CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS (
-   -   ICEM: Thin Wall problem (

just-right June 15, 2011 13:39

ICEM: Thin Wall problem
I am meshing a wing like shape in close proximity to the ground. The geometry is 2 mm thick with an endplate on the side.

I am having problems with patch dependent meshing in relation to the points and the edges. Some nodes do not obay the surface, edges or nodes and actually penetrate the opposite surface.

I am currently using the following approach:

- Import geometry
- Name surfaces
- Keep curves and points
- Set global mesh sizes and enable curvature/proximity
- Change edge criterion

The mesh is full of errors because of these overlapping surface nodes. Could somebody recommend how to approach this problem with the exception of just creating 2 mm element all over the geometry?

just-right June 16, 2011 12:33

I have seperated the large surfaces and the small surfaces. There seems great sensitivity to the edge criterion. However, even though the curves that define the edges as well as the points are still there no consistency in the meshing process is present.

I have enclosed two pictures that should represent some of the problems I am having

The surface on the right is the symmetry plane it can be seen that it interchanges with the other surfaces.

The surface mesh is so deformed that it penetrates the other surface.

just-right June 17, 2011 07:29

I have been using the "define thin cuts" approach with improvements. It resolves the small areas a lot better, however, I still encounter various difficulties. Primarily the area where the small faces and the larger faces of the diffuser meet the edge is maintained. Near the edge where the horizontal and vertical face of the diffuser meet it is not maintaind:

This problem occurs in the same region every time and it leads me to believe it is a function of the algorithm.

The current settings also don't obay the red lines between the symmetry plane and the actual geometry. Sometimes triangular faces of the symmetry plane collaps to the geometry in question and vice versa. It also regularly create piramid shapes by not connecting nodes to the curve:

I would imagine there is a solution to this problem. I have tried edge criterion but in combination with the thin cuts it results in a lot of holes in the mesh.

swiss_zhang August 4, 2011 07:46

first thing to check is the quality of your geometry.

Does any line exsist between 2 adjacet surfaces?. If not, the problem you shown will occur.

just-right August 6, 2011 20:35

Yeah lines exist between the adjacent surfaces.

A good way to approach the problem is to use thin cuts and the ignore wall thickness option. Also I found slightly decreasing the edge criteria to further help with node allignment. The combination of all of this resulted in a good quality mesh for CFD analysis.

All times are GMT -4. The time now is 02:03.