Neural networks to locate and quantify fugitive natural gas leaks for a MIR detection system

Travis, B. J.; Dubey, M. K.; Sauer, Jeremy A.

doi:10.1016/j.aeaoa.2020.100092

Cited by 30 publications

(29 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These sensors provide near-continuous concentration measurements without needing a human operator. While reliable site-level or equipment-specific emission quantification is still an open problem, current CEMS can act as an indicator for methane emission events. − See Figure S13 for an example of emission events on an enrolled asset that were captured by the CEMS. Figures S2–S7 and S13 indicate that CEMS can detect small methane concentration enhancements on the order of 1 ppm.…”

Section: Resultsmentioning

confidence: 99%

Multiscale Methane Measurements at Oil and Gas Facilities Reveal Necessary Frameworks for Improved Emissions Accounting

Wang

Daniels

Hammerling

et al. 2022

Environ. Sci. Technol.

View full text Add to dashboard Cite

Methane mitigation from the oil and gas (O&G) sector represents a key near-term global climate action opportunity. Recent legislation in the United States requires updating current methane reporting programs for oil and gas facilities with empirical data. While technological advances have led to improvements in methane emissions measurements and monitoring, the overall effectiveness of mitigation strategies rests on quantifying spatially and temporally varying methane emissions more accurately than the current approaches. In this work, we demonstrate a quantification, monitoring, reporting, and verification framework that pairs snapshot measurements with continuous emissions monitoring systems (CEMS) to reconcile measurements with inventory estimates and account for intermittent emission events. We find that site-level emissions exhibit significant intraday and daily emission variations. Snapshot measurements of methane can span over 3 orders of magnitude and may have limited application in developing annualized inventory estimates at the site level. Consequently, while official inventories underestimate methane emissions on average, emissions at individual facilities can be higher or lower than inventory estimates. Using CEMS, we characterize distributions of frequency and duration of intermittent emission events. Technologies that allow high sampling frequency such as CEMS, paired with a mechanistic understanding of facility-level events, are key to an accurate accounting of short-duration, episodic, and high-volume events that are often missed in snapshot surveys and to scale snapshot measurements to annualized emissions estimates.

show abstract

Section: Resultsmentioning

confidence: 99%

Multiscale Methane Measurements at Oil and Gas Facilities Reveal Necessary Frameworks for Improved Emissions Accounting

Wang

Daniels

Hammerling

et al. 2022

Environ. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Methane (CH 4 ) is a greenhouse gas 28 times more potent than carbon dioxide considering its warming potential over 100 years (Travis et al, 2020). Anthropogenic CH 4 emissions account for 60% of global emissions (Saunois et al, 2016).…”

Section: Introductionmentioning

confidence: 99%

Reconstruction of high-frequency methane atmospheric concentration peaks from measurements using metal oxide low-cost sensors

Martinez

Santaren

Olivier

et al. 2022

Preprint

View full text Add to dashboard Cite

Abstract. Detecting and quantifying CH4 gas emissions at industrial facilities is important goal for being able to reduce these emissions. The nature of CH4 emissions through 'leaks' is episodic and spatially variable, making their monitoring a complex task, being partly addressed by atmospheric surveys with various types of instruments. Continuous records are preferable to snapshot surveys for monitoring a site, and one solution would be to deploy a permanent network of sensors. Deploying such a network with research-level instruments being expensive, low-cost and low-power sensors could be a good alternative. However, low cost entails usually lower accuracy and the existence of sensors drifts and cross-sensitivity to other gases and environmental parameters. Here we present four tests conducted with two types of Figaro TGS sensors on a laboratory experiment. The sensors were exposed to ambient air and peaks of CH4 concentrations. We assembled four chambers, each containing one TGS sensor of each type. The first test consisted in comparing parametric and non-parametric models to reconstruct the CH4 peaks signal from observations of the voltage variations of TGS sensors. The obtained relative accuracy is higher than 10 % to reconstruct the maximum amplitude of peaks (RMSE ≤ 2 ppm). Polynomial regression and multilayer perceptron (MLP) models gave the highest performances for one type of sensor (TGS 2611C, RMSE = 0.9 ppm) and for the combination of two sensors (TGS 2611C + TGS 2611E, RMSE = 0.8 ppm) with a training set size of 70 % of the total observations. In the second test, we compared the performance of the same models with a reduced training set. To reduce the size of the training set, we have employed a stratification of the data into clusters of peaks that allowed us to keep the same model performances with only 25 % of the data to train the models. The third test consisted of detecting the effects of age in the sensors after six months of continuous measurements. We observed performance degradation through our models, of between 0.6 and 0.8 ppm. In the final test, we assessed the capability of a model to be transferred between chambers on the same type of sensor, and found that it is possible to transfer models only if the target range of variation of CH4 is similar to the one on which the model was trained.

show abstract

“…Song and Jang, 2018; Y. Song and Li, 2021;Travis et al, 2020;Wang et al, 2021). These focus mainly on pipeline leak detection and involve various types of neural networks, such as convolutional neural networks, recurrent neural networks, hybrid networks, etc.…”

Section: Introductionmentioning

confidence: 99%

“…The trained model achieved high accuracy but required a series of filters and signal transformation techniques. In another work, artificial neural network was used to predict near real-time gas leakage at a testing site using both simulated and field data (Travis et al, 2020). However, as in many cases with machine learning, this model was highly dependent on sensor data, and interference of unexpected winds caused the model to overestimate the leak rates by a significant factor.…”

Section: Introductionmentioning

confidence: 99%

Gas leakage detection using spatial and temporal neural network model

Kopbayev

Khan

Yang

et al. 2022

Process Safety and Environmental Protection

View full text Add to dashboard Cite

Neural networks to locate and quantify fugitive natural gas leaks for a MIR detection system

Cited by 30 publications

References 13 publications

Multiscale Methane Measurements at Oil and Gas Facilities Reveal Necessary Frameworks for Improved Emissions Accounting

Multiscale Methane Measurements at Oil and Gas Facilities Reveal Necessary Frameworks for Improved Emissions Accounting

Reconstruction of high-frequency methane atmospheric concentration peaks from measurements using metal oxide low-cost sensors

Gas leakage detection using spatial and temporal neural network model

Contact Info

Product

Resources

About