A probabilistic approach to the quantification of methane generation in sewer networks

Abstract

Quantifying greenhouse gas (GHG) emissions from the conveyance of wastewater is an essential part of emissions reduction as it can identify areas of high emissions that can be targeted for mitigative action. In this study, a Monte Carlo algorithm that employs a reach-based methane generation sub-model was developed and applied to a full-scale municipal sewer system in Ontario, Canada. The algorithm employed eight categories of random variables including sewage temperature, slope, and coefficients described in the sewer reach model. Using best estimates for the employed distributions and algorithm design choices, it was estimated that 2.1–3.0 g CH4/m3 (of total wastewater conveyed) is generated in the sewer system. Gravity reaches contributed 1.3–2.2 g CH4/m3, and force main reaches contributed 0.6–0.9 g CH4/m3, or 30% of total sewer-generated methane despite contributing only 4.4% of total network length. The results suggest that addressing force main methane generation (such as employing chemical addition) could reduce a large fraction of sewer-generated methane while only requiring action on a small fraction of sewer reaches which is consistent with literature. Extending the results from this study to all sewage generated in Canada indicates that anthropogenic emissions from the wastewater sector are increased by 28–40% if sewer-generated methane is included in the assessment. After testing alternative distributions and model designs, it was determined that replacing the fullness and temperature distributions with constant (no distribution) average conditions yielded identical results to that of the base case assessment, suggesting that these random variables can be excluded from future modelling exercises. It was also observed that treating model coefficients as random variables resulted in a significant increase in the standard deviation of estimates, indicating that much of the uncertainty in the results is due to the uncertainty associated with the model coefficients. The results were sensitive to the temperature correction coefficient in the methane generation model and the Manning's n used in flow calculations; indicating that dedicating resources to accurately characterize these values will increase model accuracy.