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/)
-   -   [ANSYS Meshing] Questions regarding hex mesh (http://www.cfd-online.com/Forums/ansys-meshing/130159-questions-regarding-hex-mesh.html)

oj.bulmer February 20, 2014 12:03

Questions regarding hex mesh
 
I have few questions regarding hex mesh:

1) Is it possible at all to have a prism inflation layers with structured hex mesh in ICEM CFD?
2) I understand that though ICEM CFD is generally used commercially for structured hex mesh, how robust is Workbench meshing for creating structured block meshes? Since ICEM CFD is a bit costly, we decided on Workbench meshing. I tried to create hex mesh for simple geometries like pipe etc by dividing the geometry at CAD level and I could get the structured mesh to work, but I am stumped when it comes to using this approach even for the easiest tutorial for ICEM CFD, viz. 3D pipe junction. It is not easy to visualize and create those blocks by dividing the geometry itself...

flotus1 February 20, 2014 12:24

Recently there were some nice features added to the Ansys mesher like MultiZone (http://www.cfd-online.com/Forums/ans...multizone.html), but the capabilities are indeed limited when it comes to block-structured hexahedral meshes.
Those workarounds with splitting geometries to mimic the behavior of ICEM in the Ansys mesher are only for very simple geometries. Nobody would want to do this with anything more complex than an array of cylinders.
Even If you managed to split a complex geometry correctly, you would still be limited by the fact that you dont have enough control over the node distribution at the edges.

To conclude: For the kinds of meshes you have in mind, ICEM (or similar meshing tools) are the only reasonable choice.

oj.bulmer February 20, 2014 12:57

Thanks Alex. I can guess that when the geometry gets a bit complicated, the multizone will produce an unstructured hex mesh surrounded by prism instead of a structured one as shown in your last pic. From what I read, such unstructured hex mesh doesn't have much benefit over an adequate tet mesh, especially when it comes to complex industrial models.

Any thoughts?

flotus1 February 20, 2014 18:06

I would still prefer such an "unstructured" hexahedral mesh over a tet mesh.
Although it may not look as neat as a block-structured hexahedral mesh, it still has some of its advantageous properties.
Most importantly, the mesh is still "streamlined" with the geometry/main flow direction reducing numerical cross-diffusion to a minimum.
Even if that fails, the element quality still is superior to any tetrahedral mesh.
For low-end industrial applications like calculating pressure drops a good quality tet mesh may perform just as good,
but in my opinion this is because such kind of task could be performed with almost any kind and quality of mesh.

Then there is another new feature which I didnt try too much until now, but it looks very promising.
It is called HexCore and basically does the same as SnappyHexMesh.:rolleyes:

Far February 21, 2014 07:23

Ansys meshing is combination of CFX mesh, gambit and some other algorithms and added multizone from ICEM CFD hexa.

The primary motivation of this all was to give a user integrated environment by implementing their (gambit, cfx mesh etc) best tools under one umbrella and also to provide it under the modern software architecture.

Main purpose of ansys meshing is to get you quick mesh with few clicks of mouse. Thats all. You should not expect more from it.

RicochetJ February 23, 2014 13:01

Surely you can decompose your geometry in Ansys Meshing using the virtual topology option which will enable local control that you need?


All times are GMT -4. The time now is 20:32.