[mbeaudoin]

Hello,

I am looking for some experienced opinions about OpenFOAM and the presence of concave faces or concave poly cells in OF meshes.

What are the major problems or major side effects related to the presence of concave cells/faces/polygons in OpenFOAM meshes, if any?

Are there any known solver numerical issues related to the presence of concave cells or faces in OF meshes?

snappyHexMesh seems to allow face concavity angles of up to 80 degrees by default. This seems like quite a large value.
In general, how will lowering the value of this parameter called maxConcave in snappyHexMeshDict affect the overall mesh quality?
Any rule of thumb?

With snappyHexMesh, is it possible to generate good quality meshes without the presence of any concave cells in the final output (using maxConcave < 0)?

Finally, checkMesh will only report the presence concave faces in a mesh if the face concave angle is greater than 10 degree.
This 10 degree value is hard-coded in the file checkMesh/checkGeometry.C
Wouldn't it be useful to have this angle value specified through a controlDict switch instead, so that checkMesh could report the presence ALL concave faces?

Thank you in advance for helping me get a better understanding of these issues.

Martin

[eugene]

Cell concavity is not really relevant in a face based mesh, in any case, this concavity parameter refers to face concavity and not cell concavity.

Face concavity does not mean the cell is concave. This is the edge angle between adjacent edges in a face, not between adjacent faces. This property does not influence the discretisation directly. The most important factors for this are:
-non-orthogonality
-volume Ratio
-delta ratio
-skewness

Although extreme instances of all the other criteria can cause problems too (otherwise we wouldn't need them), especially when face areas tend to zero.

Lowering the maxConcave value will "improve" the cell quality, but the mesh might not conform as well to the surface as what might otherwise be the case.

[holodeck10]

as an addition:

checkMesh -allGeometry

would write all concave cells into a set.

[mkhm]

Is it possible that the simulations run fine and converge with concave cells ?

I have this issue. I did my simulations and I did not even realize that there was a problem related to concave cells. Because normally, I do checkMesh and not checkMesh -allGeometry. However, when I wanted to mapFields the results to a finer mesh, it fails due to the concave cells. However, the refined mesh was constructed without problem from the "coarse" mesh having concave cells. In another case, the construction of a refined mesh from the mesh with concave cells is impossible. Is there any mean to circumvent this issue ?



Regards,

Mary

[mrishi]

Hi,
Did you understand or resolve the situation? I am facing similar concavity fails on the merging of fine and coarse zones of hex mesh.

This thread seems to suggest that it is not such a big deal:
how should we remove the Concave cells from the mesh?


Are there some new insights on this?

Originally Posted by mkhm

Is it possible that the simulations run fine and converge with concave cells ?

I have this issue. I did my simulations and I did not even realize that there was a problem related to concave cells. Because normally, I do checkMesh and not checkMesh -allGeometry. However, when I wanted to mapFields the results to a finer mesh, it fails due to the concave cells. However, the refined mesh was constructed without problem from the "coarse" mesh having concave cells. In another case, the construction of a refined mesh from the mesh with concave cells is impossible. Is there any mean to circumvent this issue ?



Regards,

Mary