CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [GAMBIT] Help needed with meshing! (

fluentnoob June 3, 2009 06:27

Help needed with meshing!
Hey everyone!

I have been trying my a** off trying to mesh this geometry (

But to no avail. I tried meshing the full thing with fixed size function attached to all the strut edges. Didnt work. I tried assigning separate size functions. "Curavature" for the upper and lower bodies and "Fixed" for the struts. Still cant do it properly. It always shows that a large number of cells have high skewness and also, that there is one inverted volume.

Can anyone please please help me out with meshing this? How should I approach this problem.

I am using Gambit for meshing purposes.

Thank you in advance!!

-mAx- June 4, 2009 02:38

Check the warning in Gambit.
If you have, then check the geometry in the mentionned surface.
Typically this is small edge/angle problems, which you can solve with merge tools

fluentnoob June 4, 2009 04:18

Hey, Thanks Max. I shall do this.

One more thing. I have meshed the faces separately. Should the mesh size be same at the edges where the faces meet? Like face x has fixed triangular pave meshing starting from edge a with size .05, while face y has uniform triangular pave meshing starting from edge a with size 0.2. Should these two size be same at edge a?
And if I mesh the faces separately, am I supposed to link them after meshing is done, before attaching the 3D boundary layer (aspect ratio(last), internal continuity) ?

Thanks for the help, in advance :)

-mAx- June 4, 2009 05:10

that could be the source of your issue:
if you have an element with fixed size 0.05 and the other is 0.2, then it coud generate cells with small angle.
I would create a size function for controlling the gradient between 0.05 and 0.2.

fluentnoob June 4, 2009 09:48

hello again!

I am stuck again. I tried what you said. It worked fine too. But still the mesh quality is poor in one particular part of the geometry. I can post a photograph of the part if you want me to. It shows 2 highly skewed elements in the part. If I remesh the part with a smaller mesh sizes, the number of skewed cell increases, and some inverted cells are also created. :( Can you please suggest what should I do with that part?

The picture of the part is:

You can probably see the red spots marked. Theres a fourth one too (symmetric, hidden in the picture). All the highly skewed cells are distributed around these four vertices. What should I do?

Thank you have been a great help.

fluentnoob June 4, 2009 09:49

-mAx- June 4, 2009 13:09

go to the examine mesh tool, zoom the area you mentionned and send a picture

fluentnoob June 4, 2009 13:25

ok..I'll post it in a couple of minutes once my Gambit.exe stops crashing. :( :(

I was able to reduce the skewness a little by remeshing and altering the boundary layer. Though not by much.

Also, after I was done donig that, I tried to mesh the volume using tetrahedral elements/TGrid..and the meshing fails while refining the mesh. It says that couldnt allocate 61 Mb of Dynamic memory. Does it have something to do with RAM? or Paging File Size? What should I do now?

I'll post the image online in a few minutes as soon as I get gambit working.

Thanks a lot. I reallly appreciate your patience with me. :)

fluentnoob June 5, 2009 05:51

I am sorry. Gambit just wouldnt work. Kept crashing. So I had to reinstall it. Uploaded the picture. The link is:

What should I do to reduce the skew at these points?

-mAx- June 5, 2009 06:08

i don't see anything
same picture but closer
I think you have 2 close vertex

fluentnoob June 5, 2009 06:17

The highly skewed cells are shown in yellow in the picture. Their number is really low, about .03%, but probably you can see them. I am sorry for the picture quality. :|

-mAx- June 5, 2009 06:48

Yes I can see the skewed cells, but the view is to far.
Make the same picture but with focus on one of the 4 high skewed cell's area

fluentnoob June 5, 2009 07:08

This is the picture with a close-up of the skewed cells. I am sorry for the last few pictures. I had misunderstood what you were trying to say..

Thank you. :)

-mAx- June 5, 2009 07:22

:D still to far
check the picture

fluentnoob June 5, 2009 08:28

Hey, thanks. Got the picture. I am uploading pictures of the point with full zoom in. I did check. There is only one vertex over there, so that I think is not the problem.

Thanks again.

P.S.: if the pictures are a too small, theres a zoom button also. Its not very useful, but still upto the original image size it can go, I guess.

-mAx- June 5, 2009 08:39

ok, then remesh your domain without the BL.
If you don't have those skewed cells, then your problem lies on the BL

fluentnoob June 5, 2009 09:11

I just remeshed all the faces again, and there was no skewness reported. So, it's got to be something wrong with the B.L. Thanks :).

Any suggestions, as to how could I make the boundary layer better, would be welcome. :|

-mAx- June 5, 2009 09:57

remesh the volume, not only the surfaces, and recheck the skewness

fluentnoob June 5, 2009 10:08

That is another obstacle. When I try to mesh the volume, it says that it cannot allocate enough dynamic memory, and the meshing fails. :( :(

-mAx- June 5, 2009 10:27

you have to divide (split) your domain. Meshing volume one after one will ask less memory

All times are GMT -4. The time now is 10:29.