Simulação estocástica e métodos heurísticos paralelizados para a resolução do problema de roteamento de veículos capacitados com base na estratégia “cluster first route second”

Abstract

The Capacitated Vehicle Routing Problem (CVRP) seeks to determine a set of routes for a homogeneous fleet of vehicles, from a central warehouse bound to a set of customers who demand a certain product. As one of the most important and complex problems related to the distribution logistics, heuristic approaches to this problem are constantly presented. In this way, this work addresses the heuristic approach based on a strategy "Cluster First Route Second" which consists primarily of grouping customers according to their demands for later build routes for the formed groups. The approach uses Monte Carlo simulation in conjunction with the classic algorithm Gillett and Johnson, 1976 to solve the clustering problem, while the routes for each group are constructed by the nearest neighbor heuristic with refinements coming from 2-opt exchanges. It is also employed parallel processing techniques to reduce the computational time.