Diego Leon, Project Canary
Methane emissions measurement technologies are evolving rapidly and becoming increasingly efficient over the last few years. The purpose of this paper is to introduce recent technological advancements that have helped operators in the US with more in-depth methane leak insights, improving the performance of emissions mitigation programs, ensuring proper management of associated risks, and delivering measurement-based methane emissions inventories. Technological advancements include both measurement hardware and emissions data processing algorithms and software tools. However, emission source detection, localization, and quantification are still areas of ongoing research and need further improvement.
A recently developed novel model allows the detection, localization, and quantification of the total site emissions from oil and gas production facilities using continuous monitoring data. This model uses real-time and historical data to quantify emissions from various intermittent and continuous sources while differentiating any offsite emissions. A machine learning model is employed to build a unique model for each methane monitoring device to determine how the wind direction affects the concentration readings, simulating plumes from all potential emission sources and matching the plumes to the device model with a mixture model. This model is currently used to quantify emissions on hundreds of operating well pads across the United States. These models are complemented with operator notification and alerting systems to ensure timely actions by operators that result in reducing their environmental footprint and help keep the gas in pipelines. The most recent updates to the operator notification systems, called Smart Alerts, employ machine learning algorithms to eliminate unnecessary notifications to avoid alert fatigue.