Crowd-Z

Crowd-Z (CZ) is a tangible and accessible solution for crowd simulations (CSs). It is a simple, easy to use, user-friendly, but also very flexible tool for CS on any given floor plan. Because working with CZ is very intuitive, it could be used not only by specialists, but also users like security staff e.g. at a football stadium. The dynamic animations of the simulations run smoothly on CZ and the user can control a number of parameters and interact with a running simulation. The model implemented in this framework is capable of reproducing realistic behavior. Due to its procedural architecture other CS models can be easily implemented in CZ. The information on each individual agent, such as position, local speed at each position, average speedsetc., can be stored for further analysis of the simulations. Below is a collection of links to selected CZ demonstrations.
The installation of the free Mathematica Player is required to run the demos.


Bottleneck

This demo runs the simulations in a bottleneck-scenario described in [1]. The set-up: pedestrians move from the holding area through the bottleneck of variable width w. An adjacent area is added to consider the backward effect of leaving pedestrians on those still in the bottleneck. The size of the holding area can be changed systematically to allow different initial densities of the agents.

Bottleneck


Rogsch Bottleneck Evacuation
In CZ two types of update schemes are separately implemented in the basic model: ordered sequential update (front to back - OSU) and random shuffled update (RSU). To examine the influence of these update schemes on the evacuation time the experiment described in [2] has been performed in CZ: 200 agents to clear a 25x25 cell room with a 3-cell wide exit. In the case of homogeneous crowd, Conservative agents with OSU clear the room faster than the Perky ones. Lazy agents are the slowest and demonstrate the same queuing behavior as in [2]. The evacuation time, however, can be prolonged by adding stochasticity to the agents' movements. This is implemented by error rate eR. For eR = 0 there is no additional randomness, whereas for eR = 1 the agents perform a random walk. By adjusting this parameter it is possible to extend the evacuation time so it matches the results from [2].

Bottleneck Evacuation


School
The size and character of the building are similar to the empirical case described in [3]. Three groups of 200 agents are initially distributed in different sections of the School (as shown below in magenta, green and blue) and are to leave the space through two exits (shown in red) at the shortest paths.

School Evacuation


St Peter's Basilica (BETA!)
In this demo the evacuation study can be perfimed on St Peter's Basilica. It is a well-known, truly large building; the documentation including digital images of the plan are easily acquirable from the Internet, there are multiple exits and it has already been used for validation by other researchers [4]. A digital image has been downloaded and resampled so that each pixel corresponds to a 40x40 cm^2 square area as shown below. The width at the transept is approximately 138 m, there are 15 assumed exits.

San Pietro


References
[1] J. Liddle, A. Seyfried, W. Klingsch, T. Rupprecht, A. Schadschneider, A. Winkens, An experimental study of pedestrian congestions: Influence of bottleneck width and length. ArXiv e-prints, 2009, 0911.4350. Conference proceedings for Traffic and Granular Flow 2009.
[2] C. Rogsch, A. Schadschneider, A. Seyfried, Simulation of Human Movement by Cellular Automata Models Using Different Update Schemes, Human Behaviour in Fire 2009 - Conference Proceedings, pp. 543--548, 2009
[3] H. Klupfel, T. Meyer-Konig, M. Schreckenberg, Comparison of an Evacuation Exercise in a Primary School to Simulation Results, in: S. F. et. al. (Ed.), Traffic and Granular Flow '01, Springer, Berlin, 2002.
[4] R. Lohner, On the modelling of pedestrian motion, Appl. Math. Model. 34 (2) (2009)366--382.