Fuzzy Logic-based Economic Model for Water Main Renewal

Abstract

Drinking water mains are failing at accelerating rates nationwide, leaving managers without robust, consistent, validated tools to prioritize renewal projects in uncertain conditions. The research leveraged empirical pipeline data and advanced modeling technique to address the issue, beginning with the creation of a standard cost data framework developed using nearly 200 industry case studies from over 30 North American utilities. The data standard enabled an economic modeling study of common renewal tactics via life cycle costing and equivalent uniform cost comparisons of competing interventions over time. The modeling was tested through both deterministic and probabilistic methods to reveal the sensitivity of the models to variations/errors in key inputs such as discount rate, project costs, and pipe service life. Despite these efforts the economic models proved unreliable in accurately choosing the best intervention and timing when compared with best practices as defined by AWWA per the in-situ conditions, seeing only an average 63% match. To overcome the limitations, fuzzy modeling techniques were developed and applied using critical physical, spatial, and economic parameters driving pipe performance (St. Clair, 2015)(Ge 2017). The Type-2 model was trained and calibrated using nearly 37,000 break and advanced condition assessment records collected from participating utilities. The fuzzy model greatly outperformed the deterministic economic decision-making heuristics by achieving an 85% match with best practices in intervention selection and timing. The dissertation finally provides guidance on the deterministic vs fuzzy modeling techniques for practical implementation, addressing the benefits of integration with GIS for advanced asset management. The research provides a novel, results-backed approach that enables utilities to make defensible decisions under uncertainty amidst rapidly deteriorating critical water infrastructure.