What would you do?
The psychology of human behaviour in times of disaster has become an essential discipline used by engineers to reduce the risk that people will die when a building is hit by fire, earthquake or terrorist attack, assures E&T.
In the weeks and months after the 11 September 2001 terrorist attacks on the World Trade Center, there was much agonising over whether more people would have survived had the buildings been designed differently. Later, researchers started considering another factor - one that turned out to be just as crucial: how the people in the buildings behaved after the planes struck.
What became clear is that more lives may well have been saved had people reacted differently. Rather than dash for the nearest exit, most did the opposite: they prevaricated. A study by the US National Institute of Standards and Technology (NIST) found that even those who managed to escape from the towers waited an average of six minutes before heading for the stairs. Some dallied for half an hour. These findings lent credence to a theory already gaining support among risk experts: that when disaster strikes, people behave in unexpected and irrational ways.
"The tacit assumption used to be that people respond instantaneously when an alarm goes off," says Ed Galea, director of the Fire Safety Engineering Group (FSEG) at the University of Greenwich. "That doesn't happen."
The study of human behaviour in times of crisis has become an essential research field. Its findings are being used by engineers to reduce the risk that people will die when a building is struck by fire, earthquake or terrorist attack. It is also leading to new technologies that can be used to improve evacuation procedures.
"Twenty years ago not many people wanted to listen to this kind of research," says Galea. "But recently it has become much more popular and is influencing safety regulation and design not just in buildings but with a whole range of things such as aircraft, ships and trains."
Galea and others have carried out interviews with 271 survivors of the World Trade Center attacks to try to understand the social and organisational factors that influenced the way people evacuated. Their results, due to be published later this year, mirror those of the NIST researchers.
About half of the interviewees said they hung around before deciding to escape, either waiting for more information about the situation from colleagues or officials, or collecting things to bring with them, changing their shoes, locking things into safes, going to the toilet, finishing emails, filing papers, making phone calls or shutting down their computers. When they did leave, they descended the stairs without great urgency and considerably more slowly than the building's safety engineers had forecast, apparently because of congestion from the large number of people trying to descend, even though the building was far from full.
How can this kind of behavioural analysis help engineers design skyscrapers that can be more easily evacuated?
To help answer that, a team at FSEG used a computer simulation model to re-run the World Trade Center disaster under various 'what if' scenarios. The model, known as Exodus, is designed to test evacuation times for large numbers of people trying to escape from complex spaces, tracking the movements of individuals and how they are affected by, for example, heat, smoke and toxic fumes from a fire. It can be programmed to account for how different people respond and move at different speeds, depending on their age, gender, whether they are disabled, how familiar they are with the layout of the building or whether they are in a group, and also depending on the type of building they are in.
The FSEG team found that, had the stairways been better reinforced, so that at least one had remained intact throughout the north tower of the World Trade Center, everyone who survived the initial impact could have escaped before the building collapsed. They also found that, had the building been fully occupied, the congestion on the stairs would have been so heavy that an extra 6,000 or so people would have been killed, unable to get out in time. "A mass evacuation of the fully occupied building would lead to ... a highly inefficient evacuation," concluded the researchers.
This begs a tantalising question for safety engineers and designers: how do you evacuate people from tall buildings when stairs and other exit routes are rendered useless, or when people are too slow to react to use them safely? Several ideas are in development or commercial use.
One is the Spider Rescue System, designed by Moseroth Technologies in Israel and available in the US, which consists of a steel cable and harness that enables people to lower themselves to the ground from the window of a burning building up to 50 storeys up. It is controlled by a brake mechanism that limits the speed of descent to six feet per second. Another Israeli company, Escape Rescue Systems, has designed an evacuation elevator consisting of five collapsible cabins that can climb up a building stopping opposite five floors simultaneously, allowing rescuers to enter and occupants to leave. A prototype is already in use on a 21-storey building in Ramat-Gan, Israel.
Other entrepreneurs are working on flying rescue craft that can pluck people from stricken buildings. David Metreveli, a Georgian-Canadian aerospace engineer, has proposed a vertical-take-off flying life-raft that can hover alongside a building while people evacuate through the windows. Metreveli says the aim is to rescue people trapped in places too high for ladders and out of reach of helicopters with rescue baskets, which are often impractical anyway because of flames, smoke or roof obstructions. "It can access all narrow canyons, city streets, smokestacks, bridges, television and other tower constructions," he says.
The US Department of Homeland Security has drawn up standards to guide the development of search and rescue robots which can locate and retrieve people trapped or injured in environments too hazardous for human rescuers, such as collapsed or contaminated buildings. The DoHS and NIST hold a search and rescue robot evaluation exercise each year at Disaster City, a disaster training facility at College Station, Texas, when engineers from across the world gather to test their robots in simulated emergency environments such as rubble piles, collapsed houses or a derailed train. There's also the annual RoboCup Rescue competition, organised by an international group of robotics experts to stimulate the development of robots for use in emergencies. At last year's event, held in China, the robots had to find their way around a maze without help from their human handlers and sense out toy dolls that were giving off heat or carbon dioxide or making noise.
These are all 'last step' technologies, for use when standard evacuation procedures have failed or people have failed to follow them. Yet many researchers are confident that they will remain a last resort as we come to understand better how people behave during disasters. Better understanding of behaviour, they reason, will lead to better escape planning and more appropriate building design. We are already some way down that road. As many risk experts have pointed out, the kind of unresponsive behaviours seen among the victims and survivors of the World Trade Center is now known to be fairly typical of people in a crisis of that kind.
For example, many people have died in burning aircraft because they sat in their seats too long after impact before trying to escape. A detailed computer simulation by FSEG of the 2003 Rhode Island nightclub fire, in which 100 people died, found that more people could have survived had they responded just a few seconds earlier. Since the ones who get out of such situations quickly tend to be those who have rehearsed an escape plan in their heads, one solution would be to use virtual reality models of aircraft, ships and buildings to allow passengers or inhabitants to test emergency evacuation plans. Computer scientists at the University of Durham have co-opted the software in video games to create a 3D model of their department to do just this (see E&T Vol 4 #3).
Computer and real-life simulations are also being used to try to predict what will happen in emergencies and how people will behave. Dirk Helbing, a physicist at the Swiss Federal Institute of Technology in Zurich, has found that one way to ensure more people escape from a room through a single doorway is to place an obstacle in front of it. Rather than disrupt the flow, it encourages the crowd to spontaneously adopt a more orderly exit pattern.
However, using simulations can be troublesome, because it is difficult to create realistic conditions. A typical example was the aircraft evacuation experiment by Cranfield University, held on behalf of the Civil Aviation Authority after the 1985 British Airtours crash at Manchester in which 55 people died. Researchers paid a group of students £10 each to take part, and promised them £5 if they managed to be among the first to escape from a "stricken" plane.
The ensuing chaos, with students wedged in aisles and crushed under seats as they fought for the exits, was judged as typical of how people might behave if their lives were threatened - mass panic driven by self-preservation. But it is now clear that panic during such situations is extremely rare, and that in most disasters crowds of strangers behave with remarkable cooperation and social responsibility - the World Trade Center evacuation is a case in point. It seems the psychological mechanisms at work when you are escaping for your life are different to those involved when the incentive is purely financial.
Solidarity in crisis
John Drury, a psychologist at the University of Sussex, suggests in a study to be published in the British Journal of Social Psychology that the cohesive behaviour of crowds during emergencies derives from the collective group identity that appears during such situations. His team interviewed survivors of 11 tragedies, such as the 1989 Hillsborough football stadium crush that killed 96 football fans and the 1983 IRA bomb that killed six outside Harrods in London, and found that when threatened with death the common behavioural trait is solidarity rather than competitiveness. In another study, due to run in the International Journal of Mass Emergencies and Disasters, they found that this sense of solidarity, in which people frequently put themselves at risk to help strangers, was much in evidence during the 7 July 2005 suicide bombings in London.
What is going on here? Drury argues that members of a group who share an extreme experience of a common fate tend to identify strongly with each other and form strong, albeit temporary, social bonds that transcend even those attachments that already exist (such as between friends or relatives). This makes it easy to coordinate behaviour and even act in other people's interests, "even if these others are not personally liked or even known", he says.
"Self-interest has shifted upwards from the personal self to the level of the crowd." The upshot of this is that acting individualistically in a crowd of people in a dangerous situation could be a big mistake. In the examples that Drury's team analysed, "a safe exit was more likely the more people thought of themselves and acted as a group. The more that everyone acts as an individual, the more likely it is that exits will be blocked as people compete."
How can this understanding of group dynamics translate into better design? Ed Galea points out that safety engineers tend to assume people evacuate "as individuals. This belief is implicit in all building design. This key assumption has an important influence on the unfolding evacuation dynamic and potentially on the overall efficiency of the evacuation."
The realisation that during a crisis group psychology comes to the fore and that crowds adopt a shared social identity should prompt designers to rethink. For example, there should be much greater emphasis on communication during an emergency, says Drury.
"The UK is the most surveillanced country in the world. It is time the same level of technology was applied to giving us information" as it is to collecting it. So, use giant LED screens and PA systems rather than antiquated sirens and alarms which people find easy to ignore.
Above all, he says, when disaster strikes consider the crowd as "the fourth emergency service". With a bit of guidance, people are more than likely to do the right thing.