Cross Entropy multiobjective optimization for water distribution systems design

A methodology extending the Cross Entropy combinatorial optimization method originating from an adaptive algorithm for rare events simulation estimation, to multiobjective optimization of water distribution systems design is developed and demonstrated. The single objective optimal design problem of a water distribution system is commonly to find the water distribution system component characteristics that minimize the system capital and operational costs such that the system hydraulics is maintained and constraints on quantities and pressures at the consumer nodes are fulfilled. The multiobjective design goals considered herein are the minimization of the network capital and operational costs versus the minimization of the maximum pressure deficit of the network demand nodes. The proposed methodology is demonstrated using two sample applications from the research literature and is compared to the NSGA-II multiobjective scheme. The method was found to be robust in that it produced very similar Pareto fronts in almost all runs. The suggested methodology provided improved results in all trails compared to the NSGA-II algorithm.