Administrative toolbar

6.4. Escape Dynamics

Helbing and colleagues have extended their work to model the evacuation of rooms (Helbing et al., 2000) where they assume agents start showing uncoordinated behavior when in panic. As a result agents start clogging at exits and the flow through the exit is slowed down. Models based on this principle are used design stadiums, theatres etc, in order to reduce probability of fatalities in crisis situations.

The first simulation assumes that agents move in a slow pace to the door. The agents avoid each other and don’t bump into each other. This leaves enough places of people to leave the room in an orderly way (Figure 8). If agents panic and have a higher desired speed to leave the room, they bump into each other and slow down the speed in which agents can leave the room (Figure 9). Thus with a higher desired speed it takes longer before all people left the room. “Don’t panic” is therefore the motto in times of crisis.

If agents are hurt because other agents bump into each other, they can become victims, lay on the ground and become additional obstacles for the panicking agents (Figure 10). This leads to even more delays in evacuating the room.

Figure 8: Agents leaving the room at a quite pace.

Figure 9: Agents in panic who want to leave the room quickly.

  

Figure 10: Agents in panic who want to leave the room quickly and victims of stampede become obstacles.

How to solve the problem of evacuation of agents in panic? With the simulation model Helbing et al. showed a very counter intuitive solution. Put a column in front of the door, and less agents are clogging in front of the door, and the throughput of agents is higher. Agents will clog around the column, which spread the pressure over a larger distance. Although we don’t see people implementing the insights of this experiment, blocking emergency doors, the principle of using simulation models to design buildings such as stadiums and theatres is used.

    

Figure 11: Improving the flow of agents leaving the room when a column is put in front of the door.

Another counter intuitive result is that adding some additional space in corridors can lead to clogging. Suppose that you have a corridor like in Figure 12 where in the middle there is some extra room. Agents are going from left to right. Since agents increase their distance a bit when they are in the enlargement – they don’t want to be too close to each other – this leads that agents have to change back towards the middle of the corridor when the corridor narrows. This start reducing the speed of those outward agents moving back in, and after a while it leads to clogging.

  

Figure 12: The occurrence of delays due to wider space in the middle of corridor.

Finally some results of escape dynamics from a room where the visibility is limited, for example due to smoke. What will agents do? There are two doors, but the agents do not know the exact direction. If they all individually look for the door we see random movement of agents who in the end one by one find the door. However, agents may follow each other thinking that the other may know the right direction. But this herding behavior can lead to a dramatic event where agents are stuck in a corner, such as depicted in Figure 13.

  

Figure 13: Consequence of herding behavior in panic situation. 

Comments

Netlogo escape panic model available

This is not sort of a coding, but just wanted to know if there a model of “Simulating Dynamical Features of Escape Panic” (http://arxiv.org/pdf/cond-mat/0009448.pdf) available. I have searched every on internet but could not find it.(modelling commons, openabm, netlogo libraries etc). It seems to be very pivotal paper in the field of behaviour crowd. It has been also mentioned under(http://www.openabm.org/page/vote-models-you-would-see-replicated).

entropicity