[snappyHexMesh] non-orthogonal faces in layers

Eloise · September 25, 2012, 10:31

Hi Foamers,

I have a hull geometry on which I try to insert layers. In order to reach a reasonable yPlus, I need to insert about 20 layers. When I do so, I nearly don't have any layer cells added on my geometry.
In order to understand why the cells are not inserted, I relaxed the mesh quality in sHM. I obtained a mesh in which the layers are inserted and many non-orthogonal faces are present. They are probably the reason why the layers are not inserted when I keep un-relaxed mesh quality requirement.

I do not understand how to get rid of those non-orthogonal faces. They are located inside the layer, between polyhedral cells which are perfectly piled up, and I don't see any reason why those faces are detected as non-orthogonal.

In attached image, the layer's cells are in grey and the non-orthogonal faces in red.

Help needed please!
Regards,
Eloïse

Eloise · October 12, 2012, 10:08

Hi,

I'm still stuck with this issue of non-orthogonal faces. Here is a typical output of checkMesh of the snapped mesh (Time2) and for the layer mesh (Time 3):

Code:

Time = 2

Mesh stats
    points:           2354368
    faces:            6789693
    internal faces:   6645664
    cells:            2219640
    boundary patches: 6
    point zones:      0
    face zones:       0
    cell zones:       0

Overall number of cells of each type:
    hexahedra:     2170238
    prisms:        1958
    wedges:        0
    pyramids:      0
    tet wedges:    75
    tetrahedra:    0
    polyhedra:     47369

Checking topology...
    Boundary definition OK.
    Cell to face addressing OK.
    Point usage OK.
    Upper triangular ordering OK.
    Face vertices OK.
    Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
    Patch               Faces    Points   Surface topology                  
    frontAndBack        15060    15624    ok (non-closed singly connected)  
    outlet              4650     4836     ok (non-closed singly connected)  
    inlet               4650     4836     ok (non-closed singly connected)  
    upperWall           38905    39312    ok (non-closed singly connected)  
    lowerWall           40342    41260    ok (non-closed singly connected)  
    HULL        40422    43904    ok (non-closed singly connected)  

Checking geometry...
    Overall domain bounding box (-900 -500 -300) (600 500 0)
    Mesh (non-empty, non-wedge) directions (1 1 1)
    Mesh (non-empty) directions (1 1 1)
    Boundary openness (1.07829e-15 4.53941e-16 4.62619e-14) OK.
    Max cell openness = 3.80173e-16 OK.
    Max aspect ratio = 23.3478 OK.
    Minumum face area = 0.0110912. Maximum face area = 997.836.  Face area magnitudes OK.
    Min volume = 0.00975568. Max volume = 30402.1.  Total volume = 4.49867e+08.  Cell volumes OK.
    Mesh non-orthogonality Max: 60.4405 average: 4.73595
    Non-orthogonality check OK.
    Face pyramids OK.
    Max skewness = 3.25394 OK.

Mesh OK.

Time = 3

Mesh stats
    points:           2773606
    faces:            7985817
    internal faces:   7842477
    cells:            2608303
    boundary patches: 6
    point zones:      0
    face zones:       0
    cell zones:       0

Overall number of cells of each type:
    hexahedra:     2508558
    prisms:        2220
    wedges:        0
    pyramids:      0
    tet wedges:    113
    tetrahedra:    1
    polyhedra:     97411

Checking topology...
    Boundary definition OK.
    Cell to face addressing OK.
    Point usage OK.
    Upper triangular ordering OK.
    Face vertices OK.
    Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
    Patch               Faces    Points   Surface topology                  
    frontAndBack        15060    15624    ok (non-closed singly connected)  
    outlet              4650     4836     ok (non-closed singly connected)  
    inlet               4650     4836     ok (non-closed singly connected)  
    upperWall           38905    39312    ok (non-closed singly connected)  
    lowerWall           40342    41260    ok (non-closed singly connected)  
    HULL        39733    43297    ok (non-closed singly connected)  

Checking geometry...
    Overall domain bounding box (-900 -500 -300) (600 500 0)
    Mesh (non-empty, non-wedge) directions (1 1 1)
    Mesh (non-empty) directions (1 1 1)
    Boundary openness (1.70792e-15 7.25352e-16 4.68483e-14) OK.
    Max cell openness = 1.16192e-14 OK.
    Max aspect ratio = 506.676 OK.
    Minumum face area = 3.9423e-05. Maximum face area = 997.836.  Face area magnitudes OK.
    Min volume = 6.24616e-05. Max volume = 30402.1.  Total volume = 4.49867e+08.  Cell volumes OK.
    Mesh non-orthogonality Max: 88.4737 average: 5.90217
   *Number of severely non-orthogonal faces: 3562.
    Non-orthogonality check OK.
  <<Writing 3562 non-orthogonal faces to set nonOrthoFaces
    Face pyramids OK.
    Max skewness = 2.99817 OK.

Mesh OK.

As you can see, I have a good quality mesh ("Mesh OK") at time 2 and 3, with however a warning on the non-orthogonality at time 3. Those non-orthogonal faces are located on flat faces, rounded faces (large radius) and along an edge which is covered by the layer (edge angle ~30deg). They lead to a crash during the computation, so I really need to avoid them.
I add layers while relaxing the mesh quality in the following way:

Code:

    relaxed
    {
    maxBoundarySkewness 1000; 
    maxInternalSkewness 40;
    minTetQuality -1e30; 
    minDeterminant -1;
    minTwist -1e30;
    maxNonOrtho 180;
    }

Of course, the relaxed maxNonOrtho allows those non-orthogonal faces to appear. If I don't relax this parameter, a too large part of the patches is not covered by layers. Does anyone has an idea about a way to improve the mesh quality in the layers so that the layers cover the whole patch without all those non-orthogonal errors?

Regards,
Eloïse

Tobi · October 25, 2012, 05:35

Hi,

there is a good thread here, but I do not find it now.
You can switch off all mesh quality parameters to create the layers on the whole patch.

What sHM are you using? I have no relaxation{} in my sHMD

Eloise · October 25, 2012, 05:59

Hello Tobi,

If you find the link to that thread, I'd really be interested! I've been browsing quite a lot around but haven't read about that topic.

I am using OF2.0.1. The relaxation appears under the meshQualityControls in the following way:

Code:

meshQualityControls
{
    maxNonOrtho 65;
    maxBoundarySkewness 20;
    maxInternalSkewness 4;
    maxConcave 80;
    minVol 1e-13;
    minTetQuality 1e-30;
    minArea -1;
    minTwist 0.02;
    minDeterminant 0.001;
    minFaceWeight 0.02;
    minVolRatio 0.01;
    minTriangleTwist -1;

    // Advanced
    nSmoothScale 4;
    errorReduction 0.75;

    relaxed
    {
    maxBoundarySkewness 1000; 
    maxInternalSkewness 40;
    minTetQuality -1e30; 
    minDeterminant -1;
    minTwist -1e30;
    }
}

If I relax the maxNonOrtho, I indeed cover the whole surface with layers, but I get non-orthogonal faces in the layers which lead to a crash when solving.
If I don't relax, I get holes in my layers.

Regards,
Eloïse

Tobi · October 25, 2012, 06:36

Quote:

Originally Posted by Eloise

Hello Tobi,

If you find the link to that thread, I'd really be interested! I've been browsing quite a lot around but haven't read about that topic.

I am using OF2.0.1. The relaxation appears under the meshQualityControls in the following way:

Code:

meshQualityControls
{
    maxNonOrtho 65;
    maxBoundarySkewness 20;
    maxInternalSkewness 4;
    maxConcave 80;
    minVol 1e-13;
    minTetQuality 1e-30;
    minArea -1;
    minTwist 0.02;
    minDeterminant 0.001;
    minFaceWeight 0.02;
    minVolRatio 0.01;
    minTriangleTwist -1;

    // Advanced
    nSmoothScale 4;
    errorReduction 0.75;

    relaxed
    {
    maxBoundarySkewness 1000; 
    maxInternalSkewness 40;
    minTetQuality -1e30; 
    minDeterminant -1;
    minTwist -1e30;
    }
}

If I relax the maxNonOrtho, I indeed cover the whole surface with layers, but I get non-orthogonal faces in the layers which lead to a crash when solving.
If I don't relax, I get holes in my layers.

Regards,
Eloïse

Hi,

thats nice to know. I am using 2.1.x and in my sHMDict I have no entry of the relaxation factors ...

If I get it I ll send it to you

Tobi · October 25, 2012, 16:38

Quote:

Originally Posted by Tobi

Hi,

thats nice to know. I am using 2.1.x and in my sHMDict I have no entry of the relaxation factors ...

If I get it I ll send it to you

Hi Eloïse,

do you get a output if you are setting layers that you use relaxed things?
I insert the relax entry but its not working

My output

Code:

Shrinking and layer addition phase
----------------------------------

Using mesh parameters 
{
    maxNonOrtho     65;
    maxBoundarySkewness 20;
    maxInternalSkewness 4;
    maxConcave      60;
    minVol          1e-13;
    minTetQuality   1e-30;
    minArea         1e-30;
    minTwist        0.03;
    minDeterminant  0.001;
    minFaceWeight   0.03;
    minVolRatio     0.01;
    minTriangleTwist -1;
    nSmoothScale    4;
    errorReduction  0.75;
}

Hmmm :/

SOLVED

I insert the lines into the wrong section

Eloise · October 29, 2012, 04:01

Hi Tobi,

Great to read that it is still an option in OF 2.1

Have a nice week,
Eloïse

October 25, 2012, 05:59		#4
Eloise Senior Member Eloïse Join Date: Jul 2012 Location: Trondheim, Norway Posts: 113 Rep Power: 14	Hello Tobi, If you find the link to that thread, I'd really be interested! I've been browsing quite a lot around but haven't read about that topic. I am using OF2.0.1. The relaxation appears under the meshQualityControls in the following way: Code: meshQualityControls { maxNonOrtho 65; maxBoundarySkewness 20; maxInternalSkewness 4; maxConcave 80; minVol 1e-13; minTetQuality 1e-30; minArea -1; minTwist 0.02; minDeterminant 0.001; minFaceWeight 0.02; minVolRatio 0.01; minTriangleTwist -1; // Advanced nSmoothScale 4; errorReduction 0.75; relaxed { maxBoundarySkewness 1000; maxInternalSkewness 40; minTetQuality -1e30; minDeterminant -1; minTwist -1e30; } } If I relax the maxNonOrtho, I indeed cover the whole surface with layers, but I get non-orthogonal faces in the layers which lead to a crash when solving. If I don't relax, I get holes in my layers. Regards, Eloïse

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Foam::error::PrintStack	almir	OpenFOAM Running, Solving & CFD	92	May 21, 2024 08:56
decomposePar problem: Cell 0contains face labels out of range	vaina74	OpenFOAM Pre-Processing	37	July 20, 2020 06:38
[snappyHexMesh] SHM is not extruding/adding Layers everywhere	matthiasd	OpenFOAM Meshing & Mesh Conversion	2	October 16, 2016 17:45
DecomposePar unequal number of shared faces	maka	OpenFOAM Pre-Processing	6	August 12, 2010 10:01
Unaligned accesses on IA64	andre	OpenFOAM	5	June 23, 2008 11:37

October 25, 2012, 05:35		#3
Tobi Super Moderator Tobias Holzmann Join Date: Oct 2010 Location: Bad Wörishofen Posts: 2,711 Blog Entries: 6 Rep Power: 52	Hi, there is a good thread here, but I do not find it now. You can switch off all mesh quality parameters to create the layers on the whole patch. What sHM are you using? I have no relaxation{} in my sHMD

October 29, 2012, 04:01		#7
Eloise Senior Member Eloïse Join Date: Jul 2012 Location: Trondheim, Norway Posts: 113 Rep Power: 14	Hi Tobi, Great to read that it is still an option in OF 2.1 Have a nice week, Eloïse