Multiple depot vehicle routing problem with split delivery

Abstract

© 2019, Computers and Industrial Engineering. All rights reserved. In this paper we present a modified mathematical model for the multiple depot vehicle routing problem with split delivery (MDSDVRP). We also present an algorithm for clustering and solving the model. Each cluster includes customers who have the smallest total distance to its depot and is solved under the condition that a customer can be visited by multiple vehicles. Computational experiments are carried out on 15 benchmark instances that have 50-360 customers and are taken from six related works that proposed different heuristics for solving MDVRP. The aim is to test the effectiveness of split delivery in MDVRP by comparing our results with existing best solutions from all six authors. Our approach is able to improve the solution of 9 instances including 5 better solutions compared with the latest work and 4 best solutions.