United States National Risk Management

Environmental Protection Research Laboratory

Agency Research Triangle Park NC 27711
Research and Development EPA/600/SR-96/080 June 1997

Project Summary

Methane Emissions from the

Natural Gas Industry
Matthew R. Harrison, Theresa M. Shires, Jane K. Wessels, and R. Michael
Cowgill

Gas Research Institute (GRI) and search Triangle Park, NC, to announce
the U.S. Environmental Protection key findings of the research project that
Agency’s (EPA’s) Office of Research is fully documented in 15 volumes com-
and Development cofunded a major prising a report of the same title (see
study to quantify methane emissions Project Report ordering information at
from U.S. natural gas operations. For back).
the 1992 base year, total methane
emissions were estimated at 314 ± Introduction
105 Bscf (6.04 ± 2.01 Tg), which is This report summarizes a major study
equivalent to 1.4% ± 0.5% of gross conducted by GRI and EPA to quantify
natural gas production. methane emissions from U.S. natural gas
Since 1992, many companies have operations. The goal was to determine
participated in voluntary programs de- these emissions to within ± 0.5% of natu-
signed to reduce emissions. Methane ral gas production, starting at the well-
emission reductions from these pro- head and ending immediately downstream
grams are not reflected in the report. of the customer’s meter. The study was
However, methane emissions from a conducted because this information is
future incremental increase in gas needed to determine if natural gas can be
sales were evaluated. Depending on used as an integral part of a fuel switch-
the size of the potential increase in ing strategy to reduce the potential of glo-
sales, estimated emissions would be bal warming, and to provide data for a
between 0.5% and 1.0% of the incre- global methane inventory.
mental increase. Carbon dioxide (CO2) contributes nearly
This study provides data from the as much to global warming as all other
U.S. natural gas industry needed for greenhouse gases combined. Since natu-
constructing global methane invento- ral gas produces much less CO2 per unit
ries and for determining the relative of energy when combusted than either
impacts of coal, oil, and natural gas coal or oil, the Intergovernmental Panel
use on global warming. Using this on Climate Change (IPCC), EPA, and oth-
study’s emissions estimate and some ers have suggested that, by promoting
key assumptions, an analysis showed the increased use of natural gas, global
that the impact on warming from the warming could be reduced. However,
use of oil and coal per unit of energy methane, which is the major constituent
generated is much larger than that from of natural gas, is also an important green-
the use of natural gas. house gas and, on a weight basis, meth-
This study is documented in 15 vol- ane is a more potent greenhouse gas
umes. Volume 2 is a technical sum- than CO2. For this reason, it was impor-
mary that includes what was done and tant to determine if emissions from the
how the measurements and calcula- natural gas industry are large enough to
tions were performed. substantially reduce or even eliminate the
This Project Summary was developed advantage that natural gas has because
by EPA’s National Risk Management of its much lower CO2 emissions during
Research Laboratory’s Air Pollution combustion.
Prevention and Control Division, Re- This study, like other efforts to develop
emission inventories, had to address sev-
eral difficult problems. Most of these prob- Method for Estimating sources were identified by equipment type,
lems were primarily associated with the Emissions mode of operation, and type of emission.
size and diversity of the natural gas in- This summary briefly describes the Equipment types included individual de-
dustry and the number of sources that method used to estimate methane emis- vices, such as a pneumatic operator; large
must be considered. This industry com- sions from the natural gas industry. pieces of equipment, such as a compres-
plexity, combined with the lack of both sors; or a grouping of equipment, such as
equipment populations and methods for Accounting for All Emission an offshore platform. Modes of operation
estimating emissions, meant that early in Sources are start-up, normal operations, mainte-
the program, resources were devoted to The natural gas industry (shown in Fig- nance, upsets, and mishaps. Emission
developing comprehensive methods for ure 1) was divided into four segments: types are fugitive, vented, and combus-
estimating and extrapolating emissions. production, processing, transmission/stor- tion.
This also included selecting an accuracy age, and distribution. The project estab- For this project each emission source
goal that could reasonably be achieved lished boundaries for each industry was accounted for by carefully examining
but was sufficiently accurate to examine segment to specify the equipment included the operating mode for each equipment
the fuel switching strategy. in the study. The guideline used for set- category. This differentiation ensured that
Considering these issues, a method of ting the boundary was to include only the all emission sources were accounted for
approach was developed that equipment in each segment that is re- and that all types of emissions from the
• Accounted for all emission sources; quired for marketing natural gas. source were considered. For example,
• Measured and calculated emis- To fully characterize the natural gas compressor engines can be a significant
sions; industry and account for all potential source of fugitive, vented, and combus-
• Extrapolated emissions data; and sources of methane, the four industry seg- tion emissions that result from a variety of
• Assessed the accuracy of the final ments were divided into facilities, equip- operating modes. During normal opera-
estimate. ment, and components; and emission tions, unburned methane is emitted in the

Production Processing Transmission/Storage Distribution

Main and
Direct service pipelines
sales
Compressor M
stations M&PR stations
M M
C M
Pipelines
Surface Gas plant M
facilities M
Gas

M
Pipelines M
C M
Liquids
M
Storage M
Liquids Customer meters
LNG C

Underground
storage C Compressor
reservoir Meter
M
Pressure regulator

Figure 1. Gas industry flow chart.

2
engine exhaust, and fugitive emissions can culating emissions is that it may not be oping activity factors, as in conducting
result from leaks in valves and pressur- practical to measure emissions from some emission measurements, care was taken
ized connections. Also, natural gas is sources. Since annual emissions are in developing sampling protocols, detect-
vented during engine start-ups if natural needed for the study, it is not practical to ing and eliminating bias, and developing
gas is used to power the starter turbine. try to measure highly variable, unsteady methods for calculating precision.
During upsets, natural gas is released from emissions. In developing engineering mod- The accuracy goal of the project was to
compressor blowdown and pressure relief els for calculating these types of emis- determine emissions from the natural gas
valves, and natural gas is vented during sions, it is necessary to first understand industry to within ± 0.5% of gross natural
compressor blowdown for maintenance. the equipment and the nature of the pro- gas production. This goal was established
cess causing the emissions and then to based on the accuracy needed for con-
Measuring and Calculating collect field data on the frequency of the structing emission inventories for use in
Emissions event. global climate change models and for as-
Initially, few methods were available for sessing the validity of the proposed fuel
measuring and/or calculating emissions Extrapolating Emissions switching strategy.
from natural gas facilities. Therefore, the A considerable amount of field data was The first step in achieving the accuracy
early stages of this study were spent de- collected during this study. In addition to goal was to develop accuracy targets for
veloping measurement techniques and measuring emissions and collecting infor- each source category. Accuracy targets
demonstrating them in the field before us- mation on operating characteristics of were assigned so that a higher degree of
ing them to gather data for the study. On equipment and frequency of events, a sub- accuracy would be required for the largest
the basis of these proof-of-concept tests, stantial effort was required to collect infor- sources while achieving the overall pro-
three measurement methods were even- mation on equipment populations. gram goal. This had the additional advan-
tually chosen for use in this study. For Equipment counts are needed to extrapo- tage of automatically assigning more
pipeline leaks, the emission rate was mea- late measured and calculated emissions program resources to the most important
sured by isolating the leaking section of to other similar sources in the industry. source categories.
pipe and measuring the amount of gas Data were collected on each source Accuracy is made up of precision and
needed to maintain operating pressure in category identified during initial stages of bias. Precision can be calculated, but bias
the line. For fugitive leaks from above- the project. However, because of the large can only be minimized. To minimize bias,
ground facilities, either a tracer gas method number of sources in each source cat- a sampling approach similar to dispropor-
or a component emission factor approach egory, data were collected on a relatively tionate stratified random sampling was
was used. small percentage of all sources in each developed. A project review committee was
For the tracer gas method, a tracer gas category. Therefore, these data had to be established and industry advisory groups
such as sulfur hexafluoride is released at extrapolated to account for the sources were formed for production, transmission,
a known constant rate near the methane that were not measured in order to de- and distribution to review the program and
source. The emission rate was determined velop a national emissions estimate. To ensure that any potential for bias was
by measuring the concentration of the extrapolate the emission data, emission identified and eliminated. Also the data
tracer and methane downwind; since the and activity factors were defined so that were analyzed to ensure that data were
ratio of emission rates is equal to the ratio their product equals the annual nation- not sampled disproportionately with respect
of concentrations, the methane emission wide emissions from a given source cat- to the parameters that had a large impact
rate can be calculated. egory. Typically, the emission factor is on emissions. This not only minimized bias
The component emission measurement defined as the average annual emissions but also reduced the impact that outlying
approach develops average emission rates from a piece of equipment or event. The data points had on the result. The preci-
for the basic components (valves, flanges, activity factor would then be the national sion of the activity and emission factors
seals, and other pipe fittings) that com- population (i.e., the total equipment count was calculated for a 90% confidence level
prise natural gas facilities. The total emis- or total number of events). For example, if from the number of data points collected
sions from the facility are the product of fugitive emissions from compressor en- and the standard deviation. The precision
the number of components times the cor- gines is the source category, then aver- of the emission estimate for each source
responding emission factor. age emissions per engine would be the category as well as the national estimate
New component emission factors were emission factor, and the number of en- was also calculated in a statistically rigor-
developed as a result of this study for gines would be the activity factor. ous fashion.
natural gas production and processing fa- Although this approach is straightfor-
cilities, compressor stations, and residen- ward, the application proved to be difficult Results
tial and commercial meters. Also a new due to the lack of data on equipment
populations and operational events. Lim- 1992 Baseline Emissions
“Hi-Flow” instrument was developed that Total methane emissions from the natu-
can measure emissions quickly and accu- ited information is available on a national
basis. Collecting data on activity factors, ral gas industry for the 1992 baseline year
rately from pneumatic control devices, are 314 ± 105 Bscf (6.04 ± 2.01 Tg). This
valves, flanges, and other pipe fittings. e.g., number of separators, pneumatic con-
trol devices, miles of gathering lines, is approximately 1.4% ± 0.5% of gross
In some cases it is more accurate and natural gas production, a result that meets
less complicated to calculate, rather than blowdown events, required a large num-
ber of site visits and was therefore a ma- the project accuracy goal. This represents
measure, emissions. An example is emis- approximately 19% of total U.S. anthropo-
sions from a “blowdown” to make a pipe jor part of the study.
genic emissions, based on methane emis-
repair. Knowing the temperature and pres- Assessing Accuracy sion estimates reported by the EPA for
sure of the gas, the volume of the pipe, The accuracy of the emissions estimate major anthropogenic sources (see Figure
and the frequency of the event, emissions depends on the precision and bias of both 2).
can be calculated. Another reason for cal- the activity and emission factors. In devel-

3
 pneumatic control devices and fugitives.
Prior to this study, pneumatic devices were
not considered a major emission source.
Landfills 31% Approximately a third of these devices
bleed natural gas to the atmosphere con-
Domesticated Livestock 19% tinuously. Pneumatic devices are the larg-
est source of methane emissions in the
production segment, accounting for 31 Bscf
Other 6% (0.60 Tg). Total fugitive emissions from
Coal Mining 15% production equipment are large even
though the average leak rate is small,
because of the large number (approxi-
Livestock Manure 9%
Natural Gas Systems 20% mately 80 million) of valves, connectors,
and other pipe fittings on equipment lo-
cated at production sites across the coun-
try.
Emissions from Incremental
Figure 2. Contribution of major methane sources to total U.S. anthropogenic emissions. Increases in Gas Sales
Consumption of natural gas has in-
creased since the 1992 base year. To
Figure 3 presents methane emissions Bscf (0.19 Tg). Compressor engine ex- determine the effect that this increase and
for the natural gas industry by industry hausts are responsible for slightly more future increases will have on emissions, a
segment. The transmission/storage seg- than 25 Bscf (0.48 Tg) of methane emis- study was conducted to determine the per-
ment accounts for the largest portion of sions. cent increase in emissions resulting from
emissions (37%) with the processing seg- Fugitive emissions from pipelines are an incremental increase in natural gas
ment contributing the least (12%). approximately 48 Bscf (0.93 Tg), of which production and sales. The study found
The largest emission sources for each 42 Bscf (0.80 Tg) is from distribution pip- that increases in throughput would, in
industry segment are presented in Table ing. Distribution piping systems actually many cases, produce increases in emis-
1. Fugitive emissions are the largest con- emit 51 Bscf (0.98 Tg), but approximately sions. However, the average increase in
tributor to methane emissions from natu- 18% of the natural gas leaked is oxidized emissions would be proportionally smaller
ral gas processing, transmission, and in the soil by methanotrophs. Approxi- than the increase in system throughput.
storage. Nearly 90% of these emissions mately 22 Bscf (0.42 Tg) is leaked from The study examined the consequences
result from leaks on compressor compo- cast iron mains that constitute only 6% of of increasing gas sales by 5%, 15%, and
nents such as the suction, discharge, the total length of distribution main pipe- 30% under three scenarios: uniform, win-
blowdown, and pressure regulator valves lines. However, most cast iron leaks are ter peak, and summer peak load profiles.
and compressor seals. Fugitive emissions very small and, since the oxidation rate All segments of the gas industry were
from all compressor components are ap- varies inversely with leak rate, only 60% examined to determine the percent in-
proximately 80 Bscf (1.6 Tg), while fugi- of the leaks (13 Bscf or 0.25 Tg) reach crease in equipment that would be needed
tive emissions from all other compressor the surface. to meet the increased demand. The per-
station components, such as yard piping The two largest methane emission cent increase in emissions was then esti-
and filter-separators, are approximately 10 sources in natural gas production are mated based on changes in the current
system that would be required to accom-
modate the increase in gas sales. The
GRI/EPA emission estimate was used to
calculate the percent increase in emis-
sions that would result from an incremen-
tal increase in natural gas sales for several
scenarios examined in the study.
Transmission/Storage 37% The most realistic scenario assumed
that the system would be expanded using
the latest technologies, whereas the most
Distribution 24% conservative scenario assumes that the
expanded system mirrors the existing sys-
Processing 12% tem. Generally, as the system expands,
the emission rate for the expansion would
be less, as a percent of throughput, than
Production 27% for the base system. Emissions from a
system load increase (an increase in con-
sumption of gas) of 30% would emit at
only one- to two-thirds of the base emis-
sion rate. For example, if gas production
increased by 30% (6 to 7 trillion cubic feet
per year), emissions from the system ex-
Figure 3. Summary of methane emissions. pansion would be between 30 and 70

4
Table 1. Largest Emission Sources by Industry Segment the total gas production, compared to 1.4%
of production for the baseline case.
Results from this study were used to
compare greenhouse gas emissions from
Annual Methane Percent of
the fuel cycle for natural gas, oil, and coal
Emissions Segment
Segment Source (Bscf) (Tg) Total using the GWPs recently published by the
IPCC. The analysis showed that natural
Production Pneumatic devices 31 0.60 37
gas contributes significantly less to global
Fugitive emissions 17 0.33 21 warming per unit of energy than coal or
Dehydrators 14 0.28 17 oil, which supports the fuel switching strat-
Other 21 0.41 25 egy suggested by IPCC and others.
This study, like other efforts in develop-
Processing Fugitive emissions 24 0.47 67 ing emission inventories, had to address
Compressor exhaust 7 0.13 19 the following typical but nevertheless diffi-
Other 5 0.10 14 cult problems:
• Collecting demographic information;
Transmission/ Fugitive emissions 68 1.30 58 • Developing methods for measuring
Storage Blow and purge 19 0.36 16 and calculating emissions;
Pneumatic devices 14 0.27 12 • Extrapolating a limited amount of
Compressor exhaust 11 0.22 10
data to a large, diverse national
Other 5 0.10 4
population; and
• Determining the accuracy of the fi-
Distribution Underground pipeline leaks 42 0.80 54
Meter and pressure regulating stations 27 0.53 35 nal estimates.
Customer meters 6 0.11 8 The most difficult of these is evaluating
Other 2 0.04 3 the accuracy. Accuracy targets were es-
tablished for each source category that
Total1 314 Bscf 6.04 Tg would be needed to achieve the overall
accuracy goal of the study. A sampling
1 Individual sources may not sum exactly to total shown due to roundoff errors. procedure with checks for bias was then
established, data were collected, and the
Bscf. These emissions, when divided by sions to equivalent CO2 using the Global precision of the emission estimate was
the incremental production, are equivalent Warming Potential (GWP). rigorously calculated for each category,
to an emission rate between 0.4% and The GWP is an index that relates the as well as for the national estimate.
1.0% of incremental production. This is impact of releasing quantities of the vari- During the course of the study, equip-
much lower than the 1.4% of production ous greenhouse gases to the release of ment population in the gas industry was
emitted from the current base system for an amount of CO2 that would produce the collected and new methods were devel-
1992. same impact on global warming. Currently, oped for measuring emissions from a va-
Emissions are lower for an incremental there is a great deal of uncertainty in the riety of sources. Unique methods were
increase in gas sales because the current time period associated with the GWP of developed using tracer gas techniques,
system has excess capacity and any ad- methane. Typical time periods range from and a new “Hi-Flow” instrument was de-
ditional equipment that would have to be 50 to 500 years, which correspond to GWP veloped that provides a quick, cost-effec-
installed to meet increased demand would values of 34 and 6.5, respectively. This tive method for measuring the leak rate of
use current and lower emitting technol- means that 1 lb of methane is equivalent valves, seals, pneumatic devices, and con-
ogy. A few examples of these technolo- to between 6.5 and 30 lb of CO2. nectors.
gies are no-bleed pneumatic devices, Equivalent CO2 emissions from the fuel In addition, results from this study are
turbine compressor engines, and plastic cycle of natural gas were calculated to be being used by the natural gas industry to
pipe instead of steel and cast iron mains. 132 lb/106 Btu (57 kg/GJ) for a GWP of reduce operating costs while reducing
6.5 and 152 lb/106 Btu (66 kg/MJ) for a emissions. Some companies are also par-
Emissions and Fuel Switching GWP of 34. Even for a GWP of 34, the ticipating in the Natural Gas-Star program,
The estimate of methane emissions from analysis showed that, compared to natu- a voluntary program sponsored by EPA’s
natural gas operations was used in an ral gas, oil has 1.2 times the impact on Office of Air and Radiation in cooperation
analysis to determine if the potential for global warming and coal has 1.5 times with the American Gas Association to
global warming could be reduced by the impact. implement cost-effective emission reduc-
switching from coal or oil to natural gas.
tions and to report reductions to the EPA.
Emissions from coal and oil were esti- Conclusions Since this program was begun after the
mated from other sources. Other than CO2 Based on data collected, methane emis- 1992 baseline year, any reductions in
and methane, emissions from other green- sions from natural gas operations are es-
methane emissions from this program are
house gases from the fuel cycle of fossil timated to be 314 ± 105 Bscf (6.04 ± 2.01
not reflected in this study’s total emis-
fuels are negligible. Methane, however is Tg) for the 1992 baseline year. This is sions.
a more potent greenhouse gas than CO2. approximately 1.4% ± 0.5% of gross natu-
In conclusion, the project reached its
The approach used was to determine the ral gas production. This study also deter-
accuracy goal and provides an accurate
emissions of methane and CO2 for the mined that the percentage of methane estimate of methane emissions for 1992
complete fuel cycle of natural gas, oil, and emitted per gas production rate for an
gas industry practices. The results can be
coal, and to convert the methane emis- incremental increase in natural gas sales
used to construct U.S. methane invento-
would be between 1.19% and 1.38% of ries and analyze fuel switching strategies.

5
 Matthew R. Harrison, Theresa M. Shires, Jane K. Wessels, and R. Michael Cowgill are
with Radian International LLC, Austin, TX 78720-1088.
David A. Kirchgessner is the EPA Project Officer (see below).
The complete report, consisting of 15 volumes, is titled “Methane Emissions from the
Natural Gas Industry” and has the following order numbers and costs:
The set (Order No. PB97-142913; Cost: $331.00, subject to change)
Volume 1: Executive Summary (Order No. PB97-142921; Cost: $19.50, subject to
change)
Volume 2: Technical Report (Order No. PB97-142939; Cost: $31.00, subject to change)
Volume 3: General Methodology (Order No. PB97-142947; Cost: $38.00, subject to
change)
Volume 4: Statistical Methodology (Order No. PB97-142954; Cost: $31.00, subject to
change)
Volume 5: Activity Factors (Order No. PB97-142962; Cost: $31.00, subject to change)
Volume 6: Vented and Combustion Source Summary (Order No. PB97-142970; Cost:
$21.50, subject to change)
Volume 7: Blow and Purge Activities (Order No. PB97-142988; Cost: $25.00, subject to
change)
Volume 8: Equipment Leaks (Order No. PB97-142996; Cost: $31.00, subject to change)
Volume 9: Underground Pipelines (Order No. PB97-143002; Cost: $25.00, subject to
change)
Volume 10: Metering and Pressure Regulating Stations, Etc. (Order No. PB97-143010;
Cost: $21.50, subject to change)
Volume 11: Compressor Driver Exhaust (Order No. PB97-143028; Cost: $25.00, subject
to change)
Volume 12: Pneumatic Devices (Order No. PB97-143036; Cost: $25.00, subject to
change)
Volume 13: Chemical Injection Pumps (Order No. PB97-143044; Cost: $21.50, subject
to change)
Volume 14: Glycol Dehydrators (Order No. PB97-143051; Cost: $21.50, subject to
change)
Volume 15: Gas-Assisted Glycol Pumps (Order No. PB97-143069; Cost: $21.50, subject
to change)
These volumes will be available only from
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711

BULK RATE
United States POSTAGE & FEES PAID
Environmental Protection Agency EPA
Center for Environmental Research Information PERMIT No. G-35
Cincinnati, OH 45268

Official Business
Penalty for Private Use
$300

EPA/600/SR-96/080