Metaheuristic approaches to a vehicle scheduling problem in sugar beet transportation

Abstract

A variant of vehicle scheduling problem (VSP) arising from the sugar beet transportation in a sugar factory in Serbia is introduced. The objective of the considered VSP is to minimize the required transportation time under problem-specific constraints. The problem is formulated as a mixed integer linear program (MILP). Within the framework of commercial CPLEX solver the proposed MILP model was able to produce optimal solutions for small size problem instances. Therefore, two metaheuristic methods, variable neighborhood search (VNS) and greedy randomized adaptive search procedure (GRASP), are designed to solve problem instances of larger dimensions. The proposed GRASP and VNS are evaluated and compared against CPLEX and each other on the set of real-life and generated problem instances. Computational results show that VNS is superior method with respect to the solution quality, while GRASP is able to find high quality solutions within very short running times.