Again and again, disasters strike and cause widespread panic -whether it’s fireworks lobbed into crowds during a sports event or a fire at an airport, nightclub or hotel. With a view to understanding crowd dynamics, researchers are examining whether people react to events in specific patterns and whether threats can be recognized and perhaps headed off. One such researcher is Dr. Wolfram Klein, a mathematician who works at Siemens Corporate Technology (CT) in Munich.

Together with his team, Klein has developed a model that simulates crowd behavior, thus helping researchers to predict where and when a critical situation may arise. Klein’s model can simulate the way in which crowds of tens of thousands of people behave. What’s interesting, according to Klein, is that “they move very similarly to liquids or gases.” Like molecules, people either attract or repel each other.

In addition, when people move through buildings they have to navigate around walls and other obstacles; and, of course, small, narrow spaces can lead to congestion. “Based on the principles of alternately attracting and repelling forces, we can chart human behavior and produce predictions in terms of mathematical equations,” says Klein.

The software could help architects plan safer buildings because it can identify which spots might give rise to dangerous situations. Klein is certain that comparatively simple procedures and planning steps could prevent many disasters.

In order to illustrate human behavioral patterns even more realistically, his team has continuously refined its simulation model. For instance, the software now not only uses statistical methods to depict the effects of a person’s age and health on their walking behavior, but also takes group interaction into consideration as a factor. In addition, the Munich-based researchers have improved their mathematical calculations significantly.

According to Klein, the system is now so fast that their crowd simulations can be used to make short-term predictions. “We can tell up to five minutes beforehand what is likely to happen assuming that no one intervenes. This way, the head of operations at a facility could act quickly.

This method of crowd control has already been tested in various research projects, including one carried out at Frankfurt’s central train station. Based on surveillance camera footage, the software was able to accurately predict the flow of pedestrian traffic — as well as congestion — several minutes before it occurred. The program has also been successfully used in and around the soccer stadium in Kaiserslautern.

Evacuating the city’s stadium would be a dramatic challenge for the police and fire department. Although the stadium accommodates up to 40,000 people when it is full, it offers only a few escape routes. And to make matters more difficult, all of them lead through the surrounding residential areas.

 Safe, Quick Evacuations

In the future, the researchers also want to use this knowledge to support their colleagues in Siemens’ Building Technologies Division. To this end, in the Swiss town of Zug experts are developing dynamic fire protection solutions for buildings — so-called intelligent response systems. Christian Frey, who is responsible for innovations in Zug, explains: “These are highly professional systems that can react immediately and effectively to dangerous situations or incidents.”

Frey points out that in order to get people out of a burning building safely and quickly, the usual green signs along hallways indicating escape routes are not sufficient. In public buildings such as hospitals and hotels, he says, most people aren’t familiar with their surroundings. “If you’re in a panic, the next emergency exit isn’t that easy to find.”

Studies also show that many people fail to react appropriately to conventional warning signals such as honking or sirens. They often think it’s just a fire drill or a false alarm — or else they don’t know what to do. This is where information technology can help. For instance, office workers could receive automatic warnings and updates on their personal computer screens. At the same time, large electronic screens in the hallways and smartphones would display arrows showing people how to get out of a building. In addition, sensors in ceilings and floors would be able to measure the stream of people.

Based on this information, an intelligent building software system would be able to recognize early on when a particular escape route is in danger of becoming overcrowded. It would then respond by directing people to the fastest and best alternative route out of a building and into the open. Visual systems would also be complemented by voice alarms and mass text messages.

Fire Department App

What’s more, such systems will be able to improve building management and support rescue workers. “The system analyzes data from a building, recommends immediate measures to defuse the situation, generates dynamic, up-to-date instructions, and helps rescue workers manage the evacuation and direct people to escape routes,” says Frey, describing the idea behind the software concept. In the future, he adds, when a fire breaks out, the building management system will immediately link up with the fire department’s computer system. Rescue teams and fire fighters would then receive a blueprint of the building on their smartphones. Such a plan would not only display the source of the fire, but also monitor how it is spreading. In addition, intelligent movement sensors would indicate where people are located in the building.

Together with other companies and institutes, Siemens researchers are developing these technologies as a part of the EU DESSiRE (Designing Safe, Secure and Resilient Large Building Complexes) project. Siemens’ simulation experts from Munich are also assuming an additional role. Specifically, they have developed a method that allows them to predict the spread of fire in different kinds of buildings. Klein explains how it works: “We can light a virtual fire in order to see how it will affect each building.” The researchers can simulate fire in various surroundings and different interior fittings — for example, with or without furniture, or with flammable or flame-resistant materials. By trying out these different scenarios, the heads of operations can learn to predict the spread of a fire more accurately and to thus act promptly and effectively according to a given situation.