Implementation of a Novel Natural Gas Distribution Screening Approach for High Flow Rate Below Ground Leaks Integrated with Company-Specific Emission Factors for Measurement-Informed Annual Emission Inventories

In this study, a novel screening method was implemented by a local distribution company (LDC) to identify below ground pipeline leaks that have high leak flow rates (≥10 scfh CH₄, 3.19 g/min) for the purpose of prioritizing repairs and reducing methane emissions. This decision tree (DT) method correlates methane concentration measurements as a function of defined ground surface conditions to leak flow rate. Established threshold methane surface concentrations at each defined surface condition category is used to identify leaks with high flow rates. Direct leak flow rate measurements at more than 400 leak sites were used to evaluate the method and determine the frequency of correctly classifying leak rate bins. These data were used in conjunction with annual leak inventory data to provide robust company-specific CH₄ emission factors (C-SEFs) with 90% confidence intervals (CI) of ∼±20%. State-of-the-art statistical analyses, bootstrap resampling, Monte Carlo, and Bayesian probabilistic analyses were used to estimate the DT errors, calculate the C-SEFs and confidence bounds, and estimate annual system emissions with CIs. These C-SEFs explicitly treat the skewed distributions of high flow rate vs low flow rate leaks: 13.3 scfh CH₄ (CI 10.4 to 16.6) for leaks ≥10 scfh CH₄ and 1.82 scfh CH₄ (CI 1.52 to 2.16) for leaks <10 scfh CH₄. Furthermore, these C-SEFs do not depend on classification of pipeline types and avoid issues with assigning a pipeline type for each leak. National methane emission estimates from natural gas distribution systems are outdated and have high uncertainty; however, C-SEFs can solve this problem. For the Southern California Gas Company system, annual leak emissions for 2015, before the method was implemented, were 38.3 Gg CH₄/yr, similar to an estimate using EPA emission factors. More importantly, the postimplementation 2023 emission estimate of 8.98 Gg CH₄/yr (CI 7.33 to 10.8) was 75% less than estimated for 2015. This emission reduction resulted from aggressive improvement in leak management practices implemented since 2015, including increased leak surveys, reduction of leak inventory, and application of the DT method to prioritize high flow leaks for repair.