CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] Mesh reflection to obtain perfectly symmetric grid (

Chipsgrottel September 15, 2011 14:33

Mesh reflection to obtain perfectly symmetric grid
Hello everybody,

  • I am working with ICEM CFD on a complex 3D mesh that has two symmetry planes. Since my blocking is very complex I would like to focus on half of the geometry, then generate the mesh for this first half and reflect it around the mid plane to obtain symmetric mesh around it.
  • The problem is that I have errors at the interface between the first half and the reflected mesh and I would like them to be nicely merged.

Is anyone able to give me some advice on this?

Thank you very much in advance

brunoc September 15, 2011 21:41

What type of errors?

BrolY September 16, 2011 03:50

Do you try to merge the blocking or the mesh ?
Because you can rotate your blocking and avoid to merge two meshes.

Anyway, if you merge two meshes, be careful to the mesh law of your edges.
Because if you specified a "geometric 1" law on one edge of your symmetry plane, you would have issues to merge the 2 meshes once you have make a symmetry of your mesh 1.
The edge of your mesh 1 would have to be merged with the same edge of your mesh 2, but with the symmetry operation, the orientation of your edge would have changed. And with a geometric 1 law, the nodes on those edges won't be aligned anymore !

PSYMN September 16, 2011 10:11

Are you mirroring the blocking or just the mesh? If it is topologically symmetrical, but not geometrically symmetrical, you should mirror the blocking and then adjust to fit the actual geometry... But if it is geometrically symmetrical, you can just mirror the mesh, then merge nodes with a tolerance and delete the symmetry shells.

I don't think you should have a problem with your mesh distribution laws (as BroleY was talking about) if you mirror, but only if you copy rotate by 180 (which I hope is not what you are doing).

For more help than this, you will need to tell us what the errors were.

Chipsgrottel September 21, 2011 17:54

Thanks everybody for your interest and your help,

  • The errors I get are mainly duplicate elements at the joint in the mirror plane. Then I also get multiple edges and non-manifold vertices. I do not know how much it is going to affect my calculation with Fluent.
  • I decided to do as you advised me to do, namely mirroring the blocking. I just think that it is really complicated since I am working with more than 300 blocks and doing twice the associations and being sure that I have done exactly the same on both sides is really complex.
I have other issues that I would like to discuss on the forum. Do you think I should open a new thread?

Thank you again!

PSYMN September 21, 2011 20:18

When you mirrored the mesh, you have a duplicate symmetry plane. You need to merge nodes with a tolerance (set to something very small like 1% of you mesh size) and then delete the elements in your symmetry part.

Then run your checks again and you should be fine.

Yes, other issues deserve other threads...

Chipsgrottel September 22, 2011 12:00

Ok I am doing this, thank you so much.

Sawi December 11, 2014 07:47

As i have almost the same problem how do i delete the symmetric shell? i already chose the option of merge nodes and delete duplicated elements while using the mirror option? is this not enough and what should i do more to get rid of multiple edges and non-manifold vertices. thank you very much

PSYMN December 11, 2014 11:04

Assuming the symmetric shell is in a unique part name (such as "SYMM") you can turn off all the other parts and just delete the elements.

There are a bunch of element selection options, so if you don't want to drag a box to select, you could delete visible or delete by part or draw a polygon around the elements you want to delete.

This will just delete the shells and not the volume elements on either side.

diamondx December 11, 2014 11:07

I usually run a mesh check for duplicated elements, I fix it by clicking delete. and When exporting, I define the rest as interior.

PSYMN December 11, 2014 11:22

Yup that works too... And by keeping those shells as interior, you can more easily post process on them.

All times are GMT -4. The time now is 12:53.