[giorgianig]

Dear all,

I am working on a radiation problem. I am using the viewFactor model, and the faceAgglomeration utility for speeding up the computation. I get the following error on one of the patches:

agglomeration does not create a single, non-manifold face for agglomeration 41 on patch 2

If I don't do agglomeration, the viewGen routine works but then I have a huge number of coarse faces and it takes ages to complete.
Does anyone knows why this error happens?

Thanks in advance.

[giorgianig]

No one?

Just to add some more info, if I run viewFactorGen without previously generating the agglomeration, it starts running but

Total number of coarse faces: 659510 

So, impossible to make it.
Here is my viewFactorsDict

writeViewFactorMatrix     true;
writeFacesAgglomeration   true;
writePatchViewFactors     true;

"v_.*"
{
nFacesInCoarsestLevel 100;
featureAngle 10;
}

If I increase a lot the nFacesInCoarseLevel it starts working, but again with a huge number of faces, so no way.

Also, my mesh is very good, I really don't see why the agglomeration should fail. Here is my checkMesh

Checking geometry... 
Overall domain bounding box (-0.92900002 -0.88300002 0) (0.92900002 0.88300002 2.154)
Mesh has 3 geometric (non-empty/wedge) directions (1 1 1)
Mesh has 3 solution (non-empty) directions (1 1 1)
Boundary openness (-6.0884129e-15 1.4447539e-14 -1.6253704e-14) OK.  Max cell openness = 3.6583714e-16 OK.
Max aspect ratio = 13.975798 OK.                                                                                                                                                                                                             Minimum face area = 6.6415428e-06. Maximum face area = 0.00064417908.
Face area magnitudes OK.
Min volume = 2.0805547e-08. Max volume = 1.7985865e-05.  Total volume = 7.0677655.  Cell volumes OK.
Mesh non-orthogonality Max: 65.691038 average: 5.0074554
Non-orthogonality check OK.
Face pyramids OK.                                                                                                                                                         Max skewness = 0.9386004 OK.
Coupled point location match (average 0) OK.
Mesh OK.  

[dlahaye]

My understanding is that checkMesh does not include checks on faceAgglomeration to work properly. See e.g. https://openfoamwiki.net/index.php/CheckMesh on what checkMesh does do.

What do mean by "impossible to make it"?

Do we understand what non-manifold faces are?

Best.

[giorgianig]

Originally Posted by dlahaye

My understanding is that checkMesh does not include checks on faceAgglomeration to work properly. See e.g. https://openfoamwiki.net/index.php/CheckMesh on what checkMesh does do.

What do mean by "impossible to make it"?

Do we understand what non-manifold faces are?

Best.

Yes, indeed, checkMesh doesn't do that. Still, the face mesh didn't seem visually bad, so I don't know.
Impossible to make a dense matrix n x n with n=659510.
I searched around trying to understand "non-manifold" faces. Not sure what it means except that they are not avoidable with complex surface. Maybe someone else can clarify.

Anyway. I solved the problem finally. I had to refine drastically the .stl defining the surfaces implied in the faceAggregation. Like now I have a .stl of 2 Gb for each surface.