Fast Fire Simulation
While fire simulation has developed a great deal in the wake of various fire-related catastrophies in the human built environment, it is still a very challenging topic for engineers. With ffire our fast fire simulation we are adding another powerful tool to design fire resistant environments.
Contrary to many established fire solvers, ffire was designed to produce results fast: it is easily 10x faster than many common solvers! This and its ease of use makes it easy for engineers to simulate more different scenarios, widening the options for what can be done within ever tighter deadlines. Solvers like fireFoam, FDS (Fire Dynamics Simulator) and ANSYS-CFX model the full combustion kinetics in addition to the flow and therefore suffer from geometric limitations and/or extremely high computational intensity.
Before looking into more details of our approach, let's look at an example: The following video shows the combustion of a small car (measured heat released curve) in a bus terminal (65m x 115m x 8m) with complex geometry - curved walls, tight spaces under buses and between ducts, etc.; the geometry was kindly contributed by Tian Green Building Engineering Singapore. The number of cells was ~3 Mio, the total simulation wall clock time was ~80 hours on 96 cores for ~20 minutes of fire simulation:
In ffire we simulate realistic, validated flow with temperature and buoyancy, radiation effects and also tracers like soot and carbon monoxide - the major indicators for survivability coupled with arbitrary, unsteady heat release curves that are readily available for many scenarios (a small car in the animation above).
The main advantages of our solver are:
- Extremely short solution times, allowing overnight simulations and testing of many scenarios
- Very good parallel speedup for huge cases
- Support for complex geometry (non-rectangular) and arbitrary (polyhedral) meshes
- Fully transient, measured (and also arbitrary) heat release rates supported, from a smoldering piece of paper over a sofa up to an entire truck
- Multiple file locations within one simulation
- Support for radiative heat transfer including conjugate heat transfer
- Smoke extraction through ducts, vents and fans
- Smoke visibility calculations
- Fuel agnostic - CO and soot release from conversion factors
Solver validation was done against the classic Steckler room experiment (Steckler, Quintiere and Rinkinen: Flow Induce by Fire in a Compartment, U.S. Department of Commerce). The following image shows the comparison of measured and simulated temperatures for the Steckler room experiment.
Follow the link to a pdf with a more comprehensive overview of our fast fire simulation and validation against the Steckler room experiments.
ffire is currently in beta stage and will be available soon, please contact us per email if you'd like to know more!