forrtl: severe (174): SIGSEGV, segmentation fault occurred

[jfpe37]

good morning
I got the attached error forrtl: severe (174): SIGSEGV, segmentation fault occurred, after 228.53 seconds.
(Blowers start working in 93.5 seconds)
I am attaching the FDS file after removing the build artifacts due to the need to maintain confidentiality.
I'm running Linux with 256 cores and 1 TB of memory.
(attaches the data).
I have a similar run (different simulation) on a Windows machine with 20 cores, which did not crash.

dose the line of code: &DUMP RENDER_FILE='XXX_-4A_004.ge1', DT_RESTART=25.0, DT_SLCF=2.0 could be related to this error

I would appreciate your help in understanding the error.
thanks
myrrh

XXX_-4A_004.txt

[Er9y714]

I see lots of HOLE elements. That indicates that the obstructions likely were originated to the removed geometry (which you mentioned as build artifacts). Can the problem be due to those removed elements?

Are the FDS and MPI versions the same on Linux and Windows machines?

If you investigate the results, do you see anything odd? High pressure, velocity error, number of iterations etc.?

It seems you are not running it on the latest version. Note that investigating an older version might not be a good idea since any possible issue here might have been fixed in the new versions. I recommend running it on the newer version.

[drjfloyd]

Without the complete input there is little that can be done to help. Would removing any identifying information in the input (e.g. any text giving the building name, location, etc.) be sufficient to anonymize the inputs? Unless this is a very visually unique building, it is unlikely it is going to be recongnized simply by its FDS geomtry inputs.

I see that you have both supply and exhaust fans. If you are completely specifying all suppply and exhaust with fixed volume flow fans and have no leakage or open doors/windos, then you are not going to conserve mass in the domain and over time the mass error can cause crashes. Equal volume flows of cold ambient air in and warm smoke layer air out will not have the same mass flow as they will differ in density.

[mcgratta]

When I run the input file, I get the error message

ERROR(820): VENT outlet SMOKE MAP _FXE-07 has no solid backing or an orientation index (IOR) is needed. (MPI Process:  9, CHID: XXX_-4A_004)

I am using the latest version of the source code. I suggest that you use the latest nightly version because we have added in the past few months some traps to stop the code when it detects a potentially problematic input.

[drjfloyd]

@mcgratta I think the OP removed some portions of the input file before posting.

[jfpe37]

Hi,
thank you for your answer.

what's the "nightly version", where can i find it ?
about the error -
when I upload the text file/fds, I have removed the OBST in the code. but in my complete fds file I have a solid map box up to the vent.
maybe you have another idea as to why the run is stuck and I get an error message: forrtl: severe (174): SIGSEGV, segmentation fault occurred.

thank you

[drjfloyd]

https://github.com/firemodels/test_bundles