GC085-0020
The Fugitive Emissions Abatement Simulation Testbed- Distribution (FEAST-D): A Stochastic Model to Assess Methane Emissions Mitigation in Urban Distribution Systems
The Fugitive Emissions Abatement Simulation Testbed- Distribution (FEAST-D): A Stochastic Model to Assess Methane Emissions Mitigation in Urban Distribution Systems
Monday, 14 December 2020
Poster
Abstract:
Methane leaks from aging leak-prone urban distribution systems pose a safety threat to communities and are a leading contributor to urban methane emissions . Estimating the number and size of methane leaks and directing appropriate repair action is critical to improve safety and address climate change concerns and will also help utilities to evaluate the adoption of leak detection and repair (LDAR) and pipe replacement programs. In this work, we extend the FEAST platform originally developed to analyze upstream emissions to the downstream distribution sector, thereby providing a comprehensive policy analysis tool throughout the natural gas supply chain.
FEAST-D is a data-intensive, discrete-time, Markov process-based model which enables users to estimate the number and size of methane leaks in downstream natural gas pipelines and evaluate the costs and benefits of various LDAR programs. Historical data from the Pipeline and Hazardous Material Safety Administration (PHMSA) representing the continental U.S. distribution sector pipelines and services are used to develop empirical emissions distributions. Using a multivariable linear regression model, we estimate the number of leaks as a function of pipeline material (copper, steel, ductile iron, and plastic) and year of installation (1940-2019). A predictive model was then developed using a supervised machine learning strategy to generate a distribution of number of leaks. Emissions size distributions and parameterization of leak detection technologies were obtained from peer-reviewed literature. Using this model, we evaluate potential emissions reducing when LDAR programs prioritize fixing ‘super-emitter’ leaks. Using the cost-benefit framework in the model, we compare the costs and benefits and detection effectiveness of new methane detection platforms such as drones, trucks, and handheld monitoring systems. Finally, we discuss the use and implications of this model for state- and city-level methane mitigation regulations.
FEAST-D is a data-intensive, discrete-time, Markov process-based model which enables users to estimate the number and size of methane leaks in downstream natural gas pipelines and evaluate the costs and benefits of various LDAR programs. Historical data from the Pipeline and Hazardous Material Safety Administration (PHMSA) representing the continental U.S. distribution sector pipelines and services are used to develop empirical emissions distributions. Using a multivariable linear regression model, we estimate the number of leaks as a function of pipeline material (copper, steel, ductile iron, and plastic) and year of installation (1940-2019). A predictive model was then developed using a supervised machine learning strategy to generate a distribution of number of leaks. Emissions size distributions and parameterization of leak detection technologies were obtained from peer-reviewed literature. Using this model, we evaluate potential emissions reducing when LDAR programs prioritize fixing ‘super-emitter’ leaks. Using the cost-benefit framework in the model, we compare the costs and benefits and detection effectiveness of new methane detection platforms such as drones, trucks, and handheld monitoring systems. Finally, we discuss the use and implications of this model for state- and city-level methane mitigation regulations.