With smoke following close behind, hundreds of people crowded down the stairwells of the high-rise office building. Ed Beyea, however, sat in his wheelchair on the 27th floor, waiting patiently.

Beyea, a C3 quadriplegic, wanted to avoid the broken bones that would come with being carried down the stairs. He also didn’t want his descent to create an obstruction for others trying to get out. Beyea’s co-worker and best friend of 12 years, Abe Zelmanowitz, volunteered to wait with him until help came.

Both Beyea and Zelmanowitz perished when World Trade Center One collapsed on Sept. 11, 2001.

Incidents like that are fresh in the minds of Utah State University researchers Judith Holt and Keith Christensen, who are studying how and why some people with disabilities don’t make it out of buildings during evacuations.

“September 11 really brought up some huge concerns,” says Holt. “There were many people in the two towers with disabilities. Some got out, and some didn’t. It has become clear that in past decades we’ve been working hard on making it so disabled people can get into buildings, but we haven’t dealt much with how to get them out, especially in emergency situations.”

Holt and Christensen’s latest project is responding to those concerns raised by September 11 and hopes to improve the exit of individuals with disabilities from buildings during emergencies.

“After 9/11, the federal government began asking for ways to improve building evacuation, and what was initially offered was a lot of technological band-aids, like parachutes, hovercrafts, and slides,” says Christensen. “We’re trying to figure out what’s really going on and how we can create passive systems in accommodates evacuation without a ‘patch.’”

“In a nutshell, we are asking how well accommodations for getting disabled people into buildings work when lots of people are trying to get out,” says Christensen. “In an emergency situation, the pedestrian dynamic changes when 40 or more people are trying to get out of one door at once. We are looking at what happens in that type of situation.”

One difficulty during building evacuations noted by Holt and Christensen is that disability-accessible entrances seem like the best route—for all people. So, in an emergency, many people will choose the most accessible exit, making it unavailable for the people who have no other alternative. In other cases, the path between a wheelchair-accessible door and a stair ramp may be blocked by the main route of travel everyone else is taking to exit, making people with wheelchairs wait for everyone else to get out.

Holt and Christensen are studying several different facilities—a sport/concert venue, a high school, a high-rise office building, a major airport, and Federal Triangle in Washington, D.C.—to determine what happens during an evacuation and how exiting can be improved.

“The main problem with this study is that you can’t practice with people,” says Christensen. “You can’t put 10,000 people in a stadium, declare an emergency, and then watch what happens. The fact is getting large groups out of a building fast can’t be studied in real time with real people.”

Instead of studying actual building evacuations, Holt and Christensen are using computer models to predict how big groups of people will act in an emergency. The research utilizes a method called “agent-based modeling,” which creates thousands of individual computer people, or agents, each with their own tendencies and behaviors, such as how fast they move, whether they will follow a crowd or not, how they perceive exits, and their aversion to narrow hallways. Some of these agents are programmed with disabilities, and their exits are watched especially closely.

“The computer model we are working with is based on a natural resources program used to model the movements of butterflies and ants,” says Christensen. “Although it looks like a colony of ants is always following an overall set of orders, it is really just a complex decentralized system, with individuals behaving according to rules. Egress events are very similar.”

The computer agents are placed into a simulated facility, which is a replica of a real building, and are then told to evacuate. When the actions of all the agents are studied together, overall patterns of behavior emerge.

“We are testing for lots of things: rate of travel, the time it takes to exit a building, who the stragglers are, and how long it takes for them to get out,” says Christensen.

Holt and Christensen’s current study is of an airport, with 5,000 people in the model. By studying real local public buildings, they hope to be able to make recommendations that can be generalized to most public buildings.

“The neat thing about computer modeling is that you have the opportunity to study all types of variation, and to run a situation over and over again,” says Christensen.

Accessible evacuation is a big issue, and it doesn’t just affect the 10 percent of the population with diagnosed disabilities, says Holt. “As we get older, we will all end up disabled. It’s a population that everyone gets to belong to if they live long enough.”

Family members of people with disabilities, parents with small children, and elderly people also may take longer than normal to exit a building. Thoughtful, realistic steps taken to help people with disabilities exit faster may also make a big difference for these other groups as well.

“When there is an emergency, it can be critically important to get people out of buildings in a hurry,” says Christensen. “We are trying to make sure all people are able to get out safely.”

- Anna McEntire