A three-echelon multi-commodity location-routing problem

Abstract

This research studies the problem of distribution network design. The purposes of this study are 1) to develop the mathematical model for redesign of current distribution network by focusing on reducing total distribution cost and 2) to develop new solution approach for large-scale complicated problem. This research formulates mixed integer linear programming for the three-echelon two-commodity Location Routing Problem (LRP). The objective function is to minimize facility operating and closure cost and distance cost. Due to large-scale of LRP, which is NP-hardness, this research proposes new sequential solution approach as following steps; 1) decomposing the LRP into two subproblems based on its structure, 2) establishing route by clustering algorithm and identifying shortest route from Traveling Salesman Problem (TSP), and 3) allocating proper warehouse and service center. The proposed model is tested by using data of distribution network from actual case, which consists of five distributed zones. In analyzed phase, both of the distribution network redesign for each particular zone and allowing the distribution across all zones are considered. Additionally, this research conducts sensitivity analysis on each problem. Totally, there are 66 scenarios to perform. The result indicates that solving all zones simultaneously contributes lower cost than solving each zone separately. The proposed solution provides better answer than solving the problem by exact approach, which gives lower cost with shorter computation time. Moreover, the solutions of the proposed model are tested in stochastic environmental by simulation technique. The results show that the redesign of distribution network by proposed model can provide better quality of answer than the current network.