 controllers per wellhead, 1 controller per separator
at gas sites without liquids, 2.06 controllers per separator at sites with liquids production, 1.5 controllers per process
heater, 4.3 controllers per compressor, and 2.5 controllers per dehydrators.
38
-------
aggregate, the per-station counts for all three types of controllers in the base year match the per-
station counts from the GHGI in 2019.
(c) Pneumatic Pumps
The GHGI provides information on the number of pneumatic pumps in the production
and gathering and boosting segments. For well sites, we assume that 30 percent of gas sites with
equipment (and no sites without equipment) have chemical injection pneumatic pumps, based on
analysis of the data underlying Allen et al. (2013).39 Likewise, we assume 25 percent of oil sites
with equipment have chemical injection pneumatic pumps, based on an analysis of the API
survey data. For each site assumed to have pumps, we initially assign one pump to the site.
Additional pumps are assigned in proportion to the number of pneumatic controllers at each site
such that number of pumps per-well matches the 2019 data from the GHGI. For the gathering
and boosting segment, we calculate the number of pumps per station implied by the GHGI in
2019 and apply the value to all stations for all vintages in all years.
(d) Reciprocating Compressors
The GHGI contains estimates of the number of reciprocating compressors in the
gathering and boosting, processing, transmission, and storage segments. In all cases, we calculate
the number of reciprocating compressors per site using the 2019 values from the GHGI and
apply those ratios to the cumulative and new station counts for all vintages and years. In the case
of gathering and boosting stations, the GHGI only includes a total count of compressors; we
assume that 89 percent of those are reciprocating.40
(e) Centrifugal Compressors
The GHGI contains estimates of the number of wet-seal and dry-seal centrifugal
compressors in the gathering and boosting, processing, and transmission segments. For the
transmission and storage segments, we assume that no wet seal compressors have been installed
since the NSPS OOOOa due to the routing requirements in that rule, and that no wet seal
compressors will be installed at NSPS OOOOb-affected stations either. Taking that into account,
39 The data can be downloaded from http://dept.ceer.utexas.edu/methane/study/datasets3.cfm. The workbook used
for this analysis is finalSITES.xlsx.
40 This assumption is based on data summarized on page 28 of Zimmerle et al. (2019).
39
-------
wet and dry seal compressors are allocated on a per-station basis such that the estimated counts
of wet and dry seal compressors per station in the base year match the GHGI data from 2019. For
gathering and boosting stations, the process is similar except we allocate wet seal compressors to
all vintages since this segment/affected facility type has yet to be regulated. Also, the GHGI only
includes a total count of compressors for this segment; we assume that 3 percent of those are
centrifugal,41 and that the proportion of wet-seal to dry-seal centrifugal compressors is the same
as it is in the transmission segment.
(j) Liquids Unloading
For the purposes of the RIA, liquids unloading affected facilities are defined at the event
level and apply only to natural gas well sites. To estimate impacts more accurately, we divide
natural gas wells into two categories: those with plunger lifts and those without plunger lifts. The
GHGI contains activity data for the number of wells in each category that perform liquids
unloading events, so we divide that number by the total number of natural gas wells in the
inventory in 2019 to generate fractions of wells performing liquids unloading for each category.
Those fractions are applied to our projections of well sites with equipment for all years and
vintages. In the case of wells with plunger lifts, we assume that 76 percent of wells perform
manual unloading.42 Finally, we convert from wells to events by multiplying by events per well
values from the BSER analysis.43
(g) Storage Vessels
Storage vessel-affected facility projections are generated for well sites only; projections
of tanks at centralized production facilities and in the gathering and boosting segment are an area
of ongoing development. As described in Section 2.2.1.1(a), proportions of sites with tanks and
tank counts per oil and gas well are generated from the 2016 ICR data and merged into our well
site projections. For each site assumed to have tanks, the total count of tanks is assumed to
41 Ibid.
42 Memorandum. Analysis of Greenhouse Gas Reporting Program Liquids Unloading Data. Prepared by SC&A
Incorporated for Amy Hambrick, SPPD/OAQPS/EPA. October 14, 2021. Docket ID No. EPA-HQ-OAR-2021-
0317-0143. As summarized in the memo, analysis of well-level data from the GHGRP for reporting years 2015-
2019 suggested that 76% of plunger lifts were manually operated.
43 See page 12 of Chapter 11 of the November 2021 TSD. We assume that wells without plunger lifts have 5.6
events per year, and wells with manually operated plunger lifts have 7.7 events per year.
40
-------
comprise a single tank battery. All liquids production at those sites (crude at oil sites, condensate
at gas sites) is assumed to be throughput to the tank battery.
2.2.1.3 Incrementally Impacted Facilities
Estimates of incrementally impacted facility counts by year and regulated facility for the
proposed rule are presented in Table 2-5. The counts for well sites and compressor stations
represent fugitives requirements at those sites and the counts for natural gas processing plants
represent VOC and non-VOC service.
41
-------
Table 2-5 Projection of Incrementally Impacted Affected Facilities under the Proposed NSPS OOOOb and EG OOOOc
Option, 2023 to 2035
Gathering Transmission Natural
and and Storage Gas
Boosting Compressor Processing
Well Site
Station
Station
Plants
Pneumatic
Reciprocating
Centrifugal
Liquids
Storage
Year
Fugitives
Fugitives
Fugitives
Leaks
Devices
Compressors
Compressors
Unloading
Vessels
2023
14,000
0
0
23
76,500
2,700
320
6,800
2,100
2024
22,000
0
0
34
115,500
4,000
480
11,000
3,200
2025
31,000
0
0
45
167,600
5,400
650
15,000
4,500
2026
520,000
5,200
1,900
1,200
1,400,000
36,000
4,800
260,000
5,800
2027
510,000
5,200
1,900
1,200
1,500,000
38,000
4,900
260,000
7,000
2028
500,000
5,200
1,900
1,200
1,500,000
39,000
5,100
260,000
8,300
2029
490,000
5,100
1,900
1,200
1,500,000
40,000
5,200
250,000
9,600
2030
480,000
5,100
1,800
1,200
1,499,000
41,000
5,400
250,000
11,000
2031
480,000
5,000
1,800
1,200
1,499,000
42,000
5,500
250,000
12,000
2032
470,000
5,000
1,800
1,200
1,498,000
43,000
5,600
250,000
13,000
2033
460,000
5,000
1,800
1,200
1,497,000
45,000
5,800
250,000
15,000
2034
450,000
4,900
1,800
1,200
1,497,000
46,000
5,900
240,000
16,000
2035
440,000
4,900
1,800
1,200
1,496,000
47,000
6,000
240,000
17,000
42
-------
2.2.2 Model Plant Compliance Cost and Emissions Reductions
The cost and emissions characteristics of the site projections used to estimate the impacts
of the proposed rule are derived from the technical analyses underpinning the BSER
determination. In some cases, we define our affected facilities' projections to be identical to the
model plants found in the Supplemental TSD or the November 2021 TSD, and so the cost and
emissions estimates can be directly applied. In other cases, however, our model plants leverage
the underlying data from the TSDs and other data sources to better fit the activity data.
We use cost and emissions information directly from the November 2021 TSD for
compressor station fugitives, natural gas processing plant leaks, and reciprocating compressors,
and from the Supplemental TSD for centrifugal compressors. Compressor station fugitives are
represented by a single model plant for each of the gathering and boosting, transmission, and
storage segments.44 Processing plant leaks are divided into four different model plants: all
combinations of large and small plants, and VOC and non-VOC service.45 Reciprocating
compressors are represented by a single model plant for each of the gathering and boosting,
processing, transmission, and storage segments.46 Wet-seal and dry-seal centrifugal compressors
are each represented by a single model plant for each of the gathering and boosting, processing,
and transmission segments.47
The methodology for projecting of costs and emissions impacts from OGI monitoring
programs of different frequencies uses counts of major equipment in well site groups (described
in Section 2.2.1.2(a)) and the results of the BSER technical analysis performed in support of this
action. The BSER analysis uses simulations produced by the Fugitive Emissions Abatement
44 See Chapter 12 of the November 2021 TSD for details on costs and emissions reductions associated with quarterly
OGI monitoring, which represents the proposed BSER for compressor station fugitives in both the NSPS OOOOb
and EG 0000c.
45 See Chapter 10 of the November 2021 TSD for details on costs and emissions reductions associated with NSPS
W Method 21 (the BSER established in NSPS KKK), NSPS Wa Method 21 (the BSER established in NSPS
0000), and bimonthly OGI (the BSER proposed in NSPS OOOOb and EG 0000c).
46 See Chapter 7 of the November 2021 TSD for details on costs and emissions reductions associated with rod-
packing replacement on a fixed schedule (the BSER established in NSPS 0000 and NSPS 0000a) and rod-
packing replacement based on emissions monitoring (the BSER proposed in NSPS OOOOb and EG 0000c).
47 See Chapter 2 of the Supplemental TSD for details on costs and emissions reductions associated with a direct
inspection and maintenance/repair program to maintain emissions below 3 scfm (the BSER proposed in NSPS
OOOOb for dry seals and EG 0000c for wet and dry seals).
43
-------
Simulation Tool (FEAST), which are descriped in Appendix D.48 FEAST calculates simulated
costs and emissions reductions of LDAR programs at model well sites under different
assumptions. The BSER analysis performs FEAST simulations using four model well sites (a
single-well site with no major equipment (MW1), a multi-well site with no major equipment
(MW2), a multi-well site with a separator, an in-line heater, and a dehydrator (MW3), and a
multi-well site with a separator, an in-line heater, a dehydrator, and a controlled storage tank
battery (MW4)) and five OGI frequencies (annual, semiannual, quarterly, bimonthly, and
monthly). Each model well site contains an assumed number of components based on the
number of wells and the type of major equipment present at the site. A FEAST simulation for a
model well site produces average annualized cost and emissions reduction percentage for each
OGI monitoring frequency along with a baseline emissions rate in the absence of an LDAR
program. For the impacts analysis presented in this document, we used emissions rates and
control efficiencies generated from the FEAST runs without large emission events described in
Section 4 of the FEAST memo (see Appendix D). As noted on pages 23-24 in Section 5 of the
FEAST memo, incorporating large emission events could change the results substantially.
The calculation of costs and emissions impacts from a well site group consists of five
main steps. First, a well site group is assigned major equipment as described in Section
2.2.1.2(a). Next, a component count per well site is determined for a well site group based on the
counts of major equipment from step 1. Component counts for each type of major equipment are
assigned based on Tables W-1B (for gas well sites) and W-1C (for oil well sites) from 40 CFR
part 98, Subpart W.49 Third, an emissions factor per component is determined based on the
FEAST simulation. The emissions factor per component is calculated by averaging baseline
emissions per component in the absence of OGI monitoring for each model well site in the
FEAST simulations. This emissions factor per component is multiplied by the number of
components per well site and summed over well sites to determine baseline emissions for a well
site group. Fourth, each well site group is matched to a FEAST model well site based on major
equipment counts. Single-well sites with no major equipment or one piece of major equipment
are matched to MW1, multi-well sites with no major equipment are matched to MW2, and all
48 See also Chapter 5 of the Supplemental TSD for details on the FEAST modeling and costs and emissions
reductions associated with OGI monitoring at well sites.
49 See https://www.ecfr.gov/current/title-40/chapter-I/subchapter-C/part-98.
44
-------
other sites are matched to MW3. Finally, the cost and emissions impacts of an OGI monitoring
program of a given frequency is determined by applying the average annualized cost and
emissions reduction percentage for the matched model well site from the FEAST simulations to
the well site group.
Control of pneumatic controllers and pneumatic pumps are analyzed in a unified
framework for pneumatic devices using a combination of BSER analysis, Carbon Limits'
abatement cost tool, and the GHGI. Our analysis incorporates the impacts of replacing high bleed
with low bleed pneumatic controllers,50 which reflects the BSER established in NSPS OOOO for
well sites and gathering and boosting stations and NSPS 0000a for transmission and storage
compressor stations, as well as three zero emitting control options: electronic controllers using
grid electricity, electronic controllers powered by solar photovoltaic (PV) and battery systems,
and compressed air systems using grid electricity.51 Emissions factors for low-bleed, high-bleed,
and intermittent bleed pneumatic controllers (all segments except processing) and pneumatic
pumps (production52 and gathering and boosting) from the GHGI are converted from kg CH4 per
device to tons CH4 per device and applied directly to device counts at the site level to calculate
site-wide emissions, pre- and post-control. Control costs vary across control options and are
described in detail below.
For each control option, pneumatic device control costs are comprised of capital and
annual operations and maintenance costs, each of which is based on two main components: site-
level "base" costs that are independent of the number of devices at the site, and costs that scale
with the number of devices at the site. For replacement of high-bleed controllers with low-bleed
controllers, control costs scale linearly with the number of high-bleed controllers at the site and
are estimated using the BSER analysis from the 2011 TSD after updating to the 2019 dollar year
(U.S. EPA, 201 lc). Consistent with the Carbon Limits tool assumptions, we assume retrofit
capital costs for low-bleed controllers (not including installation labor costs) are half of the cost
50 See Chapter 8 of the November 2021 TSD for details on costs and emissions reductions associated with replacing
high bleed with low bleed pneumatic controllers.
51 See Chapter 3 of the Supplemental TSD and Attachment Q "Carbon Limits 2021 Zero Bleed Pneumatics Cost
Tool" at https://www.regulations.gov/comment/EPA-HQ-OAR-2021-0317-0845 for details on costs and emissions
reductions associated with installing zero-bleed controllers (the BSER established in NSPS OOOO for processing
plants and proposed in NSPS OOOOb and EG 0000c for all other segments).
52 Specifically, the chemical injection pump emissions factor is used.
45
-------
of new capital costs, since only the controller and not the control valve will be required. For
electronic controllers powered by solar photovoltaic and battery systems, we calculate the base
and per-device capital costs associated with installing and replacing control panels, solar PV
panels, batteries, and devices and annual (per-device) costs associated with device maintenance.
Cost calculation for electronic controllers powered by grid electricity is similar, but with solar
PV and battery capital costs replaced by base and per-device annual electricity costs. For
compressed air systems, we calculate the base and per-device capital costs associated with
installing and replacing a compressor and base and per-device annual costs associated with
compressor maintenance and grid electricity purchases.53
For the proposed rule, as well as the regulatory alternatives specified in Section 2.6,
control options are applied at the site level and compared to the baseline. Costs in the baseline
consist of purchasing and installation costs (for newly constructed sites) and maintence costs (for
newly constructed and existing sites) of natural gas-driven pneumatic devices. We assume that
electronic controllers powered by solar PV and battery systems are used to comply with the zero
emissions standard at all well sites and gathering and boosting stations. This assumption likely
overestimates the costs of compliance somewhat, since sites with access to grid electricity would
not incur the costs for the solar PV panels and batteries. In contrast, we assume that transmission
and storage compressor stations are grid-connected, with the former complying through
installation of electronic controllers and the latter, due to the large number of controllers
assumed to be located at the model plant, complying through installation of a compressed air
system.
We define two model plants for liquids unloading: events at wells without plunger lifts
and manual unloading events at wells with plunger lifts. In both cases, the costs per event are
taken directly out of the Supplemental Proposal TSD. However, whereas the BSER analysis
evaluates a range of emissions reductions levels associated with the proposed option, this
53 Based on the Carbon Limits tool, we assume that compressor costs (capital and maintenance) are a quadratic
function of horsepower requirements, which is a linear function of the number of each type of device at the site.
Therefore, we model a second-order polynomial relationship between site-level compressor costs and the numbers
of each type of device at the site.
46
-------
analysis assumes emissions reductions of 29 percent and 36 percent for events at wells without
plunger lifts and manual unloading events at wells with plunger lifts, respectively.54
Storage vessel control costs and emissions reductions are adapted from the BSER
analysis summarized in Chapter 6 of the November 2021 TSD. For each well site group assumed
to have tanks, representative site-level tank potential to emit (PTE) is calculated by multiplying
crude or condensate throughput by an average emissions factor derived from the BSER
analysis.55 To determine which post-0000 sites are assumed to have controlled tanks in the
baseline, we use the VOC PTE estimate in the base year or initial year of construction,
whichever is later; pre-0000 sites are assumed to be uncontrolled in the baseline. In the
proposal scenario, control requirements are determined by the VOC PTE estimates in the year of
construction for NSPS OOOOb-affected facilities and CH4 PTE estimates in the year that the EG
is assumed to take effect (2026) for EG OOOOc-affected facilities.56 For sites subject to control
requirements, we assume that 95% control is achieved through application of flares to the entire
tank battery. The costs of control are based on the BSER analysis, with capital and annual costs
equal to a minimum value below a 50 TPY CH4 emissions threshold and following a quadratic
cost function for sites with emissions above that threshold.57
2.2.3 State Programs
The oil and natural gas industry is subject to numerous state and local requirements.
These requirements differ greatly in scope and stringency across states. Given the difficulty in
attempting to incorporate the myriad of state regulations in the baseline, we have chosen to
incorporate state actions into the baseline for California and Colorado. Both states have
54 See Chapter 11 of the November 2021 TSD for details on costs associated with best management practices during
liquids unloading events, which is the compliance option we assume for this analysis. Additionally, see the memo
titled "Analysis of Greenhouse Gas Reporting Program Liquids Unloading Data," Docket ID No. EPA-HQ-OAR-
2021-0317-0143.
55 Emission factors are estimated by calculating the average VOC and methane emissions per barrel across the
sample tanks on the "Condensate" and "Oil" tabs from the docketed workbook, EPA-HQ-OAR-2021-0317-
003 9_attachment_21.
56 Note that V2 and V3 vintage sites are subject to the more stringent of NSPS OOOO and NSPS OOOOc, which we
assume is NSPS OOOO.
57 The cost functions above the threshold are estimated by fitting a quadratic function of methane emissions on cost
using data points for methane emissions of 50, 150, 300, and 1500 TPY in the "New" and "Existing" tabs from the
docketed workbook, EPA-HQ-OAR-2021-0317-003 9_attachment_20. The fitted cost functions imposed a constraint
that the functions be equal to the cost values from the workbook at an emissions rate of 50 TPY CH4.
47
-------
comprehensive regulatory programs for the oil and natural gas industry and contribute
significantly to national production levels. We have also incorporated fugitive monitoring
requirements in New Mexico and Pennsylvania into the baseline. By not accounting for state and
local requirements (outside of Colorado, California, New Mexico, and Pennsylvania) in the
baseline, this analysis may overestimate both the benefits and the costs of the proposed
regulation.
Specifically, we assume that California and Colorado have requirements at least as
stringent as those in the proposed rule for compressor station fugitives; natural gas processing
plant leaks; pneumatic controllers; pneumatic pumps in the production and gathering and
boosting segments; pre-0000 reciprocating and wet seal centrifugal compressors in the
gathering and boosting and processing segments; and storage vessels. In addition, we assume
California has requirements at least as stringent as those in the proposed rule for pre-0000
reciprocating and wet seal centrifugal compressors in the transmission and storage segments; and
post-0000 reciprocating and wet seal centrifugal compressors in all segments. We assume that
Colorado has requirements at least as stringent as those in the proposed rule for liquids
unloading. For well site fugitives, we assume California, Colorado, Pennsylvania, and New
Mexico have requirements at least as stringent as those in the proposed rule.
To incorporate the California, Colorado, New Mexico, and Pennsylvania rules in the
baseline, our activity data projections for sites and affected facilities need to estimate the counts
for those states. For the production segment, the processes described in Section 2.2.1.1 already
account for state level activity counts. For the other segments, midstream data from Enverus was
used to calculate the proportions of natural gas processing plants and compressor stations in
California and Colorado. We assume that those proportions hold fixed in all analysis years, and
that affected facilities are also distributed according to those proportions.
2.3 Emissions Reductions
Table 2-6 summarizes the emissions reductions associated with the proposed standards.
The emissions reductions are estimated by multiplying the source-level emissions reductions
associated with each applicable control and facility type by the number of affected sources of
that facility type. We present methane emissions in both short tons and CO2 Eq. using a global
warming potential of 25.
48
-------
Table 2-6 Projected Emissions Reductions under the Proposed NSPS OOOOb and EG
OOOOc Option, 2023-2035
Emissions Changes
Methane
Year
Methane
(short tons)
VOC
(short tons)
HAP
(short tons)
(metric tons CO2 Eq. using
GWP=25)
2023
140,000
61,000
2,300
3,300,000
2024
220,000
91,000
3,500
5,000,000
2025
300,000
120,000
4,600
6,900,000
2026
3,500,000
920,000
37,000
79,000,000
2027
3,500,000
930,000
38,000
79,000,000
2028
3,500,000
930,000
38,000
79,000,000
2029
3,500,000
940,000
38,000
79,000,000
2030
3,500,000
940,000
38,000
79,000,000
2031
3,500,000
950,000
38,000
79,000,000
2032
3,500,000
950,000
38,000
80,000,000
2033
3,500,000
950,000
39,000
80,000,000
2034
3,500,000
960,000
39,000
80,000,000
2035
3,500,000
960,000
39,000
80,000,000
Total
36,000,000
9,700,000
390,000
810,000,000
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
2.4 Product Recovery
The projected compliance costs presented below include the revenue from natural gas
recovery projected under the proposed standards. Requirements for fugitive emissions
monitoring, equipment leaks at processing plants, reciprocating and centrifugal compressors,
pneumatic devices, and liquids unloading events are assumed to increase the capture of methane
and VOC emissions that would otherwise be vented to the atmosphere, and we assume that a
large proportion of the averted methane emissions can be directed into natural gas production
streams and sold; see Chapters 2-5 of the Supplemental TSD and Chapters 7 and 10-11 of the
November 2021 TSD for details on the proportion of recovered emissions associated with the
compliance options.
Table 2-7 summarizes the increase in natural gas recovery and the associated revenue.
The AEO2022 projects Henry Hub natural gas prices rising from $3.49/MMBtu in 2023 to
$3.64/MMBtu in 2035 in 2021 dollars.58 To be consistent with other financial estimates in the
RIA, we adjust the projected prices in AEO2022 from 2021 dollars to 2019 dollars using the
58 Available at: https://www.eia.gov/outlooks/aeo/excel/aeotab_13.xlsx. Accessed July 25, 2022.
49
-------
GDP-Implicit Price Deflator. We also adjust prices for the wellhead using an EIA study that
indicated that the Henry Hub price is, on average, about 11 percent higher than the wellhead
price (Budzik, 2002). Finally, we use a conversion factor of 1.037 MMBtu equals 1 Mcf.59
Incorporating these adjustments, wellhead natural gas prices are assumed to rise from $3.09/Mcf
in 2023 to $3.22/Mcf in 2035.
Table 2-7 Projected Increase in Natural Gas Recovery under the Proposed NSPS
OOOOb and EG OOOOc Option, 2023-2035
Year
Increase in Gas Recovery (Bcf)
Increased Revenue
(millions 2019$)
2023
8.0
$25
2024
12
$35
2025
17
$45
2026
200
$520
2027
200
$540
2028
200
$570
2029
200
$590
2030
200
$610
2031
200
$630
2032
200
$630
2033
200
$650
2034
200
$650
2035
200
$650
Note: Values rounded to two significant figures.
Operators in the transmission and storage segment of the industry do not typically own
the natural gas they transport; rather, they receive payment for the transportation service they
provide. From a social perspective, however, the increased financial returns from natural gas
recovery accrues to entities somewhere along the natural gas supply chain and should be
accounted for in a national-level analysis. An economic argument can be made that, in the long
run, no single entity bears the entire burden of compliance costs or fully appropriates the
financial gain of the additional revenues associated with natural gas recovery. The change in
economic surplus resulting from natural gas recovery is likely to be spread across different
market participants. Therefore, the simplest and most transparent option for allocating these
59 For MMbtu-Mcf conversion factor, see https://www.eia.gov/outlooks/aeo/data/browser/#/?id=20-
AE02021&cases=ref2021&sourcekey=0. Accessed October 7, 2021.
50
-------
revenues would be to keep the compliance costs and revenues within a given source category and
not make assumptions regarding the allocation of costs and revenues across agents.
2.5 Compliance Costs
Table 2-8 summarizes the compliance costs and revenue from product recovery for the
evaluated emissions sources and points. Total costs consist of capital costs, annual operating and
maintenance costs, and revenue from product recovery. Capital costs include the capital costs
from the requirements on newly affected pneumatic devices, reciprocating compressors, and
storage vessels, as well as the planning costs associated with monitoring requirements for
fugitive emissions at well sites and compressor stations and equipment leaks at processing
plants; these costs are reincurred as operators are assumed to have to renew survey monitoring
plans or purchase new capital equipment at the end of its useful life. The annual operating and
maintenance costs are due to requirements on fugitive emissions and equipment leaks,
controllers at gas processing plants, compressors, liquids unloading events, and storage vessels.
Note that Table 2-8 shows a pulse of capital expenditures in 2026, the year the RIA
assumes to be the compliance year for the proposed EG OOOOc. In practice, however, the
proposed requirements give States and sources the flexibility to spread these installations over a
period of up to three years, or the 2025 to 2027 period. While we do not distribute compliance
expenditures across these years in the RIA, we believe that States and sources will avail
themselves of this flexibility.
51
-------
Table 2-8 Projected Compliance Costs under the Proposed NSPS OOOOb and EG
OOOOc Option, 2023-2035 (millions 2019$)
Compliance Costs
Year
Capital Costs
Operating and
Maintenance
Costs
Annualized
Costs
Increased Revenue
from Product
Recovery
Annualized Costs with
Increased Revenue from
Product Recovery
2023
$480
$66
$120
$25
$95
2024
$270
$100
$190
$35
$150
2025
$290
$140
$260
$45
$210
2026
$12,000
$1,300
$2,800
$520
$2,200
2027
$310
$1,300
$2,700
$540
$2,200
2028
$300
$1,300
$2,700
$570
$2,200
2029
$320
$1,300
$2,700
$590
$2,100
2030
$660
$1,300
$2,700
$610
$2,000
2031
$310
$1,300
$2,700
$630
$2,000
2032
$330
$1,300
$2,700
$630
$2,000
2033
$320
$1,300
$2,700
$650
$2,000
2034
$860
$1,400
$2,700
$650
$2,000
2035
$340
$1,400
$2,700
$650
$2,000
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
The expected lifetimes that capital and planning costs are incurred over differs across
affected facilities. The cost of designing, or redesigning, fugitive emissions monitoring programs
at well sites and compressor stations are assumed to occur every eight years, while the planning
cost associated with equipment leak surveys at processing plants are assumed to occur every five
years. Pneumatic device lifetimes are assumed to be 15 years, the lifetimes of solar PV panels
and batteries used to power electronic controllers are assumed to be 10 and four years,
respectively, and the lifetime of compressors used to power compressed air systems is assumed
to be six years. Rod-packing replacement at reciprocating compressors is assumed to happen
about every 3.3 years in the processing segment, 3.8 years in the gathering and boosting and
transmission segments, and 4.4 years in the storage segment.60 The capital costs in each year
outlined in Table 2-8 includes the estimated costs for newly affected sources in that year, plus
the costs for sources affected previously that have reached the end of their assumed economic
lifetime.
60 For the purposes of assigning unannualized capital costs of subsequent replacements to years, we round the
lifetimes for rod-packing to the nearest whole number.
52
-------
The calculation of total annualized costs proceeds as follows. Capital and planning costs
are annualized over their requisite expected lifetimes at an interest rate of 7 percent. These
annualized capital costs are then added to the annual operating and maintenance costs of the
requirements to get the total annualized costs without product recovery in each year. The value
of product recovery is then subtracted to get the total annualized costs with product recovery in
each year. Under this proposal, over 80 percent of revenue from the sale of captured natural gas
is projected to be earned by operators in the production and processing segments of the industry,
where we assume that the operators own the natural gas and will receive the financial benefit
from the captured natural gas. The remainder of the captured natural gas is captured within the
transmission and storage segment, where operators do not typically own the natural gas they
transport; rather, they receive payment for the transportation service they provide. In the RIA, we
treat these revenues as an offset to projected compliance costs, while the revenues may also be
considered as a benefit of the regulatory action. However, regardless of whether the revenue
from capture of natural gas is considered a compliance cost offset or a benefit, the net benefits
are equivalent.
We now present the compliance costs of the proposed NSPS OOOOb and EG 0000c in
a PV framework. The stream of the estimated costs for each year from 2023 through 2035 is
discounted back to 2021 using 3 and 7 percent discount rates and summed to get the PV of the
costs. The PV is then used to estimate the EAV of the estimated costs. The EAV is the single
annual value which, if summed in PV terms across years in the analytical time frame, equals the
PV of the original (i.e., likely time-varying) stream of costs. In other words, the EAV takes the
potentially "lumpy" stream of costs and converts them into a single value that, when discounted
and added together over each period in the analysis time frame, equals the original stream of
values in PV terms.
Table 2-9 shows the undiscounted stream of costs for each year from 2023 through 2035
due to the proposed standards. Capital costs are the projected capital and planning costs expected
to be incurred. Total costs are the sum of the capital costs and annual operating costs. The
revenue from the increase in product recovery is estimated using the AEO2022 natural gas price
projections, as described earlier. Total costs with revenue from product recovery equal the total
anticipated costs minus the revenue.
53
-------
Table 2-9 Undiscounted Projected Compliance Costs under the Proposed NSPS
OOOOb and EG OOOOc Option, 2023-2035 (millions 2019$)
Year
Capital Costs
Annual
Operating
Costs
Total Costs
(w/o Revenue)
Revenue from
Product
Recovery
Total Costs
(with Revenue)
2023
$480
$66
$550
$25
$520
2024
$270
$100
$370
$35
$340
2025
$290
$140
$430
$45
$390
2026
$12,000
$1,300
$13,000
$520
$12,000
2027
$310
$1,300
$1,600
$540
$1,100
2028
$300
$1,300
$1,600
$570
$1,000
2029
$320
$1,300
$1,600
$590
$1,000
2030
$660
$1,300
$2,000
$610
$1,400
2031
$310
$1,300
$1,600
$630
$1,000
2032
$330
$1,300
$1,700
$630
$1,000
2033
$320
$1,300
$1,700
$650
$1,000
2034
$860
$1,400
$2,200
$650
$1,600
2035
$340
$1,400
$1,700
$650
$1,100
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
Table 2-10 shows the discounted stream of costs discounted to 2021 using a 3 and 7
percent discount rate. The PV of the stream of costs discounted to 2021 using a 3 percent
discount rate is $19 billion, with an EAV of $1.8 billion per year. The PV of the stream of costs
discounted to 2021 using a 7 percent discount rate is $15 billion, with an EAV of $1.8 billion per
year.
54
-------
Table 2-10 Discounted Projected Costs under the Proposed NSPS OOOOb and EG
OOOOc Option, 2023-2035 (millions 2019$)
3 Percent
7 Percent
Year
Total Annual
Cost (w/o
Product
Recovery
Revenue)
Revenue
from
Product
Recovery
Total Annual
Costs (w/
Product
Recovery
Revenue)
Total Annual
Cost (w/o
Product
Recovery
Revenue)
Revenue
from
Product
Recovery
Total Annual
Cost (w/
Product
Recovery
Revenue)
2023
$100
$23
$80.0
$100
$22
$83.0
2024
$160
$32
$120.0
$150
$28
$120.0
2025
$210
$40
$170.0
$200
$34
$160.0
2026
$2,200
$450
$1,700
$2,000
$370
$1,600
2027
$2,100
$450
$1,600
$1,800
$360
$1,500
2028
$2,000
$460
$1,600
$1,700
$360
$1,400
2029
$2,000
$470
$1,500
$1,600
$340
$1,200
2030
$1,800
$470
$1,400
$1,400
$330
$1,100
2031
$1,800
$470
$1,300
$1,400
$320
$1,000
2032
$1,700
$460
$1,300
$1,300
$300
$960
2033
$1,700
$450
$1,200
$1,200
$290
$890
2034
$1,600
$440
$1,200
$1,100
$270
$840
2035
$1,600
$430
$1,200
$1,000
$250
$780
PV
$19,000
$4,600
$14,000
$15,000
$3,300
$12,000
EAV
$1,800
$440
$1,400
$1,800
$390
$1,400
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding. Costs and
revenue from product recovery in each year are discounted to 2021.
2.6 Comparison of Regulatory Alternatives
In this section, we compare the compliance cost and emissions impacts projected under
the proposal with the results of two alternative regulatory scenarios, one less stringent and one
more stringent than the proposed rule. The alternative scenarios focus on the sources that account
for the largest number of estimated emissions reductions of methane and/or VOC for the
proposed rule: well site fugitives and pneumatic devices at well sites.
The alternative scenarios are summarized in Table 2-11. The NSPS OOOOa established a
standard of performance of 95% control for diaphragm pumps at well sites with existing
combustion devices, which we do not include in our baseline due to a lack of information
regarding which sites would be subject to the requirement. We believe this results in a slight over
estimate of the impacts of the proposal and more stringent scenarios. In the less stringent
scenario, we estimate the impacts of not including the AVO requirements for fugitive emissions
monitoring at well sites, extending the current NSPS of an emissions limit for continuous-bleed
55
-------
controllers to pre-0000 (for well sites and gathering and boosting stations) or pre-OOOOa (for
transmission and storage compressor station) sources rather than requiring zero emitting
controllers, and omitting requirements for pneumatic pumps rather than requiring zero emitting
pumps. In the more stringent scenario, we illustrate the impact of the small well site OGI
exemption for by requiring well sites with a single piece of major equipment to perform
semiannual OGI in addition to quarterly AVO. These alternatives reflect key regulatory design
alternatives that the EPA grappled with while developing the proposal.
56
-------
Table 2-11 Summary of Regulatory Alternatives
Source
Applicable
NSPS
NSPS Baseline
Less Stringent
Proposal
More Stringent
Fugitive Emissions at Well Sites
Wellhead only, single wellsite
OOOOa
No requirement
No requirement
Quarterly AVO
Quarterly AVO
Wellhead only, multiple well site
OOOOa
No requirement
Semiannual OGI
Quarterly AVO +
Semiannual OGI
Quarterly AVO +
Semiannual OGI
Single well site with a single price of major
equipment and no tank battery
OOOOa
Semiannual OGI
No requirement
Quarterly AVO
Quarterly AVO +
Semiannual OGI
Multiple well site with a single piece of major
equipment, or any site with two or more pieces
of major equipment or one piece of major
equipment and a tank battery
OOOOa
Semiannual OGI
Quarterly OGI
Bimonthly AVO +
Quarterly OGI
Bimonthly AVO +
Quarterly OGI
Pneumatic Controllers
Natural gas bleed
Natural gas bleed
Well Sites and Gathering and Boosting Stations
0000
rate no greater than
6 scfh
rate no greater than
6 scfh
Zero emissions
Zero emissions
Natural gas bleed
Natural gas bleed
Transmission and Storage Compressor Stations
OOOOa
rate no greater than
6 scfh
rate no greater than
6 scfh
Zero emissions
Zero emissions
Pneumatic Pumps
Well Sites
OOOOa
No requirement3
No requirement
Zero emissions
Zero emissions
Gathering and Boosting Stations
None
No requirement
No requirement
Zero emissions
Zero emissions
a The NSPS OOOOa established a standard of performance of 95% control for diaphragm pumps at well sites with existing combustion devices, which we do not
include in our baseline due to a lack of information regarding which sites would be subject to the requirement. We believe this results in a slight over estimate of
the impacts of the proposal and more stringent scenarios.
57
-------
A comparison of estimated costs and emissions reductions is presented in Table 2-12 for
three years: 2023 (the first year of NSPS OOOOb impacts), 2026 (the first year of EG 0000c
impacts), and 2035 (the last year of analysis). Overall, the table demonstrates that we estimate
the impacts of EG 0000c to be much greater than those of the NSPS OOOOb for all regulatory
alternatives. By the time the EG 0000c is assumed to begin having an effect in 2026, we
estimate that the less stringent option would result in roughly one-third of the methane and VOC
emissions reductions of the proposed option, while reducing costs by about half. On the other
hand, we estimate that the more stringent option would result in slightly more methane and VOC
emissions reductions and slightly higher costs than the proposed option. Note that since the EG
0000c regulates emissions of methane, additional benefits to the regulation result from
associated reductions in VOC emissions.
58
-------
Table 2-12 Comparison of Regulatory Alternatives in 2023, 2026, and 2035 for the
Proposed NSPS OOOOb and EG OOOOc (millions 2019$)
Regulatory Alternative
Less Stringent
Proposal
More Stringent
Emissions reductions
Methane (short tons)
VOC (short tons)
30,000
30,000
Total Impacts. 2023
140,000
61,000
140,000
61,000
Costs
Annualized Costs without
Product Recovery (3%)
Annualized Costs with
Product Recovery (3%)
$76
$72
$110
$85
$110
Emissions reductions
Methane (short tons)
VOC (short tons)
1,300,000
320,000
Total Impacts. 2026
3,500,000
920,000
3,500,000
930,000
Costs
Annualized Costs without
Product Recovery (3%)
Annualized Costs with
Product Recovery (3%)
$1,200
$970
$2,500
$2,000
$2,600
$2,100
Emissions reductions
Methane (short tons)
VOC (short tons)
1,200,000
330,000
Total Impacts. 2035
3,500,000
960,000
3,500,000
970,000
Costs
Annualized Costs without
Product Recovery (3%)
Annualized Costs with
Product Recovery (3%)
$1,300
$1,100
$2,400
$1,800
$2,500
$1,800
59
-------
3 BENEFITS
The proposed NSPS OOOOb and EG 0000c are projected to reduce methane, VOC,
and HAP emissions.61 The total emissions reductions over the 2023-2035 period are estimated to
be about 36 million short tons of methane, 9.7 million tons of VOC, and 0.39 million tons of
HAP. The decrease in methane emissions in C02-equivalent (CO2 Eq.) terms is estimated to be
about 810 million metric tons using a global warming potential of 25.
We monetize the impacts of methane reductions in this RIA. We estimate the climate
benefits under the proposal using interim estimates of the social cost of methane (SC-CH4), as
presented in Section 3.2.
In addition to presenting monetized estimates of impacts from methane reductions, we
also provide a qualitative discussion of potential climate, human health, and welfare impacts of
emissions reductions we are unable to quantify and monetize. Table 3-1 summarizes the
quantified and unquantified benefits in this analysis. We also present a supplemental illustrative
screening analysis of quantified and monetized ozone-related health impacts of VOC reductions
based on a national benefit-per-ton methodology in Appendix C. Additional benefits to EG
OOOOc, which regulates methane emissions, result from associated reductions in VOC
emissions.
61 Some control techniques of the proposed action, such as routing emission to combustion devices, are also
anticipated to have minor disbenefits resulting from secondary emissions of carbon dioxide (C02), nitrogen oxides
(NOX), PM, carbon monoxide (CO), and total hydrocarbons (THC).
60
-------
Table 3-1 Climate and Human Health Effects of the Projected Emissions Reductions
from this Proposal
Effect
Effect
More
Category
Effect
Quantified
Monetized
Information
Environment
Climate impacts from
methane (CH4)
a
~
Section 3.2
Climate
effects
Other climate impacts (e.g.,
ozone, black carbon,
aerosols, other impacts)
—
—
IPCC,
Ozone ISA,
PM ISA
Human
Health
Mortality
from
Premature respiratory
mortality from short-term
exposure (0-99)
—
—
Ozone ISA
exposure to
ozone62
Premature respiratory
mortality from long-term
exposure (age 30-99)
—
—
Ozone ISA
Hospital admissions—
respiratory (ages 65-99)
—
—
Ozone ISA
Emergency department
visits—respiratory (ages 0-
99)
—
—
Ozone ISA
Asthma onset (0-17)
—
—
Ozone ISA
Asthma
symptoms/exacerbation
(asthmatics age 5-17)
—
—
Ozone ISA
Allergic rhinitis (hay fever)
symptoms (ages 3-17)
—
—
Ozone ISA
morbidity
Minor restricted-activity
days (age 18-65)
—
—
Ozone ISA
exposure to
School absence days (age
5-17)
—
—
Ozone ISA
UZiUllv
Decreased outdoor worker
productivity (age 18-65)
—
—
Ozone ISAb
Metabolic effects (e.g.,
diabetes)
—
—
Ozone ISAb
Other respiratory effects
(e.g., premature aging of
lungs)
—
—
Ozone ISAb
Cardiovascular and nervous
system effects
—
—
Ozone ISAb
Reproductive and
developmental effects
—
—
Ozone ISAb
Premature
mortality
Adult premature mortality
from long-term exposure
(age 65-99 or age 30-99)
—
—
PM ISA
exposure to
pm25
Infant mortality (age <1)
—
—
PM ISA
Heart attacks (age >18)
—
—
PM ISA
62 We present a supplemental illustrative analysis of quantified and monetized ozone-related health impacts of VOC
reductions based on a national benefit-per-ton methodology in Appendix C.
63 Ibid.
61
-------
Category
Nonfatal
morbidity
from
exposure to
PM25
Effect
Effect
More
Effect
Quantified
Monetized
Information
Hospital admissions—
cardiovascular (ages 65-99)
—
—
PM ISA
Emergency department
visits— cardiovascular (age
PM ISA
0-99)
Hospital admissions—
respiratory (ages 0-18 and
65-99)
—
—
PM ISA
Emergency room visits—
respiratory (all ages)
—
—
PM ISA
Cardiac arrest (ages 0-99;
excludes initial hospital
and/or emergency
department visits)
—
—
PM ISA
Stroke (ages 65-99)
—
—
PM ISA
Asthma onset (ages 0-17)
—
—
PM ISA
Asthma
symptoms/exacerbation (6-
17)
—
—
PM ISA
Lung cancer (ages 30-99)
—
—
PM ISA
Allergic rhinitis (hay fever)
symptoms (ages 3-17)
—
—
PM ISA
Lost work days (age 18-65)
—
—
PM ISA
Minor restricted-activity
days (age 18-65)
—
—
PM ISA
Hospital admissions—
Alzheimer's disease (ages
PM ISA
65-99)
Hospital admissions—
Parkinson's disease (ages
PM ISA
65-99)
Other cardiovascular effects
(e.g., other ages)
—
—
PM ISAb
Other respiratory effects
(e.g., pulmonary function,
non-asthma ER visits, non-
PM ISAb
bronchitis chronic diseases,
other ages and populations)
Other nervous system
effects (e.g., autism,
cognitive decline,
dementia)
—
—
PM ISAb
Metabolic effects (e.g.,
diabetes)
—
—
PM ISAb
Reproductive and
developmental effects (e.g.,
low birth weight, pre-term
births, etc.)
—
—
PM ISAb
Cancer, mutagenicity, and
genotoxicity effects
—
—
PM ISAb
62
-------
Effect
Effect
More
Category
Effect
Quantified
Monetized
Information
Incidence of
morbidity
from
exposure to
Effects associated with
exposure to hazardous air
pollutants such as benzene
—
—
ATSDR,
IRISc-d
HAP
a The climate and related impacts of CH4 emissions changes, such as sea level rise, are estimated within each
integrated assessment model as part of the calculation of the SC-CH4. The resulting monetized damages, which are
relevant for conducting the benefit-cost analysis, are used in this RIA to estimate the welfare effects of quantified
changes in methane emissions.
bNot quantified due to data availability limitations and/or because current evidence is only suggestive of causality.
0 We assess these benefits qualitatively because we do not have sufficient confidence in available data or methods.
d We assess these benefits qualitatively due to data limitations for this analysis, but we have quantified them in other
analyses.
3.1 Emissions Reductions
Oil and natural gas operations in the U.S. include a variety of emission sources for
methane, VOC, and HAP, including wells, well sites, processing plants, compressor stations,
storage equipment, and natural gas transmission and distribution lines. These emission points are
located throughout much of the country, though many of these emissions sources are
concentrated in particular geographic regions. For example, wells and processing plants are
largely concentrated in the South Central, Midwest, and Southern California regions of the U.S.,
whereas natural gas compressor stations are located all over the country. Distribution lines to
customers are frequently located within areas of high population density.
Table 3-2 shows the emissions reductions projected under the proposed NSPS OOOOb
and EG 0000c over the 2023-2035 period. We present methane emissions in both short tons
and CO2 Eq. using a global warming potential of 25. The impacts of these pollutants accrue at
different spatial scales. HAP emissions increase exposure to carcinogens and other toxic
pollutants primarily near the emission source. VOC emissions are precursors to secondary
formation of PM2.5 and ozone on a broader regional scale. Climate effects associated with long-
lived greenhouse gases like methane generally do not depend on the location of the emission of
the gas and have global impacts. Methane is also a precursor to global background
concentrations of ozone.
63
-------
Table 3-2 Projected Annual Reductions of Methane, VOC, and HAP Emissions under
the Proposed NSPS OOOOb and EG OOOOc Option, 2023-2035
Year
Methane
(short tons)
VOC
(short tons)
HAP
(short tons)
Methane
(metric tons CO2 Eq.)
2023
140,000
61,000
2,300
3,300,000
2024
220,000
91,000
3,500
5,000,000
2025
300,000
120,000
4,600
6,900,000
2026
3,500,000
920,000
37,000
79,000,000
2027
3,500,000
930,000
38,000
79,000,000
2028
3,500,000
930,000
38,000
79,000,000
2029
3,500,000
940,000
38,000
79,000,000
2030
3,500,000
940,000
38,000
79,000,000
2031
3,500,000
950,000
38,000
79,000,000
2032
3,500,000
950,000
38,000
80,000,000
2033
3,500,000
950,000
39,000
80,000,000
2034
3,500,000
960,000
39,000
80,000,000
2035
3,500,000
960,000
39,000
80,000,000
Total
36,000,000
9,700,000
390,000
810,000,000
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
3.2 Methane Climate Effects and Valuation
Methane is the principal component of natural gas. Methane is also a potent greenhouse
gas (GHG) that, once emitted into the atmosphere, absorbs terrestrial infrared radiation, which in
turn contributes to increased global warming and continuing climate change. Methane reacts in
the atmosphere to form ozone, which also impacts global temperatures. Methane, in addition to
other GHG emissions, contributes to warming of the atmosphere, which over time leads to
increased air and ocean temperatures, changes in precipitation patterns, melting and thawing of
global glaciers and ice sheets, increasingly severe weather events, such as hurricanes of greater
intensity, and sea level rise, among other impacts.
According to the Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment
Report (IPCC, 2021), radiative forcing due to methane relative to the year 1750 was 0.54 W/m2
in 2019, which is about 16 percent of all global forcing due to increases in anthropogenic GHG
concentrations, and which makes methane the second leading long-lived climate forcer after
64
-------
C02.64 After accounting for changes in other greenhouse substances such as ozone and
stratospheric water vapor due to chemical reactions of methane in the atmosphere, historical
methane emissions account for about 0.5 degrees of warming today, or about one third of the
total warming resulting from historical emissions of well-mixed GHGs.
The oil and natural gas sector emits significant quantities of methane. The U.S. Inventory
of Greenhouse Gas Emissions and Sinks: 1990-2019 (published 2021) estimates 2019 methane
emissions from Petroleum and Natural Gas Systems (not including petroleum refineries,
petroleum transportation, and natural gas distribution) to be 187 million metric tons CO2 Eq. In
2019, total methane emissions from the oil and natural gas industry represented 27 percent of the
total methane emissions from all sources and account for about 3 percent of all CO2 Eq.
emissions in the U.S., with the combined petroleum and natural gas systems being the largest
contributor to U.S. anthropogenic methane emissions (U.S. EPA, 2021d).
We estimate the climate benefits of CH4 emissions reductions expected from this
proposed rule using the SC-CH4 estimates presented in the Technical Support Document: Social
Cost of Carbon, Methane, and Nitrous Oxide Interim Estimates under Executive Order 13990
published in February 2021 by the Interagency Working Group on the Social Cost of Greenhouse
Gases (IWG) (IWG, 2021). The SC-CH4 is the monetary value of the net harm to society
associated with a marginal increase in emissions in a given year, or the benefit of avoiding that
increase. In principle, SC-CH4 includes the value of all climate change impacts, including (but
not limited to) changes in net agricultural productivity, human health effects, property damage
from increased flood risk and natural disasters, disruption of energy systems, risk of conflict,
environmental migration, and the value of ecosystem services. The SC-CH4 therefore, reflects
the societal value of reducing emissions of the gas in question by one metric ton. The SC-CH4 is
the theoretically appropriate value to use in conducting benefit-cost analyses of policies that
affect CH4 emissions. As a member of the IWG involved in the development of the February
2021 SC-GHG TSD, the EPA agrees that the interim SC-GHG estimates represent the most
appropriate estimate of the SC-GHG until revised estimates have been developed reflecting the
64 Increased concentrations of methane and other well mixed greenhouse gases in the atmosphere absorb thermal
infrared emission energy, reducing the rate at which the Earth can cool through radiating heat to space. Radiative
forcing, measured as watts per square meter (W/m2), is a measure of the climate impact of greenhouse gases and
other human activities.
65
-------
latest, peer-reviewed science. While the IWG's SC-GHG review and updating process under EO
13990 continues, in Appendix B of this RIA the EPA presents a sensitivity analysis of the
monetized climate benefits using a set of SC-CH4 estimates that incorporates recent research
addressing recommendations of the National Academies of Sciences, Engineering, and Medicine
(2017).
The SC-CH4 estimates presented in the February 2021 SC-GHG TSD were developed
over many years, using transparent process, peer-reviewed methodologies, the best science
available at the time of that process, and with input from the public. Specifically, in 2009, an
interagency working group (IWG) that included the EPA and other executive branch agencies
and offices was established to ensure that agencies had access to the best available information
when quantifying the benefits of reducing CO2 emissions in benefit-cost analyses. The IWG
published SC-CO2 estimates in 2010 that were developed from an ensemble of three widely cited
integrated assessment models (IAMs) that estimate climate damages using highly aggregated
representations of climate processes and the global economy combined into a single modeling
framework. The three IAMs were run using a common set of input assumptions in each model
for future population, economic, and CO2 emissions growth, as well as equilibrium climate
sensitivity (ECS) — a measure of the globally averaged temperature response to increased
atmospheric CO2 concentrations. These estimates were updated in 2013 based on new versions
of each IAM.65 In August 2016 the IWG published estimates of the social cost of methane (SC-
CH4) and nitrous oxide (SC-N2O) using methodologies that are consistent with the methodology
underlying the SC-CO2 estimates. The modeling approach that extends the IWG SC-CO2
methodology to non-CC>2 GHGs has undergone multiple stages of peer review. The SC-CH4 and
SC-N2O estimates were developed by Marten, Kopits, Griffiths, Newbold, and Wolverton (2015)
and underwent a standard double-blind peer review process prior to journal publication. These
estimates were applied in regulatory impact analyses of EPA proposed rulemakings with CH4
65 Dynamic Integrated Climate and Economy (DICE) 2010 (Nordhaus, 2010), Climate Framework for Uncertainty,
Negotiation, and Distribution (FUND) 3.8 (Anthoff & Tol, 2013a, 2013b), and Policy Analysis of the Greenhouse
Gas Effect (PAGE) 2009 (Hope, 2013). Dynamic Integrated Climate and Economy (DICE) 2010 (Nordhaus, 2010),
Climate Framework for Uncertainty, Negotiation, and Distribution (FUND) 3.8 (Anthoff & Tol, 2013a, 2013b), and
Policy Analysis of the Greenhouse Gas Effect (PAGE) 2009 (Hope, 2013).
66
-------
and N2O emissions impacts.66 The EPA also sought additional external peer review of technical
issues associated with its application to regulatory analysis. Following the completion of the
independent external peer review of the application of the Marten et al. (2015) estimates, the
EPA began using the estimates in the primary benefit-cost analysis calculations and tables for a
number of proposed rulemakings in 2015 (EPA 2015b, 2015c). The EPA considered and
responded to public comments received for the proposed rulemakings before using the estimates
in final regulatory analyses in 2016.67 In 2015, as part of the response to public comments
received to a 2013 solicitation for comments on the SC-CO2 estimates, the IWG announced a
National Academies of Sciences, Engineering, and Medicine review of the SC-CO2 estimates to
offer advice on how to approach future updates to ensure that the estimates continue to reflect the
best available science and methodologies. In January 2017, the National Academies released
their final report, Valuing Climate Damages: Updating Estimation of the Social Cost of Carbon
Dioxide, and recommended specific criteria for future updates to the SC-CO2 estimates, a
modeling framework to satisfy the specified criteria, and both near-term updates and longer-term
research needs pertaining to various components of the estimation process (National Academies,
2017). Shortly thereafter, in March 2017, President Trump issued EO 13783, which disbanded
the IWG, withdrew the previous TSDs, and directed agencies to "ensure" SC-GHG estimates
used in regulatory analyses "are consistent with the guidance contained in OMB Circular A-4",
"including with respect to the consideration of domestic versus international impacts and the
consideration of appropriate discount rates" (EO 13783, Section 5(c)). Benefit-cost analyses
following EO 13783, including the benefit-cost analysis for the Oil and Natural Gas Technical
Reconsideration and Policy Review RIA,68 (U.S. EPA, 2020c) used SC-GHG estimates that
attempted to focus on the specific share of physical climate change damages in the U.S. as
captured by the models (which did not reflect many pathways by which climate impacts affect
the welfare of U.S. citizens and residents) and were calculated using two default discount rates
66 The SC-CH4 and SC-N20 estimates were first used in sensitivity analysis for the Proposed Rulemaking for
Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles-
Phase 2 (U.S. EPA, 2015).
67 See IWG (2016b) for more discussion of the SC-CH4 and SC-N20 and the peer review and public comment
processes accompanying their development.
68 The values used in the rule RIA were interim values developed under EO 13783 for use in regulatory analyses.
EPA followed EO 13783 by using SC-CO2 estimates reflecting impacts occurring within U.S. borders and 3% and
7% discount rates in our central analysis for the proposal RIA.
67
-------
recommended by Circular A-4 (2003), 3 percent and 7 percent.69 All other methodological
decisions and model versions used in the SC-GHG calculations remained the same as those used
by the IWG in 2010 and 2013, respectively.
On January 20, 2021, President Biden issued EO 13990, which established an IWG and
directed the group to develop an update of the SC-GHG estimates that reflect the best available
science and the recommendations of National Academies (2017). In February 2021, the IWG
recommended the interim use of the most recent SC-GHG estimates developed by the IWG prior
to the group being disbanded in 2017, adjusted for inflation (IWG, 2021). As discussed in the
February 2021 TSD, the IWG's selection of these interim estimates reflected the immediate need
to have SC-GHG estimates available for agencies to use in regulatory benefit-cost analyses and
other applications that were developed using a transparent process, peer reviewed
methodologies, and the science available at the time of that process. The February 2021 update
also recognized the limitations of the interim estimates and encouraged agencies to use their best
judgment in, for example, considering sensitivity analyses using lower discount rates. The IWG
published a Federal Register notice on May 7, 2021, soliciting comment on the February 2021
TSD and on how best to incorporate the latest peer-reviewed scientific literature in order to
develop an updated set of SC-GHG estimates. The EPA has applied the IWG's interim SC-GHG
estimates in regulatory analyses published since the release of the February 2021 TSD, including
in the November 2021 Proposal RIA, and is likewise using them in the primary benefit-cost
analysis calculations in this supplemental proposal RIA. While the IWG's SC-GHG review and
updating process under EO 13990 continues, the EPA also presents in Appendix B of this RIA a
sensitivity analysis of the monetized climate benefits using SC-CH4 estimates newly developed
by EPA that incorporate recent research addressing recommendations of the National Academies
of Sciences, Engineering, and Medicine (2017).
69 EPA regulatory analyses under E.O. 13783 included sensitivity analyses based on global SC-GHG values and
using a lower discount rate of 2.5%. OMB Circular A-4 (OMB, 2003) recognizes that special considerations arise
when applying discount rates if intergenerational effects are important. In the IWG's 2015 Response to Comments,
OMB—as a co-chair of the IWG—made clear that "Circular A-4 is a living document," that "the use of 7 percent is
not considered appropriate for intergenerational discounting," and that "[t]here is wide support for this view in the
academic literature, and it is recognized in Circular A-4 itself." OMB, as part of the IWG, similarly repeatedly
confirmed that "a focus on global SCC estimates in [regulatory impact analyses] is appropriate" (IWG 2015).
68
-------
The February 2021 SC-GHG TSD provides a complete discussion of the IWG's initial
review conducted under EO 13990. In particular, the IWG found that the SC-GHG estimates
used under EO 13783 fail to reflect the full impact of GHG emissions in multiple ways. First, the
IWG concluded that those estimates fail to capture many climate impacts that can affect the
welfare of U.S. citizens and residents. Examples of affected interests include direct effects on
U.S. citizens and assets located abroad, international trade, and tourism, and spillover pathways
such as economic and political destabilization and global migration that can lead to adverse
impacts on U.S. national security, public health, and humanitarian concerns. Those impacts are
better captured within global measures of the social cost of greenhouse gases.
In addition, assessing the benefits of U.S. GHG mitigation activities requires
consideration of how those actions may affect mitigation activities by other countries, as those
international mitigation actions will provide a benefit to U.S. citizens and residents by mitigating
climate impacts that affect U.S. citizens and residents. A wide range of scientific and economic
experts have emphasized the issue of reciprocity as support for considering global damages of
GHG emissions. Using a global estimate of damages in U.S. analyses of regulatory actions
allows the U.S. to continue to actively encourage other nations, including emerging major
economies, to take significant steps to reduce emissions. The only way to achieve an efficient
allocation of resources for emissions reduction on a global basis — and so benefit the U.S. and
its citizens — is for all countries to base their policies on global estimates of damages.
As a member of the IWG involved in the development of the February 2021 SC-GHG
TSD, EPA agrees with this assessment and, therefore, in this proposed rule the EPA centers
attention on a global measure of SC-CH4. This approach is the same as that taken in EPA
regulatory analyses over 2009 through 2016. A robust estimate of climate damages only to U.S.
citizens and residents that accounts for the myriad of ways that global climate change reduces the
net welfare of U.S. populations does not currently exist in the literature. As explained in the
February 2021 TSD, existing estimates are both incomplete and an underestimate of total
damages that accrue to the citizens and residents of the U.S. because they do not fully capture the
regional interactions and spillovers discussed above, nor do they include all of the important
physical, ecological, and economic impacts of climate change recognized in the climate change
literature, as discussed further below. The EPA, as a member of the IWG, will continue to review
developments in the literature, including more robust methodologies for estimating the
69
-------
magnitude of the various damages to U.S. populations from climate impacts and reciprocal
international mitigation activities, and explore ways to better inform the public of the full range
of carbon impacts.70
Second, the IWG concluded that the use of the social rate of return on capital (7 percent
under current OMB Circular A-4 guidance) to discount the future benefits of reducing GHG
emissions inappropriately underestimates the impacts of climate change for the purposes of
estimating the SC-GHG. Consistent with the findings of National Academies (2017) and the
economic literature, the IWG continued to conclude that the consumption rate of interest is the
theoretically appropriate discount rate in an intergenerational context (IWG, 2010, 2013, 2016a,
2016b), and recommended that discount rate uncertainty and relevant aspects of intergenerational
ethical considerations be accounted for in selecting future discount rates.71 Furthermore, the
damage estimates developed for use in the SC-GHG are estimated in consumption-equivalent
terms, and so an application of OMB Circular A-4's guidance for regulatory analysis would then
use the consumption discount rate to calculate the SC-GHG. As a member of the IWG involved
in the development of the February 2021 SC-GHG TSD, the EPA agrees with this assessment
and will continue to follow developments in the literature pertaining to this issue. EPA also notes
that while OMB Circular A-4, as published in 2003, recommends using 3 percent and 7 percent
discount rates as "default" values, Circular A-4 also reminds agencies that "different regulations
may call for different emphases in the analysis, depending on the nature and complexity of the
regulatory issues and the sensitivity of the benefit and cost estimates to the key assumptions." On
discounting, Circular A-4 recognizes that "special ethical considerations arise when comparing
benefits and costs across generations," and Circular A-4 acknowledges that analyses may
appropriately "discount future costs and consumption benefits.. .at a lower rate than for
70 For further discussion of EPA's focus on global estimates of SC-CH4, see the supporting material for this entitled
Report on the Social Cost of Greenhouse Gases: Estimates Incorporating Recent Scientific Advances (EPA 2022) in
the docket.
71 GHG emissions are stock pollutants, with damages associated with what has accumulated in the atmosphere over
time, and they are long lived such that subsequent damages resulting from emissions today occur over many decades
or centuries depending on the specific greenhouse gas under consideration. In calculating the SC-GHG, the stream
of future damages to agriculture, human health, and other market and non-market sectors from an additional unit of
emissions are estimated in terms of reduced consumption (or consumption equivalents). Then that stream of future
damages is discounted to its present value in the year when the additional unit of emissions was released. Given the
long time horizon over which the damages are expected to occur, the discount rate has a large influence on the
present value of future damages.
70
-------
intragenerational analysis." In the 2015 Response to Comments on the Social Cost of Carbon for
Regulatory Impact Analysis, OMB, EPA, and the other IWG members recognized that "Circular
A-4 is a living document" and "the use of 7 percent is not considered appropriate for
intergenerational discounting. There is wide support for this view in the academic literature, and
it is recognized in Circular A-4 itself." Thus, EPA concludes that a 7 percent discount rate is not
appropriate to apply to value the social cost of greenhouse gases in the analysis presented in this
analysis. In this analysis, to calculate the present and annualized values of climate benefits, EPA
uses the same discount rate as the rate used to discount the value of damages from future GHG
emissions, for internal consistency. That approach to discounting follows the same approach that
the February 2021 SC-GHG TSD recommends "to ensure internal consistency — i.e., future
damages from climate change using the SC-GHG at 2.5 percent should be discounted to the base
year of the analysis using the same 2.5 percent rate." EPA has also consulted the National
Academies' 2017 recommendations on how SC-GHG estimates can "be combined in RIAs with
other cost and benefits estimates that may use different discount rates." The National Academies
reviewed "several options," including "presenting all discount rate combinations of other costs
and benefits with [SC-GHG] estimates."
While the IWG works to assess how best to incorporate the latest, peer reviewed science
to develop an updated set of SC-GHG estimates, it recommends the interim estimates to be the
most recent estimates developed by the IWG prior to the group being disbanded in 2017. The
estimates rely on the same models and harmonized inputs and are calculated using a range of
discount rates. As explained in the February 2021 SC-GHG TSD, the IWG has concluded that it
is appropriate for agencies to revert to the same set of four values drawn from the SC-GHG
distributions based on three discount rates as were used in regulatory analyses between 2010 and
2016 and subject to public comment. For each discount rate, the IWG combined the distributions
across models and socioeconomic emissions scenarios (applying equal weight to each) and then
selected a set of four values for use in benefit-cost analyses: an average value resulting from the
model runs for each of three discount rates (2.5 percent, 3 percent, and 5 percent), plus a fourth
value, selected as the 95th percentile of estimates based on a 3 percent discount rate. The fourth
value was included to provide information on potentially higher-than-expected economic impacts
from climate change, conditional on the 3 percent estimate of the discount rate. As explained in
the February 2021 SC-GHG TSD, and EPA agrees, this update reflects the immediate need to
71
-------
have an operational SC-GHG for use in regulatory benefit-cost analyses and other applications
that was developed using a transparent process, peer-reviewed methodologies, and the science
available at the time of that process. Those estimates were subject to public comment in the
context of dozens of proposed rulemakings as well as in a dedicated public comment period in
2013.
Table 3-3 summarizes the interim SC-CH4 estimates across all the model runs for each
discount rate for emissions occurring in 2023 to 2035. These estimates are reported in 2019
dollars but are otherwise identical to those presented in the IWG's 2016 TSD (IWG, 2016b). For
purposes of capturing uncertainty around the SC-CH4 estimates in analyses, the IWG's February
2021 SC-GHG TSD emphasizes the importance of considering all four of the SC-CH4 values.
The SC-CH4 increases over time within the models — i.e., the societal harm from one metric ton
emitted in 2030 is higher than the harm caused by one metric ton emitted in 2025 — because
future emissions produce larger incremental damages as physical and economic systems become
more stressed in response to greater climatic change, and because GDP is growing over time and
many damage categories are modeled as proportional to GDP.
Table 3-3 Interim Estimates of the Social Cost of CH4, 2023-2035 (in 2019$ per metric
ton CH4)
Discount Rate and Statistic
5%
3%
2.5%
3%
Year
Average
Average
Average
95th Percentile
2023
$740
$1,600
$2,100
$4,200
2024
$770
$1,700
$2,100
$4,400
2025
$790
$1,700
$2,200
$4,500
2026
$820
$1,700
$2,300
$4,600
2027
$850
$1,800
$2,300
$4,700
2028
$870
$1,800
$2,400
$4,900
2029
$900
$1,900
$2,400
$5,000
2030
$930
$1,900
$2,500
$5,100
2031
$960
$2,000
$2,500
$5,300
2032
$990
$2,000
$2,600
$5,400
2033
$1,000
$2,100
$2,700
$5,600
2034
$1,100
$2,200
$2,700
$5,700
2035
$1,100
$2,200
$2,800
$5,900
Source: Technical Support Document: Social Cost of Carbon, Methane, and Nitrous Oxide Interim Estimates under
EO 13990 (IWG, 2021).
Note: These SC-CH4 values are identical to those reported in the 2016 TSD (IWG, 2016b) adjusted for inflation to
2019 dollars using the annual GDP Implicit Price Deflator values in the U.S. Bureau of Economic Analysis' (BEA)
NIPA Table 1.1.9 (U.S. BEA, 2021). The values are stated in $/metric tonne CH4 and vary depending on the year of
72
-------
CH4 emissions. This table displays the values rounded to the nearest dollar; the annual unrounded values used in the
calculations in this RIA are available on OMB's website: https://www.whitehouse.gov/briefing-
room/blog/2021/02/26/a-return-to-science-evidence-based-estimates-of-the-benefits-of-reducing-climate-pollution/.
Figure 3-1 presents the quantified sources of uncertainty in the form of frequency
distributions for the SC-CH4 estimates for emissions in 2030.72 The distribution of SC-CH4
estimates reflect uncertainty in key model parameters such as the equilibrium climate sensitivity,
as well as uncertainty in other parameters set by the original model developers. To highlight the
difference between the impact of the discount rate and other quantified sources of uncertainty,
the bars below the frequency distributions provide a symmetric representation of quantified
variability in the SC-CH4 estimates for each discount rate. As illustrated by the figure, the
assumed discount rate plays a critical role in the ultimate estimate of the SC-CH4. This is
because GHG emissions today continue to impact society far out into the future, so with a higher
discount rate, costs that accrue to future generations are weighted less, resulting in a lower
estimate. As discussed in the February 2021 SC-GHG TSD, there are other sources of
uncertainty that have not yet been quantified and are thus not reflected in these estimates.
72 Although the distributions and numbers in Figure 3-1 are based on the full set of model results (150,000 estimates
for each discount rate), for display purposes the horizontal axis is truncated with 0.029 percent of the estimates
falling below the lowest bin displayed and 3 percent of the estimates falling above the highest bin displayed.
73
-------
o
CM
O
w
c
o
(f)
<•*—
o
c
o
o
ID
O
o
o
o
5% Average = $930
Discount Rate
5.0%
~ 3.0%
~ 2.5%
3% Average = $2000
2.5% Average = $2500
-C
Rn_
3%
95th Pet,
$5100
Tttftff
HEEBbeebbceb
Tt r no
5th - 95th Percentile
of Simulations
0 400 1200 2000 2800 3600 4400 5200 6000 6800 7600 8400
Social Cost of Methane in 2030 [2019$ / metric ton CH4]
Figure 3-1 Frequency Distribution of SC-CH4 Estimates for 2030
The interim SC-CH4 estimates presented in Table 3-3 have a number of limitations. First,
the current scientific and economic understanding of discounting approaches suggests discount
rates appropriate for intergenerational analysis in the context of climate change are likely to be
less than 3 percent, near 2 percent or lower (IWG, 2021). Second, the IAMs used to produce
these interim estimates do not include all of the important physical, ecological, and economic
impacts of climate change recognized in the climate change literature and the science underlying
their "damage functions" — i.e., the core parts of the IAMs that map global mean temperature
changes and other physical impacts of climate change into economic (both market and
nonmarket) damages — lags behind the most recent research. For example, limitations include
the incomplete treatment of catastrophic and non-catastrophic impacts in the integrated
assessment models, their incomplete treatment of adaptation and technological change, the
incomplete way in which inter-regional and intersectoral linkages are modeled, uncertainty in the
extrapolation of damages to high temperatures, and inadequate representation of the relationship
between the discount rate and uncertainty in economic growth over long time horizons.
Likewise, the socioeconomic and emissions scenarios used as inputs to the models do not reflect
new information from the last decade of scenario generation or the full range of projections.
74
-------
The modeling limitations do not all work in the same direction in terms of their influence
on the SC-GHG estimates. However, the IWG has recommended that, taken together, the
limitations suggest that the interim SC-GHG estimates used in this proposed rule likely
underestimate the damages from GHG emissions. In particular, the Intergovernmental Panel on
Climate Change (IPCC) Fourth Assessment Report (IPCC, 2007), which was the most current
IPCC assessment available at the time when the IWG decision over the ECS input was made,
concluded that SC-C02 estimates "very likely.. .underestimate the damage costs" due to omitted
impacts. Since then, the peer-reviewed literature has continued to support this conclusion, as
noted in the IPCC's Fifth Assessment report (IPCC, 2014) and other recent scientific
assessments (e.g., IPCC (2018, 2019a, 2019b)); U.S. Global Change Research Program
(USGCRP, 2016, 2018); and the National Academies of Sciences, Engineering, and Medicine
(National Academies, 2017, 2019). These assessments confirm and strengthen the science,
updating projections of future climate change and documenting and attributing ongoing changes.
For example, sea level rise projections from the IPCC's Fourth Assessment report ranged from
18 to 59 centimeters by the 2090s relative to 1980-1999, while excluding any dynamic changes
in ice sheets due to the limited understanding of those processes at the time (IPCC, 2007). A
decade later, the Fourth National Climate Assessment projected a substantially larger sea level
rise of 30 to 130 centimeters by the end of the century relative to 2000, while not ruling out even
more extreme outcomes (USGCRP, 2018). EPA has reviewed and considered the limitations of
the models used to estimate the interim SC-GHG estimates, and concurs with the February 2021
SC-GHG TSD's assessment that, taken together, the limitations suggest that the interim SC-
GHG estimates likely underestimate the damages from GHG emissions. The February 2021 SC-
GHG TSD briefly previews some of the recent advances in the scientific and economic literature
that the IWG is actively following and that could provide guidance on, or methodologies for,
addressing some of the limitations with the interim SC-GHG estimates.
There are several limitations specific to the estimation of SC-CH4. For example, the SC-
CH4 estimates do not reflect updates from the IPCC regarding atmospheric and radiative
efficacy. Another limitation is that the SC-CH4 estimates do not account for the direct health and
welfare impacts associated with tropospheric ozone produced by methane (see the 2016 NSPS
RIA for further discussion; see also Sarofim et al. (2017), reporting that studies have found the
global ozone-related mortality benefits of CH4 emissions reductions, which are not included in
75
-------
the social cost of methane valuations, to be $800 to $1,800 per metric ton of methane emissions).
In addition, the SC-CH4 estimates do not reflect that methane emissions lead to a reduction in
atmospheric oxidants, like hydroxyl radicals, nor do they account for impacts associated with
CO2 produced from methane oxidizing in the atmosphere. See EPA-HQ-OAR-2015-0827-5886
for more detailed discussion about the limitations specific to the estimation of SC-CH4. These
individual limitations and uncertainties do not all work in the same direction in terms of their
influence on the SC-CH4 estimates.
Table 3-4 presents the undiscounted annual monetized climate benefits under the
proposed NSPS OOOOb and EG OOOOc. Projected methane emissions reductions each year are
multiplied by the SC-CH4 estimate for that year.73 Table 3-5 shows the annual climate benefits
73 According to OMB's Circular A-4 (OMB, 2003), an "analysis should focus on benefits and costs that accrue to
citizens and residents of the United States", and international effects should be reported, but separately. Circular A-4
also reminds analysts that "[d]ifferent regulations may call for different emphases in the analysis, depending on the
nature and complexity of the regulatory issues." To correctly assess the total climate damages to U.S. citizens and
residents, an analysis should account for all the ways climate impacts affect the welfare of U.S. citizens and
residents, including how U.S. GHG mitigation activities affect mitigation activities by other countries, and spillover
effects from climate action elsewhere. The SC-GHG estimates used in regulatory analysis under revoked EO 13783
were a limited approximation of some of the U.S. specific climate damages from GHG emissions. These estimates
range from $204 per metric ton CH4 (2019 dollars) using a 3 percent discount rate for emissions occurring in 2023
to $279 per metric ton CH4 using a 3 percent discount rate for emissions occurring in 2035. Applying these
estimates (based on a 3 percent discount rate) to the CH4 emissions reduction expected under the proposed rule
would yield benefits from climate impacts of $27 million in 2023, increasing to $890 million in 2035. However, as
discussed at length in the IWG's February 2021 SC-GHG TSD, these estimates are an underestimate of the benefits
of CH4 mitigation accruing to U.S. citizens and residents, as well as being subject to a considerable degree of
uncertainty due to the manner in which they are derived. In particular, as discussed in this analysis, EPA concurs
with the assessment in the February 2021 SC-GHG TSD that the estimates developed under revoked E.O. 13783 did
not capture significant regional interactions, spillovers, and other effects and so are incomplete underestimates. As
the U.S. Government Accountability Office (GAO) concluded in a June 2020 report examining the SC-GHG
estimates developed under E.O. 13783, the models "were not premised or calibrated to provide estimates of the
social cost of carbon based on domestic damages" (U.S. GAO 2020, p. 29). Further, the report noted that the
National Academies found that country-specific social costs of carbon estimates were "limited by existing
methodologies, which focus primarily on global estimates and do not model all relevant interactions among regions"
(U.S. GAO 2020, p. 26). It is also important to note that the SC-GHG estimates developed under E.O. 13783 were
never peer reviewed, and when their use in a specific regulatory action was challenged, the U.S. District Court for
the Northern District of California determined that use of those values had been "soundly rejected by economists as
improper and unsupported by science," and that the values themselves omitted key damages to U.S. citizens and
residents including to supply chains, U.S. assets and companies, and geopolitical security. The Court found that by
omitting such impacts, those estimates "fail[ed] to consider.. .important aspect[s] of the problem" and departed from
the "best science available" as reflected in the global estimates. California v. Bernhardt, 472 F. Supp. 3d 573, 613-
14 (N.D.Cal. 2020). EPA continues to center attention in this analysis on the global measures of the SC-GHG as the
appropriate estimates given the flaws in the U.S. specific estimates, and as necessary for all countries to use to
achieve an efficient allocation of resources for emissions reduction on a global basis, and so benefit the U.S. and its
citizens.
76
-------
discounted back to 2021 and the PV and the EAV for the 2023-2035 period under each discount
rate.
Table 3-4 Undiscounted Monetized Climate Benefits under the NSPS OOOOb and EG
OOOOc Option, 2023-2035 (millions, 2019$)
Undiscounted3
5%
3%
2.5%
3%
Year
Average
Average
Average
95th Percentile
2023
$97
$210
$280
$560
2024
$150
$330
$430
$880
2025
$220
$470
$610
$1,200
2026
$2,600
$5,500
$7,100
$15,000
2027
$2,700
$5,700
$7,300
$15,000
2028
$2,800
$5,800
$7,500
$15,000
2029
$2,800
$6,000
$7,700
$16,000
2030
$2,900
$6,100
$7,900
$16,000
2031
$3,100
$6,300
$8,100
$17,000
2032
$3,200
$6,500
$8,300
$17,000
2033
$3,300
$6,700
$8,500
$18,000
2034
$3,400
$6,900
$8,700
$18,000
2035
$3,500
$7,100
$8,900
$19,000
a Climate benefits are based on changes (reductions) in CH4 emissions and are calculated using four different
estimates of the SC-CH4 (model average at 2.5 percent, 3 percent, and 5 percent discount rates; and 95th percentile
at 3 percent discount rate). The IWG emphasized the importance and value of considering the benefits calculated
using all four estimates. As discussed in the Technical Support Document: Social Cost of Carbon, Methane, and
Nitrous Oxide Interim Estimates under EO 13990 (IWG, 2021), a consideration of climate benefits calculated using
discount rates below 3 percent, including 2 percent and lower, are also warranted when discounting intergenerational
impacts.
77
-------
Table 3-5 Discounted Monetized Climate Benefits under the Proposed NSPS OOOOb
and EG OOOOc Option, 2023-2035 (millions, 2019$)
Discounted back to 2021a
5%
3%
2.5%
3%
Year
Average
Average
Average
95th Percentile
2023
$88
$200
$260
$520
2024
$130
$300
$400
$810
2025
$180
$420
$550
$1,100
2026
$2,000
$4,700
$6,300
$13,000
2027
$2,000
$4,700
$6,300
$13,000
2028
$2,000
$4,700
$6,300
$13,000
2029
$1,900
$4,700
$6,300
$13,000
2030
$1,900
$4,700
$6,300
$12,000
2031
$1,900
$4,700
$6,300
$12,000
2032
$1,900
$4,700
$6,300
$12,000
2033
$1,800
$4,700
$6,300
$12,000
2034
$1,800
$4,700
$6,300
$12,000
2035
$1,800
$4,700
$6,300
$12,000
PV
$19,000
$48,000
$64,000
$130,000
EAV
$2,100
$4,500
$5,900
$12,000
Note: Totals may not appear to add correctly due to rounding.
a Climate benefits are based on changes (reductions) in CH4 emissions and are calculated using four different
estimates of the SC-CH4 (model average at 2.5 percent, 3 percent, and 5 percent discount rates; and 95th percentile
at 3 percent discount rate). The IWG emphasized the importance and value of considering the benefits calculated
using all four estimates. As discussed in the Technical Support Document: Social Cost of Carbon, Methane, and
Nitrous Oxide Interim Estimates under EO 13990 (IWG, 2021), a consideration of climate benefits calculated using
discount rates below 3 percent, including 2 percent and lower, are also warranted when discounting intergenerational
impacts.
As discussed in the November 2021 proposal RIA, the IWG is currently working on a
comprehensive update of the SC-GHG estimates under E.O. 13990 taking into consideration
recommendations from the National Academies of Sciences, Engineering and Medicine, recent
scientific literature, and public comments received on the February 2021 SC-GHG TSD. EPA is
a member of the IWG and is participating in the IWG's review and updating process under E.O.
13990. While that process continues, the EPA is taking the opportunity in this RIA to present a
sensitivity analysis of the monetized climate benefits of this proposed action using an updated set
of SC-CH4 estimates, newly developed by EPA, based on newly available research and
methodological updates addressing recommendations of the National Academies of Sciences,
Engineering, and Medicine (2017). This sensitivity analysis is provided in Appendix B below.
More information about the development of these new estimates is available at:
https://www.epa.gov/environmental-economics/scghg.
78
-------
3.3 Ozone-Related Impacts Due to VOC Emissions
This proposed rulemaking is projected to reduce VOC emissions, which are a precursor
to ozone. Ozone is not generally emitted directly into the atmosphere but is created when its two
primary precursors, VOC and oxides of nitrogen (NOx), react in the atmosphere in the presence
of sunlight. In urban areas, compounds representing all classes of VOC can be important for
ozone formation, but biogenic VOC emitted from vegetation tend to be more important
compounds in non-urban vegetated areas (U.S. EPA, 2013). Recent observational and modeling
studies have found that VOC emissions from oil and natural gas operations can impact ozone
levels . Emissions reductions may decrease ozone formation, human exposure to ozone, and the
incidence of ozone-related health effects.
Calculating ozone impacts from changes in VOC emissions requires information about
the spatial patterns in those emissions changes. In addition, the ozone health effects from the
proposed rule will depend on the relative proximity of expected VOC and ozone changes to
population. In this analysis, we have not characterized VOC emissions changes at a finer spatial
resolution than the national total due to data and resource constraints. In light of these
limitations, we present an illustrative screening analysis of ozone-related health benefits in
Appendix C based on modeled oil and natural gas VOC contributions to ozone concentrations as
they occurred in 2017 and do not include the results of this screening analysis in the estimate of
benefits (and net benefits) projected from this proposal.74 To more definitively analyze the
impacts of VOC reductions from this proposed rule on ozone health benefits, we would need
credible projections of spatial patterns of expected VOC emissions reductions. Similarly, due to
the high degree of variability in the responsiveness of ozone formation to VOC emissions
reductions, we are unable to determine how this rule might affect air quality in downwind ozone
nonattainment areas without modeling air quality changes. However, we note that in future
regulatory impact analyses supporting other regulations, the EPA plans to account for the
emissions impacts of the oil and natural gas NSPS OOOOb and EG 0000c in the baseline for
the analysis.
74 Note that this illustrative analysis does not reflect the health and welfare benefits from reductions in tropospheric
ozone production resulting from CH4 emissions.
79
-------
3.3.1 Ozone Health Effects
Human exposure to ambient ozone concentrations is associated with adverse health
effects, including premature respiratory mortality and cases of respiratory morbidity (U.S. EPA,
2020a). Researchers have associated ozone exposure with adverse health effects in numerous
toxicological, clinical, and epidemiological studies (U.S. EPA, 2020a). When adequate data and
resources are available, the EPA has generally quantified several health effects associated with
exposure to ozone (U.S. EPA, 2010, 201 le, U.S. EPA, 2021c). These health effects include
respiratory morbidity, such as asthma attacks, hospital and emergency department visits, lost
school days, and premature respiratory mortality. The scientific literature is also suggestive that
exposure to ozone is associated with chronic respiratory damage and premature aging of the
lungs.
3.3.2 Ozone Vegetation Effects
Exposure to ozone has been found to be associated with a wide array of vegetation and
ecosystem effects in the published literature (U.S. EPA, 2020a). Sensitivity to ozone is highly
variable across species, with over 66 vegetation species identified as "ozone-sensitive," many of
which occur in state and national parks and forests. These effects include those that cause
damage to, or impairment of, the intended use of the plant or ecosystem. Such effects are
considered adverse to public welfare and can include reduced growth and/or biomass production
in sensitive trees, reduced yield and quality of crops, visible foliar injury, changed to species
composition, and changes in ecosystems and associated ecosystem services.
3.3.3 Ozone Climate Effects
Ozone is a well-known short-lived climate forcing GHG (U.S. EPA, 2013). Stratospheric
ozone (the upper ozone layer) is beneficial because it protects life on Earth from the sun's
harmful ultraviolet (UV) radiation. In contrast, tropospheric ozone (ozone in the lower
atmosphere) is a harmful air pollutant that adversely affects human health and the environment
and contributes significantly to regional and global climate change. Due to its short atmospheric
lifetime, tropospheric ozone concentrations exhibit large spatial and temporal variability (U.S.
EPA, 2009b). The IPCC AR5 estimated that the contribution to current warming levels of
increased tropospheric ozone concentrations resulting from human methane, NOx, and VOC
80
-------
emissions was 0.5 W/m2, or about 30 percent as large a warming influence as elevated CO2
concentrations. This quantifiable influence of ground level ozone on climate leads to increases in
global surface temperature and changes in hydrological cycles.
3.4 Ozone-Related Impacts Due to Methane
The tropospheric ozone produced by the reaction of methane in the atmosphere has
harmful effects for human health and plant growth in addition to its climate effects (Nolte et al.,
2018). In remote areas, methane is a dominant precursor to tropospheric ozone formation .
Approximately 50 percent of the global annual mean ozone increase since preindustrial times is
believed to be due to anthropogenic methane (Myhre et al., 2013). Projections of future
emissions also indicate that methane is likely to be a key contributor to ozone concentrations in
the future (Myhre et al., 2013). Unlike NOx and VOC, which affect ozone concentrations
regionally and at hourly time scales, methane emissions affect ozone concentrations globally and
on decadal time scales given methane's long atmospheric lifetime when compared to these other
ozone precursors (Myhre et al., 2013). Reducing methane emissions, therefore, will contribute to
efforts to reduce global background ozone concentrations that contribute to the incidence of
ozone-related health effects (USGCRP, 2018). The benefits of such reductions are global and
occur in both urban and rural areas. As noted above, these effects are not included in estimates of
the social cost of methane and are not otherwise quantified or monetized in this anlaysis.
3.5 PM2.5-Related Impacts Due to VOC Emissions
This proposed rulemaking is expected to result in emissions reductions of VOC, which
are a precursor to PM2.5, thus decreasing human exposure to PM2.5 and the incidence of PM2.5-
related health effects, although the magnitude of this effect has not been quantified at this time.
Most VOC emitted are oxidized to CO2 rather than to PM, but a portion of VOC emissions
contributes to ambient PM2.5 levels as organic carbon aerosols (U.S. EPA, 2019a). Analysis of
organic carbon measurements suggest only a fraction of secondarily formed organic carbon
aerosols are of anthropogenic origin. The current state of the science of secondary organic
carbon aerosol formation indicates that anthropogenic VOC contribution to secondary organic
carbon aerosol is often lower than the biogenic (natural) contribution (U.S. EPA, 2019a). The
potential for an organic compound to partition into the particle phase is highly dependent on its
81
-------
volatility such that compounds with lower volatility are more prone to partition into the particle
phase and form secondary organic aerosols (SOA) (Cappa & Wilson, 2012; Donahue, Kroll,
Pandis, & Robinson, 2012; Jimenez et al., 2009). Hydrocarbon emissions from oil and natural
gas operations tend to be dominated by high volatility, low-carbon number compounds that are
less likely to form SOA (Helmig et al., 2014; Koss et al., 2017; Petron et al., 2012). Given that
only a fraction of secondarily formed organic carbon aerosols is from anthropogenic VOC
emissions, and the relatively volatile nature of VOCs emitted from this sector, it is unlikely that
the VOC emissions reductions projected to occur under this proposal would have a large
contribution to ambient secondary organic carbon aerosols. Therefore, we have not quantified the
PM2.5-related benefits in this analysis. Moreover, without modeling air quality changes, we are
unable to determine how this rule might affect air quality in downwind PM2.5 nonattainment
areas. However, we note that in future regulatory impact analyses supporting other regulations,
the EPA plans to account for the emissions impacts of the oil and natural gas NSPS OOOOb and
EG OOOOc in the baseline for the analysis.
3.5.1 PM2.5 Health Effects
Decreasing exposure to PM2.5 is associated with significant human health benefits,
including reductions in respiratory mortality and respiratory morbidity. Researchers have
associated PM2.5 exposure with adverse health effects in numerous toxicological, clinical, and
epidemiological studies (U.S. EPA, 2019a). These health effects include asthma development
and aggravation, decreased lung function, and increased respiratory symptoms, such as irritation
of the airways, coughing, or difficulty breathing (U.S. EPA, 2019a). These health effects result in
hospital and ER visits, lost workdays, and restricted activity days. When adequate data and
resources are available, the EPA has quantified the health effects associated with exposure to
PM2.5 (U.S. EPA, 2021f).
When the EPA quantifies PIVh.s-related benefits, the Agency assumes that all fine
particles, regardless of their chemical composition, are equally potent in causing premature
mortality because the scientific evidence is not yet sufficient to allow differentiation of effect
estimates by particle type (U.S. EPA, 2019a). Based on our review of the current body of
scientific literature, the EPA estimates PM-related premature mortality without applying an
assumed concentration threshold. This decision is supported by the data, which are quite
82
-------
consistent in showing effects down to the lowest measured levels of PM2.5 in the underlying
epidemiology studies.
3.5.2 PM Welfare Effects
Suspended particles and gases degrade visibility by scattering and absorbing light.
Decreasing secondary formation of PM2.5 from VOC emissions could improve visibility
throughout the U.S. Visibility impairment has a direct impact on people's enjoyment of daily
activities and their overall sense of wellbeing. Good visibility increases the quality of life where
individuals live and work, and where they engage in recreational activities. Previous analyses
(U.S. EPA, 2006, 201 la, 201 Id, 2012) show that visibility benefits are a significant welfare
benefit category. However, without air quality modeling of PM2.5 impacts, we are unable to
estimate visibility related benefits.
Separately, persistent and bioaccumulative HAP reported as emissions from oil and
natural gas operations, including polycyclic organic matter, could lead to PM welfare effects.
Several significant ecological effects are associated with the deposition of organic particles,
including persistent organic pollutants and polycyclic aromatic hydrocarbons (PAHs) (U.S. EPA,
2009a). PAHs can accumulate to high enough concentrations in some coastal environments to
pose an environmental health threat that includes cancer in fish populations, toxicity to
organisms living in the sediment and risks to those (e.g., migratory birds) that consume these
organisms. Atmospheric deposition of particles is thought to be the major source of PAHs to the
sediments of coastal areas of the U.S. (U.S. EPA, 2012).
3.6 Hazardous Air Pollutants (HAP) Impacts
Available emissions data show that several different HAP are emitted from oil and
natural gas operations. The HAP emissions from the oil and natural gas sector in the 2017
National Emissions Inventory (NEI) emissions data are summarized in Table 3-6. The table
includes either oil and natural gas nonpoint or oil and natural gas point emissions of at least 10
tons per year, in descending order of annual nonpoint emissions. Emissions of eight HAP make
up a large percentage of the total HAP emissions by mass from the oil and natural gas sector:
toluene, hexane, benzene, xylenes (mixed), ethylene glycol, methanol, ethyl benzene, and 2,2,4-
trimethylpentane (U.S. EPA, 2011b).
83
-------
Table 3-6 Top Annual HAP Emissions as Reported in 2017 NEI for Oil and Natural
Gas Sources
Pollutant Nonpoint Emissions (tons/year) Point Emissions (tons/year)
Benzene
26,869
502
Xylenes (Mixed Isomers)
25,410
506
Formaldehyde
23,413
222
Toluene
18,054
823
Acetaldehyde
2,722
26
Hexane
2,675
886
Ethyl Benzene
2,021
113
Acrolein
1,602
18
Methanol
1,578
342
1,3-Butadiene
337
5.80E-01
2,2,4-Trimethylpentane
252
46
Naphthalene
104
1.10E+00
Propionaldehyde
102
0.00E+00
PAH/POM - Unspecified
68
2.50E-02
1,1,2-Trichloroethane
25
1.40E-03
Methylene Chloride
22
8.70E-02
1,1,2,2-Tetrachloroethane
14
1.90E-03
Ethylene Dibromide
13
1.90E-03
Methyl Tert-Butyl Ether
0
17.30
In the subsequent sections, we describe the health effects associated with the main HAP
of concern from the oil and natural gas sector: benzene (Section 3.6.1), formaldehyde (Section
3.6.2), toluene (Section 3.6.3), carbonyl sulfide (Section 3.6.4), ethylbenzene (Section 3.6.5),
mixed xylenes (Section 3.6.6), and n-hexane (Section 3.6.7), and other air toxics (Section 3.6.8).
This proposal is projected to reduce 280,000 tons of HAP emissions over the 2023 through 2035
period.75 With the data available, it was not possible to estimate the change in emissions of each
individual HAP.
Monetization of the benefits of reductions in cancer incidences requires several important
inputs, including central estimates of cancer risks, estimates of exposure to carcinogenic HAP,
and estimates of the value of an avoided case of cancer (fatal and non-fatal). Due to methodology
75 The projected emissions reductions from the proposed NSPS and EG, including projections of HAP reductions,
are based upon the unit-level model plant analysis supporting this rulemaking multiplied by counts of units that are
potentially affected by this proposal. The model plants and counts are built from a different basis than the oil and
natural gas sector emissions estimated in the NEI. Comparisons between the projected emissions reductions under
this proposal and the NEI should be made with caution.
84
-------
and data limitations, we did not attempt to monetize the health benefits of reductions in HAP in
this analysis. Instead, we are providing a qualitative discussion of the health effects associated
with HAP emitted from sources subject to control under the proposed NSPS OOOOb and EG
0000c. The EPA remains committed to improving methods for estimating HAP benefits by
continuing to explore additional aspects of HAP-related risk from the oil and natural gas sector,
including the distribution of that risk. This is discussed further in the context of environment
justice in Section 4.2.4.
3.6.1 Benzene
The EPA's Integrated Risk Information System (IRIS) database lists benzene as a known
human carcinogen (causing leukemia) by all routes of exposure and concludes that exposure is
associated with additional health effects, including genetic changes in both humans and animals
and increased proliferation of bone marrow cells in mice (IARC, 1982; Irons, Stillman,
Colagiovanni, & Henry, 1992; U.S. EPA, 2003a). The EPA states that data indicate a causal
relationship between benzene exposure and acute lymphocytic leukemia and suggest a
relationship between benzene exposure and chronic non-lymphocytic leukemia and chronic
lymphocytic leukemia. The International Agency for Research on Carcinogens (IARC) has
determined that benzene is a human carcinogen, and the U.S. Department of Health and Human
Services has characterized benzene as a known human carcinogen (IARC, 1987; NTP, 2004).
Several adverse noncancer health effects have been associated with chronic inhalation of
benzene in humans including arrested development of blood cells, anemia, leukopenia,
thrombocytopenia, and aplastic anemia. Respiratory effects have been reported in humans
following acute exposure to benzene vapors, such as nasal irritation, mucous membrane
irritation, dyspnea, and sore throat (ATSDR, 2007a).
3.6.2 Formaldehyde
In 1989, the EPA classified formaldehyde as a probable human carcinogen based on
limited evidence of cancer in humans and sufficient evidence in animals (U.S. EPA, 1991b).
Later the IARC (2006, 2012) classified formaldehyde as a human carcinogen based upon
sufficient human evidence of nasopharyngeal cancer and strong evidence for leukemia.
Similarly, in 2016, the National Toxicology Program (NTP) classified formaldehyde as known to
85
-------
be a human carcinogen based on sufficient evidence of cancer from studies in humans supporting
data on mechanisms of carcinogenesis (NTP, 2016). Formaldehyde inhalation exposure causes a
range of noncancer health effects including irritation of the nose, eyes, and throat in humans and
animals. Repeated exposures cause respiratory tract irritation, chronic bronchitis and nasal
epithelial lesions such as metaplasia and loss of cilia in humans. Airway inflammation, including
eosinophil infiltration, has been observed in animals exposed to formaldehyde. In children, there
is evidence that formaldehyde may increase the risk of asthma and chronic bronchitis (ATSDR,
1999; WHO, 2002).
3.6.3 Toluene76
Under the 2005 Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), there is
inadequate information to assess the carcinogenic potential of toluene because studies of humans
chronically exposed to toluene are inconclusive, toluene was not carcinogenic in adequate
inhalation cancer bioassays of rats and mice exposed for life, and increased incidences of
mammary cancer and leukemia were reported in a lifetime rat oral bioassay.
The central nervous system (CNS) is the primary target for toluene toxicity in both
humans and animals for acute and chronic exposures. CNS dysfunction (which is often
reversible) and narcosis have been frequently observed in humans acutely exposed to low or
moderate levels of toluene by inhalation: symptoms include fatigue, sleepiness, headaches, and
nausea. Central nervous system depression has been reported to occur in chronic abusers exposed
to high levels of toluene. Symptoms include ataxia, tremors, cerebral atrophy, nystagmus
(involuntary eye movements), and impaired speech, hearing, and vision. Chronic inhalation
exposure of humans to toluene also causes irritation of the upper respiratory tract, eye irritation,
dizziness, headaches, and difficulty with sleep.
Human studies have also reported developmental effects, such as CNS dysfunction,
attention deficits, and minor craniofacial and limb anomalies, in the children of women who
abused toluene during pregnancy. A substantial database examining the effects of toluene in
subchronic and chronic occupationally exposed humans exists. The weight of evidence from
these studies indicates neurological effects (i.e., impaired color vision, impaired hearing,
76 All health effects language for this section came from: U.S. EPA (2005b).
86
-------
decreased performance in neurobehavioral analysis, changes in motor and sensory nerve
conduction velocity, headache, and dizziness) as the most sensitive endpoint.
3.6.4 Carbonyl Sulfide
Limited information is available on the health effects of carbonyl sulfide. Acute (short-
term) inhalation of high concentrations of carbonyl sulfide may cause narcotic effects and irritate
the eyes and skin in humans.77 No information is available on the chronic (long-term),
reproductive, developmental, or carcinogenic effects of carbonyl sulfide in humans. Carbonyl
sulfide has not undergone a complete evaluation and determination under the EPA's IRIS
program for evidence of human carcinogenic potential (U.S. EPA, 1991a).
3.6.5 Ethylbenzene
Ethylbenzene is a major industrial chemical produced by alkylation of benzene. The pure
chemical is used almost exclusively for styrene production. It is also a constituent of crude
petroleum and is found in gasoline and diesel fuels. Acute (short-term) exposure to ethylbenzene
in humans results in respiratory effects such as throat irritation and chest constriction, and
irritation of the eyes, and neurological effects such as dizziness. Chronic (long-term) exposure of
humans to ethylbenzene may cause eye and lung irritation, with possible adverse effects on the
blood. Animal studies have reported effects on the blood, liver, and kidneys and endocrine
system from chronic inhalation exposure to ethylbenzene. No information is available on the
developmental or reproductive effects of ethylbenzene in humans, but animal studies have
reported developmental effects, including birth defects in animals exposed via inhalation. Studies
in rodents reported increases in the percentage of animals with tumors of the nasal and oral
cavities in male and female rats exposed to ethylbenzene via the oral route (Maltoni et al., 1997;
Maltoni, Conti, Cotti, & Belpoggi, 1985). The reports of these studies lacked detailed
information on the incidence of specific tumors, statistical analysis, survival data, and
information on historical controls, thus the results of these studies were considered inconclusive
by the International Agency for Research on Cancer (IARC, 2000) and the National Toxicology
77 Hazardous Substances Data Bank (HSDB), online database. US National Library of Medicine, Toxicology Data
Network, available online at https://pubchem.ncbi.nlm.nih.gov/. Carbonyl sulfide health effects summary available
at https://pubchem.ncbi.nlm.nih.gov/compound/10039#section=Safety-and-Hazards. Accessed April 26, 2020.
87
-------
Program (NTP, 1999). The NTP (1999) carried out a chronic inhalation bioassay in mice and rats
and found clear evidence of carcinogenic activity in male rats and some evidence in female rats,
based on increased incidences of renal tubule adenoma or carcinoma in male rats and renal
tubule adenoma in females. NTP (1999) also noted increases in the incidence of testicular
adenoma in male rats. Increased incidences of lung alveolar/bronchiolar adenoma or carcinoma
were observed in male mice and liver hepatocellular adenoma or carcinoma in female mice,
which provided some evidence of carcinogenic activity in male and female mice (NTP, 1999).
IARC (2000) classified ethylbenzene as Group 2B, possibly carcinogenic to humans, based on
the NTP studies.
3.6.6 Mixed Xylenes
Short-term inhalation of mixed xylenes (a mixture of three closely related compounds) in
humans may cause irritation of the nose and throat, nausea, vomiting, gastric irritation, mild
transient eye irritation, and neurological effects (U.S. EPA, 2003b). Other reported effects
include labored breathing, heart palpitation, impaired function of the lungs, and possible effects
in the liver and kidneys (ATSDR, 2007b). Long-term inhalation exposure to xylenes in humans
has been associated with a number of effects in the nervous system including headaches,
dizziness, fatigue, tremors, and impaired motor coordination (ATSDR, 2007b). The EPA has
classified mixed xylenes in Category D, not classifiable with respect to human carcinogenicity.
3.6.7 tt-Hexane
The studies available in both humans and animals indicate that the nervous system is the
primary target of toxicity upon exposure of n-hexane via inhalation. There are no data in humans
and very limited information in animals about the potential effects of n-hexane via the oral route.
Acute (short-term) inhalation exposure of humans to high levels of hexane causes mild central
nervous system effects, including dizziness, giddiness, slight nausea, and headache. Chronic
(long-term) exposure to hexane in air causes numbness in the extremities, muscular weakness,
blurred vision, headache, and fatigue. Inhalation studies in rodents have reported behavioral
effects, neurophysiological changes, and neuropathological effects upon inhalation exposure to
n-hexane. Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), the
88
-------
database for n-hexane is considered inadequate to assess human carcinogenic potential, therefore
the EPA has classified hexane in Group D, not classifiable as to human carcinogenicity.
3.6.8 Other A ir Toxics
In addition to the compounds described above, other toxic compounds might be affected
by this rule, including hydrogen sulfide (H2S). Information regarding the health effects of those
compounds can be found in the EPA's IRIS database.78
3.7 Secondary Air Emissions Impacts
The control techniques to meet the storage vessel-related standards are associated with
several types of secondary emissions impacts, which may partially offset the direct benefits of
this rule. Table 3-7 shows the estimated secondary emissions associated with combustion of
emissions as a result of these requirements. Relative to the direct emission reductions anticipated
from this rule, the magnitude of these secondary air pollutant increases is small.
Table 3-7 Increases in Secondary Air Pollutant Emissions, Vapor Combustion at
Storage Vessels (short tons per year)
Year
THC
CO
NOx
PM
CO2
2023
21
55
10
0
43,000
2024
29
78
14
1
61,000
2025
37
98
18
1
78,000
2026
44
120
21
1
91,000
2027
49
130
24
1
100,000
2028
53
140
26
1
110,000
2029
57
150
28
1
120,000
2030
60
160
29
1
120,000
2031
62
160
30
1
130,000
2032
64
170
31
1
130,000
2033
66
170
32
1
140,000
2034
67
180
33
1
140,000
2035
69
180
33
1
140,000
Total
680
1,800
330
12
1,400,000
Note: Totals may not appear to add correctly due to rounding.
The CO2 impacts in Table 3-7 are the emissions that are expected to occur from vapor
combustion at affected storage vessels. However, because of the atmospheric chemistry
78 The U.S. EPA Integrated Risk Information System (IRIS) database is available at https://www.epa.gov/iris.
Accessed April 26, 2020.
89
-------
associated with the natural gas emissions, most of the carbon in the VOCs and CH4 emissions
expected in the absence of combustion-related emissions controls would have eventually
oxidized forming CO2 in the atmosphere and led to approximately the same long-run CO2
concentrations as with controls.79 Therefore, most of the impact of these CO2 contribution to
atmospheric concentrations from the flaring of CH4 and VOC versus future oxidization is not
additional to the impacts that otherwise would have occurred through the oxidation process.
However, there is a shift in the timing of atmospheric CO2 concentration changes under
the policy case, in which case combustion controls lead to contemporaneous increases in CO2
concentrations, compared to the baseline where the CO2 concentration increase is delayed
through the oxidation process. In the case of VOC, the oxidization time in the atmosphere is
relatively short, on the order of hours to months, so from a climate perspective the difference
between emitting the carbon immediately as CO2 during combustion or as VOC is expected to be
negligible. In the case of CH4, the oxidization time is on the order of a decade, so the timing of
the contribution to atmospheric CO2 concertation will differ between the baseline and policy
case. Because the growth rate of the SC-CO2 estimates is lower than their associated discount
rates, the estimated impact of CO2 produced in the future via oxidized methane from these fossil-
based emissions may be less than the estimated impact of CO2 released immediately from
combusting emissions, which would imply a small disbenefit associated with the earlier release
of CO2 during combustion of the CH4 emissions.
In the NSPS OOOOa rulemaking, the EPA solicited comment on the appropriateness of
monetizing: (1) the impact of CO2 emissions associated with combusting methane and VOC
emissions from oil and natural gas sites; and (2) a new potential approach for approximating this
value using the SC-CO2. The illustrative analysis in the NSPS OOOOa RIA provided a method
for evaluating the estimated emissions outcomes associated with destroying one metric ton of
methane by combusting fossil-based emissions at oil and natural gas sites (flaring) and releasing
the CO2 emissions immediately versus releasing them in the future via the methane oxidation
process.80 The analysis demonstrated that the potential disbenefits of flaring (i.e., an earlier
79 The social cost of methane (SC-CH4) used previously in this chapter to monetize the benefits of the CH4
emissions reductions does not include the impact of the carbon in CH4 emissions after it oxidizes to CO2.
80 See Section 4.7 of U.S. EPA (2016).
90
-------
contribution of CO2 emissions to atmospheric concentrations) are minor compared to the benefits
of flaring (i.e., avoiding the release of and associated climate impacts from CH4 emissions).
While recognizing the challenges and uncertainties related to estimation of these
secondary emissions impacts for this proposed rulemaking, EPA has continued to examine this
issue in the context of this RIA and includes an illustrative analysis using the methodology from
the NSPS OOOOa final RIA. Specifically, for this illustrative analysis, EPA assumes the
oxidization process of CH4 to be dynamic and consistent with the modeling that underlies the
SC-CH4 estimates and assumes an average methane oxidation period of 12 years, consistent with
the perturbation lifetime-folding time used in IPCC AR4. The estimated disbenefits associated
with destroying one metric ton of methane through combustion of emissions at oil and gas sites
and releasing the CO2 emissions in 2023 instead of being released in the future via the methane
oxidation process are found to be small relative to the benefits of flaring. Specifically, the
disbenefit is estimated to be about $19 per metric ton CH4 (based on average SC-CO2 at 3
percent) or about one percent of the SC-CH4 estimate per metric ton for 2023. The analogous
estimate for 2035 is $30 per metric ton CH4 or about one percent of the SC-CH4 estimates per
metric ton for 2035.81
It is important to note that there are challenges and uncertainties related to this illustrative
method and estimates, which was developed to analyze secondary fossil-based emissions from
combustion. For example, these dollar per ton CH4 estimates cannot readily be applied to the
total CH4 emissions reductions presented in Section 3.1 without additional information about the
81 To calculate the CO2 related impacts associated the complete destruction of a ton of CH4 emissions through
flaring for this illustrative application, EPA took the difference between the SC-CO2 at the time of the flaring and
the discounted value of the CO2 impacts assuming a geometric decay of CH4 via the oxidation process with a 12
year e-folding time using the same discount rate as used to estimate the SC-CO2. This value was then scaled by
44/16 to account for the relative mass of carbon contained in a ton of CH4 versus a ton of CO2. More specifically,
the impacts of shifting the CO2 impacts are calculated as:
destroyed, r is the discount rate, and T is the time horizon of the analysis. Ideally the time horizon, T, would be
sufficiently long to capture the period in which nearly all of the CH4 is expected to have been oxidized. In this
analysis we use the 2100 as the time horizon, making the assumption that the SC-CO2 remains constant after 2050,
the last year for which the IWG provides estimates. This methodology improves upon the one presented at proposal
by updating the oxidization process of CH4 to be dynamic and consistent with the modeling that underlies the SC-
CH4 estimates.
where T is the year the CH4 is
91
-------
downstream outcomes associated with the recovered gas that is not flared — e.g., whether some
of that captured gas going to be burned or leaked somewhere down the line. The EPA will
continue to study this issue and assess the complexities involved in estimating the net emissions
effects associated with secondary fossil-based emissions, including differences in the timing of
contributions to atmospheric CO2 concentrations. Given the uncertainties related to estimating
net secondary emissions effects and that the EPA has not yet received appropriate input and
review on some aspects of these calculations, the EPA is not including monetized estimates of
the impacts of small changes in the timing of atmospheric CO2 concentration increases in the
benefits estimates in this RIA. The EPA will continue to follow the scientific literature on this
topic and update its methodologies as warranted.
3.8 Total Benefits
Table 3-8 presents the PV and EAV of the projected climate benefits across the three
regulatory options for the proposed NSPS OOOOb and EG 0000c examined in this RIA. These
values reflect an analytical time horizon of 2023 to 2035, are discounted to 2021, and presented
in 2019 dollars. Multiple benefits estimates are presented reflecting alternative discount rates.
The table includes consideration of the non-monetized benefits associated with the emissions
reductions projected under this proposal. Table 3-9 and Table 3-10 present the same information
for the proposed NSPS OOOOb and EG 0000c separately.
92
-------
Table 3-8 Comparison of PV and EAV of the Projected Benefits for the Proposed NSPS
OOOOb and EG OOOOc across Regulatory Options, 2023-2035 (millions of 2019$)
Year
5%
Average
3%
Average
2.50%
Average
3%
95th Percentile
Climate Benefits (PV)a
Less Stringent
$6,800
$17,000
$23,000
$45,000
Proposal
$19,000
$48,000
$64,000
$130,000
More Stringent
$19,000
$48,000
$65,000
$130,000
Climate Benefits (EAV)a
Less Stringent
$720
$1,600
$2,100
$4,200
Proposal
$2,100
$4,500
$5,900
$12,000
More Stringent
$2,100
$4,500
$5,900
$12,000
Non-Monetized Benefits
Climate and ozone health benefits from reducing methane emissions by (in short tons):
Less Stringent 12,000,000
Proposal 36,000,000
More Stringent 36,000,000
PM2.5 and ozone health benefits from reducing VOC emissions by (in short tons)b c:
Less Stringent 3,400,000
Proposal 9,700,000
More Stringent 9,800,000
HAP benefits from reducing HAP emissions by (in short tons):
Less Stringent 150,000
Proposal 390,000
More Stringent 390,000
Visibility benefits
Reduced vegetation effects
a Climate benefits are based on changes (reductions) in CH4 emissions and are calculated using four different
estimates of the SC-CH4 (model average at 2.5 percent, 3 percent, and 5 percent discount rates; and 95th percentile
at 3 percent discount rate. For purposes of this table, we show the benefits associated with the model average at a 3
percent discount rate. The IWG emphasized the importance and value of considering the benefits calculated using all
four estimates. As discussed in the Technical Support Document: Social Cost of Carbon, Methane, and Nitrous
Oxide Interim Estimates under EO 13990 (IWG, 2021), a consideration of climate benefits calculated using discount
rates below 3 percent, including 2 percent and lower, are also warranted when discounting intergenerational impacts.
Appendix B presents the results of a sensitivity analysis using a set of SC-CH4 estimates that incorporates recent
research addressing recommendations of the National Academies of Sciences, Engineering, and Medicine (2017).
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C of the RIA.
0 The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
93
-------
Table 3-9 Comparison of PV and EAV of the Projected Benefits for the Proposed NSPS
OOOOb across Regulatory Options, 2023-2035 (millions of 2019$)
5%
3%
2.50%
3%
Year
Average
Average
Average
95th Percentile
Climate Benefits (PV)a
Less Stringent
$800
$2,000
$2,700
$5,300
Proposal
$4,400
$11,000
$15,000
$29,000
More Stringent
$4,400
$11,000
$15,000
$29,000
Climate Benefits (EAV)a
Less Stringent
$86
$190
$240
$500
Proposal
$470
$1,000
$1,300
$2,700
More Stringent
$470
$1,000
$1,300
$2,700
Non-Monetized Benefits
Climate and ozone health benefits from reducing methane emissions by (in short tons):
Less Stringent 1,900,000
Proposal 8,100,000
More Stringent 10,000,000
PM2.5 and ozone health benefits from reducing VOC emissions by (in short tons)b c:
Less Stringent 1,400,000
Proposal 2,900,000
More Stringent 3,600,000
HAP benefits from reducing HAP emissions by (in short tons):
Less Stringent 52,000
Proposal 110,000
More Stringent 140,000
Visibility benefits
Reduced vegetation effects
a Climate benefits are based on changes (reductions) in CH4 emissions and are calculated using four different
estimates of the SC-CH4 (model average at 2.5 percent, 3 percent, and 5 percent discount rates; and 95th percentile
at 3 percent discount rate. For purposes of this table, we show the benefits associated with the model average at a 3
percent discount rate. The IWG emphasized the importance and value of considering the benefits calculated using all
four estimates. As discussed in the Technical Support Document: Social Cost of Carbon, Methane, and Nitrous
Oxide Interim Estimates under Executive Order 13990 (IWG, 2021), a consideration of climate benefits calculated
using discount rates below 3 percent, including 2 percent and lower, are also warranted when discounting
intergenerational impacts. Appendix B presents the results of a sensitivity analysis using a set of SC-CH4 estimates
that incorporates recent research addressing recommendations of the National Academies of Sciences, Engineering,
and Medicine (2017).
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C of the RIA.
0 The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
94
-------
Table 3-10 Comparison of PV and EAV of the Projected Benefits for the Proposed EG
OOOOc Across Regulatory Options, 2023-2035 (millions of 2019$)
5%
3%
2.50%
3%
Year
Average
Average
Average
95th Percentile
Climate Benefits (PV)a
Less Stringent
$6,000
$15,000
$20,000
$39,000
Proposal
$15,000
$37,000
$50,000
$98,000
More Stringent
$15,000
$37,000
$50,000
$99,000
Climate Benefits (EAV)a
Less Stringent
$640
$1,400
$1,800
$3,700
Proposal
$1,600
$3,500
$4,500
$9,300
More Stringent
$1,600
$3,500
$4,600
$9,300
Non-Monetized Benefits
Climate and ozone health benefits from reducing methane emissions by (in short tons):
Less Stringent 11,000,000
Proposal 28,000,000
More Stringent 28,000,000
PM2.5 and ozone health benefits from reducing VOC emissions by (in short tons)b c:
Less Stringent 2,300,000
Proposal 6,800,000
More Stringent 6,900,000
HAP benefits from reducing HAP emissions by (in short tons):
Less Stringent 110,000
Proposal 280,000
More Stringent 280,000
Visibility benefits
Reduced vegetation effects
a Climate benefits are based on changes (reductions) in CH4 emissions and are calculated using four different
estimates of the SC-CH4 (model average at 2.5 percent, 3 percent, and 5 percent discount rates; and 95th percentile
at 3 percent discount rate. For purposes of this table, we show the benefits associated with the model average at a 3
percent discount rate. The IWG emphasized the importance and value of considering the benefits calculated using all
four estimates. As discussed in the Technical Support Document: Social Cost of Carbon, Methane, and Nitrous
Oxide Interim Estimates under EO 13990 (IWG, 2021), a consideration of climate benefits calculated using discount
rates below 3 percent, including 2 percent and lower, are also warranted when discounting intergenerational impacts.
Appendix B presents the results of a sensitivity analysis using a set of SC-CH4 estimates that incorporates recent
research addressing recommendations of the National Academies of Sciences, Engineering, and Medicine (2017).
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C of the RIA.
0 The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
95
-------
4 ECONOMIC IMPACT AND DISTRIBUTIONAL ANALYSIS
The proposed NSPS OOOOb and EG 0000c constitute an economically significant
action. As discussed in previous section, the emissions reductions projected under the rule are
likely to produce substantial climate benefits, peaking at $3.5 to $19 billion in 2035, as well as
non-monetized benefits from large reductions in VOC and HAP emissions. At the same time, the
proposed NSPS OOOOb and EG 0000c is projected to result in substantial environmental
control expenditures by the oil and natural gas industry to comply with the rule, reaching a
maximum of $2.8 billion in 2026.
While the national level impacts demonstrate the proposal is likely to lead to significant
benefits and costs, the benefit-cost analysis does not speak directly to potential economic and
distributional impacts of the proposed rule, which may be important consequences of the action.
This section includes four sets of economic impact and distributional analyses for this proposal
directed toward complementing the benefit-cost analysis and includes an analysis of potential
national-level impacts on oil and natural gas markets, a series of environmental justice analyses,
an Initial Regulatory Flexibility Analysis that includes an analysis of projected compliance costs
of proposed NSPS OOOOb on small entities, and employment impacts.
4.1 Oil and Natural Gas Market Impact Analysis
In addition to the engineering cost analysis that produces the compliance cost and
emissions reduction projections that inform the net benefits analysis, the EPA developed a pair
of single-market, static partial-equilibrium analyses of national crude oil and natural gas markets.
The market impact analyses are intended to provide readers some information on the economic
impacts of the proposed NSPS OOOOb and EG 0000c and to inform the EPA's response to
EO 13211 "Actions Concerning Regulations that Significantly Affect Energy Supply,
Distribution, or Use." The partial equilibrium market impact estimates, however, do not inform
the projected engineering costs and emissions reductions used in the comparison of benefits and
costs.
Our partial equilibrium analyses treat crude oil markets and natural gas markets
separately. We implement a pair of single-market analyses instead of a coupled market or
96
-------
general equilibrium approach to provide broad insights into potential national-level market
impacts while providing analytical transparency.
The oil market model assumes a single, aggregate U.S supplier, a single, aggregate world
consumer, and a residual world supply. We assume the U.S. supply response to a percentage
change in costs has the same effect as a percentage change in price. We do not try to model the
residual world supply precisely. Instead, we model two extreme cases — perfectly inelastic
residual world supply and perfectly elastic residual world supply. These cases bound the residual
world supply response.
The natural gas market model assumes a single, aggregate U.S. supplier, a single,
aggregate U.S. consumer, and no international trade. We assume the U.S. supply response to a
percentage change in costs has the same effect as a percentage change in price. Existing natural
gas markets are segmented in the short-term by transmission constraints, but prices are
cointegrated across the United States (Siliverstovs, L'Hegaret, Neumann, & von Hirschhausen,
2005). Infrastructure, including new infrastructure in the long term, joins disparate markets. The
assumption of a single natural gas market is a long-term modeling assumption.
In each market, we first use a supply elasticity to solve for the supply change that results
from the imposition of regulatory costs. Given the change in supply, we then use a demand
elasticity to solve for the change in price that balances supply and demand. We use projected
crude oil and natural gas prices and production for a select set of years of analysis to
operationalize the model. In the sections that follow, we discuss the data and parameters used to
implement the models, present results of each analysis, and conclude with a discussion of caveats
and limitations of the analyses.
4.1.1 Crude Oil Market Model
The crude oil market model is a constant elasticity model that assumes a competitive U.S.
market with a rest of world residual oil supply that is either perfectly inelastic or perfectly
elastic. To find the changes in crude oil production and prices under the proposed NSPS OOOOb
and EG 0000c, we first solve for the change in production using a supply elasticity and the
regulatory cost. The year t change in U.S. oil production Ais estimated using Eq. 4-1:
A«w=Ji§ib*£o,s*
-------
where Co t is the projected regulatory cost impacting oil-producing sources in year /, Qo,t is the
baseline U.S. crude oil production in year t, Po t is the baseline crude oil price, and eos is the
supply elasticity of crude oil. The term — describes the cost change as a fraction of
revenue, akin to a percentage change in price. A key modeling assumption here is that, in
addition to a constant elasticity, a fractional change in revenue due to a cost change is equivalent
to a fractional change in output price. The term Qus°fp— * Eo,s then describes the fractional
change in production.
For the model assuming perfectly inelastic rest-of-world production, we use the change in
supply solved in Eq. 4-1 the find the change in crude oil prices using Eq. 4-2:
A/it/S ^
APo.t = 0worid * ~ * Po,t, Eq. 4-2
Qo,t £o,d
where Q]o°rld is global production of crude oil and eo d is the world demand elasticity for crude
oil.
Price does not change in the alternative model; it assumes perfectly elastic rest-of-world
production, so APo t = 0.
4.1.2 Natural Gas Market Model
We model U.S. natural gas supply and demand as a closed market. For the natural gas
market, we first find the change in quantity produced AQGt using Eq 4.-3:
A(?G,t = 7 * £g,s * Qgj, Eq. 4-3
Qc,t*Pc,t
where CGt is the projected regulatory cost impacting all segments of the natural gas industry in
year t, QG t is the baseline U.S. production forecast, PG t is the natural gas price forecast, and egs
is the supply elasticity for natural gas.
We then use the change in quantity solved in Eq. 4.3 to solve for the natural gas price
change APG t using Eq. 4-4:
APG,t=^*^-*PG,t Eq. 4-4
Qc,t SC,D
98
-------
4.1.3 Assumptions, Data, and Parameters Used in the Oil and Natural Gas Market Models
This section presents the basics assumptions applied in this analysis. The section also
presents the data and parameters used to operationalize the model, including our choice of years
of analysis, elasticity estimates, and production and price data.
4.1.3.1 Years of Analysis
We estimate the price and quantity impacts of the proposed NSPS OOOOb and EG
0000c on crude oil and natural gas markets for a subset of years within the time horizon
analyzed in this RIA. We analyze 2023 and 2025 as these years represent the first and last year
the requirements in the proposed NSPS OOOOb will be in effect for the purposes of the RIA
before the requirement of the proposed EG 0000c are assumed to go into effect. We then
analyze market impacts in 2026, 2030, and 2035 to examine the effects of the proposed EG
OOOOc in addition to the cumulative impacts of the proposed NSPS OOOOb. The year 2026 is
the year of analysis with the highest regulatory costs and, as such, will represent the year with
the largest market impacts based upon the partial equilibrium market models used here. We
analyze 2030 and 2035 in order to project impacts in later years of the time horizon, as the
projected regulatory costs decline.
4.1.3.2 Elasticity Choices
The elasticity estimates used in the analysis are based on estimates from the published
economics literature (Table 4-1). Natural gas demand elasticity is calculated as the sector-level
consumption-weighted average of demand elasticities from Hausman and Kellogg (2015). The
consumption proportions used to weight the elasticities are derived from 2019 levels of natural
consumption by the residential, commercial, industrial, and electric power sectors, as reported in
EIA.
99
-------
Table 4-1 Parameters Used in Market Analysis
Parameter
Symbol
Value
Source
Oil supply
1.2
Newell, R. G.. & B. C. Prest. 2019. The unconventional oil supply boom:
elasticity
£o,s
Aggregate price response from microdata. The Energy Journal 40(3).
Oil demand
elasticity
Coglianese, J., L. W. Davis, L. Kilian. & J. H. Stock. 2017. Anticipation.
eO,D
-0.37
tax avoidance, and the price elasticity of gasoline demand. Journal of
Applied Econometrics 32(1): 1-15.
Natural gas
supply
elasticity
Newell, R. G.. B. C. Prest, & A. B. Vissing. 2019. Trophy hunting versus
eG,S
0.9
manufacturing energy: The price responsiveness of shale gas." Journal of
the Association of Environmental and Resource Economists 6(2): 391-
431.
Natural gas
demand
elasticity
Sector-level consumption-weighted average of demand elasticities from
eG,D
-0.43
Hausman, C. & R. Kellogg. 2015. Welfare and Distributional
Implications of Shale Gas. Brookings Papers on Economic Activity:! 1-
125.
4.1.3.3 Production and Price Data
Baseline U.S. crude oil production, dry gas production, West Texas Intermediate (WTI)
crude oil prices, and Henry Hub natural gas prices are drawn from AEO2022. Prices are deflated
to 2019 dollars using the GDP-Implicit Price Deflator. As the proposed NSPS OOOOb and EG
0000c apply to onshore production but not offshore production, only onshore U.S. crude oil
production is analyzed. Dry natural gas production is the sum of onshore production from the
lower 48 states and all production from Alaska. Baseline world crude oil production is from the
Energy Information Administration's 2020 International Energy Outlook. Table 4-2 presents the
baseline crude oil and natural gas production and prices used in the market impacts analysis.
Table 4-2 Baseline Crude Oil and Natural Gas Production and Prices Used in Market
Analysis
Year
Data Resource Unit 2023 2025 2026 2030 2035
Baseline Production3
U.S. Crude Oil Production
million bbl/day
10.3
11.0
11.1
11.0
10.8
World Oil Production
million bbl/day
97.1
97.7
98.0
99.5
101.8
U.S. Onshore Production
tcf/year
35.3
35.7
35.7
36.5
37.2
Baseline Prices3
Crude Oil
2019$/bbl
55.8
61.4
62.6
67.7
72.0
Natural Gas
2019$/MMbtu
3.31
2.85
2.83
3.28
3.45
Natural Gas
2019$/Mcf
3.44
2.95
2.93
3.40
3.58
3Baseline U.S. crude oil and natural gas production and prices drawn from AEO2021. Baseline world oil production
drawn from EIA's International Energy Outlook.
100
-------
4.1.3.4 Regulatory Cost Impacts
As discussed earlier, we assume the projected regulatory costs associated with the
proposed NSPS 0000b and EG 0000c produce a fractional change in output price. We
distribute the projected regulatory costs to crude oil markets and natural gas markets according to
whether the emissions sources incurring the regulatory costs are more likely to be producing
crude oil or producing, processing, or transporting natural gas. To begin, all projected regulatory
costs for natural gas processing, storage, and transmission sources are assumed to impact the
natural gas market. Within the production segment, projected regulatory costs for natural gas-
related model plants are directed to natural gas markets and costs for oil-related model plants are
assigned to crude oil markets. For example, projected regulatory costs associated with fugitive
emissions monitoring at natural gas well sites are directed to the natural gas market, and
projected regulatory costs at oil well sites are directed to crude oil markets.
For this analysis, we use the projected regulatory costs with capital costs annualized
using a 7 percent interest rate. We also use the net regulatory costs, which include projected
revenues from natural gas recovery from emissions abatement activities. Table 4-3 presents the
results of decomposing the projected regulatory costs into crude oil and natural gas shares.
Table 4-3 Projected Regulatory Costs for the Proposed NSPS OOOOb and EG
OOOOc Option Applied in the Market Analysis (millions 2019$)
Year
Resource
2023
2025
2026
2030
2035
Crude Oil
70.7
157.2
1,094.6
1,067.2
1,112.0
Natural Gas
25.8
59.6
1,165.6
1,000.9
931.3
4.1.4 Results
The results of incorporating the projected regulatory costs into the crude oil market
model are presented in Table 4-4. At its peak, the reduction is about 20.98 million barrels in
2026 or about 0.52 percent of crude oil production.
101
-------
Table 4-4 Estimated Crude Oil Production and Prices Changes under the Proposed
NSPS OOOOb and EG OOOOc Option
Year
Variable
Change
2023
2025
2026
2030
2035
U.S. Production
million bbls/year
-1.52
-3.07
-20.98
-18.92
-18.53
%
-0.04%
-0.08%
-0.52%
-0.47%
-0.47%
U.S. Prices
Assuming Perfectly Inelastic
Rest of World Supply
$/bbl
0.01
0.01
0.10
0.10
0.10
%
0.01%
0.02%
0.16%
0.14%
0.13%
Assuming Perfectly Elastic
Rest of World Supply
$/bbl
0.0
0.0
0.0
0.0
0.0
%
0.00%
0.00%
0.00%
0.00%
0.00%
We describe two models of world oil markets that bound the market price responses.
Table 4-4 describes results. Assuming perfectly inelastic world oil markets represents an upper
bound on the crude oil price change. The maximum projected oil price change in modeled years
is 0.10 dollars per barrel in 2026, an increase of less than one sixth of one percent. The
alternative model is that world oil markets are perfectly elastic and maintain a fixed oil price. In
that case the price change would be zero. Table 4-5 presents results of entering the projected
regulatory costs in the natural gas market model. We project a maximum natural gas price
increase of about $0.07 per mcf and a maximum production reduction of about 358.0 million
Mcf per year, changes of about 2.35 percent and 1.00 percent respectively.
Table 4-5 Estimated Natural Gas Production and Prices Changes under the Proposed
NSPS OOOOb and EG OOOOc Option
Year
Variable Change 2023 2025 2026 2030 2035
U.S. Onshore Production million Mcf/year -6.8 -18.2 -358.0 -264.6 -234.2
% -0.02% -0.05% -1.00% -0.73% -0.63%
U.S. Prices
2019$/Mcf 0.00 0.00 0.07 0.06 0.05
% 0.04% 0.12% 2.35% 1.70% 1.47%
We use the results in Table 4-4 and Table 4-5 to evaluate whether the proposed NSPS
OOOOb and EG OOOOc is likely to have a significant effect on the supply, distribution, or use
of energy as defined by EO 13211. To make this determination, we compare the projected
change in crude oil and natural gas production to guidance articulated in a January 13, 2021
OMB memorandum "Furthering Compliance with Executive Order 13211, Titled "Actions
102
-------
Concerning Regulations That Significantly Affect Energy Supply, Distribution, or Use"."82 The
maximum projected annual decreases in both oil production and natural gas production exceed
benchmarks for adverse effects, so this analysis indicates the proposed NSPS OOOOb and EG
0000c constitutes a significant energy action.
4.1.5 Caveats and Limitations of the Market Analysis
The oil and natural gas market impact analysis presented in this section is subject to
several caveats and limitations, which we discuss here. As with any modeling exercise, the
market impact analysis presented here depends crucially on uncertain input parameters. These
parameters include the cost to firms of compliance, the amount of natural gas that would be
recovered and sold as a result of emissions abatement requirements compliance, baseline
projections, and elasticity estimates. We note the change in price is particularly sensitive to the
demand elasticity.
This analysis considers two residual rest-of-world supply models — perfectly elastic and
perfectly inelastic. The structure of international oil markets (both supply and demand) have
shifted historically and may shift in the future. While these models bound the minimum and
maximum price changes, there is uncertainty within those bounds. One common modeling
assumption is that world oil prices are fixed relative to policy changes. This would imply
perfectly elastic residual rest-of-world supply.
This analysis uses a single-period model which is parameterized for different years,
whereas dynamic effects are important in oil and natural gas markets. Production decisions
relating to drilling and shutting-in wells affect future production, well decline curves, and
intertemporal price arbitrage (the Hotelling Rule) (Hotelling, 1931). Consideration of dynamic
effects may shift numerical results. To the extent the proposed NSPS OOOOb and EG 0000c
may impact well drilling and shut-in decisions, the static analysis present here potentially
overlooks important distributional consequences of the proposed regulation.
This analysis does not distinguish between different regions of the United States. The
cost of producing oil and natural gas varies over the United States. Compliance costs may also
82 See https://www.whitehouse.gov/wp-content/uploads/2021/01/M-21-12.pdf.
103
-------
vary. Reductions in oil and natural gas production would be larger in regions with higher
production costs or higher compliance costs. This could result in different price changes in
different regions of the country if oil is there are bottlenecks in oil or natural gas shipping
infrastructure.
Oil and natural gas markets are linked on both the supply and demand sides. On the
supply side, individual wells generally produce a mixture of oil and natural gas, and some of the
same resources can be used to drill either oil-targeting wells or natural gas-targeting wells. On
the demand side, oil and natural gas are substitutes in some markets. Consideration of these
linkages may additionally shift numerical results.
4.2 Environmental Justice Analyses
Executive Order 12898 directs the EPA to "achiev[e] environmental justice (EJ) by
identifying and addressing, as appropriate, disproportionately high and adverse human health or
environmental effects" (59 FR 7629, February 16, 1994), termed disproportionate impacts in this
chapter. Additionally, Executive Order 13985 was signed to advance racial equity and support
underserved communities through Federal government actions (86 FR 7009, January 20, 2021).
The EPA defines EJ as the fair treatment and meaningful involvement of all people regardless of
race, color, national origin, or income with respect to the development, implementation, and
enforcement of environmental laws, regulations, and policies. The EPA further defines the term
fair treatment to mean that "no group of people should bear a disproportionate burden of
environmental harms and risks, including those resulting from the negative environmental
consequences of industrial, governmental, and commercial operations or programs and
policies".83 Meaningful involvement means that: (1) potentially affected populations have an
appropriate opportunity to participate in decisions about a proposed activity that will affect their
environment and/or health; (2) the public's contribution can influence the regulatory Agency's
decision; (3) the concerns of all participants involved will be considered in the decision-making
process; and (4) the rule-writers and decision-makers seek out and facilitate the involvement of
those potentially affected.
83 See, e.g., "Environmental Justice." Epa.gov, U.S. Environmental Protection Agency, 4 Mar. 2021,
https://www.epa.gov/environmentaljustice.
104
-------
The term "disproportionate impacts" refers to differences in impacts or risks that are
extensive enough that they may merit Agency action.84 In general, the determination of whether a
disproportionate impact exists is ultimately a policy judgment which, while informed by
analysis, is the responsibility of the decision-maker. The terms "difference" or "differential"
indicate an analytically discernible distinction in impacts or risks across population groups. It is
the role of the analyst to assess and present differences in anticipated impacts across population
groups of concern for both the baseline and proposed regulatory options, using the best available
information (both quantitative and qualitative) to inform the decision-maker and the public.
A regulatory action may involve potential EJ concerns if it could: (1) create new
disproportionate impacts on minority populations, low-income populations, and/or Indigenous
peoples; (2) exacerbate existing disproportionate impacts on minority populations, low-income
populations, and/or Indigenous peoples; or (3) present opportunities to address existing
disproportionate impacts on minority populations, low-income populations, and/or Indigenous
peoples through the action under development.
The Presidential Memorandum on Modernizing Regulatory Review (86 FR 7223;
January 20, 2021) calls for procedures to "take into account the distributional consequences of
regulations, including as part of a quantitative or qualitative analysis of the costs and benefits of
regulations, to ensure that regulatory initiatives appropriately benefit, and do not inappropriately
burden disadvantaged, vulnerable, or marginalized communities." Under Executive Order 13563,
federal agencies may consider equity, human dignity, fairness, and distributional considerations,
where appropriate and permitted by law. For purposes of analyzing regulatory impacts, the EPA
relies upon its June 2016 "Technical Guidance for Assessing Environmental Justice in
Regulatory Analysis,"85 which provides recommendations that encourage analysts to conduct the
highest quality analysis feasible, recognizing that data limitations, time, resource constraints, and
analytical challenges will vary by media and circumstance.
A reasonable starting point for assessing the need for a more detailed EJ analysis is to
review the available evidence from the published literature and from community input on what
84 See https://www.epa.gov/environmentaljustice/technical-guidance-assessing-environmental-justice-regulatory-
analysis.
85 Ibid.
105
-------
factors may make population groups of concern more vulnerable to adverse effects (e.g.,
underlying risk factors that may contribute to higher exposures and/or impacts). It is also
important to evaluate the data and methods available for conducting an EJ analysis. EJ analyses
can be grouped into two types, both of which are informative, but not always feasible for a given
rulemaking:
1. Baseline: Describes the current (pre-control) distribution of exposures and risk, identifying potential
disparities.
2. Policy: Describes the distribution of exposures and risk after the regulatory option(s) have been applied
(post-control), identifying how potential disparities change in response to the rulemaking.
EPA's 2016 Technical Guidance does not prescribe or recommend a specific approach or
methodology for conducting EJ analyses, though a key consideration is consistency with the
assumptions underlying other parts of the regulatory analysis when evaluating the baseline and
regulatory options.
4.2.1 Analyzing EJ Impacts in This Supplemental Proposal
For this proposed rulemaking, the EPA conducted limited environmental justice (EJ)
analyses focused on a baseline distribution of emissions from oil and natural gas sources. EJ
analyses described in this section evaluate only baseline scenarios; this enables us to characterize
risks due to oil and natural gas emissions prior to implementation of the proposed rule. However,
we lack key information that would be needed to characterize post-control risks under the
proposed NSPS OOOOb and EG 0000c or the regulatory alternatives analyzed in this RIA.
Therefore, the extent to which this proposed rule will affect potential EJ concerns is not
evaluated explicitly due to data limitations that prevent us from analyzing spatially differentiated
outcomes.
As policy-specific air quality scenarios corresponding to future years analyzed in this
proposal (e.g., 2023 to 2035) were not evaluated, it is unknown how the proposed rule will
impact potential EJ concerns that may relate to the distribution of oil and natural gas emissions,
as well as those related to employment. Importantly, we note that this proposal may not impact
all locations with oil and natural gas emissions equally, in part due to differences in existing state
regulations in locations like Colorado and California, which have more stringent requirements.
106
-------
Additionally, these discussions and analyses are subject to various types of uncertainty related to
input parameters and assumptions.
We present several potential vulnerabilities to climate-related stress qualitatively in
Section 4.2.2. Quantitative EJ assessments include an analysis of ozone from oil and natural gas
VOC emissions (Section 4.2.3), risk from oil and natural gas air toxic emissions (Section 4.2.4),
oil and natural gas workers and communities (Section 4.2.5), and how households may be
affected by potential energy market impacts (Section 4.2.6). Overall, there is some evidence that
certain populations may be disproportionately impacted by oil and natural gas emissions,
although data gaps remain.
4.2.2 Climate Impacts
In 2009, under the Endangerment and Cause or Contribute Findings for Greenhouse
Gases Under Section 202(a) of the Clean Air Act ("Endangerment Finding"), the Administrator
considered how climate change threatens the health and welfare of the U.S. population. As part
of that consideration, she also considered risks to minority and low-income individuals and
communities, finding that certain parts of the U.S. population may be especially vulnerable based
on their characteristics or circumstances. These groups include economically and socially
disadvantaged communities; including those that have been historically marginalized or
overburdened; individuals at vulnerable lifestages, such as the elderly, the very young, and
pregnant or nursing women; those already in poor health or with comorbidities; the disabled;
those experiencing homelessness, mental illness, or substance abuse; and/or Indigenous or
minority populations dependent on one or limited resources for subsistence due to factors
including but not limited to geography, access, and mobility.
Scientific assessment reports produced over the past decade by the U.S. Global Change
Research Program (USGCRP, 2016, 2018), the Intergovernmental Panel on Climate Change
(IPCC) (IPCC, 2018; Oppenheimer et al., 2014; Porter et al., 2014; Smith et al., 2014), and the
National Academies of Science, Engineering, and Medicine add more evidence that the impacts
of climate change raise potential environmental justice concerns (National Academies, 2017;
NRC, 2011). These reports conclude that less-affluent, traditionally marginalized, or
predominantly non-White communities can be especially vulnerable to climate change impacts
because they tend to have limited adaptive capacities and are more dependent on climate-
107
-------
sensitive resources such as local water and food supplies or have less access to social and
information resources. Some communities of color, specifically populations defined jointly by
ethnic/racial characteristics and geographic location (e.g., African-American, Black, and
Hispanic/Latino communities; Native Americans, particularly those living on Tribal lands and
Alaska Natives), may be uniquely vulnerable to climate change health impacts in the United
States, as discussed below. In particular, the 2016 scientific assessment on the Impacts of
Climate Change on Human Health found with high confidence that vulnerabilities are place- and
time-specific, lifestages and ages are linked to immediate and future health impacts, and social
determinants of health are linked to greater extent and severity of climate change-related health
impacts (USGCRP, 2016).
Per the Fourth National Climate Assessment, "Climate change affects human health by
altering exposures to heat waves, floods, droughts, and other extreme events; vector-, food- and
waterborne infectious diseases; changes in the quality and safety of air, food, and water; and
stresses to mental health and well-being" (Ebi et al., 2018). Many health conditions such as
cardiopulmonary or respiratory illness and other health impacts are associated with and
exacerbated by an increase in greenhouse gases and climate change outcomes, which is
problematic as these diseases occur at higher rates within vulnerable communities. Importantly,
negative public health outcomes include those that are physical in nature, as well as mental,
emotional, social, and economic.
The scientific assessment literature, including the aforementioned reports, demonstrates
that there are myriad ways in which these populations may be affected at the individual and
community levels. Outdoor workers, such as construction or utility workers and agricultural
laborers, who are frequently part of already at-risk groups, are exposed to poor air quality and
extreme temperatures without relief. Furthermore, individuals within EJ populations of concern
face greater housing and clean water insecurity and bear disproportionate economic impacts and
health burdens associated with climate change effects. They have less or limited access to
healthcare and affordable, adequate health or homeowner insurance. The urban heat island effect
can add additional stress to vulnerable populations in densely populated cities who do not have
access to air conditioning. Finally, resiliency and adaptation are more difficult for economically
disadvantaged communities: They tend to have less liquidity, individually and collectively, to
move or to make the types of infrastructure or policy changes necessary to limit or reduce the
108
-------
hazards they face. They frequently face systemic, institutional challenges that limit their power
to advocate for and receive resources that would otherwise aid in resiliency and hazard reduction
and mitigation.
The assessment literature cited in EPA's 2009 and 2016 Endangerment Findings, as well
as Impacts of Climate Change on Human Health, also concluded that certain populations and
people in particular life stages, including children, are most vulnerable to climate-related health
effects (USGCRP, 2016). The assessment literature produced from 2016 to the present
strengthens these conclusions by providing more detailed findings regarding related
vulnerabilities and the projected impacts youth may experience. These assessments — including
the Fourth National Climate Assessment (2018) and The Impacts of Climate Change on Human
Health in the United States (2016) — describe how children's unique physiological and
developmental factors contribute to making them particularly vulnerable to climate change.
Impacts to children are expected from heat waves, air pollution, infectious and waterborne
illnesses, and mental health effects resulting from extreme weather events (USGCRP, 2016). In
addition, children are among those especially susceptible to allergens, as well as health effects
associated with heat waves, storms, and floods. Additional health concerns may arise in low-
income households, especially those with children, if climate change reduces food availability
and increases prices, leading to food insecurity within households. More generally, these reports
note that extreme weather and flooding can cause or exacerbate poor health outcomes by
affecting mental health because of stress; contributing to or worsening existing conditions, again
due to stress or also as a consequence of exposures to water and air pollutants; or by impacting
hospital and emergency services operations (Ebi et al., 2018). Further, in urban areas in
particular, flooding can have significant economic consequences due to effects on infrastructure,
pollutant exposures, and drowning dangers. The ability to withstand and recover from flooding is
dependent in part on the social vulnerability of the affected population and individuals
experiencing an event (National Academies, 2019).
The Impacts of Climate Change on Human Health also found that some communities of
color, low-income groups, people with limited English proficiency, and certain immigrant groups
(especially those who are undocumented) live with many of the factors that contribute to their
vulnerability to the health impacts of climate change (USGCRP, 2016). While difficult to isolate
from related socioeconomic factors, race appears to be an important factor in vulnerability to
109
-------
climate-related stress, with elevated risks for mortality from high temperatures reported for
Black or African American individuals compared to White individuals after controlling for
factors such as air conditioning use. Moreover, people of color are disproportionately exposed to
air pollution based on where they live, and disproportionately vulnerable due to higher baseline
prevalence of underlying diseases such as asthma, so climate exacerbations of air pollution are
expected to have disproportionate effects on these communities.
The recent EPA report on climate change and social vulnerability examined four socially
vulnerable groups (individuals who are low income, minority, without high school diplomas,
and/or 65 years and older) and their exposure to several different climate impacts (air quality,
coastal flooding, extreme temperatures, and inland flooding) (U.S. EPA, 2021c). This report
found that Black and African-American individuals were 40 percent more likely to currently live
in areas with the highest projected increases in mortality rates due to climate-driven changes in
extreme temperatures, and 34 percent more likely to live in areas with the highest projected
increases in childhood asthma diagnoses due to climate-driven changes in particulate air
pollution. The report found that Hispanic and Latino individuals are 43 percent more likely to
live in areas with the highest projected labor hour losses in weather-exposed industries due to
climate-driven warming, and 50 percent more likely to live in coastal areas with the highest
projected increases in traffic delays due to increases in high-tide flooding. The report found that
American Indian and Alaska Native individuals are 48 percent more likely to live in areas where
the highest percentage of land is projected to be inundated due to sea level rise, and 37 percent
more likely to live in areas with high projected labor hour losses. Asian individuals were found
to be 23 percent more likely to live in coastal areas with projected increases in traffic delays
from high-tide flooding. Those with low income or no high school diploma are about 25 percent
more likely to live in areas with high projected losses of labor hours, and 15 percent more likely
to live in areas with the highest projected increases in asthma due to climate-driven increases in
particulate air pollution, and in areas with high projected inundation due to sea level rise.
Indigenous communities possess unique vulnerabilities to climate change, particularly
those communities impacted by degradation of natural and cultural resources within established
reservation boundaries and threats to traditional subsistence lifestyles. Indigenous communities
whose health, economic well-being, and cultural traditions depend upon the natural environment
will likely be affected by the degradation of ecosystem goods and services associated with
110
-------
climate change. The IPCC indicates that losses of customs and historical knowledge may cause
communities to be less resilient or adaptable (Porter et al., 2014). The Fourth National Climate
Assessment (2018) noted that while Indigenous peoples are diverse and will be impacted by the
climate changes universal to all Americans, there are several ways in which climate change
uniquely threatens Indigenous peoples' livelihoods and economies (Jantarasami et al., 2018;
USGCRP, 2018). In addition, there can be institutional barriers to their management of water,
land, and other natural resources that could impede adaptive measures.
For example, Indigenous agriculture in the Southwest is already being adversely affected
by changing patterns of flooding, drought, dust storms, and rising temperatures leading to
increased soil erosion, irrigation water demand, and decreased crop quality and herd sizes. The
Confederated Tribes of the Umatilla Indian Reservation in the Northwest have identified climate
risks to salmon, elk, deer, roots, and huckleberry habitat. Housing and sanitary water supply
infrastructure are vulnerable to disruption from extreme precipitation events. Confounding
general Native American response to natural hazards are limitations imposed by policies such as
the Dawes Act of 1887 and the Indian Reorganization Act of 1934, which ultimately restrict
Indigenous peoples' autonomy regarding land-management decisions through Federal trusteeship
of certain Tribal lands and mandated Federal oversight of management decisions.
Additionally, the Fourth National Climate Assessment noted that Indigenous peoples are
subjected to institutional racism effects, such as poor infrastructure, diminished access to quality
healthcare, and greater risk of exposure to pollutants. Consequently, Native Americans often
have disproportionately higher rates of asthma, cardiovascular disease, Alzheimer's, diabetes,
and obesity, which can all contribute to increased vulnerability to climate-driven extreme heat
and air pollution events. These factors also may be exacerbated by stressful situations, such as
extreme weather events, wildfires, and other circumstances.
The Fourth National Climate Assessment and IPCC AR5 also highlighted several impacts
specific to Alaskan Indigenous Peoples (Porter et al., 2014). Coastal erosion and permafrost thaw
will lead to more coastal erosion, rendering winter travel more risky and exacerbating damage to
buildings, roads, and other infrastructure - these impacts on archaeological sites, structures, and
objects that will lead to a loss of cultural heritage for Alaska's Indigenous people. In terms of
food security, the Fourth National Climate Assessment discussed reductions in suitable ice
111
-------
conditions for hunting, warmer temperatures impairing the use of traditional ice cellars for food
storage, and declining shellfish populations due to warming and acidification. While the Fourth
National Climate Assessment also noted that climate change provided more opportunity to hunt
from boats later in the fall season or earlier in the spring, the assessment found that the net
impact was an overall decrease in food security.
4.2.3 Ozone from Oil and Natural Gas VOC Emission Impacts86
To evaluate the EJ implications of ozone from oil and natural gas VOC emissions from
the oil and natural gas sector, we analyzed a recent baseline (pre-control) air quality scenario
comparing exposures to ozone formed from VOC emissions from the oil and natural gas sector
across races/ethnicities, ages, and sexes. We focus mainly on exposure differences because these
provide the clearest view into whether emissions from this sector may be unequally distributed
among population subgroups of interest.
4.2.3.1 Data Inputs
Input data for this ozone exposure EJ analysis included potential population
characteristics of concern, and air quality scenarios.
86 The illustrative screening analysis of projected ozone-related health benefits from VOC reductions under the
primary proposal (presented in Appendix C of the proposal RIA) was subject to uncertainties in addition to those
associated with the baseline ozone-related environmental justice analysis presented in this section. For example, the
VOC emissions contributing to baseline concentrations of ozone in the environmental justice analysis are derived
from the NEI, while the emissions reductions projected under the proposal for this RIA are based upon a mix of
model plant information used in the rulemaking and activity factors as described in Section 2.2. Importantly, the
illustrative screening analysis projects emissions reductions at a national-level while the NEI-based emissions
informing the air quality modeling underpinning the environmental justice analysis are more spatially resolved.
Importantly, insufficient scientific evidence and technical limitations prevent us from stratifying relationships
between ozone exposures and health effects, we quantitatively assessed EJ exposure impacts of oil and natural gas
ozone from VOC emissions in the baseline among certain subpopulations of interest. As noted, we stopped short of
characterizing the respiratory mortality risk among these populations, or drawing comparisons among them, due to
the impact on results caused by differences in the age distributions, and therefore baseline incidence rates of
respiratory mortality, of White and non-White populations. In addition, risk results are strongly influenced by the
age distributions of various potential EJ subpopulations. Specifically, populations with higher median ages and those
with larger populations of older adults (e.g., White populations), are associated with substantially higher baseline
incidence rates of respiratory mortality. Higher baseline mortality rates translate into higher estimates of risk that
can obfuscate impacts from small differences in ozone exposure levels. Therefore, we removed the ozone EJ
mortality analyses previously presented, to make this RIA more consistent with previous RIA's involving ozone
concentration changes.
112
-------
(a) Population Characteristics
A reasonable starting point for assessing the need for a more detailed EJ analysis is to
review the available evidence from the published literature and from community input on what
factors may make population groups of concern more vulnerable to adverse effects. The Health
Effects Institute (HEI) provided a bibliography of peer-reviewed studies published since 2015
that evaluate populations that may be disproportionately impacted by the oil and natural gas
industry.87 However, there is considerable discordance among the study results. For example,
studies differ with regards to geographic area, population of interest, and health outcome. To
broadly assess potential EJ concerns, we evaluated disproportionate exposure and risk across
racial and ethnic demographics, sexes, and ages as described in Table 4-6.
Table 4-6 Components of the Criteria Pollutant Environmental Justice Assessment
(b) Air Quality Scenarios
Here we utilize modeled baseline conditions of ozone formed from oil and natural gas
VOC emissions developed for the year 2017 (Figure 4-1) (U.S. EPA, 2021a). These air quality
surfaces were developed using source apportionment (SA) modeling estimates of ozone
concentrations attributable to certain precursors such as VOC from individual sectors, which can
provide insight into the baseline (i.e., pre-rulemaking) scenario of a historical year (Appendix C,
Section C. 1.2).88 Please note the scale, as concentrations of ozone formed from oil and natural
gas VOC emissions represent a relatively small proportion of median annual MDA8
concentrations.89 Higher concentrations of ozone formed from oil and natural gas VOC emissions
tend to localize to areas of known oil and natural gas facility locations.
87 Email to EPA staff from Janet McGovern of the Health Effects Institute on May 12th, 2021. Located at Docket ID
No. EPA-HQ-0AR-2021-0317.
88 Additional information on the SA modeling is available from U.S. EPA (2021a).
89 Median annual MDA8 ozone concentration in 2015-2017 were 40 parts per billion (ppb); see Table 1-1 in U.S.
EPA (2020b).
EJ Characteristics
Description
Race
Ethnicity
Age
Sex
White, Black, Asian, Native American
Hispanic, Non-Hispanic
0-17, 18-64, 65-99
Male, Female
113
-------
<- 0 020
0,020 - 0.050
0.050 - 0.090
0090-0.140
|| 0140-0.220
f 0.220- 0.450
| 0,450-1120
\
7
llMaililM
y "
>1,120 .jf*
Figure 4-1 Map of Baseline Ozone Concentrations from Oil and Natural Gas VOC
Emissions in 2017
4.2.3.2 Results
Results of this ozone EJ analysis include the average (Section 4.2.3.2(a)) and distribution
(Section 4.2.3.2(b)) of ozone exposures.
(a) Average Ozone Exposures
Average mean daily 8-hour maximum (MDA8) ozone concentrations from oil and natural
gas VOC emissions between April and September of 2017 are shown in Figure 4-2. Exposures
for the overall reference group, adults of all races/ethnicities and sexes aged 30-99, is shown in
the top row, with population specific comparisons available below. For example, this baseline
analysis shows that Native American populations on average may be exposed to a higher
concentration of ozone from oil and natural gas VOC emissions than White populations, who in
turn may on average be exposed to a higher concentration than the overall reference group.
Similarly, the analysis suggests that Hispanic populations on average are exposed to a higher
concentration of ozone from oil and natural gas VOC emissions than both non-Hispanic
114
-------
individuals and the overall reference group. The right column also provides information
regarding the number of people within each demographic group. For example, there were less the
2 million Native Americans and nearly 30 million Hispanics in the contiguous U.S. in 2017.
African American or Black populations and Asian populations may on average be
exposed to lower concentrations than White populations and the overall reference group.
Regarding sex, females and males are estimated to be exposed to similar concentrations as
compared to the reference group. Finally, when comparing average exposure across age ranges,
ozone concentrations from oil and natural gas VOC emissions appears to decrease as age
increases.
Ages Sex
30-99 All
Female
Male
0-17 All
18-64 All
30-64 All
65-99 All
Figure 4-2
Population and Corresponding 2017 Population Counts
(b) Distribution of Ozone Exposures
While average exposure concentrations within demographic populations can convey
some insight, distributional information, while more complex, can provide a more
comprehensive understanding of the analytical results. As such, using the same baseline scenario
described above, we provide the running sum percentage of each population plotted against the
increasing ozone concentration from oil and natural gas VOC emissions in Figure 4-3 to permit
the direct comparison of demographic populations with different absolute numbers. While the
analysis indicates that exposures to ozone from oil and natural gas VOC emissions may be
similar across all races/ethnicities in the lower 60 percent of each population, it suggests there
Race/Ethnicity
All
White
Black
Asian
Native American
Non-Hispanic
Hispanic
All
AH
All
All
AH
All
0.085 0.090 0.095 0.100 50M 100M 150M ZOOM
Ozone from Oil & Gas VOC (ppb) Population (2017)
Average Ozone Concentrations from Oil and Natural Gas VOC Emissions by
115
-------
are small differences in the 65-95 percent of populations exposed to higher ozone concentrations
from oil and natural gas VOC emissions in some populations. Notably, a subset of Hispanics and
Native American populations, shown in the dark and light orange lines, respectively, may
experience slightly higher exposures to ozone from oil and natural gas VOC emissions than
White and non-Hispanic populations.
White
Non-Hispanic
Asian
Black
Native American
Hispanic
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Ozone from Oi! & Natural Gas VOC Emissions (ppb)
Figure 4-3 Distributions of Ozone from Oil and Natural Gas VOC Emissions
Concentrations by Race/Ethnicity
Figure 4-4 shows the distribution of ozone from oil and natural gas VOC emissions
across three age ranges, 0-17 shown in blue, 18-64 shown in black, and 65-99 shown in orange.
Differences are very small between the three age groups, but the baseline analysis suggests
exposure decreases as the age range increases.
116
-------
<18
18-64
>64
Ozone from Oil & Gas VOC Emissions (ppb)
Figure 4-4 Distributions of Ozone from Oil and Natural Gas VOC Emissions
Concentrations by Age Range
Figure 4-5 shows the distribution of ozone from oil and natural gas VOC emissions
across males (orange) and females (blue) from our analysis. The distribution of exposures is
virtually identical between the two sexes.
100%
9DK-
!
80*-
!
70%-
a*
*5
60%
1
50K-
a
40*
1
30*-
I
20%-
10S-
0'„
Females
Males
oo
0,1 0 2 m 0,4 o.s
Ozone from Oil & Gas VOC Emissions {i>obl
0.6
Figure 4-5 Distributions of Ozone from Oil and Natural Gas VOC Emissions
Concentrations by Sex
117
-------
(c) Ozone EJ Summary
This recent baseline ozone EJ analysis suggests that there may be some small differences
in exposures to ozone formed from VOC emissions from the oil and natural gas sector across
races/ethnicities and certain age groups. It also suggests that a substantial portion of ozone from
oil and natural gas VOC emissions are localized to rural areas where fewer people reside.
However, we lack the data to evaluate this on a more site-specific basis. Additionally, given the
size of the sector and the number of oil and natural gas locations, it is quite possible that
localized disparities may exist that our analysis did not identify.
4.2.4 Air Toxics Impacts
To evaluate the potential EJ impacts associated with baseline HAP emissions from the oil
and natural gas sector, the EPA has assessed the cancer risks and estimated the demographic
breakdown of people living in areas with potentially elevated risk levels. Typically, when we
perform risk assessments of source categories (e.g., for Risk and Technology Review [RTR]
rulemakings), we have detailed location and emissions data for each facility to be assessed and
we estimate human health risks at the census block level. For the oil and natural gas sector we do
not have such detailed data readily available. We used the most recent National Emissions
Inventory (NEI) data from 2017, which indicates nationwide emissions of approximately
110,000 tons of HAP for that year from oil and natural gas sources (see Table 3-6).
The 2017 NEI includes emissions from the sources subject to regulation and sources
outside of the regulation. It does not contain refined emissions estimates from only the sources
subject to the regulation. The result of this is that we cannot estimate risks from the source
category alone, but rather only from the larger industry sector. Another result is that the
assessment is considered a screen — it is an estimate of potential risks over a broad area. More
refined emissions data would need to be obtained to conduct an assessment where we could draw
more accurate conclusions about risk to specific areas and populations.
Most of these emissions (97 percent) are treated as "nonpoint" emissions which are
allocated from county-level data down to grid cells (4 km in the continental U.S. (CONUS), 9
km in Alaska) based on emissions surrogates. This means that we are making assumptions about
the spatial distribution of these emissions that may not be accurate. The approximately 3 percent
118
-------
of emissions that are categorized as "point" in the NEI are emitted from about 400 facilities
across the country. For these sources, we are able to estimate potential exposures and impacts
more precisely. Also, we note that some sources categorized as oil and natural gas sources in the
NEI are not in the source category for this proposed rule.
The oil and natural gas sector was one of the sectors assessed in the 2014 National Air
Toxics Assessment (NATA). In that assessment, the nonpoint emissions were also modeled as 4
km grid cells in CONUS (9 km grid cells in Alaska) and the point emissions were modeled as
point sources in the American Meteorological Society/Environmental Protection Agency
Regulatory Model (AERMOD) using census blocks as model receptors. However, NATA risk
estimates were not presented at census block level because of uncertainties associated with the
analysis, such as not knowing exactly where in each grid cell the emissions are actually
occurring. Instead, NATA risk results were presented at census tract level by population-
weighting the block risks up to the tract level. Because census tracts can have large areas, the
tract-level risks may not reflect potential elevated risks present at a finer scale. The highest tract-
level cancer risk from nonpoint oil and natural gas emissions in the 2014 NATA was 30-in-l
million, and only about 30 tracts (out of approximately 74,000 tracts nationwide) had risks
greater than 10-in-l million. For comparison, the nationwide median total cancer risk estimate
from the 2014 NATA (considering contributions from all source types) was about 30-in-l
million across all census tracts.
Here, using updated emissions and population data, we have conducted a new analysis of
HAP-related exposures and risks across the United States. In this analysis, to assess the potential
for elevated risks at a scale finer than the census tract level, we aggregated the block-level
AERMOD results from the modeling of the 2017 NEI nonpoint HAP emissions to the same 4 km
and 9 km grid cells that nonpoint emissions are allocated to. There are about 500,000 4 km grid
cells in CONUS, compared to about 74,000 census tracts so, on average, grid cells are at a finer
scale than census tract. For each grid cell, we used the median cancer risk of all the blocks that
have their internal point (or centroid) located within the grid cell. Census block demographic
data were also aggregated to each 4 km grid cell and risks were calculated at the census blocks
from the approximately 400 sources included in the 2017 NEI as point sources and added the
119
-------
highest block-level risk for each point source "facility" to the median cell nonpoint risk for the
cell containing the block.
The data used in this analysis include spatial data of the grid cells, 2010 census block
location and population data,90 AERMOD-modeled oil and natural gas 2017 HAP concentrations
at census block level for the nonpoint and point sources, and 2015-2019 block-group
demographic data. There are separate files for the 4 km grid cells that cover CONUS and the 9
km grid cells for Alaska, each using a Lambert Conformal Conic projected coordinate system.
These are the same grid definition used for the 2014 NATA nonpoint oil and natural gas
emissions. The census data are for the year 2010, with a small number of changes made to the
locations (and sometimes deletions) of specific census blocks based on the RTR pre-modeling
review of specific source categories since the 2010 census data were first available (the current
oil and natural gas AERMOD modeling is based on the census block receptor file as of May
2019). The AERMOD modeling performed (version 19191) using 2017 NEI and meteorology
data followed the same methodology used in the 2014 NATA (U.S. EPA, 2018b). Demographic
data on total population, race, ethnicity, age, education level, low household income, poverty
status and linguistic isolation were obtained from the Census' American Community Survey
(ACS) 5-year averages for 2015-2019.91
The AERMOD-modeled census block concentrations are based on the 2017 NEI
emissions data (see Table 3-6). The process by which emissions were calculated and allocated to
grid cells in the case of nonpoint emissions is discussed in the technical support document for the
2017 NEI and the emissions modeling summary for 2017, respectively (U.S. EPA, 2020b).
Emissions data are publicly available online.92 These emissions were modeled in AERMOD
(version 19191), and the resulting block-level annual concentrations of each pollutant were used
to calculate cancer risks. The pollutant cancer unit risk estimates used to calculate risks are from
the toxicity value files available on the Human Exposure Model website.93 For each census
block, the cancer risks were summed over all pollutants to obtain a total cancer risk. The
90 Data Summary File 1 available at http://www2.census.gov/census_2010/04-Summary_File_l/. See also Technical
Documentation for the 2010 Census Summary File 1.
91 Data available at https://www2.census.gov/programs-surveys/acs/summary_file/2019/data/5_year_entire_sf/.
92 Data available at https://gaftp.epa.gov/Air/emismod/2017/AERMOD_inputs/.
93 See https://www.epa.gov/fera/download-human-exposure-model-hem.
120
-------
demographic data from the ACS were joined to each census block based on the block group ID
(the first 12 characters of the census block ID).
For nonpoint sources, the census blocks were spatially joined to the grid cells (4 km
CONUS, 9 km Alaska), and the block data were aggregated at the cell level, using the median
cancer risk of the blocks in each cell, and the sum of block populations and the individual
demographic group populations (using QGIS version 3.16.3). For point sources, the highest
modeled block risk for each facility was added to the median nonpoint risk for the cell containing
the block, to provide a measure of total point and nonpoint combined risk.
There are approximately 3 million census blocks with nonzero total risk from oil and
natural gas sources based on the AERMOD modeling of the CONUS nonpoint emissions, and
these blocks are within approximately 159,000 4 km grid cells. In Alaska, there are
approximately 3,500 census blocks with nonzero total risk from oil and natural gas sources based
on the AERMOD modeling, and these blocks are within approximately 240 9 km grid cells. In
CONUS, the 90th percentile cell risk estimate attributed to oil and natural gas sources is less than
1-in-l million (0.8-in-l million) and the 99.9th percentile estimate is 40-in-l million. The
maximum cell risk estimate from oil and natural gas sources is 200-in-l million, which occurs in
two grid cells with an estimated 10 people (3 census blocks,); Carbon County, Wyoming (with
an estimated 3 people) and Weld County, Colorado (with an estimated 7 people). The 2014
NATA results for HAP risk from all sources described above (i.e., nationwide median total
cancer risk estimate from all source types of approximately 30-in-l million), can provide context
for these risk results for 2017 HAP emissions from oil and natural gas sources. The CONUS
results are summarized in Table 4-7. There are about 9500 cells containing about 6.8 million
people where the cell risk estimate is greater than 1-in-l million. There are 122 cells containing
about 140,000 people where the cell risk estimate is greater than or equal to 50-in-l million, and
there are 36 cells containing about 40,000 people where the cell risk estimate is greater than or
equal to 100-in-l million. None of the cells in Alaska has estimated cell cancer risk greater than
1-in-l million.
It is important to reiterate that these risk estimates are based on emissions from the entire
oil and gas sector, which includes sources outside the scope of this regulation. To provide some
context for how these sources relate to sources impacted by this proposed regulation, we
121
-------
categorized the fraction of oil and natural gas HAP emissions in the 2017 NEI that were
attributed to different source types. For this exercise, we specifically focused on formaldehyde
and benzene emissions (the two pollutants that accounted for most of the calculated oil and
natural gas HAP risk) in the 36 grid cells with 2017 oil and natural gas HAP risk above 100-in-l
million. It is likely that a majority of the formaldehyde emissions and about a quarter of the
benzene emissions that were categorized as coming from oil and natural gas sources in the 2017
NEI are from sources outside of this source category. Therefore, it also follows that a majority of
the estimated risk is likely being driven by sources not impacted by this proposed regulation. It
bears repeating that this is a screening assessment and full modeling would be required to
quantitatively split out risk of sources impacted by this rule from other sources categorized in the
NEI as oil and natural gas. Risk in grid cells of interest may not scale directly to emissions
within the grid cells.
For the point sources, there were 33 sources with estimated census block maximum
cancer risk greater than 1-in-l million, and only 6 sources with estimated risk greater than 10-in-
1 million (highest was 40-in-l million). There was only a single case where the maximum census
block risk from a point source, and the median cell risk from nonpoint sources (containing the
census block), were both greater than 10-in-l million. In that case, the point risk of 20-in-l
million and the nonpoint cell risk of 40-in-l million combined for an estimated 60-in-l million
risk.
Figure 4-6 shows the cell cancer risk estimates in CONUS and Alaska. As indicated in
the map, most of the cells in the country (about 150,000 of them) have estimated risk less than 1-
in-1 million. Figure 2 is a larger-scale map that shows where the estimated cell risks are the
highest. The cells with estimated risk greater than or equal to 30-in-l million are in Colorado,
Utah, Wyoming, and North Dakota, and the cells with the highest estimated risk are all in
Colorado.
Table 4-7 also contains estimated numbers of people within various demographic groups
who live in areas above the specified risk levels. For nearly all of the demographic groups the
percentage of people in the cells with estimated risk above the specified levels is at or below the
national average. Above a risk level of 50-in-l million, the percent minority is about the same as
the national average, but the Hispanic/Latino demographic group is about 10 percentage points
122
-------
higher than the national average. The overall minority percentage is not elevated compared to the
national average because the African American percentage is much lower than the national
average. The demographic group of people aged 0-17 is slightly higher than the national
average. For people with estimated risk greater than 1-in-l million, Hispanic/Latino populations
and the age 0-17 group are below the national average, but the percentage of Native American
populations is higher than the national average.
Table 4-7
Emissions
Cancer Risk and Demographic Population Estimates for 2017 NEI Nonpoint
Number of Cells
Total Population
Risks > 100-in-l
million
Risks > 50-in-l
million
Risks > 1-in-l
million
36
38,885
(936 census blocks)
122
142,885
(3,204 census
blocks)
9,499
6,804,691
(172,878 census
blocks)
Nationwide
Population
%
Population
%
Population
%
%
Minority
13,268
34.1
52,154
36.5
2,010,161
29.5
39.9
African American
140
0.4
1,434
1
535,055
7.9
12.2
Native American
77
0.2
465
0.3
59087
0.9
0.7
Other and Multiracial
1,443
3.7
5,148
3.6
323,397
4.8
8.2
Hispanic or Latino
11,608
29.9
45,107
31.6
1,092,621
16.1
18.8
Age 0-17
10,679
27.5
37,487
26.2
1,463,907
21.5
22.6
Age >65
4,272
11
17,188
12
1,085,067
15.9
15.7
Below the Poverty Level
2,000
5.1
13,455
9.4
902,472
13.2
13.4
Over 25 Without a High
School Diploma
2,788
7.2
11,320
7.9
488,372
7.2
12.1
Linguistically Isolated
808
2.1
4,418
3.1
179,739
2.6
5.4
123
-------
Figure 4-6 National Map of Grid Cell Median Cancer Risks for 2017 Nonpoint Oil and
Natural Gas NEI Emissions
124
-------
Figure 4-7 Local-Scale Map of Grid Cell Median Cancer Risks for 2017 Nonpoint Oil
and Natural Gas NEI Emissions
4.2.5 Demographic Characteristics of Oil and Natural Gas Workers and Communities
The oil and natural gas industry directly employs approximately 140,000 people in oil
and natural gas extraction, a figure which varies with market prices and technological change, in
addition to a large number of workers in related sectors that provide materials and services.
Figure 4-8 shows employment since 2001.94 We see a dramatic increase in employment with the
rapid expansion in hydraulic fracturing from 2005 to 2014, a decrease after oil prices fell in
2014-2015, and volatility in employment.
94 Data was obtained from the Bureau of Labor Statistics Current Employment Statistics program for NAICS code
211.
125
-------
200
g 180
o
0
"c
1 160
>
o
a.
,1140
120
100
Figure 4-8 National-level Employment in Oil and Natural Gas Production
The EPA also conducted a baseline analysis to characterize potential distributional
impacts on employment. A reduction in oil and natural gas activity could have a negative effect
on employment among oil and natural gas workers. This could also reduce employment,
earnings, and tax revenues in oil and natural gas intensive communities.95 Any effect on oil and
natural gas workers or oil and natural gas intensive locations would be a local and partial
equilibrium effect. In general equilibrium, there could be other and potentially offsetting effects
in other regions and sectors.
For the distribution of employment effects, we assessed the demographic characteristics
of (1) workers in the oil and gas sector and (2) people living in oil and natural gas intensive
communities. Comparing workers in the oil and natural gas sector to workers in other sectors, oil
and natural gas workers may have higher than average incomes, be more likely to have
completed high school, and be disproportionately Hispanic. People living in some oil and natural
O
95 For this analysis, oil and natural gas intensive communities are defined as the top 20 percent of communities with
respect to the proportion of oil and natural gas workers. Some analyses break the top 20 percent into subgroups
which are the 80th-95th percentiles, the 95th-97.5th percentiles, and above the 97.5th percentile by proportion of oil
and natural gas workers.
126
-------
gas-intensive communities concentrated in Texas, Oklahoma, and Louisiana, may have
disproportionate income levels, rates of high school completion, and demographic composition.
Table 4-8 provides summaries of average income, the percentage of population that is
non-Hispanic White, the percentage of population that speaks only English in the home, and the
percentage of the population with four years of high school education, all among people with
reported income. The table lists these data for the United States, for oil and natural gas workers,
for other people, for people in oil and natural gas intensive communities, and for people in other
locations. We see that oil and natural gas workers are more highly paid, more likely to be non-
Hispanic White individuals, and have higher rates of only speaking English and more likely to
have four years of high school than workers in other sectors. People in oil and natural gas
communities are demographically similar to people in other communities. This suggests that, on
average, reductions in oil and natural gas drilling or production are unlikely to disproportionately
impact marginalized communities either via direct labor channels or spillover channels.
127
-------
Table 4-8 Demographic Characteristics of Oil and Natural Gas Workers and
Communities
Sectors
Places
Overall
Oil and
Natural Gas
Other
Oil and Natural
Other
Workers
People
Gas Communities
Communities
All U.S.
Average Income
$110,000
$42,000
$40,000
$43,000
$42,000
% Non-Hispanic White
81%
71%
68%
69%
71%
% English Only
87%
82%
80%
81%
82%
4 years of High School
97%
88%
86%
88%
88%
Note: Calculations based on United States Census Bureau American Community Survey public use microdata from
2014-2019.
This analysis uses 5-year ACS data from 2015-2019 retrieved from IPUMS. This is
approximately 16 million individual ACS responses. Oil and natural gas workers are identified
by working in industries with a NAICS code that begins with "211 " Those are "Oil and natural
gas Extraction," as well as the sub-industries "Crude Petroleum Extraction" and "Natural Gas
Extraction."
The level of communities is the Public Use Microdata Area (PUMA). PUMAs are
districts defined by the United States Census Bureau. PUMA data is procured from IPUMS.
They generally have 100,000-200,000 people with an average of about 140,000 people. The
average spatial area of a PUMA is 1,692 square miles. We analyze PUMAs because economic
spillovers in this sector occur at a multicounty scale. The oil and natural gas sector includes both
substantial intercounty commuting and regional supply chains. Additionally, PUMAs are the
smallest geographic unit for which detailed individual data are available. In Table 4-8, oil and
natural gas communities are defined as the 20 percent of PUMAs with the highest percentage of
oil and natural gas workers. Figure 4-9 shows all PUMAs in the continental United States. Oil
and natural gas communities as defined in Table 4-8 are highlighted.
128
-------
Figure 4-9 Continental U.S. Map of PUMAs and Oil and Natural Gas Intensive
Communities
Table 4-9 describes demographics by a region's oil and natural gas (O&G) intensity.
Non-oil and natural gas intensive regions (column (1)) are the bottom 80 percent by portion of
workers in the oil and natural gas industry. Most of these have no reported oil and natural gas
workers. Low oil and natural gas intensive regions (column (2)) are between the 80th and 95th
percentiles of oil and natural gas industry employment, high (column (3) are the 95th-97.5th, and
very high (column (4)) are above the 97.5th percentile. People in oil and natural gas communities
of Table 4-9 are divided between columns (2)-(4). The trimmed comparison group (column (5)
is people in non-oil and natural gas intensive regions in states that contain any PUMAs with high
or very high intensity. The group of states with high oil and natural gas intensity may be a more
appropriate comparison by removing regions of the country which do not resemble oil and
natural gas intensive areas, such as the Atlantic coast states.
We see in Block A that people in oil and natural gas intensive communities (columns
(2)-(4)) are more likely to be White and Indigenous than people in non-oil and natural gas
intensive areas (column (1)). In Block B, we see that people in O&G intensive areas' more likely
to be Hispanic than people in non-O&G intensive areas. In Block C, we see income, percentage
of population with four years of high school education, and fraction working in the oil and
129
-------
natural gas industry. Comparing people in high and very high oil and natural gas intensity
regions (columns (3) and (4)) to people in the trimmed comparison group (column (5)), we see
that people in in high oil and natural gas intensity regions are more likely to be White, non-
Hispanic, Native American, and less likely to be Asian American or Pacific Islanders.
Table 4-9 Demographic Characteristics of Oil and Natural Gas Communities by Oil
and Natural Gas Intensity
(1)
(2)
(3)
(4)
(5)
Very High
Trimmed
Non-O&G
Low O&G
High O&G
O&G
Comparison
Category
Intensive
Intensity
Intensity
Intensity
Group
Block A:
White
77%
81%
84%
78%
73%
Black and African-American
10%
8%
8%
7%
8%
Native American
1%
2%
2%
3%
1%
Asian American or Pacific
Islander
6%
3%
2%
5%
9%
Other Race
4%
3%
2%
4%
7%
Multiple races
2%
2%
2%
3%
3%
Block B:
Non-Hispanic
88%
84%
86%
81%
80%
Hispanic
12%
16%
14%
19%
20%
Block C:
Income
$43,000
$39,000
$39,000
$45,000
$43,000
Four years of High School
88%
87%
87%
86%
87%
Fraction Working in O&G
0.00006
0.001
0.004
0.01
0.00008
Note: Calculations based on United States Census Bureau American Community Survey public use microdata from
2014-2019. Totals may not appear to add correctly due to rounding.
Table 4-10 shows the percentage of people by racial group identification for Hispanics
and non-Hispanics, across oil and natural gas intensity. We see that people in high and very high
intensity communities are more likely to be Hispanic Whites and non-Hispanic Native
Americans, and less likely to be non-Hispanic Asian American and Pacific Islanders than people
in non-oil and gas intensive communities.
130
-------
Table 4-10 Hispanic Population by Oil and Natural Gas Intensity
m (2) (3) (4) (51
Category
Non-
O&G
Intensive
Low
O&G
Intensity
High
O&G
Intensity
Very High
O&G
Intensity
Trimmed
Comparison
Group
Non-Hispanic White
69%
69%
73%
65%
60%
Non-Hispanic Black and African-American
10%
8%
7%
7%
8%
Non-Hispanic Native American
1%
2%
1%
3%
0%
Non-Hispanic Asian American or Pacific
Islander
6%
3%
2%
5%
9%
Non-Hispanic Other Race
0%
0%
0%
0%
0%
Non-Hispanic Multiple Races
2%
2%
2%
2%
2%
Hispanic White
8%
12%
11%
14%
12%
Hispanic Black and African-American
0%
0%
0%
0%
0%
Hispanic Native American
0%
0%
0%
0%
0%
Hispanic Asian American or Pacific Islander
0%
0%
0%
0%
0%
Hispanic Other Race
3%
3%
2%
4%
6%
Hispanic Multiple Races
1%
1%
0%
1%
1%
Note: Calculations based on United States Census Bureau American Community Survey public use microdata from
2014-2019. Totals may not appear to add correctly due to rounding.
Marginalized communities are overrepresented in some oil and natural gas intensive
communities. Figure 4-10 highlights oil and natural gas intensive communities with substantial
EJ communities in darker blue. These communities are in the bottom twenty-five percent by
income or high-school graduate or non-Hispanic White population percentage. They are
concentrated in Texas, Louisiana, and Oklahoma.
131
-------
-
rfal
v § 1 -
v
0
^v,
Hlh
¦r
¦3
Figure 4-10 Map of Oil and Natural Gas Intensive Communities of Environmental
Justice Note
4.2.6 Household Energy Expenditures
Energy provides many services to households that are necessary for a basic standard of
living. The proposed regulatory requirements will obligate affected sources to incur costs to
reduce emissions, which impact the supply and prices of oil and natural gas and generate energy
market impacts, though these impacts are expected to be minimal (see Section 4.1). This section
characterizes how household energy expenditures vary across the income distribution and for
different racial and ethnic groups. The goal of this section is to highlight which populations and
communities may be most vulnerable to potential energy market effects caused by regulatory
impacts on the oil and natural gas industry.
Energy insecurity, poverty, and access are important concepts in the discussion of energy
burden. Energy insecurity occurs when households lack certainty that they will be able to
consume adequate and sufficient energy to meet basic needs. Energy poverty exists when
households need to pay disproportionate costs for energy use due to low income, higher energy
bills, or inefficient energy use. Energy access barriers exist when households lack access to
affordable, reliable energy. Energy insecurity and poverty are persistent problems facing many
132
-------
households across the U.S. (Bednar & Reames, 2020; EIA, 2018; Kaiser & Pulsipher, 2006) and
they have many consequences for human health and wellbeing (Hall, 2013; Jessel, Sawyer, &
Hernandez, 2019; Karpinska & Smiech, 2020). The EIA found that nearly a third of U.S.
households faced challenges paying their energy bills or could not maintain adequate heating or
cooling in 2015. For purposes of this section, "energy burden" focuses primarily on energy
poverty.
Low-income and minority households tend to face disproportionately high energy
burdens (Hernandez, Aratani, & Jiang, 2014; Wang, Kwan, Fan, & Lin, 2021) and thus are
particularly vulnerable when energy prices increase. Although these households consume less
energy, energy tends to represent a larger share of their budgets. Drehobl, Ross, and Ayala
(2020) find that low-income, Black, Hispanic, Native American, and older adult households have
disproportionally higher energy burdens than the average household. Lyubich (2020) finds that
Black households spend more on residential energy than White households even after controlling
for income, household size, city, and homeowner status. Wang et al. (2021) find that Black
households spent more on energy than other households at every point on the income
distribution, suggesting that energy efficiency issues may be more problematic in Black
households. They identify geographic location, climate, the characteristics of dwellings, and
socioeconomic characteristics as primary drivers of residential energy use and energy burden.
To investigate baseline energy expenditures and potential distributional impacts of
possible increases in energy costs, we assessed expenditure and income data stratified by pre-tax
income quintiles and race/ethnicity from the 2019 Consumer Expenditure Survey (CES) from the
U.S. Bureau of Labor Statistics. We combined expenditures in the following four categories to
approximate "energy expenditures": (1) Natural gas, (2) Electricity, (3) Fuel oil and other fuels,
and (4) Gasoline, other fuels, and motor oil (transportation). The first three categories are
residential energy expenditures, and the fourth category represents transportation energy
expenditures. These categories are assumed to potentially experience price impacts due to
regulatory costs affecting the oil and natural gas industry, though we expect impacts to be
minimal (see Section 4.1).
We examined energy expenditures, the ratio of household energy expenditures to total
household expenditures, and the ratio of household energy expenditures to after-tax income
133
-------
across income quintiles and racial groups. It is important to note that energy burden is sensitive
to the particular energy services and expenditures are included and how income is defined (e.g.,
whether transfer payments or taxes are included in income calculation; the inclusion of
transportation-related energy expenditures).
Table 4-11 shows energy expenditures by quintiles of pre-tax income. The data indicate
that the highest income group consumes the most energy and spends the most per household on
it, but energy expenditures represent a smaller percentage of their total expenditures and a much
smaller percentage of their income than the lowest income quintile. Energy expenditures as a
share of total household expenditures were 8.3 percent for the lowest income quintile and 4.9
percent for the highest income quintile. For energy expenditures as a share of average after-tax
income, the distribution is more unequal, ranging from 19.4 percent for the lowest income
quintile to 3.4 percent for the highest income quintile. This means the lowest income households
are spending over five times more of their income on energy than the highest income households.
Table 4-11 Energy Expenditures by Quintiles of Income before Taxes, 2019
Lowest
Second
Third
Fourth
Highest
Metric
All
20%
20%
20%
20%
20%
Income after taxes
71,487
12,236
32,945
53,123
83,864
174,777
Annual expenditures
63,036
28,672
40,472
53,045
71,173
121,571
Natural gas
416
259
355
367
455
644
Electricity
1,472
1,049
1,351
1,446
1,587
1,924
Fuel oil and other fuels
113
69
101
86
121
189
Gasoline, other fuels, and
2,094
998
1,601
2,079
2,593
3,193
motor oil (transportation)
Energy expenditures
4,095
2,375
3,408
3,978
4,756
5,950
Energy expenditures as share of
total expenditures
6.5%
8.3%
8.4%
7.5%
6.7%
4.9%
Energy expenditures as share of
income
5.7%
19.4%
10.3%
7.5%
5.7%
3.4%
Quintile share of all energy
expenditures
11.6%
16.7%
19.4%
23.2%
29.1%
Source: Consumer Expenditure Survey, U.S. Bureau of Labor Statistics, September 2020.
https://www.bls.gOv/cex/tables/calendar-year/mean-item-share-average-standard-error.htm#cu-income. Accessed
5/27/2021.
Note: Income includes wages, self-employment income, Social Security and retirement payments, interest,
dividends, rental income and other property income, public assistance, unemployment and workers' compensation,
veterans' benefits, and regular contributions for support.
The EPA also examined the household energy expenditure data by race and ethnicity.
The data indicate that Black households' energy expenditures represent a higher share of their
total expenditures and income than for households of other races, yet their energy expenditures
134
-------
were lower. Hispanic households' energy expenditures comprise a larger share of their total
expenditures and income than non-Hispanic households, though they spent slightly more per
household on energy than non-Hispanic households.
The CES data summarized in this section highlight the disproportionately high energy
burdens experienced particularly by low-income households, as well as Black and Hispanic
households to some extent. These households must allocate a greater share of their incomes and
expenditures to energy, reducing disposable income that could be used for other essentials (e.g.,
housing, healthcare, and food) and other non-essential preferences. Thus, low income, Black, and
Hispanic households are expected to be most likely to be adversely affected by any potential
increases in energy costs due to this proposed rule because they face higher energy burdens
under the baseline. Nonetheless, since energy cost impacts are expected to be minimal, this rule
is not expected to significantly alter existing levels of inequality in energy burden.
4.2.7 Summary
EJ concerns for each rulemaking are unique and should be considered on a case-by-case
basis. For the proposal, we quantitatively and qualitatively evaluated baseline scenarios for
several potential EJ concerns, although data availability limitations and the large number of oil
and natural gas locations make it quite possible that disparities may exist that our analysis did not
identify. This is especially relevant for potential EJ characteristics that were not evaluated, such
as lower educational attainment. It is also possible that the proposed rulemaking shifts the
distribution of impacts, but our analysis did not assess policy-specific impacts.
Some commonalities emerged across the array of EJ analyses. Notably, more Hispanic
people may reside in communities with potentially elevated cancer risk from oil and natural gas-
related toxic emissions (Section 4.2.3). Similarly, Hispanic populations may experience
disproportional exposures to air pollutants from the oil and natural gas industry (Sections 4.2.3
and 4.2.4) and may be more likely to reside in communities of higher oil and natural gas
intensity (Section 4.2.5). Additionally, Hispanic households' energy expenditures may comprise
a disproportionate share of their total expenditures and income as compared to non-Hispanic
households (Section 4.2.6). However, uncertainties associated with the input data, as well as the
meaningfulness of any differences, should be taken into consideration when interpreting these
results. Additionally, we lack key information that would be needed to characterize post-control
135
-------
risks under the proposed NSPS OOOOb and EG 0000c or the regulatory alternatives analyzed
in the RIA, preventing the EPA from analyzing spatially differentiated outcomes. While a
definitive assessment of the impacts of this proposed rule on minority populations, low-income
populations, and/or Indigenous peoples was not performed, the EPA believes that this action will
achieve substantial methane, VOC, and HAP emissions reductions and will further improve
environmental justice community health and welfare.
4.3 Initial Regulatory Flexibility Analysis
The Regulatory Flexibility Act (RFA; 5 U.S.C.§ 601 et seq.), as amended by the Small
Business Regulatory Enforcement Fairness Act (Public Law No. 104-121), provides that
whenever an agency is required to publish a general notice of proposed rulemaking, it must
prepare and make available an initial regulatory flexibility analysis (IRFA), unless it certifies that
the proposed rule, if promulgated, will not have a significant economic impact on a substantial
number of small entities (5 U.S.C. § 605[b]). Small entities include small businesses, small
organizations, and small governmental jurisdictions. An IRFA describes the economic impact of
the proposed rule on small entities and any significant alternatives to the proposed rule that
would accomplish the objectives of the rule while minimizing significant economic impacts on
small entities. Pursuant to section 603 of the RFA, the EPA prepared an IRFA that examines the
impact of the proposed rule on small entities along with regulatory alternatives that could
minimize that impact. The scope of the IRFA is limited to the NSPS OOOOb. The impacts of the
EG 0000c are not evaluated here because the EG 0000c does not place explicit requirements
on the regulated industry. Those impacts will be evaluated pursuant to the the development of a
Federal plan.
4.3.1 Reasons Why Action is Being Considered
The proposed rulemaking takes a significant step forward in mitigating climate change
and improving human health by reducing GHG and VOC emissions from the oil and natural gas
industry, specifically the Crude Oil and Natural Gas source category. The oil and natural gas
industry is the United States' largest industrial emitter of methane. Human emissions of methane,
a potent GHG, are responsible for about one third of the warming due to well-mixed GHGs, the
second most important human warming agent after carbon dioxide. According to the
136
-------
Intergovernmental Panel on Climate Change (IPCC), strong, rapid, and sustained methane
reductions are critical to reducing near-term disruption of the climate system and a vital
complement to carbon dioxide (CO2) reductions critical in limiting the long-term extent of
climate change and its destructive impacts. The oil and natural gas industry also emits other
health-harming pollutants in varying concentrations and amounts, including CO2, VOC, sulfur
dioxide (SO2), nitrogen oxide (NOx), hydrogen sulfide (H2S), carbon disulfide (CS2), and
carbonyl sulfide (COs), as well as, benzene, toluene, ethylbenzene and xylenes (this group is
commonly referred to as "BTEX"), and n-hexane.
The EPA is proposing the actions described in the preamble in accordance with its legal
obligations and authorities following a review directed by EO 13990, "Protecting Public Health
and the Environment and Restoring Science to Tackle the Climate Crisis," issued on January 20,
2021. The EPA intends for the proposed actions to address the far-reaching harmful
consequences and real economic costs of climate change. According to the IPCC, "It is
unequivocal that human influence has warmed the atmosphere, ocean and land. Widespread and
rapid changes in the atmosphere, ocean, cryosphere and biosphere have occurred." These
changes have led to increases in heat waves and wildfire weather, reductions in air quality, more
intense hurricanes and rainfall events, and rising sea level. These changes, along with future
projected changes, endanger the physical survival, health, economic well-being, and quality of
life of people living in America, especially those in the most vulnerable communities.
In the proposed action, the EPA has taken a comprehensive analysis of the most
attainable data from emission sources in the Crude Oil and Natural Gas source category and the
latest available information on control measures and techniques to identify achievable, cost-
effective measures to significantly reduce emissions, consistent with the requirements of section
111 of the CAA. If finalized and implemented, the proposed actions would lead to significant
and cost-effective reductions in climate and health-harming pollution and encourage
development and deployment of innovative technologies to further reduce this pollution in the
Crude Oil and Natural Gas source category.
4.3.2 Statement of Objectives and Legal Basis for Proposed Rules
The EPA proposes to revise certain NSPS and to promulgate additional NSPS for both
methane and VOC emissions from new oil and gas sources in the production, processing,
137
-------
transmission and storage segments of the industry; and to promulgate EG to require states to
regulate methane emissions from existing sources in those segments. The large amount of
methane emissions from the oil and natural gas industry — by far, the largest methane-emitting
industry in the nation — coupled with the adverse effects of methane on the global climate
compel immediate regulatory action.
The proposal is in line with our 2016 NSPS OOOOa Rule, which likewise regulated
methane and VOCs from all three segments of the industry. The 2016 NSPS OOOOa Rule
explained that these three segments should be regulated as part of the same source category
because they are an interrelated sequence of functions in which pollution is produced from the
same types of sources that can be controlled by the same techniques and technologies. That Rule
further explained that the large amount of methane emissions, coupled with the adverse effects of
GHG air pollution, met the applicable statutory standard for regulating methane emissions from
new sources through NSPS. Furthermore, the Rule explained, this regulation of methane
emissions from new sources triggered the EPA's authority and obligation to regulate the
overwhelming majority of oil and gas sources, which the CAA categorizes as "existing" sources.
In the 2020 Policy Rule, the Agency reversed course, concluding based upon new legal
interpretations that it was not authorized to regulate the transmission and storage segment or to
regulate methane. In 2021, Congress adopted a joint resolution to disapprove the EPA's 2020
Policy rule under the CRA. According to the terms of CRA, the 2020 rule is "treated as though
[it] had never taken effect," 5 U.S.C. 801(f), and as a result, the 2016 rule is reinstated.
In disapproving the 2020 Policy Rule under the CRA, Congress explicitly rejected the
2020 Policy Rule interpretations and embraced the EPA's rationales for the 2016 NSPS OOOOa
Rule. The House Committee on Energy & Commerce emphasized in its report (House Report)
that the source category "is the largest industrial emitter of methane in the U.S.," and directed
that "regulation of emissions from new and existing oil and gas sources, including those located
in the production, processing, and transmission and storage segments, is necessary to protect
human health and welfare, including through combatting climate change, and to promote
environmental justice." House Report at 3-5. A statement from the Senate cosponsors likewise
underscored that "methane is a leading contributing cause of climate change," whose "emissions
come from all segments of the Oil and Gas Industry," and stated that "we encourage EPA to
strengthen the standards we reinstate and aggressively regulate methane and other pollution
138
-------
emissions from new, modified, and existing sources throughout the production, processing,
transmission and storage segments of the Oil and Gas Industry under section 111 of the CAA."
Senate Statement at S2283. The Senators concluded with a stark statement: "The welfare of our
planet and of our communities depends on it." Id.
The proposed rule comports with the EPA's CAA section 111 obligation to reduce
dangerous pollution and responds to the urgency expressed by the current Congress. With the
proposal, the EPA is taking additional steps in the regulation of the Crude Oil and Natural Gas
source category to protect human health and the environment. Specifically, the agency is
proposing to revise certain of those NSPS, to add NSPS for additional sources, and to propose
EG that, if finalized, would impose a requirement on states to regulate methane emissions from
existing sources. As the EPA explained in the 2016 rule, this source category collectively emits
massive quantities of the methane emissions that are among those driving the grave and growing
threat of climate change, particularly in the near term.96 Since that time, the science has
repeatedly confirmed that climate change is already causing dire health, environmental, and
economic impacts in communities across the United States.
Because the 2021 CRA resolution automatically reinstated the 2016 rule, which itself
determined that the Crude Oil and Natural Gas Source Category included the transmission and
storage segment and that regulation of methane emissions was justified, the EPA is authorized to
take the regulatory actions proposed in the rule. In addition, in this action, we are reaffirming
those determinations as clearly authorized under any reasonable interpretation of section 111.
Further information can be found in Section VIII of the preamble.
4.3.3 Description and Estimate of Affected Small Entities
The Regulatory Flexibility Act (RFA) defines small entities as including "small
businesses," "small governments," and "small organizations" (5 USC 601). The regulatory
revisions being considered by EPA for this rulemaking are expected to affect a variety of small
businesses but would not affect any small governments or small organizations. The RFA
references the definition of "small business" found in the Small Business Act, which authorizes
the Small Business Administration (SBA) to further define "small business" by regulation. The
96 81 FR 3584
139
-------
detailed listing of SBA definitions of small business for oil and natural gas industries or sectors,
by NAICS code, that are potentially affected by this proposal is included in Table 4-12. The EPA
conducted this initial regulatory flexibility analysis at the ultimate (i.e., highest) level of
ownership, evaluating ultimate parent entities.
Table 4-12 SBA Size Standards by NAICS Code
NAICS Size Standards Size Standards
Codes NAICS Industry Description (in millions of dollars) (in no. of employees)
211120 Crude Petroleum Extraction - 1,250
211130 Natural Gas Extraction - 1,250
213111 Drilling Oil and Gas Wells - 1,000
213112 Support Activities for Oil and Gas Operations $41.5
486210 Pipeline Transportation of Natural Gas $36.5
Sources: U.S. Small Business Administration, Table of Standards, Effective July 14, 2022.
https://www.sba.gov/document/support--table-size-standards. Accessed July 27, 2022.
To estimate the number of small businesses potentially impacted by the rule, EPA
developed a list of operators of oil and natural gas wells, natural gas processing plants, and
natural gas compressor stations. The list of well operators is based on data from Enverus and
consists of all operators that completed wells producing oil or natural gas in 2019, which serves
as an approximation of the universe of operators that might be affected by the proposed NSPS.
The list of processing plant operators is from the Department of Homeland Security (DHS)
Homeland Infrastructure Foundation-Level Data.97 The compressor stations operator data is from
the Enverus Midstream database. The DHS data and Enverus Midstream data did not contain
information on when facilities were constructed, and therefore could not be restricted to only
those facilities completed in 2019. The initial list of operators included 1,451 well site operators
that completed a well in 2019, 297 processing plant operators, and 574 compressor station
operators.
The list of operators was combined with data from the D&B Hoovers and Zoomlnfo
business databases in a two-step process. D&B Hoovers and Zoomlnfo are proprietary,
subscription-based databases of business information (such as revenue, employment, and
ownership structure) gleaned from sources such as financial statements, news reports, and
industry trade group publications. Using an approximate string-matching algorithm, the list of
97 Department of Homeland Security. (2020). Homeland Infrastructure Foundation-Level Data. Found at:
https://hifld-geoplatform.opendata.arcgis.com/datasets/geoplatform: :natural-gas-processing-plants/about
140
-------
operators was first merged with business information from D&B Hoovers. The remaining
unmatched operators were matched to the Zoomlnfo business database when possible. This
matching process added information on the ultimate parent companies, NAICS codes, number of
employees, and annual revenues of the operators. The matches from D&B Hoovers and
Zoomlnfo were examined and, when necessary, manual adjustments were made to the matched
list of ultimate parent companies to standardize company names, revenue, and employment
information across the two matched lists. Each matched ultimate parent company, or firm, was
classified "small business" or "not small business" based on the SBA size classification
threshold associated with the relevant NAICS code. The results of this small business coding
exercise are displayed by NAICS code in Table 4-13. In total, 998 of the 1,451 well site
operators (69%) matched to 914 ultimate parent companies; 270 of 297 processing plant
operators (91%) matched to 149 ultimate parent companies; and 519 of 574 compressor station
operators (90%) matched to 315 ultimate parent companies.
Table 4-13 Counts and Estimated Percentages of Small Entities
Estimated
Percentage of
Estimated
Small Entities
NAICS
Number of
Number of
for Identified
Codes
NAICS Industry Description
Firms Identified
Small Entities
Firms
211120
Crude Petroleum Extraction
352
319
91%
211130
Natural Gas Extraction
19
17
89%
213111
Drilling Oil and Gas Wells
48
45
94%
213112
Support Activities for Oil and Gas
357
317
89%
Operations
486210
Pipeline Transportation of Natural Gas
31
13
42%
Many3
Other
419
297
71%
a Not all owner/operators in the Enverus well database produced a match in the D&B Hoovers database under an oil
and natural gas industry-related NAICS as presented in Table 4-12.
4.3.4 Compliance Cost Impact Estimates
To estimate the compliance cost impacts of the proposed rule on small entities, we use
the dataset of operators matched to ultimate parent companies discussed in the previous section
and apply the sum of incremental costs for all relevant affected facility categories. Because the
incremental costs depend on unknown characteristics of operator-specific well sites, processing
plants, and compressor stations, we use average equipment counts at each facility type to derive
estimates of average impacts at each facility type. Ultimately, we estimate cost-to-sales ratios
(CSR) for each small entity to summarize the impacts of the proposed NSPS.
141
-------
4.3.4.1 Methodology for Estimating Impacts on Small Entities
The two main pieces of information we use to assess impacts on small entities are
ultimate parent company revenues and expected compliance costs. For most ultimate parent
companies in the dataset described in the previous section, revenue is generated from the match
with either the D&B Hoovers or Zoomlnfo database. For owners of well site operators, we also
estimated revenues from calculating total operator-level production in 2019 from Enverus,
multiplying by assumed oil and natural gas prices at the wellhead and summing over all
operators owned by a parent company. For natural gas prices, we assumed the projected price
from AEO2022 in 2022 (adjusted to approximate a wellhead price, as described in Section 2.4)
of $3.40/Mcf. For oil prices, we used the projected AEO2022 price for Brent Crude in 2022,
$66.40/barrel. Both prices are measured in 2019$. For owners of well site operators, revenue was
calculated as the minimum of the matched revenue from D&B Hoovers/Zoomlnfo and the
estimated revenue based on production. Operators of compressor stations were divided into two
groups: those that own gathering and boosting stations, and those that own transmission and
storage stations. While there is overlap between the two segments, they are treated as distinct
groups in this analysis and results are presented by segment. Summary statistics for firm revenue
by segment are presented in Table 4-14.
Table 4-14 Summary Statistics for Revenues of Potentially Affected Entities
Segment
Production
Processing
Gathering and Boosting
Transmission and Storage
Size
No. of Firms
Mean Revenue
(million 2019$)
Median Revenue
(million 2019$)
Small
836
$230
$11
Not Small
78
$19,000
$1,300
Small
88
$180
$11
Not Small
61
$27,000
$6,100
Small
123
$510
$24
Not Small
77
$19,000
$3,200
Small
50
$260
$22
Not Small
82
$22,000
$3,200
To calculate expected compliance costs for ultimate parent companies, we first
constructed an estimate of the number of sites for each firm in each segment. For well site
142
-------
operators, the number of well sites is calculated by summing the oil and natural gas wells for an
operator completed in 2019 and dividing by the average number of wells per site in the Enverus
data (-1.33 wells per wellsite). The number of well sites owned by an ultimate parent company
is calculated by summing over the well site counts of the operators it owns. For processing
plants, gathering and boosting compressor stations, and transmission and storage compressor
stations, the number of sites is obtained by summing the number of entries of each type in the
DHS data for processing plant operators and in the Enverus Midstream data for compressor
station operators. Again, the number of facilities of each type owned by an ultimate parent
company is calculated by summing over the facility counts of the processing plant or compressor
station operators it owns. To approximate the impact of state requirements on facility level costs
relative to baseline, Colorado, California, New Mexico, and Pennsylvania facilities were
removed from the well site counts, and Colorado and California were removed from the
processing plant and compressor station counts, since these facilities are assumed to have
requirements in the baseline that are at least as stringent as the proposed rule. See Section 2.2.3
for more information about the inclusion of state programs in the baseline.
Once site counts were assigned, we estimated compliance costs for each ultimate parent
company by assigning annualized costs (both with and without expected revenue from natural
gas recovery) from all relevant affected facilities: fugitive emissions, pneumatic pumps,
pneumatic controllers, storage vessels, and liquids unloading for well sites; equipment leaks,
reciprocating compressors, and dry seal centrifugal compressors for processing plants; and
reciprocating compressors, dry/wet seal centrifugal compressors and pneumatic
pumps/controllers for gathering and boosting and transmission and storage compressor stations.
Since the precise equipment counts at the facility level were necessary to estimate compliance
costs relative to baseline, and this information was not present in the operator data, average
equipment counts per facility were used to estimate site-level compliance costs for this analysis.
Median compliance costs by segment and firm size are presented in Table 4-15, both with and
without expected revenue from natural gas recovery included.
143
-------
Table 4-15 Distribution of Estimated Compliance Costs across Segment and Firm Size
Classes (2019$)
No. of
Median Cost without
Median Cost with
Segment
Size
Firms
Product Recovery
Product Recovery
Production
Small
Not Small
836
78
$40,000
$180,000
$39,000
$170,000
Processing
Small
Not Small
88
61
-$4,800
-$9,700
-$9,100
-$18,000
Gathering and
Small
123
$4,400
-$110
Boosting
Not Small
77
$11,000
-$280
Transmission and
Small
50
$1,900
$1,300
Storage
Not Small
82
$7,600
$5,200
Note: Totals may not appear to add correctly due to rounding.
4.3.4.2 Results
This section presents results of the cost-to-sales ratio analysis for the production,
processing, gathering and boosting, and transmission and storage segments. The cost-to-sales
ratios presented approximate the impact of the NSPS requirements on ultimate parent companies
of well site, processing plant, and compressor station operators. In the processing segment,
average annualized costs relative to baseline are expected to be negative, and no entity has a CSR
greater than either 1 percent or 3 percent98. In the production segment, when expected revenues
from natural gas product recovery are included, 206 small entities (25 percent) have cost-to-sales
ratios greater than 1 percent, and of those, 79 have cost-to-sales ratios greater than 3 percent (9
percent). When expected revenues from natural gas product recovery are excluded, the number
of small entities with cost-to-sales ratios greater than 1 percent increases to 220 (26 percent); 79
of those small entities (9 percent) also have cost-to-sales ratios greater than 3 percent. In the
gathering and boosting segment, no parent companies have cost-to-sales ratios greater than 3%
regardless of whether expected revenues from natural gas recovery are included. 1 parent
companies (1%) has a cost-to-sales ratio greater than 1 percent when expected revenues from
natural gas recovery are excluded (none do when they are included). In the transmission and
storage segment, no entity has a CSR greater than either 1 percent or 3 percent regardless of
98 The net compliance costs for leak detection at natural gas processing plants decrease primarily because OGI
surveys under this proposal can be conducted much more quickly and at approximately half the cost of EPA Method
21 surveys under the current requirements in NSPS Wa, so the increased flexibility under the proposal is likely cost
saving for affected facilities.
144
-------
whether expected revenues from product recovery are included. The results for all segments are
summarized in Table 4-16.
Table 4-16 Compliance Cost-to-Sales Ratios for Small Entities
Without Product Recovery With Product Recovery
Included Included
Segment
CSR Ratio
Category
No. of Small
Entities
% of Small
Entities
No. of Small
Entities
% of Small
Entities
All
836
836
Production
Greater than 1%
220
26%
206
25%
Greater than 3%
79
9%
79
9%
All
88
88
Processing
Greater than 1%
0
0%
0
0%
Greater than 3%
0
0%
0
0%
Gathering and
Boosting
All
Greater than 1%
123
1
1%
123
0
0%
Greater than 3%
0
0%
0
0%
Transmission and
Storage
All
Greater than 1%
50
0
0%
50
0
0%
Greater than 3%
0
0%
0
0%
4.3.5 Caveats and Limitations
The analysis above is subject to several caveats and limitations, many of which we
discussed in the presentation of methods and results. It is useful, however, to present a compiled
list of the caveats and limitation here.
• Not all owner/operators could be identified in either the D&B Hoovers or Zoomlnfo
database. In addition, the matching procedure used to link the operator database to the
D&B Hoovers and Zoomlnfo database is imperfect, so there may be misspecified
matches or duplicate entries for the same entity.
• The analysis assumes the same population of entities completing wells in 2019 are also
completing wells at the same rate in 2023 and beyond, and assumes facility counts are
stable over time. These firms may operate more or fewer facilities in the future depending
on economic and technological factors that are largely unpredictable. The analysis also
assumes the population of entities operating processing plants and compressor stations in
145
-------
2019 is the same population that will construct new processing plant and compressor
stations in 2023 and beyond.
• The approach used to estimate sales for the cost-to-sales ratios might over-estimate or
under-estimate sales depending upon the accuracy of the information in the underlying
databases and the market prices ultimately faced when the proposed requirements are in
effect.
• It is unknown what equipment is present at each site. The use of equipment averages to
estimate costs may under- or over-estimate costs at the site level for any given entity,
which adds uncertainty to the calculated cost-to-sales ratios.
4.3.6 Projected Reporting, Recordkeeping and Other Compliance Requirements
The information to be collected for the proposed NSPS is based on notification,
performance tests, recordkeeping and reporting requirements which will be mandatory for all
operators subject to the final standards. Recordkeeping and reporting requirements are
specifically authorized by section 114 of the CAA (42 U.S.C. 7414). The information will be
used by the delegated authority (state agency, or Regional Administrator if there is no delegated
state agency) to ensure that the standards and other requirements are being achieved. Based on
review of the recorded information at the site and the reported information, the delegated
permitting authority can identify facilities that may not be in compliance and decide which
facilities, records, or processes may need inspection. All information submitted to the EPA
pursuant to the recordkeeping and reporting requirements for which a claim of confidentiality is
made is safeguarded according to Agency policies set forth in 40 CFR part 2, subpart B.
Potential respondents under subpart OOOOb are owners or operators of new, modified,
or reconstructed oil and natural gas affected facilities as defined under the rule. Few, if any, of
the facilities in the United States are owned or operated by state, local, tribal or the Federal
government. The regulated facilities are privately owned for-profit businesses. The requirements
in this action result in industry recording keeping and reporting burden associated with review of
the requirements for all affected entities, gathering relevant information, performing initial
performance tests and repeat performance tests if necessary, writing and submitting the
146
-------
notifications and reports, developing systems for the purpose of processing and maintaining
information, and train personnel to be able to respond to the collection of information.
The estimated average annual burden (averaged over the first 3 years after the effective
date of the standards) for the recordkeeping and reporting requirements in subpart OOOOb for
the estimated 1,844 owners and operators that are subject to the rule is approximately 881,777
labor hours, with an annual average cost of about $58 million. The annual public reporting and
recordkeeping burden for this collection of information is estimated to average about 66 hours
per respondent. Respondents must monitor all specified criteria at each affected facility and
maintain these records for 5 years. Burden is defined at 5 CFR 1320.3(b).
4.3.7 Related Federal Rules
There are two National Emission Standards for Hazardous Air Pollutants (NESHAP)
rules that apply to certain equipment and processes in the oil and natural gas sector. These rules,
listed below, address air toxics, primarily benzene, toluene, ethylbenzene, and xylenes
(collectively referred to as BTEX) and n-hexane. These two rules were promulgated under
section 112 of the Clean Air Act and are codified in 40 CFR Part 63 Subpart HH and Subpart
HHH.
Aside from the EPA, several other Federal agencies have jurisdiction over the oil and
natural gas sector.
• The Bureau of Land Management (BLM) within the Department of the Interior regulates
the extraction of oil and gas from federal lands. BLM manages the Federal government's
onshore subsurface mineral estate, about 700 million acres. BLM also oversees oil and
gas operations on many Tribal leases and maintains an oil and natural gas leasing
program. BLM does not directly regulate emissions for the purposes of air quality but
does regulate venting and flaring of natural gas for the purposes of preventing waste. An
operator may also be required to control/mitigate emissions as a condition of approval on
a drilling permit.
• The Bureau of Ocean Energy Management (BOEM) within the Department of the
Interior manages the development of America's offshore energy and mineral resources.
BOEM has certain air quality regulatory authority over activities that BOEM authorizes
147
-------
on the Outer Continental Shelf of the United States in the Gulf of Mexico, west of 87.5
degrees longitude, and adjacent to the North Slope Bureau of the State of Alaska.
• The Pipeline and Hazardous Materials Safety Administration (PHMSA) within the
Department of Transportation ensures safety in the design, construction, operation,
maintenance, and spill response planning of America's 2.8 million miles of natural gas
and hazardous liquid transportation pipelines. This includes data and risk analysis,
outreach, research and development, regulations and standards, training, inspections and
enforcement and accident investigations. Section 113 of the Protecting our Infrastructure
of Pipelines and Enhancing Safety Act of 2020 (PIPES Act of 2020) mandates that
PHMSA promulgate a final rule concerning gas pipeline leak detection and repair
programs no later than one year after the enactment of the law.
• The Federal Energy Regulatory Commission (FERC) within the Department of Energy
(DOE) regulates natural gas pipeline, storage, and liquefied natural gas facility
construction. FERC also issues environmental assessments or draft and final
environmental impact statement for comment on most projects.
• The Internal Revenue Service (IRS), in the Internal Revenue Code (IRC), defines a
stripper well property as "a property where the average daily production of domestic
crude oil and gas produced from the wells on the property during a calendar year divided
by the number of such wells is 15 barrel equivalents or less." See IRC 613A(c)(6)(E).
4.3.8 Regulatory Flexibility Alternatives
Prior to the November 2021 proposal, the EPA convened a Small Business Advocacy
Review (SBAR) Panel to obtain recommendations from small entity representatives (SERs) on
elements of the regulation. The Panel identified significant alternatives for consideration by the
Administrator of the EPA, which were summarized in a final report." Based on the Panel
recommendations, as well as comments received in response to the November 2021 proposal, the
EPA is proposing, or taking comment on, several regulatory alternatives that could accomplish
99 See document ID EPA-HQ-OAR-2021-0317-0074.
148
-------
the stated objectives of the Clean Air Act while minimizing any significant economic impact of
the proposed rule on small entities. Discussion of those alternatives is provided below.
4.3.8.1 Fugitive Emissions Requirements
As described in the preamble to the supplemental proposal,100 the EPA is proposing
certain changes to the fugitive emissions standards proposed for NSPS OOOOb in the November
2021 proposal. The EPA believes that two of these proposed changes will reduce impacts on
small businesses: (1) requiring OGI monitoring for well sites and centralized production facilities
following the monitoring plan required in proposed 40 CFR 60.5397b instead of requiring the
procedures being proposed in Appendix K for these sites and (2) defining monitoring technique
and frequency based on the equipment present at a well site. The EPA describes these two
proposed changes below.
In the supplemental proposal, the EPA is not requiring OGI monitoring in accordance
with the proposed Appendix K for well sites or centralized production facilities, as was proposed
in the November 2021 proposal. Instead, the EPA is proposing to require OGI surveys following
the procedures specified in the proposed regulatory text for NSPS OOOOb (at 40 CFR 60.5397b)
or according to EPA Method 21. This proposed change is consistent with the requirements for
OGI surveys found in NSPS OOOOa at 40 CFR 60.5397a. This proposed change is a result of
the extensive comments the EPA received from oil and gas operators and other groups on the
numerous complexities associated with following the proposed Appendix K, especially
considering the remoteness and size of many of these well sites.101 In addition, commenters
pointed out that OGI has always been the BSER for fugitive monitoring at well sites and was
never designed as a replacement for EPA Method 21, while Appendix K was designed for use at
more complex processing facilities that have historically been subject to monitoring following
EPA Method 21. The EPA agrees with the commenters and is proposing requirements within
NSPS OOOOb at 40 CFR 60.5397b in lieu of the procedures in Appendix K for fugitive
100 See preamble Section IV. A.
101 See preamble Section IV.A. and see Document ID Nos. EPA-HQ-OAR-2021-0317-0579, EPA-HQ-OAR-2021-
0317-0743, EPA-HQ-OAR-2021-0317-0764, EPA-HQ-OAR-2021-0317-0777, EPA-HQ-OAR-2021-0317-0782,
EPA-HQ-OAR-2021-0317-0786, EPA-HQ-OAR-2021-0317-0793, EPA-HQ-OAR-2021-0317-0802, EPA-HQ-
OAR-2021-0317-0807, EPA-HQ-OAR-2021-0317-0808, EPA-HQ-OAR-2021-0317-0810, EPA-HQ-OAR-2021-
0317-0814, EPA-HQ-OAR-2021-0317-0817, EPA-HQ-OAR-2021-0317-0820, EPA-HQ-OAR-2021-0317-0831,
EPA-HQ-OAR-2021-0317-0834, and EPA-HQ-OAR-2021-0317-0938.
149
-------
emissions monitoring at well sites or centralized production facilities. See section VI of the
preamble for additional information on what the EPA is proposing for Appendix K related to
other sources (e.g., natural gas processing plants). The EPA believes this will particularly benefit
small entities because it will streamline the requirements for conducting and documenting OGI
surveys at these smaller, less complex sites. Additionally, this change provides a uniform set of
requirements for regulated entities that may have assets subject to different subparts within the
same region, which leads to increased regulatory certainty and eases the compliance burden. At
the same time, the EPA believes this does not compromise the stated objectives of the Clean Air
Act because these same requirements are already allowed in NSPS OOOOa and outline many of
the same data elements required by Appendix K.
Next, the supplemental proposal includes fugitive monitoring frequencies and detection
techniques that are based on the type of equipment located at a well site, instead of the baseline
methane emissions threshold that was included in the November 2021 proposal. Specifically, the
EPA is proposing four distinct subcategories of well sites:
• Well sites with only a single wellhead,
• Small well sites with a single wellhead and only one piece of major production and
processing equipment,102
• Well sites with only two or more wellheads and no other major production and
processing equipment, and
• Well sites with one or more controlled storage vessels, control devices, natural gas-
driven pneumatic controllers or pumps, or two or more other major production and
processing equipment, including centralized production facilities.
102 Small well sites are defined as single wellhead well sites that do not contain any controlled storage vessels,
control devices, pneumatic controller affected facilities, or pneumatic pump affected facilities, and include only one
other piece of major production and processing equipment. Major production and processing equipment that would
be allowed at a small well site would include a single separator, glycol dehydrator, centrifugal and reciprocating
compressor, heater/treater, and storage vessel that is not controlled. By this definition, a small well site could only
potentially contain a well affected facility (for well completion operations or gas well liquids unloading operations
that do not utilize a CVS to route emissions to a control device) and a fugitive emissions components affected
facility. No other affected facilities, including those utilizing CVS (such as pneumatic pumps routing to control) can
be present for a well site to meet the definition of a small well site. The EPA is soliciting comment on this definition
for small well sites, including whether additional metrics should be used beyond equipment counts, as well as the
proposed standards and requirements for this subcategory of sites.
150
-------
The EPA is proposing these distinct subcategories of well sites after consideration of
comments on the November 2021 proposal that stated the proposed baseline methane emissions
threshold approach would be difficult to implement, especially for small businesses that may be
less familiar with the use of emissions factors from the EPA's Greenhouse Gas Reporting
Program. The EPA believes that owners and operators, including small entities, can readily
identify the number and types of major equipment located at a well site without the need for
complicated calculations of emissions.
Further, the EPA is proposing specific monitoring frequency and techniques as the BSER
for each well site subcategory individually. For example, the EPA is proposing the use of
sensory monitoring techniques (AVO) at well sites containing only a single wellhead and at
small well sites. This monitoring technique does not require specialized equipment or operator
training, but does allow the identification of large leaks, which are of the most concern from an
environmental standpoint. Further, AVO monitoring can easily be built into regular maintenance
activities that are designed to keep the equipment at the site in good working order. The
proposed requirements are responsive to a SER recommendation that the EPA allow AVO and
soap bubble tests as an option for finding fugitive emissions, particularly because they are low
cost and easy to implement alternatives for detecting leaks, and an Advocacy recommendation
that the EPA propose allowing AVO as an alternative in limited circumstances, such as part of an
off-ramp for facilities unlikely to emit more than insignificant methane or with a demonstrated
history of insignificant emissions. The EPA believes this will particularly benefit small entities
because AVO surveys at these types of well sites are effective at identifying the types of large
emissions from sources located at these well sites at a much lower cost than OGI surveys. For
example, the costs associated with the proposed quarterly AVO inspections are estimated at
$660/year, whereas the costs associated with an annual OGI survey for this type of well site are
estimated at approximately $2,000/yr. AVO inspections allow for more frequent inspections for
large emissions events at these well sites, which results in faster emissions mitigation, than a
single OGI survey each year.
151
-------
4.3.8.2 Alternative Technology
As described in the preamble to the supplemental proposal,103 the EPA is proposing
changes to the November 2021 proposal alternative technology requirements for NSPS OOOOb.
The proposed changes are the result of overwhelming support that the EPA received for the
inclusion of an option to use advanced technologies for periodic screenings as an alternative to
the fugitive emissions monitoring and repair program proposed in NSPS OOOOb. The EPA
believes these proposed changes will reduce impacts on small businesses.
Specifically, the EPA is proposing the use of alternative screening technologies as a
compliance option rather than an additional regulatory requirement. Through the SBAR Panel
outreach, SERs supported the use of aerial, satellite, and other forms of monitoring for fugitive
emissions requirements beyond traditional LDAR, but only as an alternative and not as an
additional requirement. In addition, the Panel recommended that the EPA consider the cost and
scope of alternative technologies and propose alternative screening technology, and that the EPA
try to minimize significant additional reporting and recordkeeping requirements. In accordance
with these recommendations, the EPA is proposing changes that are intended to support the
deployment and utilization of a broader spectrum of advanced measurement technologies and,
ultimately, enable more cost-effective reductions in emissions. These changes include a proposed
"matrix" which would specify several different screening frequencies corresponding to a range
of minimum detection levels, in contrast to the single screening frequency and detection level
permitted under the November 2021 proposal. In addition, the EPA is proposing to allow owners
and operators the option of using continuous monitoring technologies as an alternative to
periodic screening and are proposing long- and short-term emissions rate thresholds that would
trigger corrective action as well as monitoring plan requirements for owners and operators that
choose this approach. The EPA believes this approach will particularly benefit small entities
because they will be allowed flexibility to determine which screening technology works best for
their needs without the need to undertake the application of an alternative means of emissions
limitation (AMEL), which would be especially burdensome for small entities with less ability to
perform extensive field testing of technologies or conduct sophisticated modeling simulations.
103 See preamble section IV.B.
152
-------
Furthermore, this approach incorporates the use of alternative test methods, which allows for
broad application of technologies after approval, without the need for individual applications
from owners and operators for their specific sites.
4.3.8.3 Associated Gas
As described in the preamble to the supplemental proposal,104 the EPA is proposing
certain changes to the requirements for oil wells with associated gas that were proposed in
November 2021 for NSPS OOOOb. These changes include proposing adjustments to the
hierarchy of the standard and compliance options. The EPA believes these proposed changes will
especially reduce impacts on small businesses.
Specifically, the EPA is proposing to require flaring of all associated gas where a
determination has been made that it is not feasible to route the associated gas to a sales line or
use it for another beneficial purpose due to technical or safety reasons. This demonstration would
need to not only address the lack of availability or access to a sales line but would also need to
demonstrate why all potential beneficial uses are not feasible due to technical or safety reasons.
This demonstration, which would require certification by a professional engineer or other
qualified individual, would be submitted in the first annual report for the well affected facility.
The EPA is soliciting comment on what this demonstration should entail and what qualifications
constitute an "other qualified individual" in the preamble for this supplemental proposal. The
EPA believes this approach will benefit small entities because it still allows for the flaring of
associated gas. Installation of a sales pipeline or other infrastructure necessary to use associated
gas in a beneficial way is very costly, especially where well sites are located at great distances
from other necessary infrastructure, such as natural gas processing plants. These costs can
disproportionally affect small businesses who may not produce a large enough quantity of
associated gas to offset the capital necessary to install such infrastructure. The proposed
allowance of flaring in these situations provides for a way to reduce emissions of methane to the
atmosphere (in contrast to direct venting of associated gas), but at a lower cost than the cost for
new infrastructure.
104 See preamble section IV.F.
153
-------
4.3.8.4 Pneumatic Controller and Pneumatic Pump Requirements
As described in the preamble to the supplemental proposal,105 the EPA is proposing
certain clarifications and changes to the pneumatic controller and pneumatic pumps emissions
requirements included in the November 2021 proposal. The EPA is seeking specific solicitations
for comment to understand any information that may dispute the conclusions the EPA has made
with regards to technical feasibility of the proposed zero-emitting standards. The EPA believes
this information will help us further understand the impacts on small businesses.
Through the SBAR Panel outreach, SERs stated that zero emission controllers are not
feasible at wells sites or other locations without reliable electricity and installing gas-fired
compressors to provide sufficient air for instrument air systems may defeat the purpose by
ultimately increasing emissions, and the installation of electric service would be extremely
expensive. EPA and Advocacy recommended that the EPA only propose zero emission standards
for pneumatic controllers at sites with reliable and consistent onsite power available and clearly
state that the intent is not to require the installation of electric services for this purpose.106
For pneumatic controllers, the EPA maintains that there is a technically feasible option
available for zero-emitting controllers for all production, processing, and transmission and
storage sites, except for sites in Alaska without access to electricity. The EPA further identifies
compliance options for pneumatic controllers other than using electricity. Therefore, the
proposed NSPS OOOOb does not include any alternative non-zero emission standards for
pneumatic controllers.
For pneumatic pumps, the proposed rule recognizes that at sites without access to
electricity, there could be situations where it is technically infeasible to use a pump that is not
driven by natural gas. As a result, the EPA is proposing to include a tiered structure in the rule
that would allow flexibility based on site-specific conditions. At sites without access to
electricity, if a demonstration is made that it is technically infeasible to use a pneumatic pump
that is not driven by natural gas, the rule would allow the use of a natural gas-driven pump,
provided that the emissions are captured and routed to a process, which EPA understands to
105 See preamble Sections IV.D and IV.E.
154
-------
achieve 100 percent reduction of methane and VOC. Such an infeasibility determination is not
allowed if the site has access to electricity. This means the proposed rule would prohibit the use
of natural gas-driven pumps at sites with access to electricity.
The EPA is requesting information that may dispute the conclusion that there is a
technically feasible option that does not emit methane or VOC available for all sites in all
segments for pneumatic controllers and pneumatic pumps. Some commenters raised concerns
about specific situations that may make individual technologies impracticable to implement (e.g.,
the inability of solar-powered controller systems to meet the needs at certain remote locations
that do not have access to electricity). Although the EPA will consider any additional
information commenters may submit about such situations, the EPA notes that there are multiple
options for meeting the proposed zero-emission standard and that limitations on the use of one
technology at any given site does not mean that other options for meeting the standard are
unavailable. As a result, the EPA is particularly interested in understanding whether there are site
characteristics that would make every zero-emitting option (e.g., electric controllers powered by
the grid or by solar power; instrument air systems powered by the grid, a generator, or by solar
power; collecting the emissions and routing them to a process; self-contained controllers, etc.)
technically infeasible at the site.
The proposed requirements and solicitations for comment are responsive to SER's
statements and concerns about technical feasibility. The EPA believes the solicitations for
comment will help continue the dialogue with small entity stakeholders to help the EPA more
fully understand the impacts and feasibility challenges on small businesses.
4.3.8.5 Reciprocating Compressors
As described in the preamble to the supplemental proposal,107 the EPA is proposing
certain changes to the proposed requirements for reciprocating compressors in the November
2021 proposal for NSPS OOOOb. The EPA believes these proposed changes will reduce impacts
on small businesses.
Concerns were expressed regarding the EPA's November 2021 proposal that shifted rod
packing changeout requirements from a designated schedule of once every 3 years to a
107 See preamble Section IV.I.
155
-------
performance standard based on an annual flow measurement. It was further noted that this type
of performance standard is often more expensive than a fixed equipment change out standard
because of the additional monitoring and recordkeeping necessary to demonstrate compliance
with the performance standard, which could negatively impact small businesses.
The EPA is proposing changes and specific clarifications to the requirements associated
with reciprocating compressor rod packing. Specifically, we are proposing: (1) to clarify that the
standard of performance is a numeric standard (not a work practice standard) of 2 scfm, (2) to
allow for repair (in addition to replacement) of the rod packing in order to maintain an emission
rate at or below 2 scfm; (3) to allow for monitoring based on 8,760 hours of operation instead of
based on a calendar year. We are also proposing regulatory text that clearly defines the required
flow rate measurement methods and/or procedures, repair and replacement requirements, and
recordkeeping and reporting requirements. For the alternative option of routing rod packing
emissions to a process via a CVS under negative pressure, we are proposing to remove the
negative pressure requirement. These changes take into account comments received on the
November 2021 proposal.
The EPA believes this approach will particularly benefit small entities because facilities
can use monitoring data to determine emission levels at which it is necessary to repair or replace
rod packing. This approach can result in operational benefits, including a longer life for existing
equipment, improvements in operating efficiencies, and long-term cost savings. The proposed
change for monitoring based on 8,760 hours of operation will ensure that undue burden is not
placed on owners and operators where compressors are not operational for multiple months or
are used intermittently and this will allow owners and operators the flexibility to stagger
maintenance activity throughout the year.
4.4 Employment Impacts of Environmental Regulation
This section presents an overview of the various ways that environmental regulation can
affect employment.108 Employment impacts of environmental regulations are generally composed
of a mix of potential declines and gains in different areas of the economy over time. Regulatory
108 Additionally, see Section 4.2.5 for a discussion of the demographic characteristics of oil and natural gas workers
and communities.
156
-------
employment impacts can vary across occupations, regions, and industries; by labor and product
demand and supply elasticities; and in response to other labor market conditions. Isolating such
impacts is a challenge, as they are difficult to disentangle from employment impacts caused by a
wide variety of ongoing, concurrent economic changes. The EPA continues to explore the
relevant theoretical and empirical literature and to seek public comments in order to ensure that
the way the EPA characterizes the employment effects of its regulations is reasonable and
informative.
Environmental regulation "typically affects the distribution of employment among
industries rather than the general employment level" (Arrow et al., 1996). Even if impacts are
small after long-run market adjustments to full employment, many regulatory actions have
transitional effects in the short run (OMB, 2015). These movements of workers in and out of jobs
in response to environmental regulation are potentially important and of interest to policymakers.
Transitional job losses have consequences for workers that operate in declining industries or
occupations, have limited capacity to migrate, or live in communities or regions with high
unemployment rates.
As indicated by the potential impacts on oil and natural gas markets discussed in Section
4.1, the proposed NSPS OOOOb and EG 0000c are projected to cause small changes in oil and
natural gas production and prices. As a result, demand for labor employed in oil and natural gas-
related activities and associated industries might experience adjustments as there may be
increases in compliance-related labor requirements as well as changes in employment due to
quantity effects in directly regulated sectors and sectors that consume oil and natural gas
products.
157
-------
5 COMPARISON OF BENEFITS AND COSTS
5.1 Comparison of Benefits and Costs
A comparison of quantified benefits and costs is presented below. All estimates are in
2019 dollars. Also, all compliance costs, emissions changes, and benefits are estimated for the
years 2023 to 2035 relative to a baseline without the proposed NSPS OOOOb and EG OOOOc.
Table 5-1 summarizes the emissions reductions associated with the proposed standards
over the 2023 to 2035 period for the NSPS OOOOb, the EG OOOOc, and the NSPS OOOOb
and EG OOOOc combined. Table 5-2, Table 5-3, and Table 5-4 present the present value (PV)
and equivalent annual value (EAV), estimated using discount rates of 3 and 7 percent, of the
changes in quantified benefits, costs, and net benefits, as well as the emissions reductions
relative to the baseline for the proposed NSPS OOOOb, for the proposed EG OOOOc, and the
proposed NSPS OOOOb and EG OOOOc, respectively. These values reflect an analytical time
horizon of 2023 to 2035, are discounted to 2021, and presented in 2019 dollars. These tables
include consideration of the non-monetized benefits associated with the emissions reductions
projected under this proposal.
Table 5-1 Projected Emissions Reductions under the Proposed NSPS OOOOb and EG
OOOOc across Regulatory Options, 2023-2035
Emissions Changes
Regulatory Proposed
Option Requirements
Methane
(millions short
tons)
VOC (millions
short tons)
HAP (millions
short tons)
Methane
(million metric
tons CO2 Eq.
using GWP=25)
Less Stringent Option
NSPS OOOOb
1.8
1.3
0.05
42
EG OOOOc
11
2.3
0.1
250
Total
13
3.6
0.2
290
Proposed Option
NSPS OOOOb
8.1
2.9
0.11
180
EG OOOOc
28
6.8
0.28
620
Total
36
10
0.39
810
More Stringent Option
NSPS OOOOb
9.7
3.5
0.13
220
EG OOOOc
28
7
0.28
630
Total
37
10
0.41
850
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
158
-------
Table 5-2 Projected Benefits, Compliance Costs, and Emissions Reductions across
Regulatory Options under the Proposed NSPS OOOOb, 2023-2035 (million 2019$)
3 Percent Discount Rate
PV
EAV
PV
EAV
Climate Benefits3
Less Stringent
$2,000
$190
$2,000
$190
Proposal
$11,000
$1,000
$11,000
$1,000
More Stringent
$11,000
$1,000
$11,000
$1,000
3 Percent Discount Rate
7 Percent Discount Rate
PV
EAV
PV
EAV
Net Compliance Costs
Less Stringent
$3,200
$300
$2,500
$280
Proposal
$3,300
$360
$3,000
$360
More Stringent
$3,800
$360
$3,000
$360
Compliance Costs
Less Stringent
$3,400
$320
$2,500
$300
Proposal
$4,400
$460
$3,700
$440
More Stringent
$4,900
$460
$3,700
$450
Value of Product Recovery
Less Stringent
$170
$16
$14
$14
Proposal
$1,000
$99
$730
$88
More Stringent
$1,100
$99
$730
$88
Net Benefits
Less Stringent
-$1,200
-$110
-$470
-$95
Proposal
$7,600
$670
$7,900
$670
More Stringent
$7,100
$670
$7,900
$670
Non-Monetized Benefits
Climate and ozone health benefits from reducing methane emissions by (in short tons):
Less Stringent 1,800,000
Proposal 8,100,000
More Stringent 9,700,000
PM2 5 and ozone health benefits from reducing VOC emissions by (in short tons)b:
Less Stringent 1,300,000
Proposal 2,900,000
More Stringent 3,500,000
HAP benefits from reducing HAP emissions by (in short tons):
Less Stringent 49,000
Proposal 110,000
More Stringent 130,000
Visibility benefits
Reduced vegetation effects
Notes: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
a Climate benefits are based on reductions in methane emissions and are calculated using four different estimates of
the social cost of methane (SC-CH4) (model average at 2.5 percent, 3 percent, and 5 percent discount rates; 95th
159
-------
percentile at 3 percent discount rate). For the presentational purposes of this table, we show the benefits associated
with the average SC-CH4 at a 3 percent discount rate, but the Agency does not have a single central SC-CH4 point
estimate. We emphasize the importance and value of considering the benefits calculated using all four SC-CH4
estimates; see Table 3-9 for the full range of SC-CH4 estimates. As discussed in Section 3 of the RIA, a
consideration of climate benefits calculated using discount rates below 3 percent, including 2 percent and lower, are
also warranted when discounting intergenerational impacts. Appendix B presents the results of a sensitivity analysis
using a set of SC-CH4 estimates that incorporates recent research addressing recommendations of the National
Academies of Sciences, Engineering, and Medicine (2017). All net benefits are calculated using climate benefits
discounted at 3 percent.
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C.
Table 5-3 Projected Benefits, Compliance Costs, and Emissions Reductions across
Regulatory Options under the Proposed EG OOOOc, 2023-2035 (million 2019$)
3 Percent Discount Rate
PV
EAV
PV
EAV
Climate Benefits3
Less Stringent
$15,000
$1,400
$15,000
$1,400
Proposal
$37,000
$3,500
$37,000
$3,500
More Stringent
$37,000
$3,500
$37,000
$3,500
3 Percent Discount Rate
7 Percent Discount Rate
PV
EAV
PV
EAV
Net Compliance Costs
Less Stringent
$4,900
$460
$3,600
$430
Proposal
$11,000
$990
$8,700
$1,000
More Stringent
$11,000
$1,000
$9,000
$1,100
Compliance Costs
Less Stringent
$6,300
$600
$4,600
$550
Proposal
$14,000
$1,300
$11,000
$1,300
More Stringent
$15,000
$1,400
$12,000
$1,400
Value of Product Recovery
Less Stringent
$1,400
$130
$990
$120
Proposal
$3,600
$340
$2,500
$300
More Stringent
$3,600
$340
$2,600
$310
Net Benefits
Less Stringent
$9,900
$930
$11,000
$960
Proposal
$26,000
$2,500
$28,000
$2,400
More Stringent
$26,000
$2,500
$28,000
$2,400
Non-Monetized Benefits
Climate and ozone health benefits from reducing methane emissions by (in short tons):
Less Stringent 11,000,000
Proposal 28,000,000
More Stringent 28,000,000
PM2.5 and ozone health benefits from reducing VOC emissions by (in short tons)b c:
Less Stringent 2,300,000
Proposal 6,800,000
More Stringent 6,900,000
160
-------
HAP benefits from reducing HAP emissions by (in short tons)
Less Stringent
Proposal
More Stringent
Visibility benefits
Reduced vegetation effects
Notes: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
a Climate benefits are based on reductions in methane emissions and are calculated using four different estimates of
the social cost of methane (SC-CH4) (model average at 2.5 percent, 3 percent, and 5 percent discount rates; 95th
percentile at 3 percent discount rate). For the presentational purposes of this table, we show the benefits associated
with the average SC-CH4 at a 3 percent discount rate, but the Agency does not have a single central SC-CH4 point
estimate. We emphasize the importance and value of considering the benefits calculated using all four SC-CH4
estimates; see Table 3-10 for the full range of SC-CH4 estimates. As discussed in Section 3 of the RIA, a
consideration of climate benefits calculated using discount rates below 3 percent, including 2 percent and lower, are
also warranted when discounting intergenerational impacts. Appendix B presents the results of a sensitivity analysis
using a set of SC-CH4 estimates that incorporates recent research addressing recommendations of the National
Academies of Sciences, Engineering, and Medicine (2017). All net benefits are calculated using climate benefits
discounted at 3 percent.
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C.
0 The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
110,000
280,000
280,000
Table 5-4 Projected Benefits, Compliance Costs, and Emissions Reductions across
Regulatory Options under the Proposed NSPS OOOOb and EG OOOOc, 2023-2035
(million 2019$)
3 Percent Discount Rate
PV
EAV
PV
EAV
Climate Benefits3
Less Stringent
$17,000
$1,600
$17,000
$1,600
Proposal
$48,000
$4,500
$48,000
$4,500
More Stringent
$48,000
$4,500
$48,000
$4,500
3 Percent Discount Rate
7 Percent Discount Rate
PV
EAV
PV
EAV
Net Compliance Costs
Less Stringent
$8,200
$770
$6,000
$710
Proposal
$14,000
$1,400
$12,000
$1,400
More Stringent
$15,000
$1,400
$12,000
$1,400
Compliance Costs
Less Stringent
$9,700
$920
$7,100
$840
Proposal
$19,000
$1,800
$15,000
$1,800
More Stringent
$20,000
$1,800
$15,000
$1,800
Value of Product Recovery
Less Stringent
$1,600
$150
$1,000
$130
Proposal
$4,600
$440
$3,300
$390
More Stringent
$4,700
$440
$3,300
$390
161
-------
Net Benefits
Less Stringent
$8,600 $810 $11,000
$870
Proposal
$34,000 $3,200 $36,000
$3,100
More Stringent
$33,000 $3,100 $36,000
$3,100
Non-Monetized Benefits
Climate and ozone health benefits from reducing methane emissions by (in short tons):
Less Stringent
13,000,000
Proposal
36,000,000
More Stringent
37,000,000
PM2.5 and ozone health benefits from reducing VOC emissions by (in short tons)b c:
Less Stringent
3,600,000
Proposal
9,700,000
More Stringent
10,000,000
HAP benefits from reducing HAP emissions by (in short tons):
Less Stringent
160,000
Proposal
390,000
More Stringent
410,000
Visibility benefits
Reduced vegetation effects
Notes: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding.
a Climate benefits are based on reductions in methane emissions and are calculated using four different estimates of
the social cost of methane (SC-CH4) (model average at 2.5 percent, 3 percent, and 5 percent discount rates; 95th
percentile at 3 percent discount rate). For the presentational purposes of this table, we show the benefits associated
with the average SC-CH4 at a 3 percent discount rate, but the Agency does not have a single central SC-CH4 point
estimate. We emphasize the importance and value of considering the benefits calculated using all four SC-CH4
estimates; see Table 3-8 for the full range of SC-CH4 estimates. As discussed in Section 3 of the RIA, a
consideration of climate benefits calculated using discount rates below 3 percent, including 2 percent and lower, are
also warranted when discounting intergenerational impacts. Appendix B presents the results of a sensitivity analysis
using a set of SC-CH4 estimates that incorporates recent research addressing recommendations of the National
Academies of Sciences, Engineering, and Medicine (2017). All net benefits are calculated using climate benefits
discounted at 3 percent.
b A screening-level analysis of ozone benefits from VOC reductions can be found in Appendix C.
0 The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
The following table shows the total emissions reductions and the PV and EAV of net
compliance costs over the 2023 to 2035 period. The projected net compliance costs for
reciprocating compressors are negative, as the projected revenue from product recovery exceeds
the projected cost increases. This observation may typically support an assumption that operators
would continue to perform the emissions abatement activity, regardless of whether a requirement
is in place, because it is in their private self-interest. However, many of the reciprocating
compressors are in the transmission and storage segment. As discussed in previous oil and
natural gas NSPS RIAs, operators in the transmission and storage segment of the industry do not
typically own the natural gas they transport; rather, the operators receive payment for the
162
-------
transportation service they provide. As a result, financial incentives to reduce emissions may be
minimal because operators are not able to recoup the financial value of captured natural gas that
may otherwise be emitted. Alternatively, there may also be an opportunity cost associated with
the installation of environmental controls (for purposes of mitigating the emission of pollutants)
that is not reflected in the control costs. In the event that the environmental investment displaces
investment in productive capital, the difference between the rate of return on the marginal
investment displaced by the mandatory environmental investment is a measure of the opportunity
cost of the environmental requirement to the regulated entity. However, if firms are not capital
constrained, then there may not be any displacement of investment, and the rate of return on
other investments in the industry would not be relevant as a measure of opportunity cost. If firms
should face higher borrowing costs as they take on more debt, there may be an additional
opportunity cost to the firm. To the extent that any opportunity costs are not added to the control
costs, the compliance cost reductions presented above may be underestimated.
163
-------
Table 5-5 Projected Emissions Reductions, Climate Benefits, and Compliance Costs (millions 2019$) for Incrementally
Affected Sources under the Proposed NSPS OOOOb and EG OOOOc Option, 2023 to 2035
Nationwide Emissions Reductions Costs and Benefits (EAV, million 2019$)
Annualized Cost, Annualized Cost,
Methane VOCa HAP Climate Capital without Product with Product
Source (metric tons CQ2e) (short tons) (short tons) Benefitsb Cost Recovery Recovery
Well Site Fugitives
83,000,000
1,000,000
38,000
$460
$23
$510
$470
Gathering and Boosting Station
Fugitives
Transmission and Storage
Compressor Station Fugitives
Natural Gas Processing Plant
Leaks
15,000,000
190,000
7,100
$85
$1.7
$34
$28
19,000,000
3,700,000
23,000
19,000
680
710
$100
$21
$4.7
-$1.1
$19
$2.4
$12
$1.0
Pneumatic Devices
540,000,000
6,400,000
240,000
$3,000
$910
$570
$360
Reciprocating Compressors
88,000,000
740,000
28,000
$490
$17
$28
-$4.6
Centrifugal Compressors
49,000,000
400,000
42,000
$280
$0
$49
$31
Liquids Unloading
5,700,000
70,000
2,600
$32
$0
$8
$6.2
Storage Vessels
3,900,000
820,000
31,000
$22
$80
$190
$190
Note: Values rounded to two significant figures. Totals may not appear to add correctly due to rounding. Costs and revenue from product recovery in each year
are discounted to 2021. The equivalent annualized values (EAV) in the table are calculated over the 2023 to 2035 period using a 3% discount rate.
a The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated reductions in VOC emissions.
b Climate benefits are based on reductions in methane emissions and are calculated using four different estimates of the social cost of methane (SC-CH4) (model
average at 2.5 percent, 3 percent, and 5 percent discount rates; 95th percentile at 3 percent discount rate). For the presentational purposes of this table, we show
the benefits associated with the average SC-CH4 at a 3 percent discount rate, but the Agency does not have a single central SC-CH4 point estimate. We emphasize
the importance and value of considering the benefits calculated using all four SC-CH4 estimates; see Table 3-8 for the full range of SC-CH4 estimates. As
discussed in Section 3 of the RIA, a consideration of climate benefits calculated using discount rates below 3 percent, including 2 percent and lower, are also
warranted when discounting intergenerational impacts. Appendix B presents the results of a sensitivity analysis using a set of SC-CH4 estimates that incorporates
recent research addressing recommendations of the National Academies of Sciences, Engineering, and Medicine (2017). All net benefits are calculated using
climate benefits discounted at 3 percent.
164
-------
5.2 Uncertainties and Limitations
Throughout the RIA, we considered several sources of uncertainty, both quantitatively
and qualitatively, regarding the emissions reductions, benefits, and costs estimated for the
proposed rule. We summarize the key elements of our discussions of uncertainty below.
Source-level compliance costs and emissions impacts: As discussed in Section 2.2, the
first step in the compliance cost analysis is the development of per-facility national-average
representative costs and emissions impacts using a model plant approach. The model plants are
designed based upon the best information available to the Agency at the time of the rulemaking.
By emphasizing facility averages, geographic variability and heterogeneity across producers in
the industry is masked, and regulatory impacts at the facility-level may vary from the model
plant averages. This assumption is particularly important when assessing the impacts of
requirements that depend on thresholds, such as for storage vessels. For a well site group, which
represents a collection of well sites and their average characteristics, all sites within the group are
categorized as either being below or above the emissions limit, though it may be the case that
some sites within the group exceed the limit while others do not. Mispecifications of this sort
may average out across the full set of well site groups.
There may also be an opportunity cost associated with the installation of environmental
controls (for purposes of mitigating the emissions of pollutants) that is not reflected in the
control costs. In the event that investment in environmental compliance displaces other
investment in productive capital, the difference between the rate of return on the investment
displaced by the mandatory environmental investment is a measure of the opportunity cost of the
environmental requirement. To the extent that such opportunity costs of capital are not accounted
for in the estimated compliance cost reductions, the cost reductions may be underestimated.
Projection methods and assumptions: As discussed in Section 2.2.1, the second
component in estimating national impacts is the projection of affected facilities. Uncertainties in
the projections informing this RIA results include: 1) choice of projection method; 2) data
sources and drivers; 3) limited information about rate of modification and turnover of sources; 4)
behavioral responses to regulation; and 5) unforeseen changes in industry and economic shocks.
165
-------
The list of assumptions required to inform the analysis is too numerous to provide a
comprehensive list, but a few key drivers of the results warrant specific mention.
2016 ICR Data: As discussed previously, the 2016 Oil and Natural Gas ICR was
withdrawn in 2017. Therefore, the data represent an incomplete, and possibly unrepresentative,
survey of operators and well sites. Even so, we believe that it represents the best available data to
use for this analysis, as it includes additional variables beyond, and many more well site
observations than, other equipment surveys that we are aware of (e.g., the API well site survey
discussed in Section 2.2.1.2, which was used to estimate the distribution of fugitive emissions
from components at well sites for the November 2021 RIA). To date, we have not formally
analyzed the representativeness of the data collected, though we may attempt to do so in advance
of the final rulemaking. Informal benchmarks, such as the proportions of single-well versus
multi-well sites and low production versus non-low production sites and average equipment
counts, when compared to outside data sources that attempt to capture the universe of well sites
(such as Enverus and GHGI), did not suggest significant issues with the representativeness of the
ICR data.
Equipment at Well Sites: A major assumption embedded in the analysis is that equipment
at well sites remains fixed over time. This assumption simplifies the analysis, but it ignores the
possibility that as production decreases at a well site, equipment may be removed from the site.
As a result, impacts may be overstated, particularly in the later years of the analysis horizon. We
will continue to assess the validity of this assumption and contemplate alternatives in advance of
the final rule analysis.
Site/Equipment Retirement and Modification: Our assumptions on non-well site
retirement rates are based on impressions stemming from conversations with, and comments
from, industry stakeholders and are not derived from data sources due to a lack of information.
Our assumptions for well site retirement rates are based on an analysis of Enverus data, but we
are still assessing improvements to our methods for estimating those rates. In all cases, we
assume that, prior to implementation of the NSPS and EG, equipment at sites shares the same
vintage as the sites themselves. For example, if a well site was constructed prior to the
promulgation of NSPS OOOO, we assume that all controllers at the site pre-date the NSPS
OOOO as well and are not replaced until the EG goes into effect in 2026. By not accounting for
166
-------
the possibility of equipment replacements and site modifications, we may be overstating the
impacts for some sources that were constructed prior to the NSPS OOOO and/or NSPS 0000a
but are now subject to those rules.
Years of analysis: The years of analysis are 2023, to represent the full first-year facilities
are affected by this action, through 2035, to represent impacts of the rule over a longer period, as
discussed in Section 2.2. While it would be desirable to analyze impacts beyond 2035 in this
RIA, the EPA has chosen not to do this largely because of the limited information available on
the turnover rate of emissions sources and controls. Extending the analysis beyond 2035 would
introduce substantial and increasing uncertainties in the projected impacts of the proposal. That
said, some amount of both benefits and costs would likely continue after 2035, and we note that
toward the end of our analytical time horizon, undiscounted net costs are relatively steady from
year to year (Table 2-9) while undiscounted monetized climate benefits (Table 3-4) are rising
each year. It is therefore plausible that significant net benefits would continue in the years after
2035, though for the reasons given, we do not currently attempt to monetize these effects.
Treatment of sources in Alaska: The RIA does not account for instances in which all or
some sources in Alaska are subject to different proposed requirements than those in the rest of
the country, both in the baseline due to previous rulemakings and in the proposal. For example,
the 2018 amendments to the 2016 NSPS OOOOa ("Alaska Amendments") reduced fugitives
monitoring frequency requirements for well sites and compressor stations on the Alaska North
Slope.109 We do not reflect those reduced requirements in the baseline in this RIA, nor do we
reflect that the same reduced requirements are being proposed for the NSPS OOOOb and EG
0000c. In addition, for sites in Alaska, the NSPS OOOOb and EG 0000c only requires non-
emitting pneumatic controllers to be installed at sites where onsite power is available; otherwise,
the requirement is to replace high-bleed controllers with low-bleed controllers and to monitor
intermittent bleed controllers for malfunctions. In both cases, these omissions suggest that our
analysis may overestimate the impacts of the proposed regulation.
State rules and voluntary action in the baseline: As discussed in Section 2.2.3, while
we accounted for state regulations in California, Colorado, and (to a more limited degree) New
109 83 FR 10628.
167
-------
Mexico and Pennsylvania, there are many other state and local requirements that may be in the
baseline that we are unable to account for. In addition, the baseline does not reflect voluntary
actions firms may take to reduce emissions in the oil and natural gas sector. By not accounting for
state and local requirements (outside of Colorado, California, New Mexico, and Pennsylvania) and
voluntary actions in the baseline, this analysis may overestimate overestimate both the benefits and
costs of the proposed regulation.
Wellhead natural gas prices used to estimate revenues from natural gas recovery:
The compliance cost estimates presented in this RIA include the estimates of the revenue
associated with the increase in natural gas recovery resulting from compliance actions. As a
result, the national compliance cost impacts depend on the price of natural gas. As explained in
Section 2.4 natural gas prices used in this analysis are from the projection of the Henry Hub price
in the AEO2022. To the extent actual natural gas prices diverge from the AEO projections, the
actual impacts will diverge from our estimates.
Oil and natural gas market impact analysis: The oil and natural gas market impact
analysis presented in this RIA is subject to several caveats and limitations. As with any modeling
exercise, the market impact analysis presented here depends crucially on uncertain input
parameters and assumptions regarding market structure. A more detailed discussion of the
caveats and limitations of the oil and natural gas market impacts analysis can be found in Section
4.1.5.
Monetized methane-related climate benefits: The EPA considered the uncertainty
associated with the social cost of methane (SC-CH4) estimates, which were used to calculate the
monetized climate benefits of the decrease in methane emissions projected because of this action.
Section 3.2 provides a detailed discussion of the limitations and uncertainties associated with the
SC-CH4 estimates used in this analysis and describes ways in which the modeling addresses
quantified sources of uncertainty. Appendix B presents the results of a sensitivity analysis using
newly developed SC-CH4 estimates that address updating recommendations of the National
Academies of Sciences, Engineering, and Medicine (2017).
Monetized VOC-related ozone benefits: The illustrative screening analysis described in
Illustrative Screening Analysis of Monetized VOC-Related Ozone Health Benefits includes
many data sources as inputs that are each subject to uncertainty. Input parameters include
168
-------
projected emissions inventories, projected compliance methods, air quality data from models
(with their associated parameters and inputs), population data, population estimates, health effect
estimates from epidemiology studies, economic data, and assumptions regarding the future state
of the world (i.e., regulations, technology, and human behavior). When compounded, even small
uncertainties can greatly influence the size of the total quantified benefits. Below are key
uncertainties associated with estimating the number and value of ozone-related premature deaths.
The estimated number and value of avoided ozone-attributable deaths are subject to
uncertainty. When estimating the economic value of avoided premature mortality from long-term
exposure to ozone, we use a 20-year segment lag as there is no alternative empirical estimate of
the cessation lag for long-term exposure to ozone. The 20-year segmented lag accounts for the
onset of cardiovascular related mortality, an outcome which is not relevant to the long-term
respiratory mortality estimated here. We use a log-linear health impact function without a
threshold in modeling both long- and short-term ozone-related mortality. However, we
acknowledge reduced confidence in specifying the shape of the concentration-response
relationship in the range of < 40ppb and below (U.S. EPA, 2020b). Thus, estimates include
health benefits from reducing ozone in areas with concentrations of ozone down to the lowest
modeled concentrations.
Our estimate of the total monetized ozone-attributable benefits is based on the EPA's
interpretation of the best available scientific literature and methods and supported by the SAB-
HES and the National Academies of Science (NRC, 2002, 2008). Since the publication of these
reports, the EPA has continued improving its techniques for characterizing uncertainty in the
estimated air pollution-attributable benefits. Where possible, we quantitatively assess uncertainty
in each input parameter (for example, statistical uncertainty is characterized by performing
Monte Carlo simulations). However, in some cases, this type of quantitative analysis is not
possible due to lack of data, so we instead characterize the sensitivity of the results to alternative
plausible input parameters. And, for some inputs into the benefits analysis, such as the air quality
data, we lack the data to perform either a quantitative uncertainty analysis or sensitivity analysis.
Additional detail regarding specific uncertainties associated with ozone health benefit estimates
can be found in the TSD for the Final Revised Cross-State Air Pollution Rule for the 2008 Ozone
NAAQS Update titled EstimatingPM2.5- and Ozone-Attributable Health Benefits (U.S. EPA,
202 lg).
169
-------
Non-monetized benefits: Several categories of health, welfare, and climate benefits are
not quantified in this RIA. These unquantified benefits are described in detail in Section 3.
Non-quantified regulatory impacts: We do not attempt to quantify regulatory impacts
for all proposed requirements in this RIA. For a discussion of these requirements, see Section
2.1.2.
Environmental justice analyses: the EPA performed quantitative EJ assessments of
baseline HAP cancer risks, ozone exposure and health risks, employment, and household energy
expenditures. Each of these analyses are subject to various types of uncertainty related to input
parameters and assumptions. Qualitatively, assessments that further subdivide the populations
assess are subject to increased uncertainty as compared to overall exposure and risk estimates.
170
-------
6 REFERENCES
Allen, D. T., Pacsi, A. P., Sullivan, D. W., Zavala-Araiza, D., Harrison, M., Keen, K., . . .
Seinfeld, J. H. (2015). Methane Emissions from Process Equipment at Natural Gas
Production Sites in the United States: Pneumatic Controllers. Environmental Science &
Technology, 49( 1), 633-640. doi:10.1021/es5040156.
Allen, D. T., Torres, V. M., Thomas, J., Sullivan, D. W., Harrison, M., Hendler, A., . . . Seinfeld,
J. H. (2013). Measurements of methane emissions at natural gas production sites in the
United States. Proceedings of the National Academy of Sciences of the United States of
America, 770(44), 18023-18023. doi:10.1073/pnas,1318658110.
Anthoff, D., & Tol, R. S. J. (2013a). Erratum to: The uncertainty about the social cost of carbon:
A decomposition analysis using FUND (vol 117, pg 515, 2013). Climatic Change, 727(2),
413-413. doi:10.1007/s 10584-013-0959-1.
Anthoff, D., & Tol, R. S. J. (2013b). The uncertainty about the social cost of carbon: A
decomposition analysis using FUND. Climatic Change, 117(3), 515-530.
doi: 10.1007/sl0584-013-0706-7.
Arrow, K. J., Cropper, M. L., Eads, G. C., Hahn, R. J., Lave, L. B., Noll, R. G., . . . Stavins, R.
N. (1996). Benefit-Cost Analysis in Environmental, Health, and Safety Regulation: A
Statement of Principles. Washington, DC: American Enterprise Institute, the Annapolis
Center, and Resources for the Future; AEI Press.
ATSDR. (1999). Toxicological Profile for Formaldehyde. Retrieved from Atlanta, GA:
https://www.atsdr.cdc.gov/ToxProfiles/tpl 11 .pdf.
ATSDR. (2007a). Toxicological Profile for Benzene. Retrieved from Atlanta, GA:
https://www.atsdr.cdc.gov/toxprofiles/tp3.pdf.
ATSDR. (2007b). Toxicological Profile for Xylene. Retrieved from Atlanta, GA:
https://www.atsdr.cdc.gov/ToxProfiles/tp71.pdf.
Bednar, D. J., & Reames, T. G. (2020). Recognition of and response to energy poverty in the
United States. Nature Energy, 5(6), 432-439. doi:10.1038/s41560-020-0582-0.
Budzik, P. (2002). U.S. Natural Gas Markets: Relationship Between Henry Hub Spot Prices and
U. S. Wellhead Prices. Energy Information Administration Report.
Cappa, C. D., & Wilson, K. R. (2012). Multi-generation gas-phase oxidation, equilibrium
partitioning, and the formation and evolution of secondary organic aerosol. Atmospheric
Chemistry and Physics, 72(20), 9505-9528. doi: 10.5194/acp-12-9505-2012.
Donahue, N. M., Kroll, J. H., Pandis, S. N., & Robinson, A. L. (2012). A two-dimensional
volatility basis set - Part 2: Diagnostics of organic-aerosol evolution. Atmospheric Chemistry
andPhysics, 72(2), 615-634. doi:10.5194/acp-12-615-2012.
171
-------
Drehobl, A., Ross, L., & Ayala, R. (2020). How High Are Household Energy Burdens? An
Assessment of National and Metropolitan Energy Burdens across the U.S. Retrieved from
Washington, DC: https://www.aceee.org/research-report/u2006.
Ebi, K. L., Balbus, J. L., Luber, G., Bole, A., Crimmins, A., Glass, G., . . . White-Newsome, J. L.
(2018). Chapter 14: Human Health. In D. R. Reidmiller, C. W. Avery, D. R. Easterling, K. E.
Kunkel, K. L. M. Lewis, T. K. May cock, & B. C. Stewart (Eds.), Impacts, Risks, and
Adaptation in the United States: Fourth National Climate Assessment (Vol. II, pp. 539-571).
Washington, DC: U.S. Global Change Research Program.
EIA. (2018). One in three U.S. households faces a challenge in meeting energy needs. Retrieved
from Washington, DC: https://www.eia.gov/todayinenergy/detail.php?id=37072.
Hall, S. M. (2013). Energy justice and ethical consumption: comparison, synthesis and lesson
drawing. Local Environment, 75(4), 422-437. doi:10.1080/13549839.2012.748730.
Hausman, C., & Kellogg, R. (2015). Welfare and Distributional Implications of Shale Gas.
Brookings Papers on Economic Activity, 71-125. Retrieved from
https://www.brookings.edu/bpea-articles/welfare-and-distributional-implications-of-shale-
gas/.
Helmig, D., Thompson, C. R., Evans, J., Boylan, P., Hueber, J., & Park, J. H. (2014). Highly
Elevated Atmospheric Levels of Volatile Organic Compounds in the Uintah Basin, Utah.
Environmental Science & Technology, 48(9), 4707-4715. doi:10.1021/es405046r.
Hernandez, D., Aratani, Y., & Jiang, Y. (2014). Energy Insecurity Among Families with
Children. Retrieved from New York, NY: https://www.nccp.org/wp-
content/uploads/2020/05/text_1086.pdf.
Hope, C. (2013). Critical issues for the calculation of the social cost of C02: why the estimates
from PAGE09 are higher than those from PAGE2002. Climatic Change, 117(3), 531-543.
doi: 10.1007/sl0584-012-0633-z.
Hotelling, H. (1931). The economics of exhaustible resources. Journal of Political Economy,
39(2), 137-175.
IARC. (1982). Some Industrial Chemicals andDyestuffs (Volume 29). Retrieved from Lyon,
France: https://publications.iarc.fr/47.
IARC. (1987). Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs
Volumes 1 to 42 (Supplement 7). Retrieved from Lyon, France:
https://publications.iarc.fr/139.
IARC. (2000). Some Industrial Chemicals (Volume 77). Retrieved from Lyon, France:
https://publications.iarc.fr/95.
172
-------
IARC. (2006). Formaldehyde, 2-Butoxyethanol and l-tert-Butoxypropan-2-ol (Volume 88).
Retrieved from Lyon, France: https://publications.iarc.fr/106.
IARC. (2012). Chemical Agents and Related Occupations (Volume 100F). Retrieved from Lyon,
France: https://publications.iarc.fr/123.
IPCC. (2007). Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II
and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
(Core Writing Team, R. K. Pachauri, & A. Reisinger Eds.). Geneva, Switzerland:
Intergovernmental Panel of Climate Change.
IPCC. (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II
and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
(Core Writing Team, R. K. Pachauri, & L. A. Meyer Eds.). Geneva, Switzerland:
Intergovernmental Panel of Climate Change.
IPCC. (2018). Global Warming of 1.5°C. An IPCC Special Report on the impacts of global
warming of 1.5°C above pre-industrial levels and related global greenhouse gas emission
pathways, in the context of strengthening the global response to the threat of climate change,
sustainable development, and efforts to eradicate poverty (V. Masson-Delmotte, P. Zhai, H-
O. Portner, D. Roberts, J. Skea, P. R. Shukla, A. Pirani, W. Moufouma-Okia, C. Pean, R.
Pidcock, S. Connors, J. B. R. Matthews, Y. Chen, X. Zhou, M. I. Gomis, E. Lonnoy, T.
Maycock, a. M. Tignor, & T. Waterfield Eds.).
IPCC. (2019a). Climate Change and Land: an IPCC special report on climate change,
desertification, land degradation, sustainable land management, food security, and
greenhouse gas fluxes in terrestrial ecosystems (P.R. Shukla, J. Skea, E. Calvo Buendia, V.
Masson-Delmotte, H.-O. Portner, D. C. Roberts, P. Zhai, R. Slade, S. Connors, R. van
Diem en, M. Ferrat, E. Haughey, S. Luz, S. Neogi, M. Pathak, J. Petzold, J. Portugal Pereira,
P. Vyas, E. Huntley, K. Kissick, M. Belkacemi, & J. Malley Eds.).
IPCC. (2019b). IPCC Special Report on the Ocean and Cryosphere in a Changing Climate (H-
O. Portner, D.C. Roberts, V. Masson-Delmotte, P. Zhai, M. Tignor, E. Poloczanska, K.
Mintenbeck, A. Alegria, M. Nicolai, A. Okem, J. Petzold, B. Rama, & N. M. Weyer Eds.).
IPCC. (2021). Climate Change 2021: The Physical Science Basis. Contribution of Working
Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change
(V. Masson-Delmotte, P. Zhai, A. Pirani, S. L. Connors, C. Pean, S. Berger, N. Caud, Y.
Chen, L. Goldfarb, M. I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J. B. R. Matthews, T. K.
Maycock, T. Waterfield, O. Yelek<;i, R. Yu, & B. Zhou Eds.): Cambridge University Press.
Irons, R. D., Stillman, W. S., Colagiovanni, D. B., & Henry, V. A. (1992). Synergistic action of
the benzene metabolite hydroquinone on myelopoietic stimulating activity of
granulocyte/macrophage colony-stimulating factor in vitro. Proceedings of the National
Academy of Sciences, 59(9), 3691-3695. doi:10.1073/pnas.89.9.3691.
173
-------
IWG. (2010). Technical Support Document: Social Cost of Carbon for Regulatory Impact
Analysis under Executive Order 12866. Retrieved from Washington, DC:
https://www.epa.gov/sites/default/files/2016-12/documents/scc_tsd_2010.pdf.
IWG. (2013). Technical Support Document: Technical Update of the Social Cost of Carbon for
Regulatory Impact Analysis Under Executive Order 12866. Retrieved from Washington, DC:
https://www.ourenergypolicy.org/wp-
content/uploads/2013/06/social_cost_of_carbon_for_ria_2013_update.pdf.
IWG. (2015). Technical Support Document: Technical Update of the Social Cost of Carbon for
Regulatory Impact Analysis Under Executive Order 12866. Retrieved from Washington, DC:
https://www.regulations.gov/document/EPA-HQ-OAR-2015-0827-0454.
IWG. (2016a). Addendum to Technical Support Document on Social Cost of Carbon for
Regulatory Impact Analysis under Executive Order 12866: Application of the Methodology
to Estimate the Social Cost of Methane and the Social Cost of Nitrous Oxide. Retrieved from
Washington, DC: https://www.epa.gov/sites/default/files/2016-
12/documents/addendum_to_sc-ghg_tsd_august_2016 .pdf.
IWG. (2016b). Technical Support Document: Technical Update of the Social Cost of Carbon for
Regulatory Impact Analysis Under Executive Order 12866. Retrieved from Washington, DC:
https://www.epa.gov/sites/default/files/2016-12/documents/sc_co2_tsd_august_2016.pdf.
IWG. (2021). Technical Support Document: Social Cost of Carbon, Methane, and Nitrous Oxide
Interim Estimates under Executive Order 13990. Retrieved from Washington, DC:
https://www.whitehouse.gov/wp-
content/uploads/2021/02/TechnicalSupportDocument_SocialCostofCarbonMethaneNitrousO
xide.pdf?source=email.
Jantarasami, L., Novak, R., Delgado, R., Marino, E., McNeeley, S., Narducci, C., . . . Powys
Whyte, K. (2018). Tribes and Indigenous Peoples. In Impacts, Risks, and Adaptation in the
United States: Fourth National Climate Assessment (Vol. II). Washington, DC: U.S. Global
Change Research Program.
Jessel, S., Sawyer, S., & Hernandez, D. (2019). Energy, Poverty, and Health in Climate Change:
A Comprehensive Review of an Emerging Literature. Frontiers in Public Health, 7, 357.
doi: 10.3389/fpubh.2019.00357.
Jimenez, J. L., Canagaratna, M. R., Donahue, N. M., Prevot, A. S. H., Zhang, Q., Kroll, J. H., . . .
Worsnop, D. R. (2009). Evolution of Organic Aerosols in the Atmosphere. Science,
326(5959), 1525-1529. doi: 10.1126/science. 1180353.
Kaiser, M. J., & Pulsipher, A. G. (2006). Concerns Over the Allocation Methods Employed in
the US Low-Income Home Energy Assistance Program. INFORMS Journal on Applied
Analytics, 36(4), 344-358. doi:10.1287/inte,1060.0223.
174
-------
Karpinska, L., & Smiech, S. (2020). Invisible energy poverty? Analysing housing costs in
Central and Eastern Europe. Energy Research & Social Science, 70, 101670.
doi: 10.1016/j.erss.2020.101670.
Koss, A., Yuan, B., Warneke, C., Gilman, J. B., Lerner, B. M., Veres, P. R., . . . de Gouw, J.
(2017). Observations of VOC emissions and photochemical products over US oil- and gas-
producing regions using high-resolution H30+ CIMS (PTR-ToF-MS). Atmospheric
Measurement Techniques, 10(8), 2941-2968. doi:10.5194/amt-10-2941-2017.
Lyubich, E. (2020). The Race Gap in Residential Energy Expenditures (306). Retrieved from
Berkeley, CA: http://www.haas.berkeley.edu/wp-content/uploadsAVP306.pdf.
Maltoni, C., Ciliberti, A., Pinto, C., Soffritti, M., Belpoggi, F., & Menarini, L. (1997). Results of
long-term experimental carcinogenicity studies of the effects of gasoline, correlated fuels,
and major gasoline aromatics on rats. Annals of the New York Academy of Sciences, 837, 15-
52. doi: 10.1111/j. 1749-6632.1997.tb56863.x.
Maltoni, C., Conti, B., Cotti, G., & Belpoggi, F. (1985). Experimental studies on benzene
carcinogenicity at the Bologna Institute of Oncology: current results and ongoing research.
American Journal of Industrial Medicine, 7(5-6), 415-446. doi: 10.1002/ajim.4700070508.
Marten, A. L., Kopits, E. A., Griffiths, C. W., Newbold, S. C., & Wolverton, A. (2015).
Incremental CH4 and N20 mitigation benefits consistent with the US Government's SC-
C02 estimates. Climate Policy, 15(2), 272-298. doi:10.1080/14693062.2014.912981.
Myhre, G., Shindell, D., Breon, F.-M., Collins, W., Fuglestvedt, J., Huang, J., . . . Zhang, H.
(2013). Anthropogenic and Natural Radiative Forcing. In T. Case (Ed.), Climate Change
2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and
New York, NY: Cambridge University Press.
National Academies. (2017). Valuing Climate Damages: Updating Estimation of the Social Cost
of Carbon Dioxide. Washington, DC: The National Academies Press.
National Academies. (2019). Framing the Challenge of Urban Flooding in the United States.
Washington, DC: The National Academies Press.
Nolte, C. G., Dolwick, P. D., Fann, N., Horowitz, L. W., Naik, V., Pinder, R. W., . . . Ziska, L.
H. (2018). Chapter 13: Air Quality. In C. Test (Ed.), Impacts, Risks, and Adaptation in the
United States: Fourth National Climate Assessment (Vol. II, pp. 512-538). Washington, DC:
U.S. Global Change Research Program.
Nordhaus, W. D. (2010). Economic aspects of global warming in a post-Copenhagen
environment. Proceedings of the National Academy of Sciences of the United States of
America, 107(26), 11721-11726. doi: 10.1073/pnas. 1005985107.
175
-------
NRC. (2002). Estimating the Public Health Benefits of Proposed Air Pollution Regulations.
Washington, DC: The National Academies Press.
NRC. (2008). Estimating Mortality Risk Reduction and Economic Benefits from Controlling
Ozone Air Pollution. Washington, DC: The National Academies Press.
NRC. (2011). America's Climate Choices. Washington, DC: The National Academies Press.
NTP. (1999). Toxicology and Carcinogenesis Studies of Ethylbenzene (CAS No. 100-41-4) in
F344/NRats andB6C3Fl Mice (Inhalation Studies). National Toxicology Program
Technical Report Series, 466, 1-231.
NTP. (2004). Report on Carcinogens, Eleventh Edition (PB2005104914). Retrieved from
Research Triangle Park, NC:
https://ntrl.ntis.gov/NTRL/dashboard/searchResults/titleDetail/PB2005104914.xhtml.
NTP. (2016). Report on Carcinogens, Fourteenth Edition. Retrieved from Research Triangle
Park, NC: https://ntp.niehs.nih.gov/go/rocl4.
OMB. (2003). Circular A-4. (68 FR 58366). Washinton, DC: Office of Management and Budget
Retrieved from https://www.federalregister.gOv/documents/2003/10/09/03-25606/circular-a-
4-regulatory-analysis.
OMB. (2015). 2015 Report to Congress on the Benefits and Costs of Federal Regulations and
Agency Compliance with the Unfunded Mandates Reform Act. Retrieved from Washington,
DC:
https://www.whitehouse.gOv/sites/whitehouse.gov/files/omb/inforeg/inforeg/2015_cb/2015-
cost-benefit-report.pdf.
Oppenheimer, M., Campos, M., Warren, R., Birkmann, J., Luber, G., O'Neill, B., & Takahashi,
K. (2014). Emergent risks and key vulnerabilities. In C.B. Field, V.R. Barros, D.J. Dokken,
K.J. Mach, M.D. Mastrandrea, T.E. Bilir, M. Chatteijee, K.L. Ebi, Y.O. Estrada, R.C.
Genova, B. Girma, E.S. Kissel, A.N. Levy, S. MacCracken, P.R. Mastrandrea, & L.L.White
(Eds.), Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and
Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change (pp. 1039-1099). Cambridge, United Kingdom
and New York, NY: Cambridge University Press.
Petron, G., Frost, G., Miller, B. R., Hirsch, A. I., Montzka, S. A., Karion, A., . . . Tans, P. (2012).
Hydrocarbon emissions characterization in the Colorado Front Range: A pilot study. Journal
of Geophysical Research: Atmospheres, 777(D4). doi: 10.1029/2011JD016360.
Porter, J. R., Xie, L., Challinor, A. J., Cochrane, K., Howden, M., Iqbal, M. M., & Lobell, D. B.
(2014). Food security and food production systems. In C.B. Field, V.R. Barros, D.J. Dokken,
K.J. Mach, M.D. Mastrandrea, T.E. Bilir, M. Chatteijee, K.L. Ebi, Y.O. Estrada, R.C.
Genova, B. Girma, E.S. Kissel, A.N. Levy, S. MacCracken, P.R. Mastrandrea, & L.L.White
176
-------
(Eds.), Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and
Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change (pp. 485-533). Cambridge, United Kingdom
and New York, NY: Cambridge University Press.
Siliverstovs, B., L'Hegaret, G., Neumann, A., & von Hirschhausen, C. (2005). International
market integration for natural gas? A cointegration analysis of prices in Europe, North
America and Japan. Energy Economics, 27(4), 603-615. doi:10.1016/j.eneco.2005.03.002.
Sarofim, M. C., Waldhoff, S. T., & Anenberg, S. C. (2017). Valuing the ozone-related health
benefits of methane emission controls. Environmental and Resource Economics, 66( 1),
45-63.
Smith, K. R., Woodward, A., Campbell-Lendrum, D., Chadee, D. D., Honda, Y., Liu, Q., . . .
Sauerborn, R. (2014). Human Health: Impacts, Adaptation, and Co-Benefits. In C.B. Field,
V.R. Barros, D.J. Dokken, K.J. Mach, M.D. Mastrandrea, T.E. Bilir, M. Chatteijee, K.L. Ebi,
Y.O. Estrada, R.C. Genova, B. Girma, E.S. Kissel, A.N. Levy, S. MacCracken, P.R.
Mastrandrea, & L.L.White (Eds.), Climate Change 2014: Impacts, Adaptation, and
Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II to the
Fifth Assessment Report of the Intergovernmental Panel on Climate Change (pp. 709-754).
Cambridge, United Kingdom and New York, NY: Cambridge University Press.
U.S. BEA. (2021). Implicit Price Deflators for Gross Domestic Product, Table 1.1.9. Retrieved
from https://apps.bea.gov/iTable/index_nipa.cfm.
U.S. EPA. (1991a). Carbonylsulfide; CASRN463-58-1. Retrieved from
https://iris.epa.gov/ChemicalLanding/&substance_nmbr=617.
U.S. EPA. (1991b). Formaldehyde; CASRN50-00-0. Retrieved from
https://iris.epa.gov/ChemicalLanding/&substance_nmbr=419.
U.S. EPA. (2003a). Benzene; CASRN 71-43-2. Retrieved from
https://iris.epa.gov/ChemicalLanding/&substance_nmbr=276.
U.S. EPA. (2003b). Xylenes; CASRN 1330-20-7. Retrieved from
https://iris.epa.gov/ChemicalLanding/&substance_nmbr=270.
U.S. EPA. (2005a). Guidelines for Carcinogen Risk Assessment (EPA-630-P-03-001F).
Retrieved from Washington, DC: https://www.epa.gov/sites/default/files/2013-
09/documents/cancer_guidelines_final_3-25-05.pdf.
U.S. EPA. (2005b). Toluene; CASRN 108-88-3. Retrieved from
https://iris.epa.gov/ChemicalLanding/&substance_nmbr=118.
177
-------
U.S. EPA. (2006). Regulatory Impact Analysis, 2006 National Ambient Air Quality Standards
for Particulate Matter. Retrieved from Research Triangle Park, NC:
https://www.epa.gOv/sites/default/files/2020-07/documents/naaqs-pm_ria_final_2006-10.pdf.
U. S. EPA. (2009a). Integrated Science Assessment for Particulate Matter (Final Report). (EPA-
600/R-08-139F). Research Triangle Park, NC: Office of Research and Development,
National Center for Environmental Assessment. Retrieved from
https://cfpub. epa.gov/ncea/risk/recordisplay. cfm?deid=216546.
U.S. EPA. (2009b). Technical Support Document for Proposed Endangerment and Cause or
Contribute Findings for Greenhouse Gases under Section 202(a) of the Clean Air Act.
Retrieved from Washington, DC: https://www.epa.gov/sites/default/files/2016-
08/documents/endangermenttsd. pdf.
U.S. EPA. (2010). Summary of the updated Regulatory Impact Analysis, National Ambient Air
Quality Standards for Ozone. Retrieved from Research Triangle Park, NC:
https://www3.epa.gov/ttnecasl/regdata/RIAs/sl-supplemental_analysis_summaryll-5-
09.pdf.
U.S. EPA. (2011a). The Benefits and Costs of the Clean Air Act from 1990 to 2020. Washington,
DC. Retrieved from https://www.epa.gov/sites/production/files/2015-
07/documents/fullreport_rev_a.pdf.
U. S. EPA. (20 lib). Draft Residual Risk Assessment for the Oil and Gas Production and Natural
Gas Transmission and Storage Source Categories. Retrieved from Research Triangle Park,
NC: https://www.regulations.gOv/document/EPA-HQ-OAR-2010-0505-0032.
U.S. EPA. (201 lc). Oil and Natural Gas Sector: Standards of Performance for Crude Oil and
Natural Gas Production, Transmission, and Distribution. Background Technical Support
Document for Proposed Standards (EPA-453-R-11-002). Retrieved from Research Triangle
Park, NC: https://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=P 100CHTC.txt.
U.S. EPA. (201 Id). Regulatory Impact Analysis for the Federal Implementation Plans to Reduce
Interstate Transport of Fine Particulate Matter and Ozone in 27 States; Correction of SIP
Approvals for 22 States. Retrieved from Research Triangle Park, NC:
https://www3.epa.gov/ttn/ecas/docs/ria/transport_ria_final-csapr_2011-06.pdf.
U. S. EPA. (20 lie). Regulatory Impact Analysis for the Federal Implementation Plans to Reduce
Interstate Transport of Fine Particulate Matter and Ozone in 27 States; Correction of SIP
Approvals for 22 States. Research Triangle Park, NC: U.S. Environmental Protection
Agency, Office of Air Quality Planning and Standards, Health and Environmental Impact
Division. Retrieved from https://www3.epa.gov/ttn/ecas/docs/ria/transport_ria_final-
csapr_201 l-06.pdf.
U.S. EPA. (2012). Regulatory Impact Analysis for the Final Revisions to the National Ambient
Air Quality Standards for Particulate Matter. (EPA-452/R-12-005). Research Triangle Park,
178
-------
NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards,
Health and Environmental Impact Division. Retrieved from
https://www3.epa.gov/ttnecasl/regdata/RIAs/finalria.pdf.
U.S. EPA. (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants
(FinalReport) (EPA-600-R-10-076F). Retrieved from Research Triangle Park, NC:
https://cfpub.epa.gOv/ncea/i sa/recordisplay.cfm?deid=247492.
U.S. EPA. (2015 a). Proposed Rulemaking for Greenhouse Gas Emissions and Fuel Efficiency
Standards for Medium- and Heavy-Duty Engines and Vehicles-Phase 2. (EPA-420-D-15-
900). Washington, D.C.: Office of Transportation and Air Quality, Assessment and Standards
Division. Retrieved from
https://nepis. epa.gov/Exe/ZyPDF.cgi/P100MKYR.PDF?Dockey=P 100MKYR.PDF.
U.S. EPA. (2015b). Regulatory Impact Analysis for the Proposed Revisions to the Emission
Guidelines for Existing Sources and Supplemental Proposed New Source Performance
Standards in the Municipal Solid Waste Landfills Sector. (EPA-452/R-15-008). Research
Triangle Park, NC: Office of Air Quality Planning and Standards. Retrieved from
https://www.regulations.gOv/document/EPA-HQ-OAR-2014-0451-0086.
U. S. EPA. (2015 c). Regulatory Impact Analysis of the Proposed Emission Standards for New
and Modified Sources in the Oil and Natural Gas Sector. (EPA-452/R-15-002). Research
Triangle Park, NC: Office of Air Quality Planning and Standards. Retrieved from
https://www.epa.gOv/sites/default/files/2020-07/documents/oilgas_ria_proposed-nsps_2015-
08.pdf.
U.S. EPA. (2016). Regulatory Impact Analysis of the Final Oil and Natural gas Sector: Emission
Standards for New, Reconstructed, and Modified Sources. (EPA-452/R-16-002). Research
Triangle Park, NC: Office of Air Quality Planning and Standards. Retrieved from
https://www.epa.gOv/sites/default/files/2020-07/documents/oilgas_ria_nsps_final_2016-
05.pdf.
U.S. EPA. (2018). Technical Support Document: EPA 's 2014 National Air Toxics Assessment.
Retrieved from Research Triangle Park, NC: https://www.epa.gov/national-air-toxics-
assessment/2014-nata-technical-support-document.
U.S. EPA. (2019). Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2019 (EPA-
430-R-19-001). Retrieved from Washington, DC:
https://www.epa.gov/ghgemissions/inventory-us-greenhouse-gas-emissions-and-sinks-1990-
2017.
U.S. EPA. (2020a). Integrated Science Assessment for Ozone and Related Photochemical
Oxidants (FinalReport). (EPA/600/R-20/012). Washington, DC: U.S. Environmental
Protection Agency. Retrieved from https://www.epa.gov/isa/integrated-science-assessment-
i sa-ozone-and-rel ated-photochemi cal -oxi dants.
179
-------
U.S. EPA. (2020b). Preparation of Emissions Inventories for 2017 North American Emissions.
Retrieved from Research Triangle Park, NC: https://www.epa.gov/air-emissions-
modeling/2017-emissions-modeling-platform-supporting-documentation.
U.S. EPA. (2020c). Regulatory Impact Analysis for the Review and Reconsideration of the Oil
and Natural Gas Sector Emission Standards for New, Reconstructed, and Modified Sources.
(EPA-452/R-20-004). Research Triangle Park, NC: U.S. Environmental Protection Agency,
Office of Air Quality Planning and Standards, Health and Environmental Impact Division
Retrieved from https://www.epa.gov/sites/default/files/2020-
08/documents/oil_and_natural_gas_nsps_review_and_reconsideration_final_ria.pdf.
U.S. EPA. (2021a). 2017 National Emission Inventory Based Photochemical Modeling for
Sector Specific Air Quality Assessments (EPA-454-R-21-005). Retrieved from Research
Triangle Park, NC: https://www.epa.gov/system/files/documents/2021-08/epa-454-r-21-
005.pdf.
U.S. EPA. (2021b). 2017 National Emission Inventory: January 2021 Updated Release,
Technical Support Document (EPA-454-R-21-001). Retrieved from Research Triangle Park,
NC: https://www.epa.gov/sites/default/files/2021-
02/documents/nei2017_tsd_full j an2021 .pdf.
U.S. EPA. (2021c). Climate Change and Social Vulnerability in the United States: A Focus on
Six Impacts (EPA-430-R-21-003). Retrieved from Washington, DC:
https://www.epa.gov/cira/social-vulnerability-report.
U.S. EPA. (2021d). Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2019 (EPA-
430-R-21-005). Retrieved from Washington, DC:
https://www.epa.gov/ghgemissions/inventory-us-greenhouse-gas-emissions-and-sinks-1990-
2019.
U.S. EPA. (202 le). Oil and Natural Gas Sector: Emission Standards for New, Reconstructed,
and Modified Sources and Emissions Guidelines for Existing Sources: Oil and Natural Gas
Sector Climate Review Background Technical Support Document for the Proposed New
Source Performance Standards (NSPS) and Emissions Guidelines (EG). Retrieved from
Research Triangle Park, NC: https://www.regulations.gov/docket/EPA-HQ-OAR-2021-
0317-0166.
U.S. EPA. (202 If). Regulatory Impact Analysis for the Final Revised Cross-State Air Pollution
Rule (CSAPR) Update for the 2008 Ozone NAAQS (EPA-452-R-21-002). Retrieved from
Research Triangle Park, NC: https://www.epa.gov/csapr/revised-cross-state-air-pollution-
rule-update.
U.S. EPA. (202 lg). Technical Support Document (TSD) for the Final Revised Cross-State Air
Pollution Rule Update for the 2008 Ozone Season NAAQS: Estimating PM2.5- and Ozone-
Attributable Health Benefits (Docket ID No. EPA-HQ-OAR-2020-0272). Retrieved from
Research Triangle Park, NC: https://www.epa.gov/csapr/revised-cross-state-air-pollution-
rule-update.
180
-------
U.S. EPA. (2022). Report on the Social Cost of Greenhouse Gases: Estimates Incorporating
Recent Scientific Advances (Docket ID No. EPA-HQ-OAR-2021-0317).
U. S. GAO. (2020). Social Cost of Carbon: Identifying a Federal Entity to Address the National
Academies' Recommendations Could Strengthen Regulatory Analysis. GAO-20-254.
Retrieved from Washington, DC: https://www.gao.gov/products/gao-20-254.
USGCRP. (2016). The Impacts of Climate Change on Human Health in the United States: A
Scientific Assessment. Retrieved from Washington, DC:
http://dx.doi.org/10.7930/NCA4.2018.
USGCRP. (2018). Impacts, Risks, and Adaptation in the United States: Fourth National Climate
Assessment, Volume II. Retrieved from Washington, DC:
http://dx.doi.org/10.7930/J0R49NQX.
Wang, Q., Kwan, M.-P., Fan, J., & Lin, J. (2021). Racial disparities in energy poverty in the
United States. Renewable and Sustainable Energy Reviews, 137, 110620.
doi: 10.1016/j.rser.2020.110620.
WHO. (2002). Concise International Chemical Assessment Document 40: Formaldehyde.
Retrieved from Geneva, Switzerland:
https://www.who.int/ipcs/publications/cicad/en/cicad40.pdf.
Zavala-Araiza, D., Alvarez, R. A., Lyon, D. R., Allen, D. T., Marchese, A. J., Zimmerle, D. J., &
Hamburg, S. P. (2017). Super-emitters in natural gas infrastructure are caused by abnormal
process conditions. Nat Commun, 8, 14012. doi:10.1038/ncommsl4012.
Zimmerle, D., Vaughn, T., Luck, B., Lauderdale, T., Keen, K., Harrison, M., . . . Allen, D.
(2019). Methane Emissions from Gathering and Boosting Compressor Stations in the U.S.:
Supporting Volume 3: Emission Factors, Station Estimates, and National Emissions.
Retrieved from http://dx.doi.org/10.25675/10217/194541.
181
-------
APPENDIX A ADDITIONAL INFORMATION ON COST AND EMISSIONS
ANALYSIS
In this appendix, we provide additional information on topics related to the development
of well site activity data and cost estimation. Specifically, we describe in detail how the 2016
ICR data was used to calculate equipment bin proportions and average equipment factors for
well sites and how equipment counts at well sites were calibrated to the GHGI for the base year.
A.l Calculation of Equipment Bin Proportions and Average Equipment Factors for Well
Sites from 2016 ICR Data
The 2016 ICR data includes a survey of equipment at well sites and production
characteristics of the wells at those sites. The data are cleaned and processed to generate
estimates of the distribution of major equipment and storage vessels across sites which can be
directly applied to our base year well site activity data as described in Section 2.2.1. The data
processing steps, and a list of key assumptions made along the way, are summarized below.
The first step in the data cleaning procedure is to read in and remove duplicate and
incomplete entries from the raw 2016 ICR data workbook. Well-level data comes from the sheets
"PtlWellsFromWebForms" and "PtlWellsFromFileUploads", with corresponding site-level data
sheets "PtlWellSiteFromWebForms" and "PtlWellSiteFromFileUploads". Data from both types
of submissions are merged together to create two master raw data tables, one for wells and one
for sites. For both types of data, duplicate entries were removed, first on the basis of having
identical entries for all rows, and then on the basis of having identical well/wellsite IDs. For the
latter, when the only difference is the submission time entry, we assume the last entry
(chronologically) is the correct one. If submission time cannot be used to differentiate in the
well-level data, we simply pick the first entry,110 unless duplicate entries have different well
types (oil or gas), in which case we drop the wells from the sample. Finally, we remove sites that
do not produce gas or oil and choose the first entry for sites with entries that only differ by their
latitude/longitude values.
110 For non-unique wells with the same well type and submission time, the only difference is found in the
"pt lproductionsiteid" column, a distinction that we assume is meaningless for this exercise.
182
-------
The next step is to standardize key information in the well and wellsite data tables. Wells
identified in the raw data as "active" or "producing" are grouped together. Likewise, wells
identified as producing gas (wet, dry, or unknown) or coal bed methane are designated as gas
wells, while wells identified as producing oil (light, heavy, or unknown) are designated as oil
wells. Finally, wells are designated either as low production, non-low production, or unknown.
For well sites, data on equipment counts are standardized such that counts are either left blank if
valid information has not been provided, or equal to an integer (for separators, dehydrators, and
compressors) or real (for tanks) value. Sites are further designated as having full equipment
inventories if valid entries were provided for all equipment columns and partial equipment
inventories if at least one equipment column has a valid entry and the remaining columns were
left blank, in which case blank entries were designated as zeroes.
After data standardization, the well and wellsite data is merged to create as single dataset
with information on equipment and production. The well-level data is aggregated to create site-
level estimates of the number of wells for each combination of oil and gas and low and non-low
production level; this step removes any well entries for which the production type and level is
not known. The aggregated data is then merged with the wellsite data based on the native
"operatorname" and "well site id name" columns. To facilitate use with the well site activity
data used for the impacts analysis, sites are then characterized as single well or multi-well sites,
oil or gas sites (based on whether there are more oil or gas wells at the site, with ties designated
as oil sites), and low or non-low production sites (only sites with exclusively low production
wells were designated as low production sites). Then well sites are assigned to one of the six
equipment/tank categories: (1) no equipment or tanks; (2) no equipment with storage tanks; (3)
one piece of major equipment without tanks; (4) one piece of major equipment with tanks; (5)
more than one piece of major equipment without tanks; and (6) more than one piece of major
equipment with tanks.
Finally, the dataset is aggregated to calculate the proportion of sites and average
equipment counts per well in each equipment/tank category. For the aggregate calculations, we
include sites with full and partial equipment inventories. To facilitate calibration with the GHGI,
which has activity data for wells but not well sites, we calculate average equipment counts per
oil well and per gas well. To do the per-well calculation, equipment counts at each site are
allocated to wells in proportion to the number of oil and gas wells at the sites, and then
183
-------
aggregated by production level, well count bin, and equipment category. The aggregated
equipment counts are then divided by aggregate counts of oil and gas wells for all sites to arrive
at equipment count averages per well for the sites in the final data set.
The results for well site proportions, stratified by production type and level and well
count bin, are presented in Table A-l. A significant portion of sites, particularly single wellhead
oil sites, do not have any major equipment or tanks. Larger sites, both in terms of production
levels and well counts, tend to have more equipment for both site types.
Table A-l Well Site Equipment/Tank Category Proportions Estimated From the 2016
ICR
Gas Oil
Non-low Non-low
Low Production Production Low Production Production
Equipment/Tank Category
Single
Multi
Single
Multi
Single
Multi
Single
Multi
(1) No major equipment or tanks
37%
4%
33%
1%
58%
11%
44%
3%
(2) No major equipment with
tanks
5%
14%
2%
1%
9%
24%
2%
6%
(3) One piece of major
equipment without tanks
21%
4%
12%
4%
2%
1%
8%
1%
(4) One piece of major
equipment with tanks
2%
5%
1%
9%
0%
0%
1%
16%
(5) Two or more pieces of major
equipment without tanks
30%
30%
39%
6%
24%
44%
23%
7%
(6) Two or more pieces of major
equipment with tanks
5%
43%
13%
79%
6%
20%
22%
67%
The results for well site equipment averages, stratified by production level, well count
bin, and equipment category, are presented in Table A-2. Typically, non-low production sites
tend to have more equipment than low production sites, particularly when it comes to separators,
though the relationships is not unambiguous across all well type, well count bin, and equipment
category permutations. Also, for the same well type, well count bin, and equipment category,
multi-well sites tend to have fewer pieces of equipment per well.
184
-------
Table A-2 Per-Well Average Equipment/Tank Counts Estimated From the 2016 ICR
Equipment Count Per Well
Well
Type
Site
Site Well
Production Count
Level Bin
Equipment
Category
Separators Compressors Dehydrators Tanks
Single
Low
Multi
Gas
(1)
(2)
(3)
(4)
(5)
(6)
(1)
(2)
(3)
(4)
(5)
(6)
0.97
1.20
0.99
1.87
0.29
0.85
0.34
0.91
0.02
0.86
0.01
0.38
0.01
0.32
0.00
0.15
0.01
0.22
0.01
0.13
0.00
0.01
0.01
0.05
1.48
1.50
2.03
0.67
0.64
0.86
Single
Non-low
Multi
(1)
(2)
(3)
(4)
(5)
(6)
(1)
(2)
(3)
(4)
(5)
(6)
0.94
1.74
0.95
1.82
0.26
1.03
0.32
1.00
0.04
0.51
0.01
0.50
0.02
0.04
0.02
0.08
0.02
0.17
0.04
0.19
0.00
0.03
0.00
0.02
2.22
1.66
2.23
2.70
0.91
0.92
185
-------
Single
Low
Multi
Oil
Single
Non-low
Multi
(1)
(2)
(3)
(4)
(5)
(6)
(1)
(2)
(3)
(4)
(5)
(6)
(1)
(2)
(3)
(4)
(5)
(6)
(1)
(2)
(3)
(4)
(5)
(6)
0.84
1.27
0.98
1.94
0.21
0.58
0.21
0.50
0.99
1.62
0.99
2.24
0.12
0.77
0.25
1.04
0.15
0.86
0.02
0.35
0.03
0.07
0.00
0.05
0.01
0.54
0.01
0.51
0.00
0.01
0.01
0.12
0.01
0.13
0.00
0.05
0.00
0.07
0.00
0.01
0.00
0.08
0.00
0.06
0.01
0.00
0.00
0.02
2.17
2.16
3.77
0.61
0.48
0.70
3.23
4.08
4.70
1.00
1.56
2.01
A.2 Equipment Count Calibration at Well Sites
The equipment count calibration performed for this analysis ensures that base year
estimates of certain types of equipment at well sites matches, in aggregate, values from the
GHGI. Equipment count estimates from other sources, such as the 2016 ICR and the API survey
data, are used to capture important dimensions of heterogeneity in equipment across different
types of well sites not captured by the GHGI (e.g., low producing versus non-low producing
wells, single well sites versus multi-well sites, etc.). Having described the use of the 2016 ICR
data in the preceeding section, we now describe our use of the API survey data and other
elements of the calibration procedure.
Since the 2016 ICR data lacks information on process heaters and heater-treaters at well
sites, we use the API survey data to fill in the gap. The first step is to determine the proportion of
well sites that have exactly one separator, compressor, or dehydrator and either a process heater
186
-------
or a heater-treater, since the fugitive emissions monitoring requirements are different for single
well sites with one piece of major equipment and those with more than one.111 Within the API
survey data, there are 451 oil sites and 724 gas sites with exactly one separator, compressor, or
dehydrator. Of the oil sites, 67% have a heater-treater, while 40% of the gas sites have a process
heater.112 As a result, we shift 67% of oil sites, and 40% of gas sites, estimated to have one piece
of major equipment by our analysis of the 2016 ICR to the "more than one piece of major
equipment" category.113 The next step is to come up with an initial estimate of the number of
heater-treaters and process heaters per site. For API survey sites with exactly one separator,
compressor, or dehydrator and at least one heater-treater, there are an average of 1.08 heater-
treaters per well at single wellhead oil sites, and 0.73 heater-treaters per well at multi-wellhead
oil sites. For sites with more than one separator, compressor, and dehydrator, there are an
average of 0.88 heater-treaters per well at single wellhead oil sites, and 0.54 heater-treaters per
well at multi-wellhead oil sites. For sites with exactly one separator, compressor, or dehydrator
and at least one process heater, there is an average of one heater-treaters per well at gas sites. For
sites with more than one separator, compressor, and dehydrator, there are an average of 0.28
process heaters per well at single wellhead gas sites, and 0.02 process heaters per well at multi-
wellhead gas sites.
Once equipment count averages have been calculated for separators, compressors
dehydrators, heaters, and heater-treaters (stratified by well type, site production level, site well
count bin, and site equipment category), those values are merged into the base year well site
group activity data. The total, nationwide counts of equipment implied by our application of the
2016 ICR/API survey data to the activity data are then calculated. Finally, the per-well
equipment counts are scaled, uniformly across all site production level, site well count bin, and
site equipment category permutations, by the ratio of the equipment counts implied by the
aggregate GHGI per-well equipment factors applied to our base year activity data to the
111 Due to a lack of sufficient data, we assume that sites without separators, compressors, and dehydrators represent
wellhead-only sites, and therefore do not have heater-treaters or process heaters either.
112 We restrict the API survey sample to exclude sites in Alaska, which are much larger than most of the the other
sites in the sample. We also consider heater-treaters only at oil sites and process heaters only at gas sites to maintain
consistency with the GHGI, which attributes all process heaters to gas production and all heater-treaters to oil
production.
113 The API survey data does not distinguish sites by production level and the number of multi-well sites with
exactly one separator, compressor, or dehydrator is small, so we apply these proportions uniformly across all site
types in the ICR summary proportions data.
187
-------
aggregate equipment counts implied by our application of the 2016 ICR/API survey data.114 This
ensures that our aggregate per-well equipment counts match the GHGI in 2019.
114 In math notation, the calibrated per-well equipment counts can be expressed as follows:
where t denotes well type, b denotes well count bin, c denotes equipment category, cal denotes a calibrated value,
GHGI denotes a value based on the GHGI, ENV denotes a value based on analysis of the base year Enverus well
data, ICR denotes a value based on analysis of the 2016 ICR and the API survey data, ICR /ENV denotes a value
based on the application of the analysis of the 2016 ICR and the API survey data to the analysis of the base year
Enverus well data, EqCnt denotes a per-well equipment count for separators, compressors, dehydrators, heater-
treaters, and process heaters, Wells denotes the total number of wells, and TotEqCnt denotes an aggregate
equipment count across the entire collection of sites.
188
-------
APPENDIX B SENSITIVITY ANALYSIS OF MONETIZED CLIMATE
BENEFITS
In this Appendix, we present the results of a sensitivity analysis of the monetized climate
benefits of this proposal using estimates of the social cost of methane (SC-CH4) newly developed
by EPA. As described below, these new SC-CH4 estimates are based on recent research
addressing recommendations for updating estimates of the SC-GHG from the National
Academies of Sciencies, Engineering, and Medicine (National Academies, 2017). Section B.l
describes the methodological updates underlying the new estimates relative to the interim SC-
CH4 estimates used in Chapter 3 of this RIA. Section B.2 presents the monetized climate benefits
under the proposed NSPS OOOOb and EG 0000c using the updated SC-CH4 estimates.
B.l Updated Estimates of the Social Cost of Methane
As discussed in Section 3.2 of this RIA, in January 2017 the National Academies
published a final report, Valuing Climate Damages: Updating Estimation of the Social Cost of
Carbon Dioxide, that responded to a U.S. Government-requested review of the IWG's SC-CO2
estimates and request for advice on approaching future updates to ensure that the estimates
continue to reflect the best available science and methodologies. The National Academies' final
report provided a comprehensive set of recommendations for updating estimates of the social
cost of carbon, including specific criteria for future updates to the estimates, a modeling
framework to satisfy the specified criteria, and both near-term updates and longer-term research
needs for multiple components of the estimation process (National Academies, 2017). Since that
time, the research community has made considerable progress in developing new data and
methods for bringing SC-GHG estimates closer to the current frontier of climate science and
economics and addressing many of the National Academies' (2017) recommendations. In this
Appendix the EPA uses new SC-CH4 estimates derived from the recent advances in the scientific
literature on climate change and its economic impacts to conduct a sensitivity analysis of the
climate benefits of this proposed rulemaking.
The SC-CH4 estimates used in this sensitivity analysis are taken from EPA's September
2022 Report on the Social Cost of Greenhouse Gases: Estimates Incorporating Recent Scientific
Advances (EPA 2022, external review draft), which has been included as supporting material for
this RIA in the docket. The SC-CH4 estimates reflect numerous methodological updates relative
189
-------
to the SC-CH4 estimates used in Section 3.2 of this RIA. All the modeling inputs and updates are
explained at length in EPA (2022) and are briefly summarized here. Consistent with the National
Academies (2017) near-term updating recommendations, the SC-CH4 values were estimated
using a modular updating approach in which the methodology underlying each of the four
components, or modules, of the SC-GHG estimation process — socioeconomics and emissions,
climate, damages, and discounting — is updated by drawing on the latest research and expertise
from the scientific disciplines relevant to that component. The socioeconomic and emissions
module relies on a new set of probabilistic projections for population, income, and GHG
emissions developed under the Resources for the Future Social Cost of Carbon Initiative
(Rennert et al., 2022a). The climate module relies on the Finite Amplitude Impulse Response
(FaIR) model (Millar et al., 2017; Smith et al., 2018, 2021), a widely used simple Earth system
model recommended by the National Academies, which captures the relationships between GHG
emissions, atmospheric GHG concentrations, and global mean surface temperature change. The
socioeconomic projections and outputs of the climate module are used as inputs to the damage
module to estimate monetized future damages from temperature change. Based on a review of
available studies and approaches to damage function estimation, the damages module is
composed of three separate damage functions. They are:
1. a subnational-scale, sectoral damage function estimation (based on the Data-driven
Spatial Climate Impact Model (DSCIM) developed by the Climate Impact Lab (CIL
2022; Carleton et al., 2022; Rode et al., 2021)),
2. a country-scale, sectoral damage function estimation (based on the Greenhouse Gas
Impact Value Estimator (GIVE) model developed under RFF's Social Cost of Carbon
Initiative (Rennert et al., 2022b)), and
3. a meta-analysis-based global damage function estimation (based on Howard and Sterner
(2017)).
Finally, in the discounting module the projected stream of future climate damages are
discounted back to the year of emissions using a set of calibrated dynamic discount rates
following the Newell et al. (2022) calibration approach, as applied in Rennert et al. (2022a,
2022b). This approach uses the Ramsey (1928) discounting formula in which the parameters are
calibrated such that the decline in the certainty-equivalent discount rate schedule matches the
latest empirical evidence on interest rate uncertainty estimated by Bauer and Rudebusch (2020,
190
-------
2021) and such that the average of the certainty-equivalent discount rate over the first decade
matches a specified near-term consumption rate of interest. Uncertainty in the starting rate is
addressed by using three near-term target rates — 1.5, 2.0, and 2.5 percent — based on multiple
lines of evidence on observed interest rate data. This approach results in three dynamic discount
rate paths and is consistent with the National Academies (2017) recommendation to use three
sets of Ramsey parameters that reflect a range of near-term certainty equivalent discount rates
consistent with theory and empirical evidence on consumption rate uncertainty. Finally, the value
of risk aversion associated with marginal GHG emissions is explicitly incorporated into the
modeling following the economic literature.
The estimation process outlined above generates nine separate distributions of the SC-
CH4 for a given year, the product of three damage modules and three near-term target discount
rates. As described in EPA (2022), to produce a range of estimates that reflects the uncertainty in
the estimation exercise while providing a manageable number of estimates for policy analysis,
the multiple lines of evidence on damage modules was combined by averaging the results across
the three damage module specifications. The resulting SC-CH4 estimates for each year of the
analysis period for this proposed rule are presented in Table B-l. Comparing the estimates
presented in Table B-l with the average SC-CH4 estimates resulting from the constant discount
rates presented in Table 3-3, for all emissions years the range of the updated estimates is higher
in magnitude than the IWG's recommended interim SC-CH4 estimates. For example, for
emissions occurring in 2035, the updated SC-CH4 values range from $2,300 to $3,600 per metric
ton CH4 (in 2019 dollars), whereas the average SC-CH4 values using the constant discount rates
presented in Table 3-3 range from $1,100 to $2,800 per metric ton CH4 (in 2019 dollars).
191
-------
Table B-l Updated Estimates of the Social Cost of CH4, 2023-2035 (in 2019$ per metric
ton CH4)
Near-Term Ramsey Discount Rate
Year
2.5%
2.0%
1.5%
2023
$1,400
$1,900
$2,500
2024
$1,500
$1,900
$2,600
2025
$1,600
$2,000
$2,700
2026
$1,600
$2,100
$2,800
2027
$1,700
$2,200
$2,900
2028
$1,800
$2,200
$3,000
2029
$1,800
$2,300
$3,000
2030
$1,900
$2,400
$3,100
2031
$2,000
$2,500
$3,200
2032
$2,100
$2,500
$3,300
2033
$2,100
$2,600
$3,400
2034
$2,200
$2,700
$3,500
2035
$2,300
$2,800
$3,600
Source: EPA (2022).
Note: The values are stated in $/metric ton CH4 and vary depending on the year of CH4 emissions. This table
displays the values rounded to two significant figures; the annual unrounded values used in the calculations in this
RIA are available in Table A.4.1 of EPA (2022) and at: www.epa.gov/environmental-economics/scghg. These SC-
CH4 values are adjusted for inflation to 2019 dollars using the annual GDP Implicit Price Deflator values in the U.S.
Bureau of Economic Analysis' (BEA) NIPA Table 1.1.9 (U.S. BEA, 2021).
The methodological updates underlying the SC-CH4 estimates presented in Table B-l
reflect conservative methodological choices, and, given both these choices and the numerous
categories of damages that are not currently quantified and other model limitations, likely
underestimate the marginal damages from methane emissions. Detailed discussion of omitted
categories of climate impacts and associated damages and other modeling limitations is provided
in EPA (2022).
As a member of the IWG, EPA will continue to participate in the IWG's work under E.O.
13990. EPA will also continue to independently review developments in the literature, including
more robust methodologies for estimating the magnitude of the various direct and indirect
damages from GHG emissions, and look for opportunities to further improve SC-GHG
estimation going forward. Information about the forthcoming peer review of the EPA report
detailing the SC-CH4 estimates presented above can be found at: www.epa.gov/environmental-
economics/scghg.
192
-------
B.2 Results of the Climate Benefits Sensitivity Analysis
Table B-2 presents the undiscounted annual monetized climate benefits under the
proposed NSPS OOOOb and EG 0000c using the updated SC-CH4 estimates. Projected
methane emissions reductions each year are multiplied by the SC-CH4 estimate from Table B-l
above for that year. Table B-3 shows the annual monetized climate benefits discounted back to
2021 and the PV and the EAV for the 2023-2035 period under each near-term Ramsey discount
rate. In Table B-3, the future benefits in each column are discounted back to 2021 using the
corresponding near-term discount rate.115
Table B-2 Undiscounted Monetized Climate Benefits Using Updated SC-CH4 Estimates
under the NSPS OOOOb and EG OOOOc Option, 2023-2035 (millions, 2019$)a
Near-Term Ramsey Discount Rate
Year
2.5%
2.0%
1.5%
2023
$190
$240
$330
2024
$300
$390
$530
2025
$430
$550
$750
2026
$5,200
$6,500
$8,800
2027
$5,400
$6,800
$9,100
2028
$5,600
$7,000
$9,400
2029
$5,800
$7,300
$9,600
2030
$6,000
$7,500
$9,900
2031
$6,300
$7,800
$10,000
2032
$6,500
$8,100
$11,000
2033
$6,800
$8,400
$11,000
2034
$7,100
$8,700
$11,000
2035
$7,300
$9,000
$12,000
a Climate benefits are based on changes (reductions) in CH4 emissions and are calculated using updated estimates of
the SC-CH4 provided in EPA (2022).
115 Given the relatively short time period of analysis for this proposed rule, the error associated with discounting
future benefits back to 2021 using a constant discount rate instead of using the year specific certainty-equivalent
discount factor will be small (i.e., resulting in a less than 1% underestimate of the present value of the 2023-2035
emission reductions). See EPA (2022) for more discussion.
193
-------
Table B-3 Discounted Monetized Climate Benefits Using Updated SC-CH4 Estimates
under the Proposed NSPS OOOOb and EG OOOOc Option, 2023-2035 (millions, 2019$)a
Discounted back to 2021
Near-Term Ramsey Discount Rate
Year
2.5%
2.0%
1.5%
2023
$180
$230
$320
2024
$280
$370
$500
2025
$390
$510
$700
2026
$4,600
$5,900
$8,200
2027
$4,600
$6,000
$8,300
2028
$4,700
$6,100
$8,400
2029
$4,800
$6,200
$8,600
2030
$4,800
$6,300
$8,700
2031
$4,900
$6,400
$8,800
2032
$5,000
$6,500
$9,000
2033
$5,100
$6,600
$9,100
2034
$5,100
$6,700
$9,300
2035
$5,200
$6,800
$9,400
PV
$50,000
$65,000
$89,000
EAV
$4,500
$5,700
$7,600
a Climate benefits are based on changes (reductions) in CH4 emissions and are calculated using updated estimates of
the SC-CH4 provided in EPA (2022).
Note: Totals may not appear to add correctly due to rounding.
Comparing the monetized climate benefits presented in Tables B-2 and B-3 with the
results presented in Tables 3-4 and 3-5 using the average SC-CH4 estimates under each discount
rate, for all emissions years the range of the climate benefits resulting from this sensitivity
analysis is higher in magnitude than the monetized climate benefits using the IWG's
recommended interim SC-CH4 estimates.116 For example, this sensitivity analysis projects
116 The disbenefit of the secondary CO2 impacts of the rule discussed in Section 3.7 will also be somewhat larger if
using the updated SC-CO2 estimates presented in EPA (2022). However, the estimated disbenefits associated with
destroying one metric ton of methane through combustion of emissions at oil and gas sites and releasing the CO2
emissions in 2023 instead of being released in the future via the methane oxidation process are still found to be
small relative to the benefits of flaring. Updating the illustrative analysis provided in Section 3.7 of this RIA with
the SC-CO2 values in EPA (2022), we find the disbenefit is estimated to be about $78 per metric ton CH4 (based on
average SC-CO2 using the 2% near-term Ramsey discount rate) or about 4 percent of the SC-CH4 estimate per
metric ton for 2023. The analogous estimate for 2035 is $115 per metric ton CH4 or about 4 percent of the SC-CH4
estimates per metric ton for 2035. As discussed in Section 3.7, given the uncertainties related to estimating net
secondary emissions effects and that the EPA has not yet received appropriate input and review on some aspects of
these calculations, the EPA is not including monetized estimates of the impacts of small changes in the timing of
atmospheric CO2 concentration increases in the total benefits estimates of this sensitivity analysis presented in this
Appendix. Nevertheless, upon consideration of the updated SC-CO2 estimates presented in EPA (2022), EPA
continues to believe that the disbenefits of the secondary CO2 impacts will be minor compared to the rule's net
benefits. The EPA will will continue to follow the scientific literature on this topic and update its methodologies as
warranted.
194
-------
undiscounted monetized climate benefits of $7.3 billion to $12 billion (in 2019 dollars) by 2035,
whereas the undiscounted monetized climate benefits based on average SC-CH4 values in Table
3-4 range from $3.5 billion to $8.9 billion in 2035. The sensitivity analysis projects the PV and
EAV of monetized climate benefits over 2023-2035 using the central 2% near-term Ramsey
discount rate to be $65 billion and $5.7 billion, respectively, whereas the PV and EAV climate
benefits presented in Table 1-6 (using the interim SC-CH4 values under a constant 3% discount
rate) are $48 billion and $4.5 billion, respectively.
B.3 References
Bauer, M.D. and Rudebusch, G.D., 2020. Interest rates under falling stars. American Economic
Review, 110(5), pp. 13 16-54.
Bauer, M.D. and Rudebusch, G.D., 2021. The rising cost of climate change: evidence from the
bond market. The Review of Economics and Statistics, pp. 1-45.
https://doi.org/10.1 162/rest_a_01 109.
Carleton, T., A. Jina, M. Delgado, M. Greenstone, T. Houser, S. Hsiang, A. Hultgren, R. Kopp,
K. McCusker, I. Nath, J. Rising, A. Rode, HK Seo, A. Viaene, J. Yuan, and A. Zhang. 2022.
Valuing the Global Mortality Consequences of Climate Change Accounting for Adaptation
Costs and Benefits. Quarterly Journal of Economics, 137(4), pp. 2037-2105,
https://doi.org/10.1093/qje/qjac020.
Climate Impact Lab (CIL). 2022. Data-driven Spatial Climate Impact Model User Manual,
Version 092022-EPA. Available at: https://impactlab.org/research/dscim-user-manual-
version-092022-epa.
Howard, P., and T. Sterner. 2017. Few and Not So Far Between: A Meta-analysis of Climate
Damage Estimates. Environmental and Resource Economics 68:197-225.
Millar, R.J., Nicholls, Z.R., Friedlingstein, P. and Allen, M R., 2017. A modified impulse-
response representation of the global near-surface air temperature and atmospheric
concentration response to carbon dioxide emissions. Atmospheric Chemistry and
Physics, 77(11), pp. 7213-7228.
National Academies of Sciences, Engineering, and Medicine (National Academies). 2017.
Valuing Climate Damages: Updating Estimation of the Social Cost of Carbon Dioxide.
Washington, D.C.: National Academies Press.
Newell, R., W. Pizer, and B. Prest. 2022. A Discounting Rule for the Social Cost of Carbon.
Journal of the Association of Environmental and Resource Economists. Accepted version
available at: https://doi.org/10.1086/718145.
195
-------
Ramsey, F.P., 1928. A mathematical theory of saving. The Economic Journal, 35(152), pp.543-
559.
Rennert, K., B.C. Prest, W.A. Pizer, R.G. Newell, D. Anthoff, C. Kingdon, L. Rennels, R.
Cooke, A.E. Raftery, H. Sevcikova, and F. Errickson. 2022a. The Social Cost of Carbon:
Advances in Long-Term Probabilistic Projections of Population, GDP, Emissions, and
Discount Rates. Brookings Papers on Economic Activity. 2021 (2): 223-305.
doi: 10.1353/eca.2022.0003..
Rennert, K., F. Errickson, B. Prest, L. Rennels, R. Newell, W. Pizer, C. Kingdon, J. Wingenroth,
R. Cooke, B. Parthum, D. Smith, K. Cromar, D. Diaz, F. Moore, U. Miiller, R. Plevin, A.
Raftery, H. Sevcikova, H. Sheets, J. Stock, T. Tan, M. Watson, T. Wong, and D. Anthoff.
2022b. Comprehensive Evidence Implies a Higher Social Cost of CO:. Nature.
https://doi.org/! 0.1038/s41586-022-05224-9.
Rode, A, T. Carleton, M. Delgado, M. Greenstone, T. Houser, S. Hsiang, A. Hultgren, A. Jina,
R E. Kopp, K.E. McCusker, and 1. Nath. 2021. Estimating a social cost of carbon for global
energy consumption. Nature 598(7880): pp.308-314.
Smith, C.J., Forster, P.M., Allen, M., Leach, N., Millar, R.J., Passerello, G.A. and Regayre, L.A.,
2018. FAIR vl. 3: a simple emissions-based impulse response and carbon cycle
model. Geoscientific Model Development, 11(6), pp.2273-2297.
Smith, C., Nicholls, Z., Armour, K., Collins, W., Forster, P., Meinshausen, M., Palmer, M.,
Watanabe, M., The Earth's energy budget, climate feedbacks, and climate sensitivity -
Supplementary Material. In Climate Change 2021: The Physical Science Basis. Contribution
of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on
Climate Change [Masson-Delmotte, V., P. Zhai, A. Pi rani, S.L. Connors, C. Pean, S. Berger,
N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R.
Matthews, T.K. May cock, T. Waterfield, O. Yelekgi, R. Yu, and B. Zhou (eds.)]. Cambridge
University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 7SM-1 to
7SM-52.
U.S. Bureau of Economic Analysis (BEA). Implicit Price Deflators for Gross Domestic Product,
Table 1. 1.9. https://apps.bea.gov/iTable/index_nipa.cfm (accessed February 10, 2021).
U.S. Environmental Protection Agency (EPA). 2022. Report on the Social Cost of Greenhouse
Gases: Estimates Incorporating Recent Scientific Advances. September. Available at:
www.epa.gov/en vi ronmental -economi cs/scghg.
196
-------
APPENDIX C ILLUSTRATIVE SCREENING ANALYSIS OF MONETIZED VOC-
RELATED OZONE HEALTH BENEFITS
In this appendix, we present a supplementary screening analysis to estimate potential
health benefits from the changes in ozone concentrations resulting from VOC emissions
reductions under the proposed rule.117 As we describe in detail below, the distribution of the
change in VOC emissions are subject to significant uncertainties; for this reason, the estimated
benefits reported below should not be interpreted as a central estimate and thus are not reflected
in the calculated net benefits above. For this analysis, we apply a national benefit-per-ton
approach based on photochemical modeling with source apportionment paired with the
Environmental Benefits Mapping and Analysis Program (BenMAP) for years between 2023 and
2035 using an April-September average of 8-hr daily maximum (MDA8) ozone metric.
C.l Air Quality Modeling Simulations
The photochemical model simulations are described in detail in U.S. EPA (2021a) and
are summarized briefly in this section. The air quality modeling used in this analysis included
annual model simulations for the year 2017. The photochemical modeling results for 2017, in
conjunction with modeling to characterize the air quality impacts from groups of emissions
sources (i.e., source apportionment modeling) and expected emissions changes due to this
proposed rule, were used to estimate ozone benefits expected from this proposed rule in the years
2023-2035.
The air quality model simulations (i.e., model runs) were performed using the
Comprehensive Air Quality Model with Extensions (CAMx version 7.00) (Ramboll Environ,
2016). The CAMx nationwide modeling domain (i.e., the geographic area included in the
modeling) covers all lower 48 states plus adjacent portions of Canada and Mexico using a
horizontal grid resolution of 12x12 km shown in Figure C-l.
117 Note that this illustrative analysis does not consider the health and welfare benefits from reducing tropospheric
ozone production resulting from CH4 emissions, which are also not included in estimates of the social cost of
methane.
197
-------
Figure C-l Air Quality Modeling Domain
C. 1,1 Ozone Model Performance
While U.S. EPA (2021a) provides an overview of model performance, we provide a more
detailed assessment here specifically focusing on ozone model performance relevant to the
metrics used in this analysis. In this section, we report CAMx model performance for the MDA8
ozone across all days in April-September. While regulatory analyses often focus on model
performance on high ozone days relevant to the NAAQS (U.S. EPA, 2018a), here we focus on
all days in April-September since the relevant ozone metrics used as inputs into BenMAP use
summertime seasonal averages. Model performance information is provided for each of the nine
National Oceanic and Atmospheric Administration (NOAA) climate regions in the contiguous
US, as shown in Figure C-2 and first described by Karl and Koss (1984).118
Table C-l provides a summary of model performance statistics by region. Normalized
Mean Bias was within ±10 percent in every region and within ±5 percent in the Northeast, Ohio
Valley, South, Southwest, and West regions. Across all monitoring sites, normalized mean bias
was -0.2 percent. Normalized mean error for modeled MDA8 ozone was less than ±20 percent in
every region except the Northwest where it was 21 percent. Correlation between the modeled
and observed MDA8 ozone values was 0.7 or greater in five of the nine regions (Northeast,
Upper Midwest, Southeast, South, and West). In the remaining four regions correlation was 0.69
in the Ohio Valley, 0.64 in the Northern Rockies and Plains, 0.46 in the Southwest, and 0.69 in
118 Figure obtained from https://www.ncdc.noaa.gov/monitoring-references/maps/us-climate-regions.php.
198
-------
the Northwest. Across the contiguous U.S. as a whole, the correlation between modeled and
measured MDA8 ozone was 0.72.
Figure C-3 displays modeled MDA8 normalized mean bias at individual monitoring sites.
This figure reveals that the model has slight overpredictions of mean April-September MDA8
ozone in the southeastern portion of the country and along the Pacific coast and slight
underpredictions in the northern and western portions of the country. Time series plots of the
modeled and observed MDA8 ozone and model performance statistics across the nine regions
were developed.119 Overall, the model closely captures day to day fluctuations in ozone
concentrations, although the model had a tendency to underpredict ozone in the earlier portion of
the ozone season (April and May) and overpredict in the later portion of the ozone season (July-
September) with mixed results in June. This model performance is within the range of other
ozone model applications, as reported in scientific studies (Emery et al., 2017; Simon, Baker, &
Phillips, 2012). Thus, the model performance results demonstrate the scientific credibility of our
2017 modeling platform. These results provide confidence in the ability of the modeling platform
to provide a reasonable projection of expected future year ozone concentrations and
contributions.
119 Memorandum. 2017 Time Series Plots Supporting the Regulatory Impact Analysis for the Proposed Standards of
Performance for New, Reconstructed, and Modified Sources and Emissions Guidelines for Existing Sources: Oil and
Natural Gas Sector Climate Review. Prepared by Heather Simon, AQAD/OAQPS/EPA. September 29, 2021.
199
-------
U.S. Climate Regions
I * Wft
¦¦I
SB
—
MA
-0E
MO
»S Bill
Figure C-2 Climate Regions Used to Summarize 2017 CAMx Model Performance for
Ozone
Table C-l Summary of 2017 CAMx MDA8 ozone model performance for all April-
September days
Mean
Mean
Number of
observed
modeled
Mean
RMS
Normalized
Normalized
Monitoring
MDA8
MDA8
bias
E
mean bias
mean error
Region
Sites
(ppb)
(ppb)
Correlation
(ppb)
(ppb)
(%)
(%)
Northeast
189
42.4
42.5
0.71
0.1
9.1
0.3
17.2
Upper
Midwest
107
42.5
39.1
0.70
-3.4
9.1
-8.0
17.2
Ohio
Valley
236
45.4
45.8
0.69
0.4
8.3
0.8
14.7
Southeast
177
40.2
43.4
0.76
3.3
8.8
8.2
17.7
South
145
42.0
43.5
0.73
1.5
8.8
3.6
16.7
Northern
Rockies
55
46.8
43.1
0.64
-3.7
9.3
-7.9
16.4
and Plains
Southwest
117
54.3
52.5
0.46
-1.8
10.2
-3.4
15.5
Northwest
28
41.4
44.0
0.69
2.7
12.4
6.4
21.0
West
200
51.6
50.1
0.74
-1.5
10.3
-2.9
16.1
All
1258
45.4
45.3
0.72
-0.1
9.3
-0.2
16.4
200
-------
03_8hrmax NMB (%) for run CAMx_2017icenginesSO4PRI_12US2 for 20170401 to 20170930
• AQS Daily
Figure C-3 Map of 2017 CAMx MDA8 Normalized Mean Bias (%) for April-September
at all U.S. monitoring sites in the model domain
C. 1.2 Source Apportionment Modeling
The contribution of specific emissions sources to ozone in the 2017 modeled case were
tracked using a tool called "source apportionment." In general, source apportionment modeling
quantifies the air quality concentrations formed from individual, user-defined groups of
emissions sources or "tags." These source tags are tracked through the transport, dispersion,
chemical transformation, and deposition processes within the model to obtain hourly gridded
contributions from the emissions in each individual tag to hourly modeled concentrations of
ozone.
For this analysis ozone contributions were modeled using the Ozone Source
Apportionment Technique (OSAT) tool. In this modeling, VOC emissions from oil and natural
gas operations were tagged separately for three regions of the U.S. regions. The model-produced
gridded hourly ozone contributions from emissions from each of the source tags which we
aggregated up to an ozone metric relevant to recent health studies (i.e., the April-September
average of the MDA8 ozone concentration). The April-September average of the MDA8 ozone
contributions from each regional oil and natural gas tag were summed to produce a spatial field
201
-------
representing national oil and natural gas VOC contributions to ozone across the United States
(Figure C-4).
Apr-Sep MDA8 03
1
159 239 318
Min = O.OOE+O at (1,1), Max = 1.885 at (145,139)
Figure C-4 Contributions of 2017 Oil and Natural Gas VOC Emissions across the
Contiguous U.S. to the April-September Average of MDA8 Ozone.
C.2 Applying Modeling Outputs to Quantify a National VOC-Ozone Benefit Per-Ton
Value
Following an approach detailed in the RIA and TSD for the Revised Cross-State Update,
we estimated the number and value of ozone-attributable premature deaths and illnesses for the
purposes of calculating a national ozone VOC benefit per-ton value for the proposed policy
scenario (U.S. EPA, 2021f, 2021g).
The EPA historically has used evidence reported in the Integrated Science Assessment
(ISA) for the most recent NAAQS review to inform its approach for quantifying air pollution-
attributable health, welfare, and environmental impacts associated with that pollutant. The ISA
202
-------
synthesizes the toxicological, clinical and epidemiological evidence to determine whether each
pollutant is causally related to an array of adverse human health outcomes associated with either
short-term (hours to less than one month) or long-term (one month to years) exposure; for each
outcome, the ISA reports this relationship to be causal, likely to be causal, suggestive of a causal
relationship, inadequate to infer a causal relationship, or not likely to be a causal. We estimate
the incidence of air pollution-attributable premature deaths and illnesses using methods
reflecting evidence reported in the 2020 Ozone ISA (U.S. EPA, 2020a) and accounting for
recommendations from the Science Advisory Board. When updating each health endpoint the
EPA considered: (1) the extent to which there exists a causal relationship between that pollutant
and the adverse effect; (2) whether suitable epidemiologic studies exist to support quantifying
health impacts; (3) and whether robust economic approaches are available for estimating the
value of the impact of reducing human exposure to the pollutant. Detailed descriptions of these
updates are available in the TSD for the Final Revised Cross-State Air Pollution Rule for the
2008 Ozone NAAQS Update titled Estimating PM2.5- and Ozone-Attributable Health Benefits
(U.S. EPA, 202lh).
In brief, we used the environmental Benefits Mapping and Analysis Program—
Community Edition (BenMAP-CE) to quantify estimated counts of premature deaths and
illnesses attributable to summer season average ozone concentrations using the modeled surface
described above (Section C.1.2). We calculate effects using a health impact function, which
combines information regarding the: concentration-response relationship between air quality
changes and the risk of a given adverse outcome; population exposed to the air quality change;
baseline rate of death or disease in that population; and air pollution concentration to which the
population is exposed. These quantified health impacts were then used to estimate the economic
value of these ozone-attributable effects as described below. For this supplemental proposal, we
quantified counts of premature deaths and illnesses by multiplying an incidence per ton against
an updated estimate of emissions described in Section 2.3. Modeled air quality changes were not
available.
We performed BenMAP-CE analyses for each year between 2023 and 2035, using the
single model surface described above, but accounting for the change in population size, baseline
death rates and income growth in each future year. We next divided the sum of the monetized
ozone benefits in each year the April-September VOC emissions associated with the oil and
203
-------
natural gas source apportionment tags in the 2017 CAMx modeling to determine a benefit per
ton value for each year from 2023-2035. Emissions totals for the oil and natural gas sector used
in the contribution modeling are reported in U.S. EPA (2021a). Finally, the benefit per ton values
were multiplied by the expected national VOC emissions changes in each year, as reported in
Section 2.3. Since values reported in Section 2 were annual totals, we assume the emissions
changes are distributed evenly across months of the year and divide emissions changes by two to
estimate the April-September VOC changes expected from this supplemental proposed rule.
C.3 Uncertainties and Limitations of Air Quality Methodology
The approach applied in this screening analysis is consistent with how air quality impacts
have been estimated in past regulatory actions (U.S. EPA, 2019b, 2021f). However, in this
section we acknowledge and discuss several limitations.
First, the 2017 modeled ozone concentrations are subject to uncertainty. While all models
have some level of inherent uncertainty in their formulation and inputs, evaluation of the model
outputs against ambient measurements shows that ozone model performance is within the range
of model performance reported from photochemical modeling studies in the literature (Emery et
al., 2017; Simon et al., 2012) and is adequate for estimating ozone impacts of VOC emissions for
the purpose of this rulemaking.
In any complex analysis using estimated parameters and inputs from a variety of models,
there are likely to be many sources of uncertainty. This analysis is no exception. This analysis
includes many data sources as inputs, including emissions inventories, air quality data from
models (with their associated parameters and inputs), population data, population estimates,
health effect estimates from epidemiology studies, economic data for monetizing benefits, and
assumptions regarding the future state of the world (i.e., regulations, technology, and human
behavior). Each of these inputs are uncertain and generate uncertainty in the benefits estimate.
When the uncertainties from each stage of the analysis are compounded, even small uncertainties
can have large effects on the total quantified benefits. Therefore, the estimates of annual benefits
should be viewed as representative of the magnitude of benefits expected, rather than the actual
benefits that would occur every year.
204
-------
Because regulatory health impacts are distributed based on the degree to which housing
and work locations overlap geographically with areas where atmospheric concentrations of
pollutants change, it is difficult to fully know the distributional impacts of a rule. Air quality
models provide some information on changes in air pollution concentrations induced by
regulation, but it may be difficult to identify the characteristics of populations in those affected
areas, as well as to perform high-resolution air quality modeling nationwide. Furthermore, the
overall distribution of health benefits will depend on whether and how households engage in
averting behaviors in response to changes in air quality, e.g., by moving or changing the amount
of time spent outside (Sieg, Smith, Banzhaf, & Walsh, 2004).
Another limitation of the methodology is that it treats the response of ozone benefits to
changes in emissions from the tagged sources as linear. For instance, the benefits associated with
a 10 percent national change in oil and natural gas VOC emissions would be estimated to be
twice as large as the benefits associated with a 5 percent change in nation oil and natural gas
VOC emissions. The methodology therefore does not account for 1) any potential nonlinear
responses of ozone atmospheric chemistry to emissions changes and 2) any departure from
linearity that may occur in the estimated ozone-attributable health effects resulting from large
changes in ozone exposures. We note that the emissions changes between scenarios are relatively
small compared to 2017 emissions totals from all sources. Previous studies have shown that air
pollutant concentrations generally respond linearly to small emissions changes of up to 30
percent (Cohan, Hakami, Hu, & Russell, 2005; Cohan & Napelenok, 2011; Dunker, Yarwood,
Ortmann, & Wilson, 2002; Koo, Dunker, & Yarwood, 2007; Napelenok, Cohan, Hu, & Russell,
2006; Zavala, Lei, Molina, & Molina, 2009) and that linear scaling from source apportionment
can do a reasonable job of representing impacts of 100 percent of emissions from individual
sources (Baker & Kelly, 2014). Additionally, past studies have shown that ozone responds more
linearly to changes in VOC emissions than changes in NOx emissions (Hakami, Odman, &
Russell, 2003; Hakami, Odman, & Russell, 2004). Therefore, it is reasonable to expect that the
ozone benefits from expected VOC emissions changes from this proposed rule can be adequately
represented using this this linear assumption.
A final limitation is that the source apportionment ozone contributions reflect the spatial
and temporal distribution of the emissions from each source tag in the 2017 modeled case. The
representation of the spatial patterns of ozone contributions are important because benefits
205
-------
calculations depend on the spatial patterns of ozone changes in relationship to spatial distribution
of population and health incidence values. While we accounted for changes the size of the
population, baseline rates of death and income, we assume the spatial pattern of oil and natural
gas VOC contributions to ozone remain constant at 2017 levels. Thus, the current methodology
does not allow us to represent any expected changes in the spatial patterns of ozone that could
result from changes in oil and natural gas emissions patterns in future years or from spatially
heterogeneous emissions changes resulting from this supplemental proposed rule. For instance,
the method does not account for the possibility that new sources would change the spatial
distribution of oil and natural gas VOC emissions. In addition, the method does not account for
any changes in spatial patterns of ozone that would result from spatially varying emissions
change which could result from differing impacts of this proposed rule in locations with existing
state regulations. For instance, in Section 2 we describe the impact of existing regulations in
Colorado and California. Due to the stringency of current on-the-books oil and natural gas
regulations in these and other states, we do not expect large impacts from this rule of VOC
emissions in those states. We note specifically that Figure 4-2 depicts that oil and natural gas
VOC contributions to ozone are large in Colorado compared to other parts of the contiguous US.
In addition, Figure 4-2 shows that there are some modeled oil and natural gas VOC contributions
to ozone in densely populated southern California. Since VOC emissions impacts from this rule
are calculated at a national level, at this time we do not have more refined information which
could be used to spatially vary the response of ozone impacts to proposed VOC emissions
changes. We also note that while we have identified existing state regulations in California and
Colorado, we have not characterized the impacts of state regulations from other states on VOC
emissions impacts or associated ozone benefits nor have we characterized how spatially
heterogeneous emissions changes due to other factors would impact the quantified benefits.
206
-------
Table C-2 Estimated Avoided Ozone-Related Premature Respiratory Mortality and
Illnesses for the Proposed NSPS OOOOb and EG OOOOc Option in 2026a'b
Avoided premature respiratory mortality
Proposed NSPS OOOOb
and EG OOOOc
Long-term exposure
Turner et al. (2016)
73
Short-term exposure
Katsouyanni et al. (2009)b and Zanobetti et al. (2008)c d
pooled
3.3
Avoided respiratory morbidity effects
Long-term exposure
Asthma onsetd
620
Allergic rhinitis symptomsf e
3,500
Hospital admissions—respiratory13
8.4
ED visits—respiratory®
190
Short-term exposure
Asthma symptoms®
120,000
Minor restricted-activity daysb
53,000
School absence daysch
40,000
a Values rounded to two significant figures.
b The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
Table C-3 Benefit Per Ton Estimates of Ozone-Attributable Premature Mortality and
Illnesses for the Proposal in 2026
Benefit Per Ton of Reducing VOC
from the Oil and Natural Gas
Sector
Short-term mortality and morbidity health effects (discounted at 3%)
$230
Short-term mortality and morbidity health effects (discounted at 7%)
$210
Long-term mortality and morbidity health effects (discounted at 3%)
$1,800
Long-term mortality and morbidity health effects (discounted at 7%)
$1,600
Table C-4 Estimated Discounted Economic Value of Ozone-Attributable Premature
Mortality and Illnesses under the Proposed NSPS OOOOb and EG OOOOc Option, 2023-
2035 (million 2019$)a d
Proposed NSPS OOOOb and EG OOOOc Option
Year
3% Discount Rate
7% Discount Rate
2023
$6.8b to $51°
$6.0b to $46°
2024
$10b to $78°
$9.1b to $70°
2025
$14b to $110°
$12b to $96°
2026
$110b to $830°
$95b to $750°
2027
$110b to $860°
$97b to $770°
2028
$110b to $870°
$99b to $780°
2029
$110b to $900°
$100b to $800°
2030
$120b to $930°
$100b to $830°
2031
$120b to $950°
$110b to $850°
2032
$120b to $980°
$110b to $880°
2033
$120b to $990°
$110b to $890°
2034
$120b to $1,000°
$110b to $910°
207
-------
2035
$130b to $1,000°
$110b to $940°
a Values rounded to two significant figures.
b Includes ozone mortality estimated using the pooled Katsouyanni et al. (2009) and Zanobetti and Schwartz (2008)
short-term risk estimates.
0 Includes ozone mortality estimated using the Turner et al. (2016) long-term risk estimate.
d The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
Table C-5 Stream of Human Health Benefits under the Proposed NSPS OOOOb and
EG OOOOc Option, 2023-2035: Monetized Benefits Quantified as Sum of Avoided
Morbidity Health Effects and Avoided Long-term Ozone Mortality (discounted at 3
percent to 2021; million 2019$)a'b
Year
Proposed NSPS OOOOb and EG OOOOc Option
2023
$51
2024
$78
2025
$110
2026
$830
2027
$860
2028
$870
2029
$900
2030
$930
2031
$950
2032
$980
2033
$990
2034
$1,000
2035
$1,000
Present Value (PV)
$7,200
Equivalent Annualized Value (EAV)
$680
a Benefits calculation includes ozone-related morbidity effects and avoided ozone-attributable deaths quantified
using the Turner et al. (2016) long-term risk estimate.
b The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
208
-------
Table C-6 Stream of Human Health Benefits under the Proposed NSPS OOOOb and
EG OOOOc Option, 2023-2035: Monetized Benefits Quantified as Sum of Avoided
Morbidity Health Effects and Avoided Long-term Ozone Mortality (discounted at 7
percent to 2021; million 2019$)a'b
Year
Proposed NSPS OOOOb and EG OOOOc
2023
$46
2024
$70
2025
$96
2026
$750
2027
$770
2028
$780
2029
$800
2030
$830
2031
$850
2032
$880
2033
$890
2034
$910
2035
$940
Present Value (PV)
$4,600
Equivalent Annualized Value (EAV)
$550
a Benefits calculated as value of avoided ozone-attributable deaths (quantified using a concentration-response
relationship from the Turner et al. (2016) study and ozone-related morbidity effects).
b The EG OOOOc regulates emissions of methane. Additional benefits to the regulation result from associated
reductions in VOC emissions.
C.4 References
Baker, K. R., & Kelly, J. T. (2014). Single source impacts estimated with photochemical model
source sensitivity and apportionment approaches. Atmospheric Environment, 96, 266-
274.
Cohan, D. S., Hakami, A., Hu, Y., & Russell, A. G. (2005). Nonlinear response of ozone to
emissions: Source apportionment and sensitivity analysis. Environmental Science &
Technology, 39(17), 6739-6748.
Cohan, D. S., & Napelenok, S. L. (2011). Air quality response modeling for decision support.
Atmosphere, 2(3), 407-425.
Dunker, A. M., Yarwood, G., Ortmann, J. P., & Wilson, G. M. (2002). The decoupled direct
method for sensitivity analysis in a three-dimensional air quality model implementation,
accuracy, and efficiency. Environmental Science & Technology, 36(13), 2965-2976.
Emery, C., Liu, Z., Russell, A. G., Odman, M. T., Yarwood, G., & Kumar, N. (2017).
Recommendations on statistics and benchmarks to assess photochemical model
performance. Journal of the Air & Waste Management Association, 67(5), 582-598.
209
-------
Hakami, A., Odman, M. T., & Russell, A. G. (2003). High-order, direct sensitivity analysis of
multidimensional air quality models. Environmental Science & Technology, 37(11),
2442-2452.
Hakami, A., Odman, M. T., & Russell, A. G. (2004). Nonlinearity in atmospheric response: A
direct sensitivity analysis approach. Journal of Geophysical Research: Atmospheres,
109(D\5).
Karl, T., & Koss, W. J. (1984). Regional and national monthly, seasonal, and annual temperature
weighted by area, 1895-1983.
Katsouyanni, K., Samet, J. M., Anderson, H. R., Atkinson, R., Le Tertre, A., Medina, S., . . .
Committee, H. E. I. H. R. (2009). Air pollution and health: a European and North
American approach (APHENA). Res Rep Health I
-------
U. S. EPA. (2019). Regulatory Impact Analysis for the Repeal of the Clean Power Plan, and the
Emission Guidelines for Greenhouse Gas Emissions from Existing Electric Utility
Generating Units. (EPA-452/R-19-003). Research Triangle Park, NC: U.S.
Environmental Protection Agency, Office of Air Quality Planning and Standards, Health
and Environmental Impact Division. Available at:
https://www.epa.gov/sites/production/files/2019-
06/documents/utilities_ria_final_cpp_repeal_and_ace_2019-06.pdf.
U.S. EPA. (2020). Integrated Science Assessment for Ozone and Related Photochemical
Oxidants (FinalReport). (EPA/600/R-20/012). Washington, DC: U.S. Environmental
Protection Agency. Available at: https://www.epa.gov/isa/integrated-science-assessment-
i sa-ozone-and-rel ated-photochemi cal -oxi dants.
U.S. EPA. (2021a). 2017 National Emission Inventory Based Photochemical Modeling for
Sector Specific Air Quality Assessments (EPA-454-R-21-005). Retrieved from Research
Triangle Park, NC: https://www.epa.gov/system/files/documents/2021-08/epa-454-r-21-
005.pdf.
U.S. EPA. (2021b). Regulatory Impact Analysis for the Final Revised Cross-State Air Pollution
Rule (CSAPR) Update for the 2008 Ozone NAAQS (EPA-452-R-21-002). Retrieved from
Research Triangle Park, NC: https://www.epa.gov/csapr/revised-cross-state-air-pollution-
rule-update.
U.S. EPA. (2021c). Technical Support Document (TSD) for the Final Revised Cross-State Air
Pollution Rule Update for the 2008 Ozone Season NAAQS: Estimating PM2.5- and
Ozone-Attributable Health Benefits (Docket ID No. EPA-HQ-OAR-2020-0272).
Retrieved from Research Triangle Park, NC: https://www.epa.gov/csapr/revised-cross-
state-air-pollution-rule-update.
U.S. EPA. (2021d). Technical Support Document (TSD) for the Final Revised Cross-State Air
Pollution Rule Update for the 2008 Ozone Season NAAQS: Estimating PM2.5- and
Ozone-Attributable Health Benefits. (EPA-HQ-OAR-2020-0272). Durham, NC: U.S.
Environmental Protection Agency. Available at:
https://www.epa.gov/sites/default/files/2021-03/documents/estimating_pm2.5-
_and_ozone-attributable_health_benefits_tsd.pdf.
Zanobetti, A., & Schwartz, J. (2008). Mortality displacement in the association of ozone with
mortality: an analysis of 48 cities in the United States. American Journal of Respiratory
and Critical Care Medicine, 777(2), 184-189. doi:10.1164/rccm.200706-8230C.
Zavala, M., Lei, W., Molina, M., & Molina, L. (2009). Modeled and observed ozone sensitivity
to mobile-source emissions in Mexico City. Atmospheric Chemistry and Physics, 9(1),
39-55.
211
-------
APPENDIX D FUGITIVE EMISSIONS ABATEMENT SIMULATION TOOLKIT
(FEAST) MEMO
This appendix summarizes fugitive emission modeling that the EPA conducted using the
Fugitive Emissions Abatement Simulation Toolkit (FEAST) version 3.1.120 FEAST is a
customizable, open-source modeling framework developed to evaluate the effectiveness of
different methane leak detection and repair (LDAR) programs at oil and gas facilities. Model
inputs include the number and type of emission source, a leak generation rate by emission source
type, a distribution of leak rates when a leak is generated, and the probability of the selected leak
detection method detecting a leak of a given size. Separate leak distributions were developed for
typical fugitive components (valves, pumps, connectors), storage vessels, and "large-emitters."
The EPA evaluated both component-level surveys, like those conducted using an optical gas
imaging (OGI) camera, and site-level surveys such as satellite or aerial surveys for various sizes
of facilities (model plants). Monte Carlo analyses were conducted using FEAST to assess
different LDAR programs (survey type, frequency, and method detection level) with and without
large-emitters included in the analysis.
The large-emitter distribution was developed to characterize emissions from component-
level leaks using an assumed distribution of emissions and leak generation rate, ranging from
very small to very large emissions, with the large emissions occuring less frequently. The large-
emitter distribution is not directly related to the definition of "super-emitter" included in this
proposal and the emissions reported for simulations including large-emitters cannot directly be
used to assess the emissions or emission reductions related to the proposed super-emitter
program. We found the modeled FEAST emissions are highly dependent on the assumed "leak
generation rate" (i.e., frequency) of the large-emitters. The EPA selected a reasonable central
tendency value for this parameter based on aerial studies conducted within the Permian basin and
conducted limited sensitivity analyses around this input parameter. The sensitivity analysis
results suggest that there are large uncertainties in the emissions contribution by large-emitters
based on the frequency at which large emission events occur. Available data suggest that the
assumed frequency of large emission events may vary significantly across different production
basins, so any national-level impact estimates that relied on extending the large-emitter emission
120 https://github.com/FEAST-SEDLab/FEAST PtE/tree/FEAST 3.
212
-------
estimates presented in this memorandum to basins beyond the Permian basin would be subject to
significant uncertainty.
213
-------
Memorandum
Jeff Coburn and Ricky Strott, RTI International
Karen Marsh, EPA/OAQPS
EPA Docket No. EPA-HQ-0AR-2020-0317
July 27, 2022
Modeling Fugitive Emissions from Production Sites Using FEAST
1. Purpose
This memorandum documents emission estimates for fugitive emission components at oil
and gas production sites using Fugitive Emissions Abatement Simulation Toolkit (FEAST). The
first objective of the modeling effort was to identify cost effective monitoring options when
using ground-based OGT. The second objective was to identify site-wide survey methods that
were equivalent to ground-based OGI.
2. Background
The Environmental Protection Agency (EPA) has used leak detection and repair (LDAR)
programs as a means to reduce emissions from leaking fugitive emission components for a wide
range of industry sectors. These LDAR programs traditionally used Method 21 of Appendix A-7
of 40 CFR part 60 (EPA Method 21), which uses a volatile organic monitor and a small pump to
draw air through sampling probe to the monitor. EPA Method 21 requires operators to slowly
traverse likely leak points, like a valve stem or flange, in attempts to identify areas of high
hydrocarbon concentration, indicating a leak. EPA Method 21 is labor and time intensive
because it requires the operator to physically locate each fugitive emission components and to on
an individual basis, traversing each component slowly enough to allow the monitor to respond to
a leak.
In 2008, EPA promulgated the alternative work practice, which allows owners and
operators to use optical gas imaging (OGI) cameras to see a hydrocarbon leak using a
sophisticated hand-held camera. These devices allow operators to scan for leaks more quickly
than when using EPA Method 21. OGI may not be able to detect small leaks that EPA Method
21 can detect, but by deploying OGI more often, OGT monitoring can achieve emission
reductions similar to EPA Method 21-based LDAR programs.1 EPA first promulgated OGI
monitoring as the fugitive emissions detection method for fugitive emission components in the
MRTI
INTERNATIONAL
FROM:
TO:
FOR:
DATE:
SUBJECT:
1 Although, OGI monitoring can achieve the same emission reductions as EPA Method 21 -based programs in some
cases, this may not always be true. For example, OGI cameras may not be able to image the compounds contained in
some gaseous emissions, in which case, an OGI-based program could not be equivalent to an EPA Method 21-based
program. It is necessary to evaluate the specific EPA Method 21-bascd program to determine whether equivalency
with an OGI-based program is possible.
214
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
new source performance standards (NSPS) for oil and gas sites ill 2016 (81 FR 35824, June 3,
2016; 40 CFR part 60, subpart OOOOa).
Technology continues to advance and there are a wide variety of different technologies
for fugitive emissions monitoring that are either available now or shortly in the future. These
include a network of continuous monitoring systems, tower-mounted laser-based monitoring
systems, automobile-mounted monitoring systems (or simply "mobile" systems), aircraft-
mounted monitoring systems (or simply "aerial'' systems), drone-mounted systems, or satellite-
based monitoring systems. Flowever, the current regulations require individual facilities to
request an alternative means of emissions limitation (AMEL) to use these alternative
technologies. The EPA is developing a streamlined process for use of these technologies by
specifying attributes (i.e., monitoring frequency and detection sensitivity) for these site-wide
monitoring techniques that are expected to achieve the same emission reductions as the ground-
based OGI monitoring that was determined to be the best system of emission reduction (BSER).
The FEAST model was specifically developed to allow users to model the emission reductions
achieved when deploying various site-wide or ground-based surveys.
3. FEAST Model Setup
3.1 FEAST Overview
FEAST is an open-source modeling framework developed ".. .to evaluate the
effectiveness of methane leak detection and repair (LDAR) programs at oil and gas facilities."2
FEAST was initially developed at the Environmental Assessment and Optimization group at
Stanford University by Chandler E. Kemp, Arvind P. Ravikumar, and Adam R. Brandt and
released in 2016.3 The FEAST model is currently on the fourth version: FEAST 3.14 (referred to
only as FEAST for the purpose of this memo) and is based in Python 3. FEAST provides a
stochastic model of emissions at the component level occurring as the result of leaks (that can be
identified and repaired through the LDAR program) and vents (that may be "detected" by some
LDAR programs but are not subject to repair) in a natural gas field.
FEAST is highly customizable. The time step and duration of the simulation can be set by
the user. Different components can be defined with different leak production (generation) rates
and emission rate distributions for the leaks generated. Sites can be defined as a collection of
different components and a gas field can be defined as a collection of sites. FEAST supports the
following LDAR technologies and can model hybrid LDAR programs:
• Optical gas imaging (OGI) camera
• Aerial surveys (both equipment- and site-level surveys)
• Drone surveys
• Continuous monitoring systems
2 https://www.arvindravikumar.com/feast
3 C.Kemp etal. Environ. Sci. Tech. 50 4546. http://dx.doi.org/10.1021/acs.est.5bQ6Q68
4 https://github.com/FEAST-SEDLab/FEAST PtE/tree/FEAST 3.1
2
215
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
The LDAR program effects are simulated based on probability of detection (PoD) curves
(or surfaces) for each monitoring method, which indicate the probability that a leak of a given
size will be detected within a given survey (or time period for continuous monitoring
technologies), and survey times (frequencies) are accounted for as finite time periods. Based on
parameters of the repair process, emission mitigation is quantified for leaking emissions which
are repaired, resulting in a lower emission profile relative to the baseline scenario.
3.2 FEAST General Modeling Approach
We conducted all modeling runs using a 1-day time step. We ran the model for 5 years
and compared the emissions in the fifth year. This modeling approach was used to allow for the
buildup of small leaks that different monitoring methods may not be able to detect. We varied
the leak generation rates for conventional components from 0.5 to 2 percent per year. We
assessed different "auto-repair" rates, which is the rate at which leaks fix themselves or are
repaired in the absence of an LDAR program. We decided to set the auto-repair rate to zero
because the auto-repair applied to all leaks, regardless of size. That is, when using the auto-repair
approach, the operator is assumed to find and fix very small leaks at the same rate as larger leaks.
This limits the buildup of small, very difficult to detect leaks when using an auto-repair rate.
Because we do not believe these small leaks would be repaired as often as larger leaks (in the
absence of a regulatory program), we used a "baseline monitoring program" to simulate facilities
occasionally identifying larger leak sources using audio, visual, or olfactory (AVO) methods and
repairing those larger leaks - those that would result in cost-savings to repair.
Because we were interested in identifying appropriate monitoring frequencies for
different types of facilities, we always modeled only one site type at a time. We modelled the
field as 20 identical sites of the type we were evaluating. The model output provides emissions
and repair data for each site. We compiled the average site emissions and the number of repairs
made per site in the 5th year for each model run to use in our costing analysis.5 Each model run
consisted of emissions analysis for the field, which was defined as 20 model plant sites. We used
a Monte Carlo approach, conducting analysis of the field emissions under a different set of leak
generated for each Monte Carlo iteration.
3.3 Emission Leak Distributions
FEAST includes a default leak data set for production facilities. The study data included
in the FEAST default data set are collected from the following measurement studies.
• Measurements of Methane Emissions at Natural Gas Production Sites in the United
States. Supporting Information (Allen et al. 2013V For this study, 150 natural gas
production sites were surveyed, and leaks were detected at 97 sites. Equipment leaks
were assessed at compressors, well heads, and equipment in four producing regions of
the United States: Appalachia, Gulf Coast, Mid-Continent, and Rocky Mountains.
Sites were screened with an infrared camera to identify leaks, and all observed leaks
5 FEAST has cost estimating procedures, but rather than attempt to revise the cost input data files to match the
EPA's historic cost estimates for conducting OGI, it was easier to estimate monitoring costs outside of FEAST using
the number of repairs made in the fifth year and conduct the cost analysis outside of FEAST.
3
216
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
were measured with a Bacharach high-flow sampler. Emissions data are available
from the web site of the Center for Energy and Environmental Resources at the
University of Texas, Austin; the data include approximately 278 measurements of
whole gas and methane emissions from leaks at production sites.7
• City of Fort Worth Natural Gas Air Quality Study (ERG and Sage. 2011V This study
measured fugitive gas emissions from equipment in the Fort Worth Basin associated
with the Bamett Shale including: 375 well pads, eight compressor stations, one gas
processing plant, and one produced water treatment facility. Leak screening and
measurement was performed using OGI and a high flow sampler for a total of 91
methane measurements at gathering and boosting facilities and 1,200 measurements
at production facilities. These data are listed as "ERG Camera 2011'' in the default
data set. EPA Method 21 and bagging techniques were also performed on selected
sources. These data are listed separately from the OGI data as "ERG TVA 2011" in
the default data set. All production facilities in the study are located in the Fort
Worth Basin and are associated with gas production from the Bamett Shale.
• Estimation of Methane Emissions from the California Natural Gas System (Kuo et al..
2015"). FEAST documentation reports this study as Kuo, 2011; expect the study was
conducted in 2011, but not published until 2015. This study measured fugitive
emissions from 25 facilities in California representing facilities across the natural gas
industry; 12 of the facilities were in the production and processing sector. FEAST
documentation notes that only the production well equipment were included in the
FEAST data set. EPA Method 21 was used to identify leaks; Bacharach Hi-Flow
samplers were used to quantify the leaks. Study authors noted that the Bacharach Hi-
Flow sampler could not accurately quantify some of the smaller leaks. They also
noted that most sites visited were already conducting routine LDAR monitoring,
which may lead to lower emission factors as compared to the 1995 GRI/EPA study.
• Repeated leak detection and repair surveys reduce methane emissions over scale of
years (Ravikumar. et al., 2020). This study included initial and follow-up
measurements at 36 sites in Alberta, Canada: 30 well pads and 6 processing plants. A
FLIR OGI camera was used to identify leaks; Bacharach Hi-Flow samplers were used
to quantify the leaks. Study author noted that tank leaks were not measured due to
safety concerns. Emission factors were developed for venting tanks from other study
data; we used only measurement data from Ravikumar. so tank emission estimates
from this study were not used. The study authors noted that equipment leak
emissions were not well correlated with production, such that low production wells
have similar emissions as high production wells.
6 The study authors noted in the Supporting Information that the threshold for detecting a leak with the infrared
camera used by the study team was 30 grams per hour (g/hr) (0.026 standard cubic feet per minute (scf/min)),
compliant with the approved alternative work practices in 40 CFR 60.18. In practice, the threshold for leak
detection depends upon operator experience and skill in interpretation of the visual images, as well as site-specific
parameters such as the visual background for the leak image.
7 See Project Data sets for Allen (2013), available for download from Center for Energy and Environmental
Resources, University of Texas, Austin, http://dept.ceer.utexas.edu/methane/studv/datasetsl.cfm.
4
217
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
• Comparison of methane emission estimates from multiple measurement techniques at
natural sas production pads (Bell, et ah, 201 7). This study estimated methane
emissions at 268 gas production facilities in the Fayetteville shale gas play (in
Arkansas) using onsite measurements (261 facilities) and two downwind methods -
the dual tracer flux ratio method (Tracer Facility Estimate - TFE, 17 facilities) and
the EPA Other Test Method 33a (OTM33A Facility Estimate - OFE, 50 facilities).
Emission sources were first identified during a comprehensive site survey using a
combination of optical gas imaging and handheld laser methane detection. Identified
leaks were quantified using a high-flow samplers, with care to limit issues identified
by Floward. et al., 2015. Only the component leak data were used.
The default FEAST data set included all measurement data from these studies, which
may include measurements of pneumatic devices and tank thief hatch releases. While pneumatic
devices may malfunction, these are not "leaks." Continuous bleed pneumatic devices emit
continuously, so the "leak generation rate for these devices should be 100%. Intermittent bleed
devices only bleed when they are actuating. Some intermittent bleed devices used for process
control may actuate every few minutes; other intermittent bleed devices are used to actuate
isolation valves and may actuate only a few times a year. However, these are not leaks and it is
incorrect to characterize "leaks" based on pneumatic device bleed rates. Pneumatic device
emissions may be included as vented emissions (non-repairable), but they must be modeled
differently than "leaks." Consequently, we separated the data so that equipment component data
[i.e., valves, connectors, flanges, pumps, open-ended lines (OELs), and pressure relieve valves
(PRV)] were in one file and tank "leak" emissions were in another file. We also compiled other
measurement data in the default FEAST input file (for pneumatics or liquids unloading venting),
but these were not used.
In addition to the filtered study data from contained in the default FEAST leak data set,
we also augmented the FEAST leak data set with data from two other studies that were recently
reviewed. These studies are:
• Equipment leak detection and quantification at 67 oil and gas sites in the Western
United States (Pacsi. et al., 2019). This study included 67 production and gathering
and boosting oil and gas sites in the Permian, Anadarko, Gulf Coast and San Juan
basins. Equipment leaks were screened on all major equipment using OGI and EPA's
Method 21. Emissions from leaks were quantified using a high-flow sampler.
Complete equipment leak component inventories were also performed at each site.
• Methane Emissions from Gatherins Compressor Stations in the U.S. (Zimmerle. et
al.. 2020a). This study included 180 gathering stations for which information was
collected on equipment counts and types. Component level equipment leak screening
was also performed using OGI, and emissions were quantified using a high-flow
sampler. We note that the Zimmerle et al., 2020a paper nor its supporting information
provided direct results of the equipment component activity data, however, these
results were provided in Supporting Volume 3 of the Department of Energy (DOE)
report (Zimmerle et al., 2019).
5
218
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Data from these studies were similarly parsed between equipment component leak data
and tank leak data.
Note that, because almost all of these studies largely relied on high-flow samplers, most
of the leak data are limited to the minimum and maximum detection limits of the high-flow
sampler, which is about 0.01 to 8 standard cubic feet per minute (scfm) or about 0.01 to 9 kg/hr.
Aerial studies have indicated site wide emissions are often much higher and some individual
sources may also be much higher than these. In order to account for large emission events, data
from the following aerial study were used to develop a distribution for high emitting sources.
• Intermittencv of Large Methane Emitters in the Permian Basin. (Cusworth. et al..
2021). This study conducted aerial methane leak detection covering 55,000 km2 area
within the Permian basin. They used two different airborne platforms: a "next-
generation" Airborne Visible/Infrared Imaging Spectrometer (AVIRIS-NG) and the
Global Airborne Observatory (GAO). The GAO uses the same detection method
(ARVIS-NG) but is also equipped with a high-resolution digital camera. Multiple
flights were conducted of the study area between September and November 2019, to
better understand the intermittency of large emitters. The study identified 3,067
unique plumes from 1.756 distinct sources over the course of the campaign. We used
these individual plume data, which ranged in size from 15 kg/hr to 16,800 kg/hr, as
the base of the large-emitter data set.
Because the Cusworth, et al, (2021) aerial methods had an apparent lower detection limit
of 15 kg/hr during optimal conditions, we assumed that there were likely more large emission
sources in the 5 to 50 kg/hr range that may not have been identified during the aerial survey due
to lower detection limits of the aerial method used. Nonetheless, emissions in the 5 to 50 kg/hr
are still large compared to the equipment component and tank leak data. Therefore, we
augmented the Cusworth data with additional data in this range using assumed log-normal leak
distribution.
To compare and contrast the study data, the input data file used for each study, which
contains leak emissions data in units of g/sec, were evaluated. While not all of the different study
data were log-normally distributed (often due to detection limit issues with the measurement
methods used), we used natural log transform of the data to more easily compare and contrast the
data. We present the data for equipment components first, then storage tank leaks, then "large-
emitters.''
3.3.1 Equipment Component Leaks
The box and whisker plot for each study in the equipment component leak data set are
provided in Exhibit 1. Looking at the median values, the ERG Camera 2011 measurements
contain the highest average emissions, while ERG TV A 2011 has the lowest. Most of the
different study data using OGI as the primary detection limit have similar range of emissions
data: these include: Allen 2013; Ravikumar 2020; Bell 2017; and Zimmerle 2019. Kuo 2012
(published 2015) has the highest emissions of the Method 21 studies; Pacsi 2019, which used a
mix of Method 21 and OGI has a low leak distribution similar to the ERG TVA study. The ERG
Camera 2011 is also the study with the second highest number of data points (751 records,
second to the Zimmerle 2019) study, almost twice the number of records in the Ravikumar 2019
6
219
-------
Technical Memorandum -Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
or Allen 2011 studies and 7 times the number of records in Kuo2012 and Bell 2017, so the
model is more likely to pick values from the ERG Camera 2011 study data than these other
studies. Nonetheless, we specifically wanted to include EPA Method21 identified leaks in this
analysis to account for small leaks that OGI survey crews are not likely to detect.
It is evident that the Kuo 2012 (actually 2015 publication) and the Ravikumar 2020 data
sets have lower detection limits with over a quarter of the data at the lowest measurement flow
rate. A Q-Q plot of this data clearly shows that many studies use methods that identified leaks
that were at or below the quantitation limit for the high-flow samplers (see Exhibit 2).
Exhibit 1. Box and whisker plot of Equipment Component leak data.
7
220
-------
# Allen 2013
# ERG TVA 2011
O ERG Camera 2011
O Kuo 2012
Ravikumar-Measured only 2020
O Bell-2017
£ Pacsi-2019
£ Zimmerle-2019
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27,2022
Exhibit 2. Q-O plot of Equipment Component leak data.
Mean = -3.1 G8
Sd-1.833
Slope = 1.787
Intercept = -3.168
Correlation, R = 0.973
Kuo 2012
N = 90
Mean = -5.34
Sd = 1.405
Slope = 1.265
Intercept = -5.34
Correlation, R = 0.89
Ravikumar-Measured only 2020
N = 380
Mean = -4.92
Sd = 1.203
Slope = 1.1
Intercept = -4.92
Correlation, R = 0.911
Bell-2017
N = 106
Mean = -4.466
Sd = 1.457
Slope = 1.442
Intercept = -4.466
Correlation, R = 0.98
Pacsi-2019
N = 319
Mean = -6.125
Sd = 2.075
Slope = 2.045
Intercept = -6.125
Correlation, R = 0.982
Zimmerle-2019
N =834
Mean = -5.092
Sd = 2.179
Normal Q-Q Plot
-3-2-10 1 2 3
Theoretical Quantiles (Standard Normal)
8
221
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Exhibit 3 provides the aggregate distribution of the combined data set (all studies) for the
equipment components.
Exhibit 3. Histogram for- the Equipment Leak data (natural logs of emissions in g/s).
3.3.2 Tank Leaks
The box and whisker plot for each study in the tank leak data set are provided in Exhibit
3. Again, the ERG Camera 2011 measurements contain the highest median emissions, the ERG
TV A 2011 has the lowest, and the other study data have similar range of emissions in between
these data. For the tank leaks data, the ERG Camera 2011 study has the highest number of data
points (184 records), which is a factor of 2 or more times the number of records included in other
studies. Again, this means that FEAST is more likely to pick values from the ERG Camera 2011
study data than from these other studies.
9
222
-------
Technical Memorandum-Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Multiple Box Plots
c
o
CO
_Q
o
-O
a>
-12
Allen 2013
ERG TVA 2011 ERG Camera 2011 Bell-2017
Zimmerle-2019
Exhibit 4. Box and whisker plot for the Tank leak data.
Exhibit 5 provides the aggregate distribution of the combined data set (all studies) for the
tank leak data. The median emissions for tank leaks is around 0.05 g/s [ln(0.05) = -3], whereas
the median emissions rate for equipment component leaks is around 0.007 g/s [ln(0.007) = -5],
Thus, the tank leaks are significantly larger than the equipment component leaks. However, a
0.05 g/s leak converts to 0.18 kg/hr, and the largest tank leak from any study used to develop the
tank leak distribution is about 10 kg/hr, well below the lowest sitewide emissions seen in aerial
surveys reported by Cusworth, et al., 2021. This is lar gely due to the limitations on the maximum
pump (or sampling) rate of high flow samplers. High volume samplers typically have an upper
quantitation limit of about 8 scfrn, which translates into emissions of about 9 kg/hr assuming
methane is the primary constituent.
10
223
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Histogram for Tanks In
-12 -11 -10 -8-7-5-1-3-1 0 1
T anksjn
Number of Values
473
Minimum
-1251
Maximum
118
SD
234
Skewness
-a 37
Kurtosis
0.06
~ Mean
-3.61
~ Median
-3.31
~ Normal Distribution
I~1 Less Bins
~ Mote Bins
Exhibit 5. Histogram for the Tank Leak data (natural logs of emissions in g/s).
3.3.3 Large-Emitters
Exhibit 6 provides the distribution of site emissions from the Cusworth study. These
emissions are much higher than the ground survey studies, largely due to the limits of detection
of the different methods used. The smallest emission rate measured by Cusworth was 15 kg/hr
and they only saw these emissions in 1 of 3 flyovers. While Cusworth focused on the
intennittency of large-emitters, it is likely the intermittency seen at this lower level of detection
is attributable to the varying meteorological conditions impacting the lower limit of detection.
As such, rather than assume there are no or limited emissions between 10 kg/hr and 70 kg/hr
based on Cusworth data, we surmised that the limited emissions in this range were more due to
the limitations of emissions detection based on the sensitivity of the aerial methods considering
the impacts of meteorology. Therefore, we augmented the Cusworth data to have a more even
distribution of large-emitters between the maximum tank emissions and the median emissions of
the Cusworth data.s Exhibit 7 provides the aggregate distribution of the augmented data set for
3 There were 3067 unique measurements from Cusworth, et al., 2021. We added 1,762 randomly generated values
from a log-normal distribution using Excel equation =LOGNORM.INV(RAND(),2.6,0.6) and 1,233 values generated
using Excel equation =LOGNORM.INV(RAND(),1.1,0.6). Note that the units of emissions in FEAST are g/sec. The first
distribution yielded 90 percent of data between 18 and 125 kg/hr; the second distribution yielded 90 percent of
data between 4 and 28 kg/hr.
11
224
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
large emission events. Note that the log normal value of 1 on this graph is equivalent to 10 kg/hr
emission rate [e1 = 2.7 g/sec = 9.8 kg/hr].
Exhibit 6. Histogram of data from Cusworth 2021.
12
225
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Histogram for Augmented Cusworth
1300
1214
1200
1112
1100
1035
I I
-lllllll.
0 2 4 6 0
Exhibit 7. Histogram for the augmented "large-emitter" distribution.
3.4 Model Production Sites
In this evaluation, we considered four different model plants, based on the size and
complexity of different production sites. Table 1 summarizes the model production sites used in
the FEAST modeling runs. The number of "sources" listed in Table 1 indicate the maximum
number of potential leaking emission sources for a given leak distribution at the site.
While a large-emitter may be a very large leak from an equipment component or tank or another
source (such as unlit flare or cracked pipe), we assign a value of 1 or 2 to large-emitters so the
facility can have at most 1 or 2 large leaks pulled from the large-emitter leak distribution.
Table 1. Model Production Sites.
Model Site
Name
Description
Number of
Fugitive
Equipment
Components
Number of
Tanks
Number of
Large-Emitters
Model Plant 1
Single wellhead only
112
0
1
Model Plant 2
Dual wellheads only
220
0
1
Model Plant 3
Typical production site;
uncontrolled tanks
612
0
2
Model Plant 4
Typical production site;
controlled tanks
612
A
2
13
226
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Model Plant 3 contains tanks, but as they are uncontrolled, the emissions from these tanks
are allowed to vent to the atmosphere. Controlled tanks in Model Plant 4 must be vented through
a closed vent system to a control device. Here, an open thief hatch or leak in the closed vent
system could require action to repair the leak.
All FEAST model runs were performed with 20 sites of the model plant being evaluated
within the field. The average emissions per site were determined for each simulation ran of 20
sites.
3.5 Leak Detection Settings
For all survey types, we defined the leak detection curves by defining 0%, 25%, 50%,
75%, and 100% detection values. The 100 percent detection value for OGI was set based on the
regulatory requirement. In the proposed NSPS OOOOb, this 100% detection limit was set at 60
g/hr. We recognize that even trained operators may not see all large leaks 100 percent of the time
(Zimmerle, et al., 2020b), but we set the upper detection probability to 100 percent at 60 g/hr
since the proposed NSPS OOOOb requires this sensitivity at a minimum.
For required OGI surveys as well as site-level survey methods, we assumed the 50%
detection limit was a factor of 2 less than the 100% detection limit and the 0% detection limit
was a factor of 4 less than the 100% detection limit. We set the 25% value as the midpoint
between the 0 and 50% values and the 75% value was set as the midpoint between the 50 and
100% values. Thus, all detection curves for required surveys had the shape seen in Exhibit 8.
As noted previously, we set the auto-repair rate to zero and established a "baseline
monitoring program" to simulate facilities occasionally identifying larger leak sources using
AVO methods. We developed this baseline monitoring program initially before fully adopting
the convention described above for the detection curves. For the baseline monitoring detection
curve, we set the 100% detection limit at 0.034 g/s (122 g/hr), the 0% detection limit at a factor
of 2 lower (61 g/hr) and used a straight line between these two points to determine the 25%,
50%, and 75% detection values. The survey frequency for this baseline survey was set at 650
days. This frequency was selected because it provided baseline emission estimates near those
expected based on recent literature. We subsequently conducted an analysis specific to AVO and
AVO monitoring frequencies and evaluated a broad range of probability detection curves for
AVO methods. The probability detection curves assumptions and analysis conducted for AVO
monitoring are described in Section 4.2 of this memorandum.
14
227
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27,2022
1
Probability of Detection Curve
1
0.9
(TJ
u
a o.6
HI
Q 0.5
H—
° 0.4
j= 0.3
£ 0.2
o
£ 0.1
0
C
jm w
25
0.005 0.01 0.015 0.02 O.C
Leak Rate (g/s)
Exhibit 8. Detection probability curve for required ground-based OGI surveys.
4. FEAST Run Results for Fugitive Emission Components
Initial FEAST simulations were conducted to identify cost-effective, ground-based
LDAR programs for the differently sized model plants using the emission distributions for
fugitive emission components (including tank thief hatches) only. For these initial analyses, large
emission events were not included. The following fugitive monitoring frequencies were
evaluated:
• Annually
• Semi-annually
• Quarterly
• Bi-monthly
• Monthly
Initial runs evaluated both 15- and 30-day repair periods (delay between detection and
repair). It is espectedthat a 30-day repair requirement would likely have an average of 15 days
between detection and repair, with some repairs done quickly (on first attempt at repair) and
some taking longer. However, because the repair period being considered was 30-days, all
subsequent runs were conducted assuming 30-day "delay" between detection andrepair.
15
228
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
4.1 Emission Estimate Results for OGI Monitoring Options
We evaluated site-wide emissions for each model plant based on a low, medium, and
high leak generation rate as summarized in Table 2.
Table 2. Leak Generation Rate Scenarios
Emission Source
Leak Generation Rate (% of components per year)
Low
Medium
High
Equipment components
0.5%
1%
2%
Tanks
2.5%
5%
10%
The results of the FEAST simulations are summarized in Tables 3 through 6 for Model
Plants 1 through 4 for various monitoring frequencies using OGI required to be able to detect
60 g/hr leak fugitive emission components. The magnitude of emissions is significantly impacted
by the assumed leak generation rate, with emissions increasing in direct proportion to the
assumed leak generation rate.9 Even so, the percent emission reductions achieved by OGI
monitoring program was essentially identical regardless of the assumed leak generation rate or
the model plant configuration. The effectiveness of the various monitoring frequencies in terms
of percent emissions from the baseline are summarized in Table 7.
Table 3. Emission Simulations Results for Model Plant 1
Monitoring Program
Average Emissions (tons CHj/year) per Site
for Leak Generation Rate Levels
Low1
Medium1
High1
Baseline
1.27
2.97
4.83
Annual OGI
0.65
1.61
2.68
Semi-annual OGI
0.41
0.97
1.64
Quarterly OGI
0.29
0.62
1.09
Bi-monthly OGI
0.25
0.55
0.93
Monthly OGI
0.19
0.43
0.78
1See Table 2 for percent of components leaking at these ratios.
1 The initial OGI FEAST analysis was performed with only 100 runs. With 20 sites per run, this assessment
considered emissions across 2,000 sites. Given the relatively small number of sites modelled (for Monte Carlo
analyses considering highly variable emission sources like fugitive emission components) the average emissions
determined in this series of results have higher uncertainty than subsequent simulations where a higher number of
runs were made. Nonetheless, for a given model plant and leak generation rate setting, all monitoring frequencies
were run in a single simulation, with a common set of leaks generated, so the emissions for a given column in
Tables 3 through 6 are directly comparable.
16
229
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27. 2022
T able 4. Emission Simulations Results for Model Plant 2
Monitoring Program
Average Emissions (tons ChU/year) per Site
for Leak Generation Rate Levels
Low1
Medium1
High1
Baseline
2.66
4.68
9.61
Annual OGI
1.48
2.56
5.62
Semi-annual OGI
0.87
1.58
3.37
Quarterly OGI
0.60
1.07
2.26
Bi-monthly OGI
0.51
0.90
1.90
Monthly OGI
0.41
0.72
1.50
1See Table 2 for percent of components leaking at these ratios.
Table 5. Emission Simulations Results for Model Plant 3
Monitoring Program
Average Emissions (tons ChU/year) per Site
for Leak Generation Rate Levels
Low1
Medium1
High1
Baseline
7.18
14.08
28.15
Annual OGI
3.77
7.94
15.46
Semi-annual OGI
2.33
4.79
9.31
Quarterly OGI
1.56
3.24
6.22
Bi-monthly OGI
1.30
2.65
5.20
Monthly OGI
1.07
2.15
4.09
1See Table 2 for percent of components leaking at these ratios.
Table 6. Emission Simulations Results for Model Plant 4
Monitoring Program
Average Emissions (tons CH4/year) per Site
for Leak Generation Rate Levels
Low1
Medium1
High1
Baseline
8.51
15.40
31.10
Annual OGI
4.52
8.76
16.84
Semi-annual OGI
2.78
5.29
10.16
Quarterly OGI
1.90
3.53
6.74
Bi-monthly OGI
1.54
2.93
5.53
Monthly OGI
1.25
2.35
4.40
1See Table 2 for percent of components leaking at these ratios.
17
230
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27. 2022
Table 7. Typical Percent Emission Reductions by OGI Monitoring Frequency
Monitoring Program
Percent Emission
Reduction from
Baseline
Annual OGI
45%
Semi-annual OGI
67%
Quarterly OGI
77%
Bi-monthly OGI
81%
Monthly OGI
85%
4.2 Emission Estimate Results for A VO Monitoring
As an alternative to OGI monitoring, we attempted to assess the emission reduction that
could be achieved using AVO monitoring methods. Unlike OGI monitoring, for which we have
specific minimum detection quantities specified in the method (60 g/hr requirement in proposed
NSPS OOOOb), we have no specific information by which to estimate AVO leak detection
sensitivities. Therefore, we initially ran for different sets of AVO monitoring detection limits as
summarized in Table 8.
Table 8. Probability of Detection Inputs for AVO Monitoring Options
AVO Option
Leak Rate at Specified Probability of Detection (g/hr)
0%
25%
50%
75%
100%
AVOI
61
77
94
108
122
AV02
61
92
122
184
245
AV03
122
153
184
214
245
AV04
122
184
214
367
490
We evaluated emissions for the same five monitoring frequencies for which OGI
monitoring was evaluated (i.e., annually, semi-annually, quarterly, bi-monthly, and monthly).
Example results for Model Plant 1 at a leak generation rate of l°o are provided in Exhibit 9.
18
231
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Impact of AVO Detection Curve Assumptions
3 C
3.0
^ 2.5
(/>
c
o
c 2.0
0
01
10
£ „ r-
UJ 1.5
O)
c
(0
JC
OJ 1.0
2
0.5
0.0
annual
semiannual quarterly
AVOl AVO 2 AVO 3
bimontly
AV04
monthly
Exhibit 9. Emission results for Model Plant lfor different AVO survey sensitivities and
frequencies using a 1% leak generation rate.
The results presented in Exhibit 9 indicate little difference in the emission results
between AVO options 1 and 2. Similarly, emission results between AVO options 3 and 4 are
quite similar, especially as the monitoring frequency is increased. These results suggest that the
performance of a monitoring method is more strongly dependent on the lower limit of detection
than it is on the upper limit of detection.
The emissions projected for annual AVO monitoring using the detection limits for AVO
option 3 were very similar to the "baseline" emissions, so the results for AVO option 3 were
used for further comparisons. The emissions for various monitoring frequencies for AVO
option 3 for Model Plants 1, 2, and 4 are presented in Table 9.
19
232
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Table 9. Methane Emissions for AVO Monitoring
AVO Option
Annual Methane Emissions (U.S. tons/yr)
MP1;
Low LGR1
MP1;
Mid LGR1
MP2;
Low LGR1
MP2;
Mid LGR1
MP4;
Low LGR1
MP4;
Mid LGR1
Annual
1.25
2.49
2.36
4.85
7.33
14.67
Semiannual
0.97
1.92
1.83
3.72
5.54
11.14
Quarterly
0.81
1.61
1.54
3.10
4.60
9.21
Bimonthly
0.74
1.49
1.44
2.87
4,28
8.50
Monthly
0.68
1.37
1.32
2.64
3.89
7.77
1MP = Model Plant; LGR = leak generation rate. Low and mid LGRs are specified in Table 2.
4.3 Emission Estimate Results for Combined OC//A VO Monitoring Options
We also evaluated the emissions projected for a combined OGI/AVO monitoring option.
Specifically, we considered the options of adding AVO monitoring to a "baseline OGI"
monitoring option. For Model Plants 1 and 2, the "baseline OGI" monitoring frequency was
assumed to be semiannual, and we considered adding quarterly, bimonthly, or monthly AVO
monitoring. Quarterly AVO monitoring, in this example, would imply the facility would monitor
4 times a year, twice using OGI and twice using AVO monitoring. For Model Plant 4, the
"baseline OGI" monitoring frequency was assumed to be quarterly, and we considered adding
bimonthly or monthly AVO monitoring. We note that bimonthly AVO only overlaps with two of
the quarterly monitoring methods. Assuming OGI is conducted initially, which we refer to as
"time zero", AVO monitoring would be conducted on months 2, 4, 8, and 10 and OGI
monitoring would be conducted on months 3, 6, 9 and 12. For this analysis, we used the
probability of detection limits for AVO option 3 from Table 8, except that we set the maximum
probability of detection to 90%. We used the emission rate value shown in Table 8 as the 90%
90% probability. The emissions for various OGI/AVO combination monitoring options for
Model Plants 1, 2, and 4 are presented in Table 10.
Table 10. Methane Emissions for Combined OGI/AVO Monitoring Options
AVO Option
Annual Methane Emissions (U.S. tons/yr)
MP1;
Low LGR1
MP1;
Mid LGR1
MP2;
Low LGR1
MP2;
Mid LGR1
MP4;
Low LGR1
MP4;
Mid LGR1
Baseline Semiannual
OGI
0.904
0.419
0.847
1.65
N/A
N/A
Baseline Quarterly OGI
N/A
N/A
N/A
N/A
1.73
3.45
Combined Quarterly
OGI/AVO
0.69
0.315
0.640
1.24
N/A
N/A
Combined Bimonthly
OGI/AVO
0.58
0.265
0.545
1.04
1.42
2.83
Combined Monthly
OGI/AVO
0.50
0.231
0.469
0.90
1.25
2.48
1MP = Model Plant; LGR = leak generation rate. Low and mid LGRs are specified in Table 2.
20
233
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
4.4 Discussion of Modeled Emission Estimates
Rutherford, et al., 2021, conducted Monte Carlo analyses using leak frequency data, leak
emission rates, and component counts for equipment to develop average emission rates for
equipment and for production sites. The average equipment level emission reported by
Rutherford, et al., (2021; Supplemental Information) include:
• Gas wellhead: 3.35 kg CFLt/day (1.35 tons CHVyr)
• Gas meter: 2.66 kg CHVday (1.07 tons CFLt/yr)
• Gas separator: 3.72 kg CFLi/day (1.50 tons CFLi/yr)
For Model Plant 1, which is a single wellhead with meter and piping, Rutherford
equipment emission factors suggest this model plant's emissions would be 2.42 tons CHi/yr.
This is similar, but slightly lower than the emissions projected for Model Plant 1 using a leak
generation rate of 1% within FEAST. While Rutherford, et al., (2021) used a much lower percent
of equipment leaking (weighted-average across different component types) than 1% leak
generation rate assumed in FEAST, Rutherford also had much higher component counts per
major equipment than projected using the East/West component counts from Table W-1B from
40 CFR part 98, subpart W, which are from the joint study between EPA and the Gas Research
Institute (GRI) of methane emission from the natural gas industry (EPA/GRI, 1996). For Model
Plant 2, Rutherford's equipment factor for gas equipment would project emissions of 4.84 tons
CFLi/yr. Rutherford's equipment factors for oil wellheads and oil meters are about half those of
gas equipment. The net result is that Rutherford's emission estimates for gas wells are similar to
the baseline emission estimates modeled when using the 1% leak generation rate in FEAST and
Rutherford's emission estimates for oil wells are similar to the baseline emission estimates
modeled when using the 0.5% leak generation rate in FEAST.
Note that the model input leak generation rate of 1% would indicate that, on average, 6.1
leaks would be generated per year per site for Model Plant 3 (with 612 components). These leaks
are drawn randomly from the leak distribution data set. Many of these leaks are below the
detection limit of an OGI instrument that is only required to detect 60 g/hr leak. We used the
number of repairs made in the fifth year as the annual leak detection (and repair) rate. The annual
leak detection rate was, for the most part, independent of the monitoring frequency used.10 The
annual number of OGI detected (and repaired) leaks averaged 3.5 for Model Plant 3 or
approximately 57% of the leak generation rate regardless of the monitoring frequency. Thus, if
an annual OGI survey is conducted that detects leaks in 0.6 percent of the monitored
components, that would be directly comparable to the 1% leak generation rate used in the model.
10 Annual frequency generally had lower number of repairs; the number of repairs for other frequencies were
identical within the random variations of the model simulation. We believe there are two reasons why annual
number of repairs were slightly lower for annual monitoring than other monitoring frequencies. First, the repairs
made in the fifth year for the annual monitoring frequency are from leaks identified on day 1 of year 5, which
considers only the build-up of small leaks in the first 4 years. Other monitoring frequencies include build-up of
leaks (and subsequent detection and repair) for some portion of year 5. So, there was less time for annual
monitoring to reach "equilibrium." Second, for leaks in the middle of the probability of detection, having multiple
monitoring of leaks in that range in a year should increase the likelihood of detecting a leak in that range. We
expect longer simulation periods would result in build-up of leaks in that middle range so that annual monitoring
would reach the same equilibrium number of repairs made per year as the other monitoring frequencies.
21
234
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Using a FEAST leak generation rate of 0.5% would be comparable to an annual OGI survey
detecting leaks in 0.3 percent of the monitored components.
This also indicates that, based on the distribution of leak rates identified in component-
specific emission measurement studies, about 20 to 40 percent of the leaks "generated" at a site
will be below the detection limits of a typical OGI instrument. These small leaks will tend to
accumulate over time. Most simulation models use an auto-repair rate to prevent the build-up of
these small leaks. As noted previously, we decided to set the auto-repair rate to zero because the
auto-repair function would fix very small leaks at the same rate as larger leaks, which is unlikely.
Even with the build-up of these small leaks over 5 years, these small leaks do not appear to
significantly contribute to the model plant emissions. Considering the repaired leaks for monthly
monitoring frequency occur for an average of 45 days (start mid-month so present 15 days before
detection, and repair completed 30 days after detection), we project that these small, non-
repaired leaks represent only 3 or 4 percent of the baseline emissions.
5. FEAST Run Results with Large-Emitters for Sitewide Monitoring Equivalence
'Die next model simulations were conducted to determine when mobile, aerial, or drone
monitoring, collectively referred to as site-wide monitoring, can achieve the same emission
levels as ground-based OGI monitoring. For these analyses, we included large emission events
and conducted analyses to determine equivalency to semiannual ground-based OGI surveys for
small sites (specifically for Model Plant 2) and quarterly ground-based OGI surveys for larger
sites (specifically for Model Plant 4).
Initially, site-wide monitoring was conducted at the following leak detection limits
(representing 100% detection level).
• 1 kg/hr
• 5 kg/hr
• 15 kg/hr
• 30 kg/hr
• 60 kg/hr
Monitoring frequencies that were evaluated were primarily monthly and bi-monthly and
included site-wide surveys only or site-wide surveys with annual OGI monitoring "backstop."
All leaks detected via a site-wide survey deployed a ground-level OGI monitoring crew to
monitor all fugitive emission components at the site and repair all leaks identified. This OGI
monitoring was identical to the quarterly OGI monitoring "baseline" (60 g/hr 100% detection
level; repair within 30 days of initial screening survey).
We looked initially at large-emitter rates of 0.5%, 1%, and 2% to understand the impact
of large-emitter generation rates on emissions. We quickly found out that the inclusion of large-
emitters greatly increased the variability of the model results. When modeling fugitive
components, there are 612 components per site (for Model Plant 4) so the fieldwide (20 sites)
emission results included simulation of 12,240 component sources. With the large pool of
sources, the individual Monte Carlo runs were quite similar and the initial modeling runs were
conducted using only 100 iterations (for a total of 2,000 sites modeled). With the inclusion of
22
235
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
large-emitters with only 2 potential sources per site (for Model Plant 4) the variability by site and
by field (i.e., group of 20 sites) is very large, highlighting the significant impact of large-emitter
emissions on the fieldwide emissions. Due to the high variability of emissions when including
large-emitters, a significantly higher number of Monte Carlo simulations are needed to yield
consistent results.
During a given simulation, we included the baseline OGI monitoring survey, a sitewide
screening survey that would call for the ground-based survey for a particular site only if
emissions were detected ("no backstop"'), and a sitewide screening survey that would call for the
ground-based survey for a particular site both if emissions were detected and annually regardless
of when a ground-based survey was last deployed ("with backstop"). These simulations occurred
on the same set of leaks generated, so the comparison on the performance of the options is most
directly comparable. We noted that, for simulations that had high baseline emissions, the
emissions of the sitewide monitoring options would also increase. However, the relative
performance of the sitewide monitoring options tended to be better when the baseline emissions
were higher, indicating stronger impact of large-emitters. This was also evident with
comparisons of modeling runs using different large-emitter leak frequencies. The higher the
large-emitter leak frequency, the more likely it was that the more frequent (but less sensitive)
sitewide monitoring would perform better than ground-based OGI monitoring.
5.1 Equivalency Matrix at Medium Leak Generation Rates
We conducted our formal equivalency modeling using median leak generation rates for
equipment components are presented in Table 2 and using a 1% leak generation rate for large-
emitters for both Model Plants 2 and 4. We selected this rate based on data from Zavala-Araiza,
et al., (2017), which showed about 1 percent of sites emitting over 26 kg/hr.11 While the
augmented large-emitter distribution has more data in the 10 to 30 kg/hr range, data below 26
kg/hr (7.22 g/sec) represent less than 25 percent of the dataset. Also, we assume 2 large-emitter
sources per site for Model Plants 3 and 4, so the 1 percent leak generation rate combined with
these other assumptions yields a sitewide large-emitter leak frequency of about 1.5 percent of
sites with large emission events. For Model Plant 2, the sitewide large-emitter leak frequency
would be about 0.75 percent of sites with large emission events because there is only one large-
emitter source assumed at the site. In a typical field, there will be a combination of facility types,
with some small facilities (similar to Model Plant 2) and some larger facilities (similar to Model
Plants 3 and 4). Therefore, our 1% large-emitter leak generation rate assumption is, in general,
consistent with the large-emitter leak frequency observed by Zavala-Araiza, et al., (2017).
A summary of the model simulations conducted are summarized in Tables 11 and 12 for
Model Plants 2 and 4, respectively. Hie "baseline" OGI monitoring frequency for Model Plant 2
simulations was semiannual; the "baseline" OGI monitoring frequency for Model Plant 4
simulations was quarterly. Note that the inclusion of large-emitters significantly impacted the
expected emissions from the fugitive emission program. For Model Plant 2, annual emissions
with semiannual OGI monitoring without large-emitters were 1.58 tons/yr, while including large-
emitters, the annual average emissions were approximately 11.6 tons/yr, or a factor of 7 times
higher. Similarly, for Model Plant 4, annual emissions with quarterly OGI monitoring without
11 Zavala-Araiza, Daniel, et al. 16 Jan 2017, https://doi.org/10.1038/ncommsl4012.
23
236
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27. 2022
large-emitters were 3.53 tons/yr; including large-emitters, the annual emissions were
approximately 15.7 tons/yr, or more than a factor of 4 times higher.
For a given Monte Carlo iteration, leaks are randomly generated for the various sites
across the 5-year modeling period and the performance of the different monitoring methods
being evaluated in that simulation all encounter the same set of leaks. So, if a 120 kg/hr large
emission event is generated at Site 7 on month 2 of year 5 during a given iteration, all monitoring
methods being evaluated experience that leak and simulate the identification and repair of that
leak based on the monitoring frequency and sensitivity being assessed. For any given modeling
run. we ran the "baseline" OGI monitoring program (semi-annual for Model Plant 2 and
quarterly for Model Plant 4), the site-wide detection limit with an annual OGI survey and that
same site-wide detection limit by itself (with no annual OGI survey). Thus, each row in Tables
11 and 12 represent a consistent set of leaks generated by which the performance of the
monitoring methods can be evaluated.
Table 11. Results of Emission Simulations Results for Model Plant 2 Including Large-
Emitters
Sitewide Monitoring Program
No. of
Runs
Average Emissions (tons CHVyear) per Site
for Leak Generation Rate Levels
Frequency(Days
Between Surveys)
Detection
Limit (kg/hr)
Semiannual
OGI
Sitewide
Survey
Sitewide
Survey with
Annual OGI
Semiannual (182)
1
500
12.61
13.75
11.84
Triannual (121)
1
500
11,23
9.51
8.64
Triannual (121)
2
1500
8.38
9.13
7.09
Triannual (121)
5
500
12.18
18.89
13.04
Triannual (121)
15
500
9.33
23.47
10.35
Quarterly (91)
1
430
13.83
9.10
8.33
Quarterly (91)
2
500
12.96
9.90
8.04
Quarterly (91)
5
500
11.12
12.83
7.43
Quarterly (91)
15
778
14.19
24.08
10.80
Quarterly (91)
30
855
13.68
26.87
14.17
Bimonthly (60)
15
500
12,55
21.78
8.61
Bimonthly (60)
30
500
11.31
25.29
12.58
Monthly (30)
30
747
11.44
23.94
11.07
Monthly (30)
60
100
8.79
26.00
13.27
24
237
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
Table 12. Results of Emission Simulations Results for Model Plant 4 Including Large-
Emitters
Sitewide Monitoring Program
No. of
Runs
Average Emissions (tons Cm/year) per Site
for Leak Generation Rate Levels
Frequency(Days
Between Surveys)
Detection
Limit (kg/hr)
Quarterly
OGI
Sitewide
Survey
Sitewide
Survey with
Annual OGI
Quarterly (91)
1
2000
16.50
17.53
16.54
Bimonthly (60)
2
1000
16.60
15.69
13.92
Bimonthly (60)
4
2387
16.00
19.12
15.83
Bimonthly (60)
5
2000
15.60
19.90
15.85
Bimonthly (60)
10
1000
16.05
26.62
16.57
Monthly (30)
4
1000
16.02
15.54
13.26
Monthly (30)
5
1392
15.60
15.87
12.71
Monthly (30)
15
1000
16.82
30.05
15.78
Monthly (30)
30
1000
15.09
49.79
15.05
Because the baseline OGI monitoring program was included in each run (for a given
model plant), we could determine a long-term average emission rate for this monitoring option
across all of the various runs. However, because of the large and random impact of large-emitter
events on the emissions and monitoring method performance, we did not compare individual
sitewide monitoring results with this long-term average for baseline OGI. The high variability in
the large-emitter simulations would require even higher number of runs for each sitewide
simulation to be more directly comparable to this long-term average.
An additional finding when conducting these runs is that it was much more difficult to
assess equivalency when the sitewide monitoring frequency is not a direct integer division of the
ground-based survey frequency. Specifically, triannual frequency compared to semiannual
frequency and bimonthly frequency compared to quarterly. With these monitoring options, it is
possible, under certain instances, that some large-emitters will emit for longer periods under the
sitewide program than the baseline OGI option. For example, consider triannual surveys
conducted as an alternative to semiannual OGI monitoring. If a large-emitter begins at day 150,
30 days after the triannual survey was conducted, it will be identified in 30 days under the
semiannual monitoring frequency but could persist up to 90 days before detection under the
triannual monitoring option. In the long run, the shorter triannual monitoring frequency can
perform better than semiannual, but there is greater variability in the Monte Carlo results because
of these scenarios, making it more difficult to determine the long-term equivalency of the
programs. For these options, we conducted a higher number of runs to better assess the long-term
equivalency of the monitoring options. When the sitewide monitoring frequency is a direct
integer division of the baseline OGI survey frequency, the sitewide option will always perform
as well or better than the ground-based OGI survey as it pertains to detectable large-emitters.
25
238
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27. 2022
5.2 Sensitivity Analysis
We evaluated a small number of site-wide screening methods to see how the different
assumptions impact the results presented in Section 5.1 of this memorandum. Table 13 shows
sensitivity runs where the large-emitter leak generation rate was maintained at 1% but the
fugitive emission component leak generation rate was set to the low inputs in Table 2. When
lowering the leak generation rate for fugitive components, the site emissions go down. The main
exception to this is the 1 kg/hr site-wide detection limit at a quarterly monitoring frequency (with
an OGI backstop) for Model Plant 4. At the "mid" leak generation rate for fugitive components,
this monitoring option appeared to be equivalent to quarterly OGI, but when using the "low"
fugitive leak generation rate, the 1 kg/hr site-wide detection limit at a quarterly monitoring
frequency (with an OGI backstop) for Model Plant 4 performed significantly worse than
quarterly OGI.
We also conducted a limited number of runs with the large-emitter leak generation rate
for Model Plants 2 and 4 set to 0.5%. Since Model Plant 4 was assumed to have two large-
emitter potential sources, lowering the leak generation rate for Model Plant 4 should yield large-
emitting sites near the 1% of sites with emissions exceeding 26 kg/hr reported by Zavala- Araiza,
et al., (2017). The 0.5% large-emitter leak generation rate may also be more representative of
basins that may have fewer large-emitter sites (relative to the total number of sites in the basin).
Table 14 summarizes the sensitivity runs when using a 0.5% leak generation rate for large-
emitters. In general, with a lower large-emitter leak generation rate, lower minimum detection
thresholds would be needed for the site-level method (at a given monitoring frequency) to have
equivalent performance as OGI when compared to the equivalency matrix at a 1% large-emitter
leak generation rate.
26
239
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27. 2022
Table 13. Sensitivity Emission Simulations Results Using Low Leak Generation Rates
(0.5%) for Fugitive Emission Components and 1% Leak Generation Rate for Large-
Emitters
Sitewide Monitoring Program
No. of
Runs
Average Emissions (tons CHVyear) per Site
for Leak Generation Rate Levels
Frequency(Days
Between Surveys)
Detection
Limit (kg/hr)
Quarterly
OGI
Sitewide
Survey
Sitewide
Survey with
Annual OGI
Model Plant 2 Sensitivity Runs: LGRs of 0.5% Fugitive Components, 1% Large-Emitter
Semiannual (182)
1
500
8.42
9.44
8.26
Quarterly (91)
2
500
9.27
8.25
6.34
Quarterly (91)
5
500
9.44
11.15
6.94
Quarterly (91)
15
496
10.84
14.82
8.48
Monthly (30)
30
500
10.15
13.66
8.02
Model Plant 4 Sensitivity Runs: LGRs of 0.5% Fugitive Components, 1% Large-Emitter
Quarterly (91)
1
500
12.33
13.79
13.01
Twice a Quarter (45)
4
1111
15.64
16.62
13.67
Twice a Quarter (45)
10
500
14.56
23.24
14.07
Twice a Quarter (45)
15
1207
16.04
32.41
16.61
Monthly (30)
4
1000
16.02
15.54
13.26
Table 14. Sensitivity Emission Simulations Results Using Low Leak Generation Rates
(0.5%) for both Fugitive Emission Components and for Large-Emitters
Sitewide Monitoring Program
No. of
Runs
Average Emissions (tons Cl-U/year) per Site
for Leak Generation Rate Levels
Frequency(Days
Between Surveys)
Detection
Limit (kg/hr)
Quarterly
OGI
Sitewide
Survey
Sitewide
Survey with
Annual OGI
Model Plant 2 Sensitivity Runs: LGRs of 0.5% Fugitive Components, 0.5% Large-Emitter
Quarterly (91)
1
997
6.54
5.08
4.35
Quarterly (91)
2
984
4.90
5.48
3.72
Model Plant 4 Sensitivity Runs: LGRs of 0.5% Fugitive Components, 0.5% Large-Emitter
Monthly (30)
1
500
7.89
7.58
6.37
Monthly (30)
2
415
6.67
8.61
5.76
27
240
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27, 2022
6. References
References
Allen, D.T., D.W. Sullivan, and M. Harrison, 2015. "Response to Comment on 'Methane
Emissions from Proeess Equipment at Natural Gas Production Sites in the United States:
Pneumatic Controllers.Environmental Science & Technology, Vol. 49, No. 6, pp. 3983-
3984. March 3, 2015.
Allen, D.T., et al., 2013. "Measurements of methane emissions at natural gas production sites in
the United States." Proceedings of the National Academy of Sciences of the United States of
America. Vol. 110, no. 44. pp. 17768-17773, October 29, 2013. Available at
http://dept.ceer.utexas.edu/methane/studv. Docket Item No. EPA-HQ-OAR-2014-0831-
0006.
Bell, et al., 2017. "Comparison of methane emission estimates from multiple measurement
techniques at natural gas production pads" Elem SciAnth, 5: 79. DOI:
https://doi.org/10.1525/elementa.266.
Cusworth, et al., 2021. "Intermittency of Large Methane Emitters in the Permian Basin."
Environ. Sci. Techno!. Lett., 8, 567-573. DOI: https://doi.org/10.1021/acs.estlett.lc00173.
EPA (U.S. Environmental Protection Agency) / GRI (Gas Research Institute). 1996. Methane
Emissions from the Natural Gas Industry, Volume 8: Equipment Leaks. Prepared by K.E.
Hummel, L.M. Campbell, and M.R. Harrison, Radian International LLC, for Gas Research
Institute and U.S. Environmental Protection Agency National Risk Management Research
Laboratory. GRI-94/0257.25, EPA-600/R-96-080h" June 1996.
ERG (Eastern Research Group, Inc.) and Sage (Sage Environmental Consulting, LP). 2011.
"City of Fort Worth Natural Gas Air Quality Study: Final Report. " July 13, 2011.
Available at http://fortworthtexas.gov/uploadedFiles/Gas Wells/AirOualitvStudv final.pdf.
Howard, T., T. W. Ferrara, and A. Townsend-Small. 2015. "Sensor transition failure in the high
flow sampler: Implications for methane emission inventories of natural gas infrastructure."
Journal of the Air & Waste Management Association, Vol. 65, Issue 7, pp. 856-862. July
2015 (published online March 24, 2015).
Kuo, J., et al. 2015. "Estimation of methane emission from California natural gas industry.'"
JAWMA, Vol. 65, No. 7, pp. 844-855.
https://www.tandfonline.com/doi/full/10.1080/10962247.2015.1025924.
Pacsi, A. P., et al., 2019. "Equipment leak detection and quantification at 67 oil and gas sites in
the Western United States."Sew! SciAnth, 7: 29. DOI:
https://doi.org/10. 1525/elementa.368.
28
241
-------
Technical Memorandum - Modeling Fugitive Emissions from Production Sites Using FEAST
July 27. 2022
Ravikumar, et al.. 2020. "Repeated leak detection and repair surveys reduce methane emissions
over scale of years." Environ. Res. Lett. 15 034029
https://iopscience.iop.org/article/10.1088/1748-9326/ab6ael.
Rutherford, J.S., et al. 2021. "Closing the methane gap in US oil and natural gas production
emissions inventories." Nat Commun 12, 4715. DOI: https://doi.org/10.1038/s41467-021-
25017-4.
Zavala- Araiza, Daniel, et al. 2017. "Super-emitters in natural gas infrastructure are caused by
abnormal process conditions." Nat. Commun. 8. 14012.
https://doi.org/10.1038/ncomms 14012.
Zimmerle, D., et al., 2019. "Characterization of Methane Emissions from Gathering Compressor
Stations: Final Report." DOE Report http://dx.doi.org/10.25675/10217/194544. October
2019 Revision.
Zimmerle, D., et al., 2019. "Methane Emissions from Gathering and Boosting Compressor
Stations in the U.S.: Supporting Volume 3: Emission Factors, Station Estimates and
National Emissions." DOE Report http://dx.doi.org/10.25675/10217/194541. October 2019
Revision.
Zimmerle, D., et al., 2020a. "Methane Emissions from Gathering Compressor stations in the
U.S." Environmental Science & Technology. 54(12), 7552-7561, 2020. DOI:
10.1021/acs.est.0c00516
Zimmerle, D., et al., 2020b. "Detection Limits of Optical Gas Imaging for Natural Gas Leak
Detection in Realistic Controlled Conditions." Environmental Science & Technology.
54(18), 11506-11514, 2020. DOI: https://pubs.acs.org/doi/pdf/T0.1021/acs.est.0c01285.
29
242
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-452/R-22-006
Environmental Protection Health and Environmental Impacts Division November 2022
Agency Research Triangle Park, NC
243
------- |