April 2015
Inventory of U.S. Greenhouse Gas Emissions and Sinks: Potential Revisions to Liquids
Unloading Emissions Estimate
Overview of Methodology in 2013 and 2014 Inventories
Data from a 2012 report published by the American Petroleum Institute (API) and America's Natural Gas
Alliance (ANGA)1 were used beginning with EPA's 1990-2011 Inventory released in 2013 ("2013 Inventory")
to develop regional activity data and regional emission factors for gas well liquids unloading activities in
natural gas systems. For more information, see memo Overview of Updates to the Natural Gas Sector
Emissions Calculations for the Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2011, available
at http://epa.gov/climatechange/Downloads/ghgemissions/fact-sheet-oil-and-gas-estimates-in-2013-
inventory.pdf.
Potential Revisions to Liquids Unloading Estimates
Several recent studies and upcoming data from GHGRP provide new information on liquids unloading.
Below is a brief overview of the data and potential uses of the data to update the GHG Inventory estimates.
Allen et al. 2014
Allen et al. 2014 presented results of a study focused on emissions from liquids unloading.2 The Allen et al.
study measured unloading emissions from 107 natural gas wells owned by 10 study participant companies.
The 107 wells were selected from 4 regions of the country, in proportion to the population of wells that
were venting for liquids unloading in each region according to GHGRP data. The wells were selected to
represent the three most common categories of unloading methods: automatic plunger-lift, manual
plunger-lift and manual non-plunger. Using data from the emission measurements, Allen et al. developed
five emission factors for various unloading methods and well types (termed "unloading categories" in this
memorandum), in units of emissions per event. The 10 study participants were also surveyed to determine
the frequency of events per year, for each of the five unloading categories, in each of four regions of
interest. This information was combined with the 2012 GHGRP count of wells performing unloading
operations (assumed to represent all national activity) to estimate national emissions from well unloading
operations. The study estimated national methane emissions from unloading operations of 270 kilotons per
year (kt/y) with a 95% confidence interval of 190 - 400 kt/y, for the year 2012. In contrast, the GHG
Inventory estimated 2012 methane emissions of 267 kt/y from unloading activities. As shown in Table 1
below, the totals are in very close agreement, though the apportionment of emissions between non-
plunger lift and plunger lift emissions differs between the two analyses.
Table 1. Comparison of 2012 National Methane Emissions from Liquids Unloading as Reported in the
2015 GHG Inventory and the Allen et. al Study
Lift Type
2015 GHG Inventory (kt)
Allen et al. (kt)
Non-plunger
126
80
Plunger
143
190
Total
269
270
1 "Characterizing Pivotal Sources of Methane Emissions from Natural Gas Production" (September 2012). Available
online: http://www.api.Org/~/media/Files/News/2012/12-October/API-ANGA-Survey-Report.pdf
2 Allen, D.T., Sullivan, D., Zavala-Araiza, D., Pacsi, A., Harrison, M., Keen, K., Fraser, M., Hill, A.D., Lamb, B.K., Sawyer,
R.F., and Seinfeld, J.H. Methane Emissions from Process Equipment at Natural Gas Production Sites in the United
States: Liquid Unloadings, Environmental Science & Technology, 10.1021/es504016r. Available online:
http://pubs.acs.org/doi/abs/10.1021/es504016r
1
-------
April 2015
Summary of Allen et al. Approach
Allen et al. observed wide variability in measured emissions data from well unloading events. Allen et al.
conducted statistical analyses to identify well and unloading event characteristics that could explain the
variability observed and found that the variable that best explained the variability in the observed annual
well emissions was unloading frequency. Allen et al. also found a positive correlation of event frequencies
with well age (with older wells having more unloading events per year) and a negative correlation of annual
emissions with well depth (with deeper wells having lower annual emissions).
Allen et al. study results indicated consistency in per-event emissions for five unloading categories that vary
by wells with or without plunger lift and by event frequency:
• Wells without plunger lifts < 10 events per year;
• Wells without plunger lifts > 10 < 50 events per year;
• Wells without plunger lifts > 50 events per year;
• Wells with plunger lifts < 100 events per year; and
• Wells with plunger lifts > 100 events per year.
To calculate national emissions, the Allen et al. study used national per-event emission factors for the five
unloading categories above in conjunction with region-specific activity data comprising two components:
1. Regional total counts of wells in the five unloading categories. Allen et al. relied on 2012 GHGRP
data for counts of wells on a regional level that perform liquids unloading with and without plunger
lifts. The Allen et al. study then applied results from a survey of study participants to apportion the
wells into the five unloading categories (e.g., 79% of non-plunger wells in the Appalachian region
have < 10 unloading events per year).
2. Regional average annual number of events that are performed on wells in each of the five
unloading categories. Allen et al. applied results from the survey of study participants to determine
the region-specific average annual number of unloading events in each unloading category (e.g.,
non-plunger wells with < 10 unloading events per year in the Appalachian region perform 3.13
events/year on average).
The study found that the distribution of high event frequency wells (> 100 events/year) is not uniform
within the Rocky Mountain region. In particular, AAPG Basin 580 (San Juan Basin) had a much higher
fraction of high frequency plunger lift unloading than the rest of the region. Therefore, the total number of
events in the two plunger lift unloading categories was calculated for two sub-basins of the Rocky Mountain
region to avoid skewing a regional estimate with the activity observed in Basin 580.
Comparison of Allen et al. with EPA data
Currently, EPA is considering ways to better reflect regional-level emissions for liquids unloading in the GHG
Inventory. Although regional emissions were not the focus of the Allen et al. study, regional data from the
study were used to compile regional emissions. Since the Allen et al. study combines the Southwest region
with Mid-Continent and combines the West Coast region with Rocky Mountain, these same combinations
were made in order to compare the Allen et al. study results to the GHG Inventory and the GHGRP
emissions on a regional basis as shown in
Table 2 below.
2
-------
April 2015
Table 2. Regional and National Comparison of 2012 Liquids Unloading Methane Emissions
NEMS Region
2012 Liquids Unloading Emissions (MMT CO;e)
Allen et al.
2015 GHG Inventory
GHGRP
Appalachian
1.13
2.10
0.51
Gulf Coast
0.27
0.35
0.49
Mid-Continent/Southwest
1.85
2.50
1.00
Rocky Mountain/West Coast
3.78
1.78
2.97
Total
7.03
6.73
4.97
The Allen et al. study and GHG Inventory compare very closely at the national level, but not at the regional
level. At the regional level, the Allen et al. study results compare more closely with the GHGRP (which is
expected to some degree because the Allen et al. study estimates are based on regional GHGRP well count
activity data).
Potential Applicability of Allen et al. to Revising the GHG Inventory Emission Factors
The GHG Inventory applies region-specific annual emission factors to wells unloading with and without
plunger lifts. For wells without plunger lifts, the GHG Inventory regional emission factors range from 100
thousand to 2.6 million scf CH4/well/year; for wells with plunger lifts, the regional emission factors range
from 3.5 thousand to 1.4 million scf CH4/well/year.
Table 3 below presents regional emission factors developed from the Allen et al. study. The basis for these
factors is combining: (1) national average per-event emissions data (scf CH4/event) in each unloading
category that were obtained from study measurements; with (2) event frequency data (events/well/year) in
each unloading category that were obtained from the survey of study participants. The factors labeled
"Regional Average" in Table 3 take into accout the relative prevalence of wells in each unloading category
in each region based on the study participant survey.
Table 3. Regional Methane Emission Factors (scf Cm/well/year) Developed from Allen et al. Study
Unloading Category
Appalachian
Gulf Coast
Mid-Continent/
Southwest
Rocky Mountain/
West Coast
Wells without plunger lifts
< 10 events per year
67,237
62,778
63,483
60,659
10 < events < 50 per year
514,707
470,360
510,892
431,633
50 < events < 200 per year
2,352,151
3,177,691
2,611,834
2,657,585
Events > 200 per year3
-
12,105,491
14,478,647
-
Regional Average
197,954
131,489
203,466
107,402
Wells with plunger lifts
< 100 events per year
69,357
50,985
49,790
91,114
3 Allen et al. obtained activity data but not per-event emissions data for this category, and assumed the same per-
event emissions as observed for the unloading category "50 < events < 200 per year."
3
-------
April 2015
> 100 events per year
326,206
162,000
2,139,423
1,447,558
Regional Average
75,336
51,726
430,730
472,937
Table 4 below compares regional emission factors developed from the Allen et al. study to the regional
factors currently used in the GHG Inventory.
Table 4. Regional Methane Emission Factors (scf CH4/well/year) Developed from Allen et al. Study
Compared to Current GHG Inventory Factors3
Data Source
Appalachian
Gulf Coast
Mid-Continent/
Southwest
Rocky Mountain/
West Coast
Wells without plunger lifts
GHG Inventory
166,000
301,000
230,000/
97,000
2,579,000/
304,000
Allen et al.
198,000
131,000
203,000
107,000
Wells with plunger lifts
GHG Inventory
315,000
70,000
1,380,000/
4,000
154,000/
345,000
Allen et al.
75,000
52,000
431,000
473,000
a - Rounded to the nearest thousand.
In order to incorporate the Allen et al. factors into the GHG Inventory using the regional structure of the
production segment calculations, the emission factors could be assigned to the Southwest region as the
Mid-Continent region, and to the West Coast region as the Rocky Mountain region, since the Allen et al.
study did not differentiate regions to the same granularity as the current Inventory methodology.
Potential Applicability of Allen et al. to Revising the GHG Inventory Activity Data
The Allen et al. study used GHGRP data to estimate national emissions. Note that the GHGRP data do not
cover 100 percent of national activity due to the reporting threshold. Table 5 below shows activity data
from the 2015 GHG Inventory compared to GHGRP data, for year 2012.
Table 5. Regional Activity Data from Current GHG Inventory Compared to GHGRP for Year 2012
Well Type/
Data Source
NEMS Region
North
East
Gulf Coast
Mid-
Continent
Southwest
Rocky
Mountain
West Coast
National
Total
Wells without plunger lifts
GHG Inventory
18,181
2,627
5,383
10,527
1,199
179
38,096
GHGRP
7,562
4,649
3,569
3,980
6,720
33
26,513
Wells with plunger lifts
GHG Inventory
7,030
860
3,031
1,797
10,162
200
23,080
GHGRP
11,118
1,613
1,903
2,596
15,412
0
32,642
Total wells with liquids unloading
GHG Inventory
25,211
3,487
8,414
12,324
11,361
379
61,176
GHGRP
18,680
6,262
5,472
6,576
22,132
33
59,155
The results of a revision to the activity data methodology to use GHGRP data would better align the GHG
Inventory with both the Allen et al. and the GHGRP results—both on a regional level and also regarding the
overall apportionment between non-plunger lifts and plunger lifts.
Potential Impacts of Allen et al. on GHG Inventory Estimates
4
-------
April 2015
Table 6 below shows 2012 methane emissions from liquids unloading for five calculation approaches:
1. Allen et al. study estimates, which apply regional emission estimates based on unloading volumes
and frequencies observed during the study to available GHGRP activity data (note that since Allen
et al. publication, updated 2012 GHGRP data have been published);
2. Current GHG Inventory estimates, which are based on emissions estimates and event counts from a
2012 API-ANGA study4;
3. Current GHGRP emissions data;
4. Emission factors developed from Allen et al. study data as shown in
5. Table 4 applied to current GHG Inventory activity data; and
6. Emission factors developed from Allen et al. study data as shown in
7. Table 4 applied to most up-to-date GHGRP activity data (i.e., updated activity data compared to
approach 1).
Table 6. Regional and National Comparison of 2012 Liquids Unloading Methane Emissions
NEMS Region
2012 Liquids Unloading Methane Emissions (MMT CChe)
Allen et al.
Study
Current GHG
Inventory
GHGRP
EFs from Allen
et al. Study &
Inventory
Activity Data
EFs from Allen
et al. Study &
GHGRP
Activity Data
Appalachian
1.13
2.12
0.51
1.95
1.10
Gulf Coast
0.27
0.75
0.49
0.24
0.33
Mid-
Continent/Southwest
1.85
2.09
1.00
2.51
1.64
Rocky
Mountain/West
Coast
3.78
1.88
2.97
2.39
3.79
Total
7.03
6.84
4.97
7.09
6.86
API/ANGA 2012 - Analysis of National versus Regional Factors
EPA compared liquids unloading emissions data for 2011 and 2012 from the 2014 Inventory to data
reported to GHGRP as of August 18, 2014. As shown in Table 7, there were significant regional differences
between the Inventory emissions estimates and the data reported to the GHGRP for the same year. For
example, liquids unloading emissions reported to GHGRP in the Rocky Mountain and South West NEMS
regions are over twice as high as emissions for these regions in the Inventory. In contrast, Inventory
emissions for the North East and Mid-Continent regions are more than five times as high as GHGRP
emissions for these regions. EPA is considering options to address the differences between the Inventory
and GHGRP.
As discussed in the "Planned Improvements" section of the 2014 Inventory, EPA investigated the impact of
using national-average data from the API/ANGA report as opposed to the regional data. Table 8 compares
four sets of data on liquids unloading. The first column presents the data from year 2012 in the 2014
Inventory and is based on regional activity and emission factors developed from the API/ANGA study. The
second column of data is the information reported under the GHGRP for year 2012 (as of August 18, 2014).
The third column is the result of replacing regional emission factors for year 2012 in the Inventory with
4American Petroleum Institute and America's Natural Gas Alliance (API/ANGA) Characterizing Pivotal Sources of
Methane Emissions from Natural Gas Production: Summary and Analyses of API and ANGA Survey Responses, Final
Report, updated September, 2012. Available online: http://www.api.org/news-and-media/news/newsitems/2012/oct-
2012/~/media/Files/News/2012/12-October/API-ANGA-Survey-Report.pdf
5
-------
April 2015
national-average emission factors for liquids unloading with and without plunger lifts that were developed
from the API/ANGA study. The fourth column replaces the regional proportions of wells venting during
liquids unloading with the national-average proportions that were reported in the API/ANGA study in
addition to using national-average emission factors. For the national average proportions of wells venting
during liquids unloading, the wells vented with plunger lifts were tracked separately from the wells vented
without plunger lifts.
Table 7. Com
larison of 2014 Inventory and GHGRP Emissions from Liquids Unloading
NEMS Region
Methane Emissions (MT CO;e)*
GHGRP 2011
Emissions
Inventory
2011
Emissions
GHGRP 2012
Emissions
Inventory
2012
Emissions
GHGRP 2013
Emissions
North East
407,027
2,077,962
603,649
2,115,336
350,790
Mid-Continent
245,585
1,798,774
292,158
1,790,352
126,146
Rocky Mountain
4,558,417
1,857,979
3,510,830
1,837,101
2,585,559
South West
1,334,024
304,100
943,095
304,888
849,077
West Coast
153,111
44,541
26,846
43,392
92,619
Gulf Coast
634,323
747,532
539,133
748,132
431,051
TOTAL
7,332,489
6,830,887
5,915,710
6,839,201
4,435,241
*Methane GWP=25
Table 8. National Emissions Estimates for Liquids Unloading
Region
Methane Emissions (MT COze)*
2012 Emissions
(In 2014
Inventory)
GHGRP2012
Emissions
Inventory 2012
Emissions Using
National Average
EFs
Inventory 2012
Emissions Using
National Average
EFs and Activity
Assumptions
North East
2,115,336
603,649
2,967,625
2,669,870
Mid-Continent
1,790,352
292,158
843,800
1,814,631
Rocky Mountain
1,837,101
3,510,830
1,552,525
1,401,015
South West
304,888
943,095
1,110,200
696,792
West Coast
43,392
26,846
37,175
35,208
Gulf Coast
748,132
539,133
860,250
1,291,289
TOTAL
6,839,201
5,915,710
7,371,575
7,908,807
*Methane GWP=25
Moving from regional-level to national-level emission factors and activity factors did not improve regional
consistency with the GHGRP emissions.
Greenhouse Gas Reporting Program
EPA's Greenhouse Gas Reporting Program (GHGRP) has collected 3 years of emissions data on liquids
unloading. EPA anticipates that for future Inventories, it may be possible to calculate updated emission
factors and/or activity data using GHGRP data that will be reported starting in 2015. Data currently
available through GHGRP include total methane emissions from liquids unloading at the sub-basin level for
each reporter. Data that would allow EPA to develop an updated approach for liquids unloading using
GHGRP data include data on the number of wells venting for liquids unloading, each reporter's total
number of wells, and the number of those wells with and without plunger lifts.
6
-------
April 2015
Table 9. Comparison of GHGRP and GHG Inventory Liquids Unloading CH4 (kt]
Year
GHGRP
2015 GHG Inventory
2011
495
268
2012
227
267
2013
177
256
Requests for Stakeholder Feedback
• EPA seeks feedback on the utility and transparency of using event-based EFs (scf CH4/event/year)
versus well-based EFs (scf CH4/well/year) to develop national estimates in the Inventory.
• EPA seeks feedback on estimating emissions using emissions data from Allen et al. The Allen et al.
study apportioned wells into five unloading categories (e.g., 79% of non-plunger wells in the
Appalachian region have < 10 unloading events per year). To apply emissions data in this structure,
annual data (or a method to approximate annual data) would be needed on the number of wells in
each category. Reflecting changes to this distribution across the time series from 1990 through
2013 would require data on distribution of unloading events over time. EPA seeks comment on
data availability on distribution of unloading events and approaches to ensure that the Inventory
reflects changes in practices overtime.
• EPA seeks feedback on an approach that uses the Allen et al. data to develop regional emission
factors for plunger and non-plunger wells (similar to the current Inventory structure). Average
emission factors that incorporate the Allen et al. study data have a built-in assumption regarding
the distribution of unloading frequency. If these factors are applied across the time series, then the
implied assumption is that distribution of unloading frequency categories has not changed over
time. EPA seeks feedback on approaches that may be applied to ensure accuracy across the time
series.
• Allen et al. observed that older wells had more frequent unloading events than newer wells. EPA
seeks stakeholder feedback on an approach to calculating annual national emissions that
incorporates data on well age to reflect this variability.
• EPA seeks feedback on options for activity data for wells performing liquids unloading with and
without plunger lifts, across the time series. The two right-most columns of Table 6 above show the
impacts of moving from the current Inventory activity data to GHGRP activity data (while applying
the same regional EFs developed from the Allen et al. study).
• EPA seeks feedback on the sampling and measurement methods used in the Allen et al. study and
the appropriateness of the data set for calculating national emissions estimates.
• EPA seeks feedback on approaches for developing updated uncertainty estimates for this source.
The Allen et al. study utilized a bootstrapping method to calculate uncertainty because the
distributions of event emissions were not normally distributed about a mean. In the bootstrapping
procedure, the original data set of each type of well was recreated by making random event
selections, with replacement, from the data set.
• The API/ANGA study estimated nationally 1.1 million unloadings without plunger lifts and 11.6
million with plunger lifts per year. The Allen et al. study data estimated much lower national event
counts—approximately 0.2 million unloadings without plunger lift and 6.8 with plunger lifts per
year. According to the Allen et al. analysis, the API/ANGA data set also show a much higher fraction
of wells that experience very frequent unloadings (> 50 events/year) compared to the Allen et al.
study findings. EPA seeks comment on the different findings of these studies.
7
-------
April 2015
Preliminary comments EPA has received on this source category include support for the development of
emission factors and activity data on a regional as opposed to a national basis and support for the use of
measurement data for calculating emission factors.
8
------- |