Unstructured Prism Layers for Boundary Layer Resolution...
 

Hello all,

I have seen unstructured prism layers (they are rectangular prisms cut in half in 3D and rectangles cut along the diagonal into triangles for 2D) being suggested for boundary layer meshing. I havent used them but I am interested. But I am worried; dont these have horrible skew? How do they actually perform in practice?

In automobile industry use of prism inflation (aka boundary layer) in very popular. For many complex geometries this is the only solution.

If you have ever tried the Pointwise's anisotripic tetra meshing (T-Rex) http://www.pointwise.com/T-Rex/, you will notice that thier quality is inferior to prism mesh. (These are my personnel comments from my limited knowledge, If someone has more knowledge then pls correct me)

This is the one of the strong point of ANSYS meshing and ICEM CFD.


Do you have more info for this particular technique and posting few pics would help alot to comment further.

Hello Robert and Far:

Far, I'm going to have to challenge your comment regarding the relative quality of Pointwise's T-Rex method relative to what you call "prism mesh."

T-Rex generates high-aspect ratio tets. They are of good quality because one of the included angles is roughly 90 degrees. While that's OK unto itself, the T-Rex method includes an optional post-processing step that combines 3 T-Rex tets into a single prism. This prism is of good quality because all three of the lines off the base triangle are roughly 90 degrees.

Best Regards

I have been using the Gridgen since Release 15.09. And I am really inspired by the relative ease of creating and moving connectors and domains from one location to other location. Gridgen's elipitic solver is also of high quality.

As I already said these comments regarding T-Rex developed by Pointwise reflects my very limited knowledge. I am interested to explorer more about this.

Now my inclination is more towards the ICEM's prism meshing combined with other methods.