EVAC access violation

[GoogleCodeExporter]

Application Version:5.3.1
SVN Revision Number:3881
Compile Date:01.05.09
Operating System:Win XP

After the FDS starting statement an error occurs. The simulation without
EVAC meshes & lines runs just perfectly.

forrtl: severe (157): Program Exception - access violation
Image              PC        Routine            Line        Source
fds5.exe           004D286E  _DUMP_mp_GAS_PHAS        3504  dump.f90
fds5.exe           004D7B00  _DUMP_mp_DUMP_SLC        3041  dump.f90
fds5.exe           00513F75  _DUMP_mp_DUMP_MES          84  dump.f90
fds5.exe           00660542  _MAIN__                   206  main.f90
fds5.exe           00699C90  Unknown               Unknown  Unknown
fds5.exe           00670479  Unknown               Unknown  Unknown
kernel32.dll       7C817077  Unknown               Unknown  Unknown

Any idea? Already tryed solutions of:
http://code.google.com/p/fds-smv/issues/detail?id=277
http://code.google.com/p/fds-smv/issues/detail?id=732

Attached my entrys for evac, which I put on my simulation.

Original issue reported on code.google.com by magnus.w...@gmail.com on 9 Jul 2009 at 1:29

Attachments:

• evac-entrys_mm.txt

[GoogleCodeExporter]

Ah perhaps useful information: The setup with TWFIN=0.0 works fine, also the 
files
like smv, csv and slf are created before the error occurs.

Original comment by magnus.w...@gmail.com on 9 Jul 2009 at 1:32

[GoogleCodeExporter]

I ran this case with the debugger turned on, and there are a number of 
problems. I 
don't know enough about the evac routines to diagnose the problems. It will 
have to 
wait for Timo Korhonen.

Original comment by mcgra...@gmail.com on 9 Jul 2009 at 1:54

[GoogleCodeExporter]

Magnus,
can you post your complete input-file?

Original comment by gregor.j...@gmail.com on 9 Jul 2009 at 4:22

[GoogleCodeExporter]

I 'll mail it, I wouldn't like to post the complete file on the net as it's 
sensitive
for my thesis ;-)
Thx Kevin for first checks.

Original comment by magnus.w...@gmail.com on 10 Jul 2009 at 7:02

[GoogleCodeExporter]

Dear Magnus,
thanks for your Mail. I'll look at the weekend. You hear from me.
Gregor

Original comment by gregor.j...@gmail.com on 10 Jul 2009 at 9:32

[GoogleCodeExporter]

Well, I'm at the office tomorrow, so I will check this.
Next time I'm at the office on Thursday morning. If I could
get the input file, then I could check it.

TimoK

Original comment by tkorh...@gmail.com on 10 Jul 2009 at 6:18

[GoogleCodeExporter]

I have send it via mail.
Thank you.

Original comment by magnus.w...@gmail.com on 10 Jul 2009 at 9:29

[GoogleCodeExporter]

Hi, thanks for the input. I did a fast check (no fire meshes)
and noticed that there was something wrong in the evac-mesh:

 Timo: Next is old evacuation mesh line xb:
 MESH IJK=348,100,1 XB=-10.44,10.44,-3.00,3.00, 0.40,0.40, 
 Timo: Next is the new one, the z1 should not be equal to z2
&MESH IJK=348,100,1 XB=-10.44,10.44,-3.00,3.00, 0.35,0.45, 

So, the evacuation mesh can not be zero thick in the z-direction.
The program should do this check, but it does not do it yet.

Check that the case is also running with the fire meshes.
Note, that I run my own version, more or less the latest
on the SVN.

 Compilation Date : Wed, 01 Jul 2009
 Version          : 5.4.0 Serial
 SVN Revision No. : 4308

I'll keep this issue open until I (or somebody else) has made
the z1 < z2 check for the (evac)meshes.

Then some general comments on the evacuation input:

 Timo: XYZ changed, old was XYZ= -5.16, -1.26, 0.40,
 XYZ is better to be at the middle of the door and a little bit
 in front of the door (this is the default, if no XYZ is given)
&EXIT XYZ= -4.5, -1.20, 0.40, XB=-3.840, -5.160, -1.260, -1.260, 0.400, 0.400, 
      ID=2,IOR=+2, TIME_OPEN=900./
&VENT XB=-5.16,-3.85,-1.26,-1.32,0.40,0.40, SURF_ID='OUTFLOW', 
      MESH_ID='MainEvacGrid',IOR=+2, EVACUATION=.TRUE.,RGB=0,0,255/Exit Fan
&OBST XB=-5.16,-3.85,-1.46,-1.32,0.40,0.40, SURF_ID='INERT', EVACUATION=.TRUE./

Do the same change for the other door. XYZ is used to decide if the door
is visible (line of sight) or not. So, if you give an XYZ, which is inside
an OBST, then the door is never visible.

And then a comment, that probably will not deal with the svn3881
test executable, but will be there if you compile the latest
SVN source code: 

 Timo: The fds 5.4.0 will need "EVACUATION= .TRUE." for the 
 slcf files of the evacuation meshes.  This is needed to reduce
 the output to the disk. User usually does not want the slcf
 files for the fire meshes at the evacuation mesh location
 (at least not velocity vector slices).
 SLCF PBZ=0.4, QUANTITY='VELOCITY',VECTOR=.TRUE./ 
&SLCF PBZ=0.4, QUANTITY='VELOCITY',VECTOR=.TRUE., EVACUATION=.TRUE./ 

And there seems to be something else wrong in your input,
the above evacuation flow field seems to be "zero" everywhere,
so something wrong with the evacuation vents.

TimoK

Original comment by tkorh...@gmail.com on 11 Jul 2009 at 8:45

• Changed state: Fixed

[GoogleCodeExporter]

This is more than two months been 'Fixed', so I change it to 'Verified'.

TimoK

Original comment by tkorh...@gmail.com on 24 Sep 2009 at 11:06

• Changed state: Verified