Professional Documents
Culture Documents
pubs.acs.org/est
Measuring Emissions from Oil and Natural Gas Well Pads Using the
Mobile Flux Plane Technique
Chris W. Rella,* Tracy R. Tsai, Connor G. Botkin, Eric R. Crosson, and David Steele
Picarro, Inc., 3105 Patrick Henry Drive, Santa Clara, California 95054, United States
*
S Supporting Information
ABSTRACT: We present a study of methane emissions from oil and gas producing
well pad facilities in the Barnett Shale region of Texas, measured using an innovative
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
ground-based mobile flux plane (MFP) measurement system, as part of the Barnett
Coordinated Campaign.1 Using only public roads, we measured the emissions from
nearly 200 well pads over 2 weeks in October 2013. The population of measured well
pads is split into well pads with detectable emissions (N = 115) and those with
Downloaded via PEKING UNIV on March 23, 2024 at 12:07:38 (UTC).
emissions below the detection limit of the MFP instrument (N = 67). For those well
pads with nonzero emissions, the distribution was highly skewed, with a geometric mean
of 0.63 kg/h, a geometric standard deviation of 4.2, and an arithmetic mean of 1.72 kg/
h. Including the population of nonemitting well pads, we find that the arithmetic mean
of the well pads sampled in this study is 1.1 kg/h. This distribution implies that 50% of
the emissions is due to the 6.6% highest emitting well pads, and 80% of the emissions is
from the 22% highest emitting well pads.
■ INTRODUCTION
Methane, the primary constituent of natural gas, is a potent
emissions. Finally, fugitive emissions are generally unintentional
emissions, due either to human error or a process or
greenhouse gas with a global warming potential of 28−86 times component failure, making any attempt at accurate quantifica-
that of an equivalent mass of carbon dioxide (Myhre et al.2). tion via an emissions factor calculation difficult.
When fugitive emissions are kept under control, methane Recent studies have made direct emissions measurements of
represents a clean burning, high energy content fuel that can well pads to in an effort to provide quantitative estimates of
reduce carbon dioxide emissions relative to other more carbon- well pad emissions nationwide. In Allen et al.,4 measurements
rich fuels. However, when methane emissions are a significant of well pad emissions were made using two methods. 150
fraction of natural gas production, the climate benefit of production well pads were studied using the primary
methane is reduced or eliminated (Howarth et al.;3 Alvarez et quantification method, called direct source measurement, in
al.4). which leaks are individually identified using an infrared camera,
Natural gas production, transportation, and distribution and the methane emissions of all detected leaks were quantified
encompass an extensive nationwide infrastructure involving using a Hi-Flow Sampler. Because not all leaks can be found
hundreds of thousands of wells and millions of miles of with an infrared camera, the direct source measurement
pipeline, as well as processing and storage facilities. Estimating method tends to provide a lower limit on total emissions.
the fugitive methane emissions from such a complex and The secondary method of dual tracer release was performed on
heterogeneous system is highly challenging. Recognizing this a small subset (20) of the well pads. In the tracer release
challenge and the importance of reducing the uncertainty of method (Shorter et al.,8 Czepiel et al.9), tracer gases (in this
fugitive emissions estimates, several recent studies (Allen et al.,5 instance, nitrous oxide, and acetylene) are released from
Eastern Research Group6) have focused on quantifying fugitive compressed gas cylinders at a controlled rate in the vicinity of
emissions at various stages of the process, from well-drilling to potential sources on the well pad. The advantage of this
well pad production, through processing, storage, transport, and methodology is that total well pad emissions are measured,
distribution. including small leaks below the detection limit of infrared
Among the different source categories, fugitive emissions cameras and physically inaccessible sources.
from production well pads are poorly understood. The roughly Even with this limited sample set, the emissions from the
half million natural gas wells in the United States (Energy well pads show a great deal of variability. Out of 19 well pads
Information Administration7) are distributed geographically in where a clear plume was identified, the maximum emissions
dozens of distinct basins, each with unique geophysics and
resulting natural gas component mixtures that require different Received: December 19, 2014
production and processing equipment. Furthermore, the Revised: March 14, 2015
regulatory burden for emissions reduction is different from Accepted: March 16, 2015
state to state, leading to different practices and resulting Published: March 25, 2015
observed was 4.16 SCFM and the minimum was 0.08 SCFM, and u(z) = component of wind speed normal to the surface A
with a range of a factor of 50 between the two. The distribution as a function of height above the ground.
is clearly non-Gaussian, with relatively few emitters responsible The biggest challenge in implementing a surface flux integral
for the bulk of the emissions: the top 4 emitters (20%) account of a plume is in retrieving the concentration measurements on
for 55% of the emissions. As with any skewed population, the surface that intersects the downwind plume. Thoma et al.11
accurate assessment of the aggregated emissions requires demonstrated quantification of emissions from landfills using
sufficient sampling of the entire population, especially the open path measurements to obtain the concentration map.
relatively rare tail of strongly emitting sources. As the Closed path instruments are much simpler to deploy, but only
distribution becomes more skewed, larger sample sets (on the provide measurements at a single point in space at a given time.
order of hundreds) involving more well pads are required. Making measurements on a surface with closed path
Quantifying emissions from well pads presents a significant instrumentation has meant either using multiple instruments
technical challenge. Direct measurement using a Hi-Flow (which can be an expensive and ultimately impractical
sampler and infrared requires physical access, is time- solution), or multiplexing multiple spatial points into a single
consuming, and not all sources are be identified or measured. instrument using a multiport valve manifold (which has the
Tracer release also requires site access and use of compressed disadvantage that the measurements are not performed
gas bottles, which are both significant impediments to rapid simultaneously).
measurement surveys of well pads. We have circumvented this limitation in the MFP method by
Inverse methods form another general category of using a novel gas storage manifold to sequentially analyze
quantification techniques. In an inverse method, an atmos- several inlet ports. The MFP method consists of (a) deploying
pheric transport model is used in conjunction with downstream a mast with 4−6 gas inlet ports on a vehicle, (b) connecting
concentration measurements to calculate upwind emissions. each of these inlet points to long tubes that can store
Thoma et al.10 employed such an inverse method from a approximately 50 s of gas, (c) driving the vehicle through the
mobile platform to quantify emissions from well pads in Texas, downwind plume, thus tracing out a measurement surface
Colorado, and Wyoming. The technique requires knowledge of defined by the mast and the path of the vehicle, (d) measuring
the source distance and height above the ground, as well as an the gas stored sequentially with a single high-accuracy methane
accurate assessment of the atmospheric turbulence and analyzer, and (e) combining the concentration data with
advective wind flow. Furthermore, terrain and the physical measured position of the vehicle to retrieve the concentration
arrangement of tanks and other physical obstructions like trees map on the surface. This concentration map, together with real-
and fences can modify the wind field and thus bias the time measurement of the wind through the surface, allows a
emissions measurement. direct calculation of the methane flux through the surface.
In this paper, we present a new methane emission The MFP measurement has distinct advantages over other
quantification technique called the mobile flux plane (MFP) techniques. First, knowledge of the location of the source is not
method capable of rapid, efficient quantification of total well required for the calculation of the emission rate. In addition,
pad emissions. In this technique, a vehicle equipped with a only a local measurement of the wind through the surface is
GPS, an anemometer, and a methane analyzer is driven required; there is no need to employ an atmospheric transport
downwind of well pads. A complete measurement of the total model. Obstructions and terrain effects that distort the plume
emissions from the well pad can be made in 6 min and does not do not affect the flux plane measurement provided that the flux
require site access. Below, we first describe the MFP method. plane captures the full extent of the plume. Unlike inverse
We then summarize controlled release experiments that validate methods, multiple upwind sources are automatically aggregated
the accuracy and precision of the technique. We then present a into a single leak rate without affecting the accuracy. Finally, the
field demonstration of the technique, in which the emissions concentration map provides clear evidence of whether the
from about 200 well pads in the Barnett Shale, Texas were plume is fully captured.
measured during a two-week campaign1 using the MFP vehicle. A schematic of the complete MFP system is shown in Figure
We conclude the paper with a discussion of the results and the 1. The methane analyzer is modified version of a commercially
implications for well pad emissions in this basin.
■
available near-infrared laser-based Cavity Ring-Down Spec-
trometer (CRDS) (Model G2301, Picarro, Santa Clara, CA,
EXPERIMENTAL SECTION USA, described in Crosson12). The analyzer is capable of
The MFP method is a ground-based mobile methane measuring ambient methane concentrations in real-time (2 Hz
measurement system that measures the methane flux through acquisition rate) with subparts-per-billion accuracy and
a vertically oriented measurement surface to infer emissions at a precision. In addition to measuring real-time methane
distance while driving downwind on roads near oil and gas concentrations with the CRDS instrument, wind speed and
producing facilities. Given an observed concentration profile, direction were measured with a sonic anemometer (Model
the flux of methane passing through the plane may be described 102779, Climatronics, Bohemia, NY, U.S.A.). The apparent
as follows: wind speed as measured from atop the car (1 m above the
vehicle roof) is corrected in real-time for the motion of the
q= ∬A k(C(y , z) − Co)u(z) dydz (1)
vehicle to reconstruct the true wind speed and direction with
respect to the ground. A submeter-resolution GPS unit (R100,
where q = measured flux of methane passing through the plane; Hemisphere, Scottsdale, AZ, U.S.A.) is used to provide real-
A = the surface over which the measurement is made; k = factor time location and car speed data.
for unit conversion from a volume of methane to kg hr−1; Air is sampled using a gas inlet tube mounted in front of the
C(y,z) = methane concentration as a function of horizontal car and located 1.7 m above the ground. In addition to this
position, y (along the axis of the vehicle’s motion) and height monitoring gas inlet, there are either four (for the validation
above the ground, z; Co = background methane concentration; experiments) or six (for the Barnett shale study) additional gas
4743 DOI: 10.1021/acs.est.5b00099
Environ. Sci. Technol. 2015, 49, 4742−4748
Environmental Science & Technology Article
direction, only a small subsample of all the oil and gas well pads quantify the minimum leak rate, using the trapezoidal
in the region could be sampled using the MFP vehicle. That integration method described in the SI).
practical limitation impeded our ability to select a truly
representative subsample from the greater population. As a
result, the emissions of the subpopulation of measured well
■ RESULTS AND DISCUSSION
The distribution of the N = 115 measurements as a function of
pads are biased to those located near public roads and may not distance is shown in the top panel of Figure 3. The average
be representative of the methane emissions of the greater
Barnett Shale production area.
During the field campaign (Oct. 15−30, 2013), we collected
N = 207 flux measurements, distributed as follows in the nine
county region centered on the major production areas: Cooke,
Denton, Hood, Johnson, Montague, Parker, Tarrant, Wise, and
Somervell. Surveys were planned to maximize the number of
facilities that would be within the measurement range of the
instrument from public roads. To guide our survey route, we
estimated which facilities would be observable based on open-
source maps of public roads, a database of well locations
(Zavala-Araiza et al.14), and the forecasted range of wind
directions. Each day, routes were chosen based on these
predictions, but since the accuracy of the predictions was not
perfect, any well pad facility that was found to be within range
of the instrument along the planned survey route was
measured. The distance between the measurement and source
well pad was calculated using the MFP mobile system’s GPS
data and the EDF-provided database of well locations. If there
were multiple wells on a single well pad, then the average of the
GPS coordinates in the database was used as the coordinates of Figure 3. top panel. Distribution of distances between the emission
the well pad. For this set of tests, only one measurement was source and the measurement point for the N = 115 MFP
made of each well-pad; it is important to note that we therefore measurements that meet all the data quality criteria (see text for
do not have an independent assessment of the measurement more information). Bottom panel: emissions (on a log scale) vs
uncertainty. distance.
To ensure that our sample contained only measurements of
well pads and no other leak sources, we selected a subsample of distance between the source and the measurement location for
the collected measurements according to the following these samples is 79 m. Note that of the well-pads nearer than
selection criteria: 150 m (N = 150), 77% met the plume centroid and width
• The mean lateral wind speed during the plume criteria (N = 115); by comparison, 84% (101 out of 120
measurement was greater than 1.0 m/s (N = 200). measurements) of the validation experiments met the same
This limit also effectively ensured that the angle between criteria, which is not surprising, given the mean distance of 34
the advective plume and the normal vector of the flux m for the validation experiments vs 79 m for the field campaign.
plane was within the ±60 degrees employed in the We observe a tendency toward fewer detections at greater
validation experiments; distances, indicating a reduction in the number of well pad
• The measurement could be clearly attributed to a single measurements that meet the data selection criteria, as expected
well pad in both the well pad database as well as the given the increasing size of the plume with propagation
satellite image, based on the on-board recorded mean distance.
wind direction (N = 177); The bottom panel shows the individual source emissions
• The estimated distance to the source well pad was 150 m measurements on a log scale vs distance, using the GF
or less (N = 150); integration method. There is no discernible bias in the
• The centroid of the plume as obtained from the Gaussian measured emissions with distance. The largest measured flux
fit of the peaks of the individual inlet measurements was was 47.6 kg/h; the smallest nonzero flux measured was 0.027
kg/h.
below the top inlet at 4.2 m (N = 142); and
The emissions distribution for these N = 115 well pads is
• The width of the plume as obtained from the Gaussian fit
shown in Figure 4, using the Gaussian fit method described in
of the peaks of the individual inlet measurements was less
the SI. The red curve is the modeled log-normal distributions,
than 5 m (N = 115), which is approximately the height of with geometric mean of 0.63 kg/h, geometric standard
the pole. 55% of all plume measurements satisfied these deviation of 4.1, and arithmetic mean of 1.74 kg/h. This is a
five criteria. Geographically, these measurements were highly skewed distribution, with the geometric mean differing
distributed in the nine county region as follows: Cooke from the arithmetic mean by a factor of about 2.5. In addition,
(1), Denton (36), Hood (1), Johnson (32), Montague we have included the precision of the measurement (yellow
(4), Parker (4), Tarrant (10), Wise (26), and Somervell curve) as derived from the controlled release work. The
(1). controlled release work was performed primarily under Pasquill
Forty-five percent of the measurements do not satisfy these Gifford atmospheric stability classes (De Nevers15) A,B; the
data quality criteria, and therefore cannot be used for accurate Barnett work was more in the B,C range; therefore, the
quantification (although they can in principle be used to controlled release precision should represent the worst case
4745 DOI: 10.1021/acs.est.5b00099
Environ. Sci. Technol. 2015, 49, 4742−4748
Environmental Science & Technology Article
■
*
ASSOCIATED CONTENT
S Supporting Information
A detailed description of the validation of the mobile flux plane
technique using controlled release experiments. This material is
available free of charge via the Internet at http://pubs.acs.org.
■ AUTHOR INFORMATION
Corresponding Author
*Phone: (408) 962−3941; fax: (408) 962-3200; e-mail: rella@
picarro.com (C.W.R.).
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
Notes
The authors declare no competing financial interest.
■ ACKNOWLEDGMENTS
The authors would like to thank Eben Thoma at the U.S.
Figure 5. Fraction of total emissions from a well pad population with a Environmental Protection Agency, who performed controlled
geometric standard deviation of 3.5, and with 37% of all sources having releases of methane in support of this research. We would also
zero emissions, plotted against the fraction of the well pad population, like to thank Pieter Tans, Colm Sweeney, Anna Karion, and
in order of decreasing emissions. Tim Newberger of NOAA for the development of the AirCore
technology used in this work. Finally, we would also like to
• The bottom 50% of the sources contribute less than 2% acknowledge the support of the Environmental Defense Fund,
of the total emissions. who provided partial support for this research.
These measurements were made on a population of 115
emitting sources and 67 nonemitters. Given that 20% of the
emissions is due to the top 1.1% of sources, this would indicate
■ REFERENCES
(1) Harriss, R.; Alvarez, R. A., Using multi-scale measurements to
that we have only measured perhaps one or two well pads in improve methane emissions estimates from oil and gas operations in
the Barnett Shale, Texas: Campaign summary. Manuscript in
this high emission bin. In other words, we are under-sampling preparation.
the high emitting tail of the distribution; the actual distribution (2) Myhre, G., Shindell, D.; F.-M. Bréon, Collins, W.; Fuglestvedt, J.;
may deviate from the derived log-normal distribution, either Huang, J.; Koch, D.; Lamarque, J.-F.; Lee, D.; Mendoza, B., et al.
higher or lower than the predicted distribution. Further work is (2013): Anthropogenic and Natural Radiative Forcing. In Climate
needed to identify the population of high emitters which may Change 2013: The Physical Science Basis. Contribution of Working Group
further skew the distribution. I to the Fifth Assessment Report of the Intergovernmental Panel on Climate
We further note that because we were not able to sample well Change; Stocker, T. F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.
pads in a fully random fashion, there are potential biases built K., Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P. M., Eds.;
Cambridge University Press: Cambridge/New York.
into this distribution that may not be reflected in the overall
(3) Howarth, R. W.; Santoro, R.; Ingraffea, A. Methane and the
population of well pads in the Barnett Shale. Lease road and/or greenhouse-gas footprint of natural gas from shale formations. Clim.
well pad access would be required to perform a more unbiased Change 2011, 106 (4), 679−690.
well pad survey. (4) Alvarez, R. A.; Pacalab, S. W.; Winebrake, J. J.; Chameides, W. L.;
We have presented an analysis of methane emissions from oil Hamburg, S. P. Greater focus needed on methane leakage from natural
and natural gas producing well pad facilities in the Barnett Shale gas infrastructure. Proc. Nat. Acad. Sci. 2012, v109 (17), 6435−6440
region measured using an innovative mobile flux plane (MFP) DOI: 10.1073/pnas.1202407109.
measurement system. For those well pads with nonzero (5) Allen, D. T.; Torres, V. M.; Thomas, J.; Sullivan, D. W.; Harrison,
emissions, the log-normal distribution is highly skewed, with M.; Hendler, A.; Herndon, S. C.; Kolb, C. E.; Fraser, M. P.; Hill, A. D.;
et al. Measurements of methane emissions at natural gas production
an arithmetic mean of 1.72 kg/h and with 95% of the well-pads
sites in the United States. Proc. Natl. Acad. Sci. U. S. A. 2013, 110 (44),
emitting between 0.037 and 10.5 kg/h. Including the 17768−17773.
population of nonemitting well pads, we find that the arithmetic (6) Eastern Research Group; Sage Environmental, City of Fort
mean of the well pads sampled in this study is 1.1 kg/h. If our Worth Natural Gas Air Quality Study, Final Report, July 13, 2011,
subsample is representative of the larger population of well available at http://fortworthtexas.gov/uploadedFiles/Gas_Wells/
pads, then this distribution would indicate that 50% of the AirQualityStudy_final.pdf, accessed August 2014.
emissions is due to the 6.6% highest emitting well pads, and (7) Energy Information Administration (2012): Number of
80% of the emissions is from the 22% highest emitting well Producing Gas Wells 2007−2012. Available at http://www.eia.gov/
pads. This result, if verified, has important implications for dnav/ng/ng_prod_wells_s1_a.htm. Accessed August 7, 2014.
(8) Shorter, J. H.; McManus, J. B.; Kolb, C. E.; Allwine, E. J.;
infrastructure integrity management of well pads, suggesting Siverson, R.; Lamb, B. K.; Mosher, B. W.; Harriss, R. C.; Howard, T.;
that deploying MFP systems to rapidly and accurately assess the Lott, R. A. Collection of leakage statistics in the natural gas system by
largest emitters in a production field can lead to an ∼80% tracer methods. Environ. Sci. Technol. 1997, 31 (7), 2012−2019.
reduction in emissions by addressing the largest 20% of the (9) Czepiel, P. M.; Mosher, B.; Harriss, R. C.; Shorter, J. H.;
sources. McManus, J. B.; Kolb, C. E.; Lamb, B. K. Landfill methane emissions