A panicked crowd is more like sand falling through an hourglass than water flowing out of the nozzle of a hose. By making this subtle distinction, scientists have created a new computer model of crowd behavior that could be used to improve the safety of designs for malls, stadiums, and other places where large crowds gather in confined spaces.
Panics are spontaneous, unpredictable, and deadly. Last year, for example, 54 people died in Minsk, Belarus, when a sudden cloudburst triggered a panicked rush for the entrance to a rail station. Postmortem analyses of crowding disasters have shown that the pressure of throbbing mobs can bend steel girders and push over brick walls. But studying stampedes is difficult. "You can't experiment on people," says crowd dynamicist Keith Still, a consultant at Legion Crowd Dynamics Ltd. in London, "and you can't get a research team to the site of a disaster fast enough." Instead, safety experts have relied on computer simulations that assumed crowds behave like fluids. But that's a questionable assumption; fluids feel no pain, never stumble, and don't make decisions.
Now, Dirk Helbing of the Dresden University of Technology and Illes Farkas and Tamas Vicsek of Eotvos University in Budapest, Hungary, have come up with a more refined model. They look at people as if they were a collection of particles or billiard balls and treat the psychosocial forces that govern their behavior as if they were mechanical forces. As crowding people search for escape routes, walls and other people repel them; at the same time, a frictional force drags them along with the mob.
If a crowd gets so panicked that it wants to exit too fast, it starts behaving counterproductively, the scientists report in this week's issue of Nature: "The faster they try to leave, the slower they get ahead," says Helbing. The same phenomenon has been reported in real panics. Helbing and colleagues believe the simple physical model works because in a run-or-perish situation, otherwise sophisticated humans are compelled to make simple snap decisions.
"I am very impressed by it," says Still. "They have successfully modeled the fundamental elements of the crowd during the onset of a disaster." Although there is no prescription for preventing a panic, Helbing believes his work suggests effective ways to reduce their severity, including installing asymmetric columns in front of doors to relieve the pressure and hiring trained guides to calmly lead crowds to safety.