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] Merge meshes with internal wall (https://www.cfd-online.com/Forums/ansys-meshing/113454-merge-meshes-internal-wall.html)

BrolY February 20, 2013 03:14

Merge meshes with internal wall
 
2 Attachment(s)
I have 2 projects : tube and cube (.tin .blk .uns .prj attached).

The inlet is located at the down part of the tube, and the outlet at the top part of the cube. The top of the tube is open into the cube. And the walls of the tube which are inside the cube (almost half of the tube) are internal wall.

My goal is to merge those 2 meshes, and at the end, to have only 1 volume (fluid). I opened ONLY the cube_only.uns and tube_only.uns files, and I merged the nodes the wall of the tube which is inside the cube with the correspondant part, and deleted one of the interface. Next, I tried to specify the remaining tube wall (which is inside the cube) as an internal wall. But this option is gray tint.

How am I sure that this interface will be considered as an internal wall by my solver ? (let's take as an example FLUENT) Was it the right way to do it ?
I tried to specify one of the interface as an internal wall when I created the blocking, but after I imported the .uns file, the option was gray tint.

Far February 20, 2013 04:31

You can merge blocking as well.


You want to keep the interface (2 and 3) ?

BrolY February 20, 2013 04:38

1 Attachment(s)
What do you mean by 2 and 3 ?

Attached the merge I've done as I explained in the previous post. I merged nodes (Edit Mesh -> Merge Nodes) of INTERF_WALL and INTERF_WALL_01, and then delete INTER_WALL_01.

I don't know if Fluent (or another solver) would consider INTERF_WALL as a separation between the fluid from the tube and the cube because I can't specified INTERF_WALL as internal wall.

Far February 20, 2013 04:58

You want the internal wall as interior or wall? Int wall is really used to define the wall.

You are merging two hexa meshes? Which I dont think is good idea. Better idea would be to merge blocking and you have the full control over everything.

If you want to get the int wall option, do this!!!

1. Load mesh

2. Edit > Mesh >>> Facets

3. Define int-wall in part mesh setup

Far February 20, 2013 05:06

1 Attachment(s)
Please find enclosed fluent file. Do you want mesh like this?

BrolY February 20, 2013 05:16

I just want the solver to understand that this wall is a wall, not shells which should not be here (2D thickness). So the fluid can't go throught this wall.
I've done that only with Tetra and I know that the option "int wall" tells the mesher (basically) "it's a wall, dont delete those shells". After I can specify those shells as a wall when I export my project.
My problem is I'm facing a mesh created by a merging of 2 hexa meshes and 2 interfaces at the end. So I don't know if merging the nodes, and deleting one of the interface would led to what I'm looking for. If I merge nodes, would it connect the volume mesh from the 2 sides of the interface and at the end, the solver would consider the shells between the volume as a wall ? Or would it consider the shells as an error and just delete them ?

That is just a simple example, but my geometry is more complicated.
Merging blocking is a good idea, but I have something like 8 blocking.
I've noticed that at some point, it's really "heavy" to ICEM to handle it (crash, slow-down ..).

So merging unstructured meshes appeared to be the best solution.
And I want to keep Hexa, not to go with facets ... because facets means to convert your mesh to tetra, am I right ?

Far February 20, 2013 05:21

Quote:

Originally Posted by BrolY (Post 408955)
So merging unstructured meshes appeared to be the best solution.
And I want to keep Hexa, not to go with facets ... because facets means to convert your mesh to tetra, am I right ?

Here mesh > Facets is used to extract the geometry from underlying mesh. and once you have geometry you can specify wall as int wall in part mesh. May be in background triangulation is used to make the surfaces but it does not mean that you are converting your hexa mesh into tetra.

I find this thread interesting.

BrolY February 20, 2013 05:31

Ok, so ICEM needs the geometry to create internal wall ? It means that when you export your mesh with internal wall, you always need to have the geometry file in your project, and not only the mesh ?

I'm still wondering if it would works. If the process merge nodes + delete one of the interface would not have created problems.

Thanks for your help ;)

Far February 20, 2013 08:11

Did you check the fluent file I just uploaded?

BrolY February 20, 2013 08:23

There are only .prj and .uns file.

That looks like to the projet I've uploaded in my 2nd post. How did you proceed at the end ?

Far February 20, 2013 08:37

1 Attachment(s)
Oh wrong file .

diamondx February 20, 2013 10:10

@broly, i understand what you want.
I once was wondering if this was doable, but never had the chance to test it out.
Now that i did,i am missing something because it cant be so easy...
When i opened your merged project where you have merged the two meshes. You left only the inter_wall. You want that to be a wall in the output menu, i specified inter_Wall as wall. it worked and it created a shadow in fluent. isn't it what you want ??
I guess no...

BrolY February 20, 2013 10:15

i'm not really familiar with Fluent, so what does "shadow" mean ?
Is it a 2D wall surronded by fluid (but keeping them from being connected) ? If yes, that's what's I'm looking for !

Far February 20, 2013 10:16

For internal wall (inside fluid) you have two sides. one is wall and second its shadow (other side) :D

BrolY February 20, 2013 10:21

So that's good I think. For each part of the fluid, it's a wall.

Also, it means that I don't need to define the INTERF_WALL as an internal wall.
The "internall wall" option is only for the mesher. Am I right ?

diamondx February 20, 2013 10:22

yes it is a 2d wall surrounded by a fluid. i guess fluent makes a shadow of that wall because you have elements in both side of the wall. where you usually have only one side...

diamondx February 20, 2013 10:32

I feel like you are planning to do multiple separated blocks, and merge them... isn't easier to do one blocking ?
for a perfect merge, the node should be in the exact location... did you create a sub-topo for this one ?

BrolY February 20, 2013 10:33

Thank you guys !
I guess we all have learned something today ;)

Far February 20, 2013 10:39

Quote:

Originally Posted by BrolY (Post 409070)
So that's good I think. For each part of the fluid, it's a wall.

Also, it means that I don't need to define the INTERF_WALL as an internal wall.
The "internall wall" option is only for the mesher. Am I right ?

You have to explicitly tell ICEM to treat extended surface as wall, otherwise it will not recognise it and think it is by error and removes the shell elements.

For tetra mesh, you define it in part mesh setup, whereas in Hexa it is defined by making the face to surface association explicitly

Far February 20, 2013 10:41

Quote:

Originally Posted by diamondx (Post 409076)
I feel like you are planning to do multiple separated blocks, and merge them... isn't easier to do one blocking ?
for a perfect merge, the node should be in the exact location... did you create a sub-topo for this one ?

Sub topo is much better idea than merge


All times are GMT -4. The time now is 11:50.