Green horizons: Sustainable global logistics in dynamic supply chain management

Abstract

Supply chain management in a global scale involves addressing numerous uncertainties, from demand fluctuations to unforeseen disruptions. Developing advanced solution approaches is critical to manage such complexities and ensure resilience. This study presents a multi-stage stochastic–dynamic model for the global supply chain, incorporating hedging policies. The aim is to identify optimal order scheduling for bill of materials, production planning, and inventory management across warehouses (i.e., materials and finished products). Due to the dynamic nature of the global supply chain (e.g., demand fluctuations, disruptions, and lead time), a multi-stage stochastic model is developed for the stochastic–dynamic supply chain network. To address dynamic factors of real-world global supply chain, an accelerated parallel stochastic dual dynamic integer programming (SDDiP) approach is proposed to deal with disruptions (e.g., political unrest, natural disasters, and pandemics), enhancing supply chain resiliency. To validate the proposed parallel SDDiP, various scenarios with different sizes are generated using the case study and compared to the SDDiP with Benders cuts and integrated stage-wise Lagrangian dual cut (SWLDC) (i.e., SDDiP-SWLDC). According to the obtained results, the proposed parallel node strategy for accelerated SDDiP consistently outperforms the basic stochastic dual dynamic programming (SDDP) and demonstrated robust CPU scalability. Evaluation across various scenario sizes shows stochastic dual dynamic integer programming-mixed integer rounding cuts (SDDiP-MIR) achieving faster computation and a smaller 7% optimality gap compared to SDDiP-SWLDC and SDDiP in large-size instances, highlighting its superior performance in complex supply chain settings.