Robust building evacuation planning in a dynamic network flow model under collapsible nodes and arcs

Rapid urbanization has caused various social problems. One typical example is the high population density of a building, particularly in a commercial building or a mega-mall. When an emergency, such as a natural or human-made disaster, occurs in a building with a high population, establishing a proper evacuation plan is required to minimize casualties. Accordingly, the evacuation planning problem, which determines optimal routes for evacuees from disaster-prone areas to safe areas, has been actively studied in various fields. However, research considering the possibility of further collapse of a specific area or intermediate route in the building has been overlooked. We propose a robust evacuation planning problem based on a dynamic network flow model that determines the optimal routes for evacuees from a building that has the potential to collapse. Computational results show that routes passing through areas with the potential to collapse may or may not be optimal for evacuees, depending on the given timeframe. If the timeframe is sufficient, detouring around the collapsible areas could be the optimal plan; however, if the timeframe is insufficient, passing through collapsible areas, with taking the risk, could be the optimal plan.