Optimizing Maintenance and Replacement Activities for Water Distribution Pipelines

Abstract

As water pipelines reach the end of their useful lives, asset managers must develop replacement and maintenance programs that mitigate the risk of pipeline failure. A risk-based asset management program must include likelihood of failure models to predict pipeline break rates or future condition. Though over forty predictive pipeline failure/condition assessment models have been introduced in the past few decades, recent surveys of utility practice show that few utilities are adopting them. The major reasons for these poor adoption rates are rooted in the data requirements and complexity of models presented in literature, and the lack of case studies demonstrating how to incorporate likelihood of failure into risk-based planning. This research introduced a likelihood of failure model that limited the number of input parameters by introducing a surrogate parameter that describes the density of breaks across a network. It was concluded that this model, initially demonstrated using data from a large utility, resulted in improved prediction performance with respect to identifying pipe groups most likely to fail. A risk-based maintenance and replacement optimization framework was demonstrated on a subset of this large utility, showing that binning of the network allowed for better convergence of the optimization algorithm and allowed the decision maker to identify candidate maintenance and replacement assets on the street and neighborhood level. It was discovered that this likelihood of failure model was not applicable for two neighboring medium sized utilities with limited numbers of breaks in pipe material classes. It was also concluded that model transfer techniques recommended in literature did not improve prediction performance. As an alternative, a clustering analysis framework was introduced that incorporates expert knowledge of the network operations and hydraulic connectivity. This framework allowed for the identification and prioritization of maintenance and replacement activities with respect to activity costs and criticality, allowing for utilities of all sizes to make better informed asset management maintenance and replacement decisions.