problems with evacuation cases

[gforney]

Timo, I'm having trouble running the evacuation cases in the repository.  I'm using
FDS built with the latest revision.  Here is a summary of what  I’ve found. 

evac_memory_test0.fds – ERROR: SLCF QUANTITY visibility not found
evac_memory_test1.fds – ERROR: SLCF QUANTITY visibility not found
evac_memory_test2.fds – crashes

evac_smv_testcase0.fds – runs ok (but no people - I assume that is your intent?)
evac_smv_testcase1.fds –crashes
evac_smv_testcase2.fds – crashes

Any ideas?

Original issue reported on code.google.com by gforney on 2011-04-26 15:21:07

[gforney]

Glenn,

We have gotten rid of supporting legacy quantity names.  All quantities are now all
caps and have no '_' characters.  So 'visibility' should be 'VISIBILITY'.

I'll give a quick look at the crashed cases to see if we broke something with all the
species work we have been doing.

Original issue reported on code.google.com by drjfloyd on 2011-04-26 18:10:29

[gforney]

These crashes appear to be related to trying to compute physical parameters on EVACUATION_ONLY
meshes.  I'll leave those to Timo to make sure that they are fixed correctly.

evac_memory_test2.fds it looks like a number of RGB inputs have real numbers rather
than integers.

Original issue reported on code.google.com by drjfloyd on 2011-04-26 18:24:57

[gforney]

thanks.  I just changed visibility to VISIBILITY so now evac_memory_test0.fds and evac_memory1.fds
crash like other cases.  Also evac_smv_testcase3.fds has a problem with the MISC line.
 Here is an updated list

evac_memory_test0.fds – crashes
evac_memory_test1.fds – crashes
evac_memory_test2.fds – crashes

evac_smv_testcase0.fds – runs ok (but no people - I assume that is your intent?)
evac_smv_testcase1.fds –crashes
evac_smv_testcase2.fds – crashes
evac_smv_testcase3.fds – problem with &MISC line

Original issue reported on code.google.com by gforney on 2011-04-26 18:25:24

[gforney]

Issue 1391 has been merged into this issue.

Original issue reported on code.google.com by drjfloyd on 2011-04-26 18:33:00

[gforney]

Well, I'll look this issue. I should update the test cases
so that they will work with the latest fds source. There
has been some input changes. And the SVN cases are quite
old, they might be "old evacuation input format". I should
make them to be new format (EVAC_FDS6=.TRUE. on some PERS
namelist). There will be a new input format for FDS 6. 
Actually, EVAC_FDS6=.TRUE. is now the default in the 
source code.

Timo

Original issue reported on code.google.com by tkorhon1 on 2011-04-27 06:24:35

[gforney]

Hi all,

I just committed new input files for the six simple evacuation input cases.
I did use:

Fire Dynamics Simulator
Compilation Date : Sun, 24 Apr 2011
Version: 6.0.0; MPI Disabled; OpenMP Disabled
SVN Revision No. : 8178

The errors were probably due to old inputs, because I did not
have any errors (well, it was not a debug compilation, just
a normal compilation). Platform was my laptop, i.e., a win32 XP
and compiler options also for win32. I should add these test
cases to my verification suite, which is not yet in SVN, see
below. So, check that you do not get the errors anymore.
Note also that these test cases (and all my V&V cases) are
now just for the simple chemistry. I should find a nice room
fire case with other chemistry (the new FDS 6 chemistry)
and put there some evacuation calculation to test the
more advanced chemistry.

Below is the svn-log-edit stuff:

Evacuation verification: Updated the simple evacuation verification
cases input files, see Issue 1390.  Note that these cases are just a
silly test cases that only test some aspects of the evacuation
calculation.  The V&V cases in the evacuation guide are not yet in the
SVN, because some of those demand too much CPU and some too many runs
(stochastic input) and I (Timo) have not yet made automatic scripts
for all cases to plot the result pictures. Now I should open the Excel
files which update automatically the csv result files and then I
should "print" the figures manually.  But I should put my verification
cases soon to the SVN folder.

Timo

Original issue reported on code.google.com by tkorhon1 on 2011-04-27 07:39:18

[gforney]

I only had problems with my newer case when I added back in the fire information (removed
the EVACUATION_DRILL=,TRUE. and added back in the REAC lines updated to work with the
latest version.  It seems to die when trying to assign species information for EVAC
only meshes when you have a fire in the simulation. With the REAC lines removed, it
runs the evacuation fine. 

File (which should be in the updated FDS_Evac format) is attached.  Both MPI and non
MPI-versions give similar errors (though only the debug MPI give the traceback information):

forrtl: severe (408): fort: (7): Attempt to use pointer ZZP when it is not associated
with a target

Image              PC                Routine            Line        Source        

fds_mpi_intel_lin  000000000228031A  Unknown               Unknown  Unknown
fds_mpi_intel_lin  000000000227EE95  Unknown               Unknown  Unknown
fds_mpi_intel_lin  000000000222B766  Unknown               Unknown  Unknown
fds_mpi_intel_lin  00000000021CEE75  Unknown               Unknown  Unknown
fds_mpi_intel_lin  00000000021CF2C9  Unknown               Unknown  Unknown
fds_mpi_intel_lin  0000000001D341E3  velo_mp_compute_v          93  velo.f90
fds_mpi_intel_lin  0000000001D32247  velo_mp_compute_v          36  velo.f90
fds_mpi_intel_lin  00000000020851C5  MAIN__                    623  main.f90
fds_mpi_intel_lin  000000000042A4CC  Unknown               Unknown  Unknown
libc.so.6          00000037B241D994  Unknown               Unknown  Unknown
fds_mpi_intel_lin  000000000042A3D9  Unknown               Unknown  Unknown

Original issue reported on code.google.com by cfastdev on 2011-04-27 11:45:39

---

• Attachment: Evacnew.fds

[gforney]

Well, I found this error some 5 days ago and fixed it.
But I did not remember to commit it to SVN. So, I'll
do it right now.

SVN 8200 should work.

These things will continue to happen as long as the
velo and divg keep changing...

TimoK

Original issue reported on code.google.com by tkorhon1 on 2011-04-27 13:46:08

[gforney]

Since the evacuation meshes are only one cell high, is there any harm (other than a
small amount of memory) in allowing the ZZ arrays to be allocated for them?  This would
avoid these errors from cropping up. 

Original issue reported on code.google.com by drjfloyd on 2011-04-27 13:55:07

[gforney]

ZZP(I,J,K,1:N_TRACKED_SPECIES)

Well, it does not need much memory, so I could allocate it.
I'll do it next time, i.e., when the allocation of zzp is
once again the problem. Well, perhaps not. There might be
evacuation meshes for stairs that have k>1 in the future.
But for these meshes I do not calculate anything, so that
is not going to be a problem. I do not need any flow fields
in stairs for the agents, there the directions are easy: up
or down. So, I should allocate/define the zzp for evacuation
meshes also. But, as said, I'll do it the next time.

Timo

Original issue reported on code.google.com by tkorhon1 on 2011-04-27 14:04:32

[gforney]

I'll close this issue.

TimoK

Original issue reported on code.google.com by tkorhon1 on 2011-06-08 08:56:18