-
Notifications
You must be signed in to change notification settings - Fork 38
Description
When the tutorial "heatTransfer/chtMultiRegionFoam/shellAndTubeHeatExchanger" is executed and you use the checkMesh function to check the mesh (in parallel):
mpirun -np 8 checkMesh -parallel -region shell
The shown interal points are negative and a warning is thrown:
Mesh stats
points: 667415
internal points: -4
--> FOAM Warning :
From function void Foam::printMeshStats(const Foam::polyMesh&, bool)
in file checkTools.C at line 65
Some processors have their points sorted into internal and external and some do not.
This can cause problems later on.
Log files (but you can easily run the tutorial by your own):
checkMesh_whole_output.txt
log.snappyHexMesh.txt
log.splitMeshRegions.txt
Expected behavior:
The internal points are positive and no warning is thrown. Similar to what happens when you execute mpirun -np 8 checkMesh -parallel, but only the region specified should be considered of the mesh created with snappyHexMesh (not the sourounding mesh from blockMesh).
What I found out:
- In this case it doesn't matter which region you check. Only the number of (negative) points changes.
- In an own multi region case I have two fluid and one solid region and the solid region works as expected, but the others don't.
- I didn't checked it in detail, but there are indications that if you run the case / snappyHexMesh reconstructed (not in parallel) the problem doesn't show up.
- It doesn't matter if you only run the meshing part or the whole case :D
Used OS and Version:
- Ubuntu 20.04.4 LTS
- mpirun (Open MPI) 4.0.3
- OpenFOAM-9
- Build: 9-b456138dc4bc
PS: This is my first GitHub entry so please be patient with me :D