TODAY’S STUDY: NEW DATA SHOWS NAT GAS DRILLING RELEASES MORE GHGS THAN THOUGHT
University of Texas study underestimates national methane emissions at natural gas production sites due to instrument sensor failure
Touche Howard, 4 August 2015 (Energy Science and Engineering)
The University of Texas reported on a campaign to measure methane (CH4) emissions from United States natural gas (NG) production sites as part of an improved national inventory. Unfortunately, their study appears to have systematically underestimated emissions. They used the Bacharach Hi-Flow® Sampler (BHFS) which in previous studies has been shown to exhibit sensor failures leading to underreporting of NG emissions. The data reported by the University of Texas study suggest their measurements exhibit this sensor failure, as shown by the paucity of high-emitting observations when the wellhead gas composition was less than 91% CH4, where sensor failures are most likely; during follow-up testing, the BHFS used in that study indeed exhibited sensor failure consistent with under-reporting of these high emitters. Tracer ratio measurements made by the University of Texas at a subset of sites with low CH4 content further indicate that the BHFS measurements at these sites were too low by factors of three to five. Over 98% of the CH4 inventory calculated from their own data and 41% of their compiled national inventory may be affected by this measurement failure. Their data also indicate that this sensor failure could occur at NG compositions as high as 97% CH4, possibly affecting other BHFS measurement programs throughout the entire NG supply chain, including at transmission sites where the BHFS is used to report greenhouse gas emissions to the United States Environmental Protection Agency Greenhouse Gas Reporting Program (USEPA GHGRP, U.S. 40 CFR Part 98, Subpart W). The presence of such an obvious problem in this high profile, landmark study highlights the need for increased quality assurance in all greenhouse gas measurement programs.
The climatic benefits of switching from coal to natural gas (NG) depend on the magnitude of fugitive emissions of methane (CH4) from NG production, processing, transmission, and distribution [12, 13, 27]. This is of particular concern as the United States increasingly exploits NG from shale formations: a sudden increase in CH4 emissions due to increased NG production could trigger climate “tipping points” due to the high short-term global warming potential of CH4 (86× carbon dioxide on a 20-year time scale) . The United States Environmental Protection Agency (USEPA) estimates CH4 emissions from the NG supply chain by scaling up individual ground-level measurements, mostly collected by reporting from industry . However, some recent studies have questioned whether these “bottom-up” inventories are too low, since airborne measurements indicate that CH4 emissions from NG production regions are higher than the inventories indicate [5, 14, 17, 20, 21].
In order to help determine the climate consequences of expanded NG production and use, and to address the apparent discrepancy in top-down and bottom-up measurements, the University of Texas (UT) at Austin and the Environmental Defense Fund launched a large campaign to measure CH4 emissions at NG production sites in the United States . This study used both existing EPA GHG inventory data and new measurements to compile a new national inventory of CH4 emissions from production sites. Forty-one percent of this new inventory was based on measurements made by , which included measurements of emissions from well completion flowbacks as well as measurements of emissions from chemical injection pumps, pneumatic devices, equipment leaks, and tanks at 150 NG production sites around the United States already in routine operation (measurements from tanks were not used for inventory purposes). However, the measurements of emissions at well production sites already in operation (which comprised 98% of the new inventory developed by ) were made using the Bacharach Hi-Flow Sampler (BHFS; Bacharach, Inc., New Kensington, PA) and recent work has shown that the BHFS can underreport individual emissions measurements by two orders of magnitude . This anomaly occurs due to sensor transition failure that can prevent the sampler from properly measuring NG emission rates greater than ~0.4 standard cubic feet per minute (scfm; 1 scfm = 1.70 m3 h−1 or 19.2 g min−1 for pure CH4 at 60°F [15.6°C] and 1 atm; these are the standard temperature and pressure used by the U.S. NG industry). Although this failure is not well understood, it does not seem to occur when measuring pure CH4 streams, but has been observed in four different samplers when measuring NG streams with CH4 contents ranging from 66% to 95%. The sampler’s firmware version and elapsed time since last calibration may also influence the occurrence of this problem.
This paper presents an analysis of the UT  emissions measurements that were made with the BHFS, and shows that high emitters (>0.4 scfm [0.7 m3 h−1]) were reported very rarely at sites with a low CH4 content in the wellhead gas (<91%), consistent with sensor transition failure. It also details testing of the exact BHFS instrument used in that study and shows the occurrence of this sensor failure at an NG production site with a wellhead composition of 91% CH4 (the highest CH4 concentration site available during testing). Finally, the downwind tracer ratio measurements made by  at a subset of their test sites are reexamined and indicate that the BHFS measurements made at sites with low wellhead CH4…
Sensor transition failure is clearly apparent in the BHFS measurements made in the UT study by Allen et al. , as evidenced by the rare occurrence of high emitters at sites with lower CH4 (<91%) content in the wellhead gas. The occurrence of this sensor transition failure was corroborated by field tests of the UT BHFS during which it exhibited this sensor failure, as well as by tracer ratio measurements made by  at a subset of sites with lower wellhead gas CH4 concentrations. At this subset of sites, the tracer ratio measurements indicate that the BHFS measurements were too low by at least a factor of three. Because BHFS measurements were the basis of 98% of the inventory developed by  using their own measurements (and 41% of their total compiled inventory), the inventory clearly underestimates CH4 emissions from production sites. However, the extent of this error is difficult to estimate because the underreporting of emission rates due to BHFS sensor transition failure at any given site would vary depending on sampler performance and on how many high emitters were present at that site. Estimating this error is further complicated by the fact that the data set collected for pneumatic devices by  was an emitter data set; this might offset the effect of underreported high emitters in their pneumatic device emission factors. Finally, although real differences may exist in regional emission rates, the UT data set  should not be used to characterize them because the occurrence of sensor failure clearly varied between regions due to variations in wellhead CH4 compositions, which may mask any actual regional differences that existed.
Although the performance of the BHFS may vary between instruments or with sensor age or calibration vintage, this analysis of the  data set shows that measurements made using a BHFS for NG streams with CH4 content up to 97% could lead to severe underreporting of NG leaks. That this failure can occur at such high CH4 concentrations, which are close to the higher end of those found in transmission and distribution systems, indicates that past measurements in all segments of the NG supply chain could have been affected by this problem. Because the BHFS sensor transition failure phenomenon is not fully understood, it is not known how much this error may have affected past measurements of CH4 emission rates. Two factors preclude this: first, the performance of any individual BHFS may vary, and second, once sensor transition failure occurs, there is no way to determine the magnitude of the measurement error in the absence of an independent flux or concentration measurement.
If BHFS sensor transition failure has occurred during industry monitoring at transmission, storage, and processing compressor stations where the BHFS is approved for leak measurements mandated by the USEPA Subpart W Greenhouse Gas Reporting Program (GHGRP) , then these errors could be larger than those observed at production sites. Leaks at transmission, storage, and processing compressor stations commonly exceed 0.4 scfm (0.7 m3 h−1) (the approximate threshold for BHFS sensor transition failure) and in some cases may range from 10 to over 100 scfm. Because the largest 10% of leaks typically account for 60–85% of the total leak rate at a given facility [9, 25], sensor transition failure in the BHFS could bias CH4 emission inventories compiled by the USEPA GHGRP substantially low since the most significant leaks could be underreported. Additionally, leak measurements using the BHFS may be used to guide repair decisions at NG facilities, and underreporting of leaks could compromise safety if large leaks remain unrepaired as a result.
Finally, it is important to note that the BHFS sensor failure in the UT study  went undetected in spite of the clear artifact that it created in the emission rate trend as a function of wellhead gas CH4 content and even though the authors’ own secondary measurements made by the downwind tracer ratio technique confirmed the BHFS sensor failure. That such an obvious problem could escape notice in this high profile, landmark study highlights the need for increased vigilance in all aspects of quality assurance for all CH4 emission rate measurement programs.