Skip to content

EBMUD – Water Network Resiliency Evaluation

    CSI has collaborated with the EBMUD to deploy AI/ML techniques to improve water distribution network resiliency. This is a rapidly developing sector of CSI+EBMUD collaboration!

    1) Investigating the Effects of Operational and Environmental Factors on Water Distribution Pipeline Failures using Survival Analysis

    This study takes a systematic approach to investigate how operational and environmental factors influence pipe failures in a water distribution system, using data from the East Bay Municipal Utility District (EBMUD). The research integrates information about the pipes with various environmental data, including soil types, groundwater levels, geohazard zones, road quality, and traffic loads obtained from open-source databases. The pipes are then categorized based on these operational and environmental factors.

    To assess the impact of these factors on pipe failures, the study employs survival analysis techniques like Kaplan-Meier and Weibull estimations. The results of the analysis reveal the differing degrees to which operational and environmental factors contribute to pipe failures, thereby identifying the most and least influential factors in this context. The study’s findings are expected to provide valuable insights into the complex factors that affect pipe failures, which is crucial for prioritizing maintenance and replacement activities within water utility systems.

    2) Utilizing data-driven methods to evaluate likelihood of failure (LoF) in water distribution system pipelines

    This research aims to develop a data-driven model for pipeline replacement decision-making in EBMUD’s water distribution system. The objective is to determine the rate of pipeline deterioration and the migration of pipes from lower likelihood of failure (LOF) scores to higher scores. The proposed model will consider factors such as pipe characteristics, operational conditions, and environmental factors using modern data analysis techniques like machine learning and geospatial data analytics. The research plan includes data preparation, statistical analysis, model development, and combining the model with environmental data. The expected impact is to improve pipeline management practices and decision-making.

    Collaborator: East Bay Municipal Utility District (EBMUD)

    Researchers: Dayu Apoji, Shih-Hung Chiu, Kenichi Soga (UC Berkeley), David Katzev (EBMUD)

    Domains: Water Infrastructure

    Capabilities: Data Analytics & Machine Learning

    Publications: TBA!