CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   Near Wall Refinement (http://www.cfd-online.com/Forums/ansys-meshing/80228-near-wall-refinement.html)

Ridley September 20, 2010 06:38

Near Wall Refinement
 
2 Attachment(s)
Hi, I need some advice on near wall refinement. Iīm simulating a hydrocyclone which has a tetra-mesh. My problem is that I donīt know how I can do a similar refinement as shown for the structured mesh (prism doesnīt work for this). I donīt want to use a structured hexa-mesh, because I got better convergence and residuals when using the unstructured tetra-mesh. My idea is to use a hexa-mesh in the boundary-layer and a tetra-mesh for the rest, but I donīt know if this is the best solution in terms of accuracy. It would be great if anybody knew a better way to to do this.

PSYMN September 21, 2010 11:55

Prism
 
3 Attachment(s)
You just need to insert a prism layer... What tool are you using?

If it is ICEM CFD (it looks like it is), you just need to go to Params by Parts and check box "Prism" for each part that you want prism on. Similarly for volumes that you want prism to grow into.

Global Mesh Params has a spot to set the global prism params.

Then Go to Compute Mesh => Prism to generate the prism. With ICEM CFD, we use a post inflation method that best generates prisms after you have volume mesh. But this "normal" method generates Penta-6 elements from a tetra tri mesh.

If you want Hexa prisms, you need to change the order a bit. Use a surface meshing method to generate the quad meshes on the walls you want inflated and tri meshes on the side walls. Then Generate the prisms (without any volume elements). Then fill the remaining volume with a delaunay tetra fill.

You can find prism tutorials, etc. in the soft ware. You can also find a lot of Prism Q&A on CFD Online.

Ridley September 22, 2010 03:24

Hi Simon, Iīm using ICEM. Quads are exactly what I wanted to have in order to get the needed refinement. I will try to do as you described.

Thanks a lot! :)


All times are GMT -4. The time now is 13:49.