[large_eddy]

Hello folks,

for I have only little experience in CFD I was hoping if someone has good advise how to overcome large jumps in the aspect-ratio at the transition zone from the surface boundary layer to the farfield. I attached a picture where you can see the problem, especially at the sharp bend. I was using Gmsh-->enGrid 1.4 to create the mesh.

The spatial extend of my domain is very high, that's why I chose a pretty rough mesh resolution, to speed up the meshing process. Once all obstacles are overcome I will use higher resolutions for the final mesh.

So how will I create smooth transition zones with little jumps of the aspect-ratio?



aspect_problem.jpg

[ogloth]

Hi,

the mesh that you show is probably not too bad, but I think I understand your concern and it definitely looks a bit ugly.

Did you try to use enGrid right from the start, also to create the surface mesh? It looks like enGrid was not able to delete small edges in the surface mesh directly adjacent to your boundary layer. If you use enGrid as surface mesher, it will have more information about the mesh density there and it might be able to give a somewhat better transition between boundary layer and far-field.

Cheers,
Oliver

[large_eddy]

First of all thank you for the help!

I tried to use enGrid for the creation of the surface mesh but it does not work.
When I define the rules for the surface mesh generation and afterwards use the create/improve surface mesh function enGrid crashes (window shuts down immediately).

Is this a common bug, or does that indicate a wrong setting of parameters?

[ogloth]

Hmmh -- it is a bug in the sense that it should report an error instead of crashing.

Most likely, however, you have an inconsistent initial triangulation (i.e. not water-tight). Engrid needs a water-tight triangulated geometry as starting point (e.g. from an STL file or exported from Blender). You can check the mesh consistency (simply press the key C).

Cheers,
Oliver

[large_eddy]

When I check my surface mesh (consiting of 4 patches) for integrity enGrid says that the mesh is OK.
So I really don't understand what the problem is in this case.

When you take a look at my mesh at the beginning of this thread would everybody agree that this is a real simple case? It should represent a plane crossed by a steep valley. Doesn't sound to complex right?

I'm now thinking of using Blender to create STL files.
But it leaves me disappointed that it seems that even though it's simple case, Gmsh-->enGrid-->openFoam won’t do the job.

How would you deal with this simple case?

[ogloth]

Hi,

If you can post the mesh which creates the crash, I would have a look at it to see what exactly is happening.

One more thing: When you divide the boundary later, you can specify two critical angles. This "pulls in" nodes which sit on sharp feature edges (e.g. the trailing edge of an airfoil). If you do not intend to export the grid in polyhedral format, this should be avoided. In this case simply set both angles to 360 degrees which should avoid the process altogether. In the example that you posted this behaviour can be observed.

Cheers,
Oliver

[large_eddy]

Hi to all,

some updates on my problem: First of all I can announce that the crashing problem has disappeared by using a different version of Gmsh and enGrid (i.e. Gmsh 2.7.0 and enGrid 1.3.0). I installed the programs on my home computer and now enGrid works even faster, even though this machine is way older than the one I am using at work.

The problem with bad aspect ratios still holds on, that's why I atteached a picture from the mesh with boundary layer created with enGrid and the *.geo and *.msh file generated with Gmsh. So if someone has time to take a shot to overcome this sawtooth effect at the transition zone I would be really grateful.

So long,
Bennit