Risk analysis of people evacuation and its path optimization during tunnel fires using virtual reality experiments

Abstract

The number of urban tunnels has been increasing rapidly, accompanied by frequent tunnel fire accidents owing to the complex tunnel structure and large traffic flow. In this study, a full-size tunnel virtual reality (VR) scenario and computational fluid dynamics (CFD) construction model were established to investigate the evacuation behavior and corresponding risk of people in the early stage of vehicle fires considering four scenarios: normal circumstance, without VR agents, without emergency evacuation signs, and without fire extinguishers. Firstly, the cumulative values of CO, CO2, and temperature along the evacuation path were monitored using CFD. Secondly, the smoke toxicity was calculated using the N-GAS model, and the total risk value was computed based on the analytic hierarchy process which was defined as “smoke hazard: temperature hazard = 7:3.” Thirdly, a multiple regression model was created based on accident data. Finally, to minimize accidents, the design of the evacuation path was optimized using the established mathematical model and A* algorithm to verify the effectiveness of the risk assessment model. The results show that the effects of VR agents, emergency evacuation signs, and fire extinguishers on the evacuation behavior of people are mutual influence. This study combined time and route in the VR evacuation experiments to overcome the limitations of the existing control and VR experiments in quantifying the evacuation results. This research can be utilized to improve emergency evacuation plans and emergency response decision making. Furthermore, it can broaden the application of VR in the field of tunnel lifecycle safety management.