From: Chloe Cortes Balcells (CIMNE- CENIT)
Last edited: 8 June 2020
The code that will be used to compute the simulations, will be the PEDFLOW code . It is a model based on a combination of force-based and agent-based methods . PEDFLOW has its own set of pre-processing tools that allow quick and easy import of geometrical/architectural data, specification of boundary conditions and pedestrian data, specification of diagnostics/output options, as well as all other runtime parameters.
Force-based and agent-based methods
Individuals move according to first Newton’s laws to define their movement. That means that they have will forces that seek global targets that are in motion while they are operating under a local level. So as to consider obstacles or other individuals, avoidance forces are introduced, and if needed, contact forces.
Speed of the model
This code has been parallelised for both shared, via OpenMP at the loop level, and distributed memory architectures, using MPI and domain decomposition. This parallelisation has allowed to simulate real time micro modelling of one million individuals where it was the first time that such speeds were achieved for micro-modelling pedestrians behaviour.
- Modelling of each individual’s specific fitness, behaviour, culture, size, age, group behaviour, etc.
- Correct over the complete range of densities.
- Extensively validated with experimental and real-life results.
- Best-in-class for accuracy, speed, generality, extendability and large number of pedestrians.
- Capable of accounting for persons with physical disabilities, wheelchairs, and vehicles.
- Capable of modelling millions of pedestrians faster than real-time.
- In development for 15 years
 Rainald Lohner, Muhammad Baqui, Eberhard Haug, and Britto Muhamad. Realtime micro-modelling of a million pedestrians. Engineering Computations, 2016
 Rainald Lohner, Britto Muhamad, Prabhu Dambalmath, and Eberhard Haug. Fundamental diagrams for specific very high density crowds. Collective Dynamics, 2:1–15, 2018