search

firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
660 stars 622 forks source link

BACKING='EXPOSED' not working with GEOM and multi-mesh #11872

Open rmcdermo opened 1 year ago

rmcdermo commented 1 year ago

@marcosvanella is this a known issue? We need to get this in the queue of stuff to fix. The case where I noticed this is here:

/home/rmcdermo/GitHub/FireModels_rmcdermo/fds/Validation/Lattimer_Tilted_Wall/FDS_Input_Files/Lattimer_50_kW_35_degree_fine.fds
johodges commented 1 year ago

Did the model error out for you or did it output a warning? When I ran the case with 6.8 it flagged a warning but did start.

Fire Dynamics Simulator

Current Date : June 1, 2023 14:38:58 Revision : FDS-6.8.0-0-g886e009-release Revision Date : Tue Apr 18 07:06:40 2023 -0400 Compiler : Intel ifort 2021.7.0 Compilation Date : Tue 04/18/2023 03:22 PM

MPI Enabled; Number of MPI Processes: 1 OpenMP Disabled

MPI version: 3.1 MPI library version: Intel(R) MPI Library 2021.9 for Windows* OS

Job TITLE : Surface temperature on slopped wall Job ID string : Lattimer_50_kW_35_degree_fine

WARNING: BACK CFACE search, other mesh NOM not equal to working mesh NM. NM= 4 , NOM and other cell IIO,JJO,KKO= 10 2 4 7 , intersection pt= -0.182223317628544 -0.302500000000000 0.260241867853140 WARNING: BACK CFACE search, other mesh NOM not equal to working mesh NM. NM= 4 , NOM and other cell IIO,JJO,KKO= 10 2 5 7 , intersection pt= -0.182223317628544 -0.293750000000000 0.260241867853140

WARNING: BACK CFACE search, other mesh NOM not equal to working mesh NM. NM= 10 , NOM and other cell IIO,JJO,KKO= 4 31 52 8 , intersection pt= -0.217084998428775 0.293750000000000 0.265745959190895 WARNING: BACK CFACE search, other mesh NOM not equal to working mesh NM. NM= 10 , NOM and other cell IIO,JJO,KKO= 4 31 53 8 , intersection pt= -0.217084998428776 0.302500000000000 0.265745959190895 Time Step: 1, Simulation Time: 0.02 s Time Step: 2, Simulation Time: 0.03 s Time Step: 3, Simulation Time: 0.05 s

rmcdermo commented 1 year ago

Just the warning. But I take it to mean that EXPOSED is not working in this case and that some other backing condition is being applied? In general, I think this is something we need to discuss. I think it should basically function like the OBST with similar limitations.

mcgratta commented 1 year ago

There is no exchange of exposed CFACES from mesh to mesh. There is for WALL cells, but not CFACE.

rmcdermo commented 1 year ago

I think this is something we should add. If I create an OBST that is one cell thick with an exposed backing. And then turn that OBST to a GEOM, I think we should be able to get the same functionality. Not saying it is trivial. But I think this should be on the list for v7.

marcosvanella commented 1 year ago

Kevin is right. The back intersection point is computed from the input geometry (all MPI processes have it) but the infra to exchange these intersection points, figure out the back cfaces in other meshes, and exchange information of these among processes hasn't been developed. Currently, it will use 'VOID' in these cases.