A tool to predict crowd turbulence

Until now, models for crowd dynamics were mainly constructed based on analogies with physicochemical systems. The behavior of a pedestrian was formalized using a combination of the forces attracting him to his destination and the forces repelling him from other individuals and obstacles. However, these models are difficult to calibrate and only imperfectly reflect reality. To overcome these problems, Mehdi Moussaïd and Guy Theraulaz from the Centre de Recherche sur la Cognition Animale (CNRS / Université Toulouse 3-Paul Sabatier), working in collaboration with Dirk Helbing at the Ecole Polytechnique Fédérale in Zurich, have proposed a novel approach based both on cognitive sciences and the physics of complex systems that closely combines experimentation and modeling.

Their model suggests that a pedestrian seeks simply to minimize congestion in his visual field by walking towards the empty spaces he can see, while at the same time adjusting his speed in order to maintain a safe distance from the nearest obstacle. Digital simulations using this model have demonstrated that these two simple rules are sufficient to reproduce a broad range of the collective behaviors observed in crowds, such as the spontaneous formation of unidirectional lanes in opposite directions. Furthermore, as the density of pedestrians increases, the model can predict the emergence of new phenomena, such as the accordion effect characterized by successive forward waves of movement, interspersed with periods during which the pedestrians stand still (stop-and-go). Above a critical density threshold, a combination of these rules with the effect of physical contacts between pedestrians can spontaneously provoke gigantic, collective crushes. This phenomenon, referred to as turbulence, was observed during the accidents that occurred in Mecca in 2006 and characterizes the dynamics of a crowd in a dangerous situation, where pedestrians are overwhelmed by chaotic movement.

This recent work enables a clearer understanding of the dynamics of a moving crowd and opens the way to the development of new risk planning tools. For example, the model designed by these scientists can identify potentially dangerous areas in an environment where are large number of people may gather. It may therefore help urban planners in developing pedestrian precincts in town centers, or engineers to design public buildings (stadiums, concert halls, stations, etc.), or even assist security experts during the organization of major events.