Drinking Water Pipeline Failure Analysis Based on Spatiotemporal Clustering and Poisson Regression

Abstract

Extended high fluctuations of internal water pressure have often been causally linked to pipeline failures in water distribution systems, especially when this exposure has continued over many years. However, as yet, there is little solid quantitative evidence to support this in real water systems. This paper examines the impact of internal pressure and other critical parameters on the integrity of water pipelines in five districts operated by a California water utility. Eight years of data from the utility’s Annual Pressure Survey and 25 years of water pipeline failure data were collectively analyzed using spatiotemporal clustering tools and a Poisson regression model. The analytical procedures and results presented in this paper are designed to support both the water utility’s short-term operations and its long-term programmatic drinking water pipeline planning and management practices.