Multi-criteria fuel distribution: A case study

Abstract

In this paper a multi-criteria fuel distribution problem similar to the common CVRP is considered with the real-life coordinates of gas stations obtained from a certain petrol company. The optimized criteria are the total distance of all tours and the number of tours. A certain method of solution representation, which ensures a feasible solution was used along a decoding scheme. A simulated annealing (SA) metaheuristic algorithm was implemented in order to obtain the approximations of the Pareto set for a number of instances with varying number of gas stations and demands for each station. The results show 17-18% improvement over the starting solution in all cases for both criteria. Moreover, the number of found solutions per instance increases with the number of gas stations. The improvements can be translated directly into profit, which was shown as well.