CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] Prism for quad mesh (

rikio March 16, 2012 12:23

Prism for quad mesh
Hi All,

I am going to mesh the model with hybrid meshes. Part with tetra, and hexa for the other one, both of them were supposed to include boundary layers. It is easy to get the prism layers with a specified growth ratio for tetra mesh, but not that easy for hexa mesh. In order to keep the consistency of the mesh nodes on the merging interface, to be better recognized when merging, the thickness of the near wall layers of hexa need to approach that of the tetra as much as possible. Splitting edge by specifing the split ratio would work, but need to manually calculate it since only one cut could be done in one operation. More layers needed, more complicated for the boundary layers be generated. As an alternative, I checked on the Use Existing Quad Layers option to generate prsim on quad meshes, but no matter what numbers were given for the layer numbers, ratio and initial height, the first layer always be equally split. Anyone have such a problem? I am thinking that it is a BUG of ICEM.

How did you do to get the boundary layers when obtaining hybrid meshes? Detailed procedure would be very appreciated. Thanks for any information.

PSYMN March 16, 2012 13:22

Use Existing Quad layers has some issues (on the list to be worked on). You can get it to work, but it is tricky. If you go the other way (insert the prisms in the Hexa side and Tetra side at the same time) the entire boundary layer has to fit inside the first hexa layer...

The easiest way is to use a meshless interface (if your solver can handle it)... (just mesh the two sides independently and don't even try to merge them).

I will try to get back to this with some how too steps later, but I am having a very busy month and don't have as much time for CFD-Online as I would like...

PSYMN March 16, 2012 13:28

3 Attachment(s)
Here are some pics of align to quad working...

I will try to come back and explain how some other time. But I really suggest you try non-conformal. You may save yourself some hassle.

rikio March 16, 2012 21:59

Thanks a lot for your comments, Simon.

I downloaded a presentation of you, Tips of ICEM 2008, at somewhere of the forum, to start to try the Use Existing Quad Layers option. I used to think that it would be the solution to mesh boundary layer, meshing with tetra and hexa separately which including only one layer first, then merging them together, splitting the one near wall layer to multiple at last. But still always failed at merging the interface and prisming for the hexa part. As you suggested, go on with non-conformal interface will be easier, but it may introduce some model error into the calculation. So, I am really looking forward to your how-to. It may help me a lot. Thank you very much.

rikio March 16, 2012 22:11

Maybe meshing without prism first, generating prism and splitting after merging, would be a workaround. Succeeded on a simple geometry just now, and try that on a more complex model later.

All times are GMT -4. The time now is 11:20.