The parameter calibration and optimization of social force model for the real-life 2013 Ya’an earthquake evacuation in China

• Published in: Safety science 2015-11, Vol.79, p.243-253
• Main Authors: Li, Meifang, Zhao, Yongxiang, He, Lerong, Chen, Wenxiao, Xu, Xianfeng
• Format: Article
• Language: English
• Subjects: Classrooms; Computer simulation; Differential evolution; Disasters; Earthquake evacuation; Earthquakes; Evacuation; Nonlinearity; Panic; Parameter calibration; Pedestrian evacuation; Pedestrians; Social force model
• ISSN: 0925-7535; 1879-1042
• DOI: 10.1016/j.ssci.2015.06.018

[Display omitted] •The escape panics in real-life 2013 Ya’an earthquake are simulated and reproduced.•Differential evolution is used to calibrate the parameters of social force model.•Trained evacuation leader is beneficial to maintain the calm and order of crowd.•Different layouts can only change the partial flow in the left and right corridors.•The evacuation efficiency has almost doubled when the back door is open. Due to the fact that escape panics are unexpected and dangerous, which excludes real-life experiments because of technical difficulties and ethical reasons, the evacuation simulations in real-life disasters are still rare. In this paper, the escape panics of classroom evacuation in real-life 2013 Ya’an earthquake in China are simulated and reproduced using the social force model. Firstly, a robust differential evolution (DE) is employed to calibrate and optimize the parameters of social force model to achieve a desired nonlinear evacuation speed which is consistent with real-life video data. It is demonstrated the DE-calibrated social force model can reproduce the characteristics of pedestrian flow in the real-life earthquake evacuation including the nonlinear evacuation speed curve and the pedestrian position distributions at different time. Moreover, the trained evacuation leader has remarkable impact on the evacuation process which can be beneficial to maintain the calm and order of crowd, decrease the desired velocity of crowd, and thus avoid fatal accident to the most degree. Finally, different layouts of classroom have similar evacuation efficiency because different internal layouts can only change partial flow of pedestrians in the left and right corridors, and total flow aggregating in the exit of classroom is still unchanged. However, total evacuation time can be sharply reduced to 17s from above 36s, and evacuation efficiency has almost doubled when the back door is open, which means we should try to keep all the emergency exits open in order to provide the crowd with more chances to escape when disasters occur.