create boundary layer mesh by extrusion, orientation problem
 

Hi,

I was trying to create a thin layer of boundary layer mesh by extruding the surface mesh of existed mesh to outward of the fluid domain.

But it always resulted in volume orientation problem.

Is there any solution to solve this problem?
The key issue is that I have to create boundary layer or prism-type mesh base on the existed mesh.

Thanks.

L

Are you sure that the orientation of the elements that you extrude have the same orientation/normal vector? In other words, is the orientation consistent?

Hi Gert-Jan,

The orientation is consistent.

But there are several points with invalid normals ( which is defined here:
https://www.sharcnet.ca/Software/TGr...ug/node225.htm)

I guess it is impossible to create extruded inflation layer for the complex geometry with invalid normals.

So I turned to try the prism mesh with Post Inflation in ICEM.
Unfortunately, I got the warning:

Assertion Failed!
Program:\med.exe
File: improve.C
Line: 339

Expression: nngh.

You are using ICEM, but are looking in the help of TGrid. I would at least make that consistent ........ :-)

If you have a mesh with only tetrahedrals, then post creation of prisms should work. Only if you have a very small gap where your tets are large (only 1 element in the gap?) and prisms do not fit in anymore, then you get this error: Assertion failed.

Do you get this error already with 1 prism layer? Just for a check.........
Mostly, the solution is use smaller elements in the gaps.

Thanks.

lol - I also tried T-grid and also failed.

The original mesh is consisted of hex and wedge elements. Indeed, there are many small gaps and only two elements in those gaps. The error already appears with 1 prism layer. I shall try more elements in gaps now and see what will happen.

You can only build prisms from tetrahedral elements.

Is there any method to create post prims for hex elements by ICEM?

No. That is impossilbe. How would you do that? You can't even get it done with pen and paper. Because if you have hex elements, then all faces are rectangular. So, you cannot get prisms from rectangular faces using Post creation process.

If you have hex elements, and want refinement close to the wall for resolving your boundary layer, then why don't you just create a refinement using hex elements. Even better then prisms.

My geometry is very complicated.
So I cannot handle each face manually, then I have to write script to generate the surface elements.
Base on extrusion of the surface elements via some directions with ICEM, I can create the volume elements.
By this I can control the total grid numbers and the mesh quality - But I am not able to design the prism by script. So I was trying to get the prism done by ICEM or other tools.

Thanks for the suggestion. Is it possible to mark the first layer of grids in ICEM and split it into 1 thin layer and 1 larger one?

You can split the first layer of prism elements. You can split it once, twice, whatever.
You can also split the fifth layer, or the tenth. All possible.

Not sure if you can split the first hex element. I am not that familiar with hex meshing.