Documentation

Documentation

jupedsim_network estimates how long the occupants of a building need to reach safety. Rooms, corridors and stair flights are nodes with an area. Doors, openings and stair entries are links with a flow capacity. Agents move through this graph by relations adapted from the SFPE hydraulic model. A run of a ten-storey building with 600 agents takes about 0.1 s, so a scenario can be sampled hundreds of times.

One time step: node densities, pre-movement, walking to the next queue, passing links and applying all transfers at once, around the shared network state of agents, queues and carries

Each time step of length $\Delta t$ runs these five stages in order and then repeats. Passing links is where capacity, carry, counterflow and the supply limit of a full node act (Update scheme).

Version 0.5.0, a prototype. The model is verified against hand calculations and adapted IMO component tests. It is not validated against drills or experiments, and it is not intended for regulatory or design use. See Limitations.

New here. Install the package, run a first scenario, then work through a complete example. Without Python, run jupedsim-network init and then jupedsim-network run scenario.json (Command line), or edit and run the file in forms in the Terminal UI.

Getting started
Install, run one building, and check the result.
Worked example
An office wing from floor plan to bottleneck and one design change.

Building your own scenario. Describe the building, the occupants and the run, and read the results:

Networks
Rooms, stairs, safe places and the links between them.
Populations
Agents, their inputs and the sampling distributions.
Running
Seeds, Monte Carlo, time step, max_density and t_max.
Results
What a run returns and how to report it.
Command line
Run a JSON scenario file and write the results to CSV and JSON.
Terminal UI
Edit, check and run a scenario file in the terminal, with the same results as the command line.

Checking the model. How it works, how it is tested, and where it stops:

Model
Nodes, links, agents, routes and the update scheme.
Verification
Hand calculations, IMO tests and max_density sensitivity.
Case study
A ten-storey building with two stairs.
Limitations
Everything the model does not do, or does in a simple way.
Comparison with EvacuatioNZ
Where this model follows EvacuatioNZ and where it differs.
References
Sources cited on these pages.

The evacuation_time of a run is the pre-movement time plus the movement time of the simulated occupants. Detection and alarm times are not modelled unless they are added to the pre-movement time, so the output is an RSET only under that condition.

Last updated on