Integration of cumulative prospect theory in cellular automata model for building evacuation

Abstract

An inefficient evacuation system can lead to high levels of fatalities arising from natural or anthropogenic disasters. As the movements of and interactions between evacuees can greatly affect the overall evacuation patterns and times, in-depth investigations of evacuees’ behaviours and interactions are warranted. The cellular automata (CA) model, widely used to simulate building evacuations, determines the moving direction of evacuees (i.e., they either remain stationary or move to a neighbouring cell) and provides agents in the model with rational decision-making choices. However, people are not always rational in real-life risk emergencies. Addressing this inconsistency, cumulative prospect theory (CPT) was developed to make the most prospective decision under risk or uncertainty by applying weighting functions to transform the outcomes and probabilities of the available options and combine them into prospects. This approach has been reported to realistically model human decision-making behaviour. At each time step of the evacuation, each evacuee is involved in the decision-making of the moving path, which critically affects the evacuation process and efficiency. Thus, CPT can be applied to model the movement direction. In our study, CPT was integrated with the CA model (CPT-CA) to mimic exit choices and route choice behaviours. Simulation results showed that the CPT-CA model can provide near-realistic evacuation patterns and evacuation times even when the sensitivity parameters are not adjusted. The proposed CPT-CA model can be helpful for logistical planning in evacuation design and emergency risk management.