Crowd movements are not chaotic but predictable
Crowds intimidate and even terrify some people. Their movements can lead to tragedy. That's why a better understanding of them is essential. A new study shows that these movements are not as chaotic as one might imagine, but, on the contrary, almost circular and periodic.
François Gu, Massachusetts Institute of Technology (MIT) and Benjamin Guiselin, University of Montpellier
You've already had the experience of being in the middle of a compact crowd in a confined space: on crowded subway platforms at rush hour, in front of a store for the release of a best-selling author's latest book, or in front of the stage at a concert. Beyond the discomfort caused by frequent involuntary physical contact with your neighbors, these situations seem uncontrollable and potentially dangerous: you feel compelled to move according to a pattern dictated by impatience or pressure from others. But what is the true nature of individual movement in a dense crowd? And can we understand their origins, in order to anticipate tragedies?
If our intuition is anything to go by, these movements seem random and unpredictable. However, our study, conducted in the team of Denis Bartolo, professor at ENS Lyon, and recently published in the journal Nature, reveals a counter-intuitive phenomenon: instead of disorderly chaos, crowds collectively move in a regular, spontaneous pattern. Above a critical density of four people per square metre (imagine four people in a shower cubicle!), and with no external instructions, the crowd spontaneously adopts an almost circular, periodic motion.
Our experience: the festivals of Pamplona and its "Chupinazo".
Our first challenge in characterizing the dynamics of dense crowds was to set up experiments to film the dynamics of hundreds of individuals from a good angle, while avoiding accidents. Clearly, we couldn't do this in our laboratory. The ideal opportunity arose when Iker Zuriguel, a professor at the University of Navarre, told us about the San Fermín festivities in Spain. Every year, on July 6, around 5,000 people gather at the Plaza Consistorial in Pamplona for the "Chupinazo" ceremony, which marks the start of a week of festivities. The population density reaches around 6 people per square metre!
The square, which is 50 meters long and 20 meters wide, is bordered by multi-storey buildings, whose balconies provide an unobstructed view of what's going on in the square. During four editions of the event, we filmed the crowd's movements with high resolution, using eight cameras placed on two balconies. The result is a dataset that is unique in the world for the study of dense crowds.
The crowd sways in sync
Thanks to a technique used, for example, in aerodynamics, we were able to map the speeds of movement in the crowd, like following air currents around an airplane. We extracted that all individuals within a radius of around 10 meters were moving in the same direction.
During the "Chupinazo", the density of people is very high, on the order of six people per square metre: so you have to imagine around 500 people spontaneously dragged together, which represents a mass of several tens of tons in motion.
We also showed that the direction of motion of this mass rotated progressively, before returning to its starting point, every 18 seconds. In other words, individuals don't move chaotically, but follow quasi-circular, periodic trajectories.
This slow movement is explained by the fact that it's not isolated individuals who are moving, but several hundred, entrained with one another.
Finally, we have observed that the oscillating circular movements of crowds are as much clockwise as anti-clockwise, even though the majority of human beings are right-handed, ortend to avoid each other on the right in Western countries. Cognitive and biological factors are therefore no longer relevant to explain the movement of masses of individuals in dense crowds, which are driven by very large-scale movements.
The origin of spontaneous crowd oscillation
To mathematically model the dynamics of a crowd of pedestrians, it seems natural to consider individuals as interacting particles.
Take a pedestrian in this crowd. He is subject to forces that set him in motion. These forces may have a physical origin - such as the forces of contact with a wall or another pedestrian - or a cognitive origin - such as when trying to avoid another pedestrian. Unfortunately, mathematical modeling of these forces relies on many unverifiable assumptions about individual behavior, making this approach impractical.
In reality, it's not necessary to describe the dynamics of each individual to predict the dynamics of the crowd. Take the flow of water through a pipe: the laws of physics make it possible to predict the flow of water, while determining the force experienced by a single water molecule in this flow proves impossible.
We have therefore determined the equation that would describe the motion of a mass of individuals all driven together, without determining the laws that govern the motion of a single pedestrian. Our approach uses only fundamental principles of physics (conservation of mass, conservation of momentum) and makes no behavioral assumptions about the movement of individuals. It has enabled us to build a mathematical model whose resolution has shown excellent agreement with experimental observations.
A new method for preventing crowd accidents?
We also analyzed surveillance camera footage from the 2010 Love Parade in Duisburg, Germany. Although this crowd was very different from the "Chupinazo" crowd, we observed the same collective oscillations. This suggests that this mass behavior is universal, regardless of the type of event or the profile of the participants.
As we pointed out earlier, these oscillations can set tens of tons in motion. We believe that such uncontrolled mass movement can become dangerous. At the "Chupinazo", no accidents have ever been reported, probably because the event is short (one to two hours) and participants come willingly, with a certain awareness of the risks. This was not the case at the 2010 Love Parade, where an accident caused dozens of deaths and hundreds of injuries. Just before the accident occurred, we detected these oscillations.
This detection can be carried out in real time, based on direct and simple analysis of video surveillance cameras. And since this dynamic is universal, the same method could be applied to other crowds. So, in the future, our findings could inspire the development of tools for the detection and prevention of mass accidents.
François Gu, Post-doctoral fellow, Massachusetts Institute of Technology (MIT) and Benjamin Guiselin, Associate Professor of Physics, University of Montpellier
This article is republished from The Conversation under a Creative Commons license. Read theoriginal article.