KAISERSLAUTERN - It's been more than two years since 21 people were killed and more than 500 injured in the Duisburg Love Parade, a tragedy that shocked Germany and reopened wider questions about how to control large crowds.
In an article published in the EPJ Data Science journal, physicist Dirk Helbing of the Swiss Federal Institute of Technology in Zurich (ETHZ) and his colleague Pratik Mukerjee have analyzed the disaster, which they blame on what’s known as the “crowd quake” or “crowd turbulence” phenomenon.
A crowd quake is the random turbulence or shock wave that can arise when many people are packed together. At occupancies of about seven persons per square meter, the crowd becomes almost a fluid mass. Shock waves can be propagated through the mass, sufficient to propel them distances of three meters or more,” found the scientists. In Duisburg that led to people falling over and being crushed as others trampled them. For their analysis, the scientists analyzed videos filmed by people who had attended the Love Parade and used computer-simulation.
But scientists were at work on ideas to make mass events safer well before the Duisburg catastrophe. The underlying problem in a crowd situation is that people begin to push and shove, creating a dangerous or even deadly situation, albeit unexpectedly and unintentionally. Physiological processes in the brain play a role. This depends in part on how alarm at the situation is processed by those involved – a slow and controlled build-up of fear reactions, or a full on, adrenaline-rush, fight or flight panic, which is catching. At some point a switch occurs between individual psychology and the conformity of mass psychology – with mass panic as the result.
Innovative design in buildings and stadiums can help counteract the resulting effects, Helbing says – for example the use of “wave breakers,” basically crowd retaining walls.
Another tool is the "Crowd Control" simulation model created by Siemens that was applied to a Klauserslautern soccer stadium holding 40,000 people and with few evacuation routes through the surrounding streets. Using monitors and analyzing crowd flows, the system was able to outline optimum evacuation routes for a number of different situations. "The police and fire departments have already incorporated the results into their evacuation plans," says Wolfram Klein, who heads the development of Siemens’s "Crowd Control" project.
At the London 2012 Olympics, police used an app-based Crowd Monitoring Technique developed by the German Research Center for Artificial Intelligence (DFKI) in Kaiserslautern.
"The great thing about the app is that if needs be every user can be sent a different message," DFKI expert Tobias Franke explains. One problem with the method is data protection – who really wants to make their details available to a state authority, especially if they are abroad? Franke says DFKI has thought of this, since optimum use of the app depends on getting a large number of people to use it and hence in winning their trust. So users remain anonymous, position data is encrypted and use is limited to the event area and only activated in critical situations.
"One problem in Duisburg was that more people gathered in some places than were expected," explains DFKI scientist Paul Lukowicz. Sensitive smart phones can however register even light surges that could lead to powerful waves of the type that led to the Duisburg disaster. Based on user information, personalized recommended escape routes could also be sent, for example not recommending a route that involves climbing over a wall to an older user.
Drones for aerial surveillance
The DFKI is presently working with organizers of Munich’s Oktoberfest, where it may be applied. Another possibility in Munich is drones for aerial surveillance, which according to the city’s fire department has the advantage of showing swelling crowds before they actually form so that prevention barriers can be set up.
Here too data protection is an issue – evaluation of the images would have to be based on crowd density only and not individual identities, although that would theoretically not be a problem since vertical filming would not make faces identifiable. Advantages include being able to identify the fastest way for an ambulance or the police to get through.
The first test run with drones in Germany took place at the Bochum Total street festival in July, when the AirRotorMedia company working at the behest of the fire department and researchers at the University of Paderborn filmed 600,000 people from a height of around 100 meters (328 feet).
"Picture quality was very good," says Stefan Lieber of the Bochum Fire Department. Not only were the drones cheaper than police helicopters, they had more technical advantages and it is virtually impossible for them to crash – it they lose radio contact or the batteries are on the blink they return automatically to launch site.
Drones made and run by the University of Technology in Kaiserslautern and researchers in the city’s Fraunhofer Institute for Experimental Software Engineering may be in use at the world soccer championships in Brazil in 2014. The idea is to incorporate their input with that of stationary cameras, smart phone apps and more into a kind of electronic situation table where it can all be evaluated.
However intensively researchers work on new crowd control technologies, the ETHZ’s Helbing cautions, "Absolute security will never exist – you cannot categorically exclude the possibility of accident."
Globalization also means that visitors at major events are increasingly heterogeneous and that can have an effect on behavior – and hence relevant security measures. Cultural differences have to be taken into account also. Another current avenue of global research is the study of how different peoples react in crowd situations.
In Germany, a number of fire departments, the Jülich Research Center and the University of Wuppertal are focusing on how catastrophes like the one in Duisburg could be avoided in future. Everything from training methods and simulation instruments to systems to develop security concepts are being developed. The goal is the creation of a manual for organizers of large-scale events. But researchers point out that wading through all the ordinances governing public events can complicate the endeavor – not to mention that there is no single definition of “large scale event.” Getting that defined at federal level would be a big help, they say.