A bi-level programming methodology for decentralized mining supply chain network design

Since the outbreak of the COVID-19 pandemic, reshaping global supply chains for bulk mining commodities, such as coal, copper, and iron ore, has posed significant challenges. The complexity and multi-stakeholder nature of mining supply chain network design (MSCND) require innovative optimization approaches. However, traditional literature often focuses on centralized MSCND strategies, neglecting the competitive dynamics and conflicts of interest among stakeholders. To address this gap, this study introduces a bi-level programming (BLP) model for decentralized MSCND, capturing interactions between upper-level ore production and lower-level ore processing enterprises. To overcome the computational complexity of the BLP model, we develop a novel hybrid math-heuristic algorithm called Sine Cosine and Differential Evolution Algorithm with Constraint Repair Mechanism (SCDEA-CRM). The proposed SCDEA-CRM integrates the search mechanisms of sine cosine and differential evolution algorithms, along with a novel constraint repair mechanism to fix infeasible solutions caused by chemical composition imbalances between raw ores and products. Numerical experiments demonstrate the SCDEA-CRM's superior performance in solving the BLP model. A real-world case study in a decentralized iron ore supply chain validates the model's practical applicability and highlights its advantages over the centralized counterpart model. A sensitivity analysis is conducted to assess the impact of product iron content variations on supply chain costs.