Twelve next generation emission measurement (NGEM) technologies completed single-blind testing at the Methane Emissions Technology Evaluation Center in 2018. This is the first series of tests to evaluate a wide variety of NGEM solutions including handheld, mobile, and continuous monitoring methods using comparable, repeatable protocols. Results assess performance of detection, localization and quantification, albeit with limited statistical significance due to a low number of tests. Overall, a higher detection rate is observed for handheld and mobile solutions than for continuous monitoring solutions. Compared to when a single emission source is present, a decline in detection rate is observed across all methods when multiple, steady emission sources are present. Localization by handheld and mobile solutions is more accurate than continuous monitoring solutions. These results support the common perception that detections by continuous monitoring systems will need to be confirmed and pinpointed by a follow-up inspection. Finally, this and other controlled release experiments, have been performed across a limited range of environmental conditions. To develop robust probability of detection curves needed for demonstrating emission reduction potential of leak detection and repair programs, new protocols are needed to evaluate methods across a wide range of metrological conditions and emission scenarios in a cost-effective manner.

This study presents the results of a campaign that estimated methane emissions at 268 gas production facilities in the Fayetteville shale gas play using onsite measurements (261 facilities) and two downwind methods – the dual tracer flux ratio method (Tracer Facility Estimate – TFE, 17 facilities) and the EPA Other Test Method 33a (OTM33A Facility Estimate – OFE, 50 facilities). A study onsite estimate (SOE) for each facility was developed by combining direct measurements and simulation of unmeasured emission sources, using operator activity data and emission data from literature. The SOE spans 0–403 kg/h and simulated methane emissions from liquid unloadings account for 88% of total emissions estimated by the SOE, with 76% (95% CI [51%–92%]) contributed by liquid unloading at two facilities. TFE and SOE show overlapping 95% CI between individual estimates at 15 of 16 (94%) facilities where the measurements were paired, while OFE and SOE show overlapping 95% CI between individual estimates at 28 of 43 (65%) facilities. However, variance-weighted least-squares (VWLS) regressions performed on sets of paired estimates indicate statistically significant differences between methods. The SOE represents a lower bound of emissions at facilities where onsite direct measurements of continuously emitting sources are the primary contributor to the SOE, a sub-selection of facilities which minimizes expected inter-method differences for intermittent pneumatic controllers and the impact of episodically-emitting unloadings. At 9 such facilities, VWLS indicates that TFE estimates systematically higher emissions than SOE (TFE-to-SOE ratio = 1.6, 95% CI [1.2 to 2.1]). At 20 such facilities, VWLS indicates that OFE estimates systematically lower emissions than SOE (OFE-to-SOE ratio of 0.41 [0.26 to 0.90]). Given that SOE at these facilities is a lower limit on emissions, these results indicate that OFE is likely a less accurate method than SOE or TFE for this type of facility.

Methane (CH 4 ) emission rates from a sample of natural gas facilities across industry sectors were quantified using the dual tracer flux ratio methodology. Measurements were conducted in study areas within the Fayetteville shale play, Arkansas (FV, Sept–Oct 2015, 53 facilities), and the Denver-Julesburg basin, Colorado, (DJ, Nov 2014, 21 facilities). Distributions of methane emission rates at facilities by type are computed and statistically compared with results that cover broader geographic regions in the US ( Allen et al., 2013 , Mitchell et al., 2015 ). DJ gathering station emission rates (kg CH 4 hr –1 ) are lower, while FV gathering and production sites are statistically indistinguishable as compared to these multi-basin results. However, FV gathering station throughput-normalized emissions are statistically lower than multi-basin results (0.19% vs. 0.44%). This implies that the FV gathering sector is emitting less per unit of gas throughput than would be expected from the multi-basin distribution alone. The most common emission rate (i.e. mode of the distribution) for facilities in this study is 40 kg CH 4 hr –1 for FV gathering stations, 1.0 kg CH 4 hr –1 for FV production pads, and 11 kg CH 4 hr –1 for DJ gathering stations. The importance of study design is discussed, including the benefits of site access and data sharing with industry and of a scientist dedicated to measurement coordination and site choice under evolving wind conditions.

Gathering pipelines, which transport gas from well pads to downstream processing, are a sector of the natural gas supply chain for which little measured methane emissions data are available. This study performed leak detection and measurement on 96 km of gathering pipeline and the associated 56 pigging facilities and 39 block valves. The study found one underground leak accounting for 83% (4.0 kg CH 4 /hr) of total measured emissions. Methane emissions for the 4684 km of gathering pipeline in the study area were estimated at 402 kg CH 4 /hr [95 to 1065 kg CH 4 /hr, 95% CI], or 1% [0.2% to 2.6%] of all methane emissions measured during a prior aircraft study of the same area. Emissions estimated by this study fall within the uncertainty range of emissions estimated using emission factors from EPA’s 2015 Greenhouse Inventory and study activity estimates. While EPA’s current inventory is based upon emission factors from distribution mains measured in the 1990s, this study indicates that using emission factors from more recent distribution studies could significantly underestimate emissions from gathering pipelines. To guide broader studies of pipeline emissions, we also estimate the fraction of the pipeline length within a basin that must be measured to constrain uncertainty of pipeline emissions estimates to within 1% of total basin emissions. The study provides both substantial insight into the mix of emission sources and guidance for future gathering pipeline studies, but since measurements were made in a single basin, the results are not sufficiently representative to provide methane emission factors at the regional or national level.

Coordinated dual-tracer, aircraft-based, and direct component-level measurements were made at midstream natural gas gathering and boosting stations in the Fayetteville shale (Arkansas, USA). On-site component-level measurements were combined with engineering estimates to generate comprehensive facility-level methane emission rate estimates (“study on-site estimates (SOE)”) comparable to tracer and aircraft measurements. Combustion slip (unburned fuel entrained in compressor engine exhaust), which was calculated based on 111 recent measurements of representative compressor engines, accounts for an estimated 75% of cumulative SOEs at gathering stations included in comparisons. Measured methane emissions from regenerator vents on glycol dehydrator units were substantially larger than predicted by modelling software; the contribution of dehydrator regenerator vents to the cumulative SOE would increase from 1% to 10% if based on direct measurements. Concurrent measurements at 14 normally-operating facilities show relative agreement between tracer and SOE, but indicate that tracer measurements estimate lower emissions (regression of tracer to SOE = 0.91 (95% CI = 0.83–0.99), R 2 = 0.89). Tracer and SOE 95% confidence intervals overlap at 11/14 facilities. Contemporaneous measurements at six facilities suggest that aircraft measurements estimate higher emissions than SOE. Aircraft and study on-site estimate 95% confidence intervals overlap at 3/6 facilities. The average facility level emission rate (FLER) estimated by tracer measurements in this study is 17–73% higher than a prior national study by Marchese et al.