A Hybrid Optimization Algorithm for Water Production and Distribution Operations in a Large Real-World Water Network

Abstract

This paper is concerned with the optimization of the operational aspects related to water production and distribution in a large real-world water supply network. The main operational costs considered are those related to water production and pumping. The investigated network has some special features, such as buffer in- and outflows, which are taken into account in the optimization. The problem is naturally formulated as a non-convex mixed-integer nonlinear program (MINLP) involving a large number of variables and constraints. As a consequence, the resulting optimization model is not easily tackled by the current software packages for water systems optimization. To partially tackle this issue, a hybrid two-steps method that aims at finding a global near-optimal water systems operating plan is proposed in this article. In the first step of the method, a mixed integer linear programming model (MILP) is constructed, using the piecewise linear approximation approach proposed in [10], and then solved to optimality. The resulting solution is then used as a starting point for Newtons method, which repairs it to obtain a feasible solution for the original MINLP model. This hybrid two-steps method is tested on a realistic network and the results show that the method is very competitive in terms of computation time while providing a near-optimal implementable water production and distribution operating plan.