Xiaojin Zheng, Meixia Yin, Yanxia Zhang

Transportation Research Part B 2019，Vol. 121

Recommend reason

The main contributions of this paper are as follows. Firstly, researchers develop an exact method for the integrated SCND model in the framework of the Generalized Benders Decomposition (GBD). The framework of the GBD method and the way of obtaining the Benders cuts can also be applied to the problems involving location-routing problem and location-inventory problem. Secondly, researchers employ the dimension-reducing principle in our solving process. The optimum solutions obtained from the classical model may be infeasible in real-life application. Researchers introduce the real world constraints to characterize the real-life problem much more precisely. These constraints can also help us to figure out the solutions which are infeasible in real world.

About the author

Xiaojin Zheng: Professor & Doctoral Supervisor, Tongji SEM.

Meixia Yin: Postgraduate, Tongji SEM

Yanxia Zhang: Associate Professor, Tongji SEM

Keywords

Supply chain network design; Location; inventory and routing; Integrated optimization; Generalized benders decomposition

Brief introduction

The supply chain network design (SCND) decision has significant impact on the performance of the supply chain (SC) because it affects total inventory and transportation costs in a long run. To create a long-term strategy for designing and planning SC, companies target for a global and integrated system by integrating the facility location decisions with inventory decisions and routing decisions in SCND problem. There have been effective methods to solve the SCND problem with the integration of location and inventory decisions. The size of the SCND model which comprises routing decisions is too large to be solved in reasonable time with current computing power. The difficulty of solving the model hampers the integration of routing decisions together. The major obstacles of the integrated optimization are the non-linearity and the huge amount of calculation caused by routing.

In this paper, we propose an integrated model to jointly optimize the location-inventory-routing decisions when we design SC network. We introduce the real-world constraints into our model to make the model portray the real-world problem precisely. We target for solving problem with exact method efficiently. In the classical model for our computational experiment, there will be more than 3 million variables in the test example. By introducing the two kinds of real-world constraints, the number of effective constraints in our test example decreases to 600 thousands. In the algorithm design, we apply Generalized Benders Decomposition (GBD) method to solve the model. The Benders cuts in our framework can be obtained with little effort. It can also be applied to the problems involving location-routing problem and location-inventory problem. Finally, we test the efficiency of the model and the method base on a passenger car supply chain network design case.

We get the following management insights which are helpful in dealing with real-world SCND problem. (1) The ordering management cost is not an indispensable part in the integrated SCND model if the company can control it efficiently. (2) The reducing of the lead time has limited influence in the structure of the SC network. (3) Compared with the partially integrated approach, our integrated model has cost superiority.