Multiobjective optimization for multiperiod reverse logistics network design

In recent years, the ever-rising return streams for repair service have forced the electronics manufacturers to expand their reverse logistics capacities. However, most existing papers on the reverse logistics network design neglected the time sensitivity of the return flows. Moreover, most of these investigations were primarily concerned with the single objective problems of either minimizing the total cost or maximizing the profit. In this paper, we propose a biobjective mixed-integer linear programming model for the multiperiod design problem of a reverse logistics network for repair service. A multiperiod setting is taken into account to make the reverse logistics network flexible to accommodate the gradual changes in the capacity of the facilities and the network configuration. To solve the NP-hard problem with biobjective, we develop a hybrid evolutionary algorithm that combines nondominated sorting genetic algorithm II (NSGA-II) with a local search method. We compare the hybrid evolutionary algorithm with NSGA-II and ϵ-constraint method using numerical examples. The comparison results indicate that the hybrid evolutionary algorithm outperforms the NSGA-II in most cases. The ϵ-constraint method performs best for the small instances, but it cannot solve large instances within reasonable time. Finally, an extensive parametric analysis is conducted and several managerial insights are derived