Understanding Detection Success of Belowground Natural Gas Leaks in Urban Environments through Controlled Release Experiments

• Urban environment significantly reduces natural gas pipeline leak detection success. • Higher leak rates improve the probability of detection by up to 37 pp (percentage points) compared to lower leak rates. • The probability of detection drops sharply with increasing downwind distance and survey height. • Detection threshold strongly influences outcomes, especially at low leak rates. • Urban environments reduce detection probability by up to 38 pp compared to rural areas. Pipeline leak detection in urban areas is critical to safety and the mitigation of greenhouse gas emissions. However, the impact of the added complexity of key features of an urban environment to leak detection success is not well known, and therefore not currently included in leak detection practice. To address this gap, we conducted a series of controlled release experiments to assess the impact of leak rate, downwind distance, and survey height on the probability of detection (POD) for belowground natural gas (NG) pipeline leaks in a simulated urban environment, using walking, driving, and simulated unmanned aerial vehicle (UAV sim ) surveys. Results indicated a significant improvement in POD with increased leak rates, by 15 – 37 pp (percentage points) for main lines and 24 – 35 pp for service lines as the leak rate increased from low (0.5 slpm or 1 scfh) to high (10 slpm or 21 scfh). Downwind distance also influenced POD, with reductions in POD of 65 pp, 10 pp, and 49 pp at 15 m downwind for walking, driving, and UAV sim surveys, respectively, compared to surveys conducted directly over the pipeline. Survey height further impacted detection success, with walking surveys outperforming both driving and UAV sim surveys. Along the centerline, POD was reduced by 18 pp and 23 pp for driving and UAV sim surveys, respectively, compared to walking surveys. Interestingly, at 5 m and 15 m downwind, driving surveys showed improved POD over walking surveys, with increases of 23 pp and 37 pp, respectively. Additionally, the detection threshold (DT) significantly impacted POD, particularly for low leak rates and service lines. For the conditions tested, a 0.05 ppm increase in DT could reduce POD by 30 – 40 pp for main lines at low leak rates and for service lines at both 0.5 and 10 slpm leak rates. Furthermore, compared to rural environments, POD was reduced by 16 pp for main lines and 38 pp for service lines in urban areas, across various downwind distances and leak rates. These findings highlight the importance of accounting for urban environmental factors in methane (CH 4 ) leak detection and provide valuable insights for developing performance metrics that consider variations in site conditions.