The psychology and the physics of stampedes

Why bother to think about stampedes?

As we know, even poorly-organized humans in groups have tremendous capacity for destruction of property and life. The raw power of collectivism has the innate potential to take a chaotic turn in the fury of the mob and the panic of the stampede. And by appreciating the fact that individuals think and act differently when in groups than when alone, calamities can be understood better and possibly averted.

Just before dawn on September 30, 2008, events unfolded at the Chamunda Devi temple in Jodhpur’s Mehrangarh Fort that left a tragic trail of death and destruction in its wake. That morning approximately 25,000 Hindu devotees approached the temple to mark the first day of Navratri. As soon as the door of the temple opened, there was a frenzied scramble of devotees trying to get inside. Within moments a stampede ensued in which, by most accounts, close to 250 pilgrims lost their lives and more than 400 others suffered various injuries. Agonizingly, this stampede was the fourth lethal one at a religious gathering in India that year. Just in the previous month, over one hundred pilgrims died in a stampede at the Naina Devi temple in Himachal Pradesh.

Why are locations in India particularly prone to human stampedes? A quick glance at a list of all the major stampedes that have resulted in a loss of life has revealed that most occurred at religious gatherings. Some of the most revered sites in Hinduism are in remote locations on hills and mountains. Accessibility has no bearing on the auspiciousness of a site and devotees take the physical strain of getting there in their stride as part of part of the pilgrimage.

Unfortunately, even in the most-accessible of locations there is always a possibility of things turning ugly. At stadiums, trained crowd-managers at stadiums look out for potential troublemakers. But even a simple spark can turn into a riot. The situation is substantially more complex at a remotely-located religious site, especially during peak pilgrimage seasons. Inaccessibility means that it is difficult to put into place adequate crowd management and safety measures. For example, by eyewitness accounts, the entrance to the Chamunda Devi temple was very narrow and there were no exits to escape out of in case of emergency.

A second major problem is that crowd-managers at events in which freely-moving crowds participate can only make preparations based on turnout estimates. Obviously, even these numbers can go awry. Those guiding temple devotees at best have a very nebulous idea of how many will turn up on any given day. In addition, the task of managing the crowd flow at religious sites is noticeably difficult because these crowds are never homogeneous. There will always be children and the elderly who will have special needs in the unfortunate case an evacuation is required.

Finally, there are the “black-box” parameters that can perturb even the best-managed crowds. Changes in weather, accidents, fires, spreading rumors, and changes in mood have all known to cause escape panic.

For all structures hosting mass events, architects and structural engineers need to consider the psychology of individuals in a group in order to be able to design effective emergency exits to prevent avoidable calamities. Sadly, only recently has crowd psychology gained traction as a practical consideration in the design of structures. Until recently, the primary guiding principle in the design of structures for mass use was fluid dynamics, the branch of physics dealing with the flow of fluids such as water. The assumption was that crowds move in a way similar to water in a pipe – evenly distributed with an equal speed at any given time.

Put plainly, this assumption does not hold water. Individuals in a group exhibit crowding behavior, something that cannot be accurately accounted for by comparing to fluids. Once we think about it, it isn’t surprising at all. From experience, we know that often one queue is longer than another at the movie theater for no apparent reason and that people do not spread out evenly in all the compartments of a local train.

In an escape panic situation, individuals deviate from fluids even further. Under duress, individuals tend to walk faster than usual. They tend to push and shove others ahead of them in the perceived escape route. In the heat of the moment, other escape routes are frequently overlooked. In confined areas such as crowded exit points, moving becomes uncoordinated and physical forces build up to a point that steel fences and brick walls can get damaged. No wonder then that many people fall in the melee and get injured! These people then become obstructions that have to be overcome by the rest of the crowd.

Partly because it is unscrupulous to conduct experiments that force people to hurt themselves in panic situations (besides being an opportunity for massive legal action), there is little scientific data on how real people react under extreme pressure. However researchers led by Dirk Helbing at the Dresden University of Technology in Germany used computers to model crowd behavior in escape panic. Helbing simulated the exit of 200 pedestrians through a one metre-wide exit of a room 225 square meters in area.

Helbing’s landmark account was different from many other studies on crowd dynamics because he incorporated both physical and psychological parameters in his computer model. For example, physical considerations included the average mass of individuals in the crowd, the average starting velocity of the moving individuals, and a realistic estimate for acceleration time upon panicking. For example a parameter based on human psychology which should be taken into account is the tendency of individuals to avoid physical contact with one another as much as possible which, of course, is true for most of us.

From computer simulations, Helbing calculated that the exit of pedestrians from a room would be regular until the pedestrians began to rush to get out. When a threshold velocity was reached, pedestrians actually began to block the exit. Impatience caused further clogging which led to fewer people exiting in the escape panic then would have gotten out under normal circumstances.

This is something that you might have also noticed in real life when observing passengers trying to get on to a bus through a narrow door. People swarm in a while trying to get on the bus, but the number of passengers actually entering does not increase proportionally with the force applied by the crowd at the door.

The obvious answer to this problem is to increase the size of the exits, but within the constraints of Helbing’s test, increasing the exit routes could not prevent congestion completely. For all types of structures, increasing exit points is not feasible either. Imagine trying to increase the doors of buses and trains and you see the problem. Instead, Helbing recommended a radical solution – placing columns in front of exits. By placing column at strategically located asymmetrical points, crowds would not be able to build up pile up to deadly physical forces.

As we have now observed through these examples, completely falling under the sway of crowd behavior can be quite detrimental. Of course, this does not indicate that at all times individuals should strive to automatically oppose any consensus. There may be an intelligible reason why an idea is held in favor by a majority. In a burning, smoke-filled room there may be a crowd near the single known exit, but searching for another exit might also be time-consuming and ultimately futile. Taking the life of another human that has meant no harm might seem deeply individualistic, but there are also valid reasons why most societies find this sort of behavior particularly reprehensible. Therefore, we should strive to strike a reasonable balance between being individualistic and social.

How, then is such an intricate balancing act possible? There are no sweeping, black-and-white answers to this question, and unfortunately, any attempt at one is no more than a hollow platitude. What we can do is examine each scenario on a case-by-case basis to determine when we want to follow the crowd and when we want to act alone. And we can increase the likelihood for an acceptable outcome through analysis of prior outcomes and increased personal experience whenever possible.

(This is an excerpt from a much longer piece on groupthink and the psychology of crowds.)

© Text, 2010-2012, Anirban