A Hybrid Solution Approach Based on Benders Decomposition and Meta-Heuristics to Solve Supply Chain Network Design Problem

Supply Chain Network Design (SCND) is a strategic supply chain management problem that determines its configuration. This mainly focuses on the facilities location, capacity sizing, technology selection, supplier selection, transportation, allocation of production and distribution facilities to the market, and so on. Although the optimal solution of the SCND problem leads to a significant reduction in the supply chain costs, but placing this problem in the NP_Hard order makes it impossible for some networks (especially large scale) to achieve the optimal solution using classical algorithms. In this research, we present a hybrid solution method based on the benders decomposition and genetic algorithm for a four-echelon SCND problem. The proposed approach inherits the run time efficiency from the metaheuristics and decomposition method, and ensures convergence to the optimal solution using Banders method. We consider major design and planning decisions in the SCND problem, to provide a more comprehensive model and solution approach which is compatible to the real supply SCND problem. To evaluate performance and effectiveness of the proposed hybrid benders decomposition and genetic algorithm (HBDGA) approach, some random test problems are generated in various scales. Numerical results show that the proposed HBDGA solution approach can overcome the speed weakness of the classic benders decomposition. In addition, proposed HBDGA, unlike meta-heuristic methods, guarantees global optimization of the SCND problem.