A heuristic algorithm for optimal cost design of gravity-fed water distribution networks. a real case study

Abstract

In this work, we put forward a new algorithm for minimizing the total cost of gravity-fed water distribution networks (WDN) by optimally choosing pipe diameters. Our approach is based on network theory and heuristic procedures and aims at overcoming a number of limitations found in the current scientific literature. WDN optimization problems belong to those that are discrete, non-linear and non-convex, making the use of standard optimization algorithms inefficient. Furthermore, resorting to metaheuristic search techniques to test many different solutions has an exceedingly expensive computational cost, due to the inherent complexity of the problem. Here we present an efficient heuristic algorithm that circumvents these limitations, exploiting detailed aspects, such as the specific network topology. In order to check the capability of the algorithm in a realistic setting, we have developed and analyzed a mathematical model of a water distribution network in a Spanish municipality, considerably larger than those addressed in previous works. Our numerical results indicate significant potential construction savings over current design practices, which justifies the interest of our approach in addressing this class of optimization problems.