Skewed general variable neighborhood search to solve the multi-compartment vehicle routing problem

Abstract

Skewed General Variable Neighborhood Search (SGVNS) is shown to be a powerful and robust methodology for solving vehicle routing problems. In this paper we suggest new SGVNS for solving the multi-compartment vehicle routing problem (MCVRP). The problem of multi-compartment vehicle routing is of practical importance in the petrol and food delivery and waste collection industries. A comparison between our algorithm and the memetic algorithm and the tabu search is provided. It was clear that the proposed algorithm is capable of solving the available instances. Skewed General Variable Neighborhood Search was used because it makes it easy to explore the space of realizable solutions for MCVRP. As a result, the SGVNS is much faster and more effective. It is able to solve 50 to 484 customers from the literature.

Keywords

• variable neighbourhood search
• vehicle routing problem
• multi-compartment vehicle routing problem