A fast and effective insertion algorithm for multi-depot vehicle routing problem with fixed distribution of vehicles and a new simulated annealing approach

Abstract

Multi-depot Vehicle Routing Problem has great practical value to the logistics and supply chain management. The fixed distribution of vehicles variant (MDVRPFD) brings it one step closer to the practical use. Based on the simple fact that all sub-routes of an optimal route must be optimal, a new randomized best insertion (RBI) algorithm Is proposed. The proposed insertion algorithm is highly effective in minimizing number of vehicles and fast. Compared to the best known result published for MDVRPFD variant, the solutions generated by this new insertion algorihtm require 20% fewer vehicles. Adopting a generalized n-op neighborhood operator, a Simulated Annealing approach yields a reduction of 12% in total distance compared to best known results of MDVRPFD. © Springer-Verlag Berlin Heidelberg 2006.