CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] meshing a labyrinth seal (

haho June 9, 2009 05:33

meshing a labyrinth seal
Hi, my name is Hannes and I'm from germany.

I'm trying to mesh a axisymmetric labyrinthseal with ICEM.
I started with a 2D Blocking of the 2D geometry I got from SolidWorks.

My problem is Blocking the arcs between the "peaks" of the labyrinth.
I tried using the O-Grid splitting but I dont get any satisfiing results.
Can you please give me any advice in meshing this sort of geometry the fastest way because I will have to mesh several layouts.

Thanks for your help,

fauka June 10, 2009 05:28

1 Attachment(s)
I tried out this blocking. Hope it helps

haho June 10, 2009 13:54

Hi, thanks for your reply.
I'm still trying to get a C-Grid. Can you please explain to me how I will get it right, because everytime Í merge the vertices after the splitting of my blocks i get a result like this:

thanks for your help!

PSYMN June 10, 2009 21:12

Merge one more pair...
You need to merge one more pair of verts on each side...

Or you could just use the Ogrid tool to take care of it automatically...

fauka June 11, 2009 01:55

3 Attachment(s)
This is how I generated the blocking.

haho June 11, 2009 03:08

Thanks for your help.
I will have a try later on and post my results :).


haho June 28, 2009 08:24

2 Attachment(s)
Hi, it's me again...

I've now solved my Problems with the C-Grids but now i'm in trouble with Meshing an other Geometry. It seems as if the Edges are sort of linked together but I don't know how this happend.
The first vertical edges are changed when im refining the nodes in the right vertical edges, is this a bug or is there are solution for this problem?

Thanks, Hannes

PSYMN June 29, 2009 12:58

Delete Blanked blocks...
Depending on how you created this, you could end up with parallel edges that don’t at all look parallel because the hidden Vorfn edges are connecting things up behind the scenes. This is most likely if you used a series of split and merge operations. Using the OGrid tools would have been cleaner…

Anyway, what you need to do is recreate this VORFN blocking.

Turn on all the parts and blocking, etc. so you can see everything you want to keep… Usually, this means everything on except the VORFN part.

Then go to delete blocks and turn on the “delete permanently” checkbox. Then select all “blanked” blocks (hotkey b or use the selection toolbar).

What this will do is delete all the VORFN blocks (since they are off = blanked). But then Hexa needs those VORFN blocks so it automatically recreates them based on the topology of your other blocks. This should be sufficient to completely solve your problem.

haho July 1, 2009 08:59

1 Attachment(s)
thanks that worked perfectly!
But Ansys keeps me busy...
In Pre and Post, Ansys has problems displaying my surfaces, there are some fragments but I'm not getting the whole area displayed correctly.

I'm converting the PreMesh to an unstructed Mesh and them I'm extruding the 2D Mesh to an axisymmetric 3D-Mesh then I'm converting the Mesh to "Ansys CFX"-files in ICEM with the output-function and the extruded surfaces in ICEM look ok.

Is there a solution for this Problem? I'm running ANSYS 11 SP1.

PSYMN July 1, 2009 10:41

CFX Mesh display is sensitive to element normals... Without seeing an image of what you mean, that is my first guess.

While in ICEM CFD, use the Edit mesh tools to adjust the normals...

Then export to ANSYS CFX...

haho July 3, 2009 06:13

4 Attachment(s)
thanks for your reply.
I've uploaded some other pics of my problem.

I only get these problems in POST when I'm switching from "variable" to "constant" color for highlighting the surfaces.

I this the right feature for changes the normals?
Edit Mesh - Reorient Mesh - Reorient Volume

Thanks for your help!


PSYMN July 5, 2009 08:52

Try the normals...
I am not a CFX expert... (just a user like you). Perhaps ask this question on the CFX site...

But first, try the normals thing... You can know before you try by right clicking on shells in the model tree and display the normals by color. If that is the problem, it should look similar to the images you just posted. If so, we just need to make sure they are all pointing the same way... There are lots of options in there. you could make consistent, etc. That reorient volume option should take all the normals and point them out from the volume (good idea).


All times are GMT -4. The time now is 16:41.