Construction of a new robust dual-channel supply chain network with forward logistics and reverse logistics

Abstract

Dual-channel closed-loop supply chain (CLSC) is a new supply chain model that integrates the offline and online sales channels into a CLSC. Dual-channel CLSC involves more echelons, which often leads to uncertainty in the dual-channel CLSC design. Although some studies have focused on uncertain dual-channel CLSC designs, there are still research gaps in the existing literature, such as a lack of robust models, inefficiencies in existing designs, and limited attention to certain industries. To fill these research gaps, we examine the design of the CLSC network based on partially known distribution information about uncertain transportation costs and demand. We construct a moment ambiguity set for uncertain costs and a bounded unimodal ambiguity set for uncertain demand. Using these ambiguity sets, we develop a new distributionally robust optimization (DRO) model with ambiguous chance constraints (ACCs) for the dual-channel CLSC design problem, which can be reformulated as a computationally tractable mixed-integer linear programming (MILP) model. To enhance solving efficiency, we build an accelerated Benders decomposition (BD) algorithm for the MILP model. A case study on BYD automotive illustrates the effectiveness of our developed approach. The computational results demonstrate that our proposed model withstands uncertainties in transportation costs and demand with a probability of 95%, establishing a dual-channel CLSC network that balances robustness and cost-effectiveness, resulting in a total cost reduction of approximately 14% compared to the deterministic model. Additionally, the sensitivity analysis reveal that the adjustment of the ACC tolerance level, the proportion of online and offline sales, and the handling methods of products significantly influence BYD Auto.