You are here:
Bayesian Estimation of Fugitive Methane Point Source Emission Rates from a SingleDownwind High-Frequency Gas Sensor
Citation:
Albertson, J., S. Ferrari, G. Katul, T. Foster-Wittig, AND E. Thoma. Bayesian Estimation of Fugitive Methane Point Source Emission Rates from a SingleDownwind High-Frequency Gas Sensor. Air and Waste Management Association 107th Annual Conference, Long Beach, CA, June 24 - 27, 2014.
Impact/Purpose:
Abstract for platform presention. This is a contiaution of OTM 33 and33A techniques which set the stage for future automated mobile fugitve emisison destction for oil and gas and other applicaitons
Description:
Bayesian Estimation of Fugitive Methane Point Source Emission Rates from a Single Downwind High-Frequency Gas Sensor With the tremendous advances in onshore oil and gas exploration and production (E&P) capability comes the realization that new tools are needed to support environmentally responsible development of these key national energy assets. Concerns for air pollution emissions from this sector include the unintentional release of product-related fugitive emissions of methane, volatile organic compounds, and hazardous air pollutants. Recent information suggests that in a given production field, a relatively small number of high-emitters may be responsible for a disproportionately large amount of emissions due to issues such as pipeline leaks and malfunctioning well-pad equipment. These mitgable emissions have significant environmental and safety impacts while serving to unnecessarily increase costs through lost product. Development of automated mobile sensor technologies is an emerging environmental monitoring area that may provide useful tools to industry and regulators for cost-effective identification and repair of fugitive emissions. From a technical standpoint, development of automated mobile fugitive detection requires both rigorous emission estimation from single high-frequency sensors and the ability to efficiently plan sensing paths to cover the large areas economically. We have recently demonstrated a conditional sampling technique to reconstruct a Gaussian plume from single sensor data for a single source emission rate. However, even in the best case this approach would give a single estimate with no rigorous probabilistic assessment of its uncertainty. In this presentation we describe the development and testing of a Bayesian analysis technique that best defines the posterior probability density functions (PDFs) of the fugitive emission rate from stationary high-frequency methane time series. This effort involves novel representations of the probabilistic structure by which the short-time averages deviate from the true ensemble averages. The technique is demonstrated in this presentation on a dataset of “Controlled Release” experiments, where the point source emission rate is known. Additionally, the technique is applied to a collection of actual field data sets collected downwind of oil and gas production facilities. The results are quite encouraging and are currently being extended for use on sensor data collected from platforms in motion.