Water Distribution System Optimization Accounting for Worst-Case Transient Loadings

Abstract

This paper presents optimal transient design as a two-step optimization problem - identification and mitigation of the worst-case in a water distribution system. In the first step, particle swarm optimization was used to identify the set of critical nodes that result in the worst-case transient loading condition. In the second step, dual-objective optimization was used to determine the optimal pipe sizes that simultaneously minimize cost and the likelihood of damaging transient events, measured by a parameter named surge damage potential factor. Nondominated sorting genetic algorithms were combined with transient analysis to produce a set of Paretooptimal solutions in the search space of pipe cost and surge damage potential factor. The New York tunnel system was tested as a case and results show that the worst-case was not always obvious and cannot always be assumed a priori. Therefore, a comprehensive and systematic optimization is required to identify the worst-case in a network. It also confirmed that transient consideration in a design phase, in conjunction with conventional least-cost pipe size optimization, will help water utilities yield tangible cost savings along with improvement in system performance.