Inventory of US Greenhouse Gas Emissions and Sinks 1990-2018: Updates Under Consideration for Natural Gas Gathering and Boosting Station Emissions

November 2019

Inventory of U.S. Greenhouse Gas Emissions and Sinks 1990-2018:

Updates Under Consideration for Natural Gas Gathering & Boosting Station Emissions

EPA has recently considered reported data from EPA's Greenhouse Gas Reporting Program (GHGRP) and data
from recent studies for improving the Inventory of U.S. Greenhouse Gas Emissions and Sinks (GHGI) methodology
for estimating emissions from the gathering and boosting (G&B) segment, including stations and pipelines.

Two memos developed in support of EPA's 2019 GHGI described updates considered and implemented in the final
2019 GHGI:

• Inventory of U.S. GHG Emissions and Sinks 1990-2017: Updates Under Consideration for Natural Gas
Gathering & Boosting Emissions (October 2018 G&B memo)1

• Inventory of U.S. GHG Emissions and Sinks 1990-2017: Updates to Natural Gas Gathering & Boosting
Pipeline Emissions (April 2019 G&B pipelines memo)2

As summarized in the April 2019 G&B pipelines memo, improvements were made in the 2019 GHGI to incorporate
newly available GHGRP data to estimate emissions from G&B pipelines, and EPA has not identified new data
sources to consider improving such estimates. Considerations for updating the G&B station methodology were
presented in the October 2018 G&B memo, but the G&B station methodology was not updated in the 2019 GHGI
in response to stakeholder feedback, which recommended waiting for expected publication of new studies on
G&B station emissions. For the 2020 GHGI, EPA continues to consider available data for potential updates to G&B
segment methodology. Specifically, EPA has reviewed recently published results from a Colorado State University-
led field campaign conducted during year 2017 to characterize emissions from G&B stations across the U.S.
(Zimmerle et al. 2019).3 This memo discusses potential GHGI improvements for estimating emissions from G&B
stations.

1 Current GHGI Methodology

The GHGI emissions calculation methodology for G&B stations was last updated in the 2016 GHGI, wherein EPA
incorporated findings from the Marchese et al. 2015 study4 to estimate station-level emissions and national
activity data. EPA's April 2016 memo Inventory of U.S. GHG Emissions and Sinks 1990-2014: Revision to Gathering
and Boosting Station Emissions (2016 G&B memo)5 and April 2017 memo Inventory of U.S. Greenhouse Gas
Emissions and Sinks 1990-2015: Revisions to Natural Gas and Petroleum Systems Production Emissions (2017
Production memo)6 document the historical considerations and the full methodology used for G&B stations in the
current GHGI.

In summary, the current GHGI estimates G&B station emissions based on estimated station counts in each year
paired with station-level emission factors (EFs) for normal events (documented in the 2016 G&B memo) and EFs
for episodic events, i.e., blowdown sources (documented in the 2017 Production memo). The total G&B station
count in each year of the time series is estimated as the marketed onshore gas production in the given year
(published by the Department of Energy's Energy Information Administration (DOE/EIA)) divided by the year 2012
throughput per station from the Marchese et al. 2015 study. The current GHGI pairs this station count activity
data with a station-level CH4 EF for normal vented and leak emissions calculated using data from the Marchese et
al. 2015 study. The current GHGI separately estimates episodic event emissions using a station-level CH4 EF from

1 https://www.epa.gov/sites/production/files/2018-10/documents/ghgi_2018stakeholders_boosting.pdf

2 https://www.epa.gOv/sites/production/files/2019-04/documents/2019_ghgi_update_-_gb_segment_2019-04-09.pdf

3 Zimmerle, Daniel et al., Characterization of Methane Emissions from Gathering Compressor Stations. Available at
https://mountainscholar.org/handle/10217/195489. October 2019.

4 Marchese, A. J. et al., Methane Emissions from United States Natural Gas Gathering and Processing. Environmental Science & Technology,
49, 10718-10727. 2015.

5 https://www.epa.gov/sites/production/files/2016-08/documents/final_revision_gb_station_emissions_2016-04-14.pdf

6 https://www.epa.gov/sites/production/files/2017-04/documents/2017_ng-petro_production.pdf

Page 1 of 18

-------
November 2019

Marchese et al. 2015. The current GHGI estimates C02 emissions from G&B station normal and episodic events
using C02 EFs developed by applying a default production segment ratio of C02-to-CH4 gas content, and as such
does not fully account for C02 from sources such as flaring or acid gas removal units (AGRUs).

2 Available Data

This section summarizes data sources that EPA has recently reviewed to develop preliminary approaches and
considerations toward updating the GHGI methodologies for G&B stations.

2.1 GHGRP Data

Subpart W of the EPA's GHGRP collects annual activity and emissions data on numerous sources from natural gas
and petroleum systems that meet a reporting threshold of 25,000 metric tons of C02 equivalent (mt C02e)
emissions. Facilities that meet the subpart W reporting threshold have been reporting since reporting year (RY)
2011; however, certain sources, including subpart W emissions from G&B facilities, were first required to be
reported in RY2016. Subpart W activity and emissions data are currently used in the GHGI to calculate CH4, C02,
and N20 emissions for many sources throughout natural gas and petroleum systems.

Subpart W specifies facility definitions specific to certain segments. G&B facilities in subpart W are each defined
as a unique combination of owner/operator and basin of operation. Subpart W does not delineate data for G&B
stations versus pipelines. However, the data are reported on an emission source level, so each source can be
assigned as likely occurring at either G&B stations or pipelines. For the preliminary analyses in this memo specific
to G&B station estimates, EPA excluded emissions from blowdown vent stacks under the "pipeline venting"
emission source and from pipelines under the "equipment leaks" emission source, and considered all other data
(including some types of blowdown venting and numerous other sources) as occurring at G&B stations. Appendix
A documents the subpart W calculation methodologies for each G&B station emission source.

The GHGRP data used in the analyses discussed in this memo are those reported to the EPA as of August 19, 2018.
EPA is currently assessing data for RY2018 which became available in October 2019. In previous years,
stakeholders have suggested additional or alternate uses of GHGRP data, such as for certain sources using
measurement data only. Stakeholders have also suggested modifications to the reported GHGRP data for use in
the GHGI, such as through removal of stakeholder-identified outliers. In the current GHGI, EPA uses the publicly
available GHGRP data set without modification for the GHGI, to ensure transparency and reproducibility of GHGI
estimates. Prior to public release of the GHGRP data, the EPA has a multi-step data verification process for the
data, including automatic checks during data-entry, statistical analyses on completed reports, and staff review of
the reported data. Based on the results of the verification process, the EPA follows up with facilities to resolve
identified potential issues before public release.

Analyses of available GHGRP data and considerations for applying such data in GHGI updates are further detailed
in Section 3.1.

2.2 Zimmerle et udy

During June through November 2017, Zimmerle et al. carried out methane (CH4) measurements at G&B stations
aiming to better characterize emissions at the component, equipment, and national levels. In the Zimmerle et al.
study, the team obtained measurements at 180 facilities in 11 U.S. states. The study noted that these facilities
were operated by nine companies that represented 35% of G&B compressors reported to GHGRP at the time. The
study team aimed to select stations representative of the U.S. gathering sector in terms of size, geographic
distribution, gas composition, and equipment mix. For vented and leak emissions identification and measurement,
the team used optical gas imaging (OGI), Bacharach® HI FLOW® Sampler (BHFS), and bagging if flow exceeded
BHFS capacity (occurred for less than 1% of samples). The team conducted emission measurements on 1,938
major equipment units (compressors, dehydrators, separators, tanks, AGRUs, and yard piping); additionally, the

Page 2 of 18

-------
November 2019

team counted components on 1,002 major equipment units. At the same time, under the same U.S. Department
of Energy funding program, a separate team led by GSI Environmental, Inc. conducted a smaller field study that
entailed performing component-level measurements at four compressor stations in southeast Texas; the GSI
study team provided results that were incorporated into the larger Zimmerle et al. study.7

This study produced several products that are useful to inform potential GHGI updates: component-level leaker
and population EFs (analogous to those prescribed in the current GHGRP); population EFs for major equipment;
and a calculated estimate for year 2017 national emissions, using both study results and GHGRP data. This study
also developed and field-tested two measurement methods to better characterize emissions from unburned CH4
entrained in compressor engine exhaust ("combustion slip") and vented and leak CH4 emissions from gas-
powered, pneumatically actuated valves and controllers. Considerations for applying the Zimmerle et al. study
results in GHGI updates are further detailed in Section 3.2.

3 Updates Under Consideration

This section summarizes EPA's previous analyses of GHGRP data and explores additional considerations for
incorporating recent data for G&B stations into the 2020 GHGI.

3.1 Analysis of Available GHGRP Data

EPA's October 2018 G&B memo presents a detailed analysis of GHGRP data and considerations for using GHGRP
data to update the GHGI G&B station emissions calculation methodology. This section highlights certain topics
from that analysis and summarizes the latest available GHGRP data.

3.1.1 Source-Level Emissions and Activity

Table 1 shows year 2017 reported subpart W G&B station source-level emissions (sorted descending by RY2017
CH4 emissions), activity, EFs calculated from the subpart W data, and compares the total reported subpart W
emissions and 2019 GHGI emissions.

Table 1. G&B Station Source-Level Emissions and Activity Data and Calculated EFs from Subpart W and
G&B Station National Total Emissions from 2019 GHGI, Year 2017

Emission Source

Reported Emissions and Activity

Calculated EFs (mt/yr/unit activity)

CH4 Emissions
(mt)

CO2 Emissions
(mt)

Count

CH4 ef

CO2 EF

EF Activity
Basis

Pneumatic Controllers

196,590

14,571

141,796

1.4

0.10

per controller

Low-bleed Controllers

5,983

366

34,502

0.17

0.011

per controller

Intermittent-bleed
Controllers

159,903

12,215

101,071

1.6

0.12

per controller

High-bleed Controllers

30,704

1,991

6,223

4.9

0.32

per controller

Equipment Leaksa

104,596

12,300

Meters/piping -169,187
Separators - 33,151
Compressors -16,118
Dehydrators - 3,169
Heaters -1,677
Headers -110
Wellheads -17

n/a

Tanks

92,284

589,125

33,500

2.8

17.6

per tank

Blowdown Vent Stacks b

65,871

8,604

476,895

0.13

0.010

per event

7 GSI Environmental, Inc. et al., Integrated Component-specific Measurements to Develop Emission Factors for Compressors and Gas
Gathering Lines. Available at https://www.netl.doe.gov/sites/default/files/2019-01/FE0029084_Flnal.pdf. 2019.

Page 3 of 18

-------
November 2019

Emission Source

Reported Emissions and Activity

Calculated EFs (mt/yr/unit activity)

CH4 Emissions
(mt)

CO2 Emissions
(mt)

Count

CH4 EF

CO2 EF

EF Activity
Basis

Dehydrators

48,720

699,029

3,062

14.1

201.7

per dehydrator

Large Dehydrators

48,009

691,124

2,958

16.2

233.6

per dehydrator

Small Dehydrators

711

7,905

507

1.4

15.6

per dehydrator

Centrifugal Compressors

39,629

4,795

172

230.4

27.9

per compressor

Combustion Slip c

29,174

n/a

2,091

14.0

n/a

per unit

Pneumatic Pumps

21,434

1,563

12,777

1.7

0.12

per pump

Flare Stacks

8,759

2,142,787

4,383

2.0

488.9

per stack

Reciprocating Compressors

2,839

433

15,723

0.18

0.028

per compressor

AGRUs

n/a

486,213

139

n/a

3,497.9

perAGRU

Subpart W Reported Totald

609,896

3,959,422

n/a

n/a

n/a

n/a

National Total (2019 GHGI)e

2,218,773

239,459

n/a

n/a

n/a

n/a

n/a - Not applicable.

a - Includes all emissions reported by G&B facilities under the equipment leaks reporting section, except for emissions attributed to
gathering pipelines. The reported equipment counts are only for those facilities that rely on population EFs and do not include counts
from facilities that use leaker EFs.

b - Includes blowdown emissions reported by G&B facilities for: compressors, emergency shutdowns, facility piping,
scrubbers/strainers, pig launchers and receivers, all other equipment with a physical volume greater than or equal to 50 cubic feet, and
emissions reported with flow meters. For emissions reported with flow meters, facilities do not report the corresponding count of
blowdown events, thus the calculated EFs do not include the emissions reported with flow meters (5,751 mt CH4 and 3,979 mt C02).
c - Combustion C02 emissions are also reported, but such emissions are not within the scope of natural gas systems estimates in the
GHGI and are therefore not shown in this table. Emissions and activity shown are from "large" combustion units; "small" combustion
units are not required to report emissions, only counts. "Large" combustion units include internal fuel combustion units of any heat
capacity that are compressor-drivers, internal fuel combustion units that are not compressor-drivers with a rated heat capacity greater
than 1 mmBtu/hr, and external fuel combustion units with a rated heat capacity greater than 5 mmBtu/hr.

d - The G&B facility definition in subpart W does not delineate reporting by "station" versus "pipeline." Therefore, these emissions
equal the sum of reported subpart W emissions assigned to G&B stations (see footnotes a and b), as documented in Section 2.1.
e - Includes normal vented and leak emissions (2,018,566 mt CH4 and 231,123 mt C02 in 2017) and episodic event emissions (200,207
mt CH4 and 8,336 mt C02 in 2017).

3.1.2 Considerations for Using GHGRP Data in the 2020 GHGI

To estimate the degree of national coverage represented by the subpart W G&B emissions and to consider an
approach for scaling to the national level for use in the GHGI, EPA compared the quantity of gas received
(reported under subpart W by G&B facilities) to the total amount of gas produced from wells (estimated from
EPA's analysis of Drillinglnfo data8) at the basin-level. This approach would result in subpart W G&B station
emissions being scaled up by a factor of approximately 1.07 for RY2017.

EPA also analyzed the subpart W calculation methodology and reported emissions from reciprocating compressor
seal and valve leakage and compressor engine exhaust (i.e., combustion slip) to assess differences between total
reported subpart W emissions and current national GHGI estimates. EPA reviewed available data and developed
alternative EFs for these sources for consideration in updating the GHGI. The estimated national emissions based
on subpart W data, and using alternative EFs for these sources, were more consistent with GHGI estimates-
leading EPA to request stakeholder feedback and additional data on these sources for GHGI update
considerations. In the time since this analysis was presented in the October 2018 memo, the Zimmerle et al. 2019
study has provided published measurement data specific to G&B compressors and G&B segment-specific EFs (see
Sections 2.2 and 3.2).

8 The activity data methodologies for several upstream emission sources within natural gas and petroleum systems rely on EPA's analyses
of the subscription-based digital Dl Desktop raw data feed. This data set is referred to throughout this memo as "Drillinglnfo data."

Page 4 of 18

-------
November 2019

3.2 Analysis of Zimmerle et al. 2019 G&B Study Data and Approach to Calculate
National Emissions

Zimmerle et al. presented an approach in their study to calculate national emissions for potential use in the GHGI.
The Zimmerle et al. approach relies on EFs developed from the study's field measurements, EFs developed from
subpart W data, and activity data derived from study partner and subpart W data.

3.2.1 Source-level Emission Factors

Zimmerle et al. calculated major equipment EFs for six major equipment sources measured during the field
campaign: compressors, tanks, yard piping, dehydrators, separators, and AGRUs. Each of these sources generally
align with major equipment units reported to GHGRP for the G&B segment (exceptions noted below), and the
Zimmerle et al. study EFs include leak emissions and vented emissions.9 In developing EFs based on study
measurements, Zimmerle et al. accounted for contribution from emissions that were detected but were too large
to be measured in the field (referred to as "large emitters" or "super emitters" within the study) by referencing
emissions data from previous studies. Table 2 below presents the CH4 EFs (metric ton (mt) per year, per unit
activity) calculated from the study measurements data, for use in estimating national G&B station emissions.

As noted above, there are two emission sources that do not exactly align between the GHGRP and Zimmerle et al.
definitions: compressors and yard piping. For compressors, the Zimmerle et al. study encountered mostly natural
gas-fueled engine-driven reciprocating compressors and developed a single EF from all measurements
(approximately four percent of which were made on centrifugal or screw compressors) that is applied to a
GHGRP-based count of total gas-driven compressors—expecting the mix of compressors in industry to not
substantially differ from what was encountered during the study. Therefore, the "Compressor L&V" emission
source in Table 2 below aligns with the combined GHGRP categories of reciprocating and centrifugal compressors
shown in Table 1. Section 3.2.5 details considerations regarding potential incorporation of the study's approach
for estimating compressor emissions in the GHGI. For yard piping, the Zimmerle et al. study developed a per-
station EF; the "Yard Piping L&V" emission source in Table 2 below aligns with a subset of the GHGRP category for
equipment leaks (i.e., meters and piping emissions) shown in Table 1.

For blowdowns, dehydrator vents, and flares, Zimmerle et al. suggest application of EFs calculated from subpart
W data (see Table 1). For combustion slip, Zimmerle et al. collected study measurement data from which EFs were
calculated (refer to separate discussion in Section 3.2.3). For pneumatic controllers, Zimmerle et al. collected
study measurement data (see Section 3.2.4) but suggest using the subpart W rule-prescribed EFs. Zimmerle et al.
did not specifically address pneumatic pump emissions in their measurement campaign or national emissions
estimation approach, but EPA would include such emissions in the GHGI, see Section 3.2.5.

Table 2. Source-level EFs from the Zimmerle et al. Study

Emission Source

Source Description

CH4 EF
(mt/yr/unit activity)

Compressor L&V

Leaks and vented emissions from the compressor, compressor
driver, and any auxiliary equipment attached to the skid

15.93 per compressor

Tank L&V

Leaks and normally operating vents

5.67 per tank

Yard Piping L&V

Fuel gas systems, station inlet and outlet headers, meter runs,
and pig launchers and receivers

12.55 per station

Dehydrator L&V

Leaks and normally operating vents (not including dehydrator
reboiler vents)

0.50 per dehydrator

Separator L&Va

Leaks and normally operating vents

0.09 per separator

AGRUL&V

Leaks and normally operating vents (not including heater
combustion stacks)

0.61 per AGRU

9 The Zimmerle et al. study uses the acronym "F&V" (fugitive and vented) to describe these emissions data. However, for consistency with
GHGI terminology, the acronym "L&V" (leak and vented) is used throughout this memo.

Page 5 of 18

-------
November 2019

a - Accounts for emissions from separators that are not on a compressor skid, referred to as 'station' separators in the
Zimmerle et al. study. Emissions from separators that are on a compressor skid are included within the compressor L&V
EF.

3.2.2 National Activity Data

In the Zimmerle et al. study's recommended approach for estimating national emissions, the CH4 EFs are paired
with GHGRP-based activity data to estimate national G&B station emissions. To develop a national estimate using
GHGRP reported activity, the study developed two key methodological steps:

1. Estimating counts of stations and separators. Subpart W G&B segment facilities are not required to
report counts of stations. Subpart W G&B segment facilities report total separator counts, and Zimmerle
et al. examined two populations of separators (separators not on a compressor skid and separators on a
compressor skid). Therefore, to calculate activity data for stations and subsets of separators, the study
used partner data to develop activity factors (AFs). Taking into account various basin-level considerations,
the study estimated a national average AF of 2.8 compressors per station, leading to an estimated 5,683
stations reporting under subpart W in RY2017. Also considering basin-level analyses, the study data leads
to an estimate of 2.04 separators per station.10 Note, this separator AF estimates the number of 'station'
separators and does not consider separators on a compressor skid (see footnote a to Table 2).

2. Scaling reported counts to national total activity. Subpart W reporting reflects data only from facilities
that meet the reporting threshold (see Section 2.1). Therefore, Zimmerle et al. developed a factor to scale
up reported activity to estimate national total activity. Zimmerle et al. used basin-level production data
from subpart W (reported quantity of gas produced for sales) and Drillinglnfo to estimate that 7.5% of
stations are not reported to the GHGRP, for a scaling factor of 1.075. Note, this scaling factor approach is
similar to that previously developed by EPA; see Section 3.1 which documents an approach EPA
considered to scale RY2017 subpart W reported G&B station emissions by a factor of 1.07 to estimate
national emissions.

Table 3 below presents the estimated activity based on the study's approach for use in pairing with the applicable
EFs in Table 1 and Table 2 to estimate national G&B station emissions. The activity applicable to combustion slip
are discussed in Section 3.2.3.

Table 3. Source-level National Activity Based on Zimmerle et al. Study Approach (Year 2017)

Emission Source

Activity Basis

Reported
Activitya

National Activity
Estimate b

Compressor L&V

GHGRP reported # reciprocating and centrifugal compressors

15,895

17,081

Tank L&V

GHGRP reported # tanks

33,500

35,999

Pneumatic
Controllers

GHGRP reported # controllers subcategorized by bleed type:
low bleed, LB; intermittent bleed, IB; high bleed, HB

LB: 34,502
IB: 101,071
HB: 6,223

LB: 37,076
IB: 108,610
HB: 6,687

Blowdowns

GHGRP reported # unit blowdown events

476,895

512,469

Dehydrator Vents

GHGRP reported # dehydrator units

Large: 2,958
Small: 507

Large: 3,179
Small: 545

Yard Piping L&V

# Stations calculated from GHGRP reported # compressors and
study assumption of # compressors per station

5,684

6,108

Pneumatic Pumpsc

GHGRP reported # pumps

12,777

13,730

Flares

GHGRP reported # flare stacks

4,383

4,710

Dehydrator L&V

GHGRP reported # dehydrator units

3,465

3,723

Separator L&V

# Separators calculated from GHGRP reported # compressors
and study assumption of # separators per compressor

11,593

12,466

AGRUL&V

GHGRP reported ttAGRUs

139

149

10 EPA calculated 2.04 separators per station by using two AFs from the Zimmerle et al. study: 0.73 separators per compressor times 2.8
compressors per station equals 2.04 separators per station.

Page 6 of 18

-------
November 2019

a - GHGRP reported counts as also shown in Table 1. Station and separator counts are not directly reported but are estimated using
reported compressor counts and study-developed ratios,
b - Scaling factor of 1.075 is applied to estimate national total activity.

c - Pneumatic pumps were not included in the Zimmerle et al. study approach, but activity data are provided for reference because EPA
would include pneumatic pump emissions in the GHGI, see Section 3.2.5.

3.2.3 Combustion Slip

In addition to analyzing leak and vented emissions from certain sources as discussed above, the Zimmerle et al.
2019 study characterizes emissions from unburned CH4 entrained in G&B compressor engine exhaust
("combustion slip").

The Zimmerle et al. approach for characterizing combustion slip from G&B compressors included conducting
standard stack testing and a newly developed in-stack tracer measurement method. Zimmerle et al. conducted
measurements (generally "as found") on 116 reciprocating compressor drivers at 51 G&B stations, including 70
four-stroke lean burn (4SLB) engines and 46 four-stroke rich burn (4SRB) engines. These types of engines were
considered representative of the vast majority of compressor drivers at G&B stations (versus electric motors or
turbines). Zimmerle et al. compared their measured emission rates to EPA AP-42 EFs for 4SLB and 4SRB engines
and found general agreement. The Zimmerle et al. study did not provide an EF that could be directly applied in the
GHGI. As such, EPA calculated a combustion slip CH4 EF, shown in Table 4, by dividing the study's national
combustion slip emissions by its national engine estimate.

Table 4. Compressor Engine Combustion Slip EF Calculated from Zimmerle et al. Study Data

Combustion Slip EF
(mt ChU/engine)

204

The Zimmerle et al. study's approach to estimate a national engine count relied on the reported subpart W
compressor counts and two assumptions (based on partner data):11

1. 50% of centrifugal compressors are driven by engines and the remainder by turbines

2. Screw compressors are not included in the subpart W compressor counts and engine-driven screw
compressors are equal to 5.18% of reported reciprocating plus centrifugal compressors

Table 5 presents the resulting national estimate of engines that would be applied with the combustion slip EF.

Table 5. Combustion Slip National Activity (Compressor Engines) Based on Zimmerle et al. Study

Approach(Year 2017)

Emission Source

Activity Basis

Reported/Estimated
Activitya

National Activity
Estimateb

Combustion Slip

Reciprocating compressor engines (GHGRP reported
# reciprocating compressors)

15,723

17,880

Centrifugal compressor engines (50% of GHGRP
reported # centrifugal compressors)

Screw compressor engines (5.18% of GHGRP reported
# reciprocating and centrifugal compressors)

824

Total Engines

16,633

a - Uses GHGRP reported counts as shown in Table 1.
b - Scaling factor of 1.075 is applied to estimate national total activity.

11 These assumptions were detailed in Supporting Volume 3 of the original Zimmerle et al. 2019 study, but were not included in the October
2019 Revision of Supporting Volume 3. EPA will also consider information in an upcoming combustion slip article from the study team
(Vaughn et al.) to estimate a national engine count.

Page 7 of 18

-------
November 2019

3.2.4 Pneumatic Controllers

Zimmerle et al. conducted long-term measurements (average of 76 hours each) of 72 pneumatic controllers at
G&B stations to better understand emission rates. A separate journal article (Luck et al. 2019) provides additional
details and discussion of the Zimmerle et al. study results.12 Of the measured population, 42% exhibited abnormal
emissions. Average emission rates for abnormally operating controllers, normally operating controllers, and
overall average emission rates for each type of pneumatic controller are summarized in Table 6. The authors
recommended that the study data be used for qualitative understanding of pneumatic controller behavior rather
than for developing emission factors. For GHGI updates, EPA is considering applying the pneumatic controller
subpart W-based EFs in Table 1, as detailed in Section 3.2.5.

Table 6. Zimmerle et al. 2019 Study Pneumatic Controller Measurements

Controller Bleed
Type

Measured

Abnormally Operating
Controllers

Normally Operating Controllers

Overall Average
CH4 Emission
Rate (mt/yr)a

Number
[% of total]

CH4 Emission
Rate (mt/yr)a

Number
[% of total]

CH4 Emission
Rate (mt/yr)a

Low Bleed

5 [21%]

4.5

19 [79%]

0.1

1.0

Intermittent

25 [63%]

2.1

15 [37%]

0.4

1.5

High Bleed

0 [0%]

n/a

8 [100%]

2.6

n/a - Not applicable.

a - EPA converted from whole gas emission rate in scfh as reported in the study to CH4 emission rate in mt/yr using the GHGI default
onshore production segment methane content of 78.8% and 8,760 operating hours.

3.2.5 Considerations for Using Zimmerle et al. 2019 Data in the 2020 GHGI

Figure 1 shows year 2017 G&B station CH4 emissions compared across three different data sources: GHGRP
subpart W data scaled to the national level, the 2019 GHGI, and the Zimmerle et al. 2019 study.13 The station
emissions are broken out by the largest contributors according to the Zimmerle et al. study. However, note that
the 2019 GHGI estimates total station emissions, not broken out by source (so the gray bar includes all relevant
vented, leak, and combustion sources other than blowdowns/episodic events).

2.5

2.0

-1.5

1/1
C

1/5 _ „

1 10

0.5

0.0

Figure 1. Comparison of Year 2017 National Total G&B Station CH4 Emissions by Source

12 B. Luck et al., Multiday Measurements of Pneumatic Controller Emissions Reveal the Frequency of Abnormal Emissions Behavior at
Natural Gas Gathering Stations. Environmental Science and Technology Letters, Apr. 2019.

13 EPA scaled subpart W data using the Zimmerle et al. scaling factor of 1.075 (see Section 0).

¦ Compressor engine exhaust

¦ Compressor seal/valve leakage

¦ Tank leaks & venting

¦ Blowdowns/Episodic events

¦ Other sources

Scaled Subpart W 2019 GHGI Zimmerle et al. 2019

Page 8 of 18

-------
November 2019

This comparison in Figure 1 highlights several considerations for incorporating the Zimmerle et al. study results
and recommended approach into GHGI updates:

1. The national emissions estimated in the Zimmerle et al. study for year 2017 are lower than the current GHGI
estimate—which is driven by lower estimated average station-level emissions (versus, for example, a lower
estimated national station count; in fact, the Zimmerle et al. study estimates a higher station count than the
current GHGI). Zimmerle et al. suggested the following reasons for the differences in emissions, highlighting
differences between the current Zimmerle et al. study and the earlier Marchese et al. study that underlies the
current GHGI:

• The Zimmerle et al. study uses an updated and possibly more representative mix of stations in terms of
throughput and complexity (e.g., the partner population in the Marchese et al. study indicated a smaller
proportion of less complex stations: 30% compression-only stations versus 60% in the Zimmerle et al.
study; the Marchese et al. study sampled approximately 700 stations from 4 partner companies versus
over 1,700 stations from 9 partner companies in the Zimmerle et al. study).

• The Zimmerle et al. study accessed activity data from the GHGRP, which were not available to the
Marchese et al. study, and which represented data from a large set of operators for the entire U.S.

• The two studies utilized different measurement methods.

• There may have been operational improvements to G&B stations and/or construction of new lower-
emitting stations during the intervening four years between studies due to increased attention to CH4
emissions across the natural gas value chain.

2. The national emissions estimated in the Zimmerle et al. study for year 2017 are higher than as-reported
subpart W emissions and scaled up subpart W emissions—EPA therefore analyzed source-level emissions to
better explain the factors driving differences between subpart W and Zimmerle et al. study estimates:

• Compressor driver combustion slip: The Zimmerle et al. study emissions exceed GHGRP emissions for
compressor driver combustion slip by a factor of approximately 12. The GHGRP methodology relies on EFs
that are based on data more representative of turbines than internal combustion engines; internal
combustion engines generally have significantly higher emissions and are the prominent type of
compressor driver in the G&B segment. Refer to EPA's October 2018 G&B memo for more information on
this issue.

• Reciprocating compressor vented and leak emissions: The Zimmerle et al. study emissions exceed GHGRP
emissions by a factor of approximately 90. The GHGRP prescribed EF for these sources is based on small
upstream compressors measured in the 1996 GRI study, whereas G&B compressors are much larger; refer
to EPA's October 2018 G&B memo for more information on this issue. Additionally, the Zimmerle et al.
study approach estimated a significant contribution from "large emitters" at reciprocating compressors.

• Tank vented and leak emissions: The Zimmerle et al. study emissions are approximately twice the GHGRP
emissions. The Zimmerle et al. study approach estimated a significant contribution from "large emitters"
at tanks, which includes emissions caused by upstream malfunctions such as failed separator dump valves
(such events are included in GHGRP required reporting to the extent that quantitative estimates are
possible).

For pneumatic controllers, EPA's approach under consideration for the GHGI would use as-reported subpart W
data from G&B facilities to calculate year-specific EFs (see Table 1), in order to reflect gas CH4 content, operating
hours, and year-to-year variation. This approach would be consistent with the Zimmerle et al. study
recommended approach for calculating EFs from reported subpart W data for other emission sources not
measured in their study (i.e., blowdowns, dehydrator vents, and flares). EPA would additionally include pneumatic
pump emissions estimates (not addressed in the Zimmerle et al. study) in the GHGI based on reported subpart W
data.

Page 9 of 18

-------
November 2019

For compressors, as described in Section 3.2.1, Zimmerle et al. calculated a single leak and vent EF and suggested
applying the EF to the combined total of gas-driven compressors in the G&B segment. Table 7 below is a
replication of Table S3-16 from the Zimmerle et al. study supporting information that documents the types of
compressors and drivers measured during the study and underlying the suggested EF. The majority of emissions
measurement data underlying the EF were collected from reciprocating compressors, and as noted in the bullet
above, this new EF is likely more representative of G&B reciprocating compressor emissions than the subpart W
EF. However, G&B centrifugal compressors subject to subpart W apply a higher CH4 EF (230 mt/compressor/yr,
see Table 1) compared to the Zimmerle et al. compressor L&V EF of 17.7 mt/compressor/yr (see Table 2). Few
centrifugal compressors operate within the G&B segment, approximately 1 percent of compressors reporting to
subpart W (see Table 1). Zimmerle et al. created a single compressor EF as a simplification step, due to a lack of
data and the likely minimal impact that applying a centrifugal compressor-specific EF would have based on the
current G&B compressor population. In Section 6 below, EPA seeks stakeholder feedback on an approach that
uses the newly developed Zimmerle et al. study compressor EF versus an alternative approach such as that used
in the GHGRP that would assign distinct EFs for reciprocating and centrifugal compressors in the G&B segment.

Table 7. Compressor Driver Type by Compressor Type for Zimmerle et al. Measured Compressors

Compressor Driver

Compressor Type

Total

Percent of
Total

Centrifugal

Reciprocating

Screw

Electric motor

2.9%

Reciprocating engine

402

433

95%

Turbine

1.8%

Total

403

454

100%

Percent of Total

3.1%

89%

8.1%

100%

EPA is considering incorporating the Zimmerle et al. study data and general approach to estimating national
emissions into the 2020 GHGI to update G&B station emission estimates. Section 4 discusses additional
considerations regarding variability across the time series and geographic regions. Section 6 outlines requests for
stakeholder feedback on the Zimmerle et al. study-based updates under consideration.

3.3 Other Recent Research Studies with G&B Station Emissions Data

In addition to analyzing subpart W and Zimmerle et al. 2019 study data for comparison to GHGI estimates, EPA
reviewed findings from recent research studies which provide station-level EFs that can be directly compared to
the current GHGI EF:

• Vaughn et al. (2017). Comparing facility-level methane emission rate estimates at natural gas gathering
and boosting stations.

• Yacovitch et al. (2017). Natural gas facility methane emissions: measurements by tracer flux ratio in two
US natural gas producing basins.

• Alvarez et al. (2018). Assessment of methane emissions from the U.S. oil and gas supply chain.

The Vaughn, et al. (2017) study calculated two station-level EFs, shown in Table 8. Both EFs are higher than the
current GHGI EF, the degree to which depends on whether tank venting (that was observed at two stations) is
included in the Vaughn et al. station-level EF.

The Yacovitch et al. (2017) study calculated EFs for two regions, the Fayetteville shale play and Denver-Julesburg
(DJ) Basin; Table 8 presents the study results. The station-level emission rate for the DJ Basin is lower than the
Fayetteville shale play (note that the statistical mode of the EFs were presented in the study, rather than average
EFs); this emphasizes the existence of regional variation in station emissions. Yacovitch et al. (2017) also
presented confidence intervals around their study data. The confidence intervals encompass the current GHGI EF.
The Yacovitch et al. (2017) study also summarized results from prior studies (shown as "Multi-Basin: Tracer Sites"
in Table 8), which are included for reference.

Page 10 of 18

-------
November 2019

The Alvarez et al. (2018) study synthesized results from recent measurement studies to estimate national G&B
station emissions as 2,100 Gg CH4 in year 2015 (compared to the 2018 GHGI estimate of 1,968 Gg CH4). Their
approach analyzed data from the Mitchell et al. 2015 G&B study (underlying the Marchese et al. 2015 study) and
from a Zavala-Araiza et al. 2015 study to calculate an effective average EF that is approximately 10% higher than
the Marchese et al. EF used in the current GHGI, as shown in Table 8 below.

Table 8. G&B Station CH4 Emission Rates from Recent Studies Compared to the Current GHGI

CH4 Emission Rate

Parameter

(kg/h)

Vaughn et al. 2017

Station EF, excluding tank venting

50.4

Station EF, including tank venting

74.5

Yacovitch et al. 2017

Multi-basin: tracer sites mode EF

[95% confidence interval]

[12-3,300]

Fayetteville study area mode EF

[95% confidence interval]

[15-730]

DJ study area mode EF

[95% confidence interval]

[4.5-75]

Alvarez et al. 2018a

Station EF, excluding episodic events

Station EF, including episodic events

Zimmerle et al. 2019

Station EF, including episodic events

27.8

2019 GHGI

Station EF, excluding episodic events

Station EF, including episodic events

a - Station-level factors not presented in Alvarez et al. 2018, estimated
here from discussion text in Alvarez et al. 2018.

EPA seeks stakeholder feedback on whether and how to incorporate data from recent studies into the 2020 or
future GHGI methodologies; refer to Section 6 for specific questions. Additionally, Appendix A summarizes the
general approach (e.g., measurement methods, representativeness) of each study.

4 Regional Variability and Time Series Considerations

Stakeholders have previously suggested that differences due to regional and temporal variability should be
considered when updating GHGI methodologies, particularly for sources where variation is expected. The EPA is
similarly considering whether and how to represent regional variability in G&B emissions. EPA is specifically
considering the following regarding EFs and activity data to estimate G&B station emissions in the GHGI.

Station and Separator Count Ratios. Subpart W does not contain station counts and Zimmerle et al. examines two
populations of separators (see separator discussion in Section 3.2.1 and 3.2.2). To estimate the applicable counts,
Zimmerle et al. developed ratios for compressors per station and separators per compressor at the basin-level
from partner data. These ratios reflect differences in station size and configuration between basins. Zimmerle et
al. calculated a national average of 2.8 compressors per station, with a basin-level minimum and maximum of 1.8
and 5.1 compressors per station, and a national average of 0.73 separators per compressor, with a basin-level
minimum and maximum of 0.3 and 1.1 separators per compressor.14 No comparable ratio information exists

14 Reflects Zimmerle et al. partner equipment counts aggregated by basin and presented in Report Figures S3-10 and S3-11.

Page 11 of 18

-------
November 2019

across the GHGI time series. EPA is considering applying the Zimmerle et al. national average ratios to estimate
the number of stations and separators represented by subpart W data.

Scaling Factor. There are likely small-scale G&B facilities (based on the GHGRP definition of a single operator
within a single basin) that do not exceed the GHGRP emissions threshold and therefore are not reporting to
GHGRP. Zimmerle et al. conducted a basin-level analysis to develop a scaling factor to account for G&B stations
that are not reporting, and estimated that nationally, 7.5% of stations did not report to GHGRP in RY2017.
Zimmerle et al. used two key basin-level inputs in that estimate:

• Production ratio between GHGRP production and Drillinglnfo natural gas production. Considered to
estimate coverage of GHGRP data in those basins with GHGRP reporters.

• Basins with no GHGRP reports but some Drill inglnfo production. Zimmerle et al. counted 27 basins
without GHGRP reporters in RY2017, accounting for 0.63% of all G&B stations.

EPA is considering a simplified, national-level approach to implement the Zimmerle et al. scaling factor. Ideally
EPA would update the scaling factor each year by basin. The granularity has potential utility to organizations
conducting region-specific field studies. However, given that GHGRP reporting coverage is high, the approach
under consideration is to update the scaling factor calculation at the national-level. A national-level approach
would likely end up very similar to a basin-level approach, but with level of effort differing greatly. EPA is also
considering how and whether to account for basins with no GHGRP reporters, which has minimal impact on the
scaling factor.

In addition to the above considerations, there are multiple options for implementing Zimmerle et al. data into the
GHGI time series calculations to update estimates currently based on the Marchese et al. data. To determine G&B
station counts over the time series, EPA is considering: (1) applying a Zimmerle et al. derived per station volume
across the time series (this would increase station counts across the time series), (2) using data from both studies
(e.g., using the Marchese et al. data from 1990 through 2013, Zimmerle et al. data for years 2017 and forward,
and interpolating between the two for intermediate years), or (3) maintaining the production volume per station
derived from Marchese et al. Similar considerations exist for applying the Marchese et al. based-EFs versus the
Zimmerle et al. approach to EFs over the time series. In Section 6, EPA seeks stakeholder feedback on the
approaches under consideration, including how to reflect industry segment changes over the time series that
might help explain differences between the Zimmerle et al. 2019-based estimate and current GHGI.

EPA is considering using subpart W data to calculate G&B flaring C02 and N20 emissions. Flaring emissions data
are only available for recent years, and this data may not be representative of emissions over the entire 1990-
2018 GHGI time series. To estimate flaring emissions over the time series, EPA is considering whether subpart W
EFs should be applied to all years, if flaring EFs should be set equal to zero in early years and interpolated to
recent year data, or if another approach is more reasonable.

5 Preliminary National Emissions Estimates for 2017

Table 9 presents preliminary CH4, carbon dioxide (C02), and nitrous oxide (N20) emissions by the approach under
consideration for estimating national emissions in the 2020 GHGI.

Table 9. G&B Station National Emissions Estimates for the Approach Under Consideration and 2019

GHGI Total Emissions, Year 2017

Emission Source

CH4 Emissions
(mt)

CO2 Emissions
(mt)

N2O Emissions
(mt)

Combustion Slip

364,762

n/a

Compressor L&V

272,146

32,800

n/a

Tank L&V + Flaring

204,152

633,310

2.2

Intermittent Bleed Pneumatic Controllers

171,895

13,131

n/a

Page 12 of 18

-------
November 2019

Emission Source

CH4 Emissions
(mt)

CO2 Emissions
(mt)

N2O Emissions
(mt)

Blowdowns

64,629

4,972

n/a

Dehydrator Vents

52,374

751,456

6.2

Dehy Vents - Large units

51,609

742,959

3.8

Dehy Vents - Small units

764

8,498

2.4

Yard Piping L&V

76,709

9,245

n/a

High-bleed Pneumatic Controllers

33,007

2,140

n/a

Pneumatic Pumps

23,042

1,681

n/a

Flares

9,416

2,303,496

4.3

Low-Bleed Pneumatic Controllers

6,432

394

n/a

Dehydrator L&V

1,854

223

n/a

Separator L&V

1,146

138

n/a

AGRUL&V

522,679

n/a

Total

1,281,654

4,275,666

12.7

2019 GHGI Total

2,218,773

239,459

n/a - Not applicable.

EPA calculated the national CH4 emissions by emission source shown in Table 9 by applying the general approach
outlined in the Zimmerle et al. 2019 study. This approach relies on applying a combination of GHGRP-based EFs (in
Table 1, for pneumatic controllers, blowdowns, dehydrator vents, and flares) and Zimmerle et al. study
measurement-based EFs (see Table 2), and the corresponding activity data in Table 3. As discussed in Section
3.2.5, the approach under consideration results in G&B station CH4 emission estimates lower than those
estimated in the current GHGI for year 2017.

The Zimmerle et al. 2019 study did not address C02 or N20 emissions. Emissions data for these pollutants are
collected via the GHGRP. EPA developed preliminary estimates of C02 emissions (from sources with leaks and
vents, and flares) using default C02-to-CH4 gas content ratio or GHGRP data, and N20 emissions (from flares) using
GHGRP data. Table 10 documents the CH4 EF data source for each emission source by the approach under
consideration and presents approaches that EPA is considering for C02 and N20 EFs (and used to develop the
emissions shown in Table 9). For emission sources that use CH4 EFs based on Zimmerle et al. study measurements,
EPA is considering applying a default C02-to-CH4 gas content ratio to calculate the corresponding C02 EF for leak
and vent emissions (similar to the current GHGI methodology, discussed in Section 1)—with the exception of
tanks and AGRUs. For tanks and AGRUs, EPA is considering using GHGRP data to calculate C02 and N20 EFs to
account for tank flaring emissions and the significant process C02 emissions from AGRUs. For emission sources
that use CH4 EFs based on GHGRP data, EPA is considering using GHGRP data to also calculate C02 and N20 EFs, as
applicable. Note, Table 1 shows the calculated subpart W C02 EFs for each emission source, some of which would
be used in the approach under consideration according to the table below.

Table 10. C02 and N20 EF Approaches Under Consideration by Emission Source

Emission Source

CH4 EF Source

CO2 EF Source

N2O EF Source

Combustion Slip

Zimmerle study
measurements

n/a

Compressor L&V

Zimmerle study
measurements

Default C02:CH4 ratio

n/a

Tank L&V + Flaring

Zimmerle study
measurements

GHGRP

Pneumatic Controllers

GHGRP

n/a

Blowdowns

GHGRP

n/a

Dehydrator Vents

GHGRP

Page 13 of 18

-------
November 2019

Emission Source

CH4 EF Source

CO2 EF Source

N2O EF Source

Yard Piping L&V

Zimmerle study
measurements

Default CC>2:CH4 ratio

n/a

Pneumatic Pumps

GHGRP

n/a

Flares

GHGRP

Dehydrator L&V

Zimmerle study
measurements

Default CC>2:CH4 ratio

n/a

Separator L&V

Zimmerle study
measurements

Default CC>2:CH4 ratio

n/a

AGRUL&V

Zimmerle study
measurements

GHGRP

n/a

n/a - Not applicable.

6 Requests for Stakeholder Feedback

EPA seeks stakeholder feedback on the approaches under consideration discussed in this memo and the particular
questions below.

In previous memos covering this topic, EPA also sought stakeholder feedback. The October 2018 memo
summarizes feedback from two stakeholders regarding the June 2018 memo version. Two stakeholders
responded to questions raised in the October 2018 memo (which outlined potential approaches for using GHGRP
data in GHGI updates and was released before publication of the 2019 GHGI and recent studies such as Zimmerle
etal. 2019).

Summary of Stakeholder Feedback on EPA October 2018 G&B Memo

• Two stakeholders supported consideration of studies focused on G&B emission sources before developing
GHGI EFs based solely on GHGRP data. One stakeholder specifically expressed concerns that using only
GHGRP data to estimate G&B station emissions would underestimate CH4 emissions and suggested the
current GHGI approach be maintained in the 2019 GHGI.

• A stakeholder did not support using the processing or transmission segment-specific EFs to represent
compressor vented and leak emissions in the G&B segment but supported reviewing new information
from G&B source-specific measurements expected to become available in 2019.

• A stakeholder supported disaggregating emissions by source and using source-level EFs that represent
both routine emissions and large emissions caused by abnormal conditions.

• A stakeholder supported the general approach of using GHGRP data as the basis for GHGI activity
estimates (e.g., estimating station count based on reported compressor counts).

• A stakeholder acknowledged difficulty in discerning the representation of G&B stations within reported
GHGRP data since there is no national count of G&B stations including no data on equipment at G&B
stations or their throughput. The stakeholder supported continued analysis to understand why nearly all
gathering pipeline mileage is represented by GHGRP reported information while G&B stations were
believed to be under-represented in GHGRP reporting.

• A stakeholder supported taking steps to accurately reflect CH4 emissions from compressor engine exhaust
in the GHGI.

• A stakeholder supported potential use of basin-level data for the highest emitting basins for reflecting
regional variability for G&B stations, and noted that temporal variability is reflected in many of the
emission sources reported under GHGRP (e.g., those requiring event-based data such as blowdowns).

Current Questions to Stakeholders

1. EPA seeks feedback on applying the general approach outlined in the Zimmerle et al. 2019 study to
calculate G&B station emissions for the GHGI, including:

Page 14 of 18

-------
November 2019

a. Applying EFs as presented in the Zimmerle et al. study that were calculated using recent field
measurement data and an approach for incorporating large emitters (see Section 3.2.1 and Table
2).

b. Applying EFs calculated from GHGRP data for emission sources that were not included in the
Zimmerle et al. study field campaign (blowdowns, dehydrator vents, flares, and pneumatic
controllers; see Section 3.1.1 and Table 1).

c. The use of onshore production volumes to determine the coverage of reported subpart W G&B
data, used to develop a scaling factor (see Section 3.2.2).

2. EPA seeks feedback on whether it is appropriate to apply a single EF to estimate leak and vent emissions
from the total population of gas-driven reciprocating and centrifugal compressors (as suggested in the
Zimmerle et al. study; see Sections 3.2.1 and 3.2.5), versus having separate EFs for each compressor type
(as in the GHGRP and as generally used for other GHGI industry segments).

a. If a centrifugal compressor-specific EF is used, what EF should EPA apply (e.g., subpart W EF or an
EF from another data source)?

b. Few centrifugal compressors are currently used in the G&B segment, so EPA seeks feedback on
whether this is likely to change in the future to the extent that it is valuable to show centrifugal
compressors as a unique emission source in the GHGI.

3. EPA seeks feedback on how to consider regional variability for G&B stations in the GHGI, including
whether to apply a simplified, national-level approach to determine ratios and scaling factors versus the
detailed, basin-level approach the Zimmerle et al. study developed, as discussed in Section 4.0.

4. EPA seeks feedback on how to consider temporal variability for G&B station emissions in the GHGI,
including:

a. How to apply the Zimmerle et al. approach versus Marchese et al. EFs (the basis of the current
GHGI) over the time series, as discussed in Section 4.0. Differences between the Zimmerle et al.
and Marchese et al. study EFs are discussed in Section 3.2.5.

b. How to use Zimmerle et al. data versus Marchese et al. data (the basis of the current GHGI) to
determine G&B station counts over the time series, as discussed in Section 4.0.

c. How to use subpart W data to estimate flaring emissions over the time series, as discussed in
Section 4.0.

5. EPA seeks feedback on addressing any overlap between sources in the G&B segment and in other industry
segments (e.g. onshore production and natural gas processing) between the GHGI and the GHGRP.

Page 15 of 18

-------
November 2019

Appendix A - Measurement Methodologies from Data Sources Considered for Updates

Emission Source

Measurement and/or Calculation Type

# Sources

Location &
Representativeness

EF Calculation Method

GHGRP Subpart W and Subpart C

G&B Acid gas removal (AGR)
vents

Emissions calculated from the available methods: (1) CEMS for
C02 with volumetric flow rate monitors, (2) Vent meter for C02
and annual volume of vent gas, (3) measured inlet (or outlet)
gas flow rate and inlet and outlet volumetric fraction of C02, or
(4) simulation software.

Emissions data (for 2017) are
available from 54 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Centrifugal Compressors

Emissions calculated using the count of centrifugal
compressors that have wet seal oil degassing vents multiplied
by default EF (annual volumetric flow per unit).

Emissions data (for 2017) are
available from 24 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Combustion

Emission calculations depend on the type of fuel burned:

•	If burning pipeline quality natural gas or the identified fuels
and blends (i.e., coal, coke, natural gas, petroleum
products, certain other solids and gaseous fuels,
solids/gaseous/liquid biomass fuels) then use default
subpart C EFs.

•	If burning field gas, process vent gas, or a gas blend then
determine volume of fuel combusted from company
records and use a continuous gas composition analyzer to
measure mole fraction of gas.

•	These sources are exempt: (1) external fuel combustion
sources with rated heat capacity < 5 MMBtu/hr, (2) internal
combustion sources, not compressor-drivers, with a rated
heat capacity < 1 MMBtu/hr (equal to 130 HP).

Emissions data (for 2017) are
available from 312 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Dehydrators

Emissions calculations depend on the daily throughput:

•	If daily throughput is > 0.4 million scf then use simulation
software.

•	If daily throughput is < 0.4 million scf then use EFs and a
dehydrator count

•	For dessicant dehys, use the amount of gas vented from
the dessicant vessel when it is depressurized

•	When a flare or a regenerator fire-box/fire tube is used
adjust the emissions to reflect the control efficiency.

Emissions data (for 2017) are
available from 276 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Equipment Leaks

Emissions calculated using: (1) default EFs, by source type; (2)
source type counts (rule provides default counts e.g., valves
per wellhead) including miles of gathering pipelines by
material type; (3) estimated time the source was operational;
and (4) concentration of C02 and CH4.

Emissions data (for 2017) are
available from 319 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Pneumatic Controllers

Emissions calculated using: (1) counts of continuous high
bleed, continuous low bleed, and intermittent bleed

Emissions data (for 2017) are
available from 289 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported

Page 16 of 18

-------
November 2019

Emission Source

Measurement and/or Calculation Type

# Sources

Location &
Representativeness

EF Calculation Method



controllers, (2) default EFs for each controller type, (3) annual
operating hours, and (4) GHG concentrations in vented gas.





emissions and activity data to
consider use in GHGI updates.

G&B Pneumatic Pumps

Emissions calculated using: (1) counts of pneumatic pumps, (2)
default EF, (3) annual operating hours, and (4) GHG
concentrations in vented gas.

Emissions data (for 2017) are
available from 218 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Reciprocating
Compressors

Emissions calculated using the count of reciprocating
compressors multiplied by default EF (annual volumetric flow
per unit).

Emissions data (for 2017) are
available from 313 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Tanks

Emissions calculations depend on the daily throughput:

•	If oil throughput is >10 bbl/d and the gas and liquid passes
through non-separator equipment (e.g., stabilizers, slug
catchers) before flowing to the tank, calculate C02 and CH4
emissions using simulation software or by assuming all C02
and CH4 is emitted.

•	If oil throughput is >10 bbl/d and the gas and liquid flows
directly to a tank without passing through a separator,
assume all C02 and CH4 is emitted.

•	If oil throughput is <10 bbl/d then calculate C02 and CH4
emissions from (1) counts of separators, wells, or non-
separator equipment that feed oil directly to the storage
tank and multiply by EF (annual volumetric flow per unit).

•	Subtract emissions if a VRU is used and if a flare is used
then use the flare calculation methodology.

Emissions data (for 2017) are
available from 231 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Flare Stacks

Emissions calculated using: (1) gas volume sent to the flare, (2)
combustion efficiency (from manufacturer or assume 98%),
fraction of feed gas sent to an un-lit flare, and (3) gas
composition for C02, CH4, and hydrocarbon constituents.

Emissions data (for 2017) are
available from 154 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

G&B Blowdown Vent Stacks

Emissions calculated from the available methods: (1) use
blowdown volumes, the number of blowdowns, and the ideal
gas law modified with a compressibility factor, or (2) used a
flowmeter to directly measure emissions for each equipment
type or all equipment associated with a blowdown event.

Emissions data (for 2017) are
available from 262 facilities.

Facilities in the U.S. that
exceed 25,000 mt C02e
reporting threshold.

For this memo, the EPA
evaluated the reported
emissions and activity data to
consider use in GHGI updates.

GRI/EPA 1996

Compressor exhaust

An average emission rate was calculated for each model of
compressor engine and turbine in the GRI TRANSDAT
Emissions Database, which is based on compressor tests
conducted by Southwest Research

Institute (SwRI). The emission rates were calculated from the
reported methane emissions per unit of fuel and the reported
fuel use rate for each compressor model.

86 turbines and 775
reciprocating engines

Natural gas value chain

TRANSDAT data were
combined to generate
emission factors by
correlating compressor driver
type, methane emissions, fuel
use rate, and annual
operating hours

Page 17 of 18

-------
November 2019

Emission Source

Measurement and/or Calculation Type

# Sources

Location &
Representativeness

EF Calculation Method

Vaughn et al. 2017

G&B facilities

Dual-tracer measurements, aircraft measurements, and on-site
component-level measurements (direct measurements and
simulated direct measurements) coupled with engineering
estimates using Monte Carlo model.

36 gathering stations

•	Measurements conducted
September-October 2015

•	Eastern portion of the
Fayetteville shale play
(Arkansas)

Dual-tracer measurements,
including and excluding
significant tank venting

Yacovitch et al. 2017

Production, gathering,
processing, and transmission
facilities

Dual tracer flux ratio method

•	DJ study area: 12 gathering
stations, 5 wellpads, and 4
processing plants
measured.

•	FV study area: 31 gathering
stations, 18 wellpads, and 4
transmission stations
measured.

•	Two natural gas production
regions: Denver-Julesberg
(DJ) basin and Fayetteville
shale play (FV) in Arkansas

•	Nov 2014 for DJ basin

•	Sep-Oct 2015 for FV play

Dual-tracer measurements to
calculate facility-level
emission rates and
throughput-weighted
emissions

Alvarez et al. 2018

G&B stations

Synthesized data from 3 studies: Zavala-Araiza et al. 2015,
Mitchell et al. 2015, Marchese et al. 2015

National activity estimated as
5,122 stations in year 2015

Synthesized data from 3
studies: Zavala-Araiza et al.
2015, Mitchell etal. 2015,
Marchese et al. 2015
(measurements in multiple
U.S. basins)

Adjusted the Marchese et al.
central estimate loss rates by
the ratio of the Zavala-Araiza
et al. and Mitchell et al. EFs
(59.6/54) to better account
for heavy-tail emissions

Zimmerle et al. 2019

G&B stations

For vented and leak emissions identification and
measurement, optical gas imaging (OGI), Bacharach® HI
FLOW® Sampler (BHFS), and bagging if flow exceeded BHFS
capacity (occurred for «1% of samples).

•	Measurements taken at
180 facilities on 1,938
major equipment units

•	Components counted on
1,002 major equipment
units

•	National activity estimated
at a source-level based on
GHGRP data, for a total of
6,108 stations in year 2017

•	Study aimed to select
stations representative in
terms of size, geographic
distribution, gas
composition, and
equipment mix.

•	Measurements conducted
June - November 2017.

•	180 facilities in 11 U.S.
states.

•	Nine partner companies
represented 35% of G&B
compressors reported to
GHGRP at the time

•	Study measurements for
combustion slip,
compressors, tanks, yard
piping.

•	GHGRP data for pneumatic
controllers, blowdowns,
dehydrator vents, flares,
leaks from non-compressor
equipment (e.g.,
separators).

Page 18 of 18

-------