Center for Computational Fluid Dynamics

Pedestrian Flow and Crowd Simulation

St. Peter’s, Rome:

The safe operation of large pilgrimage centers is a key requirement for event planners and security personnel. This simulation shows a possible scenario at St. Peter’s in Rome. Different views of the same simulation are shown below.

NIST 10-Story Building:

This simulation shows a detailed simulation of a fire drill in the Johnson Center at George Mason University. The input data was taken from camera footage in the facility. The results obtained are shown here.

Evacuation of a Theater:

The safe egress from theaters and cinemas is a building requirement that must be verified. This simulation shows a possible scenario for the evacuation of a theater.

Evacuation of a Stadium:

Given the long history of fatalities, the safe egress from stadiums has always been of primary concern to architects, civil engineers and stadium operators. This simulation shows a typical section of a stadium. One can clearly see the queueing behavior of the pedestrians, as well as the resulting regions of high density.

Post-Blast Evacuation in Metro Station:

The safe evacuation of metro stations is of eminent importance. This simulation shows the evacuation of the Atocha Metro Station in Madrid, Spain, moments after the terrorist attack. The self-explanatory movie is shown here.

Evacuation of Metro Station During Fire Alarm:

This simulation shows the combination of Computational Fluid Dynamics (CFD) and Computational Crowd Dynamics (CCD). It is assumed that a fire has started in one of the stairs. The flow- and temperature field is computed with the CFD code FEFLO. This data is then imported by PEDFLOW, so that the pedestrian motion takes into account both visibility and smoke inhalation. The results obtained are shown below.

Evacuation of a Complete City:

This simulation shows a near real-time simulation of the city of Barcelona. Approximately 1.4 million pedestrians are considered. The street-plan is very accurate, as well as the initial population distribution. The run was performed on a massively parallel machine using nearly 400 processors. The results obtained are shown here.