Rehabilitation of a water distribution system using sequential multiobjective optimization models

Identification of the optimal rehabilitation plan for a large water distribution system (WDS) with a substantial number of decision variables is a challenging task, especially when no supercomputer facilities are available. This paper presents an initiative methodology for the rehabilitation of WDS based on three sequential stages of multiobjective optimization models for gradually identifying the best-known Pareto front (PF). A two-objective optimization model is used in the first two stages where the objectives are to minimize rehabilitated infrastructure costs and operational costs. The optimization model in the first stage applies to a skeletonized WDS. The PFs obtained in Stage 1 are further improved in Stage 2 using the same two-objective optimization problem but for the full network. The third stage employs a three-objective optimization model by minimizing the cost of additional pressure reducing valves (PRVs) as the third objective. The suggested methodology was demonstrated through use of a real and large WDS from the literature. Results show the efficiency of the suggested methodology to achieve the optimal solutions for a large WDS in a reasonable computational time. Results also suggest the minimum total costs that will be obtained once maximum leakage reduction is achieved due to maximum possible pipeline rehabilitation without increasing the existing tanks