[gdbaldw]

To All,

I'm modeling external aerodynamic flow about a body. I understand that for accurate drag assessment the mesh should extend from the body by 20 times the reference length, y+ should be less than 5, and the cell height should grow by a factor of about 1.15.

My question is that the above constraints inevitably lead to smallEdges and highAspectRatioCells, yes? And is this ok?

For instance, for a given geometry of air I can generate a mesh with no errors by increasing y+ and increasing the cell height growth factor. But, for a more refined mesh with the above constraints, checkMesh issues warnings and errors.

checkMesh outputs are...

For very coarse mesh:

Mesh stats
points: 58116
edges: 159043
faces: 143936
internal faces: 114112
cells: 43008
boundary patches: 1
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 43008
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 0
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
all 29824 29826 ok (closed singly connected surface)

Checking geometry...
Domain bounding box: (-150.156 -0.743357 -174.114) (149.844 150 125.886)
Boundary openness (1.62991e-16 -3.78198e-16 -2.14744e-16) OK.
Max cell openness = 3.20853e-15 OK.
Max aspect ratio = 129.482 OK.
Minumum face area = 1.91668e-05. Maximum face area = 1135.09. Face area magnitudes OK.
Min volume = 3.67946e-07. Max volume = 3217.21. Total volume = 6.45353e+06. Cell volumes OK.
Mesh non-orthogonality Max: 80.4968 average: 27.0736
*Number of severely non-orthogonal faces: 599.
Non-orthogonality check OK.
<<Writing 599 non-orthogonal faces to set nonOrthoFaces
Face pyramids OK.
Max skewness = 3.82417 OK.
Min/max edge length = 0.00129032 172.098 OK.
All angles in faces OK.
Face flatness (1 = flat, 0 = butterfly) : average = 0.999757 min = 0.943597
All face flatness OK.

Mesh OK.

End

But, for the more refined mesh...

Mesh stats
points: 1452900
edges: 4324771
faces: 4291136
internal faces: 4224448
cells: 1419264
boundary patches: 1
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 1419264
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 0
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
all 66688 66690 ok (closed singly connected surface)

Checking geometry...
Domain bounding box: (-150.156 -0.74336 -174.114) (149.844 150 125.886)
Boundary openness (5.17672e-17 7.27314e-16 -1.80348e-16) OK.
***High aspect ratio cells found, Max aspect ratio: 3723.68, number of cells 519
<<Writing 519 cells with high aspect ratio to set highAspectRatioCells
Minumum face area = 5.83592e-09. Maximum face area = 275.784. Face area magnitudes OK.
Min volume = 6.86967e-11. Max volume = 1096.84. Total volume = 6.45353e+06. Cell volumes OK.
Mesh non-orthogonality Max: 89.4941 average: 26.6997
*Number of severely non-orthogonal faces: 27612.
Non-orthogonality check OK.
<<Writing 27612 non-orthogonal faces to set nonOrthoFaces
Face pyramids OK.
Max skewness = 2.99001 OK.
*Edges too small, min/max edge length = 4e-06 22.4771, number too small: 307632
<<Writing 322167 points on short edges to set shortEdges
All angles in faces OK.
Face flatness (1 = flat, 0 = butterfly) : average = 0.999975 min = 0.98297
All face flatness OK.

Failed 1 mesh checks.

End

Any and all advice welcomed.

Thank you,
Doug

[vega]

Doug,

Typically, in automotive industry people use prism cells on the body and then Hex in the rest of flow domain to avoid quality issues. For my external aero simulations, with Tgrid/Fluent surface mesh is pure
trias followed by several layers of prism layers and then tetra or hex in rest of the flow domain. Quality is pretty much independent of the mesh size. Surface mesh quality is of bigger concern.

what kind ( low/high speed, streamlined/bluff body) simulation are you doing?
--
Rajneesh

[gdbaldw]

Rajneesh,

Thank you for offering your experience. I'm simulating Re=1e7 streamlined flow. My mesh is pure quad on the surface wall, with a uniform column of hexs from each surface quad all the way out to the boundary patch. The boundary patch has exactly the same number of quads as the surface wall.

checkMesh is perfect if I go out to say 3x reference length. But at up to 20x reference length the ratio between short edges and long edges produces short edge errors, and aspect ratio errors are also generated. The potentialFoam solver works anyway, and the results appear correct.

I'm interested in knowing if I have uniformly short edges near the body and uniformly long edges far away from the body will this degrade the quality of the results.

My basic question is if I know the mesh quality is perfect in checkMesh for a mesh of say 3x reference length, and I simply add every larger cells beyond 3x which eventually causes short edge and aspect ratio errors in checkMesh, should I be concerned? I could probably chase the cause of error and eventually have a perfect checkMesh out to 20x, but I'd like to first know if this is worth the time and effort.

Doug

[pbo]

Doug,

I've run simpleFoam on some wing cases, using the SA model with a y+ of about 1 at the wall-adjacent cells and a stretching factor of 1.1 (hex in the boundary layer, tet in the far field). Checkmesh did not complain about too high aspect ratio though. here's the output of checkmesh for one of my case:
/*---------------------------------------------------------------------------*\
| ========= | |
| \ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \ / O peration | Version: 1.4.1 |
| \ / A nd | Web: http://www.openfoam.org |
| \/ M anipulation | |
\*---------------------------------------------------------------------------*/

Exec : checkMesh . stfwSym_RF00_RA00
Date : Oct 16 2007
Time : 18:03:58
Host : kittyhawk
PID : 27593
Root : /users/aexpb/OpenFOAM-1.4.1/aexpb-1.4.1/run/simpleFoam
Case : stfwSym_RF00_RA00
Nprocs : 1
Create time

Create polyMesh for time = constant

Time = constant

Mesh stats
points: 1114235
edges: 4231194
faces: 5316839
internal faces: 5268373
cells: 2199879
boundary patches: 7
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 853440
prisms: 30440
wedges: 0
pyramids: 48376
tet wedges: 0
tetrahedra: 1267623
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
symmetry-4 15074 11062 ok (not multiply connected)
outflow-5 1108 602 ok (not multiply connected)
velocity-inlet-6 1126 611 ok (not multiply connected)
slip-wall-7 1238 658 ok (not multiply connected)
aftWinglet-8 6330 5616 ok (not multiply connected)
frontWinglet-9 6318 5610 ok (not multiply connected)
wing-10 17272 17496 ok (not multiply connected)

Checking geometry...
Domain bounding box: (-20 -5.721684313e-13 -20) (20 20 20)
Boundary openness (-3.50775689274e-17 1.03061502414e-15 4.01765248617e-17) OK.
Max cell openness = 1.99760162922e-14 OK.
Max aspect ratio = 454.261326444 OK.
Minumum face area = 2.01963520532e-10. Maximum face area = 3.65068621543. Face area magnitudes OK.
Min volume = 1.85425452264e-15. Max volume = 2.39605975495. Total volume = 31999.9963006. Cell volumes OK.
Mesh non-orthogonality Max: 89.2519784873 average: 27.8654976223
*Number of severely non-orthogonal faces: 100029.
Non-orthogonality check OK.
<<Writing 100029 non-orthogonal faces to set nonOrthoFaces
Face pyramids OK.
Max skewness = 2.91618646831 OK.
*Edges too small, min/max edge length = 1.09452909612e-05 3.42547059472, number too small: 270345
<<Writing 285910 points on short edges to set shortEdges
All angles in faces OK.
Face flatness (1 = flat, 0 = butterfly) : average = 0.999985939551 min = 0.977107638673
All face flatness OK.

Mesh OK.

End

Despite the warning about the small edges, the run went fine.

Patrick

[gdbaldw]

Patrick,

Great! I now know to focus only on the Aspect Ratio, which I can readily correct.

Question, is there a benefit to you using tet in the far field rather than hex? For my meshing method, pure hex is easier. I could use tet in the far field but that would require two meshes then meshMerge and meshStitch. Thanks for sharing your experience.

Doug

[pbo]

Doug,

My using tet in the far field was more a matter of convenience (my geometry was quite complex and didn't allow me to extrude the structured surface grid up to the far-field boundary of the computational domain). If you can stick to a pure hex grid, then stick to it (hex are better than tet as far as accuracy is concerned).

Patrick

[solique]

Hi Patrick,

I am not an expert but I just want to post a question: the max Aspect ratio of 454 is OK?

I was told that the maximum Aspect ratio should be kept within 150, although I totally know not why this number is like this.

Should you give me any suggestions?

Thanks!

Vic

Originally Posted by pbo

Doug,

Max aspect ratio = 454.261326444 OK.

Patrick

[saucy178]

Mine is even worse! CheckMesh gives:
***High aspect ratio cells found, Max aspect ratio: 8.41538e+193, number of cells 8460
<<Writing 8460 cells with high aspect ratio to set highAspectRatioCells
Minumum face area = 5.24051e-07. Maximum face area = 1.11681e-06. Face area magnitudes OK.
Min volume = 2e-300. Max volume = 2e-300. Total volume = 1.692e-296. Cell volumes OK.
Mesh non-orthogonality Max: 180 average: 169.488
***Number of non-orthogonality errors: 22956.
<<Writing 22956 non-orthogonal faces to set nonOrthoFaces
***Error in face pyramids: 50760 faces are incorrectly oriented.
<<Writing 27804 faces with incorrect orientation to set wrongOrientedFaces
Max skewness = 0.704293 OK.


Dont know why Im getting this. I have very benign mesh with all cells near to 1 cubic mm. Can it be because some of my edges in blockMesh are arc. In fact many of them.

[rikio]

It may caused by the too slim cells, check you mesh and modify the node distribution at the problematic regions. Besides, edges are following arcs as you said, orientation problem is very prone to be there.

[saucy178]

Thanks Rikio I will try it

[KateEisenhower]

Thank you Doug for this interesting post,

I am experienceing similar problems. To be honest I am not sure if it's a problem at all, because my case is solving nicely. Additionally, I get underdeterminedCells. If this case is still on your HD, could you run checkMesh -allTopology -allGeometry on your case to see if you get this error too?

Best regards,

Kate

Originally Posted by gdbaldw

To All,

I'm modeling external aerodynamic flow about a body. I understand that for accurate drag assessment the mesh should extend from the body by 20 times the reference length, y+ should be less than 5, and the cell height should grow by a factor of about 1.15.

My question is that the above constraints inevitably lead to smallEdges and highAspectRatioCells, yes? And is this ok?

For instance, for a given geometry of air I can generate a mesh with no errors by increasing y+ and increasing the cell height growth factor. But, for a more refined mesh with the above constraints, checkMesh issues warnings and errors.

checkMesh outputs are...

For very coarse mesh:

Mesh stats
points: 58116
edges: 159043
faces: 143936
internal faces: 114112
cells: 43008
boundary patches: 1
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 43008
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 0
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
all 29824 29826 ok (closed singly connected surface)

Checking geometry...
Domain bounding box: (-150.156 -0.743357 -174.114) (149.844 150 125.886)
Boundary openness (1.62991e-16 -3.78198e-16 -2.14744e-16) OK.
Max cell openness = 3.20853e-15 OK.
Max aspect ratio = 129.482 OK.
Minumum face area = 1.91668e-05. Maximum face area = 1135.09. Face area magnitudes OK.
Min volume = 3.67946e-07. Max volume = 3217.21. Total volume = 6.45353e+06. Cell volumes OK.
Mesh non-orthogonality Max: 80.4968 average: 27.0736
*Number of severely non-orthogonal faces: 599.
Non-orthogonality check OK.
<<Writing 599 non-orthogonal faces to set nonOrthoFaces
Face pyramids OK.
Max skewness = 3.82417 OK.
Min/max edge length = 0.00129032 172.098 OK.
All angles in faces OK.
Face flatness (1 = flat, 0 = butterfly) : average = 0.999757 min = 0.943597
All face flatness OK.

Mesh OK.

End

But, for the more refined mesh...

Mesh stats
points: 1452900
edges: 4324771
faces: 4291136
internal faces: 4224448
cells: 1419264
boundary patches: 1
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 1419264
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 0
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
all 66688 66690 ok (closed singly connected surface)

Checking geometry...
Domain bounding box: (-150.156 -0.74336 -174.114) (149.844 150 125.886)
Boundary openness (5.17672e-17 7.27314e-16 -1.80348e-16) OK.
***High aspect ratio cells found, Max aspect ratio: 3723.68, number of cells 519
<<Writing 519 cells with high aspect ratio to set highAspectRatioCells
Minumum face area = 5.83592e-09. Maximum face area = 275.784. Face area magnitudes OK.
Min volume = 6.86967e-11. Max volume = 1096.84. Total volume = 6.45353e+06. Cell volumes OK.
Mesh non-orthogonality Max: 89.4941 average: 26.6997
*Number of severely non-orthogonal faces: 27612.
Non-orthogonality check OK.
<<Writing 27612 non-orthogonal faces to set nonOrthoFaces
Face pyramids OK.
Max skewness = 2.99001 OK.
*Edges too small, min/max edge length = 4e-06 22.4771, number too small: 307632
<<Writing 322167 points on short edges to set shortEdges
All angles in faces OK.
Face flatness (1 = flat, 0 = butterfly) : average = 0.999975 min = 0.98297
All face flatness OK.

Failed 1 mesh checks.

End

Any and all advice welcomed.

Thank you,
Doug

[KateEisenhower]

Hi Patrick,

can you remember what your minimal cell height was? Mine is circa 6e-6 m and checkMesh complains about underdeterminedCells.

Best regards,

Kate

Originally Posted by pbo

Doug,

I've run simpleFoam on some wing cases, using the SA model with a y+ of about 1 at the wall-adjacent cells and a stretching factor of 1.1 (hex in the boundary layer, tet in the far field). Checkmesh did not complain about too high aspect ratio though. here's the output of checkmesh for one of my case:
/*---------------------------------------------------------------------------*\
| ========= | |
| \ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \ / O peration | Version: 1.4.1 |
| \ / A nd | Web: http://www.openfoam.org |
| \/ M anipulation | |
\*---------------------------------------------------------------------------*/

Exec : checkMesh . stfwSym_RF00_RA00
Date : Oct 16 2007
Time : 18:03:58
Host : kittyhawk
PID : 27593
Root : /users/aexpb/OpenFOAM-1.4.1/aexpb-1.4.1/run/simpleFoam
Case : stfwSym_RF00_RA00
Nprocs : 1
Create time

Create polyMesh for time = constant

Time = constant

Mesh stats
points: 1114235
edges: 4231194
faces: 5316839
internal faces: 5268373
cells: 2199879
boundary patches: 7
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 853440
prisms: 30440
wedges: 0
pyramids: 48376
tet wedges: 0
tetrahedra: 1267623
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
symmetry-4 15074 11062 ok (not multiply connected)
outflow-5 1108 602 ok (not multiply connected)
velocity-inlet-6 1126 611 ok (not multiply connected)
slip-wall-7 1238 658 ok (not multiply connected)
aftWinglet-8 6330 5616 ok (not multiply connected)
frontWinglet-9 6318 5610 ok (not multiply connected)
wing-10 17272 17496 ok (not multiply connected)

Checking geometry...
Domain bounding box: (-20 -5.721684313e-13 -20) (20 20 20)
Boundary openness (-3.50775689274e-17 1.03061502414e-15 4.01765248617e-17) OK.
Max cell openness = 1.99760162922e-14 OK.
Max aspect ratio = 454.261326444 OK.
Minumum face area = 2.01963520532e-10. Maximum face area = 3.65068621543. Face area magnitudes OK.
Min volume = 1.85425452264e-15. Max volume = 2.39605975495. Total volume = 31999.9963006. Cell volumes OK.
Mesh non-orthogonality Max: 89.2519784873 average: 27.8654976223
*Number of severely non-orthogonal faces: 100029.
Non-orthogonality check OK.
<<Writing 100029 non-orthogonal faces to set nonOrthoFaces
Face pyramids OK.
Max skewness = 2.91618646831 OK.
*Edges too small, min/max edge length = 1.09452909612e-05 3.42547059472, number too small: 270345
<<Writing 285910 points on short edges to set shortEdges
All angles in faces OK.
Face flatness (1 = flat, 0 = butterfly) : average = 0.999985939551 min = 0.977107638673
All face flatness OK.

Mesh OK.

End

Despite the warning about the small edges, the run went fine.

Patrick