Introduction: Human crowd crushes are catastrophic events where dense gatherings can lead to severe injuries and fatalities. Understanding crowd behavior and biomechanics is crucial for developing effective prevention strategies, such as eliminating bottlenecks and ensuring ingress and egress. Given the infeasibility of real-world trials, simulations are useful to examine crowd disasters. This study reports the results of a simulation that analyzes human behavior and injury patterns during a crowd crush by incorporating variable crowd densities and trajectories using ragdoll physics principles. Methods: This simulation was developed with ragdoll physics using the Unity Simulation Engine (Unity Technologies, 2023, San Francisco, USA). A custom-built virtual environment modeled a crowd crush incident and included three groups, each consisting of 500 individuals, converging from the left, right, and perpendicularly in a “T” junction. Pedestrians walking at speeds averaging 1 meter per second were randomly assigned. A series of 30 simulation trials was run with 1,500 pedestrians per trial. The number of injuries and fatalities was recorded. Results: A crowd simulator was successfully generated, allowing users to create groups with specific sizes, routes, and speeds, with visualization of humanoid subjects. The mean number of injuries was 191.6 (±18.75 SD), with a median of 191.5 injuries per trial and an interquartile range (IQR) of 182 to 206.5 injuries. For fatalities, the mean was 63.1 (±8.56 SD), with a median of 62 fatalities per trial and an IQR of 55.25 to 68.75 fatalities. Conclusion: The ragdoll physics simulation provides valuable insights into biomechanics, injuries, and fatalities associated with human crowd crushes. This tool has the potential to help improve crowd management protocols. Future research will focus on refining the model and its application in real-time crowd monitoring systems.
Chang et al. (Sun,) studied this question.