Multi-objective optimization of cost and resilience of water distribution system design

The presented study features a multiobjective optimization model that aims to achieve the best tradeoff between low cost and high reliability of a water distribution system design. The algorithm is novel in merging the use of Genetic Algorithm (GA); features an effective exploration of the search space, with the so-called «split pipes» method; and enables the use of more than one diameter for each pipe section. This was achieved by a new formulation that takes advantage of a theoretical old finding that multidiameter pipes, if they exist in a solution, comprise at most two adjacent discrete pipe diameters. The method was applied on the well studied problems of the two-looped and Hanoi network and achieved the best known least-cost design solutions. In addition the study presents an interesting insight regarding the resilience measurement. Analysis of the system response to diverse failure cases (involving demand increase and pipe breakage) showed that an improvement in resiliency does not necessarily imply a reliability improvement.