Metaheuristic-based optimization of treated water distribution in a Mexican City

Abstract

Nowadays, urban areas are composed of logis- tic networks that form a compact and complex entity whose integration has a meaningful impact on the sustainability of the urban system. Currently, there are several models for the optimization of these networks mainly focused on the trans- port of products and people. One of these models is the Vehi- cle Routing Problem (VRP). The scientific literature provides evidence that Genetic Algorithms (GAs) find acceptable solu- tions to VRP. Besides, local searches optimize the GA solutions and hence reduce computing runtime. In this study, three al- gorithms for local search are compared: (1) Tabu search [11], (2) a threshold accepting approach, and (3) a multi-start lo- cal search based on cross exchanges [1]. Instances taken from a real-world case study of water distribution from the “Jun- ta Municipal de Agua y Saneamiento de Ciudad Jua ́rez” are considered for this analysis. Further results show that imple- mented local search algorithms reduce the traveled distance of the initial solutions.