A new multi-objective evolutionary algorithm for the optimization of water distribution networks

Abstract

Abstract This work introduces a new bi-objective optimization model for the design of a water distribution network (WDN). Reliability and design cost are two contradicting objectives that are optimized. The evolutionary technique multi-objective (MO) Jaya, hereafter referred as MO-JAYANET, is developed for the optimal design of WDN. EPANET is used as a hydraulic network solver in this study. The concept of constraint violation, dominance, non-dominance and crowding distance is used to develop MO-JAYANET. As the Jaya algorithm is a parameter-less technique, and with the concept of constraint dominance involved in the MO model, the penalty parameter required to penalize the non-feasible solution is also dropped. This makes MO-JAYANET a completely parameter-less technique, requiring no tuning at all. MO-JAYANET is applied on two benchmark networks from the literature using three different reliability indices that are Resilience Index, Network Resilience Index, and Modified Resilience Index. MO-JAYANET is found to be very easy and efficient for WDN optimization. The present work obtains a smooth Pareto Front (PF) that overlaps the best-known PF. The maximum reliability diameters obtained by the MO-JAYANET for the three indices are also simulated for both hydraulic and mechanical failures under different scenarios to identify a better-performing index.