Machine learning techniques to increase the performance of indirect methane quantification from a single, stationary sensor.

Researchers are searching for ways to better quantify methane emissions from natural gas infrastructure. Current indirect quantification techniques (IQTs) allow for more frequent or continuous measurements with fewer personnel resources than direct methods but lack accuracy and repeatability. Two IQTs are Other Test Method (OTM) 33A and Eddy Covariance (EC). We examined a novel approach to improve the accuracy of single sensor IQT whereby the results from both OTM and EC were combined with two machine learning (ML) models, a random forest (RF) and a neural network (NN). Then, models were enhanced with feature reduction and hyper-parameter tuning and compared to traditional quantification methods. The NN and RF improved upon the default OTM by an average of 44% and 78%, respectively. When compared to traditional OTM estimates with low Data Quality Indicators (DQIs), RF and NN models reduced 1σ errors from ±66% to ±13% and ±34%, respectively. Models also reduced the standard deviation of estimates with 93% and 85% of estimates falling within ±50% of the known release rate. This approach can be deployed with single sensor systems at well sites to improve confidence in reported emissions, reducing the number of anomalous overestimates that would trigger unnecessary site evaluations. Additional improvements could be realized by expanding training datasets with more methane release rates. Further, deployment of such models in a variety of situations could enhance their ability help close the gap between bottom-up inventory and top-down studies by enabling continuous monitoring of temporal emissions that could identify with improved confidence, atypically higher emissions. Accurate remote single sensor systems are key in developing an improved understanding of methane emissions to enable industry to identify and reduce methane emissions.

Methane emissions from oil and gas production sites and their storage tanks in West Virginia.

A measurement campaign characterized methane and other emissions from 15 natural gas production sites. Sites were surveyed using optical gas imaging (OGI) cameras to identify fugitive and vented emissions, with the methane mass emission rate quantified using a full flow sampler. We present storage tank emissions in context of all site emissions, followed by a detailed account of the former. In total, 224 well pad emission sources at 15 sites were quantified yielding a total emission rate of 57.5 ± 2.89 kg/hr for all sites. Site specific emissions ranged from 0.4 to 10.5 kg/hr with arithmetic and geometric means of 3.8 and 2.2 kg/hr, respectively. The two largest categories of emissions by mass were pneumatic devices (35 kg/hr or ~61% of total) and tanks (14.3 kg/hr or ~25% of total). Produced water and condensate tanks at all sites employed emissions control devices. Nevertheless, tanks may still lose gas via component leaks as observed in this study. The total number of tanks at all sites was 153. One site experienced a major malfunction and direct tank measurements were not conducted due to safety concerns and may have represented a super-emitter as found in other studies. The remaining sites had 143 tanks, which accounted for 42 emissions sources. Leaks on controlled tanks were associated with ERVs, PRVs, and thief hatches. Since measurements represented snapshots-in-time and could only be compared with modeled tank emission data, it was difficult to assess real capture efficiencies accurately. Our estimates suggest that capture efficiency ranged from 63 to 92% for controlled tanks.

On the Long-Term Temporal Variations in Methane Emissions from an Unconventional Natural Gas Well Site.

Understanding methane emissions from the natural gas supply chain continues to be of interest. Previous studies identified that measurements are skewed due to "super-emitters", and recently, researchers identified temporal variability as another contributor to discrepancies among studies. We focused on the latter by performing 17 methane audits at a single production site over 4 years, from 2016 to 2020. Source detection was similar to Method 21 but augmented with accurate methane mass rate quantification. Audit results varied from ∼78 g/h to over 43 kg/h with a mean emissions rate of 4.2 kg/h and a geometric mean of 821 g/h. Such high variability sheds light that even quarterly measurement programs will likely yield highly variable results. Total emissions were typically dominated by those from the produced water storage tank. Of 213 sources quantified, a single tank measurement represented 60% of the cumulative emission rate. Measurements were separated into four categories: wellheads (n = 78), tank (n = 17), enclosed gas process units (n = 31), and others (n = 97). Each subgroup of measurements was skewed and fat-tailed, with the skewness ranging from 2.4 to 5.7 and kurtosis values ranging from 6.5 to 33.7. Analyses found no significant correlations between methane emissions and temperature, whole gas production, or water production. Since measurement results were highly variable and daily production values were known, we completed a Monte Carlo analysis to estimate average throughput-normalized methane emissions which yielded an estimate of 0.093 ± 0.013%.

Temporal Variations in Methane Emissions from an Unconventional Well Site.

Studies have aimed to quantify methane emissions associated with the growing natural gas infrastructure. Quantification is completed using direct or indirect methods-both of which typically represent only a snapshot in time. Most studies focused on collecting emissions data from multiple sites to increase sample size, thus combining the effects of geospatial and temporal variability (spatio-temporal variability). However, we examined the temporal variability in methane emissions from a single unconventional well site over the course of nearly 2 years (21 months) by conducting six direct quantification audits. We used a full flow sampling system that quantified methane mass emissions with an uncertainty of ±10%. Results showed significant temporal variation in methane mass emissions ranging from 86.2 to 4102 g/h with a mean of 1371 g/h. Our average emissions rate from this unconventional well pad tended to align with those presented in the literature. The largest contributor to variability in site emissions was the produced water tank which had emissions rates ranging from 17.3 to 3731 g/h. We compared our methane mass emissions with the total production for each audit and showed that relative methane loss rates ranged from 0.002 to 0.088% with a mean of 0.030%, typically lower than reported by the literature, noting that our data excluded well unloadings. We examined natural gas production, water production, and weather conditions for trends. The strongest correlation was between methane emissions and historical water production. Our data shows that even for a single site, a snapshot in time could significantly over-predict (3×) or under-predict (16×) methane emissions as compared to a long-term temporal average.

Estimated Emissions from the Prime-Movers of Unconventional Natural Gas Well Development Using Recently Collected In-Use Data in the United States.

Natural gas from shale plays dominates new production and growth. However, unconventional well development is an energy intensive process. The prime movers, which include over-the-road service trucks, horizontal drilling rigs, and hydraulic fracturing pumps, are predominately powered by diesel engines that impact air quality. Instead of relying on certification data or outdated emission factors, this model uses new in-use emissions and activity data combined with historical literature to develop a national emissions inventory. For the diesel only case, hydraulic fracturing engines produced the most NO x emissions, while drilling engines produced the most CO emissions, and truck engines produced the most THC emissions. By implementing dual-fuel and dedicated natural gas engines, total fuel energy consumed, CO2, CO, THC, and CH4 emissions would increase, while NO x emissions, diesel fuel consumption, and fuel costs would decrease. Dedicated natural gas engines offered significant reductions in NO x emissions. Additional scenarios examined extreme cases of full fleet conversions. While deep market penetrations could reduce fuel costs, both technologies could significantly increase CH4 emissions. While this model is based on a small sample size of engine configurations, data were collected during real in-use activity and is representative of real world activity.

Development of engine activity cycles for the prime movers of unconventional natural gas well development.

With the advent of unconventional natural gas resources, new research focuses on the efficiency and emissions of the prime movers powering these fleets. These prime movers also play important roles in emissions inventories for this sector. Industry seeks to reduce operating costs by decreasing the required fuel demands of these high horsepower engines but conducting in-field or full-scale research on new technologies is cost prohibitive. As such, this research completed extensive in-use data collection efforts for the engines powering over-the-road trucks, drilling engines, and hydraulic stimulation pump engines. These engine activity data were processed in order to make representative test cycles using a Markov Chain, Monte Carlo (MCMC) simulation method. Such cycles can be applied under controlled environments on scaled engines for future research. In addition to MCMC, genetic algorithms were used to improve the overall performance values for the test cycles and smoothing was applied to ensure regression criteria were met during implementation on a test engine and dynamometer. The variations in cycle and in-use statistics are presented along with comparisons to conventional test cycles used for emissions compliance. IMPLICATIONS: Development of representative, engine dynamometer test cycles, from in-use activity data, is crucial in understanding fuel efficiency and emissions for engine operating modes that are different from cycles mandated by the Code of Federal Regulations. Representative cycles were created for the prime movers of unconventional well development-over-the-road (OTR) trucks and drilling and hydraulic fracturing engines. The representative cycles are implemented on scaled engines to reduce fuel consumption during research and development of new technologies in controlled laboratory environments.