CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (https://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Merging two hexes with different mesh density (https://www.cfd-online.com/Forums/ansys-meshing/94349-merging-two-hexes-different-mesh-density.html)

leekj92 November 12, 2011 18:53

Merging two hexes with different mesh density
 
Hi guys,

So I've been trying to merge two hexahedral meshes that are rectangular but have 3:4 mesh density difference. I.E one mesh has 48 nodes in x and y direction and the other mesh has 64 nodes in x and y direction. I am trying to merge the meshes so that there are no interfaces (all faces are 1 to 1). Is there any way to do this automatically? (I've been doing this manually, but my mesh has gotten too big to do it manually).

Thanks!

rikio March 16, 2012 00:10

Merge meshes would meet your requirements. But pyramid would be generated also.

Pacot March 21, 2012 08:31

Hi all,

I am also trying to merge two hexa meshes with different mesh density with the purpose to get a 100% hexa mesh.

Thus, I first merged the two hexa meshes. Then, to have elements connected node by node I used the "Resolve refinements" in Edit Mesh\Merge Nodes. However, icem report that the mesh has no couplings...

Does anyone has a hint? Is there some special precaution to take care of when generating the two topologies?

Thanks a lot!

rikio March 21, 2012 10:20

Resolve refinement is used to make the non 1:1 mesh to be 1:1. Generally, it is a reverse operation of the refinement which could be accessed at Premesh Parameter/Meshing Parameter/Refinement. If your mesh is not a connected, N to 1 mesh, you would get the error message.

Pacot March 21, 2012 23:05

Hi Rikio,

Thanks for your answer! So, do you know if there is any way to merge two hexa meshes directly into icem ? I mean, if the merge mesh does not produce a connected mesh, how should we do to benefit from the resolve refinement function?

So far I split all my blocking into two parts to be able to generate my mesh using "standard" computer. However, the inconvenience of this is that my two blocking have to have the same topology at the interface.

I don't know if Leekj92 could manage this issue?

Thanks for any suggestion!

rikio March 22, 2012 02:45

Mesh at the interface could be merged by using the build-in merge node tool. But it may not always work well. As an alternative, a non-conformal interface could be created in solver, like CFX for fluid flow, to deal with an averaged algorithm.
Someone in ANSYS used to suggest me not to rely on strictly conformal mesh, mismatched one would work well too. But the interface should not be lieing in the sensitive positions.

I am not sure whether I answered your question. You can send your project files to me at ri7ja@163.com, or just right here, for a better understanding.

Pacot March 22, 2012 22:17

1 Attachment(s)
Hi Rikio,

Basically, what I am trying to do is to merge two hexa mesh to obtain one hexa mesh. As the in-house software does not, so far, allow non-conformal interface, like CFX, and no multiple elements, I have to find a way to generate my full hexa mesh.

I am actually meshing a centrifugal pump. I meshed one part consisting in the guide and stay vanes and a second part consisting in the spiral casing. As you can see on the picture, I merge these two meshes to obtain one. You can see the stay vane in brown, the top of the volume of fluid in violet and the spiral casing in cyan. However, I don't have a conformal interface between my two meshes. That's why, I tried to use the resolve refinements function to generate a conformal interface. But it seem it is not applicable for such operation.

So, I don't know if this kind of merge can be performed directly using ICEM. However, I will be able to generate my mesh using my two blocking, with conformal interface, and merge them to obtain my complete mesh. But definitively, if I could merge my two hexa meshes would considerably save me a lot of time :)

Olivier

P.S. My research is from the industry and I am thus not allow to provide my project file... I hope that the explanation above would make my problem more clear. Anyway, thank you very much for all your help!

rikio March 22, 2012 23:08

Hi Pacot,

If conformal mesh at the interface is mandatory, blocking may be the best way, or you will take another effort to handle the interface. Mesh nodes could be merged too, but if all hexa is required, node numbers on each side of the interface should be exactly equal. Otherwise, pyramid would be there for transnition. In you current model, neither the mesh size nor the node number is equivalent. So, the result of merge would be something like that presented in the post below.
http://www.cfd-online.com/Forums/ans...g-problem.html

On the other hand, as far as I know, non-conformal interface is allowed in CFX by using MFR or Frozen Rotor model. And it is widely used in the simulation of compressor/turbine.

Wish it helps you a little.

Pacot March 23, 2012 08:22

Hi Rikio,

Thanks a lot for your suggestions. I will thus definitively stop to work on merge my meshes and go with the blocking. It will take more time but it will produce exactly what I want.

I decomposed my whole blocking into several sub-topologies, allowing me to work easily on the different part of the mesh. Once I'm satisfied of the different meshes it remains to me to merge the different topologies to generate the whole mesh. It takes some time but it will generate a full hexa mesh :)

Thanks again!

Olivier


All times are GMT -4. The time now is 22:23.