The unpredictability of crowds causes problems for scientists studying their behaviour. The challenge is to make public spaces safer, especially when considering the impact of a disaster but, experts have discovered, you can't measure crowd fear in real-time.
Nevertheless, built environment professionals need to understand how crowds behave when faced with fire, flood or terrorist attack so that they can design and manage spaces that will avoid lethal crushes.
One team of scientists at the Technical Institute of Buenos Aires recognised that the annual bull-running in Pamplona, Spain provided a perfect natural crowd experiment to be observed and analysed. Setting up cameras from rooftops, the team studied people running at top speed to evade the bulls.
They concluded that runners sped up when the bulls got closer but, unexpectedly, the speed of runners increased with the crowd's density – this was at odds with the assumption among professionals that people slow down as group density increases to lower the risk of collision. It suggests that, in the moment, people fear what they are running from more than the potentially lethal risk of falling. The conclusion was that designers and developers might need to consider wider passages, tunnels and bridges in public spaces and venues of mass gatherings.
Many factors feed into how a crowd behaves. “The traditional assumption in engineering models is that people are like billiard balls and mindless,” says John Drury, professor of social psychology at the University of Sussex. “The consequences are that the only architectural solution for that is the width of exit.”
The width of the exit matters, but what people also need, Drury says, is information they can respond to. Most people die in fires, for example, not because of rushing for exits but because they don't take the threat seriously enough.
The 9/11 World Trade Centre terrorist attack is one of the most researched disasters in this field. Drury says analysis concluded that there was little panic in the sense that few people abandoned their norms or acted selfishly in the moment.
“It is a myth that people tend to panic,” he says. “When a crush occurs, it is often spoken of as a stampede or panic. Most crushing incidents are due to mismanagement of space. If you have a crowd event you need oversight of the capacity and the numbers coming in.”
“It is a myth that people tend to panic. Most crushing incidents are due to mismanagement of space” Professor John Drury, University of Sussex
Keith Still is a visiting professor of crowd science at the University of Suffolk and has consulted for some of the world's leading venues and high-profile projects, including the 2000 Sydney Olympics. He often acts as an expert witness in legal cases related to crowd safety and says that when looking at significant incidents, it is always a density-related problem. He believes built environment professionals need a greater understanding of crowd flow and the operational efficiency of public places where people assemble in large numbers.
He believes that putting crowd science knowledge up front in the design process helps build more efficient places of public assembly. This means a more comprehensively safe environment, reducing the need to fix problems once the space becomes operational.
“With crowd safety, you need to know the limits,” he says. “There is always a point where safety becomes compromised. Raising awareness of that needs to be incorporated into architectural thinking. Whenever we've worked with major consultants or an event like Dubai Expo 2020, primarily dealing with crowd throughput, it was extremely useful for built environment professionals to understand that.”
Regarding evacuations, the building codes tend to be designed for fires. Still believes regulations need to consider other threats. In Singapore, for example, his team studied the parliament buildings to assess behaviour during a terrorist attack involving detonating toxic weapons. They calculated that it would take four to five minutes for the lethal contamination to spread, less time than it would to evacuate all the people.
There are only five responses to a threat: total evacuation, direct evacuation, phased evacuation, stay put, or ‘invacuation’ in the case of an active shooter outside the building. “Not all these scenarios fit within building codes and regulations or the architectural research,” says Still. “It would be useful if we started to think about places of public assembly, where shelter might be easy to facilitate, and what strategies could be in environments that are at greater risk. It is about awareness during the design and knowing what architecturally might facilitate a safer space by understanding those threats.”
James Bulley FRICS heads Trivandi, a consultancy that provides strategy, assurance, planning and implementation services for venues and mega-events worldwide. Having been involved with redesigning and rebuilding UK football stadiums in response to the 1989 Hillsborough disaster, which cost 97 lives, Bulley later moved to the London Organising Committee of the Olympic and Paralympic Games for the venues and Olympic Park for London 2012.
He says a critical element to planning venues and complex built environments is thinking through the spaces around the facility and how they function operationally and in an emergency. This includes the ‘last mile’, the stretch between transport facilities and the venue.
“Where do you put tens of thousands, possibly 100,000 people? You need to keep some distance from transport, because if you're right on top of them, you've got no space for all those people to dissipate and get away,” he explains. “For example, at the London Olympic Park, we didn't give people the direct route from the Park, we sent them in a roundabout way to get there.”
Developers won't get a safety certificate for a stadium unless they can safely evacuate the number of people it can contain and that means, Bulley says, planning large spaces to dissipate people and hold them.
An event like the Olympics, with huge numbers of people and multiple contests, needs extensive crowd modelling to understand the scenarios. “We did plus 10% on everything because we knew we would need it at peak moments. The vast majority of the time, we didn't need it, but we always planned above what we thought would be the peak in case we were under pressure from an operational perspective,” he says.
“We designed the widths of concourses and bridges, for example, to be over what we believed we would need and made sure that there was no hard gate to hold people. People would be held by a line of stewards and police and not held by gates.”
On this point, Bulley attributes many crowd control failures to hard gates. Most recently, in Paris at the Champions League Final, hundreds of fans were pushed up against barriers at the Stade de France and forced to climb over when authorities channelled tens of thousands of fans behind them.
“I've witnessed it myself in the Middle East in major events where large numbers are held back by solid gates,” adds Bulley, “and that's where crush happens. If you cannot hold people back from arriving at those gates, that is where dangers appear.”
The planning principle involves modelling and testing, creating large spaces for holding and managing people, particularly the egress when everybody's leaving simultaneously.
Much of minimising the risk of danger is about communication, which is essential between the authorities, stewards, and police, but also critical to managing crowd behaviour. “We would rarely hold people for more than 10 minutes, because after that behaviours turn and they may try and jump the queues or climb barriers,” says Bulley. “You must communicate what is happening and why they're being held and ensure everyone can hear it.”
Managing crowd safety requires an all-round approach that starts with understanding crowd dynamics, venue design and capacity and ends with communication, robust processes and contingency planning.
“Managing crowds is about awareness during the design and knowing what might facilitate a safer space” Professor Keith Still, University of Suffolk