CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] Pipebranch hexa meshing (

darookie July 21, 2009 10:02

Pipebranch hexa meshing
I have a problem to mesh a pipe branch. The pipes have the same diameter and should be meshed with an O-Grid. After creating the intersection between both faces, I don't know in which sequence i should create the o grids?
Even if it is a usual and typical problem, the details to gain a good mesh are tough.
Thank you,

PSYMN July 21, 2009 17:01

All at once?
With ICEM CFD Hexa, we would usually ogrid it all at once. Select all the blocks and put faces on the inlets and outlets.

The best reason for doing it in stages is if you want the ogrids (and mesh) to criss cross.

Maybe an image would help us understand how you have blocked this ypipe.

darookie July 23, 2009 04:47

Thanks for the fast reply
Here is a picture of the geometry
and here is my solution for blocking with an o-grid.
I first did an ogrid for the whole geometry and then cutted blocks for the pipes and applied an ogrid for each pipe separately
I will try it out to mesh it all at once

PSYMN July 23, 2009 08:57

That is a T-pipe actually. They are easier than a Y pipe (no messy collapsing, etc.).

I could quickly block it for you (5 or 10 minutes) if you sent me your tetin file (you could send it privately).

I would start with an OGrid to capture the concentric cylinders and then just Hblock out the side pipe. After projecting everything to the pipe ends, I would run an oGrid thru the whole thing.

While I would OGrid the T-Pipe section all at once, it may be a good idea (from a physics capture point of view) to boundary layer Ogrid the tank first and then punch thru that with a pipe ogrid that includes some blocks within the core of the tank. I might also do things a little differently if the flow was going into the tank vs out of it.

PSYMN July 24, 2009 12:12

Bad associations...
1 Attachment(s)
I will try to block it between now and Monday (should be quicker than it takes to write up the explanation of what I did).

As for your current blocking, it should give you positive determinants if you fix the associations in the t-section area. The three sides away from the T should just be surface projected. And don't forget to fix that curve projected vertex (should be surface projected...)

I will make a more efficient and physics capturing blocking for you also... Basically, I will do one ogrid for the large pipe and a second for the smaller pipe that penetrates only a short distance into the large pipe (and then diffuses).

PSYMN July 27, 2009 17:43

Icem cfd...
4 Attachment(s)
Sorry, I don't have time for much of a write up, but I will post pics...

I didn't end up doing it the way I planned (no Boundary layer ogrid in the combustion chamber), but I think this should work well since the flow will not be running along the end of the large pipe anyway...

I accidentally messed up the replay file, but decided to send it anyway in hopes that it will help you understand what I did. Also look at the blocking file separately, perhaps you can figure it out from thatů These both should work with your starting tetin file...

This was only about ten minutes to get the blocking, but I didn't spend any time improving the quality or edge distributions, etc. Hopefully you can do that. I would guess you could easily put another hour into matching edges and distributions, etc. to get a nice smooth mesh.



darookie July 29, 2009 05:49

Wow, thanks a lot the blocking is very good. You really helped me a lot. I can reproduce every step except this instruction: ic_hex_create_composite {tmp00.2 CURVE00} What does it mean?
Thanks for the extraordinary help!!!

PSYMN July 29, 2009 09:53

This is the command for grouping curves...

I didn't do it explicitly, but probably selected one edge (or multiple edges) to associate to more than one curve. Before that can happen, the code automatically combines the curves into a composite curve and then performs the association...

{tmp00.2 CURVE00} were the names of the two curves that were combined.


All times are GMT -4. The time now is 00:36.