CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ANSYS Meshing] Meshing Strategy for inside geometry (

powpow January 15, 2013 10:25

Meshing Strategy for inside geometry
2 Attachment(s)

I have a question regarding to create mesh in a tube with geometry inside.

I have to do a simple CFD-Simulation how an additional tube in a test chamber influences the air flow.
So I build a CAD-Model of the test chamber with CATIA V5. The CAD-Model is a surface model, no solid.

I imported the .igs-file into ICEM CFD 13.0.
Then I build topology.
Next I created parts.
Then I created a BODY called Fluid inside my tube.

The result is shown in Image 001

I'd like to create a mesh with 10 layers of prism around the tube and the small nozzle, the rest of the volume should be filled with tetra.

So far, I tried to create the mesh on a lot of different ways, e.g. like it is described in the tutorial with the helicopter:
- first create a mesh using the Robust (Octree) method
- smooth the surface mesh using the Laplace algorithm
- recreate the volume mesh now using the Delaunay method and keeping the surface mesh previously generated
- smooth the new mesh again
- generate the prism layers
- smooth the mesh several times freezing the prismatic elements
- do one last smoothing step lowering the desired quality by one or two orders of magnitude now including the prisms

or another method:
- Register Mesh - Surface Mesh Setup - Surfaces: I selected all surfaces, entered my values and applied.
- Register Mesh - Compute Mesh - Volume Mesh - Compute.
- Then, I wanted to add 10 layers of prism:
- Register Mesh - Compute Mesh - Prism Mesh.
- There I selected the parts for the prism layers and clicked "ok".
- Then I clicked surface mesh up, chose the surfaces of the tube and the small nozzle and applied.
- Register Mesh - Compute Mesh - Volume Mesh. I enabled create prism layers and computed.

I always checked the mesh, too!

Each result I exported with Register Output - Select Solver. I chose Ansys CFX and ANSYS and applied.
Then Register Output - Write Input and enabled BINARY instead of ASCII.

Then I started CFX-Pre - New case - Simulation Type: General
Click with the right mouse button on mesh - import mesh - ICEM CFD and chose the created .cfx5-file.

No matter which way I created the mesh, I alway "loose" my inside geometry as you can see in the image 002. The parts aren't listed at the structure tree, too.

I'm really desperate and hope someone can help me! :)

diamondx January 15, 2013 12:07

by loose do you mean some part don't show in the structure tree ?

powpow January 15, 2013 12:19

Yes, they aren't neither shown in the structure tree nor they "exist" in my model on the right side (see Image 002)...:confused:
It exports always only the utter geometry elements!

diamondx January 15, 2013 12:25

you procedure in ICEM was good. you just forgot to mention those part as internal wall.
The reason they don't appear in your solver is because there is no element (shell) on them, so your solver skips them.
in order to generate a volume mesh without skipping those part, you need to mention them as internal wall you can do that in mesh part setup... use the search forum for these keyword, you will find more info

powpow January 16, 2013 04:02

Thanks a lot! This seems to be the solution :)

I just defined the parts as internal walls by ticking them at the part mesh set up.

powpow January 16, 2013 05:06

Multiple edges
1 Attachment(s)
After defining the parts as internal walls, I created a first mesh using the Robust (Octree) method. After that, I did a Check Mesh and got the error shown in Image 003.

What can I do to fix this Problem?

powpow January 16, 2013 05:32

2 Attachment(s)
Since I found here in the forum the hint that "There are certain solvers that can't handle multiple edge T connections, but major commercial solvers don't mind at all." ( I decided to click "Ignore".

The next error was "Single edges". I clicked "delete" (image005)

If I than export it into CFX-Pre it looks like image004. Is this a Problem?

All times are GMT -4. The time now is 00:32.