Link cutcells across mesh boundaries

[rmcdermo]

Describe the bug When a ZVALS terrain crosses a mesh boundary it looks like we have some old code that blocks cells.

To Reproduce Run the attached case and you will see the terrain looks fine with the 32 mesh set, which does not have a mesh boundary cutting the geometry. Comment the 32 mesh set and uncomment the 64 mesh case, which splits the lower set of meshes and then you see the blocked cells.

Expected behavior Should look like this one. The grid resolution does not change. I think this case (ZVALS) slipped through the cracks when the other mesh to mesh errors where fixed for GEOM.

Screenshots

Desktop (please complete the following information):

• OS: linux FDS, macOS Smokeview
• Version: latest source for FDS, SMV-6.8.0-2080-gd3d300ce6-test

[rmcdermo]

Sorry, here is the test case.

topo_demo.fds.txt

[marcosvanella]

I'll take a look

---

From: Randy McDermott @.> Sent: Thursday, March 7, 2024 04:26 PM To: firemodels/fds @.> Cc: marcosvanella @.>; Assign @.> Subject: Re: [firemodels/fds] Issue at mesh boundaries using ZVALS for GEOM (Issue #12596)

Sorry, here is the test case.

topo_demo.fds.txthttps://github.com/firemodels/fds/files/14530269/topo_demo.fds.txt

— Reply to this email directly, view it on GitHubhttps://github.com/firemodels/fds/issues/12596#issuecomment-1984529556, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABY23VLT4HV5IDRD2K6ORJTYXDLQJAVCNFSM6AAAAABELX24JCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBUGUZDSNJVGY. You are receiving this because you were assigned.Message ID: @.***>

[marcosvanella]

Try this with latest source. You'll still see some cut-cells blocked (unlinked), but not the line of cubes.

[rmcdermo]

When I try to run with the latest code I get a seg fault. Here is the debug output.

 Number of MPI Processes:  64

 MPI version: 3.1
 MPI library version: Intel(R) MPI Library 2021.7 for Linux* OS


 Job TITLE        : Demo of FF3D with geom at Fishlake National Forest
 Job ID string    : topo_demo

forrtl: severe (408): fort: (3): Subscript #1 of the array BOUNDARY_COORD has value 0 which is less than the lower bound of 1

Image              PC                Routine            Line        Source             
fds_impi_intel_li  00000000011EB7F6  complex_geometry_        8952  geom.f90
fds_impi_intel_li  0000000001F7E6E1  cc_scalars_mp_ini        8001  ccib.f90
fds_impi_intel_li  000000000418E76E  MAIN__                    412  main.f90
fds_impi_intel_li  0000000000409B1D  Unknown               Unknown  Unknown
libc-2.17.so       00002B9157074555  __libc_start_main     Unknown  Unknown
fds_impi_intel_li  0000000000409A36  Unknown               Unknown  Unknown
srun: error: burn007: task 27: Exited with exit code 152
srun: launch/slurm: _step_signal: Terminating StepId=141144.0
slurmstepd: error: *** STEP 141144.0 ON burn003 CANCELLED AT 2024-03-15T13:20:53 ***
srun: error: burn003: tasks 0-10: Segmentation fault
srun: error: burn011: tasks 44-53: Segmentation fault
srun: error: burn012: tasks 54-63: Segmentation fault
srun: error: burn007: tasks 22-26,28-32: Segmentation fault
srun: error: burn004: tasks 11-21: Segmentation fault
srun: error: burn008: tasks 33-43: Segmentation fault

[marcosvanella]

Same case from the op? This is what I see in blaze (dv mode):

Fire Dynamics Simulator

 Current Date     : March 15, 2024  13:34:19
 Revision         : FDS-6.8.0-1696-gf37b64a-master
 Revision Date    : Thu Mar 14 11:42:57 2024 -0400
 Compiler         : Intel(R) Fortran Intel(R) 64 Compiler Classic for applications running on Intel(R) 64, Version 2021.7.1 Build 20221019_000000
 Compilation Date : Mar 15, 2024 13:30:27

 Number of MPI Processes:  64

 MPI version: 3.1
 MPI library version: Intel(R) MPI Library 2021.7 for Linux* OS


 Job TITLE        :
 Job ID string    : topo_demo_64

 Time Step:      1, Simulation Time:      0.10 s

STOP: FDS completed successfully (CHID: topo_demo_64)

[marcosvanella]

Let me try the debug target.

[rmcdermo]

OK, I still had RESTART=T. The case starts now.

[rmcdermo]

@marcosvanella The main problem is solved. I think we should leave this open until we get linking working across meshes. We should be able to set the CELL_BLOCK_IOR=-3 for this case and not have any blocked cells on the flow side of the terrain.

[rmcdermo]

Additionally we should set CELL_BLOCK_IOR=-3 by default for any terrain (ZVALS or IS_TERRAIN) case.

[rmcdermo]

Related to the cell linking, this case has an instability if CCVOL_LINK=0.1.

debug_instability.fds.txt

[marcosvanella]

Ok will look into this. Just pushed setting default cell blocking IOR to -3 for terrains.