Multiple entry points in the design of the future Jamarat bridge should help to reduce crowding even further.For more images and video, visit Helbing's website

The annual pilgrimage of Muslims to Mecca in Saudi Arabia, known as the Hajj, has on occasion been marred by deaths from trampling in the huge crowds that gather for the rituals. But scientists studying how pedestrians move around think they have made such crowd disasters much less likely.

In 2006, 362 people died in the crush that developed in the town of Mina, where pilgrims gather to perform a ritual stoning of pillars representing the devil as part of the Hajj. This year's ritual, which happened in late December and early January, went off without incident. Although there have been plenty of other accident-free years, this time the reason owes more to sound planning than to luck, says Dirk Helbing of the Dresden University of Technology in Germany. Hebling and his coworkers used the science of crowd dynamics to introduce a raft of new crowd-control measures.

"This Hajj, in contrast to many previous ones, was very safe, without any panics or incidents, even though it was expected to be the most critical ever and there were about 800,000 more pilgrims than the expected 3 million," he says. "This great success was due to a completely different organization of pilgrim flows."

Easing congestion

The activity in Mina, about four miles from Mecca, centres on the ritual stoning of three pillars known as the jamarat, which draws huge crowds. To ease congestion problems, the old pillars have been replaced by larger, elliptical ones, and a building called the Jamarat Bridge gives pilgrims two tiers of access to the jamarat. But as the number of pilgrims has increased steadily over the years, even these precautions have not prevented disasters.

Last year, after being consulted by the government on how to improve crowd safety, Helbing and his colleagues were allowed to analyse video recordings of the crowd at Mina. "The Saudis invested a lot of money in putting up cameras to gather data," he says.

Helbing and his co-workers had previously analysed how people move past each other in corridors or intersections, and how jams may occur when many people try to exit quickly through a single door. These effects, which can be mimicked in simple computer models where the people are represented as moving particles that repel one another, can account for how some tragedies happen when a crowd panics.

But their work on the 2006 Hajj showed a totally new type of behaviour, Helbing and colleagues report in a paper on the arXiv preprint server¹.

Pilgrims directed along one-way routes can follow the flow, avoiding dangerous crush points.

As the mêlée thickened, first the throng stopped passing steadily onto the bridge and instead moved in waves, so that individuals would be repeatedly stopping and starting. But then, as the crowd became even denser, it changed to another mode in which clumps of people were jostled in all directions, apparently at random and against their wish to move steadily towards the jamarat.

"Pilgrims were being pushed around," says Helbing. If they stumbled and didn't get back on their feet quickly enough, they were trampled. The movements look like those in a fluid when it becomes turbulent, which hasn't been seen before in human motion.

Re-routed

Helbing says that there are warning signs for the development of this type of behaviour. For a given point on the route, for example, the average number of people passing per minute falls below a critical threshold up to half an hour before turbulence sets in.

For the 2007 Hajj, Helbing consulted with the Saudi authorities to plan a new route and schedule that pilgrims would be compelled to follow, rather than meandering at will to the jamarat. "All 1.5 million registered pilgrims got a timetable and a route in order to distribute them uniformly in space and time," says Helbing. In case the more-than-a-million unregistered participants confounded this plan, they also had the capacity to use real-time data from surveillance cameras to alter the schedule, guided by their models of crowd behaviour. That capacity wasn't needed this year, but the scheme is in place for future.

"The science was very important," says Salim Al Bosta, a civil engineer at the Saudi Ministry of Municipal and Rural Affairs in Riyadh, who managed the consultation exercise. "We learnt a lot about how to organize the flow from the crowd simulations."

Helbing says that several of the new measures were controversial, with some experts worrying it would make things worse. But the scheme was a success. An important step was to introduce a one-way system, with roads designated only for walkers coming from the stoning back to the camp. "Last year you had to push a lot to get to the camp," says Helbing. "This year you could comfortably follow the stream all the way. Everyone was very happy." Al Bosta agrees, saying that the new organization was "very effective."

Helbing says that most of the critical organizational measures are now in place, so that future years should be safer. Moreover, from next year on, the Jamarat Bridge will have more floors, easing the flow even more. "Now other experts can take over," he says.

He confesses that he was nervous about taking on such a high-risk project, where lives were clearly at stake. But in the end he worried that the danger of doing nothing was even greater. "Could I feel comfortable if people had died and I'd declined to help?" he says. "It was a matter of responsibility."

Visit our researchersmake_pilgrim.html">newsblog to read and post comments about this story.  

• References

  1. Helbing D., et al. preprint available at http://arxiv.org/abs/physics/0701203.