A case-driven simulation-optimization model for sustainable medical logistics network

The supply chain industry represents one of the largest and most critical sectors worldwide, and it is undergoing substantial transformation with the increasing integration of Electric Vehicles (EVs). In particular, EVs are being adopted within healthcare logistics networks to substantially mitigate carbon emissions and counteract escalating fuel costs, thereby enhancing the alignment of supply chain operations with broader public health and environmental sustainability objectives. This study proposes a novel Sustainable Healthcare Supply Chain Network (SHSCN) model that explicitly incorporates the deployment of EVs for the distribution of medical products and the optimal siting of Charging Stations (CSs) to support their operation. To quantitatively assess the queuing behavior of EVs at these charging facilities, an M/M/c queuing model is employed, providing insights into system performance in terms of vehicle waiting times. Additionally, the Simulation Method (SM) is utilized to estimate optimal fleet sizes and operational parameters. The validity and practical applicability of the proposed mathematical framework are demonstrated through a case study conducted within the medical industry context, employing the augmented ε-constraint method to handle the model's multi-objective nature. Given the NP-hardness of the formulated optimization problems, two novel hybrid metaheuristic approaches are introduced: Hybrid Simulated Annealing integrated with K-Medoids clustering (HKMSA), and Hybrid Tabu Search combined with K-Medoids clustering (HKMTS). Computational results indicate that both HKMSA and HKMTS exhibit superior performance relative to alternative methods, particularly in terms of solution quality and computational efficiency across problem instances of varying scales. Sensitivity analyses further reveal that a 30 % reduction in demand results in increases in all three objective function values, reaching 458,369, 894,100, and 761,790 units, respectively. Conversely, a 30 % improvement in service rate leads to a reduction in the first objective function's cost from 450,984 to 407,369 units.