CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] Hybrid mesh (

lostinicem July 6, 2012 05:54

Hybrid mesh
4 Attachment(s)
Hello community,
since 2 weeks Im fighting with my problem and now I came to a point, where I dont know what to do, so I ask you and hope for your help:
I have to create a geometry ( I want to create a cone ) which contains/ is made of structured mesh. Here a pic:Attachment 14241
After it is created I need a tetra-mesh around it, so it looks like this when I create it:Attachment 14243
So in the next step is merge it and smooth it, it looks not really good but this is what it looks like:Attachment 14244
So the quality is not that good but this is where the problem starts; I need hexa layers around my geometry and those merged with tetra and hexa mesh, something like this:Attachment 14245
But this is not merged. If I merge the tetra mesh before making the layers I cant make them afterwards with the prism layers option. and if I create them first, like in the last pic I also cant merge them with my geometry, because the computer crashes.
Can anyone help me with this:
- I need to know how to make better hexa mesh and also control the ratio ( so it gets bigger from cone to the walls and it is small around my geometry)
- how can I merge the hexa and tetra mesh and have those tetra layers around it.
I googled, searched through this and other forums but still nothing, that could help me.

best regards

diamondx July 6, 2012 08:31

for a good merging size elements matters, did you take a look at this thread:
if you have the possibility to mesh it all in hexa, will you do that? hexa is much better than tetra, because of size and computation errors i found.
If you don't mind, you are welcome to share you project with us... we can dig more into the problem

Ananthakrishnan July 7, 2012 03:07

Hi lostinicem,
Just a small note.. I just found out that, if you use fluent you dont have to merge the meshes.
You just have to define the "interface" in "mesh interfaces" in fluent and it will run normally. Even the size of the structured and unstructured elements at the interface does not matter..

Far July 7, 2012 03:14

in that case you have write the both meshes separately and then use read and append commands to combine them.

Ananthakrishnan July 7, 2012 03:19

I dont think thats required. in case of fluent, after both meshes exist under the mesh tree in ICEM (ie both structured and unstructured mesh exists at the interface), you just have to take the output as usual into .msh type. then define the interface separately in fluent..

diamondx July 7, 2012 07:35

you mean using the interface command without appending the meshes. loading the meshe earlier in one shot ?? i didn't know that...

lostinicem July 9, 2012 04:27

uh sorry to watch here late!
@diamondx: i saw that thread and actually this was my first choice to make this kind of steps, but it didnt worked out; i can see now why, i guess my tetra-mesh was too large, because i couldnt merge it, like in your picture.

@anan; also thanks for this hint, Ill try it when mesh is finished

all in all thanks for the answers; I just thought i could also work here with those layers/layer-option, so it would make easier with other geometries; because in tutorials you can see alot of those meshes, but I cant handle to create it ....

@diamondx regarding my problem: there is conical geometry, which is in a flow that includes also the temperature. my problem is to generate some structured-mesh-layers around the cone, to have a better quality of what is happening there. the rest is tetra-mesh/tri-mesh (unstructured mesh), because its not that interesting for us and since making it as a whole as structured, would cost too much calculating time. and if I take only unstrucutred mesh, it seems its not accurate enough. but you can correct me on my opinions here, since Im new to this kind of problems and maybe I misunderstood something.

best regards

lostinicem July 16, 2012 05:51

by the way Im trying now to create a cone around cone and to control the structured mesh with this; but it seems like I cant adopt the idea from cube, because if I split here blocks at the end the mesh of the bigger cone looks pretty strange.
I guess I need to play with the settings of associatins ...

anyone got any hints?

lostinicem July 16, 2012 06:05

2 Attachment(s)
sorry for tripple post ...
here some pics: before i make volume mesh ( so if I have only surface mesh) ;it looks ok:Attachment 14461

and after I create the volume mesh:Attachment 14462

energy382 July 16, 2012 06:07


Originally Posted by lostinicem (Post 370410)
my problem is to generate some tetra-layers around the cone, to have a better quality of what is happening there. the rest is hexa-mesh, because its not that interesting for us and since making it as a whole tetra would cost too much calculating time. and if I take only hexa mesh, it seems its not accurate enough.


why do you need tetra layers? i really don't understand.

i think you confuse tetra with hexa!? you need a hex mesh with hex layers and tetra around the body!? and you want to match the hex layer with tetra

sorry, but i don't understand your issue

lostinicem July 17, 2012 04:59

2 Attachment(s)
@energy, thanks for your attention! maybe this is why nobody answers:p I edited my post:
yes, its about cone which has structured mesh-layers around it and after those layers there is tetra mesh :D
so I thought, maybe if I create around one geometry same but bigger geometry, I can influence those structured mesh layers. but blocking seems not be right here, because the volume mesh gets messed up ... and so Im looking for new ideas..

and here what Ive got till now, its a square form around cone:Attachment 14541

and at the walls:Attachment 14542

lostinicem July 31, 2012 04:44

1 Attachment(s)
I managed to create the mesh of the cone ,finally, how I wanted, but the problem now is merging the mesh:Attachment 14868

after creating surface and body, it doesnt want to merge hexa and tetra mesh. if I create normal geometry and dont do alot about it, the merging is no problem, but as soon as I create this kind of cone its says: there is currently only 1 region.
does anyone knows whats the cause of this problem????

Ananthakrishnan July 31, 2012 04:53

can you put a screen shot of the "mesh tree" which is available on the left hand side of the window.. please expand it completely before taking the photo..

lostinicem July 31, 2012 05:08

2 Attachment(s)
this one ?
Attachment 14869

Ananthakrishnan July 31, 2012 05:19

1. did you try using the merge option in edit mesh tab along with interface as the required surface.. Pls upload your block and tetin files..
2. try taking the output directly without merging.

lostinicem July 31, 2012 05:31

1 Attachment(s)
1) yes I tried this; actually this is the only option I know :)
2) I am not sure what do you mean..

btw here are all the files I could attach ...:Attachment 14872
tell me if something is missing =)

Ananthakrishnan July 31, 2012 06:05

i am able to merge it at the interface cone but with some errors.. i dont get the "there is only one region " error..
2. Convert to .msh directly without "merging". In case of fluent you can directly tell the existence of this interface in fluent itself. It will be able to solve without a problem..

lostinicem July 31, 2012 07:19

to your second point: ah yes you suggested it already and its good idea for fluent; but the problem is, that I need it for openfoam; though it would be also great to know how to merge it anyway.
for the one region error: I also dont get this, but it seems, that if I do ogrid blocking, icem does something while these steps; the regions are seen as one or made as one; because if its normal block, there are no problems with merging

but nevetheless: thank you for your help !!

EDIT: so I noticed now; the merge function doesnt work, because I put on the down surface a circle ( edge ) to associate it with the surface of the down part of cone, so I can control the ratio and size of the middle part of the mesh. because if I remove it and try to merge it afterwards it works, nevetheless it looks crippled ...

lostinicem August 3, 2012 04:04

so found the error!! if its interesting for someone:
if the error occurs with "you have only 1 region" be sure to check whether on your surface are 2 mesh types ( recalculate or make again unstructure-mesh again)

cyln July 30, 2016 14:00

ICEM Merging Tetra Mesh with Hexa Mesh
5 Attachment(s)

I am working on generating an hybrid mesh (tetra+hexa) and having a problem with merging. I went through 'Merged Tetra-Hexa Mesh in a Hybrid Tube' tutorial many times; however, I couldnt figure out why I had such a problem.

Edit Mesh > Merge Nodes > Merge Meshes. When I follow this procedure, ICEM says 'there is currently only 1 region' and interface meshes remain without a merge.

I believe it can be better to write the main steps I have followed in order to illustrate the problem.

I have a quarter cylinder flow domain and hexa mesh is generated inside a triangular prism which is within the flow domain. You can see it better in topology figure. To obtain such a flow domain, in Design Modeler, I firstly generated a frozen quarter cylinder and then generated a frozen triangular prism. After that, I subtracted the triangular prism from the quarter cylinder and the triangular prism was kept after the boolean operation. Finally the geometry was imported to ICEM. Here, I built the topology with a tolerance of 10 percent of the smallest geometry feature. As you can see in topology figure, there are some blue edges along the interface. I am not sure if it is ok to proceed that way since red curves are usually prefered in color coding.

When it comes to generating the mesh, I created the part names first and named the front, top and back surfaces of the triangular prism as interface.
The other steps are as follows:

Tetra mesh generation:

-create a material point
-define part mesh setup
-generate tetra mesh
-save mesh and close it

Hexa mesh generation:

-create block and edit it (Y block)
-define edge parameters
-open pre-mesh

Merge operation:

-open tetra mesh > mesh exists > merge
-close all the parts except interface
-edit mesh > merge nodes > merge meshes
-merge surface mesh parts > select part 'interface' and Apply

Result: 'there is currently only 1 region',
No Merge.

So what can be the problem?

I will appreciate any help. Thanks in advance.

niablis July 30, 2016 15:23

Hi Cyln,

Try saving the hex mesh blocking, then creating the unstructured tetra mesh. Once the tetra mesh is complete, use File-->Mesh-->Load from Blocking,select your saved hex mesh blocking and opt for merging the mesh. Then run through the process of merging the surface mesh parts.

Hope that helps!!

cyln July 31, 2016 08:38

Thanks Niablis, it works..

All times are GMT -4. The time now is 22:08.