Modeling Crowd Turbulence by Many-Particle Simulations

TL;DR

This paper enhances a social force model to simulate turbulent crowd behavior, successfully reproducing empirical features observed during high-density pedestrian flows such as pilgrimages.

Contribution

The authors extend the social force model's repulsive term to better simulate crowd turbulence, capturing transitions and empirical statistical features.

Findings

Successfully reproduces turbulent crowd features

Captures transition from laminar to turbulent flow

Aligns well with video data analysis

Abstract

A recent study [D. Helbing, A. Johansson and H. Z. Al-Abideen, {\it Phys. Rev. E} 75, 046109 (2007)] has revealed a "turbulent" state of pedestrian flows, which is characterized by sudden displacements and causes the falling and trampling of people. However, turbulent crowd motion is not reproduced well by current many-particle models due to their insufficient representation of the local interactions in areas of extreme densities. In this contribution, we extend the repulsive force term of the social force model to reproduce crowd turbulence. We perform numerical simulations of pedestrians moving through a bottleneck area with this new model. The transitions from laminar to stop-and-go and turbulent flows are observed. The empirical features characterizing crowd turbulence, such as the structure function and the probability density function of velocity increments are reproduced well, i.e. they are well compatible with an analysis of video data during the annual Muslim pilgrimage.