Impact of risk preferences on evacuee behavior and attention distribution in urban underground space evacuations

• Main Authors: Wei, Yixuan, Zhang, Yixing, Xu, Yifei, Wang, Shu, Liu, Jianguo, Jin, Longzhe, Ou, Shengnan, Pan, Song, Liu, Yiqiao
• Format: Article
• Published: Elsevier 2024
• Subjects: TA Engineering (General). Civil engineering (General); TH Building construction
• Online Access: http://eprints.um.edu.my/45416/; https://doi.org/10.1016/j.physa.2024.129698
• Institution: Universiti Malaya

Evacuation and conducting emergency rescue operations in underground spaces pose substantial challenges. However, limited work has been done to investigate how evacuees' heterogeneity influence the evacuation efficiency. In this study, we conducted virtual environment experiment and on-site experiment to analyze the physical behavior and attention distribution of evacuees with different risk preferences during route selections. A total of 30 participants were clustered into three groups: Risk Averter (RA), Risk Neuter (RN) and Risk Seeker (RS), based on the proportion of choosing safe detours. Eye metrics analysis revealed that RA allocated more time to gaze at risk elements, whereas RS tended to pay more attention to the route distance (P =0.003). The total fixation time for RA and RS was about 300 ms longer than RN (P =0.007), indicating that RN exhibited more cautious behavior pattern when selecting an evacuation route. In addition, all participants displayed a noticeable gaze bias towards potential choices during the decision-making process. Specifically, gaze likelihood increased and exceeded 50% at 300-400 ms before decisions were made. Additionally, evacuation models were established based on AnyLogic to investigate the influence of evacuees' risk preference on evacuation efficiency. The evacuation simulation results demonstrated a significant improvement in evacuation efficiency when all evacuees adopted a risk-averse approach. This resulted in a reduction of evacuation time by 12.07-28.84% compared to the benchmark condition. The findings offer valuable insights into the behavior prediction in the evacuation simulation and emergency management in urban underground spaces.