[ICEM] Problem with unstructured meshes

eRzBeNgEl · January 20, 2011, 04:28

Hi

As you can see in picture 1+ picture 2 in the attachement I am working on a unsymetric (there is a opening on the side) cooler + housing. My idea was to create Interfaces ( two Inlets (Top and Bottom) and one Outlet on the side). The problem is that Icem just meshed the inner part of the green housing. I need also the whole outer part meshed up (Picture 3 - Big Cylinder around the cooler housing and the green housing itself).

What did I do wrong?

Thanks for help

BrolY · January 20, 2011, 05:15

Have you created bodies everywhere you should?

About your 3rd picture, if you close your small cylinder with interfaces, ICEM won't mesh the big cylinder if you didn't tell him to do it (by creating bodies). And if the fluid can't move from the big cylinder to the small cylinder, why do you need the big one ?

For you 2nd picture, the problem is linked with this body issue I explained before.

eRzBeNgEl · January 20, 2011, 06:49

@BrolY

Thanks for your help. There were not enough material points. That was the reason.

Regarding your other question. My task is to simulate the fluid flow from the Big Cylinder, over the blades to the outlet on the side. There are 2 blade sides, one on the top, the other on the bottom. So I am havin two Fluid Inlets (the pink circle is one) - on top and bottom. On the side you have the oulet Area

My first model was a rotating -counter rotating model (with the big cylinder) - but it failed, cause the housing is not closed.

Then my next idea is to work with interfaces. I have never done this before (CFD Newbie) and still working on it. Maybe u have some ideas to improve or find a model ?:-)

Thanks for help

BrolY · January 20, 2011, 07:39

I think ICEM "understands" what you named "Inlets" as a wall.

Inlets or Outlets should be the entry(ies) or exit(s) of your domain. Do not create those surface inside your domain. Or it would prevent the flow going from your big cylinder to your small cylinder, which is not what your are looking for if I'm right.

Interfaces should be used if you want to break your model in two parts (or more), and then merge them together.

From your pictures, I would say you need only one material point (body) in your fluid domain.

PSYMN · January 21, 2011, 22:22

I would agree with Alexandre if the model were stationary... But if you are spinning the disk (like a hockey puck), you need those interface walls to separate the spinning cell zone from the surrounding one and to create a nice smooth interface that you can spin...

Just add the second material point and consider that inlet and outlet as cell zone interface (you can keep the name, it won't do any harm).

eRzBeNgEl · January 24, 2011, 07:27

Thanks a lot for your help.....

PSYMN was right. I used the "Inlet and Oulet surfaces as interface connection domains and my model is working now, and it also worked with one material point. So everything is fine now :-)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ICEM] Problem vt unstructured meshing	raghav	ANSYS Meshing & Geometry	1	October 8, 2010 12:21
Divergence problem on different meshes	Harry	Main CFD Forum	2	September 26, 2006 00:23
artifical dissipation on unstructured meshes	Frederic	Main CFD Forum	0	October 31, 2001 17:42
unstructured grid	sreekanth	Main CFD Forum	1	August 6, 2001 15:09
Finite Volume Approach For Unstructured grid	APURVA SHUKLA	Main CFD Forum	2	March 15, 2001 00:14

January 20, 2011, 05:15		#2
BrolY Senior Member AB Join Date: Sep 2009 Location: France Posts: 323 Rep Power: 21	Have you created bodies everywhere you should? About your 3rd picture, if you close your small cylinder with interfaces, ICEM won't mesh the big cylinder if you didn't tell him to do it (by creating bodies). And if the fluid can't move from the big cylinder to the small cylinder, why do you need the big one ? For you 2nd picture, the problem is linked with this body issue I explained before.

January 20, 2011, 06:49		#3
eRzBeNgEl Senior Member Join Date: Dec 2010 Posts: 135 Rep Power: 15	@BrolY Thanks for your help. There were not enough material points. That was the reason. Regarding your other question. My task is to simulate the fluid flow from the Big Cylinder, over the blades to the outlet on the side. There are 2 blade sides, one on the top, the other on the bottom. So I am havin two Fluid Inlets (the pink circle is one) - on top and bottom. On the side you have the oulet Area My first model was a rotating -counter rotating model (with the big cylinder) - but it failed, cause the housing is not closed. Then my next idea is to work with interfaces. I have never done this before (CFD Newbie) and still working on it. Maybe u have some ideas to improve or find a model ?:-) Thanks for help

January 20, 2011, 07:39		#4
BrolY Senior Member AB Join Date: Sep 2009 Location: France Posts: 323 Rep Power: 21	I think ICEM "understands" what you named "Inlets" as a wall. Inlets or Outlets should be the entry(ies) or exit(s) of your domain. Do not create those surface inside your domain. Or it would prevent the flow going from your big cylinder to your small cylinder, which is not what your are looking for if I'm right. Interfaces should be used if you want to break your model in two parts (or more), and then merge them together. From your pictures, I would say you need only one material point (body) in your fluid domain.

January 21, 2011, 22:22		#5
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	I would agree with Alexandre if the model were stationary... But if you are spinning the disk (like a hockey puck), you need those interface walls to separate the spinning cell zone from the surrounding one and to create a nice smooth interface that you can spin... Just add the second material point and consider that inlet and outlet as cell zone interface (you can keep the name, it won't do any harm).

January 24, 2011, 07:27		#6
eRzBeNgEl Senior Member Join Date: Dec 2010 Posts: 135 Rep Power: 15	Thanks a lot for your help..... PSYMN was right. I used the "Inlet and Oulet surfaces as interface connection domains and my model is working now, and it also worked with one material point. So everything is fine now :-)