[AleXX87]

Dear CFD users,


I am trying to simulate the turbulent 3D flow over a thin periodic wing element, with a 3d damage on the leading edge.

The mesh is unstructured and automatically generated by ansys meshing (for me this is a mandatory aspect). This makes the grid to have 1 cell thickness in the far field.

I have to impose a periodic BC spanwise, but this seems to have a very bad effect on the convergence for the mesh zone with one layer elements spanwise (far field). Sometimes I cannot even initialize the run.


Does anyone knows if there is any issue or particular setting that can help to solve this problem?


(I wish to not come back to the mesh and try to make two layers at least everywhere)


Thanks

[vinerm]

By one-cell thickness do you mean span length? Why do you want to use one-cell thickness?

[AleXX87]

Yes, but it is not really a choice, it is more a consequence of using the automatic meshing tool. Because of the geometry of the damage I want to simulate, I must use an unstructured tetrahedral mesh. Being the far field cells very large with respect to the spanwise thickness of the domain, the mesh tool reduces this region to a one layer cell mesh.

[vinerm]

Your description is not very clear to me. Could you share a picture showing what you mean by one-cell? There is nothing called automatic meshing; there are default parameter values and there are user provided values. Tet mesh is not bad but one-cell thickness anywhere is mostly useless, except under certain specific conditions.

[AleXX87]

You can find an example attached for what I intend as one cell span. That is a view of the airfoil lower boundary. The red arrow shows the spanwise direction.

With the proper settings I can obtain a mesh with more than 1 cell spanwise in the area of the figure, but not in the far field.



I hope this helps to understand the configuration I am trying to simulate.

[vinerm]

If the shape of the damage on the airfoil is 3D, then you should generate proper 3D mesh. With one cell thickness, the simulation would essentially be 2D. I understand that meshing tool may not be doing a good job of put more cells in the far-field region but that user needs to control by using sizes or some other option. This mesh will not give you good results. I suppose someone in Meshing forum can help you with this.

If the damage is not really 3D, then use a 2D domain.

[AleXX87]

Thank you for your reply.

Yes, the damage is 3D but is concentrated in a small region, anyway I need to solve a 3D periodic problem.
I just write here what I understood and close the topic:

Giving a mesh with one cell span in the far field region (as in the figure before) and periodic condition does not produce good results in fluent, no solver configuration can help this.

The only possible solution is to try to put at least two layers of cells between the periodic faces everywhere in the mesh (which is possible but requires more work from the definition of the geometry of the domain).

[vinerm]

Do note that periodicity requires at least four cells across the thickness. With one-cell, periodicity does not mean anything, if periodicity is meant across the span length.