United States
Environmental Protection
Agency
         National Risk Management
         Research Laboratory
         Research Triangle Park NC 27711
Research and Development
         EPA/600/SR-96/080  June 1997

                                                         the
Matthew R. Harrison, Theresa
Cowgill
1. Shires, Jane K. Wessels, and R. Michael
  Gas  Research  Institute (GRI) and
the U.S.  Environmental  Protection
Agency's (EPA's) Office of Research
and Development cofunded a major
study to quantify methane emissions
from U.S. natural gas operations. For
the 1992  base year, total  methane
emissions were estimated  at 314 ±
105       (6.04 ± 2.01  Tg),  which  is
equivalent to  1.4% + 0.5%  of gross
natural  gas production.
  Since 1992,  many  companies  have
participated in  voluntary programs de-
signed to reduce emissions. Methane
emission  reductions  from these pro-
grams are not reflected in the  report.
However,  methane emissions from a
future  incremental increase in gas
     were evaluated.  Depending on
the size of the potential increase  in
sales, estimated emissions would be
between 0.5%  and 1.0% of the incre-
mental increase.
  This  study provides  data  from the
U.S. natural gas industry  needed for
constructing global methane invento-
ries and for determining the relative
impacts of coal, oil,  and natural gas
use on  global warming.  Using this
study's  emissions estimate and some
key assumptions, an  analysis showed
that the impact on warming from the
use of oil and  coal per unit of energy
generated is much larger than that from
the use  of natural gas.
  This study is documented  in  15 vol-
umes. Volume 2 is a technical sum-
mary that includes what was done and
how the measurements and calcula-
tions were performed.
  This Project Summary was developed
by EPA's National Risk Management
Research Laboratory's Air  Pollution
Prevention and Control Division, Re-
         search Triangle Park, NC, to announce
         key findings of the research project that
         is fully documented in 15 volumes com-
         prising a report of the same title (see
         Project Report ordering information at
         back),

         Introduction
           This report summarizes a major study
         conducted by GRI and  EPA to quantify
         methane emissions from U.S. natural gas
         operations.  The goal was to determine
         these emissions to within + 0.5% of natu-
         ral  gas production, starting at the  well-
         head and ending immediately downstream
         of the customer's  meter. The study was
         conducted  because this information  is
         needed to determine if natural gas can be
         used as an integral part of a fuel switch-
         ing  strategy to reduce the potential of glo-
         bal  warming, and  to provide  data for a
         global methane inventory.
           Carbon dioxide (CO2)  contributes nearly
         as  much  to  global warming as all other
         greenhouse gases combined. Since natu-
         ral gas produces much  less CO2 per unit
         of energy when  combusted than either
         coal or oil,  the Intergovernmental Panel
         on Climate Change (IPCC), EPA, and oth-
         ers  have suggested that,  by promoting
         the  increased use of natural gas, global
         warming  could  be reduced. However,
         methane, which is the  major constituent
         of natural gas, is also an important green-
         house gas and, on a weight basis, meth-
         ane is a more potent   greenhouse gas
         than CO2. For this reason, it was impor-
         tant to determine  if emissions from the
         natural gas industry are large enough  to
         substantially reduce or even eliminate the
         advantage that natural  gas  has because
         of its  much lower CO2  emissions during
         combustion.
           This study, like other  efforts to develop
         emission inventories, had to address sev-

-------
eral difficult problems. Most of these prob-
lems were primarily associated with the
size  and diversity of the natural gas in-
dustry  and the number of sources that
must be considered. This industry com-
plexity,  combined with the lack of both
equipment populations and  methods for
estimating emissions, meant that early in
the program,  resources  were devoted  to
developing comprehensive methods  for
estimating  and extrapolating emissions.
This  also  included selecting  an  accuracy
goal  that  could reasonably be  achieved
but was sufficiently  accurate to examine
the fuel switching strategy.
  Considering these issues,  a method  of
approach was developed that
      Accounted for all emission sources;
      Measured and calculated emis-
      sions;
      Extrapolated emissions data; and
      Assessed the accuracy of the final
      estimate.
Method for Estimating
Emissions
  This summary briefly describes the
method used to estimate methane emis-
sions from the natural gas industry.

Accounting for All Emission
Sources
  The natural gas industry (shown in Fig-
ure 1) was  divided  into four segments:
production,  processing, transmission/stor-
age, and distribution. The project estab-
lished  boundaries  for each  industry
segment to specify the equipment included
in the study. The guideline  used for set-
ting the boundary was to include only the
equipment  in each  segment  that is re-
quired for marketing natural  gas.
  To  fully  characterize  the  natural gas
industry  and account  for  all  potential
sources of methane,  the four industry seg-
ments were divided  into facilities, equip-
ment, and  components; and  emission
sources were identified by equipment type,
mode of operation, and type of emission.
Equipment types  included individual de-
vices, such as a pneumatic operator; large
pieces of equipment,  such as a compres-
sors; or a grouping of equipment, such as
an offshore platform.  Modes of operation
are start-up, normal  operations,  mainte-
nance, upsets, and  mishaps.  Emission
types are  fugitive, vented, and combus-
tion.
  For this project each  emission source
was accounted for by carefully examining
the operating mode for  each equipment
category. This  differentiation ensured that
all emission sources  were accounted for
and that all  types of emissions from the
source were considered.  For  example,
compressor  engines can be  a significant
source of  fugitive, vented, and combus-
tion emissions that result from a variety of
operating  modes. During normal opera-
tions,  unburned methane is emitted in the
             Production
                                                                                         Distribution
                                                                                                  Main and
                                                                                                  service pipelines
             Surface
             facilities
                                                               Underground
                                                               storage
                                                               reservoir
                                                                                                Customer meters
                                                C |  Compressor

                                                    Meter

                                                    Pressure regulator
Figure 1. Gas industry flow chart.

-------
engine exhaust, and fugitive emissions can
result from leaks in valves and pressur-
ized connections. Also,  natural  gas is
vented during  engine  start-ups  if  natural
gas is  used to  power the  starter turbine.
During upsets, natural gas is released from
compressor blowdown and pressure relief
valves, and natural gas is vented during
compressor blowdown for maintenance.

Measuring      Calculating
Emissions
  Initially, few methods were available for
measuring and/or calculating  emissions
from natural gas facilities. Therefore, the
early stages of this study were spent de-
veloping   measurement techniques  and
demonstrating them in the field before us-
ing  them to gather data for the study. On
the  basis  of these proof-of-concept tests,
three  measurement methods were even-
tually  chosen for use  in this study. For
pipeline leaks, the emission rate was mea-
sured by  isolating the  leaking  section of
pipe and  measuring  the amount  of gas
needed to maintain operating  pressure in
the  line.  For fugitive  leaks from  above-
ground facilities, either a tracer gas method
or a component emission factor approach
was used.
  For the tracer gas method, a tracer gas
such as sulfur hexafluoride is  released at
a known  constant rate  near the methane
source. The emission rate was determined
by  measuring   the concentration  of the
tracer and methane downwind; since the
ratio of emission rates  is equal to the ratio
of concentrations, the methane emission
rate can be calculated.
  The component emission measurement
approach develops average emission rates
for the basic components (valves, flanges,
seals,  and other  pipe fittings) that  com-
prise natural gas facilities. The total emis-
sions  from the  facility are the product of
the  number of components times the cor-
responding emission factor.
  New component emission factors were
developed  as  a result of this study for
natural gas production and processing fa-
cilities, compressor stations, and residen-
tial  and commercial meters. Also  a new
"Hi-Flow"  instrument was developed that
can measure emissions quickly and accu-
rately  from pneumatic  control devices,
valves, flanges, and other pipe fittings.
  In some cases  it is  more accurate and
less complicated to calculate,  rather than
measure, emissions. An example is emis-
sions  from a "blowdown" to make a pipe
repair.  Knowing the temperature and pres-
sure of the gas, the volume of the  pipe,
and the frequency of the event, emissions
can be calculated. Another reason  for cal-
culating emissions is  that it may not be
practical to measure emissions from some
sources.  Since annual  emissions  are
needed for the study,  it is not practical to
try to measure highly variable,  unsteady
emissions. In developing engineering mod-
els  for calculating these types of emis-
sions,  it is necessary to  first understand
the  equipment and the nature of the  pro-
cess causing the emissions and then to
collect field  data on the  frequency of the
event.


  A considerable amount of field data  was
collected during this study.  In addition to
measuring emissions and collecting infor-
mation  on  operating characteristics of
equipment and frequency of events, a sub-
stantial effort was required to collect infor-
mation  on  equipment  populations.
Equipment counts are needed to extrapo-
late measured and  calculated emissions
to other similar sources in the industry.
  Data  were collected  on  each  source
category identified during initial  stages of
the project. However, because of the large
number of sources  in each source  cat-
egory,  data were collected on a relatively
small  percentage of all  sources in each
category. Therefore, these data  had to be
extrapolated to account  for the sources
that were not measured  in  order to  de-
velop a national emissions estimate. To
extrapolate the emission data,  emission
and activity factors were defined  so  that
their product equals  the annual  nation-
wide emissions from a given source  cat-
egory.  Typically,  the  emission  factor is
defined as the average annual emissions
from a piece of equipment  or event.  The
activity factor would then be the national
population (i.e., the total  equipment count
or total number of events). For example, if
fugitive  emissions from  compressor  en-
gines is the source category, then aver-
age emissions  per engine  would be  the
emission factor,  and  the number of  en-
gines would be the activity factor.
  Although  this  approach  is straightfor-
ward, the application proved to be difficult
due to the  lack  of  data on equipment
populations  and  operational events.  Lim-
ited information is available on a national
basis. Collecting data on activity factors,
e.g., number of separators, pneumatic con-
trol  devices,  miles  of  gathering lines,
blowdown events, required  a large num-
ber of site visits and was therefore a  ma-
jor part of the study.

             Accuracy
  The accuracy of the emissions estimate
depends on the precision  and bias of both
the activity and emission factors. In devel-
oping activity  factors,  as in  conducting
emission measurements, care was taken
in  developing sampling protocols,  detect-
ing and eliminating  bias, and developing
methods for calculating precision.
  The accuracy goal of the project was to
determine emissions from the natural gas
industry to within + 0.5% of gross  natural
gas production. This goal was established
based  on  the  accuracy needed for con-
structing emission  inventories for use in
global climate change  models and for as-
sessing the  validity of the proposed fuel
switching strategy.
  The first step in achieving the accuracy
goal was to  develop accuracy targets for
each source category. Accuracy  targets
were assigned so that a higher degree of
accuracy would be required for the largest
sources while  achieving the overall pro-
gram goal. This had the additional  advan-
tage of automatically  assigning more
program resources to  the most important
source categories.
  Accuracy  is  made up of precision and
bias. Precision can  be  calculated, but bias
can only be  minimized. To minimize bias,
a sampling approach similar to dispropor-
tionate  stratified random sampling was
developed. A project review committee was
established and industry advisory  groups
were formed for production, transmission,
and distribution to review the program and
ensure  that  any potential for bias was
identified  and  eliminated. Also  the data
were analyzed to ensure that data  were
not sampled  disproportionately with  respect
to  the parameters that had a large impact
on emissions. This not  only minimized bias
but also reduced the impact that outlying
data points had on the result. The preci-
sion  of  the activity and emission  factors
was calculated for a 90% confidence level
from the number of data points collected
and the standard deviation. The precision
of the emission estimate for each  source
category as  well as the  national estimate
was also calculated in  a statistically rigor-
ous fashion.

Results

1992
  Total  methane emissions from the natu-
ral gas industry for the  1992 baseline year
are 314 +  105  Bscf (6.04 + 2.01  Tg). This
is  approximately 1.4% + 0.5% of gross
natural gas production, a result that meets
the project accuracy goal. This represents
approximately 19% of total U.S. anthropo-
genic emissions,  based on methane emis-
sion  estimates reported by the EPA for
major anthropogenic sources (see Figure
2).

-------
    Domesticated Livestock 19%
  Coal Mining 15%
                                                               Other 6%
                                                    .; ;.S"-,'\ Livestock Manure 9%
Figure 2. Contribution of major methane sources to total U.S. anthropogenic emissions.
  Figure 3 presents  methane emissions
for the  natural  gas industry by industry
segment. The transmission/storage seg-
ment  accounts  for the  largest portion  of
emissions (37%) with the processing seg-
ment contributing the least (12%).
  The largest emission sources  for each
industry segment  are presented  in Table
1. Fugitive emissions are the largest con-
tributor to methane emissions from natu-
ral  gas  processing, transmission,  and
storage.  Nearly 90% of these emissions
result from leaks  on compressor compo-
nents such  as the  suction,  discharge,
blowdown, and  pressure regulator  valves
and compressor seals. Fugitive emissions
from all  compressor components are ap-
proximately 80  Bscf (1.6 Tg), while fugi-
tive emissions from all  other compressor
station components, such as yard  piping
and filter-separators, are approximately 10
Bscf (0.19  Tg). Compressor engine  ex-
hausts are  responsible for  slightly more
than 25 Bscf (0.48 Tg) of methane emis-
sions.
  Fugitive emissions from  pipelines  are
approximately 48 Bscf (0.93 Tg), of which
42 Bscf (0.80 Tg) is from distribution pip-
ing.  Distribution  piping systems  actually
emit 51  Bscf (0.98 Tg), but approximately
18% of the  natural gas leaked is oxidized
in  the soil  by  methanotrophs.  Approxi-
mately 22 Bscf (0.42 Tg) is leaked from
cast iron mains that constitute only 6% of
the total length of distribution main pipe-
lines. However, most cast iron leaks  are
very small  and, since the oxidation  rate
varies inversely with leak rate,  only 60%
of the leaks (13 Bscf or 0.25 Tg) reach
the surface.
  The two  largest methane  emission
sources  in   natural  gas  production  are
              Distribution 24%
                                  Transmission/Storage 37%
Figure 3. Summary of methane emissions.
pneumatic control devices and fugitives.
Prior to this study, pneumatic devices were
not considered a major emission  source.
Approximately a third of these  devices
bleed natural gas to the atmosphere con-
tinuously. Pneumatic devices are the larg-
est source of methane emissions in the
production segment, accounting for 31 Bscf
(0.60 Tg). Total fugitive emissions from
production equipment are  large  even
though the  average leak  rate is small,
because  of  the  large number (approxi-
mately 80 million) of valves, connectors,
and  other pipe  fittings on  equipment lo-
cated at production sites across the coun-
try.

Emissions from Incremental
Increases in Gas Sales
  Consumption of  natural  gas  has  in-
creased since the 1992  base year.  To
determine the effect that this increase and
future increases will have on emissions, a
study was conducted to determine  the per-
cent increase in emissions resulting from
an  incremental  increase  in  natural  gas
production and  sales. The study found
that  increases  in  throughput would,  in
many cases,  produce increases in emis-
sions. However, the average increase  in
emissions would be proportionally smaller
than the increase in system throughput.
  The study examined the consequences
of increasing  gas sales by 5%,  15%, and
30% under three scenarios: uniform, win-
ter peak, and summer peak load  profiles.
All  segments of the gas  industry  were
examined  to determine  the percent  in-
crease in equipment that would be needed
to meet the  increased demand. The per-
cent increase in emissions was then esti-
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.
  The  most  realistic scenario  assumed
that the system would be expanded using
the latest technologies, whereas the most
conservative  scenario assumes that the
expanded system mirrors the existing sys-
tem. Generally,  as the system expands,
the emission rate for the expansion would
be less, as a percent of throughput, than
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-
pansion would  be between 30  and  70

-------
Table 1. Largest Emission Sources by Industry Segment
Annual Methane
Emissions
Segment
Production

Processing
Transmission/
Storage

Distribution

Source
Pneumatic devices
Fugitive emissions
Dehydrators
Other
Fugitive emissions
Compressor exhaust
Other
Fugitive emissions
Blow and purge
Pneumatic devices
Compressor exhaust
Other
Underground pipeline leaks
Meter and pressure regulating stations
Customer meters
Other
(Bscf)
31
17
14
21
24
7
5
68
19
14
11
5
42
27
6
2
(Tg)
0.60
0.33
0.28
0.41
0.47
0.13
0.10
1.30
0.36
0.27
0.22
0.10
0.80
0.53
0.11
0.04
Percent of
Segment
Total
37
21
17
25
67
19
14
58
16
12
10
4
54
35
8
3
Total1
                                                     314 Bscf    6.04 Tg
1 Individual sources may not sum exactly to total shown due to roundoff errors.
Bscf. These emissions, when divided by
the incremental production, are equivalent
to an emission rate  between 0.4% and
1.0%  of  incremental  production. This is
much lower than the 1.4%  of production
emitted from the current base system for
1992.
  Emissions are lower for an incremental
increase in gas sales  because the current
system has excess capacity and any ad-
ditional equipment that would  have to be
installed to meet increased demand would
use current and lower emitting technol-
ogy. A few examples of these technolo-
gies are  no-bleed  pneumatic  devices,
turbine compressor engines,  and  plastic
pipe instead of steel and cast  iron mains.

Emissions      Fuel Switching
  The estimate of methane emissions from
natural gas operations was used  in an
analysis to determine if the potential for
global  warming could  be reduced by
switching from coal or oil to natural gas.
Emissions  from  coal and  oil  were  esti-
mated from other sources. Other than CO2
and methane, emissions from other green-
house gases from the fuel cycle of fossil
fuels are negligible. Methane,  however is
a more potent  greenhouse gas than  CO2.
The approach  used was to determine the
emissions  of methane  and CO2 for the
complete fuel cycle of natural gas, oil, and
coal, and  to convert  the  methane  emis-
sions to equivalent CO2 using the Global
Warming Potential (GWP).
  The  GWP is  an  index that relates the
impact of releasing quantities of the vari-
ous greenhouse gases to the release of
an amount of CO2 that would produce the
same impact on global warming. Currently,
there is a great deal  of uncertainty in the
time period associated with the GWP of
methane. Typical time periods range from
50 to 500 years, which correspond to GWP
values  of 34 and 6.5,  respectively. This
means that 1 Ib of  methane is equivalent
to between 6.5 and 30 Ib of CO2.
  Equivalent CO2 emissions from the fuel
cycle of natural  gas were calculated to be
132  lb/106 Btu (57  kg/GJ) for a GWP of
6.5 and 152 lb/106 Btu  (66 kg/MJ) for a
GWP of 34.  Even for a GWP of 34, the
analysis showed that, compared to natu-
ral gas, oil  has 1.2 times the impact on
global  warming and  coal has 1.5 times
the impact.

Conclusions
  Based on data collected, methane emis-
sions from natural gas operations are es-
timated to be 314 + 105 Bscf (6.04 + 2.01
Tg) for the  1992 baseline  year. This is
approximately 1.4% + 0.5% of gross natu-
ral gas production.  This  study also deter-
mined  that the percentage  of  methane
emitted per  gas production rate  for an
incremental increase  in natural gas sales
would  be between  1.19% and 1.38% of
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
the fuel cycle for natural gas, oil, and coal
using the GWPs recently published by the
IPCC.  The analysis showed that natural
gas contributes significantly less to global
warming per unit  of energy than coal  or
oil, which supports the fuel switching strat-
egy suggested by IPCC and others.
  This study, like  other efforts in develop-
ing emission inventories,  had to  address
the following typical but nevertheless diffi-
cult problems:
      Collecting demographic information;
      Developing methods for measuring
      and calculating  emissions;
      Extrapolating  a  limited amount  of
      data to  a  large, diverse  national
      population;  and
      Determining the accuracy of the fi-
      nal estimates.
  The  most difficult of these is evaluating
the accuracy. Accuracy targets were es-
tablished for each source category that
would  be needed to achieve the overall
accuracy goal  of  the  study. A sampling
procedure with  checks for bias was  then
established, data were collected,  and the
precision of the emission estimate  was
rigorously calculated  for  each category,
as well as for the  national estimate.
  During the course of the study, equip-
ment population in the gas industry was
collected and new methods were devel-
oped for measuring emissions from a va-
riety  of sources.  Unique methods  were
developed  using  tracer gas techniques,
and  a  new "Hi-Flow" instrument was de-
veloped that provides  a quick, cost-effec-
tive method for measuring the leak rate of
valves, seals, pneumatic devices, and con-
nectors.
  In  addition, results from this study are
being used by  the natural gas industry to
reduce  operating costs  while  reducing
emissions. Some companies are also par-
ticipating in the Natural Gas-Star program,
a voluntary program sponsored by EPA's
Office of Air and Radiation in cooperation
with  the American  Gas  Association  to
implement cost-effective emission reduc-
tions and to report reductions to the EPA.
Since this program was begun  after the
1992 baseline  year,  any reductions  in
methane emissions from this program are
not reflected  in this  study's total emis-
sions.
  In  conclusion, the project reached  its
accuracy goal  and provides an accurate
estimate of methane emissions for 1992
gas industry practices. The results can  be
used to construct U.S. methane  invento-
ries and analyze fuel switching strategies.

-------
  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
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati,  OH 45268
      BULK RATE
POSTAGE & FEES PAID
         EPA
   PERMIT No. G-35
Official Business
Penalty for Private Use
$300
EPA/600/SR-96/080

-------