Methane to Ethane Ratios of Emissions at Natural Gas Compressor Stations and Storage Facilities
Aethane emissions from natural gas production continue to be a targeted research area. Two methods to measure methane emissions include top-down and bottom-up approaches. However, reconciliation between these approaches requires further research . In October 2013, a collection of assorted top-down and bottom-up research teams converged on the Barnett Shale region to participate in the Environmental Defense Fund’s Barnett Coordinated Campaign (BCC) . A goal of this campaign was to address reconciliation between the approaches in order to improve emissions estimates from the oil and gas sector . Top-down approaches typically focus on determining a regional flux of emissions based on mass balance approaches and typically use full-scale fixed wing aircraft;
however, site level top-down approaches with unmanned aerial vehicles (UAV) are possible. Regional top-down approaches were completed for both methane and ethane emissions across the Barnett shale region and their ratios were us ed to attribute 71 to 85% of methane emissions to fossil sources [4, 5]. A facility level top-down approach used a UAV to estimate methane emissions from a natural gas compressor station and reported emissions higher than down-wind sampling and was likely due in part to lofted exhaust emissions [6, 7]. Bottom-up approaches focus on discrete measurements either at component or facility scales. These approaches may use downwind sampling and dispersion modeling, tracer ratio methods, or direct quantification. Bottom-up measurements have been combined with facility/equipment counts, activity factors, or emissions factors to estimate regional emissions rates. These analyses were also completed for the BCC by integrating multile studies, measurements, and inventories to estimate regional methane emissions for natural gas production sites and the Barnett shale region in general [8, 9]. Depending on measurement approach, total methane emissions represent those from both oil and gas production along with emissions from landfills, livestock, and other natural sources. To distinguish between sources, hydrocarbon ratios are commonly used in source apportionment processes as anthropogenic sources include higher alkanes. The ratio of ethane to methane (C2/C1) and propane to methane (C3/C1) can be used to corroborate apportionment for top down approaches. Aside from methane, ethane is the most abundant hydrocarbon. Research had previously shown a long-term decline in atmospheric ethane emissions (decline of 3.0 ± 0.4 Tg/yr from 1984 to 2010). This decline may have been attributed to decreased fugitive emissions from fossil fuels . Co-measurement of C2 and C1 was recommended and scrutiny in changes of their atmospheric growth rates was encouraged . However, recent studies have shown annual increases of 3-5% per year for the Northern Hemisphere which was shown to correlate with increased oil and natural gas production from shale plays [11, 12]. This recent rise in ethane was used to estimate net increases in methane emissions from the oil and gas sector from 20 Tg/yr in 2008 to 35 Tg/yr in 2014 . Similarly, atmospheric concentrations of propane also increased . Of particular interest, a study showed significantly increased ethane emissions within the Bakken shale region . The downwind C2/C1 ratio for the Bakken (40%) correlated well to the high C2/C1 ratio of the produced gas [42%] as opposed to processed gas . C2/C1 ratios for natural gas range from 0.1-14.5% by volume while C3/C1 ratios are typically lower and range from 0-6.4%. The C2/C1 ratio of biogenic sources is significantly less than 1% . There are currently six major natural gas shale plays and even more prospective plays that may produce in the fu-ture. . Table 1 includes a breakdown of the reported gas
composition from six of the major shale plays . Composition is shown as the average. Four different measurements were included for the Barnett shale to show regional variability. Not shown is the Bakken shale, which had a C2/C1 ratio of nearly 42% as it is dominated by oil production.
During the BCC researchers from Aerodyne, NOAA, and University of Colorado measured ground-based C2/C1 ratios throughout the region. The mode of their C2/C1 ratio measurements was 1.5% while the measurements varied between 1.1 and 17% . Another approach in the same region but prior to the BCC, used aircraft measurements to examine methane and ethane emissions and regions of correlated enhancements . Researchers at the University of Cincinnati and University of California – Irvine collected downwind canister samples throughout the BCC from a variety of locations and examined both hydrocarbon ratios and isotopic signatures for use in apportionment . Hydrocarbon ratios showed wide variability and correlations to δ13C showed poor correlation with methane concentration. It was suggested that future apportionment studies may benefit from measurement of δD-CH4.
Pétron et al focused on the Denver-Julesburg Basin with ground level and tower sampling for methane and C3+ alkanes . They found strong correlation of C3 to C1 among natural gas sources with C3/C1 ratios that ranged from 0.085 – 0.105%. Kort et al conducted airborne measurements of methane and ethane fluxes over the same region and period. Their C2/C1 ratios varied from 3.6 to 7.4% . Another bottom-up approach ses tracer release methods to estimate site level emissions rates. A recent study by Aerodyne, Carnegie Mellon University, and Colorado State University used this method to establish ethane and methane emissions as well as C2/C1 ratios for oil and gas facilities . The study identified varied C2/C1 ratios at sites that were attributed to different sources such as engine, tanks, and leaks. Subramanian et al used the same technique to measure emissions from transmission and storage (T&S) facilities . Though it was been shown that C2/C1 ratios vary by component,  showed excellent agreement between C2/ C1 ratios derived from tracer measurements and those ratios provided by site operators based on pipeline composition.
This research focused on site-to-site and on-site variability in C2/C1 ratios of five T&S facilities on which methane leak and loss audits were conducted . C3/C1 data have also been included and show similar trends but were not analyzed rigorously within this manuscript.
Emissions samples were collected from five natural gas facilities, which included three compression facilities and two storage facilities. Hydrocarbon ratios were examined for main contributors to emissions along with average leak ratios and both were compared to pipeline hydrocarbon ratios as provided by site operators. Reciprocating engine exhausts referred to samples collected from the raw engine exhausts at pre-installed sampling ports. Crankcases (CC) vents were vents that allowed pressure fluctuations from the engine volume containing the crankshaft to be vented such that blowby did not cause increased pressure on the backside of the pistons. Wet seals referred to the sealing mechanisms used on the centrifugal compressors that were mated to gas turbines. Packing vents were much like the CC vents for engines but were for the low-pressure side of the compressors where blowby occurred. Tanks/ slop tanks were liquid storage tanks onsite that were vented to the atmosphere. These tanks collected water, oils, and other liquids from onsite components for storage until removed by a third party. Leaks referred to unintended methane emissions from faulty components. Pipeline referred to the natural gas being transported by each facility, site operators provided these data, and the uncertainty in their measurements was not known. Table 2 includes site details.
Table 2. Site details including function, prime mover totals, and type.
Leak or loss samples were collected in a Tedlar bag sampling system. The bags were housed inside a plastic ‘brief-case’ as to be shielded from sunlight. Sampling times ranged from 5-10 minutes to ensure a sufficient sample (5-10 liters (L)) was collected in order to perform a 30-second analysis where the sampling flowrate was restricted to 1-2 L per minute. A heated filter and heated pump assembly fully evacuated sample bags to ensure that hydrocarbon condensation did not occur. The Fourier transform infrared (FTIR) spectrometer was an MKS Multigas 2030-HS analyzer (MKS Instruments, Andover, MA). The FTIR was a high-speed unit that reported at 5Hz. The unit used a 5.11-meter high-optical-throughput 200 mL gas sample cell and liquid nitrogen for cooling. The FTIR uncertainty was ± 5%. The FTIR was used to determine the C2/C1 and C3/C1 ratios. The uncertainty in ratio measurements was calculated using the sum of squares method , which yielded 7.07%. Prior to processing bag samples, a new reference background was collected using ultra-high purity nitrogen. After the new reference, bottled calibration gases of methane, ethane, and propane (each at +/-1% accuracy) were sampled by the FTIR for verification against the known value. The minimum detection limits of this FTIR for methane, ethane, and propane were 1, 2, and 5, respectively. However, the sampling flowrate was decreased to ensure a longer residence time within the cell. Each of the component level measurements was corrected for near-field ambient concentrations (local background). Near-field ambient samples were collected within the fence line of the facility but far from known leaks and lose. Sampling height was approximately 0.3 meters. All leaks or losses were 1-3 orders of magnitude higher than near-field ambient concentrations. A fractional analysis of moved gas was obtained from the station operators for the sites. It should be noted that Sites 1 and 2 used the same sampling point but analyses were completed on different days for respective comparisons to variability. The uncertainty of the gas chromatograph used for speciation was not known but the day-to-day variability was used for the statistical analysis. The dilute emissions samples were measured by the FTIR and their volume concentrations were reported in parts per millions (ppm). Equation 1 defined hydrocarbon ratios.
Site 1 was a compressor station. Engine exhaust was the main contributor of methane emissions accounting for 60% of site emissions . The C2/C1 ratio for the exhaust was 2.55% and was 10% lower than the average crankcase ratio, 13% lower than compressor packing ratio, and 20% lower than the average leak ratio. When multiple measurements or components were included in the average, the standard deviation is shown as error bars on the respective striped columns. Figure 2 shows the site wide variations of C2/C1 and C3/C1 ratios. Also included are the site average and pipeline data. Where single measurements were used and for the site averages, the relative measurement uncertainty of 7.07% is shown with solid col-].umns. The measurement uncertainty of the reported pipeline analysis was not known.
Figure 2. Site 1 C2/C1 and C3/C1 ratio distribution.
Site 2 was a storage facility and the two main contributors of methane emissions were engine exhausts (39%) and slop tanks (43%) . The engine exhaust had the highest C2/C1 ratio at this facility but it was not statistically different from the C2/C1 ratio of the slop tank. The C2/C1 ratio of the reciprocating engine exhaust was higher than the ratios of crankcase, packing, and slop tank ratios by 2%, 7%, and 2%, respectively. Figure 3 shows the distribution of C2/C1 and C3/C1 ratios for Site 2. This site included after treatment catalysts for the engine exhausts. The calculated C2/C1 ratio was not statistically different from that reported from pipeline data.
Site 3 was a compressor station and methane emissions from reciprocating engine exhaust were the main contributor at 50% . The C2/C1 ratio of the reciprocating engine exhaust was lower than the ratios of crankcase, packing, and slop tank ratios by 23%, 30%, and 23%, respectively. The C2/C1 ratio of the reciprocating engine exhaust was higher than wet seals and leaks by 57% and 7%, respectively. Figure 4 shows the C2/ C1 and C3/C1 distributions. Note that turbine exhaust concentrations of propane and ethane were below the detection limit (BDL) of the FTIR and are not presented.
Site 4 was a compressor station. The major emitters of methane emissions were the engine exhausts (43%) and compressor packing vents (47%) . Figure 5 shows that the C2/C1 ratio of the engine exhaust was lower than crankcase, packing, and leaks by 20%, 29%, and 15%, respectively.
Figure 5. Site 4 C2/C1 and C3/C1 ratio distribution.
Site 5 was a storage facility similar to Site 2 but used older twostroke lean-burn (2SLB) engines. Site 5 reciprocating engine exhaust showed the lowest C2/C1 ratio when compared to ther components. The C2/C1 ratio of the exhaust was lower than crankcase and leaks by 62%. The methane emissions at this site were more evenly distributed among leaks, losses, engine exhaust, packing, and crankcase . Figure 6 shows the istribution of C2/C1 and C3/C1 ratios.
Figure 6. Site 5 C2/C1 ratio distribution.
Near-field Ambient Samples
Bag samples were collected for near-field ambient measurements (sampler mounted approximately 0.3 m above groundlevel) within the fence line perimeter at all sites. Figure 7 shows the distribution of C2/C1 and C3/C1 ratios. In the caseof Site 1 and 3, the near-field ambient sample concentrations of propane, and propane and ethane were below the detectionlimit (BDL) of the FTIR.
Comparison of All Sites
The average C2/C1 ratios were applied to the respective methane volumetric emissions to estimate an average C2/C1 signaturefor each site. Figure 8 shows the estimated site-wide C2/C1 ratios by site relative to type of regional production .
The ratios matched well with the transition from dry to wet gas and the relative increase in C2/C1 ratios of the regionalnatural gas composition. Yacovitch et al reported a C2/C1 ratio of 2.8% for Site 1 , while the calculated C2/C1 of this workwas 2.7 ± 0.19%, well within our uncertainty. A separate ambient sample from within the facility was analyzed by  andyielded a C2/C1 of 2.58%, also within the uncertainty of the calculated site value. However, pipeline samples yielded a lowerC2/C1 ratio of 1.13%, well below the site average. We have estimated a C2/C1 ratio of 0.93 ± 0.07% for Site 2. The resultsof a downwind ambient sample of  yielded 0.73%. Figure 8 shows that estimated site signatures correlated well withthe spatial distribution of dry and wet gas regions; however, the Site 1 estimate was much higher than the pipeline data asdiscussed below and was determined to be the only site average that was statistically different from the reported pipelinevalue. Site 4 and 5 signatures were lower than their respective pipeline ratios and beyond the measurement uncertainty.Comparison of Components As shown in Figure 8, the average C2/C1 ratios varied by region but these ratios also varied by type of components within a given site. The C2/C1 ratio for engine exhaust was 17% lower compared to leaks, yielding the largest variation at Site1. The C2/C1 ratio for exhausts was 8% higher compared to compressor packing vents, yielding the largest variation at Site2. At Site 3, the C2/C1 ratio of reciprocating engine exhaust was similar to slop tank and leak ratios but lower than crankcaseand packing vents. However, the ratio of reciprocating engine exhaust was nearly 17% higher than compared to wet sealemissions. It is noted that other studies have highlighted difficulty in obtaining wet seal emissions rates using methods suchas calibrated bags . However, our sampling system utilized a dilution sampling method presented in  and where multiplepoints were measured, their emissions were combined. At Site 4, the lowest C2/C1 ratio was attributed to the engineexhaust. The ratio was 29% lower than packing emissions and 15% lower than leaks. Site 5 utilized older 2SLB technologyand its C2/C1 ratio was 62% lower than that of leaks and the pipeline sample.
Figure 9 and 10 shows the C2/C1 and C3/C1 ratios depicted on a box and whiskers plot along with sample size. Engine crankcasesnd leaks showed the highest variance in both C2/C1and C3/C1 ratios.
Comparison with Pipeline Gas
Natural gas fractional analyses were obtained from the operator. Sites 1 and 2 utilized the same pipeline sample locationas they fed the same regional system; however, our audits occurred on different days. As such, the respective daily fractionalanalysis was used for comparison. It is noted that the C2/C1 ratio for the gas sample varied between 0.9 and 1.14% over thefour days. Separate pipeline compositions were obtained for Sites 3-5. For Site 4, only an average value was reported and weassigned the largest variation from the other four sites to this value for the statistical analysis.To examine if data were statistically different, we calculated a 95% confidence interval (CI) for these small sample sizes us-ing one-tail cumulative probability from a t-Table. The largest number of samples for any component at a site was five while the minimum number was one. In the case of a sample size of one, the average of a minimum of 30 seconds of FTIR data was used. The standard deviation was calculated by bootstrapping five random measurements based on the uncertainty of the FTIR at one standard deviation.
Figure 8. Average C2/C1 ratios for sites measured in the Barnett Shale region
Figure 9. Variation of C2/C1 ratios for components among all sites for T&S facilities of the Barnett Shale.
Figure 10. Variation of C3/C1 ratios for components among all sites for T&S facilities of the Barnett Shale.
This random standard deviation was used to calculate the 95% CI for a sample size of five. Data were found to be statistically different if the averages ± 95% CI did not overlap. Table 2 shows the variations in the average site C2/C1 ratios as compared to the pipeline samples. It is noted that with this 95% CI method, only the Site 1 average was statistically different from the pipeline sample. Table 3 shows the site-specific data that were statistically different from the pipeline samples. It is noted that all data at Site 1 were found to be statistically different from the pipeline. At three of five sites, the C2/C1 ratio of the engine exhaust was statistically different from pipeline. All of these sites utilized either four-stroke or two-stroke, leanburn natural gas fuel engines without exhaust after treatment. However, the low pipeline C2/C1 ratio for Site 1 was statistically different the engine exhaust, engine crankcase, compressor packing, and leaks. Regarding the other sites, we saw that the only statistically different ratio at Site 3 were the wet seals of the centrifugal compressors which were 15% lower than that of the pipeline. It is noted that only Site 3 utilized this type of compressor as mated with natural gas turbines. Site 4 utilized four-stroke, lean burn natural gas engines, which had a C2/ C1 ratio that was 32% lower than pipeline. Site 5 utilized twostroke, lean burn natural gas engines, which yielded an even lower C2/C1 ratio compared to pipeline (-62%). At four of five sites, the engine exhausts were consistently lower than pipeline ratios.
Table 3. Pipeline C2/C1 compared with site-specific ratios.
Table 4. Statistically different C2/C1 ratios. * = bootstrapped sample.
Causes of Variability
It is believed that the C2/C1 ratios of leaks would best compare to that of the natural gas being transmitted. However, remote gas sampling is still an evolving technology. New systems utilize conditioning systems to ensure that gas samples are not affected by temperature and pressure changes, which could alter the measured composition . Beyond pressure and temperature effects of processes, there are other factors that affect the C2/C1 ratios of components including gas densities, solubility of gases in process liquids (engine oil, compressoroil, liquid hydrocarbons, and water), and the reactivity of hydrocarbon species within the combustion and after treatment processes.
The density of methane, ethane, and propane at 20°C are 0.668, 1.265, and 1.867 g/L, respectively . Not only is methane lighter than higher alkanes but it is also lighter than air. Noting that at most sites the main contributor of methane was from the hot engine exhaust, its buoyancy is further increased. Dependent upon sampling height and stability class, the downwind C2/C1 could vary. This effect is seen when comparing site-wide C2/C1 average ratios of Figure 8 as compared to near-field ambient C2/C1 ratios presented in Figure 7. In all cases where the ratios were measured, the near-field ambient ratios were higher than the estimated site wide ratio. It is noted that near-field ambient samples were taken 0.3 meters above the ground.
In addition to density, light alkanes also have variable solubility in water and oils, with methane being the least soluble in water and propane being more soluble than ethane in oils [30-32]. Solubility is strongly correlated with temperature and less so with pressure. Compressor packing emissions are at elevated temperatures compared to slop tanks, which were at or near ambient temperature. This likely contributes to the high variability in the C2/C1 ratios of the slop tank measurements. The C2/C1 ratios of engine crankcases are due to both elevatedtemperatures of fuel gas that leaks past the ring and due to the C2/C1 ratios of exhaust gases that may also be present within the sample.
Of greatest concern are the variable C2/C1 ratios created by the main methane emitters at these sites, the engine exhausts. All engines operating at these sites and most sites in general utilize lean-burn operation. The lower flammability limits for methane, ethane, and propane are 5, 3, and 2.2% volume in air, respectively . The combustion process of light alkanes is complex and comprised of a number of reactions. Ethane and propane also exhibit higher peak flame velocities (40-50 cm/s) as compared to methane (<40 cm/s) . At the same time, ethane and propane may exhibit lower ignition delays in air compared to methane dependent on temperature and pressure [34, 35]. Ignition delays of mutli-component mixtures, as in the case of natural gas, also lower ignition delays of higher hydrocarbon mixes as opposed to those with higher methane content [35, 36]. Ethane and propane also exhibit lower minimum ignition energy as compared to methane . The effects of fuel composition and equivalent methane number have been studied by industry utilizing natural gas engines for mobile applications (transportation). Methane number is a fuel quality metric similar to octane ratings used for gasoline. The methane number represents a gaseous fuels tendency to knock as compared to pure methane and hydrogen. Research has shown that total hydrocarbon emissions tend to increase with increased methane number (methane content) while non-methane hydrocarbons further decreased with an increase in methane number. This effect yielded increased methane emissions compared to C2+ hydrocarbons or lower C2/C1 ratios .
In fact, the low reactivity of methane is facing new research regarding catalyst developments necessary to oxidize methane at low temperatures. Catalysts (Pt/Rh/Pd) have shown high propane and ethane conversions (near 100%) with only limited methane conversion (~50%) [39, 40]. However, thesetypes of catalysts can and are used in exhaust after treatment systems at compressor stations to reduce emissions such as carbon monoxide. All these effects can explain the variable C2/ C1 ratios as compared to other components. Data shows clear trends of lower C2/C1 ratios for reciprocating engine exhausts at Site 1, 2, and 5. Due to the combustion chemistry, reactivity, ignition delay, and flammability of the various components of natural gas, it is likely that engine exhausts will exhibit lower C2/C1 ratios, especially if catalysts are not employed. This could significantly affect the overall C2/C1 signature of T&S facilities since the main contributor to methane inventories is due to engine operation. Table 4 shows the methane number as calculated for each of the Sites . No correlation was found between methane number and equivalent exhaust C2/ C1 ratio – likely due to varied engine age, load, and type.
Table 4. Methane number values for sites – based on pipeline analysis.
We conducted direct source quantification of methane emissions at T&S facilities as part of the Environmental Defense Fund Barnett Coordinated Campaign. The ratios of ethane and propane to methane were examined for different sources at these facilities. C2/C1 ratios were shown to vary within sites by up to 62%. The overall C2/C1 signatures were estimated for each of the sites and correlated well with regional variation in wet/dry gas composition. The major emitters of methane emissions at these facilities were the reciprocating engine exhausts. At three sites, the C2/C1 ratios of reciprocating engine exhausts were lower than all other components, and at Site 3 the C2/C1, exhaust ratios were lower than other main contributors were, except tanks. Only in the case of Site 2 were the C2/ C1 ratios higher from the exhaust as compared to other components. It is noted that this site included an after treatment catalyst which added another layer of variability to site wide C2/C1 ratios. Engine exhausts are the largest source of methane emissions and their C2/C1 signature tended to be lower than other components for both four-stroke, lean burn, and two-stroke, lean burn engines without exhaust after treatment. As new regulations take effect, industry may use oxidation catalysts, which could affect the average C2/C1 ratio of sites.
When compared to local pipeline gas composition, only Sites1 and 3 had a higher calculated C2/C1 site signature while the remaining had lower signatures compared to pipeline samples. The closest site signature, as compared to pipeline, was Site 3 where C2/C1 was 0.4% higher than pipeline. In the cases of reciprocating engine exhausts at Sites 4 and 5 and wet seals at Site 3, the C2/C1 ratios were found to be statistically lower than the pipeline samples. In all measureable cases, near-field ambient samples taken within the fence line of the facility exhibited igher C2/C1 ratios compared to pipeline and most components.
Due to the variable nature of the C2/C1 ratios from a T&S site, the correlation of atmospheric C2/C1 ratios with source apportionment should be conducted with due diligence at sites that could skew the atmospheric ratios when compared to the pipeline sources. Specifically, researchers should take cautionwhen using near-field ambient samples to classify the overall C2/C1 ratio of the site or of the pipeline gas. Samples taken nearer a specific component or at different sampling heights within the fence line could yield varied results. Building on this, the main contributor to methane emissions at these five T&S stations were the reciprocating engine exhausts. As thesesources are elevated in height, elevated in temperature, and have upward momentum, their effects on C2/C1 ratios require specific attention on mixing ratios when using ground-level,
downwind sampling methods.
The authors thank the West Virginia University Research Corporation, George Berry Chair Endowment, and the Environmental Defense Fund for supporting this program. The authors also thank Jason England, Chris Rowe, Zac Luzader, David McKain, and Richard Atkinson of WVU for their technical support.
WVU:West Virginia University;
FTIR: Fourier transform infrared;
kg/hr: kilograms per hour;
g/hr: grams per hour;
C2/C1: ethane to methane ratio;
C3/C1:propane to ethane ratio;
cm/s: centimeters per second;
BDL: below detection limit;
T&S: transmission and storage;
4SLB: four-stroke, lean-burn;
Supporting Information. Site specifics and brief component descriptions. This material is available free of charge via the Internet at http://pubs.acs.org.
1. http://www.netl.doe.gov/ File%20Library/Research/Energy%20Analysis/Life%20 Cycle%20Analysis/LCA-XIV-NG-Leakage–Littlefield-.pdf, Web. 15 November 2015.
3. Harriss.R,Alvarez.R,Lyon.D,Zavala-Araiza.R,Nelson, Dand Hamburg, S. Using Multi-Scale Measurements to Improve Methane Emission Estimates from Oil and Gas Operations in the Barnett Shale Region, Texas. Environ. Sci. and Technol. 2015. DOI: 10.1021/acs.est.5b02305.
5. Smith.M., et al. Airborne Ethane Observations in the Barnett Shale: Quantification of Ethane Flux and Attribution of Methane Emissions. Environ. Sci. and Technol. 2015. DOI: 10.1021/ acs.est.5b00219.
11. Franco.B.et al.Evaluating Ethane and Methane Emissions Associated with the Development of Oil and Natural Gas Extraction in North America. Environmental Research Letters. 2016. DOI:10.1088/1748-9326/11/4/0440101.
18. http://www2.mmm.ucar. edu/wrf/users/workshops/WS2013/posters/p20.pdf, Web. 15 November 2015.
19. BridgeTownsend-Small.A.Marrero.J,Lyon.D,Simpson.I,Meinardi. S,Blake.D. Integrating Source Apportionment Tracers into a Bottom-up Inventory of Methane Emissions in the Barnett Shale Hydraulic Fracturing Region. Environ. Sci. and Technol. 2015. 49 (13), 8175-8182; DOI: 10.1021/acs.est.5b00057.
21. Kort. E.et al. Airborne Ethane Observations in the Barnett Shale: Quantification of Ethane Flux and Attribution of Methane Emissions, Environ. Sci. and Technol. 2015. (49) 8158- 8166. DOI: 10.1021/acs.est.5b00219.
22. Roscioli.J.R.et al. Measurements of Methane Emissions from Natural Gas Gathering Facilities and Processing Plants: Measurement Methods. Atmos. Meas. Tech. Discuss. 2014. (7) 12357-12406. DOI: 10.5194/amtd-7-12357-2014.
23. Subramanian.R.et al. Methane Emissions from Natural Gas Compressor Stations in the Transmission and Storage Sector: Measurements and Comparisons with the EPA Greenhouse Gas Reporting Program Protocol. Environ. Sci. and Technol. 2014. (49) 3252-3261. DOI: 10.1021/es5060258.
24. BridgeJohnson.D,Covington.A,Clark.N. Methane Emissions from Leak and Loss Audits of Natural Gas Compressor Stations and Storage Facilities. Environ. Sci. and Technol. 2015. 49 (13), 8132-8138; DOI:10.1021/es506163m.
26. http://oilshalegas.com/ barnettshale.html. Web. 15 November 2015.
27. Natural Gas Industry Methane Emission Factor Improvement Study Final Report: Cooperative Agreement No. XA- 83376101; http://www.utexas.edu/research/ceer/GHG/ files/FReports/XA_83376101_Final_Report.pdf. Web. 15 February 2016.
32. Khadikar.P,Mandloi.D,Bajaj.A,and Joshi.S. QSAR Study on Solubility of Alkanes in Water and Their Partition Coefficients in Different Solvent Systems Using PI Index. Bioorganic and Medicinal Chemistry Letters. 2003. (13) 419-422. DOI: 10.1016/ S0960-894-X(02)00953-8.
35. Kuwahara.K.et al. Chemical Kinetics Study of Small-Alkane Ignition Process to Design Optimum Methane-Based Blend for HCCI. Society of Automotive Engineers International. 2014. DOI:10.4271/2014-01-1281.
38. Feist.M,Landau.M.Harte.E. The Effe3ct of Fuel Composition on Performance and Emissions of a Variety of Natural Gas Engines. Society of Automotive Engineers International. 2010. DOI: SAE 2010-01-1476.
39. Gelin.P,Urfels.L,Primet.M,Tena.E.Complete Oxidation of Methane at Low Temperature Over Pt and Pd Catalysts for the Abatement of Lean-Burn Natural Gas Fueled Vehicle Emissions: Influence of Water and Sulphur Containing Compounds. Catalysis Today. 2003. (83). 45-57. DOI: 10.1016/S0920- 5861(03)00215-3.
40. Mizutani.A,Okawa.T,Matsuzaki.H,Kubota.H.and Hosogai.S. Oxygen Sensor for CNG Application as ULEV or Tighter Emission Vehicle. Society of Automotive Engineers International. 1998. DOI: 10.4271/980264.