CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ANSYS Meshing] To ask a question about O-grid generation (http://www.cfd-online.com/Forums/ansys-meshing/104435-ask-question-about-o-grid-generation.html)

lnk July 9, 2012 14:50

Why don't we use domain interface to solve this problem?
 
2 Attachment(s)
Hi,

May I ask you a question about O-grid generation? :confused: Here is a picture. I'd like to generate a picture like B. But since I can't rotate my block, I always obtain A. How to solve this problem?

Best regards and many thanks,
lnk

Far July 9, 2012 15:14

what u want to do?

Far July 9, 2012 15:49

A=B. So it does not make any difference.

lnk July 9, 2012 16:20

4 Attachment(s)
Quote:

Originally Posted by Far (Post 370529)
A=B. So it does not make any difference.

Here is the problem. The geometry I'd like to mesh is a connection of a tube with a cube. If I have the O-grid like A, I can't correct the bad quality meshes at the connection part of the tube and cube. So I try to rotate the O-grid to solve this problem.:confused:

Best regards and many thanks,
lnk

Far July 10, 2012 02:31

This is easy case for ICEM Hexa. If you can share your files, we can give a quick try.

BrolY July 10, 2012 04:38

just create points at the 2 exits of your model and associate the vertices of your o-grid to modifiy its shape. Then align the internal vertices of your o-grid with the vertices of the 2 exits of your model and you are all good !

lnk July 10, 2012 06:30

1 Attachment(s)
Quote:

Originally Posted by Far (Post 370564)
This is easy case for ICEM Hexa. If you can share your files, we can give a quick try.

Here is my file. Thanks you very much!

Best,
lnk

lnk July 10, 2012 08:10

Quote:

Originally Posted by BrolY (Post 370592)
just create points at the 2 exits of your model and associate the vertices of your o-grid to modifiy its shape. Then align the internal vertices of your o-grid with the vertices of the 2 exits of your model and you are all good !

What do you mean by 'exits'? Won't that change my geometry?:confused:

diamondx July 10, 2012 23:50

Far, Broly. this geometry is not easy...very tricky... i've been fighting with it for more than 4 hours now.i thought may be li's strategy for blocking was incorrect, so i erased his blocking and tried to came up with a better one using bottom-up approach. I couldn't, then i came back to li's blocking, realized that he can never get a good quality because he needs ygrid when the arc's circle meets the tangent line, alse noticed that geometry needs some cleanup, there are additional surfaces inside (tube), maybe that's why. I don't understand why and how li did that blocking... i need a break. i hope i can wake up the morning and find your approaches...
@li i'll come back to your geometry tomorrow to finish it... i saw another thread where you're asking about ygrid, you have to merge before, and select the triangular shape, but i just don't know how you can do that with your blocking...

diamondx July 10, 2012 23:52

i'm going on a vacation on Sunday... like you said FAR may be this is easy geometry, i just need vacation...

BrolY July 11, 2012 04:47

Those kind of geometry are hard because your circle geometry meet the corner of a rectangular shape. So I think you can avoid bad elements at the intersection of those 2 shapes.
I don't have much type to take a look at the geometry, will try this week :S

BrolY July 11, 2012 05:40

1 Attachment(s)
Ok finally, I took some time to have a look.
Attached an example of blocking. you only need to do the association and maybe improve the position of some vertices.

You can't avoid some bad elements between the corner of your rectangle and the print of your tube, but 1/4 of O-grid help to handle it.

lnk July 11, 2012 06:48

2 Attachment(s)
Quote:

Originally Posted by BrolY (Post 370767)
Ok finally, I took some time to have a look.
Attached an example of blocking. you only need to do the association and maybe improve the position of some vertices.

You can't avoid some bad elements between the corner of your rectangle and the print of your tube, but 1/4 of O-grid help to handle it.


I tried that but I can't improve the quality because of the same reason. The quality at the attachment is the best I can do. :o

BrolY July 11, 2012 07:28

You can incraese the number of nodes, move the vertices in order to optmize the blocking. But you have to know that at some point, you won't be able to improve the mesh because of your geometry.

lnk July 11, 2012 07:41

Quote:

Originally Posted by BrolY (Post 370789)
You can incraese the number of nodes, move the vertices in order to optmize the blocking. But you have to know that at some point, you won't be able to improve the mesh because of your geometry.

Yes. That's the problem. I can't make all my meshes have quality over 0.5.

BrolY July 11, 2012 08:02

Yep, but your biggest issue is not about the determinant, but about the min angle ... You can get a min determinant around 0.2, which is not that bad.
But there are few cells which min angle is between 0 and 9, which could crash your solver ...
If you work hard and moving the vertices of the 1/4 of O-grid, maybe you could improve a little bit your mesh ...

Or maybe someone else could provide another blocking ?

Have fun ;)

flotus1 July 11, 2012 10:21

You have to ask yourself one question:

Is it important that the geometric feature (the sharp angle) is represented by the mesh?

Or wouldn't the flow field be almost exactly the same if the sharp angle is not represented by the mesh? In this case: change the geometry to allow better meshing.

diamondx July 11, 2012 10:30

The blocking is wrong...
@stief, that's exactly what i was thinking. you are right. Why do we have to follow the geometry ?
how about something like this
https://dl.dropbox.com/u/35161486/manifold.JPG
someone correct if i'm wrong...

BrolY July 11, 2012 10:59

1 Attachment(s)
I correct some errors in the blocking (see attached).

And I fully agree with flotus ;)

lnk July 11, 2012 11:25

2 Attachment(s)
Quote:

Originally Posted by diamondx (Post 370848)
The blocking is wrong...
@stief, that's exactly what i was thinking. you are right. Why do we have to follow the geometry ?
how about something like this
https://dl.dropbox.com/u/35161486/manifold.JPG
someone correct if i'm wrong...


I did in this way at the first time. It didn't work so I changed to make the mesh follow the geometry.

The pipe is not inside the cube. It goes through the cube. If the mesh doesn't follow the geometry, the mesh will have problem at the connection part too which is, I think, also difficult to correct. You can see my attachments. It is even clear at the mesh picture.


Best,
lnk


All times are GMT -4. The time now is 05:01.