CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Meshing & Mesh Conversion (http://www.cfd-online.com/Forums/openfoam-meshing/)
-   -   decomposePar corrupting mesh (http://www.cfd-online.com/Forums/openfoam-meshing/109401-decomposepar-corrupting-mesh.html)

ski November 16, 2012 10:57

decomposePar corrupting mesh
 
Hi all,

I have a mesh which passes checkMesh serial no problems. However if I use decomposePar, and run checkMesh in parallel, it fails for the same mesh as follows:

====================

Face pyramids OK.
***Max skewness = 4.28888, 475 highly skew faces detected which may impair the quality of the results
<<Writing 475 skew faces to set skewFaces
Coupled point location match (average 2.60702e-16) OK.

Failed 1 mesh checks.

====================
The number of skewed faces depends on the number of processors I use (with greater #processors seemingly corrupting the mesh more)

My decomposeParDict:

====================
numberOfSubdomains $NSLOTS ;

method scotch;


preservePatches
(
cyc_half0
cyc_half1
);
====================

I have cyclic boundaries in spanwise direction (which are exactly the same on patch cyc_half0 and cyc_half1). Not sure if this could have an effect?

Any suggestions?

Thanks

ski November 19, 2012 08:25

Does anyone else see this happen at all? The mesh does seem to work OK, but it strikes me as odd that the cell quality should deteriorate by domain decomposition. decomposePar should not be affecting the coordinates of the mesh vertices, so this shouldn't happen...

Is this a bug?



Quote:

Originally Posted by ski (Post 392565)
Hi all,

I have a mesh which passes checkMesh serial no problems. However if I use decomposePar, and run checkMesh in parallel, it fails for the same mesh as follows:

====================

Face pyramids OK.
***Max skewness = 4.28888, 475 highly skew faces detected which may impair the quality of the results
<<Writing 475 skew faces to set skewFaces
Coupled point location match (average 2.60702e-16) OK.

Failed 1 mesh checks.

====================
The number of skewed faces depends on the number of processors I use (with greater #processors seemingly corrupting the mesh more)

My decomposeParDict:

====================
numberOfSubdomains $NSLOTS ;

method scotch;


preservePatches
(
cyc_half0
cyc_half1
);
====================

I have cyclic boundaries in spanwise direction (which are exactly the same on patch cyc_half0 and cyc_half1). Not sure if this could have an effect?

Any suggestions?

Thanks


vbchris November 19, 2012 14:46

Try

preservePatches (cyc_half0);
preservePatches (cyc_half1);

preservePatches may only take 1 augment.
I detailed my cyclic BC issues in the following thread:

http://www.cfd-online.com/Forums/ope...esh-error.html

ski November 20, 2012 07:51

Quote:

Originally Posted by vbchris (Post 393044)
Try

preservePatches (cyc_half0);
preservePatches (cyc_half1);

preservePatches may only take 1 augment.
I detailed my cyclic BC issues in the following thread:

http://www.cfd-online.com/Forums/ope...esh-error.html

Thanks for the tip. I was not aware of this. Although it has not solved the problem sadly. checkMesh is still reporting highly skewed cells.


All times are GMT -4. The time now is 18:54.