Fire and evacuation problem

[Guillem]

Hi, once more.

We have a problem with a simulation. We simulate evacuation and fire at the same time, and when the time arrive at 7 seconds, the fire simulation stops, and the evac side, go on. it's so strange, there's no error messages... We launch several times the simulation and occurs the same. Can somebody help us?

Regards,

Guillem.

[TimoK]

Well, can not help, I need more intormation. Are you running a MPI parallel job?
If yes, try to run it using serial version (fds5.exe). If you have problems with
your computer memory size, reduce the number of grid cells so that the
calculation fits a single machine.

Read the file: http://virtual.vtt.fi/virtual/proj6/fdsevac/documents/Readme_InputErrors.txt
and follow the steps there.

1) MISC-namelist: "EVACUATION_DRILL=.TRUE." so you can
   have the fire meshes there. They are just not read in.

Is your calculation running nicely when just calculating
the evacuation part? (fire drill mode)

If this is true, put NO_EVACUATION=.TRUE. and EVACUATION_DRILL=.FALSE.
at the MISC namelist, i.e., run just the fire calculation. If this
is  running nicely, then there is, indeed, something strange out there...

TimoK

[Antoni Balaguer]

Thanks Timo,

I'm working with Guillem. We are running an OpenMP job, but it seems that the problem is fixed. We had SYNCHRONIZE=.FALSE., but now we have put it to TRUE and it's working.

Although now we have another doubt. We think that we can do just one fire + evacuation simulation and then repeat more evacuation simulations using the FED output but we can change some evac parameters. The doubt is if we can see in Smokeview the first fire simulation with another later evac simulation case.

[TimoK]

Well, you can do a fire+evacuation simulation (say, CHID='caseA') to get caseA_evac.fed (and .eff also...).
Then you do evacuation simulation (CHID='caseB') putting EVACUATION_MC_MODE=.TRUE. on the
MISC namelist and doing a copy caseA_evac.fed => caseB_evac.fed and run the caseB.

The caseB has less meshes (no fire meshes) than the caseA, so the FDS output files are
named a little bit different, the mesh numbers are changed accordingly. But you should
be able to figure out which mesh number of caseB corresponds to caseA evacuation meshes.
Then you have two choises how to see the caseB results with caseA smoke:

 1) copy caseB_####.prt5 => caseA_####.prt5 for all main evacuation meshes. The
    prt5 files (of the main evacuation meshes) contain the agent information. There might
   also be caseA prt5 files for the fire meshes, these contain fire related "particles".
   The "####" are the mesh numbers that are a little bit different in caseA and caseB,
   because there are no fire meshes in caseB, so do the copying correctly.  After this,
   open the caseA.smv and see the results.

   Note: Do a "backup" copy of your caseA results before the copying actions...

 2) Edit (with some text editor) the file caseA.smv.  You have there lines like:

EVA5     4    1.0000
 Atrium_0004.prt5
  1
  1
EVA5     5    1.0000
 Atrium_0005.prt5
  1
  1

 So change the CHID (caseA => caseB) and also the "####" (mesh number) like
 in the case 1 above (caseB has no fire meshes). Note that you should not touch
 the lines "EVA5  5  1.000". The "5" is there the mesh number, but you should not
 change this, you just change the file name. The caseA.smv has still all the
 meshes there (fire+evacuation).

A good strategy for a FDS+Evac input file and mesh definitions:
Define first all fire meshes and then the evacuation meshes. This
is needed for the MPI calculation, serial does not need this, but
this is still a nice choise. Then you mesh number changes are
like #### (caseA) => #### - NumberOfFireMeshes (caseB).

TimoK

[Guillem]

Thanks, Timo!

[Antoni Balaguer]

Well, it doesn't work very well. Perhaps I have done something wrong.

I have copied the "caseA.fed" and the "caseA.eff" and renamed them into "caseB.fed" and "caseB.eff" respectively (I'm not sure if I need to copy the .eff). I have put EVACUATION_MC_MODE=.TRUE. on the MISC namelist. The agents in the caseB evacuation simulation doesn't decrease his walking speed due to the smoke effects and his FED is zero.

I don't understand if I have to change something in the caseA.smv, I renamed the prt5 files from the caseB changing his name to caseA and increasing the #### by the number of fire meshes. I think that the EVA5 lines are correct without changing anything in the .smv file. In fact in Smokeview results there are the caseA smoke and fire and the agents of the caseB, but without FED and smoke effects.

[Antoni Balaguer]

Ok, I have seen that I have EVACUATION_DRILL=.TRUE. 

I'll try putting it to .FALSE.

[TimoK]

EVACUATION_DRILL=.TRUE. means that no fire (e.g., smoke) information
is used, so no FED file  is  read in nor (re)calculated. To read in the FED
file, you should have EVACUATION_DRILL=.FALSE., i.e., you do want
to do fire+evacuation calculation. But if you want use an existing fire
calculation, then you are doing "Monte Carlo" type of calculation, 
many different evacuation runs with one smoke information.
You have some stochastic stuff in the evacuation part, so for
each fire calculation you should do many evacuation calculations.
So, set EVACUATION_MC_MODE=.TRUE.. If no FED file is
found, then it is recalculated, i.e., the fire meshes are also done.
If the FED file is found, then it is read in and no fire meshes are
calculated.

Well, Antoni got this already, but I wrote it here just to remind
other users how it works.

TimoK