Fire simulation with FDS

Last November I downloaded something called Fire Dynamics Simulator, which is made by NIST (the US government’s National Institute of Standards and Technology).  If there’s one thing I admire about the United States government, it’s that it actually does useful stuff!  I’ve found other great sciency stuff on the NIST web site in the past (LaTeX documentation, physics data, etc.).

A couple of weeks after getting FDS, my house burned down.  So, today I finally got round to installing it and firing it up (har har!).

I’ve not read any manuals or tutorials.  All I did was notice some executable files and an examples directory.  It works pretty seamlessly.

  • fds5.exe an_example.fds runs the simulation, generating several files of output, including an_example.smv.
  • smokeview.exe an_example.smv loads the output and displays it in 3D.  There are options to enable particles, smoke, temperature diagrams, and other generated artifacts.

I’ve run a burning couch simulation, and a bucket simulation.  I’ve moved up to a room-on-fire simulation, which is grinding away as I write.  The default simulator executable does not use multiple cores, but apparently there are OpenMP enabled builds available.  I’ve got one of them going now.  It identifies my six cores but doesn’t appear to use more than 18% of the processor — and it runs noticeably slower than the default version. :/

After about 1000 steps, the simulation has got to this point:

1000 steps into the room_fire.fds example

This represents about 45 seconds into the simulation, which is clearly only enough time for the chair itself to start to burn — though long enough for it to get pretty smokey (using a wholly subjective assessment of how bad the smoke looks in the picture :p).

Now, in case you’re wondering, I’m not seriously planning to re-enact my own calamity with FDS…  I’m just resuming my general interest in physics simulation.

This entry was posted in Science, Simulation and tagged . Bookmark the permalink.

3 Responses to Fire simulation with FDS

  1. Halitus says:

    As per our irc chat:

    the openmp build is just as easy to run as the standard version. But if you get to the openMPI stuff running over ethernet its difficult and you need to fix geometry to work with n number of meshes. Smoke view is buggy but does the job reasonably well.

    I can send you more scripts to run if you are interested. Also feel free to discuss the software with me. I have been using the software during my Masters in Fire Engineering.

    • ejrh says:

      Well it was you who got me interested in FDS in the first place as I recall!

      OpenMP didn’t seem to accelerate anything much. There is any number of possible reasons for that though so I’m adding it to my list of things to try again later. I also tried OpenMPI but it appears to require at least two steps to get running which is too much for me today. :)

    • ejrh says:

      The house sim ran with the OpenMP version for about 21 hours of CPU time, and ended up using six cores after all. Perhaps the first few time steps are too simple for parallelism to kick in or something like that.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s