Modified social force model for emergency evacuation: Non-adaptive behaviors shaped by heterogeneous emotional intensity

During unconventional emergencies, significant psychological and behavioral heterogeneity exists in crowds, where high emotional intensity may trigger non-adaptive behaviors (e.g., colliding and pushing). To investigate the impacts of heterogeneous emotional intensities and non-adaptive behaviors, this study proposes a modified social force model that, while preserving individual emotional heterogeneity, uses the “initial emotional mean” to capture the crowd’s overall emergency emotional intensity and couples non-adaptive behaviors induced by high emotional intensity. Velocity entropy is employed to assess the orderliness of crowd movement, and a computational approach to “spatial hazard degree (SHD)” is developed to reveal the dynamic changes in high-risk areas. Furthermore, the correlation mechanism among “evacuation orderliness, potential injury risk, and high-risk area evolution” is comprehensively analyzed. The simulation results show that emotional intensity exerts a dual-directional effect: moderate emotions are conducive to improving evacuation efficiency, while excessive emotions are detrimental to it, and pedestrian potential injury risk rises with increasing emotional intensity. Non-adaptive behaviors significantly exacerbate crowd chaos, reduce evacuation efficiency, increase potential injury risk, expand the scope of high-risk areas, and shift the maximum risk area from both sides of the exit to the interior of the system. Density is the critical factor that amplifies the effects of emotions and behaviors. Additionally, this study identifies an emotion-behavior-driven dynamic transmission chain, namely “impaired orderliness → aggravated congestion → increased potential injury risk → formation, expansion, and migration of high-risk areas,” which clarifies the relationship among emotional states, individual behaviors, and emergency evacuation. This research provides targeted theoretical support for the emergency safety management and guidance of heterogeneous pedestrians in emergency scenarios.