Dynamic Evacuation Path Planning for Multi-Exit Building Fire: Bi-Objective Model and Algorithm

Building fires are one of the most common fire types and tend to cause heavy casualties. During the evacuation, the shortest path makes the escapees leave the accident scene faster. Meanwhile, the safety of escapees in the evacuation process cannot be ignored. Optimal path should both ensure evacuation efficiency and maximize the safety of escapees. A bi-objective path planning model, which considers both path risk and path length, is presented. Path risk is based on five risk factors that are likely to cause casualties in a fire: carbon monoxide (CO), hydrogen cyanide (HCN), temperature, visibility, and crowding. The linear weighted sum method is used to convert the risk-objective model and the length-objective model into a dynamic bi-objective path planning model. The Dijkstra algorithm is modified to solve the model. The modified algorithm outperforms the traditional Dijkstra algorithm in terms of both efficiency and adjustment ability. The simulation analysis of a building fire shows that the bi-objective model and algorithm can plan a combined optimal evacuation path for escapees considering risk and path length, which avoids the area with high risk level and optimizes the evacuation path length.