Comments Received during the Public Review Period on the
Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-
2014
Commenter: Pamela Lacey
American Gas Association (AGA)
Comment: Meter and Regulator Stations
AGA is pleased that EPA has followed through on its proposal in the Distribution Memo to revise
estimated emissions from metering and regulating (M&R) stations by incorporating updated station
counts and emission factors from the Lamb et al. study. As we commented before, AGA believes that
EPA's proposal to use the updated emission factors and the above grade and below station counts our
members report to EPA under 40 C.F.R. Part 98, Subpart W and scaled for national representation results
in a more accurate estimate of the actual number of M&R stations.
As we noted in our January comments, we agree that it makes sense to estimate M&R emissions across
the time series by using the new updated emission factors for years after 2011 when the Subpart W data
became available, to use the 1992-vintage GRI emission factors for early years beginning in 1990, and to
use interpolation for the years in between. We agree this is the best approach to more accurately reflect
net emissions without the need to subtract Gas STAR program emission reductions.
Comment: Pipeline Leaks
For estimated pipeline leaks in the Draft Inventory, EPA used the previous activity data sources for miles
of pipeline by material and for leaks per mile, and the Lamb et al. data on emissions per leak. AGA agrees
with this approach, and particularly supports EPA's incorporation of the Lamb et al. pipeline emission
factors. As AGA noted in its prior comments, numerous regulatory developments and voluntarily operator
actions have resulted in significant reductions in leak rates and incidents, reflected in the overall lower
emissions found in the Lamb et al. study. AGA also agrees with EPA's approach to use interpolation
between GRI/EPA emission factors in early years and Lamb et al. emission factors in recent years.
Comment: Residential Customer Meters
AGA supports EPA's inclusion in the Draft Inventory of revised emission factors for residential customer
meters by combining data from the 1996 GRI/EPA study with newer data from a GTI 2009 study and
Clearstone 2011 study. As noted in previous comments, the newer data sources, and in particular the GTI
2009 study, include a robust data set composed of numerous data points representing a variety of
residencies, including single family homes, duplexes, townhouses, and apartment buildings. Given the
homogeneity of the residential meters found at all the distribution companies sampled through the GTI
2009 study, incorporating the new residential meter factor into the GHGI is appropriate.
AGA also supports EPA's update of its customer meter activity data for residential meters to incorporate
customer data reported to the U.S. Energy Information Administration (EIA). The customer data is
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
1
-------
reported to EIA on its Form EIA-176. EIA does not collect data on meters specifically. Rather, EIA
instructs respondents to report the average number of consumers served directly from facilities during the
year. For residential consumers, this includes master-metered apartments, mobile homes, multi-family
dwellings (individually metered), and single-family dwellings. Using data reported to the EIA will
improve accuracy compared to the previous GHGI methodology of using 1992 counts driven by gas
consumption.
Comment: Commercial & Industrial Meters
AGA is pleased to see that for commercial and industrial meters, EPA has applied the GTI2009
commercial customer meter emission factor to the total count of commercial and industrial meters in the
GHGI. As AGA noted in its prior comments, consistent with EPA's approach in the Draft Inventory, the
GTI 2009 industrial meter data should not be incorporated into the GHGI. The GTI 2009 study only took
industrial meter measurements from a limited number of sites (46 meters). Due to limited resources,
measurements of industrial meters were intended to represent the broad range of meters in this sector, but
do not provide a statistical sampling indicative of the industrial meter national inventory, nor does it
account for the significant variance in equipment type and size in industrial meters. For this reason, AGA
agrees with EPA not to include this data into the GHGI.
AGA also supports EPA's update of its customer meter activity data for commercial and industrial meters
to incorporate customer data reported to the EIA. As explained above, the customer data is reported
directly to the EIA. Using this data will improve accuracy compared to the previous GHGI methodology
of using 1992 counts driven by gas consumption.
Comment: Blowdowns and Mishaps/Dig-Ins
For pipeline blowdowns and mishaps/dig-ins, in the Draft Inventory EPA used PHMSA data of
distribution main and service miles for the activity data to calculate the estimate of emissions. Although
AGA appreciates EPA's attempt to update the methodology used to calculate emissions from pipeline
blowdowns and mishaps/dig-ins, AGA does not believe that EPA's approach provides an accurate
representation of the emissions from these sources.
As EPA recognizes, the current approach taken in the GHGI for both sources, which relies on 1992
distribution main and service miles and is scaled by residential gas consumption, results in a mileage
estimate that is influenced by factors that would impact natural gas usage, but are unrelated to pipeline
miles. AGA agrees with EPA that PHMSA data is a more accurate data source of pipeline miles. Pipeline
operators are required to report data directly to the Department of Transportation on an annual basis,
which renders the PHMSA data on pipeline mileage an accurate representation of installed pipeline
mileage and is superior to the current methodology of estimating pipeline mileage.
However, AGA is concerned with EPA's use of pipeline miles to estimate emissions from blowdowns
and mishaps/dig-ins. These sources of emissions are discrete events and there is no available data that
suggests a correlation between the number of miles in a pipeline system and the number of mishap events
on that system. The number of reported pipeline incidents on gas distribution systems has been flat or
down during the past five years; during that time, from 2010 to 2014, the number of miles of installed
distribution main in the U.S. has increased by nearly 60,000 miles or 5%.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
2
-------
AGA encourages EPA to use activity data that reflects the reality that an emission blowdown or
mishap/dig-in is a discrete event that is not correlated to the number of miles in a pipeline system. AGA
recognizes the difficulty in obtaining a comprehensive set of data for these sources of emissions.
However, because data associated with both will be reported through EPA's proposed Methane Challenge
for companies selecting this best practice, EPA will have more data for possible use in the future to
generate activity data for the GHGI. In addition, for mishaps/dig-ins, AGA notes that significant incidents
are reported to PHMSA, where significant is defined as an incident above a certain size or impact
threshold. [Incidents on natural gas distribution systems are defined as an event that involves a resale of
gas from a pipeline that results in a death or significant personal injury, property damage of $50,000 or
more, or 3 million cubic feet of lost gas. 49 C.F.R. § 191.3.] AGA recommends consideration of incident
data reported to PHMSA and data collected through the Methane Challenge as possible alternative data
sources for development of more representative activity data for mishaps/dig-ins.
Comment: New Methodology Obviates Need to Subtract Gas STAR Reductions
In the past, EPA used emission factors based on data collected in 1992 in an EPA-Gas Research Institute
(GRI) Study. The agency recognized that practices and materials changed over time, as companies
modernized their systems and implemented best practices shared through the Gas STAR program. EPA
thus considered the 1992 vintage emission factors to reflect the potential emissions sources could emit in
the absence of modernization, and the agency attempted to reflect the effect of continuing modernization
by subtracting voluntary reductions reported under the Gas STAR program to calculate net emissions
from the sector.
AGA agrees that the new methodology - using new data, including that collected in 2013 from the March
2015 Lamb et al. study and Subpart W reporting - results in a more accurate representation of current
operations practices and emissions levels. We agree this obviates the need to continue subtracting
voluntary emission reductions achieved through the Gas STAR program to estimate current emission
levels for M&R stations, pipeline leaks, and customer meters, since the new data already reflects current
practices and emission levels.
Comment: AGA Also Generally Supports the Use of New Data and Methodology for Estimating
Methane Emissions from Natural Gas Transmission and Storage
EPA's revisions to the GHGI for the natural gas transmission and storage segment primarily rely upon
Zimmerle et al. and an interpolation of existing and new data between the early and current inventory
years. Although AGA believes that these approaches can serve as an interim step in EPA's GHGI, AGA
encourages EPA to recognize the significantly larger data set available from measurements conducted at
transmission and storage compressor stations subject to Subpart W of the GHG Reporting Program. For
example, the Subpart W data could be evaluated to assess the relative population of wet seal versus dry
seal centrifugal compressors. AGA also believes that Subpart W data can provide a more accurate
representation of activity data and device type for pneumatic controllers. AGA encourages EPA to
commit to additional updates to the 2017 GHGI report that would integrate Subpart W data.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
3
-------
Commenter: Cynthia A. Finley
National Association of Clean Water Agencies (NACWA)
Comment: NACWA has submitted comments on each of the previous nine Inventories, and we appreciate
the clarifications that EPA has made to clarify the emissions calculations and the factors that are used in
the calculations. Although the wastewater treatment section has not yet been updated for the 2014
Inventory, EPA states that the same methodology will be used as in the previous Inventory. NACWA
previously stated its concern that potentially outdated data was used in the emissions calculations (e.g.,
the 2004 Clean Watershed Needs Survey). If the same data is used in the 2014 Inventory, our concern
remains that the calculations may not accurately reflect current wastewater utility practices. NACWA also
believes that more specific emissions factors could be developed for U.S. wastewater treatment.
NACWA understands that EPA will be looking at possible improvements for the wastewater treatment
calculations in the next year. NACWA is willing to assist EPA in any way with these improvements, such
as providing general information about current wastewater practices or collecting specific data from our
member utilities.
Commenter: Evan Weber, William Snape, Lydia Avila, Colette Pichon
Battle, Joan Brown, Andres Restrepo, Alan Journet, Erik Schlenker-
Goodrich
U.S. Climate Plan, Center for Biological Diversity, Energy Action Coalition, Gulf Coast
Center for Law & Policy, New Mexico Interfaith Power and Light, Sierra Club, Southern
Oregon Climate Action Now, Western Environmental Law Center
Comment: We respectfully submit these comments on the Draft U.S. Greenhouse Gas Inventory Report:
1990-2014. Our comments are intended to encourage EPA to examine gross U.S. greenhouse gas (GHG)
emissions using the most updated values of the Global Warming Potential (GWP) of methane and nitrous
oxide. Given recent international news on China's underreporting of its coal consumption and,
accordingly, GHG emissions (aNovember 3, 2015 New York Times article estimates the undercounting
at over 900 million metric tons), we believe that the U.S. should place additional importance on
accurately quantifying its own GHG emissions.
Our comment states that the Inventory Report, in Annex 6.1, uses an alternative set of GWPs [from the
Intergovernmental Panel on Climate Change's 5th Assessment Report (AR5)] that exclude carbon cycle
feedbacks, resulting in emissions estimates lower than if EPA were to include these feedbacks. While we
understand that EPA excludes these feedbacks to align methodology with the GWPs used in the main text
of the Inventory Report, we believe that these higher emissions estimates, which represent the full climate
impact of methane and nitrous oxide, must be presented to the public.
It is our goal to increase the transparency by which the EPA reports U.S. GHG emissions to the global
community. We believe that using GWPs inclusive of carbon cycle feedbacks accomplishes this goal.
Comment: In Section 6.1 of the U.S. GHG Inventory, Table A-282 presents alternative scenarios of
greenhouse gas emissions estimates if EPA used GWPs from the IPCC Fifth Assessment Report (AR5),
rather than the Fourth Assessment Report (AR4). However, this analysis underestimates the GWP of CH4
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
4
-------
and N2O, based on Table 8.7 (page 714) of the AR5 Working Group I report. This underestimation results
from excluding "carbon cycle feedbacks" previously not quantified in AR4. Table 1 shows that by
including these feedbacks for AR5 100-year GWPs, the emissions increase (relative to AR4 values) is far
higher than EPA presents. While EPA reports this increase to be 22.6 (0.3% higher than AR4 total
emissions) MMTCC^e, the true value is 238.0 (3.5% higher) MMTCChe. According to WRI's CAIT tool,
this additional 215.4 MMTC02e is roughly equal the gross emissions of Norway, Sweden, Denmark, and
F inland—combined.
Our analysis does not include the following factors, which we believe indicate that our upward
adjustments are actually conservative:
IPCC indicates that the GWP of biogenic methane is 34, whereas fossil methane is 36, over a
100-year time horizon. Given that over one-third of U.S. methane emissions are fossil (from
natural gas systems, coal mining, and petroleum systems), the change in methane from AR4 to
AR5 should be greater than our value of 254.8.
EPA's also underestimates the GWP of HFC-134a, which represents 40% of Emissions from
Substitution of Ozone Depleting Substances - the AR5 value EPA uses is 1,300, whereas IPCC,
including carbon cycle feedback, uses 1,550. Other high-GWP gases, whose carbon cycle
feedbacks are not quantified in Table 8.7, very likely have higher GWPs than EPA uses in Annex
6.1, though the lack of IPCC data prevents us from quantifying this.
These emissions must be presented to the public. We do understand that EPA has chosen not to include
the carbon cycle feedbacks from CH4 and N2O for the AR5 GWPs in order to align methodologies with
AR4. However, given that the GWPs highlighted yellow in Table 1 below are the "true" values, we see no
reason to keep the lower AR5 numbers, as changes in methodology to quantify carbon cycle feedbacks
are precisely the goal of updated scientific research. If consistency between methodologies really is
necessary (though again, this shouldn't be a reason not to use the higher values), then we recommend
communicating these findings in addition to the previous ones. Table 2 presents the GHG emissions totals
by gas, for further transparency as to how we calculated the differences between emissions for each GWP
accounting method.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
5
-------
Table 1 - Changes in Emissions using AR-t GYVPs. AR5 GWPs Excluding Climate Feedbacks, and AE5 GWPs
Including Climate Feedbacks
GHG
AR4GWP
Inventory
AR5GWP
Inventory
2014 Change
from AR4 to
AR5 - no
Carbon Cycle
Feedback
%
Change
AR5
GWP -
IPCC
2014 Change
from AR4 to
AR5 - Carbon
Cycle Feedback
Included
%
Chang
e
COz
1
1
0.0
0%
1
0.0
0%
ch4
25
28
84.9
12.0%
34
254.8
36%
n2o
298
265
-45.6
-11.1%
298
0.0
0%
HFCs
MIXED
MIXED
-16.4
-9.3%
MIXED
-16.4
-9.3%
PFCs
MIXED
MIXED
-0.6
-9.6%
MIXED
-0.6
-9.6%
SF6
22,800
23,500
0.2
3.1%
23,500
0.2
3.1%
NFb
17,200
16,100
0.0
-6.4%
16,100
0.0
0%
Total 22.6 0.3% 238.0 3.5%
Table 2 - GHG Emissions Totals by Gas using LPCC's AR4, EPA's AR5, and IPCC's AR5 GWPs
GHG
2014 Emissions
{AR4)
2014 Emissions (AR5, excluding
carbon cycle feedbacks)
2014 Emissions (AR5, including
carbon cycle feedbacks)
C02
5,564.3
5,564.3
5564.3
CH4
707.9
792.8
962.7
N20
411.4
365.8
411.4
HFCs
175.8
159.4
159.4
PFCs
5.8
5.2
5.2
SF6
6.9
7.1
7.1
NF3
0.6
0.6
0.6
Total 6,872.7 6,895.3 7,110.7
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
6
-------
Commenter: Brad Upton
National Council for Air and Stream Improvement, Inc. (NCASI)
Comment: The estimated forest ecosystem carbon stock changes reported in the draft 1990-2014 national
inventory are significantly different than those reported previously. The text in the report explains that
this is due, at least in part, to new estimation methods (described in Woodall et al. 2015) and
reclassification of land in Alaska. It is our understanding that the new estimates rely more heavily on
measured data (compared to model-generated data) than earlier estimates and, as a result, are likely to be
more accurate. It would be helpful for the text in the report to elaborate on the benefits of greater reliance
on measured vs. modeled data in the updated estimates.
Comment: While the report contains a summary of the recalculations of forest ecosystem carbon, it is
unfortunate that the annexes have not been updated to provide a full explanation of the sources of the
difference between the new and previous estimates. We encourage the agency, in future years, to make
the annexes available for comment at the same time the report is made available.
Comment: Changes in carbon stocks in products-in-use are also significantly different than in previous
inventories, but this is not acknowledged or explained in the report or the annexes. This should be
discussed in the report and examined in more detail in the annexes.
Comment: In Chapter 7 Waste on page 7-11, line 1, EPA states that the degradable organic carbon (DOC)
value for landfilled pulp and paper waste was revised from 0.20 to 0.15 based on a literature review and
data reported under 40 CFR Part 98 (referred to as the Greenhouse Gas Reporting Program, GHGRP, the
new DOC value is also discussed in Chapter 9 Recalculations and Improvements on page 9-1, line 39, and
in Annex 3.14 on page A-391, line 38). The new value of 0.15 corresponds to a weighted average of all
DOC values reported to the GHGRP within subpart TT by pulp and paper facilities in 2013. It is stated in
a reference supporting the draft inventory (RTI 20152) that 72% of the pulp and paper facilities that
reported to subpart TT used only the default DOC values from Table TT-1 and that 49% of the reported
waste quantities were associated with the default DOC value for general pulp and paper industry waste
other than industrial sludge (0.20). Therefore, the new DOC value used in the draft inventory (0.15) is
heavily influence by use of the default value of 0.20 in Table TT-1.
The current default DOC for general pulp and paper industry waste other than industrial sludge in Table
TT-1 (0.20) is based on an erroneous interpretation of IPCC guidance, as documented by NCASI in prior
communications with EPA (NCASI 20113). Therefore, it is inappropriate to include data elements
corresponding to the default value of 0.20 when developing a new DOC value for use in the inventory.
As noted in RTI 2015, 28% of pulp and paper facilities that reported to subpart TT developed DOC
values specific to their landfilled waste streams by analysis using methodologies specified by EPA. It is
more technically appropriate (and accurate) to develop a DOC value for pulp and paper industry waste
from a weighted average of these waste stream-specific DOC values reported to the GHGRP, as these
values represent the characteristics of the actual waste placed in industrial landfills at pulp and paper mills
and would not be influenced by the erroneous general DOC value of 0.2. RTI 2015 presents such a
weighted average DOC value for pulp and paper industry wastes, which is 0.10. EPA should use a DOC
value of 0.10 rather than 0.15 in developing estimates of methane emissions from industrial landfills at
pulp and paper mills.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
7
-------
Comment: In Annex 3.14 on page A-391, line 38, EPA incorrectly associates the new DOC value for
pulp and paper industry waste (0.15) with an Lo value of 49 m3/MT. An Lo value of 49 m3/MT correlates
with a DOC value of 0.10, which is the technically appropriate DOC value to use in the agency's top
down analysis as explained above. On line 47 the agency states that "data were available through the
GHGRP to warrant a change to the Lo (DOC) from 99 to 49 m3/MT..." Note that the previous DOC
(0.20) is correlated with an Lo of 99 m3/MT, and further note that DOC is directly proportional to Lo.
Therefore, halving L0 (from 99 to 49 m3/MT) would result in DOC also being halved (i.e., from 0.20 to
0.10).
Comment: As conveyed in our comments on the public review Draft US Inventory of Greenhouse Gas
Emissions and Sinks: 1990-2013 (included herein as Appendix A), production statistics developed by
EPA for use in waste-related GHG emissions calculations for the pulp and paper sector are too high.
Table 7-12 lists 2013 production of the pulp and paper sector at 131.5 million metric tons, based on data
from the Food and Agriculture Organization of the United Nations (FAO), and includes a note that this
figure represents the sum of woodpulp production plus paper and paperboard production. The same
production figures are presented in RTI 2006, which describes EPA's method for estimating industrial
landfill emissions. Summing woodpulp, paper, and paperboard production results in double counting,
because the majority of woodpulp production is used to produce paper and paperboard at integrated mills
(an integrated mill includes both pulping and papermaking at the same facility).
A more appropriate method for characterizing total pulp and paper sector production would be to sum
paper production, paperboard production, and market pulp production [Market pulp is produced at a pulp
mill and then sold rather than being used at the same mill to produce paper or board]. For 2013, the
American Forest and Paper Association reported total production of paper and paperboard to be
approximately 73 million metric tons and total production of market woodpulp to be approximately 8
million metric tons (AF&PA 2014). Based on these statistics, total pulp and paper sector production in
2013 was approximately 81 million metric tons.
EPA's method of using the FAO statistics overstates the pulp and paper industrial sector's production,
which in turn results in estimates of pulp and paper sector industrial wastewater treatment and landfill
methane emissions being far too high. On page 7-28 of the Draft US Inventory of Greenhouse Gas
Emissions and Sinks: 1990-2014, lines 42-47, EPA notes that the agency is evaluating new approaches to
estimating industry-level production (and other values) used in estimating industrial wastewater treatment
GHG emissions. The agency should use production data from AF&PA's Statistical Summary reports in
calculating both wastewater treatment and landfill emissions from the pulp and paper sector, which will
result in more accurate characterization of industrial waste-related methane emissions from this sector.
Commenter: Michael Schon
Portland Cement Association (PCA)
Comment: The Draft Inventory's approach to accounting for emissions associated with cement
production does not consider available data, however, or determine whether those data are consistent with
the conclusions reached by the Draft Inventory. In addition, the Draft Inventory does not present a
comprehensive and easily discernible estimate of the industry's total GHG emissions. This issue makes
verification of the total emissions associated with cement production impossible and also masks
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
8
-------
efficiency improvements by the sector. In these comments, PCA suggests areas for improvement to
address these concerns.
Comment: As the Draft Inventory acknowledges, GHG emissions are released at two points in the
production of cement—an essential component of concrete. First, the combustion of fuel to heat cement
kilns and to enable necessary chemical reactions produces GHG emissions. Thanks to efficiency
improvements, including use of carbon-neutral alternative fuels, cement production plants reduced
combustion-related emissions per unit of production in recent years. Second, emissions are generated
through calcination, a chemical reaction that produces calcium oxide—a foundational component of
cement. Calcium carbonate is converted to calcium oxide and carbon dioxide: CaCC>3 -> CaO + CO2.
There is little opportunity to reduce the calcination process- related CO2 emissions per unit of production.
Comment: EPA developed a Greenhouse Gas Reporting Program (GHGRP) for cement plants to
inventory both of these types of emissions on a facility-specific basis. Under Subpart H to 40 C.F.R. Part
98, all cement production plants in the United States must report both their combustion-related and
process-related emissions. 40 C.F.R. §§ 98.80, 98.82. EPA now has five years of reported data from those
facilities on file. In 2010, an Intergovernmental Panel on Climate Change (IPCC) task force encouraged
the consideration of GHGRP data in the development of the annual inventory of domestic GHG emissions
that EPA submits to the United Nations in accordance with the United Nations Framework Convention on
Climate Change (UNFCCC).
Yet this year's draft domestic inventory, like its predecessors, still does not consider the GHGRP data for
cement production, including whether those data points are in line with the GHG estimations presented in
the Draft Inventory. Rather, EPA punts on considering those data. This is a missed opportunity to
evaluate facility-specific data, as EPA itself acknowledges.
Comment: In the Draft Inventory, EPA also misses an opportunity to analyze emissions associated with
cement production in a comprehensive manner. While the process-related emissions of cement
production are addressed in the Industrial Processes and Product Use chapter of the Draft Inventory, the
combustion-related emissions of cement production are not disaggregated from other industries'
combustion-related emissions in the Energy chapter. The Draft Inventory estimates total process-related
cement production emissions at 38.8 MMT CC^e in 2014, but presents no equivalent figure for the
combustion-related cement production emissions. This makes it impossible to determine the total
emissions generated by the industry.
Thus, PCA cannot comment on whether the Draft Inventory's accounting of cement production emissions
is defensible or accurate. We encourage EPA to calculate and present an overall emissions figure
associated with cement production so that it can be compared to the total reported cement production
emissions of 67.6 MMT CChe in 2014 under the GHGRP.
Comment: PCA also encourages EPA to consider cement production emissions not only on a total mass
basis but also on a production rate basis so that efficiency improvements are apparent. As economic
conditions have improved, demand for cement has increased, resulting in an increase in the total tons of
emissions. Importantly, however, significant efficiency improvements, on an emissions per unit of
production basis, have also occurred.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
9
-------
Commenter: Kerry Kelly
Waste Management (WM)
Comment: We have gained considerable experience by implementing the Mandatory GHG Reporting
Rule (GHG MRR) since 2010, reporting emissions for active and closed Municipal Solid Waste (MSW)
landfills and associated renewable energy projects. The landfill sector has significant interest in the Draft
Inventory since EPA, for the first time has used annual waste disposal data reported by MSW landfills
under Subpart HH of the GHG MRR, in its Draft Inventory emissions estimates. We very much want to
work with you to ensure that GHG MRR data are used correctly to refine the Draft Inventory for MSW
landfill emissions.
We commend EPA for using GHG MRR data to refine the inventory estimates of emissions. As EPA
states in Chapter 7 -Waste, of the Draft Inventory (at 7-7), the EPA rigorously verifies data provided by
reporters subject to the GHG MRR. Moreover, reporters certify the data as true and accurate before
submitting it to the Agency, and must collect data and ensure its quality in accordance with GHG MRR
requirements and the facility's GHG Monitoring Plan. Thus, data developed for the GHG Reporting
Program (GHGRP) is of known quality and has far greater certainty than other databases EPA has relied
upon. Using reporting data and emissions calculations prepared for the GHGRP should enhance the
quality and validity of the nationwide inventory.
Comment: Because of the emphasis on accuracy and verification with GHG MRR data, we were
surprised with the changes to MSW landfill emissions estimates in the Draft Inventory. We believe that
thorough evaluation of the databases must be undertaken before EPA can confidently express 2015
emissions using the GHG MRR data. The changes in net emissions, and amounts of methane flared and
used for energy appearing in the draft inventory are very significant and negative. The 24-year methane
reduction performance achieved by MSW landfills working to comply with EPA control standards
dropped from a projection of 38% reduction to a mere 1.4% reduction. We could not replicate the
Agency's calculations, and they appear to be in contravention with other data all agree to be reliable.
Specifically, there appears to be a fundamental disconnect between the estimated emissions reported by
MSW landfills subject to the GHGRP and the estimated emissions reported in the Draft Inventory. The
GHGRP emissions from MSW Landfills in 2014 were 91.5 MMT CChe. EPA designed the GHGRP to
obtain the highest possible percentage of emissions from each reporting sector, while minimizing the total
number of facilities that would be required to report. EPA selected a reporting threshold for MSW
landfills based on estimated methane generation of 25,000 MT CC^e or greater, and estimated that the
MSW landfills reporting under GHGRP comprise 82% of total national emissions of MSW landfills for
both active and closed landfills.
The inconsistency in the emissions reported becomes evident when comparing the 2014 emissions from
the GHGRP to those estimated in the Draft Inventory for the same year. If 91.5 MMT represents 82% of
MSW landfill emissions, then logically, the total from all MSW landfills will be approximately 111.5
MMT CChe. Instead, total emissions from MSW landfills are 167 MMT CC^e, and emissions for the
landfill sector (both MSW and industrial landfills) are 181.8 MMT CC^e.
Comment: The landfill sector representatives appreciated your meeting with us to describe the process
used to integrate GHGRP annual waste disposal figures into the Draft Inventory. Since we first reviewed
these estimates, we have been attempting to discover what factors led to a total methane generation of
almost twice as much as what was in the GHGRP data. This is a challenging exercise because the
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
10
-------
database has been structured in such a way to make accessing all of the relevant information very
difficult.
Comment: We found a significant source of error in the use of GHGRP annual waste disposal figures in
the Draft Inventory because the waste was not properly differentiated between degradable waste and inert
materials. Since only degradable waste produces methane, applying the degradation factor (or DOC) for
bulk MSW to all waste disposed (even separate inert waste streams that do not degrade) significantly over
predicts methane generation.
We looked first at the public database for the GHGRP (Envirofacts) to assess how many reporters in 2014
characterized their annual waste receipts to identify inert materials. Because Envirofacts does not capture
the waste type descriptor provided by reporters, one must query the database to identify reporters using
various DOC values for different waste streams and sum those fractions to one. For 2014, 944 landfill
sites reported accepting waste. Of those 944 reporting annual waste receipts, 42% reported receiving
inert waste, using the waste composition option to delineate inert wastes (DOC=0), and combining
separate C&D waste streams with MSW under the bulk waste category, or by using the modified bulk
waste option showing (MSW DOC=0.31 C&D DOC=0.08, Inert DOC=0).
In fact, because it is so difficult to identify reported waste types in Envirofacts, we turned to the SCS
Engineers database, which contains all required reporting elements from 2010-2014 for 544 MSW landfill
GHGRP reporters, or 44% of the total number of reporters, and 50% of the annual waste receipts. The
landfills in this database include both private and municipal sites located across the country. Looking at
GHGRP annual disposal amounts for the 544 sites in 2014, 23% of waste disposed was reported as inert.
The prior reporting years 2010-2013, had similar percentages of waste reported as inert (ranging from
17% in 2010 to 22.5% in 2013), with the amount of inert waste growing in each year. This is consistent
with the current emphasis on diversion of organic wastes from landfills, and efforts by landfills to make
up the difference with inert waste streams such as ash and soils.
We also evaluated the GHGRP waste disposal history for these 544 sites (including total waste in place ~
WIP). Of the total WIP, 8.1% is inert. However, WIP data is far less definitive than annual waste
disposal information because most reporters did not have historical data, or chose to estimate historical
waste in place as MSW and did not characterize the different waste streams (MSW, C&D, inert) disposed
in the landfill.
Waste Management did report well-characterized waste back to 1999 for most sites. A review of this
information showed that from 1999 through 2015, there has been a 21.5% drop in the amount of MSW
waste disposed in landfills, a 21% increase in inert wastes, and an 11% increase in C&D waste. These
findings comport with the experience of public and private landfills across the country. Increased
recycling and organics diversion initiatives have resulted in a decline in MSW landfill disposal, yet for
many landfills receipt of inert waste streams has steadily increased.
Based on our analysis of the three datasets, we believe the annual waste disposal volumes used in the
Draft Inventory to calculate methane generation were likely assigned inappropriately high DOC values,
resulting in an over prediction of methane generation. This in turn led to inflated estimates of methane
emissions from MSW landfills.
Comment: We know that you share our interest in assuring the final Inventory is as accurate as possible.
The information in the Draft Inventory presents major, adverse policy implications for the Administration
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
11
-------
and EPA. The current draft could be interpreted to contradict White House and Agency regulatory
statements, plans and documents with regard to methane controls, vitiate the effectiveness of the EPA's
twenty-year old New Source Performance Standards (NSPS) and Emission Guidelines (EG) Rules, and
undermine the accomplishments of the Landfill Methane Outreach Program (LMOP). These very serious
impacts must certainly be avoided if they result from a misinterpretation of GHGRP waste disposal data
because the inventory database simply does not fully characterize waste types and their potential to
generate methane overtime.
The landfill sector wants to work with you to ensure that the GHGRP data are appropriately used, and the
resulting estimated emissions are representative of MSW landfill disposal and gas collection and control
practices. We are concerned that there is limited time for the Agency to conduct a thorough reevaluation
of the data and make the necessary changes. If the Agency were to publish the Draft Inventory results as
they appear in the current draft, public officials and community residents would be misinformed about
landfill emissions, and there could be significant policy and economic repercussions for the sector.
To allow sufficient time for correction of the draft estimate, in the short-term, we urge EPA to use the
2015 Inventory data and protocols for estimating MSW landfill emissions. For future inventories, we
encourage the Agency to make use of the emissions calculations developed and certified by GHGRP
reporters under the force of law. The Agency has been proactive in improving the estimation of landfill
methane emissions by updating GHGRP protocols. Use of these verified emissions data could only
enhance the U.S. Inventory, while reducing administrative burdens on Agency staff. We urge EPA to
work with the landfill sector to develop a methodology to incorporate GHGRP results and the growing
body of measured methane emissions into the nationwide inventory - much as you are doing with the
natural gas sector. We believe this is a wise practice, and we commit to do everything possible to assist
your review.
Commenter: Luis Orlindo Tedeschi
Texas A&M University
Comment: I know this is past the date of March 23, 2016, but I really wanted to make sure this is
addressed. I noticed in Table 5-3, the order of Horses, Sheep, and Swine might be incorrect. Looking at
previous reports, you had Swine, Horses, and Sheep, and the numbers for the current inventory don't
match my expectations. I'd think that Swine is greater than horses and sheep, and sheep is greater than
horses.
DRAFT, 1990-2014 Inventory Report:
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
12
-------
Table 5-3: cm Emissions from Enteric Fermentation (MMT CO2 Eq.)
Livestock Type
1990
2005
2010
2011
2012
2013
2014
Beef Cattle
119.1
125.2
124.6
121.8
119.1
118.0
116.7
Dairy Cattle
39.4
37.6
40.7
41.1
41.7
41.6
41.9
Horses
2.0
2.3
2.4
2.5
2.5
2.5
2.4
Slieep
1.0
1.7
1.7
1.7
16
1.6
1.6
Swine
2.3
1.2
1.1
11
1.1
1.1
1.0
Goats
0.3
0.4
0.4
0.3
0.3
0.3
0.3
American Bison
0.1
0.4
0.4
0.3
0.3
0.3
0.3
Mules and Asses
+
0.1
0.1
0.1
0.1
0.1
0.1
T otal
164.2
168.9
171.3
168.9
166.7
165.5
164.3
Note: Totals may not sum due to independent rounding.
+ Does not exceed 0.05 MMT CO2 Eq.
1990-2012 Inventory Report:
Table 6-3: CH4 Emissions from Enteric Fermentation (Tg CO2 Eq.)
Livestock Type
19.90
2005
2008
2009
2010
2011
2012
Beef Cattle
100.0
105.8
107.5
106.3
105.4
103.1
100.6
Dairy Cattle
33.1
31.6
34.1
34.4
34.1
34.5
35.0
Swine
1.7
1.9
2.1
2.1
2.0
2.1
2.1
Horses
0.8
1.5
1.6
1.6
1.6
1.6
1.7
Sheep
1.9
1.0
1.0
1.0
0.9
0.9
0.9
Goats
0.3
0.3
0.3
0.3
0.3
0.3
0.3
American Bison
0.1
0.4
0.3
0.3
0.3
0.3
0.3
Mules and Asses
+
+
0.1
0.1
0.1
0.1
0.1
Total
137.9
142.5
147.0
146.1
144.9
143.0
141.0
Notes: Totals may not sum due to independent rounding.
+ Does not exceed 0.05 Tg CO; Eq.
Commenter: Jean Bogner
University of Illinois - Chicago
Comment: The purpose of this letter is to, first, document the deficiencies of the current IPCC (2006)
FOD model for landfill methane generation, recovery, and emissions as currently applied to U.S. sites
under the GHGRP HH- methodologies [Spokas et al., 2011, 2015; Bogner et aL 2010, 2014, 2016], In
general, IPCC (2006) relies on 40-year old science using a 1970's landfill gas generation model as well as
a default 10% oxidation value based on a 20-year old study for oxidation at one U.S. site (Czepiel et al.,
1996a,b). Importantly, neither IPCC (2006) nor the recent modifications for oxidation and emissions
added to the GHGRP methodologies explicitly model the major climate drivers for emissions now known
from literature.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
13
-------
Comment: In addition, these model applications lack comprehensive field-validation for emissions. See
Appendix A for more detailed discussion.
Comment: A second purpose is to introduce an existing, freely-available [www.ars.usda.gov], fully-
documented, user-friendly JAVA tool for landfill methane emissions inventory reporting. This model
[CALMIM] was developed using established relationships for gaseous & heat transport, then
independently field-validated.
Comment: Instead of relying on a landfill gas generation model, CALMIM explicitly models landfill
methane emissions based on 1-dimensional gaseous, heat, and water transport in each cover material for a
typical annual cycle of 365 days. The major drivers are: 1) the individual cover thicknesses and physical
properties at a specific site; 2) the annual climate cycle for each cover as it affects soil moisture and
temperature at various depths and, in turn, methane transport and oxidation rates; and 3) the physical
effect of engineered gas recovery on soil gas concentration gradients.
Comment: A third purpose is to initiate discussion regarding the application of CALMIM as an
alternative to IPCC (2006) for landfill methane emissions inventory reporting under the GHGRP. As
stated in IPCC (2006), "higher order validated" models are permitted under IPCC national GHG
inventory guidelines.
Comment: In general, very wide ranges for methane emissions and oxidation had been quantified, often
not aligning with the 10% value and ranging from negligible to >100% (uptake of atmospheric methane).
Comment: It is reasonable to point out that, in the intervening years, the expected temporal variability of
oxidation rates over an annual cycle in site-specific cover materials has often been overlooked. In short,
oxidation is a variable, not a constant, for each specific cover material at a specific global location.
Comment: Regarding b), potential improvements to the underlying IPCC (2006) FOD gas generation
model, there were many problems with trying to fit this conceptual model to a growing database of site-
specific field measurements for emissions. Those problems included large mismatches between modeled
& measured emissions, a primary dependency for FOD-modeled methane emissions on waste in place for
the California inventory [Appendix A] irregardless of waste composition data & k values, and
observational data from current California sites where measured gas recovery rates were robustly &
linearly related to WIP only [Appendix A; Spokas et al., 2015], Thus CALMIM was developed as a new
"emissions-only" model as discussed in Appendix B.
Commenter: Karin Ritter
American Petroleum Institute (API)
Comment: In lieu of a formal expert review process of the Preliminary Draft of the national GHG
Inventory (GHGI), as was customarily done in past years, EPA released several memos between
December 2015 and February 20161 outlining revisions under considerations for estimating GHG
emissions from the Distribution, Transmission & Storage, Gathering & Boosting and Petroleum &
Natural Gas production segments of the Petroleum and Natural Gas Systems sector. API's comments on
those memos are provided herein as an attachment starting on page 6.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
14
-------
Comment: While the last set of memos on Production and Gathering and Boosting were still under expert
review, EPA released the Public Review Draft of the GHGI, already incorporating the revisions that were
dubbed "under consideration" in EPA's memos, without providing industry the opportunity to comment
on these proposed revisions, or for EPA to incorporate industry's expert comments, prior to releasing the
Draft GHGI for public review. In addition, the released Public Review Draft does not provide specifics
on the revised methodological changes for specific sources and lacks the normal methodological details
usually provided in the applicable Annexes.
Comment: Based on information provided in the memo Inventory of U.S. Greenhouse Gas Emissions and
Sinks: Revisions under Consideration for Natural Gas and Petroleum Production Emissions (February
2016, Table 4), API attempted to recreate the production sector emission data reported in Table 3-43 of
EPA's Public Review Draft Inventory of U.S. Greenhouse Gas Emissions and Sinks. The following table
summarizes API's comparison of 2013 source level emissions published in the April 2015 GHGI and the
2013 emission estimates from Table 3-43 of the recent Public Review version of the GHGI.
T;iMt i. Comp/u hon of 2'U* F mi "ion F tor Natural Gas Production
i iin Im'Shio in.iinl -riu'
**rfrtf ff/n*-
j/fii % Wjfjhj fjrJ
\"*f ("Hi
Lini-
MM i (. o:<
2Ml3 N\-r (. H4 Lini- ion-..
MMI (. 02e
Pneumatic Controllers
13.5
26 0
Major Equipment! Fugitives
8,6
Q~
Chemical Injection Piunps
1.5
3,7
Dehydrator Pumps Vents
12.2
12,2
Compressor Starts
0.1
0,1
Large Gathering Compressor
Station Fugitives
0,4
43 3
Gathering Pipeline Leaks
4.2
Gas Ensines
2,7
"J T
Condensate Tanks
7.8
7,8
Bloivclo YC11S
0,2
0,2
Upsets
0.1
0.1
Wellpacl Fugitives Yenuns
11,5
11,5
Offshore
3.S
3,8
Other Voluntary Reductions
-16 5
-16,0
Regulatory Reductions
-3,0
TOTAL
47.0
105,1
As is shown in the table above, total emissions for Natural Gas Production operations are estimated to
increase from 47 million metric tonnes (MMT) C02e as published in last year's GHGI, to 105 MMT
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
15
-------
CC>2e, which indicates more than a doubling of emissions. It appears that EPA intends to include
approximately 16 MMT CC^e in emission reductions from voluntary activities, although it is unclear to
which sources these emission reductions will apply. It is also unclear if fugitive emissions from wells are
included under "Wellpad Fugitive Venting" or under "Major Equipment Venting". API is concerned that
these additional details are not available for review and comment ahead of the final GHGI that is
scheduled to be published in April 2016.
Comment: For Petroleum and Natural Gas Systems, EPA provides "computed" emission values for
calendar year 2013, using the proposed, revised methodologies from EPA's sector specific memos.
Emissions for the years 1990-2012 are not back-cast or updated, and EPA does state in the Public Review
draft that the 2013 emissions estimates are preliminary and subject to revision in the final GHGI. As a
result, it looks like a large step-change in estimated emissions for 2013 resulting from EPA's
methodological changes. The new methodology used by EPA, especially for the Petroleum and Natural
Gas production segments of the industry, does not reflect a "real" increase in emissions but rather
improved availability of some industry activity data as reported to the GHGRP. The improved industry
activity information provided by larger facilities, which are above the GHGRP reporting threshold, is
being used by EPA for scaling up to the nationwide inventory without recognizing that the smaller (non-
reporting) facilities likely have very different activity characteristics and thus should not be included in
the scaled up activity factors proposed by EPA.
Comment: The estimated Petroleum Systems emissions for 2013 indicate a 151% increase as compared
to what was previously reported for 2013 and is driven by an assumed increase of 157% in Petroleum
Production emissions. This assumed emissions increase from Petroleum Production is due to EPA's
scaling up the count of pneumatic controllers and process fugitive components as reported through the
GHGRP. This does not reflect the fact that smaller production sites, which are not subject to GHGRP
reporting, have much smaller component counts per wellhead and many of them use little - if any -
pneumatic controllers, particularly in petroleum systems . Most importantly, EPA did not revise the
emission factors used for characterizing overall emissions from pneumatic controllers and fugitive
sources, despite repeated comments from industry that these factors are outdated and overestimate
emissions from properly functioning pneumatic controllers and typical process components.
Comment: For Natural Gas Systems, EPA estimates that 2013 emissions would increase 23% after
applying EPA's new estimation methodology. The data for individual segments such as production,
processing, transmission & storage and distribution show a respective emissions change of 136%, 0%, -
47% and -64%. Again, the change of 136% in the production segment is due to extrapolation of
pneumatic controllers and process fugitive component counts from the GHGRP to a nationwide basis, as
well as using the same overestimation of component counts for smaller production sites that do not report
to the GHGRP. The change in Natural Gas Production also includes a new and very large estimate for
Gathering and Boosting compressor stations based on limited, short-duration, downwind measurements.
API does not believe the data used to derive emissions for Gathering and Boosting stations are sufficient
for determining national emissions from these operations due to the large uncertainty associated with the
measurement method on which they are based.
Comment: For some activity data, larger equipment counts would be expected for the types of sites that
are more likely to be reported in the GHGRP. However, applying data from GHGRP sites to the entire
population of U.S. wells is inappropriate. For example, emergency shut-down devices (ESDs) may be
counted as pneumatic controllers in the GHGRP but have very different emission characteristics
(infrequently emitting) than the types of pneumatic controllers that are assumed in the GHGI.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
16
-------
Comment: EPA's approach appears inconsistent. First, EPA notes that Subpart W GHGRP data covers
32% of the active wellheads for 2013 and proposes to use this percentage to "scale" some emission
sources to a national level. Simultaneously EPA states that the GHGRP Subpart W data covers the
majority of national oil and natural gas production sources. Separately, EPA has also determined that
Subpart W covers about 85% of the GHG emissions from the onshore oil and natural gas production
sector as indicated in the Subpart W Technical Support Document.
Comment: Clearly, if Subpart W covers 85% of the GHG emissions from theoil and natural gas
production sector, then there is no basis for changing the GHGI in a manner that estimates 90% higher
overall GHG emissions (based on the recalculated 2013 inventory). This discrepancy of GHGRP Subpart
W emissions coverage must be fully explored and explained prior to making the proposed changes to
derive GHG emissions for this sector in the GHGI. Given that the GHGRP Subpart W reported GHG
emissions are substantially less than in the estimated GHGI emissions for 2013, the resultant scaling of
the GHGRP data to national GHG emissions should be less than the 15% of emissions EPA previously
determined are not covered by GHGRP Subpart W.
Comment: API agrees that updated GHGI activity factors and emissions data are warranted and as such
recommends that EPA form a multi-stakeholder working group comprised of industry, governmental, and
environmental organizations active in GHG emissions measurements and estimation to evaluate recently
published data that may be considered for updating the national GHGI prior to rushing to implement the
proposed revisions that are based on invalid extrapolation of GHGRP data from large facilities to non-
reporting smaller installations.
Comment: API recognizes that emerging data from recent field studies have raised concerns about
measurements uncertainty, and recognizes the need for a thorough discussion of means of improving the
methodology to ensure collection of robust measurement data. API proposes that a working group - as
discussed above - be convened following the completion of the 2014 GHGI (April 2016) to provide a
structured framework for consultation and review of GHGI updates. An early start (April 2016) and
frequent meetings (every 1-2 months) would provide sufficient time to review and consolidate
information in an informed process for updating the 2015 GHGI (that would be published in April 2017)
and beyond.
Commenter: Giles Ragsdale
Comment: My 2 cents - Figure ES-15 (I look at this figure every year) -1 think the majority of people
forget that when comparing current greenhouse gas emissions to 1990, the population has risen steadily
which drives demand for and emissions from most categories of greenhouse gases, e.g. electricity,
transportation, etc. I think this figure tells a great story - emissions per capita are down to flat compared
to 1990. I'd say EPA is doing good work that the general population does not recognize and some
politicians chose to not recognize.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
17
-------
Commenter: Bridget Chadwick
Comment: Page 3-4:
Clarify the definition of energy as "the capacity for doing work as measured by the capability of doing
work (potential energy) or the conversion of this capability to motion (kinetic energy)" [EIA Monthly
Energy Review, MER] and identify the types of energy sources: fossil fuels, nuclear, and renewables.
Emphasize that some fossil fuels are consumed for non-energy purposes (e.g. feedstock, reducing agents
and non-energy products) but are inventoried separately in Section 3-2.
Comment: Page 3-7, Figure 3-4 U.S. Energy Consumption (Quadrillion Btu):
change the scale of the graph to provide more detail; (2) add gridlines so that energy consumption can be
read more easily from the graph; (3) It appears that data for energy consumption + consumption of fossil
fuels for non-energy use have been graphed with a peak of about 100 qBtu in 2007. From my estimates,
using fossil fuel energy data provided in Table A-18 of EPA's draft Inventory and nuclear and renewable
energy provided in the EIA's MER, total energy consumption in 2007 peaked at about 93.5 qBtu.
Comment: Pages ES-19, 3-6, 3-7 "In the United States, 82 percent of the energy consumed in 2014 was
produced through the combustion of fossil fuels..." (page 3-6). :
From my estimates, in 2014, total fossil fuel energy amounted to 73.6794 qBtu (using data in Table A-l 1
of the EPA's Inventory). Nuclear and renewable energy (including geothermal energy) and imported
electricity amounted to 18.143 qBtu (using data in EIA's February 2016 MER Tables 1.3 and 2.6). So
fossil fuel energy was about 80% of total energy consumed in 2014. My calculation of energy
consumption for specific energy sources will differ from EPA's calculation, too.
Comment: Page 2-11, Figure 2-5: 2014 Energy Chapter Greenhouse Gas Sources (MMT C02 Eq.):
The scale of the bar chart deemphasizes the significance of fossil fuel combustion. The scale should be
expanded so that readers can see fossil fuel combustion produces the greatest portion (about 92%) of
energy-chapter emissions. Furthermore, the adjacent piechart should show the breakdown of fossil fuel
combustion in the energy chapter "slice".
Comment: Page 2-3, "Energy-related C02 emissions also depend on the type of fuel or energy consumed
and its carbon (C) intensity. Producing a unit of heat or electricity using natural gas instead of coal, for
example, can reduce the C02 emissions because of the lower C content of natural gas". :
(1) Explain that the carbon intensity of an energy mix (e.g. electricity) is the energy-weighted average of
the C02 emission factors of the energy sources in the mix; (2) Provide a table of C02 emission factors
for all energy sources including nuclear and renewable energy and/or refer readers to Table A-39.
Comment: Page 3-14, (a) "Recently an increase in the carbon intensity of fuels consumed to generate
electricity has occurred due to an increase in coal consumption, and decreased natural gas consumption
and other generation sources", (b) "Total U.S. electricity generators used natural gas for approximately 27
percent of their total energy requirements in 2014 (EIA 14" 2015b)". :
Please correct the above statements: (a) Using the EPA Inventory for fossil fuel data (Table A-l 1) and the
EIA MER (Table 2.6) for C-free/neutral energy data, the c-intensity of electricity has DECREASED
steadily since 2005, from 60.579 MtC02/qBtu in 2005 to 52.785 MtC02/qBtu in 2014. (b) In 2014,
natural gas was 22% of the total primary energy consumed for generating electricity and C-free/neutral
energy was 35% of the total primary energy.
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
18
-------
Supplemental Material Received
Appendix A
Appendix A from the University of Illinois at Chicago comment on the U.S. Greenhouse Gas Emissions
and Sinks: 1990-2014
Appendix B
Appendix B from the University of Illinois at Chicago comment on the U.S. Greenhouse Gas Emissions
and Sinks: 1990-2014
Appendix C
Appendix API comments on EPA's Memos on the updates being considered for the Transmission and
Storage, the Production and the Gathering and Boosting segments of the Petroleum and Natural Gas
Systems Sector in the GHG Inventory from the American Petroleum Institute comment on the U.S.
Greenhouse Gas Emissions and Sinks: 1990-2014
Appendix D
Appendix API Comments on Updates under Consideration for Natural Gas and Petroleum Production
Emissions, and Gathering and Boosting Emissions from the American Petroleum Institute comment on
the U.S. Greenhouse Gas Emissions and Sinks: 1990-2014
Comments Received on Public Review Draft of Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2014
19
-------
Appendix A
-------
APPENDIX A.
Shortcomings of current IPCC (2006) methodology for landfill methane emissions.
To summarize the shortcomings of the current IPCC (2006) model, below are listed the
major deficiencies with supporting references and datasets:
~ This model was never systematically field-validated for ChU emissions. Rather,
the historic "validation" consisted of comparing measured recovery to modeled
generation at 9 Dutch landfill sites (Oonk & Boom, 1995; Van Zanten and
Scheepers, 1995; Oonk, 2010)
~ Model results do not systematically replicate results from a growing database of
field measurements for CH4 emissions (Spokas et al., 2011, 2015; Bogner et al.,
2010, 2011, 2016).
~ GHG inventories [e.g., California GHG inventory] often do not consider actual
landfill gas recovery data at specific sites, only an assigned "recovery efficiency"
percentage applied to modeled generation. Typically, the assigned landfill gas
recovery can differ substantially in both magnitude and direction (+ or -) from
measured recovery. (Bogner et al., 2010, 2016)
~ Actual measured landfill gas recovery can be directly related related to waste in
place (WIP) using a simple linear relationship. Fig. 1 below demonstrates this
relationship for 129 California sites using data from Walker et al., (2012). The
relationship shown in this figure was independent of climate, status (open or
closed), age, or size (WIP).
Historical Note: In general, landfill gas modeling began in California during the
mid-1970's at the time of the first commercial landfill gas utilization projects.
Then, At that time, a multiplicity of site-specific models were applied to the early
project sites in order to predict future LFG recovery from waste-in-place (WIP),
climate, waste composition, and other factors. [See further discussion in
Findakakis and Leckie, 1979; EMCON, 1980; Halvadakis et al., 1983; Findakakis
et al., 1988.] In those days, the choice of a particular model format for a specific
site depended on optimizing the match between predicted annual LFG recovery
and actual LFG recovery from the monitoring data available at that time. The
models ranged from simple empirical relationships to complex, multicomponent
multiphase kinetic models, some with lag times prior to the initiation of LFG
generation. For the kinetic models, there was no unique solution for a specific
site as multiple parameters were adjusted to improve model fit. The kinetic
models (IPCC, 2006; LandGEM) were primarily adapted from the anaerobic
digestion literature and accelerated laboratory decomposition studies on the
premise that, conceptually, the annual mass of waste buried in a landfill may
degrade similarly to waste in a digester but over longer timeframes.
What might be a better idealized model for landfill biodegradation? Landfills also
have significantly lower liquid contents than even "dry" or high solids digestion
systems and, indeed, would be impossible to manage if digester values were
applied to field settings. Taking a broader view, a better analogy for landfills is
comparison to terrestrially-derived organic matter buried at shallow depths over
5
-------
longer-term "geologic" timescales. Initially, after burial, some portion of the
organic carbon undergoes anaerobic decomposition with biogas generation.
However, a significant portion of the organic carbon in the buried waste does not
degrade over decadal timeframes (Bogner, 1992; Barlaz, 1998) and is
available for future transformations via deeper geologic burial under conditions of
increased heat and pressure. That process is termed "diagenesis" with
endpoints over geologic timescales expected to be similar to peaty/humic coal
materials.
In spite of variable waste input data and climate-related k values for LFG
generation using IPCC (2006), the primary dependency for emissions is on
waste-in-place (WIP). This can be demonstrated [Fig. 2 below] using the 2011
California GHG inventory data (372 full-scale landfill sites). [See also Spokas et
al., 2015; Bogner et al. 2016.] Using this methodology, larger landfills [having
high WIP] cannot reduce emissions below a certain threshold as defined by this
relationship. Moreover, this relationship tends to reward larger sites with non-
optimized gas recovery strategies [due to the relatively constant relationship for
emissions to WIP]. Conversely, this relationship tends to reduce incentives for
sites to improve gas recovery systems to achieve emission reductions as those
reductions are not credited.
As discussed above, the default assumption of 10% annual oxidation in IPCC
(2006) is based on a single study at one landfill (Czepiel et al.. 1996). Oxidation
is a variable, not a constant, with unique seasonal trends in each cover soil at
each site. [See discussion and data in Spokas et al., 2011; Spokas and Bogner,
2011; Bogner et al., 2011.]
The 3 major drivers for emissions are excluded. These are:
1) The area, composition, and thickness of site-specific cover soils as the
major engineered barrier for emissions.
2) Climate trends unique to both the specific global location (e.g,
latitude/longitude) & individual cover soils with seasonally variable
gaseous transport & CH4 oxidation rates due to temporally and spatially
variable soil moisture & temperature.
3) The physical effect of the engineered LFG system to recover CH4 and
concurrently reduce soil gas CH4 concentrations at the base of
the cover, reducing the CH4 concentration gradient and thus reducing
diffusive flux [see Spokas et al., 2011].
6
-------
Dutch data
(large circles):
y = 0.000312x
40000
35000
JZ
1 30000
z
t
% 25000
Q
u
GJ
™ 20000
00
0
3
1 15000
c
ta
UJ
no
10000
~ 2010 CALRECYLE Average Landfill
Gas Recovery Rate (Nm3 h-1)
Normalized to 50% CH4
• Dutch LFG recovery (IPCC FOD
"validation")
0 30,000,000 60,000,000 90,000,000 120,000,000
Waste in Place (Mg)
California data
(small diamonds):
y = 0.0002SSX
R2 = 0,90
Fig. 1. Comparison between WIP and average biogas recovery rate for: (a) 2010
data from Calrecycles for 129 California sites (Walker et al., 2012): blue diamonds; and
(b) IPCC FOD model field validation data from 9 Dutch landfills (1986-1993) (Oonk &
Boom, 1995): red circles. Figure reprinted from Bogner et al., 2016.
7
-------
16,000
14,000
10,000
6,000
4.000
uj O)
2,000
T^070Ut)220x
f?==4>^0
0 <£0?*
0.E+00 1.E+07
2.E+07
3.E+07
4.E+07
5.E+07
6.E+07
2011 Waste in Place (Mg waste)
Fig. 2. (a) ABOVE: Relationship between estimated 2011 site-specific landfill CH4
emissions using IPCC (2006) and WIP for 371 California landfills, (b) BELOW: Same
relationship including the large Puente Hills Landfill [N=372]. Data from California Air
Resources Board [ARB] (Hunsaker, 2012). NOTE: Predicted emissions from WIP using
regression coefficients are 190-220 Mg CH4/million Mg WIP. Figure reprinted from
Bogner et al., 2016.
30,000
20,000
y = 0.000191x
r2 = 0.88
10,000
5.00E+07
1.00E+08
1.50E+08
8
-------
Appendix B
-------
Appendix B.
Description and Overview of the CALMIM 5.4 Model.
[See Spokas et al.. 2015; Bogner et al.. 2014; Spokas et al.. 2011; Spokas and Bogner,
2011; Bogner et al.. 20111
Developed over the last decade, CALMIM, or CAIifornia Landfill Methane Inventory
Model, is a 1-dimensional finite difference model for the simultaneous simulation of heat,
water, and gaseous transport through landfill cover soils. The model consists of a
process-based methane emissions model which simulates emissions using 10-min time-
steps and 2.5 cm depth increments in user-specified landfill cover materials at any global
location. Table 1 at the end of this appendix provides an overview of the model structure,
components and default boundary conditions. CALMIM is a freely available
[www.ars.usda.gov] JAVA program which integrates site-specific data (location and
cover design) with climatic simulation and one-dimensional soil microclimate and gas
diffusion models for daily, intermediate, and final cover areas inclusive of CH4 oxidation
over a typical annual cycle. The model has proven to be user-friendly at sites where it
has been applied to date (e.g., Cambaliza et al., 2015).
CALMIM includes: (1) the effect of engineered gas extraction; (2) the physical effect of
daily, intermediate, and final cover materials to retard emissions; and (3) seasonal
moisture and temperature effects on both gaseous transport and methanotrophic CH4
oxidation in cover soils. The empirical relationship for oxidation used in the CALMIM
model is derived from a series of over 900 laboratory incubations of landfill cover soils to
determine relationships between methanotrophic activity and soil temperature & moisture
(See Spokas and Bogner, 2011).
CALMIM was independently field-validated, first for v. 4.3 for California in the initial
CALMIM project for the California Energy Commission [Bogner et al., 2011]. The original
field validation for the CEC project (>800 measurements using static chambers) was
conducted over two years on daily, intermediate, and final covers at two California sites,
including the northern coastal Marina Landfill (Monterey County, CA) and the southern
Scholl Canyon Landfill (Los Angeles County, CA). Also included were continuous
measurements of soil temperature, moisture, and selected meteorological variables.
Additional limited field validation was conducted for intermediate covers at the Lancaster,
Kirby Canyon, and Tri-Cities Landfills through the cooperation of Waste Management,
Inc. Oxidation was quantified through the use of a stable carbon isotopic method
developed by J. Chanton which relies on the preference of CH4-oxidizing
microorganisms for the isotope of smaller mass (12C) versus the heavier isotope (13C).
Subsequently, the improved CALMIM 5.4 developed under the EREF project was globally
field-validated using 40 covers at 29 sites on 6 continents [Bogner et al., 2014], using
data supplied directly by international research groups, published data, and data
collected by the CALMIM team. A wide variety of methods (chamber, gradient, tracer,
micrometeorological, vertical radial plume mapping, aircraft-based) were applied over
scales ranging from <1m to km. CALMIM comparisons to field measurements resulted
in a d-index of 0.765 using site-specific data (Willmott Index of Agreement; Wilmott,
1981), a Pearson r value > |0.8| for modeled vs. measured comparisons at 25 of 29 sites,
and an average mean error across all covers of 12 g CH4 m2 d 1. Figure 3 below shows
the main CALMIM input screen.
-------
Figure 3. Main CALMIM input screen.
1551 © c+-
•1^1 *1
© e>:
About New Site Open Site Last Site Exit
CALMIM
California Landfill Methane Emission Inventory fvlodel
I
CALMIM estimates typical annual, site-specific landfill CH; emissions based on the
respective areas and properties of daily, intermediate, and final cover materials, as well
as the extent of engineered gas extraction. A major change from the IPCC (2006)
method is that emissions are decoupled from a CH4 generation model; instead, the
emission processes at the top of the landfill are modeled directly. Another major change
is that seasonal CH4 oxidation is also modeled directly rather than relying on a %
oxidation "default." In terms of the IPCC structure, CALMIM is an IPCC "validated, higher
quality" methodology for typical annual CH4 emissions from landfills. CALMIM consists of
four major integrated components:
(1) Data-Input Template;
(2) Meteorological Model;
(3) Soil Microclimate Model;
(4) 1-D Emissions/Oxidation Model.
With regard to (1), site locations are linked to latitude and longitude information. Input
data are required on the surface area, thickness, and properties of the various cover
materials for a particular site. Also, the extent of gas extraction and seasonal vegetation
for each cover type are also required (both as % of surface area). With regard to (2) and
(3), the meteorological and soil microclimate models rely on modified versions of the
following globally-validated USDA models: Global TempSIM, Global RainSIM, Solarcalc,
and STM . In particular, the soil temperature functions for STM' (Soil Temperature and
Moisture2) were modified to accommodate the landfill heat source. The latitude and
longitude of the site are used to extrapolate the daily climatic conditions, as well as the
soil microclimate conditions for 10-min, intervals for (minimum) 2.5-cm. depth increments
for any landfill cover soil. With regard to (4), the emissions model is based on 1-
dimensional diffusional transport of CH4 and 02 through each specified cover material.
10
-------
The driving force is the CH4 concentration gradient through the cover materials, which is
dependent on the presence or absence of engineered gas recovery, the thickness and
properties of the cover materials, and seasonal CH4 oxidation rates. Methane oxidation is
modeled through the use of scaled results relative to maximum rates for the full range of
soil temperature and moisture conditions based on extensive laboratory studies for
California landfill cover soils (>2000 incubations) and published literature. Oxidation is
quantified by the difference in separate CALMIM model runs with and without oxidation
for each cover type. CALMIM also calculates total annual site emissions by summing
the emissions for all cover types. A standard substraction is also applied for 02 uptake
by heterotrophic respiration [competition for 02 with CH4 oxidation].
Below is shown (Fig. 4) some typical CALMIM output comparing 30 cm to 90 cm loamy
sand intermediate covers at a southern California site. Note both the large variability in
emissions at this site between the two thicknesses and differences for each thickness
between the oxidized and unoxidized emissions. The highest emissions were associated
with the mid-year dry season, diminishing in the later part of the year when the rainy
season begins.
Day of Year
Permitted Minimum Cover Thickness Re|d Resu|ts ()C and FC; 7/2010 to 4/2011):
Modeled Surface emissions without oxidation(g/m2/day) ^ 95% UCL Mean
Modeled Surface emission with oxidation (g/m2/day)
Triple Permitted Minimum Cover Thickness Mean Modeled Surface emissions with oxidation
Modeled Surface emissions without oxidation(g/m2/day) H Permitted Minimum Cover Thickness
Modeled Surface emission with oxidation (g/m2/day) ¦ Triple Permitted Minimum Cover Thickness
Fig. 4. Typical CALMIM output for southern California intermediate cover material.
Comparison of 30 cm to 90 cm thickness over typical annual cycle. See text for
additional explanation.
CALMIM relies on well-researched and accepted theoretical relationships, previous field
and laboratory studies, existing globally-validated U.S. Dept. of Agriculture models, and
extensive supporting laboratory studies on CH4 oxidation using a variety of landfill cover
soils over the full range of temperature and moisture conditions. Because the CALMIM
model uses average climatic and soil microclimate data to calculate typical annual
emissions, results may not be representative for atypical climate conditions (e.g., drought
years) or where there are large differences in relief relative to regional weather stations.
The site-specific application of CALMIM can be significantly improved through the use of
11
-------
"advanced" functions and site-specific data, including field measurement of the CH4
concentration at the base of the cover.
To demonstrate the strong climate dependency of emissions, we remodeled the 2010
California landfill CH4 emissions inventory for 372 sites using Calrecycles data from
Walker (2012) and the field-validated CALMIM5.4 model (Spokas et al., 2015, Spokas et
al., 2011), then compared the results to the existing 2010 California inventory from the
California Air Resources Board (ARB) using the IPCC (2006) FOD model with regional
California waste data and k values. See Fig. 5 below. It is important to note that the
ARB method applies a 75% gas recovery efficiency to estimate the residual emissions,
regardless of actual gas recovery. Importantly, the IPCC methodology does not consider
either soil or climate drivers for gaseous transport nor seasonal methanotrophy in cover
soils, allowing only the 10% annual oxidation per Czepiel et al. (1996 a,b).
Highest Emitting Sites: Different Drivers for Different Methods:
ARB vs. CALMIM
CARB 2010 CH4 Emission Estimates
(Mg/yr)
Intermediate Cover only
[96% of emissions]
Climate
Mean Annual Precipitation
(mm/yr)
CALMIM
Fig. 5. Comparison of major dependencies for estimated California landfill CH4 emissions
using:
TOP: 2010 ARB inventory based on IPCC (2006) model showing dependency on WIP.
BOTTOM: 2010 inventory using CALMIM 5.4 showing dependency on climate for
intermediate cover [96% of estimated state emissions]. Cover areas from Walker et al.,
(2012). The typical intermediate cover was modeled as 90 cm loamy sand with emission
rates normalized to g CH4 m"2 d"1. See Spokas et al. (2015) for additional discussion and
details. Also shown at left are the 11 highest emitting sites from each inventory.
Note that, in Fig. 5 the intermediate cover emissions for a typical 90 cm loamy sand are
<20 g CH4m"2d-1 when the mean annual precipitation (MAP) is >500 mm y"1. Moreover,
comparing the highest-emitting sites between the ARB and CALMIM inventories, those
12
-------
sites shift from landfills containing the largest mass of waste in the ARB inventory to sites
with large areas of thinner intermediate cover and reduced oxidation rates during the
annual cycle (e.g., too hot, too dry). These climate dependencies have important
implications for developing more realistic, science-based GHG inventories for landfill
CH4.
Finally, we also directly compared CALMIM modeling using site-specific inputs for cover
materials and areas to field measurements at 10 California sites [Fig. 6]. Field methods
ranged from meter to kilometer scales, including chamber techniques, vertical radial
plume mapping (VRPM), and aircraft plume methods. In this figure, we show standard
CALMIM outputs for CH4 emissions with oxidation and CH4 emissions without oxidation
for a "typical annual cycle" of 365 days. The plots shown in this figure include both single
cover materials and whole site measurements over several years, depending on the
methodology, scale, and date of the individual campaigns cited in the figure caption. See
Spokas et al. (2015), Bogner et al. (2014), and references cited therein for additional
details.
In Fig. 6., please also note the high seasonal variability and the large seasonal
differences between the upper blue lines (emissions without oxidation) and the lower
black lines (emissions with oxidation). Especially note that the lines for emissions with
and without oxidation become merged at several sites during the mid- to late-year dry
season due to negligible oxidation (too hot, too dry). Thus, modeled emissions inclusive
of oxidation readily respond to dynamic soil moisture and temperature effects on
oxidation rates during an annual cycle. Moreover, when examining results from any
short-term field measurement campaigns at a specific global location, it is important to
consider those results within the larger expected temporal variability of emissions over an
annual cycle. In short, consistent with other soil sources of CH4, climate effects on both
oxidation and gaseous transport can vary greatly between cover soils at any one site, as
well as seasonally and spatially between sites (Cambaliza et al., 2015).
In general, the CALMIM modeled emissions align with the field values and, as a
minimum, are within the same order of magnitude. Differences can be attributed mainly
to: (1) cover thickness and/or composition not modeled correctly (may not be rigorously
tracked at specific sites except to confirm "permitted minimum" thickness or materials);
(2) whether daily cover area emissions were realistically modeled (i.e., whether the
working area overlies new waste only with expected low fluxes or fully methanogenic
older waste driving high fluxes, with or without gas recovery); and (3) annual weather
variability compared to 30-year average weather with 0.5 degree reliability.
13
-------
01-AA-0008
200-
f
A
100 -
• *
IL
llli|I^IIS|
&
01-AA-0O09
8 O-
€
3
CO
I II ^ I 4 ^ I S i )
19-AA-0012
19—AA—O012
U I ? I I ^ ! I S i ^
19—AA—0050
¦Max-
100 -
19-AA-0053
Ml!
50 -
0 -
t
U I I I ^ ? I 8 i ^
27-AA-0010
19-AA-0056
400 -
150 ~
300 -
ii i _¦ mi r
200 -
WW
100 -
100 -
/ \
50 -
0 -
/ i
0-!
^ S I Ifl ^ I 1 S ! II I ! I l| I ^ I IS
30-AB-0035
43—AN—0008
s g
2 M
Figure 6. TOP: Comparison of typical annual cycle of emissions using CALMIM at 10
California sites to field measurements using a variety of techniques, CALMIM results
indicate the "typical annual cycle" of 365 days where the black line is predicted emissions
with soil oxidation and the blue line represents surface emissions without oxidation. The
region between is shaded in light blue. Field results are plotted for the month of the
measurement using different symbols for different techniques: Red plus sign indicates
surface chambers (Spokas et al. 2011; Shan et al, 2012), black diamond/triangles
indicates aircraft plume measurements (Peischl et al, 2013; Tratt et al, 2014), and the
green circle indicates vertical radial plume mapping [VRPM] methods (Goldsmith et al,
2012). All units are g CH4 m"2 d" . Figure reproduced from Spokas et al., 2015; please
consult for further details. BOTTOM: Location map for California sites.
14
-------
CALMIM is written entirely in JAVA and currently consists of 531 Java Classes and is
written in the NetBeans Integrated Developer Environment (IDE). NetBeans IDE and
NetBeans Platform are based on software from netbeans.org, which has been dual
licensed under the Common Development and Distribution License (CDDL) and the GNU
General Public License Version 2 with Classpath exception. For more information, please
visit www.netbeans.org.
CALMIM uses a total of 21 integrated libraries, with the most significant ones being:
• jFreeChart - Provides the graphical display of the generated data - see
http://www.jfree.org/
• Liquid-Look-n-Feel - Overall look-n-feel of the program
• PTPLOT 5.6 - plotting program to display data -
http://ptolemy.eecs.berkeley.edu/java/ptplot/
• NanoXML - Embedded XML parser for the CMM preference files
http://nanoxml.sourceforge.net/orig/
• XStream - simple library to aid in saving and loading XML class library
files (CMM preference file) - http://xstream.codehaus.org/
• MigLayout - layout manager for GUI windows http://miglayout.com/
As stated above, CALMIM is a 1-dimensional finite difference model for the simultaneous
simulation of heat, water, and gas transport through the landfill soil cover. Table 1 below
provides an overview of the model structure, components and default boundary
conditions:
Table 1. Overview of CALMIM input parameters, bundled models, and outputs.
Description
Value/Units /Reference
Model Site
Inputs
Latitude
Longitude
Waste Footprint
Decimal degrees (+N , -S)
Decimal degrees (-W, +E)
Acres
Cover
Characteristics
Coverage
Organic Matter
Vegetation Presence
0-100% of waste footprint
Low-high (0-5%)
0-100% cover (slider bar)
Modifies incoming solar
radiation
Gas Recovery System
[Si = (l-Veg%)*Si]
0-100% coverage (slider bar)
Reduces the lower methane
concentration in default cover
scenarios
Cover Type Selection
Temperature Upper
Lower
Cm Upper
Lower
Oxygen Upper
Lower
CH4 oxidation rate
Temperature Upper
Lower
Air temperature simulation
Daily
25 °C
2 ppmv
0.3 % (v/v)
20% (v/v)
5 % (v/v)
Intermediate
400 ng CH4 gsoii^d-1
Air temperature simulation
15
-------
CH-t Upper
2 ppmv
Lower
45 % (v/v)
Oxygen Upper
20% (v/v)
Lower
1 % (v/v)
CH-t oxidation rate
400 ng CH-t gaoii^d"1
Temperature Upper
Air temperature simulation
Lower
40 °C
CH-t Upper
2 ppmv
Final
Lower
55 % (v/v)
Oxygen Upper
20% (v/v)
Lower
0 % (v/v)
CH-t oxidation rate
400 ng CH-t gaoii^d"1
Custom
User selectable boundary conditions
Layer
Characteristics
Material
Various materials (Table 2)
Thickness
Variable: 2.5 cm to 2.5 m
(1 to 100")
16
-------
Table 1. (Continued)
Description Value/Units/Reference
Bundled
Models
GlobalTempSIM
Global RainSIM
Air temperature simulation
Precipitation simulation
Spokas and Forcella, 2009
Spokas and Forcella, 2009
SolarCalc
Solar radiation simulation
Spokas and Forcella, 2006
STM2
Soil temperature and moisture model
Spokas and Forcella, 2009
Gas Diffusion
Oxygen and methane diffusion
Campbell, 1985
Model
Outputs
Model outputs are written directly to Excel compatible files for each cover type
Daily Surface CH4
emissions
With oxidation
Without oxidation
g CH4 m^d"1
g CH4 m^d"1
Soil Temperature
°C
Soil Moisture
Volumetric (cm3 cm 3)
Air-filled porosity
cm3 cm-3
Oxygen Concentration
%02
Soil Nodes
(2.5 cm layer in
cover)
With oxidation
ch4
Concentration Without oxidation
% ch4
% CH4
CH4 oxidation rate
CH4 oxidation percentage
Bulk density
Fraction of time oxidizing
g CH4 m^d"1
%
g cm-3
0 to 100% (0-1)
Simulated Weather
Data
Maximum air temperature
Minimum air temperature
Precipitation
°C
°C
mm
17
-------
References Cited: Letter and Appendices A & B:
Barlaz, M.A. 1998, Carbon storage during biodegradation of municipal solid waste
components in laboratory-scale landfills, Global Biogeochemical Cycles, 12, 373-380.
Bogner, J.E., 1992, Anaerobic Burial of Refuse in Landfills: Increased Atmospheric
Methane and Implications for Increased Carbon Storage, Ecological Bull. 42:98-108.
Bogner, J.E., and Spokas, K.A., 2016, No More California Dreaming: Realistic Modeling
of Landfill Methane Generation and Emissions Inclusive of Climate. Extended abstract
for Global Waste Management Symposium, Indian Wells, California.
Bogner, J.E., Spokas, K.A., Chanton, J.P., 2011, Seasonal greenhouse gas emissions
(methane, carbon dioxide, nitrous oxide) from engineered landfills: Daily, intermediate,
and final California cover soils. J. Environmental Quality, 40, 1010-1020.
Bogner, J.E., Spokas, K.A., Chanton, J.P., 2010, CALMIM: California Landfill Methane
Inventory Model - A New Field-Validated Inventory Methodology for Landfill Methane
Emissions, Final Report to California Energy Commission Public Interest Energy
Research Program, Contract No. 500-05-039, G. Franco, Program Manager. September
2010. 96 p.
Bogner, J., Spokas, K., and Corcoran, M., 2014, International Field Validation of
CALMIM: A Site-Specific Process-Based Model for Landfill Methane (CH4) Emissions
Inclusive of Seasonal CH4 Oxidation, Final Report to Environmental Research and
Education Foundation (EREF), 406 p. Available at http://erefdn.org/index.php/grants/
fundedresearchinfo/international_field_validation_of_a_new_ipcc_model_for_landfill_met
hane_emi/
Cambaliza, M.O., Shepson, P.B., Bogner, J., Daulton, D., Stirm, B., Sweeney, C.,
Montzka, S., Gurney, K., Spokas, K., Salmon, O., Lavoie, T., Hendricks, A., Mays, K.,
Turnbull, J., Miller, B., Lauvaux, T., Davis, K., Karion, A., Moser, B., Miller, C.,
Obermeyer, C., Whetstone, J., Prasad, K., Crosson, E., Miles, N., and Richardson, S.,
2015, Quantification and source apportionment of the methane emission flux from the city
of Indianapolis, Elemental Science of the Anthropocene, 3, paper 000037, doi:
10.12952/journal.elementa.000037. Available at: https://www. elementascience. org/
articles/37/
Campbell, G.S., 1985, Soil Physics with Basic: Transport Models for Soil-Plant Systems,
Elsevier, NY. 150 p.
Czepiel, P., Mosher, B., Crill, P., Harriss, R., 1996a. Quantifying the effect of oxidation on
landfill methane emissions. Journal of Geophysical Research: Atmospheres (1984-2012)
101, 16721-16729.
Czepiel, P., Mosher, B., Harriss, R., et al., 1996b. Landfill methane emissions measured
by enclosure and atmospheric tracer methods, Journal of Geophysical Research:
Atmospheres (1984-2012) 101, 16711-16719.
18
-------
EMCON, 1980, Methane Generation and Recovery from Landfills. Ann Arbor Science,
Ann Arbor, Ml, USA, CRC Press, 150 p.
Findikakis, A. N. and J. O. Leckie, 1979, Numerical simulation of gas flow in sanitary
landfills, J. Environ. Eng, 105, 927-945.
Findikakis, A. N., C. Papelis, C. P. Halvadakis, and J. O. Leckie, 1988, Modeling gas
production in managed sanitary landfills, Waste Manage. Res. 6, 115-123.
Halvadakis, C.P., Robertson, A.P., and Leckie, J.O., 1983, Landfill Methanogenesis:
Literature Review and Critique. Stanford University Dept. of Civil Engineering,
Environmental Engineering and Science Technical Report No. 271. Supported by Pacific
Gas & Electric and Southern California Gas Co., 157 p.
Hunsaker, L., 2012, Site-specific California landfill CH4 emissions inventory database
2009-2011. EXCEL Spreadsheet.
IPCC, 1996, 2006, National GHG Inventory Guidelines Vol 5: Waste. Available at:
http:/fwww. ipcc-nggip. iges. or.jp/public.
Oonk, H., 2010. Literature Review: Methane From Landfills—Methods To Quantify
Generation, Oxidation, and Emissions. Report for the Sustainable Landfill Foundation.
Assendelft, Netherlands.
Oonk H, and Boom T. 1995. Validation of landfill gas formation models. Studies in
Environmental Science 65, 597-602.
Peischl, J., Ryerson, T., Brioude, J., Aikin, K., Andrews, A., Atlas, E., Blake, D., Daube,
B., Gouw, J., Dlugokencky, E., 2013. Quantifying sources of methane using light alkanes
in the Los Angeles basin, California. J. Geophysical Research: Atmospheres, 118, No.
10 27,4974-4990.
Scheutz, C., Kjeldsen, P., Bogner, J., deVisscher, A., Gebert, J., Hilger, H., Huber-
Humer, M., and Spokas, K., 2009, Microbial methane oxidation processes and
technologies for mitigation of landfill gas emissions, Waste Management and Research,
27, 409-455.
Shan, J., Jacoboni, M., and Ferrante, R., 2013, Estimating greenhouse gas emissions
from three Southern California landfill sites. Proceedings 2013 SWAN A Landfill Gas
Symposium. Published by SWANA, Silver Spring, MD.
Spokas, K., and Bogner, J., 2011, Limits and dynamics of methane oxidation in landfill
cover soils, Waste Management 31, 823-832.
Spokas, K., Bogner, J., and Chanton, J., 2011, A process-based inventory model for
landfill CH4 emissions inclusive of soil microclimate and seasonal methane oxidation, J.
Geophys. Research Biogeosciences, 116, paper G04017.
Spokas K, Bogner J, Corcoran M, and Walker S., 2015, From California dreaming to
California data: Challenging historic models for landfill CH4 emissions. Elemental Science
19
-------
of the Anthropocene, 3, paper 000051, doi: 10.12952/journal.elementa.000051. Available
at: https://www. elementascience. org/articles/51/
Spokas, K., and Forcella, F. 2006. Estimating Hourly Incoming Solar Radiation from
Limited Meteorological Data. Weed Science 54: 182-189.
Spokas, K. and F. Forcella, F. 2009. Software Tools for Weed Seed Germination
Modeling. Weed Science 57, 216-227.
Tratt, D. M., Buckland, K.N., Hall, J.L., Johnson, P.D., Keim, E. R., et al., 2014, Airborne
visualization and quantification of discreet methane sources in the environment, Remote
Sens. Environ., 154, 74-88.
van Zanten, B. and Scheepers, M., 1995, Modeling of Landfill Gas Potentials. In
Proceedings from the SWANA 18"'Annual Landfill Gas Symposium, New Orleans, LA,
published by SWANA, Silver Spring, MD.
Walker, S., et al., 2012,, California Dept. of Resource Recovery & Recycling. EXCEL
Database: Landfill Data Compilation. With the assistance of W. Gin (Senior Eng.;
deceased), M. Holmes, and H. Hansra,
Willmott, C.J., 1981, On the validation of models, Physical geography 2, 184-194.
20
-------
Appendix C
-------
API Comments on Updates under Consideration for Natural Gas Transmission and Storage Emissions
Karin Ritter
Manager
Regulatory and Scientific Affairs
1220 L Street, NW Washington, DC 20005-4070 USA
Telephone: 202-682-8472
Fax: 202-682-8031
Email: ritterkffiapi.org
www.api.org
February 5, 2016
Ms. Melissa Weitz
Climate Change Division, Office of Atmospheric Programs
U.S. Environmental Protection Agency - Office of Air and Radiation
1200 Pennsylvania Ave., NW
Washington, DC 20460
weitz.melissa@epa. gov and ghginventory.gov
Re: Updates under Consideration for Natural Gas Transmission and Storage Segment
Emissions in the 1990-2014 GHG Inventory
Dear Melissa,
The American Petroleum Institute (API) appreciates the opportunity to provide comments on
proposed updates to the 1990-2014 U.S. Greenhouse Gas (GHG) inventory for the Natural Gas
Transmission and Storage segment.
API continues to compile and analyze emissions data for petroleum and natural gas operations and
is open to working with EPA on utilizing data provided through EPA's mandatory GHG reporting
program (GHGRP). API has provided comments and recommendations to the U.S. EPA on the
draft Natural Gas Systems and Petroleum Systems sections of the national inventory since 2002,
including at the recent stakeholder workshop in November 2015 regarding GHG data for Petroleum
and Natural Gas Systems.
For this current review, API provides general comments and also addresses several specific
questions raised in EPA's transmission and storage memo. Our review, however, is limited due to
the short response time, overlapping comment periods for other proposed changes to the GHGRP,
and the approaching March deadline for reporting 2015 GHGRP data.
General Comments
EPA's proposed updates for compressor station components rely primarily on two studies published
12
by Colorado State University in 2015 . Substantial new data are available from measurements at
1 Subramanian R.; Williams, L.L.; Vaughn, T.L.; Zimmerle, D.; Roscioli, J.R.; Herndon S.C.; Yacovitch, T.I.;
Floerchinger, C.; Tkacik, D.S.; Mitchell, A.L.; Sullivan, M.R.; Dallmann, T.R; Robinson A.L. Methane Emissions
from Natural Gas Compressor Stations in the Transmission and Storage sector: Measurements and Comparisons with
the EPA Greenhouse Gas Reporting Program Protocol. Enviromnental Science and Technology, 49, 3252-3261. 2015.
API
7 of 87
-------
API Comments on Updates under Consideration for Natural Gas Transmission and Storage Emissions
transmission and storage compressor stations that report through Subpart W. API agrees that
updated GHGI emissions data are warranted and as such recommends that EPA form a multi-
stakeholder working group comprised of industry, governmental, and environmental organizations
active in GHG emissions measurements and estimates to evaluate recently published data that may
be used for updating the national GHG inventory. API proposes that such a working group be
convened following the completion of the 2014 GHGI to provide a structured framework for
consultation and review of GHGI updates. An early start (April 2016) and frequent meetings (every
1-2 months) would provide sufficient time to review and consolidate information in an informed
process for updating the 2015 GHGI and beyond.
API reiterates that the EPA should carefully analyze and screen GHGRP reported data in order to
improve the validity of data used in the national GHGI. Obvious data errors and/or outliers should
be assessed, corrected or excluded to prevent disproportionately impacting the derivation of
emission factors (EFs) or extrapolation of information for the national GHGI.
Responses to EPA Questions
Transmission and Storage Station Fugitive Emissions
•S (Question #1 from. EPA's memo) As EPA considers options for applying EFs for this
source, the EPA seeks stakeholder feedback on the timing of changes in transmission and
storage stations non-compressor fugitive sources that may result in different emissions in
recent years from those in the GR1/EPA study. The EPA could use GR1/EPA factors for
earlier years in the time series, and Zimmerle factors for ecent years. Alternatively,
the EPA could apply the Zimmerle EF to all years of the time series. The EPA seeks
stakeholder feedback on these options.
API Comment: GRI/EPA emission factors should be used for initial estimates in the time series
and EPA should use updated emission factors for the current estimate.
(Question #3 from EPA's memo) The EPA seeks stakeholder feedback on how to
incorporate information on super emitters into estimates for transmission and storage
stations. For example, the Zimmerle study estimated a fraction of the population that may
be super emitters at a given time, and estimated super emitter emissions from these sources
(incremental to those estimated for the non-super emitter population). The EPA also seeks
stakeholder feedback on which G urces are more likely than others to act as super
emitters and whether and how to apply a super emitter factor or other methodology to those
sources.
API Comment: Recent measurement studies have shown skewed "long tail" distributions for
source level measurements, where a few emission sources may contribute a disproportionately
high fraction of emissions. As the Zimmerle study points out, large data sets are needed to
accurately characterize the "long tail" distributions. Although the Subramanian study
contributes new measurement data for 45 compressor and storage stations, it represents just a
2 Zimmerle, D.J.; Williams L.L.; Vaughn, T.L.; Quinn. C.; Subramanian, R.; Duggan, G.P.; Willson. B.; Opsomcr. J.D.;
Marchese, A.J.; Martinez D.M.; Robinson. A.L. Methane Emissions from the Natural Gas Transmission and Storage
System in the Unitcd States. Environmental Science and Technology. 49, 9374-9383. 2015.
8 of 87
-------
API Comments on Updates under Consideration for Natural Gas Transmission and Storage Emissions
subset of all measurements conducted as part of Subpart W reporting, which provides a
substantially larger data set of emissions that are characteristic of the entire distribution.
As the Zimmerle study indicates, the identified "super emitters" fraction of the population is
dynamic and may vary each time a measurement is taken. Therefore the approach being
proposed by EPA in the question - which implies that EPA is considering to separately adjust
the national inventory for super-emitters - is not appropriate for extrapolation of the data to the
national GHG1. This approach would be incorrect and would essentially double count the effect
of super-emitters since they are already accounted for in the Zimmerle emission factors and in
the Subpart W reported data.
The Pipeline Research Council International (PRCI) is conducting a research project to compile
and analyze Subpart W data. The dataset includes 2011 through 2013 measurement data
collected from members who have also provided supplemental data on equipment, operations,
and measurement methods. Although a subset of data reported to EPA, it represents well over
half of the reporting facilities. These measurement data should be assessed and can be used to
calculate compressor station emission factors and evaluate the frequency and size of the larger
leaks from key sources - compressor seals, compressor valves and storage tank dump valves.
The report is expected to be available in the second quarter of 2016.
API advises that an alternative approach would be to develop new average emission factors that
integrate data from both the recent measurement study results and Subpart W measurements.
Such average emission factors should incorporate the range of emissions observed in current
operations without artificially superimposing on them a "super emitter" adjustment which is
highly uncertain. The emission factors should be updated periodically based on additional
Subpart W data that become available with each future reporting year and potentially new,
relevant and independent measurement programs.
**"" (Question #4 from EPA's memo) The EPA seeks stakeholder feedback on how to
incorporate Subpart W data into the GHGI methodology, such that the transmission station,
and storage station activity dai: . ' d/or J ": uld be updated annually to reflect
ongoing trends in the industry. For example, I v could consider combining the
Zimmerle et ah data and Subpart W data in some way.
API Comment: A significant amount of information is reported to EPA through Subpart W.
EPA now has four years of fugitive measurement data for specific emission sources and activity
data regarding the distribution of centrifugal versus reciprocating compressors as well as the
fraction of wet seal versus dry seal centrifugal compressors. API encourages EP A to make use
of this information and integrate Subpart W based emission factors as an update to the GHGI.
Activity data and emission factors should be updated periodically based on additional Subpart
W data that become available with each future reporting year and potentially new, relevant, and
independent measurement programs.
EPA's memo on revisions under consideration for transmission and storage emissions indicates
that EPA intends to use the emission factors for compressor fugitive emissions, non-compressor
fugitive emissions, and pneumatic controllers from the Zimmerle study. API supports the use of
this recent measurement data, which accounts for the presence and random nature of super-
emitters. However, API strongly encourages EPA to also make use of the substantial amount of
9 of 87
-------
API Comments on Updates under Consideration for Natural Gas Transmission and Storage Emissions
measurement data available from Subpart W. The PRC I report is an example of additional
information that should be considered by EPA and a multi-stakeholder workgroup.
(Question #5 from EPA's memo) In fall 2015 * '1 in a California storage field began
leaking methane at an estimated rate of 50 Ml »er clay. The EPA is considering how to
include this emission source in its 2(317 GHG _ estimates from 1990-2015), For
example, the EPA could review and potentially incorporate estimates of the leak developed
by the California Air Resources Board (CAI1B).
API Comment: The storage field leak in California is a one-off failure event. If EPA believes
the emissions from this event warrant inclusion in the 2015 national GHG emissions for Natural
Gas Systems, then API contends that the emissions should be estimated for this single event
with an annotation in the inventory which references the event and the emission estimation
method. The emissions from this singular event should not be back-cast to prior years, nor
should the emissions be projected to future years.
Reciprocating and Centrifugal Compressors
For Storage, EPA is not considering changes to the method used to count compressors. EPA
plans to report a combined number and will not differentiate between reciprocating and
centrifugal compressors to be consistent with planned updates to the emission factor. EPA's
memo notes that the Zimmerle study found most storage stations employ reciprocating
compressors. However, this is inconsistent with the Subramanian study which observed that the
compressor type can impact emissions and centrifugal compressors have become much more
common at transmission and storage stations. For compressor emission factors applied to
Storage, API recommends utilizing storage station compressor measurement data reported for
Subpart W to develop emission factors separately for reciprocating and centrifugal compressors,
and also report compressor emissions separately by compressor type. This provides greater
transparency and enables trends in compressor counts and emissions to be tracked over time.
Pneumatic Controllers
**"' (Question #11 from EPA's memo) The EPA seeks stakeholder feedback on use of the
Zimmerle et ah estimates of pneumatic controller counts per transmission or storage station
to develop national AD across the time series. For example, the EPA could use GM/EPA
pneumatic controller counts for earlier years in the time series and Zimmerle et ah counts for
more recent years. Alternatively, the EPA could apply th nerle et ah pneumatic
controller counts to all years of the GFIG1 time series. It stakeholder feedback
on these options,
API Comment: Subpart W provides a comprehensive, annual data set for determining the
number of pneumatic controllers by station and the distribution by type of controller. API
recommends using the Subpart W activity data for recent years in the GHGI, the GRI/EPA data
for early years in the time series, and interpolating between the two for intermediate inventory
years rather than using activity data that is based on the Zimmerle or Subramanian study.
•S (Question #13 from EPA's memo) The EPA seeks stakeholder feedback on approaches to
stratify pneumatic controller estimates into specific bleed rate categories (e.g., basin .n
the number of low-bleed, intermittent bleed, and high bleed devices and applying an EF
10 of 87
-------
API Comments on Updates under Consideration for Natural Gas Transmission and Storage Emissions
specific to each type). For example, the EPA could use the Subpart W data on the number of
pneumatic controllers of specific controller types per station, and their associated specific
EFs. In addition, the EPA seeks comment on use of GHGRP data to represent national
transmission and storage station pneumatic controller activity and emissions.
API Comment: API recognizes that the stratification of pneumatic controllers into specific bleed
rate categories can be challenging. API has recently engaged in technical assessments of
pneumatic controllers" categories and their leakage vs. engineered venting characteristics^. Over
the past year, through API's standard development process including a stakeholders group, API
has been working to establish a process for categorizing properly functioning pneumatic
controllers and to address fugitive emissions from mal-functioning controllers. API hopes that
this standard, when complete, will go a long way towards addressing the issue raised by EPA
above.
Hi-Flow Sampler Measurements
**"' (Question #14 from EPA's memo) Much of the available measurement data on transmission
and storage segment emissions were developed using Hi-Flow Samplers. A recent study,
Howard 2015, highlights potential malfunctions in certain Hi-Flow instruments under
certain conditions that can lead to underestimates. The EPA is seeking stakeholder feedback
on the impacts of the Hi-Flow sampler issue on the results of studies highlighted here and
whether are there methods for recalculating some of the data points to correct for it.
API Comment: The Subramanian study showed good agreement between the concurrent site
level emission source measurements and down-wind tracer flux measurements. The study
report indicates that the dominant uncertainty in the study onsite estimate is due to
uncharacterized emission sources (undetected or identified as inaccessible) rather than
"parametric uncertainty associated with individual measurements or instruments " Based on
this observation by the researcher/author, it might be concluded that the issues identified by
Howard did not appear to have occurred in the measurements conducted during the
Subramanian study.
The June 2015 article by Howard (Energy Science and Engineering 2015; 3(5):443—455, doi:
10.1002/ese3.81) focusses on measurements conducted in the production sector ("UT Phase 1"
Study) and has drawn attention to a sensor response issue that may be averted to a large extent
with a firmware update, careful calibration, and repeated quality control checks during the
measurement process. Allen responded to Howard's article, providing information that extra
steps were undertaken during to ensure the validity of the measurements from the UT Phase 1
study.4
The Hi-Flow instrument is one of a very few existing devices for cost-effectively quantifying
natural gas emissions from fugitive and venting at the emission source, and it is an approved
measurement device under Subpart W. As with any measurement device, uncertainties in
measured data exist and the experience gained by additional field studies is enabling the
31 Simpson. 2014] Pneumatic Controllers in Upstream Oil and Gas. Oil & Gas Facilities Volume 3 Number 5, October,
2014
4 Allen. D.T., Sullivan. D.W., and Harrison, M. Response to Comment on "Methane Emissions from Process Equipment
at Natural Gas Production Sites in the United States: Pneumatic Controllers". Environmental Science & Technology,
49, 3983-3984, doi: 10.1021/acs.est.5b00941 (2015).
11 of 87
-------
API Comments on Updates under Consideration for Natural Gas Transmission and Storage Emissions
research community to alert instrument manufacturers and industry to operation and calibration
problems that ought to be fixed.
API's comments above are based on our long term engagement in reviewing and providing
information for the U.S. GHG Inventory. It includes observations and recommendations for careful
QA/QC of data extracted from the mandatory GHGRP to improve the validity and
representativeness of data used for the U.S. GHG Inventory. We reiterate our recommendation for
EPA to form a multi-stakeholder workgroup to discuss updating the national GHG1 to incorporate
information from recent measurement study results and Subpart W data.
API appreciates the opportunity to provide comments on the proposed revisions to the U.S. national
GHG Inventory and EPA's willingness to work with industry to improve the data used for the
national inventory. API encourages EPA to continue these collaborative discussions and is
available to work with EP A to make best use of the information available under the GHGRP to
improve the national emission inventory. We look forward to continuing our collaborative work in
the GHG1 development process.
Sincerely,
Karin Ritter
cc: Alexis McKittrick, Climate Change Division
12 of 87
-------
Karin Ritter
Manager
Regulatory and Scientific Affairs
1220 L Street, NW
Washington, DC 20005-4070
USA
Telephone
202-682-8340
202-682-8270
March 2, 2016
Fax
Email
ritterk@api.org
www.api.org
Ms. Melissa Weitz
Climate Change Division, Office of Atmospheric Programs
U.S. Environmental Protection Agency - Office of Air and Radiation
1200 Pennsylvania Ave., NW
Washington, DC 20460
weitz.melissa@epa. gov and ghginventory.gov
Re: Updates under Consideration for Natural Gas and Petroleum Production Sector Emissions
and Gathering and Boosting Emission in the 1990-2014 GHG Inventory
Dear Melissa,
The American Petroleum Institute (API) appreciates the opportunity to provide comments on
proposed updates to the 1990-2014 U.S. Greenhouse Gas Inventory (GHGI) for the Natural Gas and
Petroleum Production Sectors, and for Gathering and Boosting emissions.
API continues to compile and analyze emissions data for petroleum and natural gas operations and
appreciates the opportunity to work with EPA on utilizing data provided through EPA's mandatory
greenhouse gas reporting program (GHGRP). API has provided comments and recommendations to
the U.S. EPA on the draft Natural Gas Systems and Petroleum Systems sections of the national
inventory since 2002, including at the recent stakeholder workshop in November 2015 regarding
greenhouse gas (GHG) data for Petroleum and Natural Gas Systems.
For this current review, API provides general comments and also addresses several specific
questions raised in the two EPA memos:
• Inventory of U.S. Greenhouse Gas Emissions and Sinks: Revisions under Consideration for
Natural Gas and Petroleum Production Emissions, February 2016; and
• Inventory of U.S. Greenhouse Gas Emissions and Sinks: Revisions under Consideration for
Gathering and Boosting Emissions, February 2016.
Our review, however, is as comprehensive as is possible within the short response time, overlapping
comment periods for other proposed changes to the GHGRP and the approaching March deadline
for reporting 2015 GHGRP data. On top of our response to these memos, API intends to also
comment on the "public review" version of the 1990-2014 preliminary Draft Inventory of U.S.
Greenhouse Gas Emissions and Sinks that was released on February 22, 2016.
General Comments
• EPA's current methodological updates for natural gas and petroleum production
operations rely primarily on Subpart W reported activity data with a focus on fugitive
emission sources and pneumatic devices. Of note is that the production memo does not
13 of
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
address new measurement studies or updates that were previously outlined in two memos
EPA issued in April 2015.1'2
• EPA's logic, presented in these memos appears inconsistent. First, EPA notes that Subpart
W GHGRP data covers 32% of the active wellheads for 2013 and proposes to use this
percentage to "scale" some emission sources to a national level. Simultaneously EPA states
that the GHGRP Subpart W data covers the majority of national oil and natural gas
production sources. Separately, EPA has also determined that Subpart W covers about 85%
of the GHG emissions from the onshore oil and natural gas production sector - see the
Subpart W Technical Support Document (Table 5, Threshold Analysis for Petroleum and
Natural Gas industry Segment; https://www.epa.gov/sites/production/files/2015-
05/documents/subpart-w tsd.pdf). Clearly, if Subpart W covers 85% of the GHG emissions
from this sector, then there is no basis for changing the GHG1 in a manner that estimates
90% higher overall GHG emissions (based on the recalculated 2013 inventory). This
discrepancy in GHGRP Subpart W emissions coverage must be fully explored and explained
prior to making the proposed changes to derive GHG emissions in the GHG1 for this sector.
Given that the GHGRP Subpart W reported GHG emissions are substantially less than in the
GHG1 for 2013, the scaling to national GHG emissions for the GHG1 should also be less
than the 15% of emissions EPA previously determined are not covered by GHGRP Subpart
W.
• EPA's methodological updates for Gathering and Boosting relies solely on data from the
Mitchell et al.3 and Marchese et al.4 studies. However, the study focused on downwind, site-
level ambient concentration measurements that are not appropriate nor designed to
characterize activity data or emission factors for the Gathering and Boosting sector sources.
• API suggests that EPA review the work of Eben Thoma et al. with the EPA's Office of
Research and Development (ORD) pertaining to off-site ambient concentration type studies,
and the criteria necessary to obtain useful information from such a study as well as the
limitations to the accuracy and usefulness of the information developed.5 The conclusions
are similar to the conclusions from an Australian government commissioned study
conducted by CSIRO.6 (For EPA's convenience, copies of both papers are provided in the
appendix to these comments, beginning on page 20)
1 "Inventory of U.S. Greenhouse Gas Emissions and Sinks: Potential Revisions to Liquids Unloading Emissions
Estimate" April 2015.
2 "Inventory of U.S. Greenhouse Gas Emissions and Sinks: Potential Revisions to Pneumatic Controller Emissions
Estimate (Production Segment)" April 2015.
3 Mitchell, A. L.; Tkacik. D. S.; Roscioli, J. R.; Herndon. S. C.; Yacovitch. T. I.; Martinez, D. M.; Vaughn, T. L.;
Williams, L.L.; Sullivan. M R.; Flocrchinger, C.; Omara, M.; Subramanian, R.; Ziininerle. D.; Marchese, A.J.;
Robinson, A.L. Measurements of Methane Emissions from Natural Gas Gathering Facilities and Processing Plants:
Measurement Results. Environmental Science & Technology. 49, 3219-3227. 2015.
4 Marchese. A. J.; Vaughn. T. L.; Ziininerle. D.J.; Martinez. D.M.; Williams. L. L.; Robinson, A. L.; Mitchell, A. L.;
Subramanian, R.; Tkacik. D. S.; Roscioli. J. R.; Herndon. S. C. Methane Emissions from United States Natural Gas
Gathering and Processing. Environmental Science & Technology. 49, 10718-10727. 2015.
5 Halley L. Brantley,'!",# Eben D. Thoma,*,f William C. Squier.'i' Birnur B. Guven.J and David Lyon§; Assessment of
Methane Emissions from Oil and Ga.s Production Pads using Mobile Measurements
6 Day. S.. Dell'Amico, Fry, R.. Javanmard Tousi. H.. (2014). Field Measurements of Fugitive Emissions from
Equipment and Well Casings in Australian Coal Seam Gas Production Facilities. CSIRO. Australia
2 of 75
14 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
• API is concerned about EPA's intent to utilize the Mitchell et al. and Marchese et al. studies
to develop a station-level emission factor which would significantly limit any evaluation of
source-level emission trends over time. The small population size of the underlying
Mitchell et al. study, the lack of emission source detail, and the numerous compounding
assumptions made in the Marchese et al. study to "scale" the modeled results, may not
provide sufficient certainty to use the study results for GHGI revisions to the Gathering and
Boosting sector.
• Conversely, significant activity data will be available through the GHGRP in coming years.
API urges EPA to delay significant revisions to the GHGI related to Gathering and Boosting
until the GHGRP data are available. At that time, API recommends that EPA provide a
separate accounting of activity data and emissions for Gathering and Boosting sources as a
separate sector or as a subset of the Production sector.
• As stated previously in our comments on EPA's Transmission/Storage memo, API agrees
that updated GHGI emissions data are warranted and as such recommends that EPA form a
multi-stakeholder working group comprised of industry, governmental, and environmental
organizations active in GHG emissions measurements and estimation to evaluate recently
published data that may be used for updating the national GHG inventory. API proposes
that such a working group be convened following the completion of the 2014 GHGI (April)
to provide a structured framework for consultation and review of GHGI updates. An early
start (April 2016) and frequent meetings (every 1-2 months) would provide sufficient time to
review and consolidate information in an informed process for updating the 2015 GHGI and
beyond.
• Additionally, API reiterates that the EPA should carefully analyze and screen GHGRP
reported data in order to improve the validity of data used in the national GHGI. Obvious
data errors and/or outliers should be assessed, corrected or excluded to prevent
disproportionately impacting the derivation of emission factors (EFs) or extrapolation of
information for the national GHGI.
Responses to EPA Questions for Revisions under Consideration for the Production
Sector
General Use of Subpart W Data
S (Question #1 from. EPA's Production memo) The EPA seeks feedback on how to take into
account the reporting threshold when using Subpart W data, and the appropriateness of
using Subpart W-based AFs for the national population of major equipment and pneumatic
controllers.
a. Are other data sources available that would help the EPA determine characteristics
of the non-reporting population9
b. Are other approaches available for scaling up this data for use in the GHGI9
API Comment: Although Subpart W does not capture all U.S. production operations, it is the
most significant source of activity data available. We would expect that production operations
not reporting through Subpart W are likely much smaller facilities, such as those associated with
3 of 75
15 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
stripper wells. It is reasonable to expect a difference in major equipment and pneumatic
controller counts in these smaller facilities compared to facilities that meet the Subpart W
reporting threshold. However, sufficient information for major equipment and pneumatic
controller counts, for emission estimates, is lacking for the facilities that fall below the reporting
threshold. Therefore, although API supports EPA's use of information available through the
GHGRP to update equipment counts in the national inventory, a note of caution is advised when
using the GHGRP pneumatic device count to characterize stripper wells or other smaller
production well types, which tend to typically have fewer, if any, pneumatic controllers for their
operations. As a result, the use of activity factors ( AFs) based solely on average reporting data
in the GHGRP will likely over-estimate equipment counts from non-GHGRP wells.
In addition, estimates of the coverage of the GHGRP would be expected to be different in each
production basin depending on the characteristics of ownership (many small operators vs. larger
companies), historical development trends, and type of production in the region. For example, a
recent analysis of available data in the Barnett Shale7 in 2013 found that the oil and gas well
count in the GHGRP (15,900 wells) only represented 46% of the well count (34,800) derived
from GHGI methods. In that same study, the author estimated 29,900 oil and gas wells from
other available data. This discrepancy highlights the need for more transparency in GHGI well
count methods, as API has previously commented (see Question #7).
The correlation between GHGRP and GHGI well counts would be expected to be worse in other
o
production regions since much of the Barnett Shale development has occurred over the last 8
years for shale oil and gas production, which typically includes more on-site production
equipment and may be more likely to be reported under the GHGRP. In addition, some
operators have begun to move towards multi-well pads and shared production equipment for
multiple wells. Properly-scaling GHGRP and other activity factors to a national level is a
difficult technical challenge that will require substantial data analysis and a multi-stakeholder
group for proper implementation. Such a group should be convened in order to ensure that
future changes to the GHGI represent a true and robust national emissions estimate.
Furthermore, under the GHGRP, companies report devices that do not emit as typical pneumatic
controllers so the population of controllers in the GHGRP data is very different than the
population measured in the GRI/EPA study (conducted in 1992-1993 and published in 1996)
and it is erroneous to take the count of all such devices and scale them up to the national
inventory by using the wellhead count and the emission factors from the GRI/EPA study. For
example, emergency shutdown devices (ESD) are largely designed to emit only during a process
upset in order to shut-in production. Given the infrequency of this type of event, it would be
improper to characterize these controllers in the same way as the continuous vent pneumatics
that are assumed as part of current GHGI inventory factors.
7 Lyon, D R., Zavala-Arai/a. D.. Alvarez, R A., Harriss, R.. Palacios. V., Lan, X., Talbot, R., Lavoie, T., Shcpson. T„
Yacovitch, T. I., Herndon, S. C., Marchese, A.J., Ziininerle. D.. Robinson. A. L. and Hamburg. S. P. Constructing a
spatially resolved methane emissions inventory for the Barnett Shale Region, Environmental Science and Technology.
49. 8147-8157, 2015
8 http://www.rrc.state.tx.iis/oil-gas/maior-oil-gas-format.ioiis/bariietI-steile-iiifomiatioii/
4 of 75
16 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
**"' (Question #2 from EPA's Production memo) The EPA seeks feedback on other data sources
(e.g., Allen et al. 2013 and 2(314, the Prasino Group 2013) that could be considered for the
development of emission factors for equipment leaks and/or pneumatic controllers,
a. Allen et al. 2014 study did not differentiate between controller type possible to
disaggregate the Allen emissions data in a way that would allow th to calculate
emissions for various control types9
API Comments oil Pneumatic Controllers: API commented previously9 that the emission
factors used for quantifying pneumatic controller emissions, especially the intermittent-bleed
controller factor, largely overestimates these emissions. Therefore, if EP A intends to update the
count of pneumatic controllers in the national inventory then EP A must also in parallel (or at the
same time) update the emission factors.
EP A's current memo outlining methodological changes under consideration for estimating
methane (CH4) emissions from production operations does not refer to, nor draw on information
EPA presented in its April 2015 memo on potential revisions to pneumatic controller emission
estimates . In the April 2015 memo, EPA summarized the following studies:
• Allen, D.T., Pacsi, A., Sullivan, D., Zavala-Araiza, D., Harrison, M., Keen, K., Fraser,
M., Hill, A.D., Sawyer, R.F., and Seinfeld, J.HMethane Emissions from Process
Equipment at Natural Gas Production Sites in the United States: Pneumatic Controllers,
Environmental Science & Technology. 10.1021 /es5040156.
• Oklahoma Independent Petroleum Association (OIPA), Pneumatic Controller Emissions
from a Sample of 172 Production Facilities, November 2014.
• The Prasino Group, Final Report- For Determining Bleed Rates for Pneumatic Devices
in British Columbia, December 18, 2013.
• The Independent Petroleum Association of Mountain States (IPAMS) and Western
Regional Air Partnership (WRAP), 2006.
• Central States Air Resources Agencies (CenSARA), 201 1.
In the April 2015 memo, EPA noted that the Allen et al. 2014 study (a.k.a UT/EDF Phase 2
Study) did not differentiate between controller types. However, supplemental information for
the Allen et al. 2014 study does provide classification of pneumatic controllers by Subpart W
types, for a subset of controllers and also determined classification based on gas flow time-
series measured during the study for all measured controllers (refer to Table S4-2 from the Allen
et al. 2014 study1"). EPA could examine this information for updating emission factors for
intermittent-bleed controllers. However, it may be more difficult to analyze the data for high-
bleed versus low-bleed controllers since malfunctioning low-bleed controllers could exhibit
characteristics of high-bleed controllers. It is our understanding that the Allen et al. 2014 study
also collected meta-data for each controller that includes the manufacturer and model number of
each controller and that this information is available upon agreeing to confidentiality provisions.
9 Shires. T.; "Onshore Oil and Gas Production - Pneumatic Controllers". Presented at the Stakeholder Workshop on
EPA GHG Data on Petroleum and Natural Gas Systems, November 19, 2015.
10 Allen. D.T., Pacsi. A., Sullivan, D.. Zavala-Araiza, D„ Harrison, ML, Keen. K„ Fraser. ML, Hill. A.D., Sawyer, R.F.,
and Seinfeld, J.H.. Methane Emissions from Process Equipment at Natural Gas Production Sites in the United States:
Pneumatic Controllers Supporting Information, Environmental Science & Technology. 10.1021, Pneumatics
es5040156_si_001.pdf
5 of 75
17 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
The manufacturer and model number would enable classification of each controller into the
appropriate EPA "bucket" on the basis of the controller design rather than the measured
emission profile.
Generally, the Allen et al. 2014 data showed lower emission rates per controller than the current
emission factors in the GHG1. For example, the current GHG1 emission factor for gas wells is
15.4 scf/hr/controller. On average, the estimate from the Allen et al. 2014 study was 5.5
scf/hr/controller, even accounting for emissions from malfunctioning controllers or related
systems (i.e. a pinhole leak in the control valve) that were included in the emission factor for
pneumatic controllers. There are reasons to believe that the current GHG1 emission factor over-
estimates the emissions from current controllers in operations. For example, many operators
have changed out or retrofitted continuous high-bleed controllers as part of voluntary and
regulatory programs.
The Oklahoma Independent Petroleum Association (OlPA) conducted an analysis of the Allen
et al. 2014 pneumatic data to complement the data from the OlPA study, by including emissions
from leaking or malfunctioning intermittent-bleed controllers. In the Allen et al. 2014 study, 10
of 320 intermittent-bleed controllers (3%) were "high emitters;" (i.e., were either leaking or
malfunctioning and had an average "malfunctioning" emissions factor of 50 scf/hr). The OlPA
study calculated an emission factor for vented emissions from intermittent-bleed pneumatic
controllers of 0.4 scf/hr based on physical observations of actuation frequency and calculated
volume of gas released per actuation. The distinction is that "vented" emissions from pneumatic
controllers represent the gas released due to normal operation of the controller, while
"malfunction" emissions from pneumatic controllers represent leaking or malfunctioning
controllers. Applying the OlPA "vented" emissions factor of 0.4 scf/hr to 3 10 of the properly
functioning intermittent-bleed controllers in the Allen et al. 2014 study, while applying the
"malfunction" emissions factor of 50 scf/hr to the 10 leaking or malfunctioning intermittent-
bleed controllers gives a weighted average emissions factor of 2.0 scf/hr for all intermittent-
bleed controllers ([(3 10 x 0.40 scf/hr) + (10 x 50 scf/hr)]/320controiiers = 2.0 scf/hr). The OlPA
study also provides information on the count of pneumatic controllers for new well sites and old
well sites (including stripper wells and smaller conventional well pads). As shown in the OlPA
study, a robust emission estimate must include understanding the characteristics of both of these
types of wells.
Regarding the Prasino study, API cautions EPA in using data from that study as the focus was
only on pneumatic controllers with manufacturer bleed rates > 6 scfh and thus the Prasino study
is intentionally biased toward high emitting pneumatic controllers.
Overall, while all these recent studies present the most current data available, they likely should
not be EPA's primary source of data due the variability from study to study. Addressing the use
of new measurement data to update the GHG1 would benefit from further evaluation of all
available data by a multi-stakeholder working group. Such an approach would provide for a
structured update of the applicable emission factors to complement the revi sed counts being
obtained from Subpart W. If the EP A decides to update the inventory without such a
stakeholder engagement, API recommends the use of the Allen et al. 2014 study emission
factors for pneumatic controllers, as the best available current data set, which can also provide
improved understanding of these emissions. As an area with expected future studies, EPA
6 of 75
18 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
should consider that understanding of emission rates from this source is likely to evolve in the
near term as new data sets and measurement techniques are considered. API is interested in
maintaining an on-going dialogue of emission sources in this sector.
**"' (Question #3 from EPA's Production memo) The EPA seeks feedback on how to take into
account reported emissions data under Subpart W for major equipment fugitives in the
GHG1. For reporters using equipment leak methodology 1 (98% of reporters in RY2014),
emissions data are reported at the facility level based on use of component-level EFs
specified in the rule, not at the equipment lev< EPA seeks feedback on how to use such
data in developing equipment-specific fugitiv hat could be applied in the natural gas
and petroleum systems sectors of the GHGI. 1. ne suopart W specified EF for reporting
vented emissions from CIPs uses the same basis (GRI/EPA) as the current GFIGF The EPA
is considering adjusting the G lission factor for CIP using Subpart W reported data,
which takes into account operating hours.
API Comment: Existing GHGRP data on fugitive emissions reported for the production sector
is of limited value for the GHGI since it relies on a set of average emission factors per
component counts as prescribed by EPA and does not contain measurement information that
may be useful to update the emission factors. Equipment counts reported through Subpart W
could be useful for updating activity data for the GHGI, but such extrapolations would be
technically challenging as discussed in Question #1. As a result, API strongly encourages a
detailed stakeholder process related to determining the best method for this extrapolation given
the different populations of wells expected to be covered and not covered under the GHGRP.
However, EPA should refrain from using the default component level emission factors specified
for Subpart W to develop equipment-based fugitive emission factors for the GHGI.
Subpart W provides counts of chemical injection pumps (CIPs) and operating hours that can be
used to scale up GHGRP data to a national emission estimate. However, Subpart W does not
provide information to support updating the emission factor for CIPs. The Allen et al. 2013
study (a.k.a. UT/EDF Phase 1 study) provides measurement data for 62 CIPs with an average
emission rate of 0.192 scf CHVmin/device. EPA should consider evaluating this information for
updating both the default emission factor available in Subpart W and the emission factor
currently used in the GHGI.
Calculations Using Subpart W Data
S (Question #4 from EPA's Production memo) The EPA seeks feedback on the methodology
for allocating Subpart W data between the natural gas and petroleum production sectors. Are
other approaches available for allocating Subpart W equipment and pneumatic controller
counts between production types9 For example, one limitation in the current methodology is
that for facilities covering both oil and gas sub-basins and having separators, the count of
separators-per-gas well is equivalent to separators-per-oil well.
7 of 75
19 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
API Comment: Following IPCC guidance, EPA separately reports production operations for
Natural Gas Systems and Petroleum Systems in production operations", while oil and gas
production activities are combined in the GHGRP. EPA's current approach of separating
GHGRP data based on the ratio of oil production basins to high permeability gas, shale gas, coal
seam, or other tight reservoir rock, although somewhat arbitrary is reasonable.
To aid in comparing the GHGI to GHGRP data, API suggests that EPA resolve differences in
emission source types between the two reporting programs and between natural gas and
petroleum production activities. For example:
• Production operators report emissions from associated gas venting and flaring in the
GHGRP, but this source is not included in the GHGI;
• Well drilling emissions are a vented source in the GHGI under Natural Gas Systems, but
combustion and fugitive emissions from well drilling are tracked under Petroleum
Systems;
• "Wellheads" are an equipment category for reporting fugitive emissions in the GHGRP,
but the GHGI reports emissions for associated gas wells, non-associated gas wells (less
wells with hydraulic fracturing), gas wells with hydraulic fracturing, oil wellheads
(heavy crude) and oil wellheads (light crude).
These are just a few examples where inconsistencies in terminology complicate comparing
emissions between the GHGRP and Natural Gas Systems and Petroleum Systems in the GHGI.
•S (Question #5 from EPA's Production memo) The EPA seeks feedback on whether and how
to use Subpart W data to reflect geographic variation of activity factors and/or emission
factors. In the current GHGI, emissions from natural gas systems are calculated separately
for six NEMS regions, and emissions from petroleum systems do not have geographic
variation. The update under consideration is applied at the national level. The EPA plans to
explore options to reflect geographic variation in future GHGIs.
API Comment: In the Natural Gas Systems production sector, EPA reports emission factors
and activity factors by National Energy Modeling System (NEMS) regions. Except for fugitive
emission factors, emission factors vary from year to year due only to slight changes in the
methane composition between each NEMS oil and gas supply region. The methane
compositions are derived from a 2001 GT1 study12 and adjusted year to year using gross
production for NEMS oil and gas supply modelled regions from the E1A.
Distinctions made between eastern and western fugitive emission factors, derived from the 1996
GR1/EP A study were based on operational differences and the extent of production of sour
crude, and are no longer relevant to operations today.
API recommends that EPA drop the breakout of natural gas production data by NEMS region.
This breakout gives a false sense of data accuracy, as most of the emission factor variability is
based on methane concentration and not on different operating practices. In addition, regional
11 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Volume 1, Section 8.0 Reporting Guidance and
Tables, Table 8.2
12 GTI (2001) Gas Resource Database: Unconventional Natural Gas and Gas Composition Databases. Second Edition.
GRI-01/0136.
8 of 75
20 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
data is not needed for the GHGI, as evidenced by the other natural gas and petroleum sectors
that are only reported at the national level.
•S (Question #6 from. EPA's Production memo) The EPA seeks stakeholder feedback on year-
to-year trends in reported Subpart W data, and whether it is more appropriate to recalculate
activity factors and/or emission factors separately for each RY, or to use another approach
(e.g., combine data from multiple early RYs such as the current methodology for
hydraulically fractured gas well completions which uses combined RY2G11 through
RY2013 data to calculate the emission factor).
API Comment: For Subpart W, the 2011 and 2012 GHGRP data include estimates due to the
use of B AMM, and for pneumatic controllers due to the option to estimate counts initially. In
addition, data tend to improve over time as reporters become more familiar with the
requirements and establish more robust reporting processes. API does recognize the value in
using Subpart W data to reflect year to year trends. However, API suggests that early-year
reporting data may not be as accurate as data reported in the third year and beyond. For
production operations, API recommends that EPA use an average of 2013 and 2014 GHGRP
data to update activity factors. As data become available for the Gathering and Boosting sector,
EPA should recognize that reporting year 2016 will include the use of B AMM and even
reporting year 2017 may reflect the learning curve in establishing reporting programs for this
new sector.
•S (Question #7 from. EPA's Production memo) The EPA seeks feedback on how to address
time series consistency in using AFs derived from Subpart W data—i.e., calculating activity
in years between the early 1990s base year and recent Subpart W-era years. As discussed
under "Time Series Considerations" the EPA might use the count of active production wells
as an activity data driver for major equipment and total pneumatic controller counts in
natural gas systems, and simple linear interpolation for petroleum systems. The EPA could
consider taking into account other factors (e.g., year to year production changes). The EPA
seeks stakeholder feedback on other factors that impact equipment counts and potential,
methods to incorporate these factors into the GHGI calculations.
API Comment: API examined the Drillinglnfo (DI) Desktop data over the 1990-2014 period
to determine if there are any unusual peaks or valleys in oil or gas well counts or production
data. The trends for well counts and production data are generally the same, with no apparent
outliers. Therefore, it seems reasonable for EPA to use national well count and production data
to estimate emissions over the inventory time series.
However, API notes that obtaining accurate and replicable well counts is a complex issue. API
is engaged in ongoing discussions with EPA about how to estimate well counts using the
Drillinglnfo (DI) database. At a primary level, these discussions revolve around differences in
how the EPA accesses the DI data versus how API accesses the data. While EPA starts with
actual raw data files, API accesses the data through a desktop application of the data that only
allows for certain search parameters. This means that there are significant differences in how
users can access and search the data, which makes it very difficult to replicate well counts. For
example, because EPA has access to all raw well data, they are able to easily classify wells as
either "oil" or "gas" based on a GOR that they calculate. Through the desktop application
however, wells are classified as "oil" or "gas" based on state definitions that are not consistent
9 of 75
21 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
across all wells. The following table illustrates the differences in well counts accessed by API
through the DI database, compared to well counts reported by EPA for 2013 in the previous
GHGI13
DI Database Well Counts for
2013 (accessed by API) EPA Reported 2013 GHGI Well Counts13*
Gas Wells
417,277
Non-associated gas wells
207,279
Gas wells with hydraulic fracturing
244,017
Gas and Oil Wells
70,679
Associated gas wells
477,023
Oil Wells
455,243
Heavy crude oil wells
38,682
Light crude oil wells
510,005
TOTAL
943,199
TOTAL
1,477,006
* Including 315,000 crude oil stripper wells (<15 Bbls per day); Reference 13 Table A-126
Unless one downloads all of the well data, which is not a feasible solution, the desktop
application does not allow a user to calculate a GOR and use it as a search parameter. API
urges EPA to be transparent in describing how EPA utilizes information in Drilling Info for the
GHGI in order to facilitate comparisons and ensure that there is no undercounting or
overcounting of wells.
We would also like to point out that the noted discrepancies in the well counts are not a new
issue. For example, the U.S. Energy Information Administration (EIA) reports 514,637
producing gas wells for 2011 (as compared to 604,681 in the GHGI published in 2013) and
536,000 producing oil wells (as compared to 220,787 crude oil wells and 315,213 crude oil
stripper wells in the GHGI published in 2013). For 2013, the EIA reports 484,994 producing
gas wells (with gas-oil ratio > 6000 scf/barrel) but does not furnish equivalent information for
oil wells.
The well counts provided in EPA's Production sector memo equal 1,315,196 (Table 4: 2013
wellheads for petroleum & natural gas combined). This value is different from the sum one
derives (per table above) from the respective petroleum and natural gas tables in Annex 3 of the
2013 GHGI. Since EPA is proposing to use the number of wellheads (well count) as the
normalization factor for scaling Subpart W data, it is imperative that the well count be accurate.
API is providing all of these examples to highlight the discrepancies in the data used to update
the emissions estimates for the production sector and the need to have them reconciled by a
transparent and structured process via a multi-stakeholders group, as previously stated.
Other Emission Sources
•S (Question #8 from EPA's Production memo) The EPA discusses potential revisions to the
GHGI production sector structure in a companion memo titled "GHGI of U.S. Greenhouse
Gas Emissions and Sinks: Revisions under Consideration for Natural Gas Gathering and
13 U.S. EPA, 2015, "Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2013", EPA 430-R-15-004, April
15, 2015; Tables A-126 and A-133.
10 of 75
22 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
Boosting Emissions" (February 2016), Potential revisions would include updating some of
the production emission calculation methodologies based on Marchese et al. (2015)
measurement data for centralized production and gathering-only facilities. With such
revisions, certain emission sources would overlap with the Marchese et al. facility-level EF
if current methodology were retained: dehydrator vents, Kinray pumps, arid storage tanks.
The EPA seeks feedback on how to improve GFIGI activity, emissions, and controls data for
sources located at non-gathering production sites based on available Subpart W data,
API Comment: EPA's memo on proposed revisions to the GHGI for Gathering and Boosting
focuses entirely on utilizing information from the Mitchell et al. and Marchese et al. studies.
However, the Mitchell et al. measurements are limited in their use because only downwind, site-
level short-duration "snapshot" measurements were conducted. This approach does not provide
sufficient information to properly characterize emissions at individual sites in gathering and
boosting operations, much less individual sources within the sites.
API recommends that EPA postpone major updates to the GHGI for gathering and boosting
emissions until GHGRP data are available. The GHGRP will provide additional activity data
for gathering operations and will enable EPA to properly characterize equipment populations
and distinguish between production and gathering. When this new information and
characterization become available, API recommends that EPA revise the GHGI to present,
separately, gathering emission estimates from production emission estimates, even if they
ultimately have to be combined for reporting under the IPCC categories. This will align the
inventory with the GHGRP, provide greater transparency, and enable trends to be evaluated. As
stated above, API requests that EPA delay making any significant changes to the methodology
until GHGRP data are available in 2017. At that time, EPA will have facility specific data for a
significant number of Gathering and Boosting facilities in the country, including population
information, activity data, and actual emission data for some sources.
**"' (Question #9 from. EPA's Production memo) The EPA seeks stakeholder feedback on
production sector sources not discussed in this memorandum.
a. For sources whei data are currently available, the EPA seeks stakeholder
feedback on horn a may be used to revise current GFIGI methodologies. For
example, the EPA seeks stakeholder feedback on whether similar methods to those
discussed in this memorandum could be used to scale up sub""** w activity data for
sources such as liquids unloading and hydraulically fracture gas well completions
b. For sources wh- •' GRP data are not currently available, 1 » A seeks stakeholder
feedback on data sources available for updates to those methodologies. The EPA is
considering including emissions from hydraulically fractured oil well completions and
workovers in the GFIGI, using information from the 2015 NSPS OOOOa proposal. In
addition, the EPA seeks stakeholder feedback on any currently available or upcoming
activity and/or emissions data on abandoned wells,
API Comment: (a) For emission sources with data available through the GHGRP, API
recommends that EPA make use of GHGRP information to update the national inventory. As
mentioned in our responses above, the exception to this is where the GHGRP does not collect
new emissions data but utilizes default emission factors, such as for fugitive emissions in
production, pneumatic controllers, pneumatic pumps, compressors in production, and small
dehydrators.
11 of 75
23 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
API previously commented on the use of GHGRP data for gas well completions and workovers
to update emission estimates in the GHGI.14 EPA incorporated updated emission factors for
these sources, although API continues to believe that the emissions data can be well represented
by only two emission factors (completions and workovers vented without REC, and all other
completions and workovers) rather than the four categories used by EPA. These two categories
maximize the use of GHGRP data, will be more straightforward to back cast for previous
reporting years in the GHG1, and are consistent with current practices.
API cautions EPA against using the ratio of well completions and workovers to overall well
counts in the GHGRP, in order to scale up completion and workover counts to the national
level. Completions, by definition, only apply to new wells, although not all new wells are
hydraulically fractured. Information on new wells should be available through EI A or DI
Desktop. Determining an appropriate method of scaling GHGRP data may be best achieved
through discussions and consideration by the multi-stakeholder group suggested by API.
(b) Although not currently required under the GHGRP, some companies have reported
emissions data for oil well completions and workovers with hydraulic fracturing. API
commented previously on the use of GHGRP data to derive emission factors for the GHGI.14
API previously identified 149 reported data sets, providing emissions data for 1675 completions
and 226 workovers for the years 201 1 through 2013 combined (we have not examined the 2014
GHGRP data to update this analysis). API believes the GHGRP provides sufficient data to
include these emissions in the GHGI, and that much more information will be available in the
next few years to update the national emission estimates.
The DI database provides activity data for abandoned wells. A 2014 study (Kang et al.15)
provides information on emissions from abandoned wells in the Appalachia region. However,
many of these wells are very old, predate any abandonment criteria, were not properly
abandoned and were limited to a single geographic region. Therefore, while the study did
provide new information, the findings should not be considered as representative nor used as the
basis for national extrapolation. A proper data set is needed that reflects geographical
variability and well-age to represent emissions from abandoned wells on a national basis.
**"' (Question #10 from EPA's Production memo) Recent production sector studies have
detected the presence of super emitters in the product!on sector. The EPA seeks stakeholder
feedback on how to incorporate information on super emitters into estimates for the
production sector. The EPA also seeks stakeholder feedback on which GHGI sources are
more likely than others to act as super emitters and whether and how to apply a super emitter
factor or other methodology to those sources.
14 Letter to Lcif Hockstad and Melissa Weitz, API Expert Review Comments on EPA's Draft U.S. GHG Inventory:
1990-2013. January 9. 2015.
15 Kang et al. (2014) "Direct Measurements of Methane Emissions from Abandoned Oil and Gas Wells in
Pennsylvania". Proceedings of the National Academy of Sciences of the United States of America. Available at:
http://www.pnas.org/content/lll/51/18173.Ml.pdf]
12 of 75
24 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
API Comment: Recent measurement studies have shown skewed "long tail" distributions for
source-level measurements, where a few emission sources may contribute a disproportionately
high fraction of emissions. This is a common and expected statistical distribution for random
events, such as fugitive emissions from process components and equipment malfunctions. A
combination of variability in production and non-steady state emissions may result in a 'fat-tail"
distribution even in the absence of operational upsets. Emission factors derived from such
measurements already account for the emission distributions throughout the range of
observations for each of the sources, including the emissions at the high range of the tail.
The approach raised by EPA, of potentially, separately adjusting the national inventory for the
so called 'super emitters," is not appropriate. API contends that there should not be any
consideration of using downwind offsite measurements - especially those that depend on short
duration, snapshot measurements - to characterize emissions in the GHG1. Recent studies in the
Barnett Shale region indicate that there might be several order of magnitude differences in
repeated emissions from a given set of sites, probably due to stochastic variables that are
transient in nature. In particular, a study of 22 separate flights around the same compressor
station16 indicated that facility-level emissions ranged from 0.3 to 73 g CH4/sec with highly
skewed distributions (mean=14 g/sec and median = 7.4 g/sec). Again, API suggests that the
EPA inventory team consult with the EPA ORD's Eben Thonia regarding the adequacy of
downwind ambient concentration measurements in determining emissions.
All the studies aiming to quantify fugitive emissions indicate that the distribution of emissions
and the shape of its tail are not well understood. API insists that both EP A and the scientific
community do not have enough information to identify the reasons for the variability of some
emission sources. All measurements have some degree of uncertainty. This is especially true
for short duration snapshot measurements conducted offsite, which fail to differentiate between
routine episodes of high emissions, operating conditions, or operators errors that may lead to
17
periodic higher emissions. For example, one study focused on "super-emitter" quantification
in the Barnett Shale and relied on measurements of 1-5 minutes in duration at distances of up to
several kilometers downwind in a region with high oil and gas site density.
EPA's ORD research5 that was conducted with strict data quality control parameters, longer
sampling times, and nearer pad sampling, indicated that, at best, downwind measurements
provide screening level accuracy with ±60%. Insufficient research exists to validate high
downwind measurements with on-pad emission sources such that it could be used to
characterize national emission estimates for a program like the GHG1. API concurs with EPA's
ORD that in order to properly quantify emissions measurements, they should be taken over a
long period of time in order to capture the full range of variability, rather than rely on just peak
emissions. Assuming that peak emissions occur all the time would lead to biased results.
16 Nathan, B.J.. Colston, L. ML, O'Brien. A.S., Ross, K. Harrison, W. A., Tao. L., Larv. D. J., Johnson. D. R..
Covington. A. N., Clark, N. N., and Zondlo. M. A., Near-field characterization of methane emission variability from a
compressor station using a model aircraft. Environmental Science & Technology. 49,7896-7903 2015
17 Yacovitch. T. I., Herndon, S.C.. Petron, G.. Kofler. J., Lyon, D„ Zahniser. M. S. and Kolbacovitch. C. E. et al. Mobile
laboratory observations of methane emissions in the Barnett Shale region. Environmental Science and Technology. 49,
7889-7895, 2015
13 of 75
25 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
In summary, API maintains that adjusting emissions for what EPA terms "super emitters" may
lead to gross overestimation due to the unpredictable nature of such high emissions events and
may also lead to duplicative counting, since these events are already part of the emission
distribution that is used to derive emission factors. For example, if a connection failure is
posited as the cause of a theoretical site being deemed as a theoretical "super-emitter", emission
factors developed from in-field measurements of a population of connectors already account for
some of these components emitting at a high rate. Consequently, API insists that since EP A
does not have sufficient information to characterize and understand this then no such adjustment
to the GHGI inventory approach should be considered.
Responses to EPA Questions for Revisions under Consideration for Gathering and
Boosting Emissions
Data Availability
**"' (Question #1 from. EPA's Gathering and Boosting memo) The EPA is seeking stakeholder
feedback on additional, data available to consider in revising G&B emission estimates at this
time. The EPA seeks stakeholder feedback on the proposed approach to use Marchese et al.
estimates for national, activity data. Are additional data sources or approaches available to
estimate national. G&B activity9
API Comment: The Marchese et al. study results are based on facility level, downwind short-
duration "snapshot" measurements conducted during the Mitchell et al. study. Marchese et al.
used that data to model the total methane emissions from approximately 120 facilities. The
modeled results are then "scaled" - using multiple assumptions - to a national level to represent
the methane emissions from over 4,500 Gathering and Boosting facilities.
As indicated in our general comments, API urges EPA to wait on any significant revisions to the
GHGI related to Gathering and Boosting until the GHGRP data are available. Significant
activity data will be reported through the GHGRP, including throughput volumes and equipment
counts. This information will be superior to the Marchese et al. study for developing national
Gathering and Boosting activity data.
•S (Question #2 from. EPA's Gathering and Boosting memo) Replacing current GHGI. EFs for
large reciprocating compressors and stations with the EF based on Marchese et ah G&B
station emissions may introduce double counting of the "mixed category" sources based on
current GHGI methodology. The EPA's updates under consideration for the G&B sector
(this memorandum) and production sector (Inventory/ of U.S. Greenhouse Gas Emissions
and Sinks: Revisions under Consideration for Natural Gas Production Emissions (February
2 i combination avoid potential double counting issues by calculating emissions for
each as distinct sectors. Please comment on the overall approach, under consideration for
production and G&B.
API Comment: The Mitchell et al. study relies on offsite, downwind measurements, using
inverse flux methodology to derive emissions over short durations. These types of
measurements have significant uncertainty, which has been documented by EPA's ORD5.
EPA's proposed approach to segregate Gathering and Boosting emissions from Production is
specifically designed to utilize data from the Marchese et al. study4, which is a desktop
-5
modeling study based on the Mitchell et al. measurements but is inconsistent with the Mitchell
14 of 75
26 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
et al study. API does not support the use of the emissions data from the Mitchell et al. or
Marchese et al. studies for updating the GHG1.
The API further cautions the EPA on the development of new national emissions factors based
on the Mitchell et al. study due to the large degree of variability and small sample size for the
study. For the 1 14 facilities, emission rates ranged more than 4 orders of magnitude (from 0.6
to 600 scf CH4/minute). Part of this variability is inherent in the short sample durations for the
plumes in the study (30-120 seconds). Given the wide variation in facility emission rates from a
study of 22 separate flights around the same compressor station16, which indicated that facility
level emissions ranged from 0.3 to 73 g CHVsec with highly skewed distributions (mean=14
g/sec and median = 7.4 g/sec), more context is needed for understanding emission rates in the
Mitchell et al. study before considering application to national emission estimates.
In attempting to avoid double counting of emissions sources, EPA is artificially defining
Production versus Gathering and Boosting equipment. For example, EPA is proposing to assign
emissions from all pneumatic controllers, chemical injection pumps, dehydrator vents, and
Kim ray pumps to the Production sector. This will give the false impression that these sources
only occur in Production.
API recommends that EPA wait until data are available through the GHGRP for the Gathering
and Boosting sector. We believe this information will better represent the emission sources
associated with Gathering and Boosting (recognizing that some Gathering and Boosting
operations will continue to be reported under the Production sector due to the location of a well
at the Gathering/Boosting site). In addition, we recommend that EPA report emissions from
Gathering and Boosting separate from the Production sector, or as a subset of the Production
sector. This will provide greater transparency and comparison to the GHGRP than combining
Gathering as part of the Production sector, as is currently reported in the GHGI.
**"' (Question #3 from EPA's Gathering and Boosting memo) As discussed in this
memorandum, G&B data will be available in 2017 through GHGRP. GHGRP data could
allow the EPA to calculate emissions for individual equipment types as opposed to using
emission factors and activity data at the station level The EPA seeks stakeholder feedback
on the two approaches. The EPA could considering using the station level approach for the
IGI, and then re-evaluating and potentially revising the approach with new GHGRP
data in the 2017 GHGI, or could consider implementing updates to the G&B sector starting
with the 2017 GHGI and using GHGRP and/or the Marchese et al. data at that time.
API Comment: API does not believe the Marchese et al. study results are appropriate for
updating the national inventory and encourages EPA to wait until the Gathering and Boosting
data are available through the GHGRP. As EPA indicates, the GHGRP data will allow the EPA
to calculate emissions for individual emission source types as opposed to using emission factors
and activity data at the station-level. Data for individual equipment types will be significantly
more useful and transparent than emission factors and activity data at the station level. There is
no need to introduce a significant revision to the GHGI now to accommodate the Marchese
study information, only to later have to significantly revise the methodologies again to utilize
the GHGRP data.
15 of 75
27 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
**"' (Question #4 from EPA's Gathering and Boosting memo) The EPA seeks feedback on
whether and how to use the Marchese et ah data to reflect geographic variation of activity
factors and/or emission factors,. In the current GHGI, emissions from G&B sources are
calculated separately for six NEMS regions along with production sources. The update
under consideration would be applied at the national level. The EPA plans to explore options
to reflect geographic variation in future GHG inventories.
API Comment: The small population size of the underlying Mitchell et al. study, the lack of
emission source detail, and the numerous compounding assumptions made in the Marchese et al.
study to extrapolate the modeled results do not provide sufficient certainty to use the study
results to characterize the Gathering and Boosting Sector. Nor does the Marchese study provide
sufficient information to characterize geographic variability. As mentioned above, in response
to questions raised in the Production memo, API recommends that EPA discontinue breaking
out natural gas production data by NEMS region and instead report Production sector emissions
data at the national level only, as EPA does for the other sectors under Natural Gas Systems and
Petroleum Systems. Similarly EPA should not attempt to calculate emissions from the
Gathering and Boosting sector for individual NEMS regions.
Time Series Considerations
**"' (Question #5 from. EPA's Gathering and Boosting memo) The EPA seeks feedback on the
appropriateness of using the Marchese et al. based G&B station EF across all years of the
time series, or whether there are approaches that may be considered for reflecting changing
industry trends impacting emissions over time,
API Comment: The Marchese et al. study, which is based primarily on drive-by, snap-shot
measurements from the Mitchell et al. study, does not provide useful data for characterizing
current national emissions, nor does it provide sufficient information to reflect emission trends
over time.
**"' (Question #6 from. EPA's Gathering and Boosting memo) The EPA seeks stakeholder
feedback on the activity driver (volume of marketed onshore gas production) under
consideration. Other options for the activity driver could include well count data or other gas
production categories. Please comment on which activity driver would be the most
appropriate to show trends in G&B.
API Comment: EPA will have significant activity data reported for the Gathering and
Boosting sector through the GHGRP starting in 2017. API recommends that EPA evaluate this
information when it's available to identify activity drivers for scaling Gathering and Boosting
emissions data to a national level. API also points out that it may take more than one reporting
cycle to work through data quality concerns associated with the first year of reporting for a new
sector.
S (Question #7 from. EPA's Gathering and Boosting memo) The EPA seeks stakeholder
feedback on trends in G&B activity data that would result in more or fewer stations per
volume of marketed onshore gas production during any point in the GHGI time series. The
EPA requests stakeholder feedback on how upcoming subpart W G&B activity data
(available in 201.7) could be used to inform the time series activity data to reflect ongoing
trends.
16 of 75
28 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
API Comment: As noted in our comment to question #6, API expects the activity data reported
through the GHGRP for the Gathering and Boosting sector to provide significant information
for developing national scaling factors and similarly will be appropriate data for informing
activity data over the time series.
**"' (Question #8 from EPA's Gathering and Boosting memo) Since the EI A does not publish
separate values for the onshore portion of marketed natural gas production prior to 1992, the
EPA is considering using the relationship of onshore marketed production to onshore gross
withdrawals in 1992 to estimate marketed onshore production in 1990 and 1991, based upon
onshore gross withdrawals for these two years. Are there alternatives to addressing this
missing AD9
API Comment: API supports EPA's proposal to relate onshore marketed production to
onshore gross withdrawals in 1992 in order to estimate marketed onshore production in 1990
and 1991.
**"' (Question #9 from EPA's Gathering and Boosting memo) Although it is not possible to
directly compare the G&B emissions estimate developed with GR1/EPA study data to the
Marchese et al results, it is evident that the G&B emissions from Marchese et al. are
significantly higher than estimates in the current GHGI. The EPA seeks stakeholder
comment on this discrepancy.
API Comment: It is not appropriate to compare the Marchese et al. modeling information
which is based on short-duration, off-site ambient concentration measurements, which rely on
inverse flux methods to derive emissions; to source specific emission estimates. The site level
measurements conducted in the Mitchell et al. study significantly limit the use of the data for
updating the national inventory, which is compiled from source level emission estimates. API
urges EPA to delay revising the emission estimation methods for the Gathering and Boosting
sector until more data is available for this sector through the GHGRP.
Gas Processing
S Marchese et al also measured the methane emissions from 16 natural gas processing plants
using a similar approach as described above for G&B stations. The results of the Marchese
et al. testing were scaled to the estimated 600 national gas processing plants using a similar
Monte Carlo simulation as was used for G&B stations. The results of the Marchese et al.
simulation was a national methane emission estimate for gas processing plants of 506 Gg.
As with the G&B stations, Marchese et al. estimated that the emission results were biased
low for several factors. The brief sampling period did not capture routine maintenance and
upset emissions. In addition the sampling method did not capture a significant portion of the
compressor exhaust emissions, Marchese et al. compared their findings to the E GI
of 2012 emissions. The net GHGI methane emissions for 2012 from processing plants were
891 Gg. The net GHGI emissions from processing plants, excluding compressor exhaust
and blowdown/venting emissions were estimated to be 666 Gg. EPA seeks stakeholder
comment on the potential use of Marchese et al. results for the processing sector.
API Comment: As mentioned previously, measurement data from the Mitchell et al. study are
not particularly useful for updating the GHGI because the data lack emission source detail.
Substantial new activity data and some measurement data are available for gas processing
17 of 75
29 of 87
-------
API Comments on Updates under Consideration for Natural Gas and Petroleum Production Emissions, and Gathering
and Boosting Emissions
facilities that report through Subpart W. EPA now has four years of fugitive emission surveys
and measurement data for specific emission sources and activity data that can be used to update
the GHGI. API encourages EP A to make use of the survey results and actual measurements
reported in GHGRP. In the November 2015 stakeholders" workshop, API presented a
preliminary comparative analysis of methane emissions from equipment leaks from natural gas
processing, showing that it is about six times larger in the GHGI as compared with the GHGRP.
Although the number of gas plants reporting to the GHGRP is different than the number of gas
plants in the GHGI, this difference cannot fully account for the emission differences. API would
welcome further collaboration with EPA to address these differences and develop a procedure
that incorporates the GHGRP measurement data in the GHGI.
API's comments above are based on our long term engagement in reviewing and providing
information for the U.S. GHG Inventory. It includes observations and recommendations for careful
QA/QC of data extracted from the mandatory GHGRP to improve the validity and
representativeness of data used for the U.S. GHG Inventory. API recognizes that emerging data
from recent field studies have raised concerns about measurements uncertainty, and recognizes the
need for a thorough discussion of means of improving the methodology to ensure collection of
robust measurement data. We reiterate our recommendation for EPA to form a multi-stakeholder
workgroup to discuss updating the national GHGI to incorporate information from recent
measurement study results and Subpart W data.
API appreciates the opportunity to provide comments on the proposed revisions to the U.S. national
GHG Inventory and EPA's willingness to work with industry to improve the data used for the
national inventory. API encourages EPA to continue these collaborative discussions and is
available to work with EP A to make best use of the information available under the GHGRP to
improve the national emission inventory. We look forward to continuing our collaborative work in
the GHGI development process.
Sincerely,
Karin Ritter
cc: Alexis McKittrick, Climate Change Division
18 of 75
30 of 87
-------
Appendix D
-------
This is an open access article published under an ACS AuthorChoice License, which permits
copying and redistribution of the article or any adaptations for non-commercial purposes.
fponpu
twcesTedinoogij
Assessment of Methane Emissions from Oil and Gas Production Pads
using Mobile Measurements
Halley L. Brantley/ '' Eben D. Thoma,*'' William C. Squier,' Birnur B. Guven,: and David Lyon®
'Office of Research and Development, U.S. Environmental Protection Agency, 109 T.W. Alexander Drive, Research Triangle Park,
North Carolina 27711, United States
^Houston Advanced Research Center, The Woodlands, Texas 77381, United States
^Environmental Defense Fund, 301 Congress Ave., Suite 1300, Austin, Texas 78701, United States
© Supporting Information
¦CSAjttaOicict
Article
pubs.acs.org/est
ABSTRACT: A new mobile methane emissions inspection Wind Direction
approach, Other Test Method (OTM) 33A, was used to .—V
quantify short-term emission rates from 210 oil and gas
production pads during eight two-week field studies in Texas,
Colorado, and Wyoming from 2010 to 2013. Emission rates
were log-normally distributed with geometric means and 95%
confidence intervals (CIs) of 0.33 (0.23, 0.48), 0.14 (0.11,
0.19), and 0.59 (0.47, 0.74) g/s in the Barnett, Denver-
Julesburg, and Pinedale basins, respectively. This study focused
on sites with emission rates above 0.01 g/s and included short-
term (i.e., condensate tank flashing) and maintenance-related
emissions. The results fell within the upper ranges of the
distributions observed in recent onsite direct measurement
studies. Considering data across all basins, a multivariate linear
regression was used to assess the relationship of methane emissions to well age, gas production, and hydrocarbon liquids (oil or
condensate) production. Methane emissions were positively correlated with gas production, but only approximately 10% of the
variation in emission rates was explained by variation in production levels. The weak correlation between emission and
production rates may indicate that maintenance-related stochastic variables and design of production and control equipment are
factors determining emissions.
¦ INTRODUCTION
Environmentally responsible development of oil and gas assets
requires an understanding of atmospheric emissions of methane
(CH4) and other organic pollutants as well as their potential
impact on local and regional air quality and greenhouse gas
budgets. Emissions are associated with many different processes
in upstream (well development and production) and midstream
(transportation and storage) oil and gas activities.1,2 Although
differing in profile, emissions occur in all phases of well
construction, drilling, and completion, and continue as part of
the ongoing production processes.3 Oil and gas production
pads (pads) typically consist of well heads, separation units, and
storage tanks. Emissions from pads can be difficult to measure
and model due to temporal variability and the large number of
potential sources.4,5 Pad emission profiles depend on a variety
of factors including the geological formation, equipment design
and maintenance state, and on operational procedures. For
example, depending on engineering and control strategies,
atmospheric-pressure condensate storage tanks are a significant
potential source of emissions and can be challenging to
measure.6,' Pad emissions can also vary over time as wells age
and production levels and pressures change. Improving our
understanding of emissions from production sites requires a
combination of approaches, including estimating emissions
using engineering calculations for inventories,2,8,9 direct
measurements for refinement of emission and activity factors,10
and new inspection techniques to inform departures from
routine operations and support compliance activities.11
Direct (onsite) measurements can provide information on
component-level emissions, but are resource intensive,
requiring site access and special safety considerations.
Furthermore, the high site-to-site variability decreases the
probability of obtaining a representative sample from a small
number of sites. To complement direct measurement
approaches, a number of research groups are investigating the
use of mobile inspection techniques to locate and assess
emissions from off-site observing locations.4,12 14 These
emerging approaches vary with respect to execution require-
ments and emission estimation techniques; however, their
mobile nature facilitates identification of unknown emission
sources (e.g., pipeline leaks) and anomalous operating
Received: June 24, 2014
Revised: September 23, 2014
Accepted: November 6, 2014
Published: November 6, 2014
^ ACS Publications * 2014 American Chemical Society
14508
dx.doi.org/10.1021/es503070q I Environ. Sci. Technol. 2014, 48, 14508-14515
2° of 75 32 of 87
-------
Environmental Science & Technology
conditions (e.g., malfunctions). Unlike direct measurements,
mobile approaches typically cannot isolate specific emitting
components and are generally less precise than direct measures
but are comparatively easier to implement, enabling emission
assessments to be made at a greater number of locations on a
more routine basis.
This paper describes a novel mobile inspection approach,
EPA Other Test Method (OTM) 33A,17 and its use to generate
CH4 emission rate data from oil and gas production sites in the
Denver-Julesburg (Dj) Basin, the Barnett Shale, Pinedale, and
Eagle Ford from 2010 to 2013. OTM 33A uses a combination
of mobile sampling to identify sources and stationary
measurements to quantify emissions. In addition to the analysis
of repeated measurements at nine sites, the emission estimates
from the OTM 33A field studies were compared with recent
on-site studies led by the Eastern Research Group (ERG)15 and
Allen et al.16 The ERG study,15 conducted for the City of Fort
Worth, TX, used both direct measurement and source
estimation methods to characterize CH., and volatile organic
compound emissions at 388 production sites containing wells,
produced water storage tanks, separators, and compressors.
Component-level source identification in the ERG study15 was
accomplished by infrared camera observations and direct source
measurements were conducted using Hi Flow samplers
(Bacharach Inc., New Kensington, PA), toxic vapor analyzers,
and evacuated canisters. The measurements were used by the
City of Fort Worth to evaluate the adequacy of setback
provisions for pads and compressor stations. The results of the
ERG study15 indicated that compressors, leaking tank thief
hatches, and pneumatic valve controllers are the most
frequently encountered and significant emissions sources of
CH.,. Using similar on-site measurement techniques, Allen et
al.16 measured CH, emissions from ISO production sites in four
regions of the United States to evaluate engineering estimates
of CH., emissions from natural gas production that are used in
national inventories. Their results indicated that emissions from
pneumatics and equipment leaks were higher than estimated in
the EPA greenhouse gas (GHG) emissions inventory.16
M MATERIALS AND METHODS
OTM 33A17 is a mobile inspection approach used to locate
sources and determine real-time emission rates with screening-
level accuracy (±60%), without the need for site access or
location-specific modeling. The technique is applicable to select
oil and gas sources such as roadway proximate pads located in
relatively open areas. In addition to downwind vehicle access
and favorable plume transport conditions required for all
mobile assessment methods, the emission characterization
portion of OTM 33A relies on relatively consistent
meteorological conditions, obstruction-free line of sight
observation, and a knowledge of the distance to the source.17
Sampling Platform Design and Protocol. The OTM
33A equipment configuration, further described in OTM33A
Appendix A,17 used either a G1301 -fc cavity ring-down
spectrometer (Picarro, Inc., Santa Clara, CA) or a GG-24-r
off-axis integrated cavity output spectrometer (Los Gatos
Research Inc., Mountain View, CA) as CH, concentration
measurement instruments (CMIs). The mobile measurement
platforms were sports utility vehicles containing the CM I,
computer control system, and battery systems allowing engine-
off instrument operation during stationary observations to
prevent self-sampling of vehicle exhaust The vehicles were
fitted with rotatable front-mounted masts with a height of 2.7 m
allowing the CMI probe and meteorological instruments to be
located away from the body of the vehicle. Primary wind field
data were acquired using a model 81000 V Ultrasonic
Anemometer (RM. Young, Inc., Traverse City, Ml). A
collocated compact weather station (model AIO 102780,
Climatronics Corp., Bohemia, NY) provided secondary wind
data along with temperature, atmospheric pressure, and relative
humidity measures. Location was recorded using a Hemisphere
Crescent R100 Series GPS system (Hemisphere GPS, Calgary,
AB Canada). A LabView (National Instruments, Inc., Austin
TX) computer program time-aligned the data stream while
allowing user control of the system.
The accuracy, linearity, and range of the CH., CMIs were
confirmed in predeployment testing with in-field accuracy
verified to be within ±5% of actual using nominal 20 ppm CH,
(air balance) gas standard challenges as per OTM 33 Section
9.4.17 The CMI readings were not corrected for atmospheric
water vapor (OTM 33A Appendix A)17 which introduces an
approximate 1.5% average negative bias to CH., emission
determinations for the conditions encountered in this study.
For a typical pad assessment, emissions were located through
downwind, drive-by inspection, keying on sharply elevated CH,
spikes indicative of proximate source plumes. Maximizing real-
time CH., concentrations measured by the CMI, the vehicle was
positioned in the plume at a safe and appropriate downwind
observing location with the probe facing the source, and the
engine was turned off. Distance from the measurement vehicle
to the emission source ranged from 10 to 200 m with an
average distance of 57 m. Data were acquired for a 15 to 20 min
time period with the vehicle remaining stationary. Auxiliary data
from infrared cameras (FLIR Systems, Inc., Boston MA), when
available, helped identify the source location, facilitating laser
rangefinder measurements of the distance from the mobile
platform to the source. Distances were later confirmed through
Google Earth images coupled with wind-concentration rose
data. The vehicle was positioned to minimize line-of-sight wind
flow obstructions.
Emission rate estimates were calculated using a point source
Gaussian (PSG) approach with a custom MAT LAB (Math-
Works, Natick, MA) analysis program (OTM 33A Appendix
Fl).17 This approach relies on variations in wind direction to
move the plume around the observation location in three
dimensions; further assumptions include a point source and
Gaussian plume dispersion. The analysis software time-aligned
the measurements to correct for sampling line delay, rotated
the 3-D sonic anemometer data to polar coordinates centered
on the predominant wind direction, and binned the CH.,
concentrations by wind direction data in ten degree increments.
The results were fitted with a Gaussian function to determine
the average peak CH., concentration in the plume. Background
concentrations were determined by the program during time
periods with no plume-probe overlap (OTM 33A Section
8.7).17 The program calculated the representative atmospheric
stability indicator (ASl) from an average of the turbulence
intensity (Tl), measured by the 3D-sonic anemometer and the
standard deviation in 2-D wind direction (off), acquired by the
compact meteorological station. By defining a seven unit ASI
scale with steps of equal increments (TI = 0.025, 00 = 4.0°), an
ASI value for each measurement was assigned which ranged
from 1 (TI > 0.205, o0 > 27.5°) to 7 (TI < 0.08, a9 < 7.5°),
roughly corresponding to the Pasquill stability classes A
through D.18 For the PSG emission estimate, the values of
horizontal (ay) and vertical (oz) dispersion are determined
14509
dx.doi.org/10.1021/es503070q I Environ. Sci. Techno!. 2014, 48, 14508-14515
21 of75 33 of 87
-------
Environmental Science & Technology
from an interpolated version of point source dispersion tables
using the measured source distance and the ASI (OTM 33A
Section 12, Appendix Fl).17 The PSG emission estimate (q) is
a simple 2-D Gaussian integration (no reflection term)
multiplied by mean wind speed (u) and the peak concentration
(c) determined by the Gaussian fit: (q = In-ay-oz-u-c).17
Method Validation Using Controlled Release Experi-
ments. A set of 107 controlled CH4 release experiments were
conducted to investigate data quality indicators and the
expected accuracy range for the PSG approach in relatively
obstruction-free, open areas as encountered in this study
(OTM 33A Section 9).17 The experiments used single point
releases from slightly dispersed, mass flow-controlled cylinders
of 99.9% CH4, performed at a variety of site locations,
observation distances, and under a range of atmospheric
conditions. Release rates ranged from 0.19 g/s to 1.2 g/s with
60% at approximately 0.6 g/s. Based on these experiments, a
primary set of three data quality indicators was identified: (l)
fitted peak CH4 concentration centered within ±30 degrees of
the source direction; (2) an average in-plume concentration
greater than 0.1 ppm; and (3) a Gaussian fit with an R2 > 0.80.
The plume centering indicator helps ensure the identity of the
upwind source and can protect against off-axis interfering
sources and poor plume advection conditions. The concen-
tration limit helps protect against insufficient plume transport
and the R2 indicator helps identify interfering sources and
obstructed wind flow conditions (non-Gaussian transport).
The percent error ([estimated emission rate-release rate]/
[release rate]) of the controlled release experiments that met
the data quality criteria ranged from —60% to 52% with 72% of
the measurements within ±30%. Without application of the
data quality indicators, the set of release experiments produced
accuracy values ranging from —87% to 184% of actual. The
184% overestimate was believed to be due to pooling and
release under partially stagnant conditions and a trial wind
variance indicator was developed for this case (not observed in
field trials). Factors affecting accuracy can include insufficient
plume advection and nonrepresentative concentration profiles
caused by near-field obstructions or poor plume-probe overlap.
Potential data quality indicators such as wind speed and plume
concentration statistics are being investigated as part of OTM
33A method development.17 For the current analysis, only
measurements that met the three primary criteria were included
(representing 77% of the controlled release measurements and
71% of the field measurements).
Description of Field Studies and Production Data.
OTM 33A was used in eight two-week field campaigns in four
oil and gas production basins: Colorado D| Basin, July 2010
and 2011; Texas Barnett shale, September 2010 and 2011;
Texas Eagle Ford Shale, September 2011; and Wyoming
Pinedale, which includes the Pinedale Anticline and Jonah
fields, June 2011, July 2012, and June 2013. Data sets for each
individual basin were combined as the methods of data
collection were similar, although there were some software and
hardware improvements in later studies. All measurements were
collected in the daytime on days with no significant
precipitation.
Oil and gas production information for the counties sampled
was obtained from DI Desktop (Drillinginfo, Austin, TX).
Included in the data set were well type, operator, first
production date, spatial coordinates of the well, and annual
and monthly hydrocarbon liquids, gas, and water production
levels. OTM 33A measurements were spatially matched with
production data using aerial imagery (Google Earth19 and
ArcGIS20 base maps). When coordinates did not align with
aerial imagery, additional data sets provided by the State of
TX21 and State of CO22 were used to cross-reference location
information. Monthly production values were available for 81%
of the measurements. When monthly production was not
available, annual values were converted to monthly estimates.
The matched data set was analyzed using R23 and ArcGIS 10.20
Both emissions estimates and production values were log-
normally distributed and for this reason, data in figures are
shown on a log scale. The mean and 95% CI of the log-
transformed data were calculated using a nonparametric
bootstrap24'25 and then transformed back into the original
scale. The nonparametric bootstrap involved resampling with
replacement 1000 times, the mean of each of the samples was
taken and the 95% CIs were calculated from the resulting
normally distributed means. The nonparametric bootstrap was
chosen because it does not assume the underlying data comes
from a normal distribution. To compare OTM 33A emissions
estimates with the direct measurement studies conducted by
ERG15 and Allen et al.,16 direct measurements were converted
from CH4 scfm into g/s using a molar volume of 40.87 mol m~3
and summed by site. Measurements from the ERG study15
were matched with the corresponding monthly production
values from DI Desktop (Drillinginfo, Austin, TX) based on the
recorded Entity ID. Production values for the sites measured by
Allen et al.16 were reported by the well operators to the study
team.
B RESULTS AND DISCUSSION
Description of Sites with Repeat Measurements. The
OTM 33A mobile inspection approach was used to identify and
assess CH, emissions from roadway proximate well pads with
an average in-plume concentration enhancement over back
ground >0.1 ppm. No attempt was made to measure or
statistically account for well pads with apparently low (and thus
difficult to measure) emissions. In many cases, infrared camera
videos (examples in Supporting Information (Si) Supplemental
B) acquired from off-site observing locations, simultaneously
with the CH4 measurements, helped to identify specific
emission sources. Storage tank-related emissions were
frequently observed. The emission rates and video examples
presented here may not be representative of current conditions
due to engineering advancements, changes in work practices,
and the implementation of new state regulations.
To improve understanding of both technique and source
variability, repeat measurements (three or more) were made at
nine sites in the Pinedale Basin, with the number of
measurements per site ranging from 3 to 21 (SI Table Si).
The consistent winds and lack of obstructions in the Pinedale
Basin create favorable conditions for OTM 33A Measurements
were made in different years at four of these sites (Figure l),
and the time between measurements ranged from < 1 day to
732 days (SI Table Si). For sites A—G, the 95% CI for the
geometric mean was less than 1 g/s while at sites H and I, large
variations in emissions were observed, resulting in a CI > 2 g/s
(SI Table Si).
The results indicate that while relatively low emissions (<2
g/s) frequently persist over time, the larger emissions observed
using OTM 33A are likely episodic in nature. One source of
persistent low-level emissions observed with the infrared
camera is believed to be a vented produced water tank at Site
C (SI Video Si). Previous studies have shown that flashing
14510
dx.doi.org/10.1021/es503070q I Environ. Sci. Techno!. 2014, 48, 14508-14515
220175 34 of 87
-------
Environmental Science & Technology
Article
10-
3
T
X
O 5-
0-
Figure 1. CH4 emission rates (g/s) measured at repeated sites in
Pinedale, WY by year.
from a condensate tank after a separator dump can result in
episodic large emissions.6 CH4 emissions greater than 2 g/s
were observed at 13% of the 210 unique sites measured. The
variability of emission rates at sites H and 1 indicates that these
larger emissions may be episodic events that cannot be used to
infer annual emission rates without a greater understanding of
their frequency and duration (Figure l).
Site I was measured on four separate days in 2012. On each
of the days, the emissions appeared to originate from the same
tank. Infrared videos indicate that all of the emissions >3.0 g/s
occurred during the time period that a thief hatch on a
condensate tank was open (SI Video S4, Video S5, and Video
S6). On the last day the site was measured, the thief hatch was
closed and the measured emissions seemed to originate from a
pressure relief device and were <3.0 g/s (SI Video S7).
Another potential cause of variation in emissions levels is the
variability in plume capture. Depending on meteorological
conditions, the plume measured can include all of the sources
on the pad or only some of the sources (Figure 2).
Measurements were made at Site H on 3 days in 2012 and 1
day in 2013 (four and two independent emission measure-
mentSj respectively). The higher emissions observed were only
present on one of the days in 2012 and originated from the
tank on the north side of the pad (SI Video S2), whereas the
smaller emissions seemed to originate from the southern edge
of the pad (SI Video S3).
Comparisons of CH4 Emissions by Basin and with
Direct Measurement Studies. A total of 318 OTM 33A
measurements that met the data quality criteria were collected.
Of these measurements, 31 were excluded from the analysis
because the measured emissions either did not originate from
routine pad operations (e.g., evidence of active pad
maintenance, pipeline leaks, gas processing plants, etc.) or no
current production data were available, resulting in a total of
210 unique sites. The sites were classified into gas or oil pads
based on the TX Railroad Commission definition of a gas
well20 (>100 Mscf of gas per barrel of hydrocarbon liquids).
Gas pads constituted 93%, 2%, 75%, and 84% of the sites
measured in the Barnett, DJ, Eagle Ford, and Pinedale basins,
respectively. Methane emissions were averaged by site and
month, resulting in a total of 228 combinations of emission and
production values. Due to the small sample size in the Eagle
Ford (n = 4), these measurements were excluded from the
basin comparison (Figure 3). CH4 emissions were log-normally
distributed with geometric means and 95% confidence intervals
(CIs) of 0.33 (0.23, 0.48), 0.14 (0.11, 0.19), and 0.59 (0.47,
0.74) g/s in the Barnett, Denver-Julesburg, and Pinedale basins,
respectively. Emissions by basin were compared using a
Kruskal—Wallis one-way analysis of variance test and pairwise
Wilcoxson rank-sum tests and were found to be significantly
different (p < 0.05). The differences in emissions between
basins are likely a result of a combination of factors, including
but not limited to variations in gas and oil production,
emissions control devices, and natural gas and oil composition.
The OTM 33A measurements were compared with the
results of the direct measurement studies of routine pad
operations conducted by ERG1;i and Allen et at16 (Figure 3).
The studies encompass a range of pads that vary with respect to
oil and gas composition, production levels, amount and type of
production equipment, age, and emission control measures,
resulting in a broad distribution of emissions. The mean of the
CH4 emissions measured using OTM 33A in the Barnett Shale,
0.33 (0.23, 0.48) g/s, is more than twice the mean of the
emissions measured by ERG14 0.14 (0.11. 0.18) g/s. Never-
theless, the interquartile range of the OTM 33A measurements
in the Barnett falls within the interquartile range of the ERG
emissions estimates despite the differences in the measurement
methods and the bias toward higher-emitting sites in the OTM
33A measurements.
Both onsite and remote measurement techniques can
provide important information on emissions. Whereas direct
measurements can accurately quantify component-level emis-
sions, they are less amenable to locating and assessing
malfunction-related or large short-term emissions such as
condensate tank flashing. The measurements by Allen et al.1
were limited primarily to equipment leaks, pneumatic
controllers, and chemical injection pumps. Condensate tank
emissions were measured at some sites but rarely could all of
the emission points be accessed. In the ERG study,15 due to
A 2011 + 2012 * 2013
-f
+
-4-
1 1 1 1 1 1 1 T T"
A8CDEFGH I
Site
cm (g/s)
01-10 3.1-5.0
11-20 ^ 51
4< 10.1 -13 4
2 1 -3.0
Arrows indicate mean
wind direction during
measurement.
0 100
i Meters
Figure 2. Map of repeated measurements at sites H and I. The directions of the colored arrows indicate mean wind directions and the locations
indicate the locations of the mobile platform during the measurement
ta I.G.Sobe,
GeoEye' Earilistar •
Geograpnicsi S^ESA'fbus
tD§^USDA U.SG|!^X
*
14511
dx.doi:org/10.1021/es503070q I Environ. Sci. Techno!. 2014, 48, 14508-14515
23 of 75 35 of 87
-------
Environmental Science & Technology
Article
100.0000'
1.0000
0.0100
0.0001 -
n = 43
Allen etal. (2013)
0.1'
n = 58
n = 21
015<|
n = 17
0.03f
1 1
Appalachian Gulf Coast
ERG (2011)
n = 295
0.14*
This study (OTM 33A)
n = 43
: 74
0.33*
M incontinent Rocky Mountain Barnett
Basin
Barnett
l
DJ
n = 107
¦
0.59*
0.14*
Rinedale
Figure 3. Comparison of measured CH4 emissions per pad (g/s) from Allen et al.,16 ERG,15 and OTM 33A by basin. Boxes represent the 1st and
3rd quartiles of the data, while whiskers extend to the largest measurement that is within 1.5 times the interquartile range (IQR). Means and 95% CIs
are shown in black and were calculated using a nonparametric bootstrap.
lack of condensate production, flash emissions were not
represented. Although both studies measured fugitive compo-
nent leaks, neither identified or measured potentially larger
maintenance-related emissions (e.g., open thief hatch or failed
pressure relief value). In contrast, OTM 33A measurements
generally represent an integrated plume including all potential
sources on a pad. Supporting infrared camera footage from the
OTM 33A studies indicated that emissions often originate from
condensate storage tanks which have previously been shown to
comprise a significant source ,s (SI Supplemental B). OTM
33A is also more likely to capture malfunction-related CH4
releases than direct measurement methods because of its
mobile and off-site measurement capabilities.
However, the remote nature of the OTM 33A method and
its application in these studies to only sites with downwind
average in-plume concentrations greater than 0.1 ppm result in
an effective lower sampling limit of approximately 0.010 g/s,
compared with <0.001 g/s limits for the on-site measurement
techniques (Figure 4a). As a result, the OTM 33A measure-
ments only represent the upper end of the distribution in this
comparison (Figure 4b).
Comparison of Measurements with Production
Values. CH4 emissions from the direct measurement studies
and OTM 33A were compared to monthly gas production
using a linear regression on the log transformed data (Figure
5). Sites with gas production <1 Mscf/day or CH4 emissions
<0.0005 g/s were excluded from the analysis (five sites in the
ERG study15). Gas production values explained more of the
variation in the OTM 33A measurements than the measure-
ments from the on-site studies, although variation in gas
production still accounted for only 8.3% of the total variation in
emissions (R2 = 0.083) (Figure 5).
The OTM 33A CH4 emission estimates were also compared
with hydrocarbon liquids and water production and the
(arithmetic) mean age of active permitted wells on the site
using Pearson correlation coefficients (Table l) and a
multivariate linear regression.
Approximately 23% and 15% of the pads measured using
OTM 33A reported no hydrocarbon liquids or water
production, respectively. To use these pads in the log-
transformed model, pads with no reported oil or water
production were assigned 0.01 bbl/day. Several values were
Allen et al. (2013) — ERG (2011) — This study (OTM 33A)
0.6
S" 0.4
u>
0.0
>< 1.00
'55
§ 0.75
0
> 0.50
aj
1 025
O 0,00
Figure 4. Density (a) and cumulative density (b) of measurements of
CH4 emission rates (g/s) from this study (OTM 33A), Allen et al.,16
and ERG.15 Note the logarithmic ,v-axis.
tested and the choice of this value did not significantly affect
the results. When considering the correlation between
production and emissions individually, CH4 emissions were
most strongly correlated with gas production (R = 0.29). CH4
emissions were also positively correlated with water production,
negatively correlated with mean age, and not correlated with
hydrocarbon liquids production (Table I).
A multivariate linear regression was conducted to determine
the effect of gas and hydrocarbon liquids production and age of
the well on CH4 emissions simultaneously. Water production
was not included in the model because it was so highly
correlated with gas production (R > 0.7) that the effects could
not be separated. The following model was used:
l°g(CH4) = ^log(gas) + //.log(oil) + fygt (x)
0.001 0.010 0.100 1.000 10.000
T
~r
T
T
T
0.001 0.010 0.100 1.000 10.000
CH4 Emitted Per Site (g/s)
14512
dx.doi.org/10.1021 /es503070q I Environ. Sci. Technol. 2014, 48, 14508-14515
24 of 75 36 Of 87
-------
Environmental Science & Technology
Article
Allen et al. (2013)
ERG (2011)
This study (OTM 33A)
R2 = 0.055
R2 = 0.031
R2 = 0.083
* .•»<££*
*>> /•*.* • •
t •
J "
• * • • ••
•
•
10 1000 100000 10 1000 100000 10 1000 100000
Gas Production (Mscf/day)
Figure 5. CH4 emissions (Mscf/day) versus reported monthly gas production (Mscf/day). Blue lines represent the linear regression lines.
Table 1. Pearson Correlation Coefficients (R) of Emissions
and Production
ch4
emissions
(Mscf/
day)
gas
production
(Mscf/ day)
hydrocarbon
liquids
production
(bbl/day)
water
production
(bbl/day)
CH4 emissions
(Mscf/day)
1.00
gas production
(Mscf/day)
0.29
1.00
hydrocarbon
liquids
production
(bbl/day)
-0.01
0.44
1.00
water production
(bbl/day)
0.22
0.77
0.40
1.00
mean age (years)
-0.20
-0.59
-0.34
-0.57
where CH4 represents measured emissions in g/s, gas is total
reported production in Mscf/day, oil is total reported
hydrocarbon liquids production in bbl/day, and age is the
mean age of the wells in years. Age was not significantly
correlated with CH4 emissions, while gas production was
significantly positively correlated, and oil production was
significantly negatively correlated (SI Table S2). The negative
correlation with oil production is consistent across the basins
(SI Figure Si). This negative correlation with oil production is
likely due to the lower fraction of CH4 in wet gas compared to
dry gas. Furthermore, emissions from condensate tanks, which
are more prevalent in wet gas areas, typically contain a lower
fraction of CH4 and higher fraction of heavier hydrocarbons
such as VOCs when compared with produced gas.6 The
inclusion of hydrocarbon liquids and age in the model did not
explain much more of the variation in emissions resulting in an
adjusted R2 of only 0.096, in contrast to an R2 of 0.083 when
only gas production was included (Figure 5).
Other important sources of variation not accounted for in
this analysis include emissions controls and equipment present
on the pads. Further uncertainty is introduced by the
production data: daily or hourly production levels may not
be consistent with monthly production.
Although the OTM 33A CH4 emissions data include episodic
features (e.g., flash emissions), it is instructive to compare
emission rates as a ^percent of production with the measure-
ments by Allen et al. and ERG. 5 The differences between the
CH4 emissions estimates of the three studies are amplified
when emissions are considered as a percentage of total
production rather than in mass emission rate (SI Figure S2).
For the sites measured using OTM 33A, approximately 0.72
14513
(0.44, 1.17)%, 1.36 (0.97, 1.95) %, and 0.58 (0.39, 0.86) % of
production was emitted on average (with 95% Cl) in the
Barnett, DJ, and Pinedale basins, respectively, compared with
0.11 (0.09, 0.16)% of production measured by ERG15 in the
Barnett shale and 0.01 (0.01, 0.01) % and 0.09 (0.04, 0.20)%
measured by Allen et al.16 in the Appalachian and Rocky
Mountain basins, respectively (SI Figure S2). As evidenced in
the statistical analysis, differences in production rate explain
only a fraction of the variation in emissions. The percentages
from this study only represent emissions from routine well pad
operations and thus cannot be directly compared to other
estimates of total CH4 emitted as a percent of production such
as those by Brandt et al.' that include emissions from many
other processes.
Mean gas production at the OTM 33A sites was significantly
lower than mean gas production at the sites measured in the
direct measurement studies (SI Figure S4). Gas production at
the OTM 33A sites ranged from 3.7 (Mscf/day) to 9021
(Mscf/day) with 37% of the sites producing <100 Mscf/day. In
contrast, Allen et al.16 reported a gas production range of 20 to
47 690 (Mscf/day) with only 10% of the sites producing <100
Mscf/day and with approximately 20% of the measured sites
producing >10,000 Mscf/day. The gas production values of the
ERG15 sites ranged from 0.06 to 9085 Mscf/day in the Barnett
with 10% of the sites producing <100 Mscf/day (SI Figure S4).
The OTM 33A results indicate that sites with very low gas and
oil production can emit a much greater fraction of the gas
produced than sites with higher production levels. Maintenance
issues (e.g., fugitive leaks, open or leaking thief hatches, failed
pressure relief devices, malfunctioning separator dump valves)
could be more prevalent at smaller older production sites than
at higher producing sites that are potentially better maintained
and may have fundamentally different engineering designs (e.g.,
use of buffer tanks to suppress flash emissions). Furthermore,
many of the fugitive processes can emit at levels that are not
linearly associated with production rates as is evidenced by the
lack of correlation between emissions and production and the
finding by Allen et al.16 that equipment leaks are under-
estimated by the 2011 EPA national inventory.
In summary, the OTM 33A mobile inspection method can
be used to complement direct measurement techniques and
expand our knowledge of the upper range of the distribution of
CH4 emissions. OTM 33A was successfully applied to quantify
CH4 emissions at 210 oil and gas well pads with an accuracy of
±60% determined by controlled release tests. Well pad
emissions were log-normally distributed and differed signifi-
cantly by basin with geometric means ranging from 0.14 g/s in
dx.doi.org/10.1021/es503070q I Environ. Sci. Technol. 2014, 48, 14508-14515
25 of 75 37 of 87
-------
Environmental Science & Technology
the Denver-Julesburg to 0.59 g/s in the Pinedale basin. Repeat
measurements at 9 sites indicated consistent low emission rates
at seven sites and highly variable emissions at two sites, one a
documented malfunction. The production rates accounted for
approximately 10% of the variation in sampled emission rates in
a multivariate linear regression on age, hydrocarbon liquid and
gas production. Normalizing emissions by gas production
amplified the differences between the remote and onsite
measurements. Compared to the direct measurements in the
Barnett, the mean of the remote measurements was
approximately twice as large in terms of mass emissions rate,
but approximately seven times as large when considered as a
percentage of production, indicating that sites with lower
production levels can emit a much greater percentage of
production. Infrared camera videos indicate that emission rates
may be strongly affected by stochastic variables. In particular,
equipment malfunctions or operator error may cause emission
rates to increase substantially compared to routine operating
conditions. Accurately estimating site emissions on a regional
scale likely will require determining the average magnitude and
frequency of these stochastic events.
m ASSOCIATED CONTENT
Q Supporting Information
Supplemental figures, tables, and IR videos are supplied This
material is available free of charge via the Internet at http://
pubs.acs.org.
OR INFORMATION
Corresponding Author
*Phone: 1-919-541-7969; fax: 1-919-541-0359; e-mail: thoma.
eben(2> epa.gov.
Notes
The authors declare no competing financial interest.
#On Oak Ridge Institute of Science and Education Fellowship.
B ACKNOWLEDGMENTS
We thank Shahrooz Amin and Mark Modrak with ARCADIS,
Inc., for field and data analysis support for this project. We
would like to thank EPA colleagues Bill Mitchell, Adam Eisele,
Mike Miller, Jason DeWees, Robin Segall, and Ken Garing and
his team for ongoing support in development of OTM 33.
Special thanks to Eric Crosson, Chris Rella, and Tracy Tsai
with Picarro for ongoing collaboration on mobile measure-
ments. Primary funding for this effort was provided by U.S.
EPA ORD's Air, Climate, and Energy (ACE) and Regional
Applied Research Effort (RARE) programs. Funding for the
Environmental Defense Fund's methane research series,
including this work, is provided for by Fiona and Stan
Druckenmiller, Heising-Simons Foundation, Bill and Susan
Oberndorf, Betsy and Sam Reeves, Robertson Foundation,
Alfred P. Sloan Foundation, TomKat Charitable Trust, and the
Walton Family Foundation. The views expressed in this article
are those of the authors and do not necessarily represent the
views or policies of the U.S. Environmental Protection Agency.
M REFERENCES
(1) Moore, C. W.; Zielinska, B.; Petron, G.; Jackson, R. B. Air
impacts of increased natural gas acquisition, processing, and use: A
critical review. Environ. Sci. Technol. 2014; DOI: 10.1021/es4053472.
(2) Roy, A. A.; Adams, P. J.; Robinson, A. L. Air pollutant emissions
from the development, production, and processing of Marcellus Shale
natural gas. J. Air Waste Manage. Assoc. 2014, 64 (l), 19—37
DOI: 10.1080/10962247.2013.826151.
(3) CenSARA, 2011 Oil and Gas Emission Inventory Enhancement
Project for CenSARA. States prepared for: Central States Air Resources
Agencies 2011. http://www.censara.org/filedepot_download/S6064/
14 (accessed September 17, 2014).
(4) Field, R A.; Soltis, J. J.; Murphy, S. Air quality concerns of
unconventional oil and natural gas production. Environ. Sci.: Processes
Impacts 2014, 16, 954-969 DOI: 10.1039/C4EM00081A
(5) Brandt, A.; Heath, G.; Kort, E.; O'Sullivan, F.; Petron, G.;
Jordaan, S.; Tans, P.; Wilcox, J.; Gopstein, A; Arent, D. Methane leaks
from North American natural gas systems. Science 2014, 343 (6172),
733-735 DOI: 10.1126/science.l247045.
(6) Hendler, A.; Nunn, J.; Lundeen, J.; McKaskle, R VOC Emissions
from Oil and Condensate Storage Tanks-, Houston Advanced Research
Center, 2006; http://files.harc.edu/Projects/AirQuality/Projects/
H051C/H051 CFinalReport.pdf.
(7) Gidney, B.; Pena, S. Upstream Oil and Gas Storage Tank Project
Flash Emissions Models Evaluation Final Report-, Texas Commission on
Environmental Quality: Austin, TX, 2009; http: //www.bdlaw.com/
assets/htmldocuments/
TCEQ%20Final%20Report%200il%20Gas%20Storage%20Tank%20
Project.pdf.
(8) U.S. EPA Greenhouse Gas Reporting Rule Subpart W, Petroleum
and Natural Gas Systems; Public Law 78 FR 71904, 2013; http://www.
gpo.gov/fdsys/pkg/FR-2013-ll-29/pdf/2013-27996.pdf.
(9) U.S. EPA Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2011; EPA 430-R-13-001, 2013; http://www.epa.gov/
climatechange/Downloads/ghgemissions/US-GHG-Inventory-2013-
Main-Text.pdf.
(10) U.S. EPA EPA Needs to Improve Air Emissions Data for the Oil
and Natural Gas Production Sector; Report No. 13-P-0161; U.S.
Environmental Protection Agency Office of Inspector General:
Washington DC, 2013; http://www.epa.gov/oig/reports/2013/
20130220-13-P-0161.pdf.
(11) Snyder, E. G.; Watkins, T. H.; Solomon, P. A; Thoma, E. D.;
Williams, R W.; Hagler, G. S.; Shelow, D.; Hindin, D. A; Kilaru, V. J.;
Preuss, P. W. The changing paradigm of air pollution monitoring.
Environ. Sci. Technol. 2013, 47 (20), 11369-11377 DOI: 10.1021/
es4022602.
(12) Caulton, D. R; Shepson, P. B.; Santoro, R L.; Sparks, J. P.;
Howarth, R W.; Ingraffea, A R; Cambaliza, M. O.; Sweeney, C.;
Karion, A.; Davis, K. J. Toward a better understanding and
quantification of methane emissions from shale gas development.
Proc. Natl. Acad. Sci. U. S. A. 2014, 111 (17), 6237-6242
DOI: 10.1073/pnas.l316546111.
(13) Karion, A; Sweeney, C.; Petron, G.; Frost, G.; Michael
Hardesty, R; Kofler, J.; Miller, B. R; Newberger, T.; Wolter, S.; Banta,
R Methane emissions estimate from airborne measurements over a
western United States natural gas field. Geophys. Res. Lett 2013, 40
(16), 4393-4397 DOI: 10.1002/grl.50811.
(14) Thoma, E.; Squier, B.; Olson, D.; Eisele, A; DeWees, J.; Segall,
R; Amin, M.; Modrak, M. Assessment of methane and voc emissions
from select upstream oil and gas production operations using remote
measurements, interim report on recent survey studies. In Proceedings
of 1 OSth Annual Conference of the Air & Waste Management Association,
Control No. 2012-A-21-AWMA, 2012, 298-312.
(15) ERG. City of Fort Worth Natural Gas Ar Quality Study Final
Report; Fort Worth, TX, 2011. http://fortworthtexas.gov/gaswells/
?id=87074.
(16) Allen, D. T.; Torres, V. M.; Thomas, J.; Sullivan, D. W.;
Harrison, M.; Hendler, A; Herndon, S. C.; Kolb, C. E.; Fraser, M. P.;
Hill, A D. Measurements of methane emissions at natural gas
production sites in the United States. Proc. Natl Acad. Sci. U. S. A.
2013, 110 (44), 17768-17773 DOI: 10.1073/pnas.l304880110.
(17) U.S. EPA Other Test Method (OTM) 33 and 33A Geospatial
Measurement of Ar Pollution-Remote Emissions Quantification-
Direct Assessment (GMAP-REQ;DA). 2014. (http://www.epa.gov/
ttn/emc/prelim.html).
14514
dx.doi.org/10.1021/es503070q I Environ. Sci. Technol. 2014, 48, 14508-14515
260175 38 of 87
-------
Environmental Science & Technology
(18) Golder, D. Relations among stability parameters in the surface
layer. Bound.-Lay. Meteorol 1972, 3 (l), 47-58 DOI: 10.1007/
BF00769106.
(19) Google Inc. Google Earth Pro (Version 7.1.2.2041), 2014.
(20) ESRI. ArcGIS Desktop: Release 10; Environmental Systems
Research Institute: Redlands, CA, 2011.
(21) Public GIS Map Viewer for Oil, Gas, and Pipeline Data, http://
wwwgisp.rrc.state.tx.us/GISViewer2/ (April 22, 2014),.
(22) COGCC GIS Online, http://dnrwebmapgdev.state .co.us/
mg2012app/ (accessed April 22, 2014).
(23) R Core Team R: A Language and Environment for Statistical
Computing; R Foundation for Statistical Computing: Vienna, Austria,
2013.
(24) Wickham, H. ggplotl: Elegant Graphics for Data Analysis;
Springer: New York, 2009.
(25) Harrell Jr, F. E.j Dupont, C. Hmisc R package version 3.14-1.
2007.
(26) TX RRC. Texas Administrative Code. Title 16 Part 1 Chapter 3
Rule §3.79, Railroad Commission of Texas, Oil and Gas Division:
Austin, TX, http://info.sos.state.tx.us/pls/pub/readtacSext.
TacPage?sl=R&app=9&p_dir=&p_rloc=&p_tloc=&p_ploc=&pg=
l&p_tac=&ti=l6&pt=l&ch=3&rl=79.
14515
dx.doi.org/10.1021/es503070q I Environ. Sci. Techno!. 2014, 48, 14508-14515
270175 39 of 87
-------
CSIRO ENERGY TECHNOLOGY
www.csiro.au
Field Measurements of Fugitive
Emissions from Equipment and
Well Casings in Australian Coal
Seam Gas Production Facilities
Report to the Department of the Environment
Stuart Day, Mark Dell'Amico, Robyn Fry and Hoda Javanmard Tousi
June 2014
28 of 75 40 of 87
-------
l.-llll i. ion
Day, S., Dell'Amico, Fry, R., Javanmard Tousi, H., (2014). Field Measurements of Fugitive Emissions from
Equipment and Well Casings in Australian Coal Seam Gas Production Facilities. CSIRO, Australia.
© 2014 CSIRO To the extent permitted by law, all rights are reserved and no part of this publication
covered by copyright may be reproduced or copied in any form or by any means except with the written
permission of CSIRO.
CSIRO advises that the information contained in this publication comprises general statements based on
scientific research. The reader is advised and needs to be aware that such information may be incomplete
or unable to be used in any specific situation. No reliance or actions must therefore be made on that
information without seeking prior expert professional, scientific and technical advice. To the extent
permitted by law, CSIRO (including its employees and consultants) excludes all liability to any person for
any consequences, including but not limited to all losses, damages, costs, expenses and any other
compensation, arising directly or indirectly from using this publication (in part or in whole) and any
information or material contained in it.
29 of 75 41 of 87
-------
Contents
Executive Summary iii
1 Introduction 1
? National Greenhouse Gas Reporting Practices 4
3 tixperirrtental Methods /
3,1 Selection of Wells /
3,? Methane Analysis System 8
3.3 Plume I reverses 9
3.4 teak and Vent lestini- 10
3,;j Surface Emissions 1?
4 Results 14
4,1 Controlled Release 14
4,? Well Measurements 14
4,3 Casing leaks ?8
Discussion 30
5, 1 Emission I actors 34
6 Conclusions 36
References 3/
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Productioij^acj-li^es ^
-------
3
We gratefully acknowledge the CSIRO Energy Flagship and Department of the Environment for providing
financial support for this project. The project would not have been possible without the cooperation of the
CSG industry and the numerous company staff who provided assistance and advice. In particular we would
like to thank Sally Oelerich, Tom Lawler and Aaron Clifton of AGL, Jenny Bright and Ben McMahon of Arrow
Energy, Graeme Starke of QGC, Liz Beavis, Michael Boyle, Mark Rider and Dan O'Sullivan of Origin Energy,
and Anshul Jain, Michael Roberts, and Rob Dunsmuir of Santos.
ii | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities 31 of 75 43 f 87
-------
Execic
The Australian coal seam gas (CSG) industry has developed rapidly over the last decade and as several
liquefied natural gas (LNG) plants currently under construction in Queensland are completed, gas
production will increase significantly over the next few years. Fugitive emissions of methane from gas
production and processing have the potential to diminish the greenhouse benefits of CSG utilisation
compared to other fossil fuels but at present the extent of fugitive emissions from the CSG industry and
unconventional gas production more generally is not well understood. Recent reports from the United
States have suggested that fugitive emissions from unconventional gas production, especially shale and
tight gas, are much higher than previously estimated. However, because of significant differences in
production methods and other factors, it is unlikely that emission estimates from U.S. shale and tight gas
production are indicative of emissions from Australian CSG operations. To provide quantitative information
on emissions from CSG operations, CSIRO and the federal Department of the Environment initiated a
project to measure emissions from a range of production wells in Queensland and NSW.
Methane emissions were measured at 43 CSG wells - six in NSW and 37 in Queensland. Measurements
were made by downwind traverses of well pads using a vehicle fitted with a methane analyser to determine
total emissions from each pad. In addition, a series of measurements were made on each pad to locate
sources and quantify emission rates.
Of the 43 wells examined, only three showed no emissions. These were two plugged and abandoned wells
and one suspended well that had been disconnected from the gas gathering system. The remainder had
some level of emission but generally the emission rates were very low, especially when compared to the
volume of gas produced from the wells. The principal methane emission sources were found to be:
• venting and operation of gas-powered pneumatic devices,
• equipment leaks and
• exhaust from gas-fuelled engines used to power water pumps.
The median methane total emission rate (from all sources) for the 43 wells was approximately 0.6 g min"1,
and the mean about 3.2 g min"1. Thirty seven wells had total emissions less than 3 g CH4 min 1 and 19 less
than 0.5 g min \ There were however, a number of instances where much higher emission rates were
found. The highest emission rate of 44 g min 1 was from a vent on a water line at one well although this
represented a very minor proportion of gas production. These emission rates are very much lower than
those that have been reported for U.S. unconventional gas production.
Gas operated pneumatic devices were installed at some well sites and were occasionally found to be
emitting small amounts of methane. These emissions were small (mean emissions rate of 0.12 g min"1) and
may reduce even further as gas operated pneumatic systems are replaced by air or electrically operated
devices.
Equipment leaks were found on 35 wells with emission rates ranging from less than 1 mg min 1 up to about
28 g min"1. The median and mean emission rates from these wells were 0.02 g min 1 and 1.6 g min"1, which
correspond to emission factors of about 0.1 kg C02-e t"1 and 2.4 kg C02-e t"1, respectively. This range is
consistent with the current emission factor of 1.2 kg C02-e t"1 commonly used throughout the CSG industry
to account for equipment leaks for the purposes of reporting emissions under the National Greenhouse and
Energy Reporting legislation.
Several of the larger equipment leaks were found at seals on water pump shafts on some wells. However,
once identified, well maintenance staff were able to repair some of these leaks on site, which effectively
eliminated methane emissions.
Fifteen of the well sites had gas fuelled engines operating at the time measurements were made. The
exhaust from most of these engines was found to be contributing to the well site emissions, in several cases
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production^dlj.ti^s | ^
-------
comprising the bulk of methane emissions. From a greenhouse gas accounting perspective, methane in
exhaust is not considered to be a fugitive emission but is counted as a combustion emission.
During the field measurements, no evidence of leakage of methane around the outside of well casings was
found at any of the wells included in this sample.
Although the well pad emissions were low, a separate, larger source of methane was found on a gas relief
vent on a water gathering installation close to one of the wells examined during this study. An indicative
estimate of the emission rate from this vent suggested that the source was at least three times higher than
the largest well pad emission rate. Similar installations are widespread through the Queensland gas regions
and hence further examination is needed to determine the extent of this potential emission source.
The results obtained in this study represent the first quantitative measurements of fugitive emissions from
the Australian CSG industry; however, there are a number of areas that require further investigation.
Firstly, the number of wells examined was only a very small proportion of the total number of wells in
operation. Moreover, many more wells are likely to be drilled over the next few years. Consequently the
small sample examined during this study may not be truly representative of the total well population. It is
also apparent that emissions may vary over time, for instance due to repair and maintenance activities. To
fully characterise emissions, a larger sample size would be required and measurements would need to be
made over an extended period to determine temporal variation.
In addition to wells, there are many other potential emission points throughout the gas production and
distribution chain that were not examined in this study. These include well completion activities, gas
compression plants, water treatment facilities, pipelines and downstream operations including LNG
facilities. Emissions from some of these sources are often estimated for reporting purposes using
methodology based on emission factors largely derived from the U.S. gas industry. However, reliable
measurements on Australian facilities are yet to be made and the uncertainty associated with some of
these estimates remains high.
iv | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities nn s-7k a i-
cscs or cd 45 of 87
-------
Coal seam gas (CSG) production is a major and rapidly expanding industry in Australia. During 2011-2012,
Australian CSG production was around 247 PJ, which represented about 12 % of total gas production in
Australia (BREE, 2013). Since then, production in Queensland alone has increased to more than 264 PJ in
2012-2013 (DNRM, 2014) with production likely to increase even further as several liquefied natural gas
plants under construction come on stream. Most Australian CSG is currently produced in Queensland with
only one operational project in NSW; however, there are a number of other projects planned for NSW at
various stages of approval.
One of the key drivers of increased demand for gas is that greenhouse gas emissions from gas utilisation
are usually lower than other fossil fuels (Day et al., 2012). However, because of the much higher global
warming potential of methane compared to C02, even relatively small proportions of fugitive methane
released during the production, processing and distribution of natural gas can reduce this advantage
relative to other fuels (e.g. Wigley, 2010; Alvarez et al., 2012).
In the natural gas industry, fugitive emissions are considered to include all greenhouse gas emissions from
exploration, production, processing, transport and distribution of natural gas, except those from fuel
combustion (IPCC, 2006). However certain combustion processes like flaring and waste gas incineration are
also counted as fugitive emissions.
At present the level of fugitive emissions from the Australian CSG industry is not well defined, although
individual companies estimate and report their annual emissions under the requirements of the National
Greenhouse and Energy Reporting Act 2007 (NGER, see Section 2). These data are used for compiling the
Australian National Greenhouse Gas Inventory which currently estimates fugitive emissions from the
Australian oil and gas industry to be around 12 Mt C02-e per annum (DIICCSRTE, 2013a). About 60 % of
these emissions are attributed to venting and flaring, which are in principle amenable to direct
measurement; hence the uncertainty on this component may be relatively low. However, other sources
such as equipment leaks are frequently difficult to measure so are usually estimated by methodology
characterised by very high uncertainty. Despite significant differences in production methods, the national
inventory does not at present distinguish between conventional gas production and unconventional
sources like shale gas and CSG.
In 2012, the CSIRO reviewed the available scientific and technical literature to assess the current state of
knowledge relating to fugitive emissions from unconventional gas production, especially for CSG production
in Australia (Day et al., 2012). Most of the information in the public domain at the time was concerned with
shale and tight gas production in the United States with virtually none specific to CSG. Up until then, only
one study based on actual measurements had been published (Petron et al., 2012). This group measured
methane emissions in the Denver-Julesburg Basin in Colorado and depending on the method used,
estimated that the emission rate from the gas field was equivalent to 1.7 to 7.7 % of the gas produced in
the region.
Since 2012, several other studies, also from the United States, have been published. Karion et al. (2013)
conducted an airborne survey of ambient methane in an unconventional gas field in the Uintah Basin in
Utah in the United States. The Karion et al. study yielded emission estimates of between about 6 and
almost 12 % of gas production of the region. In a detailed examination of atmospheric methane data from
airborne and fixed monitoring stations, Miller et al. (2013) determined the spatial distribution of methane
emissions throughout the United States. This study considered all sources of anthropogenic methane
emissions, including fugitive emissions from oil and gas production. For the Texas/Oklahoma region
emissions from oil and gas production were estimated to be 3.7 ± 2.0 Tg C y"1, which is 4.9 ± 2.6 times
higher than the current estimate of 0.75 Tg C y1 in the European Commission's Emissions Database for
Global Atmospheric Research (EDGAR).
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Productior^jcjJit^ |^
-------
Both the Miller et al. (2013) and Karion et al. (2013) studies used top-down methodology and did not
attempt to determine the specific sources of the methane emissions. Petron et al. (2012) also used top-
down methods which yielded the higher estimates (i.e. ~7.7 % of production) although the bottom-up
methodology used by that group gave much lower emission estimates (1.7 %). A bottom-up approach was
used by Allen et al. (2013) who examined emissions at the facility level to determine both the rate and
route of methane emission. In that study, methane emissions were measured at 190 onshore natural gas
sites within the United States, which included 489 production wells (all of which had been hydraulically
fractured), 27 well completion flowbacks, nine well unloadings, and four well workovers. One of the key
findings of this work was that the measured emissions were generally comparable to the most recent
USEPA estimates of emissions from the sources examined, although the relative proportion of emissions
from individual categories differed somewhat. For example, emissions from pneumatic devices were
significantly higher than current estimates while emissions from well completions were much lower than
estimates in the U.S. inventory. Overall, the emissions estimated from the unconventional gas industry
corresponded to about 0.42 % of production.
This bottom-up estimate contrasts with the much higher top-down estimates discussed above. The lower
emission rate estimated by Allen et al. (2013) may be explained in part by the fact that only production
facilities were considered. Emissions from downstream processing, transport and distribution were not
included so any emissions from these facilities would be expected to increase this proportion. Another
reason for the discrepancy between bottom-up and top-down estimates has been proposed by Brandt et
al. (2014) who suggested that a large proportion of emissions may be due to a small number of 'super
emitters'. If true, facility level bottom-up measurements may sometimes miss these large emission sources.
In addition to gas production facilities, other sources may be contributing to overall emissions, which are
not captured by the bottom-up methods. Tait et al. (2013), for example, proposed that drilling and
associated activity may induce fracturing of overlying strata thus providing pathways for methane to reach
the surface and escape to the atmosphere. Such landscape-scale emissions would be detected by many
top-down methods but may be difficult to measure using the bottom-up methodology applied by Allen et
al. (2013). However, the Tait et al. (2013) model was based on ambient radon measurements; methane
emission rates were not measured so this emission route remains speculative at this stage. Other possible
emission sources that could account for the apparent discrepancy between the reported top-down and
bottom-up methods are geological sources such as seeps that are often associated with oil and gas fields
(Klusman, 1993) or abandoned boreholes (Etiope et al., 2013; Day et al., 2013).
In Australia, limited investigations into fugitive methane emissions from CSG production have been
undertaken over the last couple of years. In an initial study that was widely reported, Santos and Maher
(2012) surveyed a CSG production region near Tara in Queensland using an instrumented vehicle to
measure the spatial distribution of ambient methane concentrations. They measured elevated methane
concentrations within the gas field that they suggested may be indicative of fugitive methane release from
production activities. More recently, a study of ambient methane levels in the vicinity of CSG production
facilities south of Sydney was reported (Pacific Environment Limited, 2014). This study also found elevated
methane concentrations near CSG facilities although they concluded that on average, ambient methane
concentrations within the gas field were comparable to those in a nearby urban area. However, neither
study attempted to measure emission flux and in any case, the presence of other potential methane
sources such as cattle feed lots, abandoned boreholes and landfill sites complicated the interpretation of
the results. Consequently attempts to attribute sources based on these results remain inconclusive.
Despite the level of recent activity aimed at quantifying emissions from unconventional gas production, the
situation remains unclear. The Australian studies reported to date only considered ambient methane
concentrations near gas production sites and provide no information on emission flux. While the U.S.
studies measured emission rates, widely varying estimates were reported. Moreover, they were concerned
with shale and tight gas operations, which are unlikely to be indicative of emissions from Australian CSG
production facilities. Due to the lack of quantitative emission data specific to Australian operations, the
CSIRO review recommended, among other things, that a series of measurements at CSG production
facilities was required to better understand the actual level of fugitive emissions from the Australian CSG
2 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities 35 of 75 47 f Qy
-------
industry (Day et al., 2012). A similar recommendation for emissions measurements was made by Saddler
(2012) when reviewing methodology for estimating emissions from CSG production.
As a result of these recommendations, CSIRO initiated a project with the principal aims of (1) developing
atmospheric top-down methodology for monitoring and quantifying methane fluxes from CSG production
facilities and (2) measuring methane emission fluxes from operational CSG production sites. Shortly after
this project commenced, the federal Department of the Environment (then the Department of Climate
Change and Energy Efficiency) requested that CSIRO to extend the scope of the field measurements to
include an investigation of gas leakage from well casings and equipment located on individual well pads.
In this report we present the results of field measurements made at well sites throughout NSW and
Queensland. The specific objectives of these measurements were to:
• quantify methane emissions from individual well pads,
• identify the primary routes of these emissions,
• measure leak rates from individual items of equipment located on well pads and
• determine whether or not methane was leaking from around the outside of well casings and if so,
measure the leakage rate.
While wells represent a major segment of the CSG production infrastructure, it is important to note that
there are many other components downstream of the wells which have the potential to release
greenhouse gases. These include processing and compression plants, water treatment facilities, gas
gathering networks, high pressure pipelines and several LNG production facilities currently under
construction near Gladstone. In the study reported here, we have only examined emissions from a small
sample of CSG wells; none of the other downstream infrastructure has been considered at this stage.
However, the ongoing CSIRO research into atmospheric top-down method methodology is aimed at
developing techniques for monitoring emissions across the CSG industry more broadly.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Productioa^acj|it^|s |^
-------
2 ^ ^tices
Before discussing the experimental results of the field measurements it is instructive to consider the
methodology currently used to estimate greenhouse gas emissions from CSG wells.
Australian CSG gas producers (along with conventional gas operators) are required to estimate and report
their annual greenhouse emissions in accordance with the National Greenhouse and Energy Reporting Act
2007 using methodology prescribed in the National Greenhouse and Energy Reporting (Measurement)
Determination 2008. The scope of the Act covers all sectors of the gas industry i.e. production and
processing, transmission and distribution, and includes emissions from fuel combustion (e.g. stationary
engines at well sites and compression plants) and fugitive emissions (leaks from equipment, venting and
flaring).
According to the definition used in the Determination, fugitive emissions associated with natural gas
production and processing comprise:
• Emissions from venting and flaring
o the venting of natural gas
o the venting of waste gas and vapour streams at facilities that are constituted by natural gas
production or processing
o the flaring of natural gas, waste gas and waste vapour streams at those facilities
• Emissions other than venting and flaring which include
o a gas wellhead through to the inlet of gas processing plants
o a gas wellhead through to the tie-in points on gas transmission systems, if processing of
natural gas is not required
o gas processing plants
o well servicing
o gas gathering
o gas processing and associated waste water disposal and acid gas disposal activities
The Determination specifies methodology for estimating emissions from all of these sources; the 'Methods'
are broadly classified into four generic categories of varying complexity, which are briefly described below.
• Method 1 is the simplest approach and relies on activity data and an emission factor for the
process. The emission factors used in Method 1 are generic and are usually specified in the
NGER Determination.
• Method 2 is more specific and uses emission factors based on more detailed data.
• Method 3 is very similar to Method 2 except that the methods are based on internationally
accepted standards.
• Method 4 is the direct measurement of emissions.
Some emissions can be directly measured (i.e. Method 4) but often emissions cannot be readily measured
so instead, simpler methodology based on the concept of emission factors is used.
Emission factors are average emission rates of a particular gas (i.e. methane but also C02 and N20 if
applicable) from a given source. Emissions, E, are calculated by multiplying the emission factor, EF, by the
activity of the process producing the emissions, A (Equation 2.1).
E — EF X A Equation 2.1
Examples of activity are the amount of fuel consumed or the amount of gas produced.
This methodology can yield accurate emission estimates for processes such as fuel combustion where both
the emission factor (which is based on the chemical composition of the fuel) and the activity data (i.e.
consumption rate of fuel, which is often known to a high level of accuracy) can be well defined. However,
4 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities 37 0[75 f Q7
-------
for some fugitive emissions sources such as equipment leaks, emission factors may be subject to very high
uncertainty. For instance, the American Petroleum Institute's Compendium of Greenhouse Gas Emissions
Methodologies for the Oil and Natural Gas Industry which provides emission factors for calculating
emissions from gas production and processing operations, estimates that uncertainties on some emission
factors may be as much as 1000 % (API, 2009). One of the reasons for this high level of uncertainty is that
emission factors are often based on very limited experimental data.
CSG well pads may release greenhouse gases from a range of sources, all of which are estimated for annual
reporting purposes. These sources include fuel combustion in well site engines used to drive water pumps,
and fugitive emissions from vents, gas operated pneumatic devices and leaks in equipment. Occasionally,
during maintenance operations for example, gas may be flared and this too counts as a fugitive emission
that is accounted for. Combustion emissions from engines or flaring are predominantly C02 although small
amounts of methane (unburnt fuel) and N20 (produced in the combustion process) may also be emitted.
Most of the other non-combustion emissions are methane.
Some emissions from vents can be measured according to Method 4 but because of its simplicity, many
CSG operators use the Method 1 approach for estimating most of the other greenhouse gas emissions from
well pads. The methods are summarised in Table 2.1.
Table 2.1. Summary of NGER estimation methods for various well pad sources
Classification
Method
Fuel Combustion
Fugitive Emissions
Exhaust emissions from
well site engines
Flare
Emission factor to account
for C02, CH4 and N20
emissions:
51.2 kg C02-e GJ"1 (C02)
0.1 kg C02-e GJ"1 (CH4)
0.03 kg C02-e GJ1 (N20)
Emissions factor to account
for C02, CH4 and N20
emissions:
2.7 t C02-e t1 (C02)
0.11 C02-e t"1 (CH4)
0.03 t C02-e t1 (N20)
Fugitive Emissions
Fugitive Emissions
Equipment leaks
Gas driven pneumatic
equipment
Emission factor of 1.2 kg
C02-e t"1 gas produced
Emission factors specified
in the API Compendium
(API, 2009)
Fugitive Emissions
Cold process vents
In some cases these can be
measured directly (i.e.
Method 4). Otherwise
estimated using emission
factors in API Compendium.
Although most of the methods shown in Table 2.1 are based on the use of emission factors, the level of
uncertainty associated with the estimates is quite variable. In the case of emission from engines, the
uncertainty is likely to be relatively low provided the amount of fuel consumed is known accurately (which
is usually the case). Similarly emissions from flaring can be estimated with reasonable accuracy if the gas
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production^gdjit^gs ^
-------
flow to the flare is measured. Emissions from vents are often measured using process instrumentation so
these too should be known with a high degree of certainty. Emissions from equipment leaks, pneumatic
equipment and vents estimated by emission factors, on the other hand, have higher levels of uncertainty.
However, the overall uncertainty of emission inventories is also influenced by the relative contribution of
various sources. Hence if a source with high uncertainty comprises only a small proportion of total
emissions from a particular sector, the overall level of uncertainty is not greatly influenced by the minor
component.
6 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities e
39 °f 75 51 of 87
-------
3 Experimental Methods
3.1 Selection of Wells
Five CSG companies provided access to wells in various gas fields throughout NSW and Queensland, which
are summarised in Table 3.1. Each company usually provided CSIRO with a list of their wells from which
CSIRO staff selected a subset of wells for examination. Because individual companies agreed to participate
in the project at different times during the course of the project it was not possible to make a properly
randomised selection of wells at the start of the project. Instead, wells were selected on an ad hoc basis in
the order that companies agreed to participate. In addition, access to sites due to weather and agreements
with landholders determined the selection of wells to some extent.
Factors considered when selecting wells included:
• The production region
• The age of the well, i.e. old to new
• The gas production rate, i.e. from low to high rates
• Whether or not the well had been hydraulically fractured
• The type of surface equipment installed at the well, i.e. pumped or free flowing.
Table 3.1. Participating CSG producers and the gas fields where emission measurements were made.
Company Name Project Name Basin Locality
AGL Energy Limited Camden Sydney MacArthur region, NSW
Arrow Energy Limited
Daandine
Surat
Dalby area, Qld
Kogan North
Surat
Dalby area, Qld
Tipton
Surat
Dalby area, Qld
Origin Energy Limited
Talinga
Surat
Chinchilla area, Qld
QGC Pty Limited
Bellevue
Surat
Chinchilla area, Qld
Berwyndale
Surat
Chinchilla area, Qld
Berwyndale South
Surat
Chinchilla area, Qld
Codie
Surat
Chinchilla area, Qld
Kenya
Surat
Chinchilla area, Qld
Lauren
Surat
Chinchilla area, Qld
Santos Limited
Fairview
Bowen
Injune area, Qld
Scotia
Bowen
Wandoan area, Qld
For the purpose of this report, we consider the well pad to be the (usually) fenced area around a well head
that contains the surface equipment associated with gas production. This includes the well head,
dewatering pump (if fitted), separator, pipework and associated valves and fittings. Also included are vents,
(including those installed on water gathering system components on the well pad) and engines used to
power dewatering pumps.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production^jcjJit^gs ^ q-j
-------
The 43 wells selected represent less than 1 % of the 5,000 CSG wells across Australia and therefore may not
be representative of the total well population. Nevertheless, it provides a reasonable cross section of the
industry covering a range of different producers and geographic locations within the main gas production
regions. For comparison, a recent study of well emissions in the U.S. where emissions measurements were
made at 489 wells represented only about 0.01 % of U.S. unconventional gas wells (Allen et al., 2013).
3.2 Methane Analysis System
Methane measurements were made using a Picarro Model 2301 Cavity Ring-down Spectrometer
CH4/CO2/H2.Q analyser coupled with a Picarro Mobile Measurement Kit. The resolution of this analyser is < 1
ppbv CH4 and has very low drift characteristics (Crosson, 2008) so that very small CH4 perturbations can be
reliably detected against the background concentration. Both instruments were mounted in a 19" rack in
the rear of a 4WD vehicle (Figure 3.1).
Anemometer
Figure 3.1. Photographs of the field vehicle where the GPS antenna and sonic anemometer are visible on the top of
the vehicle (left hand photograph). The methane analyser and a calibration gas cylinder are shown in the rear of the
vehicle (right hand photograph).
The Mobile Kit included a GPS receiver and software that allows the spectrometer output to be processed
and displayed in GIS software. A two-dimensional sonic anemometer (Climatronics Sonimometer) was also
fitted for measuring local wind speed during plume traversing measurements (Section 3.3).
For mobile surveys, the spectrometer was operated continuously as the vehicle was driven. Air was
sampled via a %" nylon tube from the front of the vehicle about 1 m above ground level. The normal flow
rate of sample air to the spectrometer is approximately 100 mL min"1; however, to minimise the lag time
between air entering the inlet tube and reaching the analyser, an auxiliary pump in the Mobile Kit was used
to increase the flow rate to about 5 L min"1. When used for flux chamber measurements (Section 3.5), the
auxiliary pump was bypassed using a three-way valve.
Initially, the instrument was configured to measure CH4, C02 and H20 simultaneously; however, the
sampling rate in this mode was relatively slow with measurements made approximately every 3 s. To
increase the spatial resolution during plume traverses, the sampling rate was increased to about 2 Hz by
reconfiguring the analyser to measure CH4 only.
8 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities A. , 7J_
41 of 75 53 0f 87
-------
The analyser was calibrated against a reference air sample containing 1.732 ppm CH4 prepared by the
CSIRO Marine and Atmospheric Research GASLAB (Francey et al., 2003). Additional standard gas mixtures
of 10.2 and 103 ppm CH4 in air (BOC Gases Australia) were used for multipoint calibrations.
Although the nominal range of the analyser is 0-20 ppm CH4, we found that the instrument could reliably
measure concentrations well in excess of this level. In one experiment, an Ecotech GasCal dilution system
was used to generate gas flows with known CH4 concentrations up to about 280 ppm. The results of this
experiment are shown in Figure 3.2 where the analyser output is plotted against the actual methane
concentration.
300
250
q_ 200
Q.
0
50
100
150
200
250
300
Actual [CH4] (ppm)
Figure 3.2. Calibration curves obtained for the methane analyser. Open circles correspond to points made using gas
mixtures generated with a gas diluter. Red circles represent a multipoint calibration made using reference gases
several months later.
The response of the instrument remained linear at least to 280 ppm CH4. One of the routine multipoint
calibration curves using the three reference gases made several months later (red markers) is also plotted
to demonstrate the low drift characteristics of the instrument.
Multipoint calibrations were performed before and after each field campaign and single point calibration
checks were made periodically in the field.
3.3 Plume Traverses
Methane emissions from well pads were estimated using a plume dispersion method. In this method, the
CH4 concentration profile in a plume originating from CH4 emission sources on the pad is measured at some
distance downwind of the pad by performing traverses across the plume. Since the plume comprises all CH4
released from the pad, it yields total emissions from each pad. The technique is illustrated in Figure 3.3.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production^djit^gs ^
-------
Plume Characteristics
z
x
Emission Source
y
Wind Direction
St?
Figure 3.3. Schematic representation of the piume traversing experiments.
The field vehicle with the CH4 analyser was driven 15 to 50 m downwind of each well to measure the
ground level CH4 concentration across each plume. The emission flux, F, over each traverse was estimated
by integrating the CH4 concentration enhancement (i.e. the measured concentration minus background CH4
concentration), c, of the plume in the horizontal and vertical directions and multiplying by the average wind
velocity, u, measured at each site (Equation 3.1). Background CH4 concentrations were measured by
performing upwind traverses of the well pad.
Since the traverse measurements were made at ground level only, the vertical extent was estimated by
reference to the Pasquill-Gifford curves of ov (i.e. the standard deviation of the distribution of CH4
concentration in the vertical direction) as a function of downwind distance under given atmospheric
turbulence conditions (Hanna et al., 1982). The vertical concentration profile of CH4/ within the plume was
assumed to decrease from the ground level concentration with height according to a Gaussian distribution
across the traverse plane. For each well, an average emission rate was determined from up to 10 traverses
made over about a 20-minute period.
One of the primary sources of uncertainty with the plume traversing method is associated with determining
the height of the plume because it must be estimated rather than measured. To assess the level of
uncertainty in the plume traversing results, we performed a number of experiments where CH4 was
released from a cylinder of compressed gas at a known rate while traverses were made downwind of the
source. The results of the traverses were then compared with the actual rate of CH4 release. These
controlled release measurements were made at a site near the CSIRO laboratories in Newcastle where
there were no other sources of CH4 present and to simulate field conditions, traverses were made between
15 and 50 m downwind of the controlled release point. The results of these experiments are discussed in
Section 4.1.
At each well site an initial survey for elevated CH4 concentrations was made by performing vehicle
traverses as described above to determine if CH4 emissions were present. The presence of elevated CH4
concentrations indicated some type of leak, venting or engine exhaust emission from the pump power
pack. Where CH4 was detected, more detailed examination of the facility was undertaken using a probe
connected to the vehicle mounted CH4 analyser to locate the source or sources of CH4 (Figure 3.4). On
F = U f-y Soc(y> z)dydz
Equation 3.1
3.4 Leak and Vent Testing
10 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^ ^
-------
some occasions, leaks were located by spraying a leak detection solution (Snoop, Swagelok Company) onto
individual components.
Figure 3.4. Locating equipment leaks at a CSG well pad.
When the source of the leak was identified, the leak rate was measured. During the first set of field
measurements, leak rates were measured in accordance with the USEPA Protocol for Equipment Leak
Emission Estimates (USEPA, 1995). In this procedure, the leaking component is enclosed in a plastic bag or
sleeve and an air stream is passed through the bag at a known rate while the outlet stream is analysed for
CH4 concentration. Although this is a proven method for quantifying leak rates, it was found to be very slow
and labour intensive. For later measurements (and the majority of the results reported here) we
constructed a high-flow apparatus, similar in principle to the 'Hi-Flow' device reported by Kirchgessner et
al. (1997). In this system, a high capacity fan attached to a 100 mm diameter flexible tube was used to
provide an air stream around the leak point to entrain the leaking CH4. A variable power supply was used to
allow the fan speed to be varied up to a maximum flow rate of approximately 80 L s"1 (4.8 m3 min"1).
During leak tests, the inlet of the hose was held within about 150 mm of the apparent leak point while the
CH4 concentration in the outlet air stream was measured with the CH4 analyser in the field vehicle. The leak
rate, /?,, was calculated from the volumetric flow rate of the air stream, V, and the steady state CH4
concentration, c, according to Equation 3.2
R-l = V y. C Equation 3.2
A schematic diagram of the apparatus is shown in Figure 3.5.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production l^jiNp^-l ^
-------
OT
To Methane Analyser <-
n
•{ ( Fan () 1 AirFloT^]^^ ( OH^)
Flow FlexibleTube
Straightener
I
n
Figure 3.5. Schematic diagram of the leak testing apparatus. Methane leaking from a component (red arrow) is
entrained in the airstream drawn into the tube by the fan.
Occasionally emission rates from some sources (e.g. vents and pneumatic devices) were amenable to a
simple measurement technique where the exhaust point was sealed in a plastic bag of known volume and
measuring the time required to fill the bag. In a few cases where the emission rate was reasonably
constant, emission rates were measured by attaching a flow calibrator (DryCal DR2) to the emission outlet.
3.5 Surface Emissions
Measurements were made on the ground surface near well heads to determine if CH4 was migrating
around the outside of well casings or through casing walls. These measurements were made using a surface
flux chamber, a technique frequently used to measure emission rates of soil gases. For these
measurements, a plastic cylindrical chamber 37.5 cm in diameter and 40 cm high with a total volume of
about 45 L and an area of coverage of 0.11 m2 was placed on the ground at each sampling point. A small
solar powered fan mounted in the chamber ensured that the sample within the chamber was well mixed
during each experiment. The chamber was connected to the CH4 analyser in the field vehicle via a %" nylon
tube and the CH4 concentration within the chamber, C, continuously measured over a period of several
minutes. The flow rate of the sample stream from the flux chamber to the analyser was approximately 100
mL min"1.
The CH4 emission flux, F, was calculated according to Equation 3.3
F — H-w Equation 3.3
where V is the volume of the chamber, dC/dt is the rate of change in the CH4 concentration over time, t,
and A is the area of surface covered by the chamber.
A schematic diagram of the chamber system is shown in Figure 3.6
12 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^ ^ ^
-------
Analyser
Return Flow
r | -|
Flux
Chamber
tnrtrtr
CH4 Flux
Figure 3.6. Schematic diagram of the flux chamber system used for well casing leak determinations
Typically, chamber measurements were made at four or more points within about 1 m of the well casing. In
many cases, the chamber was placed adjacent to the casing, depending on access. Occasionally, additional
measurements were made at distances up to about 20 m from the well head.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production l^cili|i^-|
-------
4 Results
4.1 Controlled Release
Controlled release experiments were conducted on several occasions with CH4 release rates of between 0.7
and 0.8 g min"1 and traversing distances between 15 and 30 m downwind of the release point. Figure 4.1
shows the results of the controlled release experiments. The black markers represent the mean value
determined by the traverses while the error bars show the minimum and maximum results determined
over each set of traverses. The red markers represent the actual release rate.
c
'E
O)
ro
Q1
c
o
I
O
2.0
1.8
1.6
1.4 -
1.2 -
1.0
$ 0.!
'E
LD
0.6
0.4
0.2
0.0
• Average Measured
• Actual
Experiment Number
Figure 4.1. Summary of the controlled release experiments showing the CH4 release rate determined by plume
traversing and the actual release rate. Downwind distances were: Exp No 1 = 20 m; Exp No2 = 30 m; Exp No 3 = 15
m; Exp No 4 = 30 m. The error bars represent the range of emission rates measured during each set of six traverses.
Two initial experiments using a higher release rate of approximately 3.5 g min"1 and up to 50 m downwind
overestimated the actual emission rate by about 100 and 60 % respectively. However, these experiments
were based on only two traverses each so the poor agreement is unsurprising. The subsequent experiments
(shown in Figure 4.1) were made using six traverses for each determination. In these cases, the agreement
was much better with the emission rate determined by the average of the six runs being within about 30 %
of the actual release rate, although there was significant variation among the individual traverses as shown
by the error bars in Figure 4.1. Measurements made at CSG wells using the plume traversing method were
therefore based on at least six and usually 10 or more individual traverses at each site.
4.2 Well Measurements
Emission measurements were made at 43 sites in NSW (six sites) and Queensland (37 sites). Most sites had
only a single well on the pad, but there were a number where up to four well heads were located on an
individual pad. The majority of wells were production wells, although 11 were not flowing at the time of the
14 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^ ^
-------
measurements due to maintenance or other activities. Two of the wells examined were plugged and
abandoned and one well had been 'suspended' where the well head was still in place but had been
disconnected from the gathering network and most of the surface equipment had been removed.
Twenty-nine wells were producing gas during the measurements, flowing at rates ranging from less than
1000 m3 day"1 to more than 186,000 m3 day"1. Eleven of the sampled wells were hydraulically fractured. The
selection also included a mix of free-flowing wells (water was not pumped from the well) and pumped wells
(water was pumped from the well to allow gas flow). Pumped wells used on-site engines to power hydraulic
pumps or generators to drive down-hole water pumps. In all but one case (which used diesel), these
engines were fuelled from gas supplied from the well. A summary of the wells is shown in Table 4.1. To
maintain commercial confidentiality, the well locations and operators of individual wells are not identified
in this report.
Table 4.1. Details of wells examined during this study.
Well Number
Completion
Date
Production Rate
(mB day"1)
Fracture
Stimulated
Type
Pump with Engine
Wells on Pad
A1
11/10/1999
1,470
Yes
Vertical
No
1
A2- Suspended
1/05/2003
0
Yes
Vertical
No
1
A3
1/07/2007
0
Yes
Vertical
Yes - not running
1
A4
20/04/2010
18,400 (total of all
4 wells on pad)
No
Horizontal
No
4
A5
8/06/2011
14,900
No
Horizontal
Yes
2
A6
11/12/2007
13,700
No
Horizontal
No
1
B1
24/09/2006
38,880
No
Vertical
Yes
1
B2
11/01/2008
0
No
Vertical
No
1
B3
06/08/2011
9,360
No
Vertical
Yes - not running
1
B4
21/09/2010
26,400
No
Vertical
Yes - not running
1
B5
08/12/2010
0
No
Vertical
No
1
B6
27/04/2003
23,760
Yes
Vertical
Yes
1
B7
09/08/2007
26,400
No
Vertical
Yes
1
B8
26/01/2008
62,400
No
Vertical
No
1
B9
23/06/2008
7,680
No
Vertical
Yes
1
B10
07/04/2007
55,200
No
Vertical
No
1
Bll
23/06/2011
94,602
No
Vertical
Yes - not running
1
B12
28/06/2011
0
No
Vertical
Yes - not running
1
B13
21/02/2005
0
No
Vertical
No
1
B14
30/08/2007
75,360
No
Vertical
No
1
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production l^iN^i^-1 ^
-------
B15
08/04/2009
70,800
No
Vertical
No
CI
15/05/2001
76,101
No
Vertical
Yes
C2
2/08/2003
853
No
Vertical
Yes
C3
4/10/2007
0
No
Vertical
Yes - not running
C4
29/03/2007
52,458
No
Vertical
Yes
C5
29/03/2007
58,594
No
Vertical
Yes - not running
C6
28/01/2008
186,464
No
Vertical
Yes
C7
17/09/2009
0
Yes
Vertical
No
C8
22/05/2010
0
No
Horizontal
No
C9
16/10/2003
78,731
Yes
Vertical
No
CIO
1/10/2003
85,556
Yes
Vertical
No
Cll
27/08/2004
0
Yes
Vertical
No
D1 -
Abandoned
8/11/2003
0
No
vertical
No
D2
1/09/2005
93,400
Yes
vertical
No
D3 -
Abandoned
29/11/2003
0
Yes
vertical
No
D4
19/04/2004
0
Yes
vertical
Yes (x2 - not
running)
D5
7/11/2009
7,900
No
vertical
Yes (x2)
D6
28/11/2009
0
No
vertical
Yes (x2)
El
16/3/2008
43,843 (total of
both wells on pad)
No
vertical
Yes
E2
7/9/2008
26,847
No
vertical
Yes
E3
16/3/2007
3,707
No
vertical
Yes - not running
E4
31/5/2009
6,598
No
vertical
Yes
E5
31/5/2005
14,498 (total of all
3 wells on pad)
No
vertical
Yes
Downwind plume traverses were made at all wells sites except Wells B7 and C3 where the wind was too
light to produce stable plumes. Of the well sites where traverses were made only three did not exhibit any
CH4 emissions. These were the two plugged and abandoned wells (D1 and D3) and the suspended well (A2).
All of the other wells examined exhibited some level of CH4 emissions although in most cases the amount
was relatively small. The plume traversing results for all wells are presented in Table A1 in the Appendix.
On-pad measurements were made at most wells except in a few cases where high ambient CH4 levels from
major leaks or vents made locating minor leak points difficult. In one case at Well B2, CH4 released from a
vent on a water gathering line was drifting over the pad components so it was not possible to determine if
16 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities
49 of 75 61 of
-------
there were other leaks against the high background. Similar conditions were encountered at Wells C3 and
E4 where variable plumes from leaks around the water pump shaft seals precluded reliable leak detection.
In one case we attempted to measure emissions from a well about 500 m downwind of a gas compression
plant but the CH4 emissions from the plant prevented any measurements being made at this site.
Most of the CH4 emissions were found to be derived from equipment leaks and venting but we also found
that exhaust from the engines used to drive the water pumps on some wells was frequently a significant
source of methane. Fifteen of the pumped wells had the engines operating during the measurements and
in most cases the exhaust was found to contain CH4 that contributed to total emissions. In a few cases, the
plume from the engine exhaust was sufficiently spatially separated from other sources of CH4 to quantify
the sources separately using the traverse method (Figure 4.2).
~
Separator
Engine
1
Wind
Direction
Figure 4.2. Methane concentration profile at Well C2 showing the separate plumes associated with the engine and
equipment leaks elsewhere on the pad.
However, in most cases the plumes were coincident and the exhaust component could not be separated.
To attempt to estimate the magnitude of engine emissions, we measured the CH4 concentration in the
exhaust outlet of the engine where this was possible. The range of CH4 concentration varied considerably;
from only a few ppm to more than 1500 ppm. The exhaust gas flow rate was estimated from the nominal
fuel consumption (often stated on the engine nameplate) or power rating and assuming a 33 % efficiency
and 17:1 air fuel ratio.
In the example for Well C2 shown in Figure 4.2, the plume traverse yielded an emission rate from the
engine of 0.8 g min 1 compared to the estimate based on the fuel consumption and exhaust CH4
concentration of 0.9 g min"1. In another example, engine emissions from Well B7 were estimated using the
exhaust method to be 0.2 g min"1. A separate measurement made by the well operator using a stack testing
method also gave 0.2 g min"1. While these two examples suggest that this method provides a reasonable
approximation of exhaust CH4 emissions, in many cases the CH4 concentration measured was well above
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production ^;ili|i^| ^ gy
-------
the calibrated range of the CH4 analyser (i.e. > 280 ppm) and hence the results can only be considered
indicative.
Although on-pad measurements provided reasonably accurate leak rate results for individual leak points,
the large number of possible emission sources including equipment leaks, vents, pneumatic devices and
engine exhaust presented a risk that some emission points on each pad would be missed during the surveys
(Figure 4.3).
Figure 4.3. CSG well pad showing some of the surface equipment and potential emission points. Note the engine in
the background for supplying hydraulic power to the water pump.
To check this we compared the emission rates determined from the on-pad measurements to those
calculated from the downwind traverses, which capture all emissions from the pad. Ideally therefore, if all
the emission sources have been accounted for, on-pad measurements should equal emission rates
determined from traverse data. Apart from one result, there was generally good agreement between the
two methods, which is shown in Figure 4.4 where the emission rate determined for each well by the on-pad
methods is plotted as a function of the traversing results. The outlier (red marker in Figure 4.4) corresponds
to Well B2 where the traverses were made under very light and variable conditions, which make accurate
quantification difficult. The mean traverse result for this well was approximately 17 g min 1 but this result
exhibited the greatest variably of all the traverses, ranging from 1 to 66 g min"1. If this result is omitted from
the plot, the slope of the line is close to 1 (0.94) confirming that the on-pad measurements generally
accounted for the main emission points i.e. there were no major sources that were missed during the leaks
surveys.
18 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities f 7J_
si or 03 of 87
-------
30
TJ
(0 20
Traverse Emission Rate (L min")
Figure 4.4. Correlation of total CH4 emissions determined by traverses with on-pad measurements
The well site results from individual companies are discussed in more detail in the following sections.
4.2.1 COMPANY A
Figure 4.5 summarises the total emissions measured at Company A's well sites using the traversing method.
At the time of the measurements only four wells were producing gas - Well A2 was suspended and Well A3
was shut-in for maintenance.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production ^ q-j
-------
3.0
T
Company A
2.5 -
c
'E
O)
2.0
03
ir.
c
0
'(/)
w
'E
LU
1
o
1.5
1.0
0.5
0.0
Suspended
A1
V
A2
Max = 13.4
Avg = 7.3
A3 A4
Well Number
A5
A6
Figure 4.5. Total CH4 emission rates estimated at Company A's well sites using the traversing method.
Apart from the suspended well (A2) emissions were detected at each site. Generally emissions were very
low rwith five of the wells having emissions below about 0.1 g min"1. On-pad measurements made at the
well sites showed that in two cases (Wells A1 and A5) the emissions were due to the operation of
pneumatic devices with emission rates of ~75 mg min"1 and 55 mg min"1, respectively.
Two other wells (A3 and A6) were also found to have minor emissions but at the time the measurements
were made, venting from pneumatic equipment was not contributing (i.e. these devices did not operate
over the few hours we were on site at each well). In the case of A6, CH4 was leaking slowly from a loose
plug on a branch pipe at a rate of 22 mg min"1. This leak was repaired by gas company personnel shortly
after it was identified and further measurements on site showed that the leak had been eliminated. At Well
A3, a leak was found in the gathering line, but again, this was very small amounting to less than 1 mg min"1.
The largest emissions were found at Well A4. Two separate sets of traverses yielded an average emission
rate of 7.3 g min"1. Methane leaks were detected at a valve and pipe joint on the well pad but the combined
emission rate from these was about 7 mg min"1 so the bulk of the methane release was from another
source. This well was on a pad with three other wells within close proximity, which were not examined in
detail during this campaign, so it is possible that some of the observed methane in the plume may have
originated from these other wells. However, the bulk of the source was traced to a buried gathering line
adjacent to the pad that serviced all four wells. We attempted to measure the emission rate using the
surface flux chamber method; however, because of the diffuse nature of the emissions through the gravel,
this was not successful.
Although the average emission rate of 7.3 g min"1 (15.5 m3 day"1 at 15 °C) determined by the traverses was
by far the largest emission source found at Company A, it represented only about 0.1 % of the indicated gas
flow of 18,400 m3 day"1 from the four wells on the pad.
A summary of the emissions determined by on-pad measurements at Company A is provided in Table 4.2.
20 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities cr> e-7c r^r- r. r\->
or co 65 OT 87
-------
Table 4.2. Summary of on-pad emission rates measured at Company A sites; nf denotes 'not found'. Note the leak
rate shown for Well A6 was determined from the traverses.
Well Number Leaks (g min"1) Vents (g min"1) Pneumatics (g min"1]
A1
A2
A3
A4
A5
A6
3.3 x 10"
0
4.5 x 10"'
7.3
0
2.2 x 10"2
nf
nf
nf
nf
nf
nf
7.5 x 10"
nf
nf
nf
5.5 x 10":
nf
4.2.2 COMPANY B
Methane emissions estimates based on the traverses for the Company B wells are summarised in Figure
4.6.
O)
CD
-t—'
03
Q1
c
0
'(/)
co
'E
LD
1
o
Company B
B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14 B15
Well Number
Figure 4.6. Total CH4 emission rates estimated at Company B's well sites using the traversing method.
These emissions were somewhat higher than measured at Company A with average emissions ranging from
less than 50 mg min"1, (B4, B5, B12 and B13) to 17 g min"1 (B2). Note however, that one individual traverse
on B2 indicated an emission rate of more than 66 g min"1. The traverses at Well B2 were made under light
and variable wind conditions so the results are subject to high uncertainty. More accurate emissions
measurements of emissions were made at B2 using an on-pad method. In this case, CH4 was found to be
predominantly released from a single vent on a water gathering pipe from the well. The flow rate from the
vent was relatively constant at 44 g min"1 (measured using a flow calibrator), which was within the range of
the traverses but higher than the traverse average of 17 g min"1. The high CH4 emission rate however,
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production £ajiNp^-| gLg ^
-------
meant that it was not possible to identify any other sources on the pad because the plume was engulfing
the surface equipment.
Well B2 was not flowing at the time of the measurements, but assuming the normal flow rate is 26,400 m3
day"1 (i.e. the median production rate of the Company B wells examined), fugitive emissions from this vent
represent about 0.4 % of the well's production.
Emissions at the other Company B well sites were much lower than B2, with emission rates generally less
than 2 g min"1. Most of the well sites exhibited a small level of leakage from certain items of equipment and
especially a particular brand of pressure regulator. These regulator leaks however, were quite low with the
maximum measured less than 25 mg min"1. Most of the CH4 emissions were, like Well B2, from vents
present on many of this company's wells. Vent emissions were significantly higher than the equipment
leaks, typically more than 1 g min"1, with the maximum of 44 g min"1.
The on-pad measurements for Company B are summarised in Table 4.3.
Table 4.3.Summary of on-pad emission rates measured at Company B sites; nf denotes 'not found'.
Well Number
B1
B2
B3
B4
B5
B6
B7
B8
B9
BIO
Bll
B12
B13
B14
B15
Leaks (g min"
2.4 x 10
nf
2.1 x 10"4
1.5 x 10"3
nf
6.4 x 10"3
9.6 x 10"4
2.1 x 10"2
2.4 x 10"3
2.3 x 10"2
2.5 x 10"2
3.0 x 10"4
1.0 x 10"3
3.94 x 10"3
2.4 x 10"3
2.9
43.8
nf
nf
nf
1.0
1.1
6.2
nf
3.6 x 10"2
1.2
< 10"4
0.9
3.3
Pneumatics
nf
nf
nf
nf
nf
nf
nf
nf
nf
nf
nf
nf
nf
nf
nf
n
. -l
¦
min
In addition to the emissions from the well pads, we found a significant CH4 emission point from a water
gathering line installation near Well B13 (Figure 4.7).
22 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^ ^
-------
Methane Release Points
Figure 4.7. Methane emission sources on a water gathering line.
Methane was being released from the two vents shown in Figure 4.7 at a rate sufficient to be audible a
considerable distance from the vents. It was not possible at the time to the site visit to directly measure the
emission rate from the vents due to restricted access, however, the CH4 concentration 3 m downwind of
the vents was 15 % of the lower explosive limit of CH4 (i.e. 7,500 ppm). Based on the prevailing wind speed,
we estimate that the CH4 emission rate from the two vents was at least 200 L min"1 (130 g min"1) or almost
300 m3 day"1. This is a factor of three more than the highest emitting well examined during this study.
4.2.3 COMPANY C
Figure 4.8 summarises the CH4 emission rates estimated by the traversing method for Company C,
Emissions were generally estimated to be below 1,5 g min"1, except for Weils CI and C4, with emission rates
of about 8.7 and 11.8 g min"1, respectively. The bulk of the emissions from wells CI and C4 were due to CH4
in the engine exhaust rather than venting or equipment leaks. Similarly, emissions from Wells C2 and C6
comprised mainly CH4 in engine exhaust although the emissions rates were much lower than CI and C4.
On-pad measurements at each of the wells showed that emissions from the wells were generally relatively
low when the engine exhaust is excluded (Table 4.4). In this case, leaks were mostly less than 0.3 g min"1.
Most of these leaks were found to be from vent pipes on equipment such as pressure relief valves or
pressure regulators similar to those on Company B's well sites. In one case (Well C10), a pneumatic device
was found to be venting at an average rate of 0.5 g min"1 in addition to the equipment leak rate of 0.3 g
min"1 to give a total emission of 0.8 g min"1.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production l^cilij;i^|.| gSg ^
-------
O)
a)
ro
a:
c
0
'(/)
co
'E
LD
1
o
30
25 -
20 -
15 -
10 -
5 -
Max = 34.8
Company C
I
4U
I
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11
Well Number
Figure 4.8. Total CH4 emission rates estimated at Company C's well sites using the traversing method.
Traverses were not made at Well C3 due to lack of wind, however, on-pad inspections revealed a significant
gas leak was on the seal of the water pump shaft. The emission rate from this leak was approximately 28 g
min"1 (measured using the high-flow apparatus), which was the second largest well emission (after B2) and
the largest equipment leak of the 43 sites examined. Since this well was shut-in at the time of
measurement, it was not flowing but using the median flow rate of Company C's wells (52,500 m3 day"1) the
leak rate corresponds to about 0.1 % of the well's production.
The water pump shaft seal was also found to be the source of CH4 leakage at Well C5 but in that case, the
emission rate was about 0.3 g min"1, about 100 times less than C3.
Table 4.4. Summary of on-pad emission rates measured at Company C sites; nf denotes 'not found'.
Well Number
CI
C2
C3
C4
C5
C6
C7
C8
C9
Leaks (g min"
5.3 x 10"
0.2
28.
8.0 x 10"2
0.3
0.2
0.1
2.1 x 10"3
8.9 x 10"3
nf
nf
nf
nf
nf
nf
nf
nf
nf
Pneumatics
nf
nf
nf
nf
nf
nf
nf
nf
nf
n
. -l
¦
min
24 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ gg f 87
-------
CIO
Cll
0.3
7.4 x 10"
nf
nf
0.5
nf
4.2.4 COMPANY D
Two of the wells at Company D were plugged and abandoned with all surface equipment removed. Detailed
traverses and flux chamber measurements made on the well sites revealed no sign of any residual
emissions from these wells. The traversing results for Company D are shown in Figure 5.9.
c
'E
O)
ro
Q1
c
0
'(/)
co
'E
LD
1
o
Company D
P & A
P & A
D3
Well Number
Figure 4.9. Total CH4 emission rates estimated at Company D's well sites using the traversing method.
Of the operating wells, D2 had the lowest emissions with on-pad measurements indicating total emissions
of less than 60 mg min"1, which were due to minor equipment leaks. Well D4 also had low emissions
totalling about 65 mg min"1. A small emission from a pneumatic actuator of approximately 14 mg min"1 was
also found on well D4.
Wells D5 and D6 had higher total CH4 emission rates and although affected by engine exhaust, significant
proportions of the observed emissions were due to equipment leaks. In the case of D5, most of the CH4 was
leaking from the water pump shaft seal at about 1.5 g min"1 (Table 5.5). For D6, we estimate that about two
thirds of the CH4 was due to engine exhaust but approximately 0.75 g min"1 was leaking from what
appeared to be a damaged diaphragm in a valve actuator (Figure 5.10). Several smaller leaks on this well
resulted in a total leak rate of about 0.9 g min"1.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Jg£iNp^.| 2^ ^
-------
Methane leaking from
damaged actuator
diaphragm
Figure 4.10. Methane leak from a valve actuator. Note the soap solution bubbles around the emission point.
Table 4.5 shows a summary of the on-pad results from Company D,
Table 4.5. Summary of on-pad emission rates measured at Company D sites; nf denotes 'not found'.
Well Number Leaks fg min'1) Vents (g min"1) Pneumatics (g min"1)
D1 0 nf nf
D2 5.7 x 10"2 nf nf
D3 0 nf nf
D4 6.4 x 10"2 nf 1.4x 10"2
D5 1.5 nf nf
D6 0.9 nf See note
Note: Although the emissions from the actuator shown in Figure 4.10 were from a pneumatic device, it appeared that this was due to a leak rather
than normal operational emissions. Hence we have classified this as a leak in Table 4.5
4.2.5 COMPANY E
The traverse results obtained for Company E are shown in Figure 4.11. The lowest emitting well of the five
examined was E5. This well was located on a pad of three wells, with a single engine providing power to all
three water pumps. Emissions from all three wells were less than 60 mg min"1, most of which were
26 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^
-------
probably associated with engine exhaust. We did not find any equipment leaks or venting emissions at this
site.
O)
CD
-t—'
03
Q1
c
0
'(/)
co
'E
LU
1
o
25
20
15 -
10 -
5 -
Max = 56.8
Company E
<
i
\-1
1 •—1
i
m-
E1
E2
E3
Well Number
E4
E5
Figure 4.11.. Total CH4 emission rates estimated at Company E's well sites using the traversing method.
The other wells, however, showed higher emissions, the largest of which was on Well E4 with an emission
rate of about 15 g min"1. This was traced to a leak on the water pump shaft seal. Like a number of other well
sites examined during this study, the seal was repaired on site once the leak had been identified and
subsequent measurements confirmed that CH4 leakage was completely eliminated.
Well site El was also found to be leaking CH4 from the water pump shaft seal. This site had two wells on the
pad and both were found to be leaking from the seal. The combined rate of leak from this source was 0.7 g
min"1. These wells also showed significant leakage from two pressure regulators, similar to those used at
various other well pads examined, with a combined emission rate of 1.7 g min"
at El were 2.5 g min"1 (Table 4.6).
, Total emissions from leaks
The next highest emitting well from Company E was E2 but most of these emissions were apparently from
the engine exhaust. For E3, a very slight leak was detected from the pump shaft seal (about 40 mg min"1)
but most of the CH4 emissions were from a leak in a filter attached to the engine fuel line (0.6 g min"1).
Table 4.6. Summary of on-pad emission rates measured at Company E sites; nf denotes 'not found'.
Well Number
El
Leaks (g min"
2.5
nf
Pneumatics
nf
m
. -i
¦
min
E2
E3
E4
E5
nf
0.6
15
0
nf
nf
nf
nf
nf
nf
nf
nf
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production l^iN^i^l ^
-------
4.3 Casing Leaks
CSG wells are designed so that gas is extracted from the seam through a well casing but if the casing is
damaged or improperly sealed into the surrounding strata, it is possible that gas can migrate to the surface
outside the casing (Figure 4.12). To determine if CH4 was escaping from the well casing, the flux chamber
method was applied at each well site to measure the emission rate of any leakage from around the outside
of the casing.
Possible leak
around casing
Water to treatment plant
Water pumped
up tubing
COAL SEAM
Gas to processing plant
Gas flows up
casing
i >
BOTTOM HOLE PUMP
Typically
300 to 1000 metres
Figure 4.12. Schematic representation of a CSG well showing a possible route for CH4 leaking outside a casing.
We anticipated that ieakage from this source may be quite low, so it was important to ensure that the
measurement technique had sufficient sensitivity to detect low level seepage. Therefore, prior to making
field measurements a series of preliminary experiments were performed to determine the lower limit of
detection of the method. Several experiments were made using a controlled release of CH4 into the flux
chamber system. Figure 4,13 (a) shows a plot of the CH4 concentration within the chamber over about 5
minutes. The actual flow rate of CH4 into the chamber was 7.76 x 105 g min"1 whereas the measured rate
was 7,42 x 10"5 L min"1 or a difference of about 4 %. While this is a very low emission rate (cf. the smallest
well leak rates of ~3 x 10" ' g min"1) the ultimate sensitivity was several orders of magnitude lower.
Measurement of CH4 emissions from natural surfaces showed that emission rates less than 1 x 10"' g min"1
could be reliably quantified (Figure 4,13 b).
28 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities f 7J-
or 73 of 87
-------
12 -
10 -
0
50
100
150
200
250
300
350
Time Cs)
1.785
1.780 -
1.775 -
1.770 -
1.765 -
1.760
0
100
200
300
400
500
600
Time (s)
Figure 4.13. Methane concentration as a function of time in the flux chambers (a) controlled release experiment; (b)
natural surface emission.
At the well sites, even with the very high sensitivity of the chamber method, we did not detect any
emissions from around the well casing. Because the flux chamber measurements were applied at discreet
points around the well it is possible that leak points were missed, however we believe that this was very
unlikely since any significant emissions would have been detected during the mobile plume traverses and
leak detection measurements made near the well heads.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production ^ q-j
-------
5 Discussion
Overall, the emission rates measured at the well sites were quite low, especially when compared to the
volume of gas produced. Of the 43 sites examined, 19 had emission rates less than 0.5 g min"1 and 37 less
than 3 g min"1; however, there were a number of wells with substantially higher emission rates up to 44 g
min"1 (Figure 5.1).
CH4 Emission Rate (g min")
Figure 5.1. Histogram of emission rates from all sources measured at the 43 well sites.
Well pad emissions were found to be derived from several sources:
• exhaust from engines used to power dewatering pumps,
• vents and the operation of pneumatic devices and
• equipment leaks.
The mean emission rate of all of these sources for all wells is 3.2 g min"1 whereas the median (middle value)
is 0.6 g min"1.
Engine exhaust is not considered to be a fugitive emission for the purposes of greenhouse accounting since
it is counted separately as a combustion source. Nevertheless, exhaust represented a significant proportion
of the total CH4 emissions at some well sites. The wide range of CH4 concentrations present in the exhaust
meant that the contribution of exhaust to overall emissions was highly variable. Some engines appear to
have very low CH4 emissions such as that at Well A5. Similarly, an unidentified well in Queensland was
found to have no detectable CH4 in the exhaust within close proximity to the pad (Day et al., 2013). On the
other hand, engine exhaust was by far the primary source of CH4 emissions at some wells (e.g. Wells CI and
C4).
As noted in Section 2, methane emissions from combustion are estimated for NGER reporting using an
emission factor of 0.1 kg C02-e GJ"1 (DIICCSRTE, 2013b), which is equivalent to 4.8 g CH4 GJ"1 using a global
warming potential for CH4 of 21. Assuming that the fuel consumption of the well site engines was 594 MJ
30 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities , 71-
oT 75 Of 87
-------
h 1 (indicated on the nameplate fitted to one make of engine commonly used throughout the industry), this
equates to a CH4 emission rate of 0.05 g min"1, which lower than some of the estimates made during the
study. Well C4 for example was estimated to be emitting CH4 at a rate of 11.8 g min \
Pneumatic devices, which are potential emission points, were installed at many wells, although during the
measurement campaign, only seven of these were releasing CH4 at the time of the site visits. Emissions
from the these pneumatic devices ranged from 3.8 x 102 to 0.47 g min 1 with a mean emission rate 0.12 g
min 1 and standard deviation of 0.18 g min \ This is somewhat lower than the emission rate for pneumatic
devices recently reported by Allen et al. (2013). They found that the average emission rate from
intermittent pneumatic devices at U.S. unconventional gas well was 5.9 ± 2.4 g min \ The result obtained
for the Australian CSG wells is also lower than the production average emission factor for pneumatic
devices provided in the API Compendium (API, 2009) of 345 ± 49.5 scf d 1 (4.6 ± 0.66 g min"1).
It is not clear why these emission rates are lower than the U.S. estimates; however, it should be borne in
mind that the results of our study represent only a very small sample. The Allen et al. (2013) study
examined 305 devices compared to only seven in our study. Another reason for the difference may be due
to the intermittent operation of the devices. Most of the CH4 emission apparently occurs when the devices
operate and hence the frequency of operation has a strong influence on the emission rate so a longer
period of sampling may have yielded different results.
Despite the uncertainty of the results for pneumatic devices, it is probable that emissions from these
systems will tend to decrease in the future. Some Australian CSG companies are now installing compressed
air operated or electrical actuators on newer well pads which will eliminate pneumatic CH4 emissions from
these pads.
Vents installed at various points on some well pad equipment were frequently found to be sources of CH4
emissions. Of the 43 well sites examined, ten had vents, all from Company B, that were emitting CH4 at the
time measurements were made. The rate of emissions varied substantially from less than 104 g min 1 up to
44 g min"1, which was the highest rate of emissions measured from any source measured during this
project. The mean vent emission rate was 6.1 g min 1 with a standard deviation of 13.4 g min"1, reflecting
the large range of values.
The third main source of well pad CH4 emissions was from equipment leaks. Most of the wells examined
were found to have some degree of leakage from equipment on the pad. Minor leaks (usually less than 60
mg min"1) were found on various items such as fuel lines to engines, valves, sight gauges on separators and
other equipment. However, there were some leak points that were consistently found across the well sites.
The first of these was a particular type of pressure regulator installed at many wells (Figure 5.2). This device
was apparently associated with the separator and was usually found to be leaking a small amount of CH4.
Mostly, these leaks were less than 150 mg min 1 but in one case (Well El) the emission rate was about 1.5 g
min"1.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production
-------
¦
£tt
Figure 5.2. Pressure regular that was a common source of CH4 leakage.
The other common leak point was the seal around water pump shafts on pumped wells (Figure 5.3). The
two largest equipment leaks detected were due to leaking seals at Welis C3 and E4. At the time of the site
visit, Well C3 was shut-in for maintenance and as a result the pressure on the seal was almost 2 MPa, which
was much higher than normal operating pressure and this is likely to have contributed to the high leak rate
from the well. This is consistent with a study of leaking wells in Queensland made in 2010 where high CH4
concentrations (up to 6 % CH4) due to leaks were often found on shut-in wells that were under high
pressure (DEEDI, 2010).
32 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^
-------
Figure 5.3. Well head showing the location of the water pump shaft and seal which was found to be a common leak
point.
At Well E4, the seal had apparently 'dried out' since the previous inspection and was allowing CH4 to leak
around the rotating pump shaft at almost 15 g min"1. After the leak was identified, however, maintenance
staff applied more grease to the seal and tightened the gland around the shaft, which effectively eliminated
the leak. A smaller leak of around 1.5 g min 1 on the shaft seal on Well D5 was also repaired on site by
simply tightening the gland.
Although the water pump shaft seal is a potentially large source of CH4 emissions, it is clear that in many
cases these leaks can be easily repaired. Regular inspection of these seals, especially during shut-ins when
the well pressure may increase substantially, is therefore likely to be important for minimising well site
emissions.
None of the wells examined during this study exhibited any sign of CH4 emissions around the well casing so
this does not appear to be a common route for CH4 release. Methane leaks have been detected at ground
ievel adjacent to weii casings on Australian CSG wells previously but these were traced to leaks in the
threaded connection between the casing and well head base (DEEDI, 2010) rather than gas leaking around
the outside of the casing.
Despite this, it has been suggested that 6 to 7 % of well completions in the United States are subject to
integrity failure that could lead to CH4 leakage (Ingraffea, 2013). Given that we surveyed less than 1 % of
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production
»«?8of87
-------
Australian CSG wells, it is possible that the small sample size is not sufficiently representative to assess the
true extent of well leakage. Further work would be required to conclusively determine the extent of casing
leaks.
Four of the wells surveyed were horizontal; the remainder were vertical. The range of emissions from the
four horizontal wells was 0.05 to 7.3 g min"1 compared to 0 to 44 g min"1 for the vertical wells. It is not
possible based on only four wells to determine if horizontal wells have different emission characteristics
compared to vertical; however, it seems unlikely that this would be the case. The emission routes were
always associated with surface equipment, some of which was common to both horizontal and vertical well
pads.
Eleven wells examined had been hydraulically fractured and as shown in Table 5.1, average emissions from
these wells were lower (0.42 g min"1) than those measured on the unfractured wells (4.2 g min"1). Because
the data are heavily skewed and it is unlikely that the sample size is statistically representative, it is
misleading to draw conclusions about the relative emission rates based on a comparison of means alone.
Methane emissions were observed from both fracture stimulated and unfractured wells but in all cases,
emissions were from surface equipment that would not be expected to be affected by the stimulation
method. Therefore, the observed difference between the emission rates of the fractured and unfractured
wells in this sample is probably unrelated to the stimulation method.
Table 5.1. Comparison of emission rates measured on hydraulically fractured and unfractured wells.
Fractured Unfractured
Number of Wells 11 32
Mean (g min"1) 0.42 4.2
Median (g min"1) 0.07 1.0
Std Deviation (g min"1) 0.66 14.3
Another parameter that was initially thought to possibly contribute to differences in emission rates was the
well production rate. The range of gas production from the wells varied substantially but there was no
observable correlation between production and leak rate. The highest emissions were from wells that were
not producing gas at the time of the measurements. In the case of one of the non flowing wells (C3) at
least, it may have been that the high well pressure due to the shut-in was contributing to the high leakage.
Conversely, Well C6, which was producing about 186,000 m3 day"1 (cf. the median production rate of 13,700
m3 day"1) had relatively low emissions, most of which were derived from the exhaust from the engine on
the well pad.
Despite the rather low well pad emissions measured during this study, a much higher emission source was
identified on a water gathering line installation. Unfortunately accurate measurements could not be made
at this site but indicative estimates suggested that the emission rate from this source was at least three
times higher than the largest emission rate measured on any of the wells. Similar installations are
widespread through the Queensland gas regions and occasionally, gas can be heard escaping from vents on
these systems. It is possible that these may be a significant source of CH4 and is an area that needs further
investigation.
5.1 Emission Factors
As discussed in Section2 emissions from equipment leaks are often estimated for NGER reporting according
to Method 1 using a generic emission factor of 1.2 kg C02-e t"1, which is equivalent to 57 g CH41"1. It is
therefore instructive to compare this emission factor to the leak emission data measured in the field. The
field measurements yielded a median leak rate 0.02 g min"1 and mean rate of 1.6 g min"1 from the 35 wells
34 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^ ^
-------
where leaks were found. The median production rate of the wells was 13,700 m3 day"1 or 9.3 t CH4 day"1
(referenced to 15 °C). Dividing the median daily leak rate by the median production rate gives an emission
factor of approximately 4 g CH41"1 or 0.1 kg C02-e t"1 (based on a global warming potential of 21). Using the
mean leak rate of 1.6 g min"1 and mean production rate of 29,600 m3 day"1 yields an emission factor of 115
g CH41"1 or 2.4 kg C02-e t"1. This range is consistent with the current NGER emission factor for general
equipment leaks and tends to confirm that equipment leaks comprise only a very small proportion of
greenhouse gas emissions from CSG production.
Similar calculations may be made to develop emission factors for vents and pneumatic equipment. A
summary of the emission data for leaks, vents and pneumatic equipment and the corresponding emission
factors calculated from these data are shown in Table 5.2.
Table 5.2. Summary of emission data from leaks, vents and pneumatic equipment. Emissions factors calculated
from the mean emission rate for each category are also shown in units of kg C02-e t"1 (GWP of 21 used in this
calculation).
Equipment Leaks
Vents
Pneumatic
Equipment
Mean (g min"1)
1.59
6.05
0.12
Median (g min"1)
0.02
1.14
0.06
Std Dev
5.36
13.40
0.18
N
35
10
7
Calculated Emission Factor from
Mean Emission Rate (kg C02-e t"1)
2.4
9.1
0.2
Although these averaged emission factors are low it should be remembered that firstly, the number of
wells examined was less than 1 % of wells in operation so may not be representative of the total well
population and secondly, there were several equipment leaks that were much higher than the average
values (Figure 5.1). The maximum leak rate measured in this study was about 28 g min"1 on Well C3 and
although this well was not flowing at the time, based on the median production rate for all wells, is
equivalent to 91 kg C02-e t"1. A high leak rate of 15 g min"1 was also found at Well E4 and based on its
production rate, equates to 102 kg C02-e t"1. These leak rates are about two orders of magnitude higher
than the current NGER emission factor for equipment leaks.
Another important point with regard to the reliability of emission factors is that they may change due to
operating conditions or maintenance. For instance, the leak from Well E4 discussed above was repaired
during the site visit and completely sealed. Several other leaks were effectively repaired during the course
of the visits once they were identified. However, since wells operate largely unattended, there may be
some time between when the leak forms and when it is repaired.
With regard to well casing leaks there is currently no emission factor representative of Australian
operations for estimating emissions. The current Method 2 emission factor is based on measurements
made at some Canadian wells during the mid 1990s (CAPP, 2002). While there have been suggestions that
well leakage may be a significant source of emissions (Somerville, 2012), the wells examined in this study
showed no evidence of emissions via this route. But again, this needs to be considered in the context of the
small number of wells examined.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production ^iN^i^l ^
-------
§
Fugitive CH4 emission rates were measured at 43 CSG well sites in Queensland and NSW. A range of
methods was applied including downwind traverses of CH4 plumes originating from well pads, and on-pad
measurements to determine leak rates from individual items of equipments and well casings.
Emission rates from production sites ranged from zero to a maximum of about 44 g min \ The highest
emission rate was due to CH4 released from a vent on the well pad while the lowest emitters were two
plugged and abandoned wells and a suspended well. All of the producing wells were found to have some
level of emissions, although in all cases these were very low compared to overall production. Emissions
were found to comprise equipment leaks, venting, pneumatic device operation and engine exhaust. The
wells examined in this study did not show any evidence of CH4 migration outside the well casing.
Overall, the median CH4 emission rate from all sources for the wells examined was approximately 0.6 g
min 1 while the mean emission rate was about 3.2 g min 1 or about 7 m3 day"1. This compares to a mean
production rate of the 43 wells of 29,600 m3 day1 and represents about 0.02 % of total production. This is
very much lower than recent estimates of CH4 emissions from unconventional gas production in the United
States.
Apart from vents, highest emissions were due to CH4 leaking from seals on water pump shafts. On several
occasions, these leaks were repaired on site once they were identified. The median emission rate of all the
equipment leaks identified was 0.02 g min 1 and the mean was 1.6 g min"1, which yield emission factors of
about 0.1 kg C02-e t"1 and 2.4 kg C02-e t"1, respectively. This range is consistent with the emission factor
currently used in the National Energy and Greenhouse Reporting Method 1 methodology for estimating
equipment leaks.
Although well pad emissions were generally found to be low, one significantly higher emission source was
found on a vent associated with a water gathering line. This source appeared to be at least three times
higher than the highest emission rate from any well examined.
The results obtained in this study represent the first quantitative measurements of fugitive emissions from
the Australian CSG industry; however, there are a number of areas that require further investigation.
Firstly, the number of wells examined was only a very small proportion of the total number of wells in
operation. Moreover, many more wells are likely to be drilled over the next few years. Consequently the
small sample examined during this study may not be truly representative of the total well population. It is
also apparent that emissions may vary over time, for instance due to repair and maintenance activities. To
fully characterise emissions, a larger sample size would be required and measurements would need to be
made over an extended period to determine temporal variation.
In addition to wells, there are many other potential emission points throughout the gas production and
distribution chain that were not examined in this study. These include well completion activities, gas
compression plants, water treatment facilities, pipelines and downstream operations including LNG
facilities. Emissions from some of these sources are often estimated for reporting purposes using
methodology based on emission factors largely derived from the U.S. gas industry. However, reliable
measurements on Australian facilities are yet to be made and the uncertainty surrounding these some of
these estimates remains high.
36 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities 09 0f 75 f 87
-------
References
Allen, D.T. Torres, V.M., Thomas, J., Sullivan, D.W., Harrison, M., Hendler, A., Herndon, S.C., Kolb, C.E.,
Fraser, M.P., A. Daniel Hill, A.D., Lamb, B.K., Miskimins, J., Sawyer, R.F., Seinfeld, J.H., 2013.
Measurements of methane emissions at natural gas production sites in the United States.
Proceedings of the National Academy of Science 110,18023-18024.
Alverez, A.A., Pacala, S.W., Winebrake, J.J., Chameides, W.L., Hamburg, S.P., 2012. Greater focus needed on
methane leakage from natural gas infrastructure. Proceedings of the National Academy of Science,
109, 6435-6440.
API, 2009. Compendium of greenhouse gas emissions methodologies for the oil and natural gas industry.
American Petroleum Institute, Washington DC.
Brandt, A.R., Heath, G.A., Kort, E.A., O'Sullivan, F., Petron, G., Jordaan, S.M., Tans, P., Wilcox, J., Gopstein,
A.M., Arent, D., Wofsy, S., Brown, N.J., Bradley, R., Stucky, G.D., Eardley, D., Harris, R., 2014.
Methane leaks from North American natural gas systems. Science 343, 733-735.
BREE, 2013. Energy in Australia 2013. Bureau of Resources and Energy Economics, Canberra
(hup ://bro«. sliced labs, com.au/sitos/dofau ll/filos/fil<;s/pul)l ications/onorj;y -in- aust/broo-
, accessed 12 February 2014).
CAPP, 2002. Estimation of flaring and venting volumes from upstream oil and gas facilities. The Canadian
Association of Petroleum Producers, Calgary.
(http://www.capp.ca/library/publications/sourGasi Iar1ni»Ventini»/pai»es/publiifo.aspx?l)ocId' >3823
4, accessed 30 May 2014).
Crosson, E.R., 2008. A cavity ring-down analyzer for measuring atmospheric levels of methane, carbon
dioxide, and water vapor. Applied Physics B 92, 403-408.
Day, S., Etheridge, D., Connell, N., Norgate, T. (2012). Fugitive greenhouse gas emissions from coal seam gas
production in Australia. CSIRO Report EP128173. 28 pp.
Day, S., DeN'Amico, M., Fry, R., Javanmard, H. (2013). Preliminary field measurements of fugitive methane
emissions from coal seam gas production in Australia. Proceedings of the 14th International
Conference on Coal Science and Technology, Pennsylvania State University 29 Sept-3 Oct 2013,
1445-1455.
DEEDI, 2010. Leakage testing of coal seam gas wells in the Tara 'rural residential estates' vicinity.
Department of Employment, Economic Development and Innovation, Brisbane.
(h up ://w w w, d n r rn. c f
accessed 21 February 2014).
DIICCSRTE, 2013a. Australian national greenhouse accounts. National inventory report 2011, Volume 1.
Department of Industry, Innovation, Science, Research and Tertiary Education, Canberra.
(hup ://www.climatochanf;o.Hov.au/sitos/climatochanf50/filos/docu monts/Ob. 7013/AUS NIK 7011
Voll.pdf, accessed 23 February, 2014).
DIICCSRTE, 2013b. Australian national greenhouse accounts. National greenhouse account factors.
Department of Industry, Innovation, Science, Research and Tertiary Education, Canberra.
(hrrti://www diiTiaf{»chafW{».s»ov.aii/sif{»s/diiTiatechani»e/fiIes/documents/0/ 7013/national •
, accessed 17 May 2014)
DNRM, 2014. Queensland's coal seam gas overview, January 2014. Queensland Department of Natural
Resources and Mines, (h - • .industry ¦¦ ¦ ' , ¦ ,
accessed 11 February 2014).
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production g' ^ gy
-------
Etiope, G., Drobniak, A., Schimmelmann, A., 2013. Natural seepage of shale gas and the origin of the
'eternal flames' in the Norther Appalacian Basin, USA. Marine and Petroleum Geology 43,178-186.
Francey, R.J., Steele, L.P., Spencer, D.A., Langenfelds, R.L., Law, R.M., Krummel, P.B., Fraser, P.J., Etheridge,
D.M.,Derek, N., Coram, S.A., Cooper, L.N., Allison, C.E., Porter, L., Baly, S., 2003.The CSIRO
(Australia) measurement of greenhouse gases in the global atmosphere, report of the 11th
WMO/IAEA. Meeting of Experts on Carbon Dioxide Concentration and Related Tracer Measurement
Techniques, Tokyo, Japan, September 2001, S.Toru and S. Kazuto (editors), World Meteorological
Organization Global Atmosphere Watch 97-111.
Hanna, S.R., Briggs, G.A., Hosker Jr., R.P., 1982. Handbook on atmospheric diffusion. U.S. Department of
Energy, Technical Information Center (page 29).
Ingraffea, A.R., 2013. Fluid migration mechanisms due to faulty well design and/or construction: an
overview and recent experiences in the Pennsylvania Marcellus play. Physicians, Scientists &
Engineers for Healthy Energy.
(1'ittpi//|• " ' y,c>«»/data/i:>Si: Cement failure. Causes and K; Analaysis Jan 703.3. 1
, accessed 24 February 2014).
IPCC, 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the National
Greenhouse Gas Inventories Programme, Eggleston H.S., Buendia L., Miwa K., Ngara T. and Tanabe
K. (eds). Published: IGES, Japan.
Karion, A., Sweeney, C., Petron, G., Frost, G., Hardesty, R.M., Kofler, J. Miller, B.R., Newberger, T., Wolter,
S., Banta, R., Brewer, A., Dlugokencky, E., Lang, P., Montzka, S.A., Schnell, R., Tans, P., Trainer, M.,
Zamora, R., Con ley, S., 2013. Methane emissions estimate from airborne measurements over a
western United States natural gas field. Geophysical Research Letters 40, 4393-4397.
Kirchgessner, D.A., Lott, R.A., Cowgill, R.M., Harrison, M.R., Shires, T.M., 1997. Estimate of methane
emissions from the U.S. natural gas industry. Chemosphere 35,1365-1390.
Klusman, R.W., 1993. Soil gas and related methods for natural resource exploration. Wiley, Chichester, UK.
Miller, S.M., Wofsya, S.C., Michalak, A.M., Kort, E.A., Andrews, A.E., Biraude, S.C., Dlugokenckyd, E.J.,
Eluszkiewicz, J., Fischerg, M.L., Janssens-Maenhouth, G., Miller, B.R., Miller, J.B., Montzka, S.A.,
Nehrkorn, T., Sweeney, C., 2013. Anthropogenic emissions of methane in the United States.
Proceedings of the National Academy of Science, 110, 20018-20022.
Pacific Environment Limited, 2014. AGL fugitive methane emissions monitoring program - technical report.
(http i//www.af*I. corn, au/~/media/ACjI./About%20ACjI./l)ocu men ts/How%2()We%7()Source%20t:ner
tit, accessed 15 February 2014).
Petron, G., Frost, G., Miller, B. R., Hirsch, A. I., Montzka, S. A., Karion, A., Trainer, M., Sweeney, C., Andrews,
A. E., Miller, L., Kofler, J., Bar-llan, A., Dlugokencky, E.J., Patrick, L., Moore Jr., C.T., Ryerson, T.B.,
Siso, C., Kolodzey, W., Lang, P.M., Conway, T., Novelli, P., Masarie, K., Hall, B., Guenther, D., Kitzis,
D., Miller, J., Welsh, D., Wolfe, D., Neff, W., Tans, P., 2012. Hydrocarbon emissions characterization
in the Colorado Front Range: A pilot study. Journal of Geophysical Research-Atmospheres, 117.
D04304.
Saddler, H., 2012. Review of literature on international best practice for estimating greenhouse gas
emissions from coal seam gas production. Report prepared for the Department of Climate Change
and Energy Efficiency. Pitt & Sherry.
(h«TS.,,, ,h.r,,,rf;!, f,,,v ail/Sjtes/climatechani»e/files/media/(v1arcih%2C)2C)13/coal -seam-
, accessed 26 February 2014)
Santos, I., Maher, D., 2012. Submission on estimation of fugitive methane from coal seam gas operations.
(httpi//www,cIimatecliani'e.i*oviai.i/sites/cljtTiatechani»e/flles/flles/cIjrnate'
38 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities 71 f 7J_
n of 83 Of 87
-------
, accessed 15
February 2014).
Somerville, W., 2012. Submission on estimation of fugitive methane from coal seam gas operations.
(http://www.climatftchanKft.Kov.au/sitfts/climatechanf5ft/filfts/filfts/clii
, accessed 31 January, 2014).
Tait, D.R., Santos, I. Maher, D.T., Cyronak, T.J., Davis, R.J., 2013. Enrichment of radon and carbon dioxide in
the open atmosphere of an Australian coal seam gas field. Environmental Science and Technology
47, 3099-3104.
USEPA, 1995. Protocol for Equipment Leak Emission Rates. U.S. Environmental Protection Agency,
Washington DC.
Wigley, T., 2011. Coal to gas: the influence of methane leakage. Climatic Change, 108, 601-608.
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production E^iNjj^-l ^
-------
Appendix
Table Al. Results of the downwind traverse measurements for each well. The average results shown for Wells B2,
B7 and C3 were measured on each well pad. All units are g min1.
Al 0.04 0.03 0.11 0.03
A2 - Suspended 0.00 0.00 0.00 0.00
A3 0.04 0.02 0.08 0.02
A4 7.28 2.75 13.42 3.38
A5 0.10 0.08 0.20 0.06
A6 0.05 0.04 0.05 0.01
B1 1.50 0.01 3.60 1.22
B2 43.8 (on pad) 1.09 66.5 22.5
B3 0.07 0.01 0.28 0.08
B4 0.04 0.01 0.22 0.06
B5 0.01 0.01 0.02 0.00
B6 1.66 0.77 3.10 0.74
B7 1.27 (on pad)
B8 1.31 0.10 2.85 0.98
B9 0.83 0.14 2.95 0.81
B10 0.15 0.07 0.28 0.07
Bll 1.79 0.09 3.65 1.07
B12 0.05 0.01 0.12 0.03
B13 0.02 0.01 0.06 0.02
B14 0.61 0.01 3.23 0.98
B15 1.61 0.11 7.78 2.35
CI 8.69 2.73 15.9 4.77
C2 1.10 0.33 2.45 0.66
40 | Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production Facilities ^ ^
-------
C3 28.0 (on pad)
C4 11.8 0.46 34.8 12.4
C5 0.93 0.21 1.82 0.56
C6 1.17 0.07 2.38 0.71
C7 0.54 0.04 0.99 0.35
C8 0.10 0.02 0.27 0.08
C9 0.05 0.01 0.10 0.03
CIO 1.75 0.76 3.52 0.82
Cll 0.07 0.05 0.10 0.02
D1 Abandoned 0.00 0.00 0.00 0.00
D2 0.12 0.03 0.16 0.04
D3 Abandoned 0.00 0.00 0.00 0.00
D4 0.32 0.17 0.57 0.13
D5 1.07 0.11 2.18 0.71
D6 2.52 0.44 5.00 1.42
El 2.17 0.63 4.08 1.19
E2 0.99 0.50 2.17 0.55
E3 0.60 0.22 1.13 0.33
E4 14.8 1.89 56.8 18.8
E5 0.06 0.01 0.19 0.06
Field Measurements of Fugitive Emissions from Equipment and Well Casings in Australian Coal Seam Gas Production gg ^
-------
CONTACT US
t 1300 363 400
+613 9545 2176
e enquiries@csiro.au
w www.csiro.au
YOURCSIRO
Australia is founding its future on
science and innovation. Its national
science agency, CSIRO, is a powerhouse
of ideas, technologies and skills for
building prosperity, growth, health and
sustainability. It serves governments,
industries, business and communities
across the nation.
FOR FURTHER INFORMATION
Energy Technology
Stuart Day
t +612 4960 6052
e stuart.day@csiro.au
w www.csiro.au
------- |