Multi-objective Priority Planning for Dam breach Flood Emergency Evacuation Routes Based on an Improved NSGA-III Model

Abstract

In recent years, global warming has intensified extreme precipitation events, significantly increasing the risk of dam breach floods. Once a dam breach occurs, the unleashed massive floodwaters can cause devastating damage to downstream populations. Therefore, simulating dam breach scenarios and formulating scientifically grounded emergency evacuation routes is of critical importance. However, existing research on evacuation route planning under extreme dam breach flood scenarios remains limited, especially in terms of evacuation optimization strategies that differentiate by risk level. This study employs the HEC-RAS hydrodynamic model to simulate dam breach floods, delineate flood affected zones, and classify them into high-, medium-, and low-priority areas based on impact severity to facilitate hierarchical evacuation planning. The approach begins by precomputing and caching the shortest evacuation paths for all affected areas using the Dijkstra algorithm. A K-means clustering initialization is then applied to preferentially assign evacuation zones to nearby shelters. Following the generation of initial multi-objective solutions via NSGA-III, a heuristic refinement strategy is introduced to further optimize evacuation paths, ensuring that populations in high-priority areas are evacuated more efficiently. To address sudden changes in shelter capacity, a secondary evacuation reassignment algorithm based on preprocessed path data is introduced, allowing for adaptive and flexible adjustment of evacuation plans. The proposed model significantly enhances evacuation efficiency and practical emergency response capacity, particularly improving the protection and dispatch of populations in high-risk areas. The findings provide valuable decision-making support for dam breach disaster emergency management.