Continuous Methane Leak Detection and Localization for Oilfield Assets

Oilfield assets comprise thousands of equipment items that are connected by pipelines to aid production and conveyance of hydrocarbons. Along this chain there is risk of methane emissions due to leaks from old equipment and component failure. As methane is a significant greenhouse pollutant, one that is 84 times more potent than carbon dioxide in the atmosphere, a continuous monitoring strategy to identify leaks is imperative for climate emissions control. This talk presents a procedure that can quickly identify and isolate large methane emission sources leading to expedient remediation. Minimizing time to leak identification and the subsequent time to dispatch repair crews can significantly reduce the amount of methane released into the atmosphere.

The procedure developed utilizes permanently installed low-cost methane sensors at an oilfield facility to continuously monitor leaked gas concentration above background levels. The sensors provide concentration readings along with prevailing meteorological data that includes wind speed, wind direction, and solar intensity with time. The data are relayed to the cloud for processing in five key steps.

First, the data are partitioned by time windows to identify records of detection above background conditions. Second, a quality measure is established for each record as a measure of meteorological stability. Third, the data are processed to identify zones of detection for all active sensors in the form of linear inequalities. Fourth, the records are inverted for the leak source subject to all stipulated constraints by minimizing the difference between the observed and predicted concentrations. Here, the widely used Gaussian plume model (GPM) is adopted as the representative forward model that can provision methane concentration at a point given a known leak source. The model assumes steady-state atmospheric conditions and is the reason why meteorological data are used to infer atmospheric stability and to assign weights to each record. Lastly, an uncertainty quantification step is employed to extract confidence intervals. The procedure then repeats after a given delay period. When the leak measure is significant, an automated request can be issued for action by a repair crew.

Results demonstrating successful leak identification are presented.