Graph-Based Decision Support System for Last-Mile Humanitarian Logistics: The case of Peru

This study focuses on optimizing last-mile routes to ensure complete node coverage while minimizing arrival times. The proposed model is validated by a case study simulating a major seismic event and subsequent tsunami in Callao, a critical port city in the metropolitan area of Lima, Peru. The optimization model was executed using the validated instance of 832 POIs. Computational behavior differs significantly between the unconstrained geometric approach and the proposed graph-based network routing model. This approach assumes an unconstrained plane and calculates simple straight-line connections that disregard the physical reality of the urban grid. The empirical findings confirm the theoretical challenges associated with combinatorial optimization in large-scale emergency management and are analyzed within the classical vehicle routing problem (VRP) optimization, which maximizes commercial profits. By embedding realistic routing distances into the system constraints, the model aligns the actual workload with localized distribution center capacities and regional population demands. This integration provides emergency coordinators with an actionable and dependable decision-support framework, ensuring that computed supply sequences remain physically viable under real-world post-disaster conditions.