Multi-objective Optimization of Vehicle Routing Problem with Time Windows

Products distribution is a process on which products are moved from sources to customers. According to Toth and Vigo [1], transportation costs typically represent 10-20% of the final price of goods on the market. Over the past 20 years, Vehicle Routing Problem (VRP) was especially solved through the use of meta-heuristics, see Jozefowiez et al. [2], Vidal et al. [3]. Eksioglu et al. [4] made a taxonomic review of VRP defining this research field and making a fully detailed classification of variants with many examples. After the review of previous classifications and taxonomies, major journals that publish articles on this Abstract: This paper focuses on the importance of multi-objective optimization in vehicle routing problem. The main objective of this paper is to do multi-objective optimization of Solomon 100 customers 56 vehicle routing problems independently of problem type and parameters. To perform this work, firstly our mathematical model is remembered; secondly for each problem our Savings-insertion builds a good initial solution and thirdly our Reactive Tabu with a Variable Threshold improves the initial solution. The three objectives are the minimization of a number of routes, the minimization of total distance and the minimization of total time by respecting the time window and the demand of all customers. Finally, the experimental results obtained with our methodology for Solomon 100 customers 56 vehicle routing problems are showed and discussed. Our results show that the multi-objective optimization is more beneficial than the total distance minimization.