All Times EDT
Virtual
Characterizing Sampling and Detection Properties of AMLD (308535)
*Emily Annette Luetschwager, Colorado State UniversityJoseph C von Fischer, Colorado State University, Dept. of Biology
Zachary D Weller, Colorado State University, Dept. of Statistics
Keywords: logistic regression, Monte Carlo simulation, greenhouse gases, mobile surveys
Advanced mobile leak detection (AMLD) is a new technology for surveying local natural gas (NG) distribution systems. Although AMLD has shown promise as a tool for managing NG systems, there are many open questions regarding the features and efficacy of these mobile surveys. We analyzed a large data set arising from mobile methane surveys in 14 U.S. cities to investigate temporal correlation of methane concentrations, quantify variability in leak emission rate estimates and estimate the probability of leak detection. Our results indicate that methane concentrations are only moderately correlated at short (<2 hr) time scales, and otherwise nearly uncorrelated. We also show how emission rate estimates and leak detection probability vary with leak size. We combined the results of these analyses to develop a Monte Carlo simulation that demonstrates the effect of sampling effort for detecting leaks and quantifying emissions for a simulated population of leaks. Our results provide insights into recommended driving procedures for using these surveys as a part of leak detection and methane emissions monitoring efforts.