Optimising Shelter Locations for Bus Evacuation and Relief Supply Under Traffic Congestion

Effective disaster management requires shelter location optimisation to enhance evacuation efficiency and ensure timely relief distribution. This study integrates human evacuation and relief logistics while accounting for traffic congestion during large-scale evacuations, thereby proposing a model that prioritises bus-based evacuation to mitigate congestion and expedite movement, particularly for transit-dependent populations. Employing a metaheuristic evolutionary algorithm with a local search process, the model is applied to a flood scenario in Ulsan, South Korea and significantly outperforms alternative methods in optimising shelter placement, transportation routes and relief supply distribution. Comparative analysis indicates that the proposed shelter locations reduce total costs by 9.4% relative to manually selected nearest shelters. Additionally, neglecting network congestion was found to underestimate evacuation time by up to 41%. The proposed approach also reduces relief transportation costs by 4.5%. Sensitivity analysis examines the impact of bus availability and evacuation demand variations. This study is the first to fully incorporate city-wide traffic congestion into shelter location optimisation under multimodal evacuation scenarios.