A11J-0204
Mobile Detection of Fugitive Emissions using Computationally Optimized Geochemical Methods
Monday, 14 December 2015
Poster Hall (Moscone South)
Alex D Marshall1, David A Risk2, Martin Lavoie3, Bjorn-Gustaf Brooks4, Chris M Macintyre5, Jennifer Baillie6, Warren Daniel Laybolt2, James P Williams1, Mathias Goeckede7 and Claire Louise Phillips8, (1)St. Francis Xavier University, Earth Science, Antigonish, NS, Canada, (2)St. Francis Xavier University, Earth Sciences, Ottawa, ON, Canada, (3)St. Francis Xavier University, Antigonish, NS, Canada, (4)US Forest Service Asheville, Asheville, NC, United States, (5)St. Francis Xavier University, Earth Sciences, Antigonish, NS, Canada, (6)St. Francis Xavier University, Ottawa, ON, Canada, (7)Max Planck Institute for Biogeochemistry, Jena, Germany, (8)Oregon State University, Crops and Soil Science, Corvallis, OR, United States
Abstract:
The grand challenge of surface leak monitoring is to detect and attribute even small leaks across large energy development sites, which often span hundreds of square kilometres. Ratio-based geochemical methods show great potential for near-surface leak detection and attribution in vehicle-based mobile surveys. Ratios are useful especially when applied to concentration anomalies that exceed the Ambient Background (ABG), because they preserve the ratio of emission, and allow for more definitive attribution. Predicting ABG is, however, difficult because its variance originates from many processes including atmospheric patterns, local vegetation, other natural factors, and human activity.
Here we present a method of vehicle-based atmospheric leak detection. We have developed a signal conditioning process for accommodating a variable ABG throughout a survey dataset. ABG is the lowest value of a species within a time interval of variable length, and anomalies are detected when ratios of excess concentration (above ABG) exceed defined ratio limits based on expected sources. We computationally iterate through many configurations of ABG time interval and other parameters to find an optimized scenario. In surveys of CH4, δ13CH4, CO2 and H2S at a large energy development with active infrastructure, we compared our technique to a concentration threshold detection technique (2 ppm CH4), and a variation of our process where ABG is assumed to be the lowest dataset value.
Across ~1500 km of survey data, our process detected 8 times more leak anomalies than did the threshold technique. The lowest value background technique detected a similar number of leak anomalies as the optimized ABG, but was oversensitive to combustion (CO2-rich) emissions. With the optimized scenarios we observed some persistent leak anomalies in as many as 50% of survey passes, throughout different seasons and wind conditions. Leak persistence showed no significant relationship to leak size. CO2-rich leaks typically ranged from 10 to 80 ppm above ABG, while CH4-rich leaks were detected typically in the 10-40 ppb range. Overall, our technique increased both the visibility of small leaks, and the persistence of leaks of all sizes at only moderate computational expense.
