A recent study has confirmed that the first-ever natural-gas leak-detection tool developed by researchers from Los Alamos National Laboratory effectively uses machine learning and sensors to detect leak points at oil and gas fields.
The new tool promises new automatic, economical sampling across large natural gas infrastructure.
Our automated leak location system finds gas leaks fast, including small ones from failing infrastructure, and lowers cost as current methods to fix gas leaks are labor intensive, expensive and slow. Our sensors outperformed competing techniques in sensitivity to detecting methane and ethane. In addition, our neural network can be coupled to any sensor, which makes our tool very powerful and will enable market penetration.
Manvendra Dubey, Study Co-Author and Lead Scientist, Los Alamos National Laboratory
The team developed the Autonomous, Low-cost, Fast Leak Detection System (ALFaLDS) to identify inadvertent discharges of methane—a powerful greenhouse gas—and received the 2019 R&D 100 award.
The ALFaLDS tool detects, locates, and measures a natural gas leak based on real-time ethane and methane (contained in natural gas) and on atmospheric wind measurements investigated by a machine-learning code trained to find leaks.
The machine-learning code is trained using high-resolution plume dispersion models from Los Alamos National Laboratory, and the training is further improved on-site through controlled discharges.
Results of tests using blind discharges at an oil and gas well-pad facility based at Colorado State University in Fort Collins, Colorado, showed that the ALFaLDS tool detects the engineered methane leaks accurately and measures their size.
The article, published in the Atmospheric Environment: X journal, concluded that this new ability to detect leaks with excellent skill, precision, and speed at reduced cost may allow new automatic and economical sampling of fugitive gas leaks that occur at gas and oil fields and well pads.
The success of the ALFaLDS tool in finding and measuring fugitive methane leaks at natural gas centers may result in a 90% reduction in methane discharges if executed by the sector.
The ALFaLDS tool uses a tiny sensor, which makes it perfect for installing on drones and cars. At present, the Los Alamos group is designing the sensors that were combined with a mini 3D sonic anemometer and the robust machine-learning code in these analyses.
But the code is autonomous and capable of reading information from any wind and gas sensors to help detect leaks quickly and decrease fugitive gas emissions from the huge network of natural gas extraction, production, and usage.
With this addition, the ALFaLDS tool provides a groundbreaking method for non-profit organizations surveying the problem, for oil and gas service providers in leak detection, and academia and national laboratories investigating the production of natural gas.
The development of the ALFaLDS tool was financially supported by the Department of Energy Advanced Research Projects Agency-Energy. Aeris Technologies and Rice University collaborated on the initial project that received the 2019 R&D 100 award from “R&D Magazine.”
Travis, B., et al. (2020) Neural networks to locate and quantify fugitive natural gas leaks for a MIR detection system. Atmospheric Environment: X. doi.org/10.1016/j.aeaoa.2020.100092.