Towards operational methane emission detection from oil and gas facilities through multi-modal sensing and advanced dispersion and atmospheric modeling

Abstract

Localizing methane emissions from oil and gas facilities is a challenging endeavor. Emissions are intermittent. Sensing systems (terrestrial, aircraft, and satellites) are sparse, and frequency of measurement is low. The computational cost of recovering emission rates from measurements can be substantial for high-fidelity frameworks. Emission detection must be precise when multiple oil and gas operators are active near one another. Reliable and timely detection of methane emissions from oil and gas facilities is imperative for successful mitigation. We address the following: which sensing technology is optimal under different emission scenarios; which computational model is most appropriate for various emission regimes; how do various sensing technologies and computational models combine to improve reliability of detection and response time? We attempt to answer these questions by modeling a well-documented methane emission event in the Permian Basin, where many oil and gas operators are active. We compare Lagrangian and Eulerian dispersion models to better understand their strength and shortcomings under different emission and atmospheric conditions, when multi-modal sensing systems are used. We compare several dispersion and atmospheric models with varying levels of grid-resolution (from 100 m to 3 km). We also study how these models perform under different emission scenarios (from 10 kg/h to 1000 kg/h), which are representative of moderate to large emissions. We compare the computational cost of different models and highlight options to reduce them. We attempt to draw conclusions on the needed level of grid resolution to model routine- to super-emitters, either at equipment- or facility-level. Equipment-level emission modeling generally requires costly, fine-resolution grids and operational models. Facility-level emission modeling can tolerate coarser grid resolution but need to be augmented with additional strategies for mitigation purposes, such as using a locally finer model, or human inspection. Statistical behavior of various equipment also provides valuable information for mitigation attempts. Lastly, we comment on challenges that need to be addressed for the deployment of a reliable, timely, and cost-efficient methane emission detection framework in an operational setting.