A stochastic decision model for emergency logistics using hybrid metaheuristic methods

Earthquakes pose major challenges for emergency logistics because of their unpredictable nature and the destruction they cause to infrastructure and demand. This study introduces a two-stage model for earthquake response, combining Particle Swarm Optimization (PSO) and Simulated Annealing (SA) in a hybrid algorithm. In the first stage, decisions like facility location and fleet allocation are made, while in the second stage, routing and inventory distribution are adjusted once disaster scenarios unfold. This two-stage approach integrates both strategic and operational decisions, ensuring preparedness while allowing for flexibility—a crucial aspect often missed in other models. The model also considers practical constraints, such as vehicle capacities, split deliveries, and damaged transportation networks. A case study in Palu City, Indonesia, shows that the hybrid PSO-SA method leads to lower logistics costs and faster computations compared to using PSO or SA alone. Sensitivity and robustness tests prove that the method can adapt to different scenarios and parameters. The unique contribution of this study is the combination of stochastic programming with a hybrid PSO-SA algorithm to manage both preparation and response under uncertain conditions. From a practical perspective, the findings provide useful insights for humanitarian agencies on how to pre-position resources and design flexible routing strategies, improving logistics adaptability and enhancing disaster response. • Propose a hybrid metaheuristic model to improve logistics decisions in earthquake relief efforts. • Develop a two-stage stochastic framework for managing uncertainty in disaster scenarios. • Integrate realistic logistics constraints into a scalable decision-support system. • Demonstrate superior performance of the hybrid method using real disaster case data. • Validate model robustness through sensitivity analysis across varying demand and damage levels.