In this fourth post on Computational Fluid Dynamics (CFD) we are going to study the evacuation of smoke in a garage when a vehicle bursts into flames. For this fire simulation case we will use the Fire Dynamics Simulator software together with the BlenderFDS add-on that allows us to use the Blender software for the preparation of the models.
Pre-processing with BlenderFDS
The study begins with the geometry of a garage created in CAD software and imported into Blender. First, the surfaces must be grouped according to their thermal properties and assigned with the appropriate FDS Keywords depending on whether they are solid surfaces or openings.
The following image shows the geometry of the garage, hiding the ceiling representation so that the interior can be seen.
The general conditions of the analysis, such as the simulation time or the parameters of the combustion reaction, can also be defined in Blender. For this example, a fuel with a CO and smoke generation efficiency of 10% and a heat release rate of 400 W/m2 at the front surface of the car is selected.
The next step is to create the mesh. FDS needs to mesh prismatic volumes with a rectangular base. Also, if two partial meshes share a face, one of the meshes must have an element size multiple of the one chosen in the other prism. BlenderFDS allows to visualise these meshes to facilitate the fitting in the pre-processing.
Definition of results to be obtained
After adjusting the meshes, the desired results to export must be defined. The most interesting ones in these simulations are usually the smoke generation, which is exported by default, and the values of temperature and species concentration in the collection of planes considered relevant. It is common to define planes parallel to the ground at different heights to see the clean air concentrations at different slices of the room.
To obtain quantitative data, FDS allows the definition of species concentration sensors at different points in the domain, which can be placed for example on defined evacuation routes to check that the amount of smoke in these areas does not exceed a certain limit. These sensors generate a CSV file with the set of results measured over time. They can be defined in BlenderFDS or externally in the generated text file, like the rest of the elements. In this way, optimisations of the smoke extraction systems of the enclosure can be studied.
Fire simulation with FDS
Once the model is created, Blender exports the text file with extension “.fds” that can be executed directly from the terminal. A good practice when creating these models is to export them with a simulation time equal to 0s, to run them in FDS and check in the post-processor that no errors are being made in the different stages. For example, you can check that the geometry definition in FDS is correct or that the fire start position is the desired one.
The default fire simulation in FDS uses the LES turbulence model, which allows to see in detail the movements of the flow generated by the fire. The software allows temperature and gas concentration maps to be studied, making it a very useful tool for studies of low velocity flow and significant thermal component with buoyancy movement. It is also highly optimised for combustion modelling and the corresponding multiphase simulation.
Visualising results with Smokeview
When the simulation is finished, the results can be displayed in the Smokeview post-processor, which is available as open source software like FDS. To display the results, the display speed, the magnitude scales in the slices and other parameters can be adjusted. As an example, a couple of images of the smoke evolution in the garage are shown.