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/)
-   -   Gambit Problem Size Function (http://www.cfd-online.com/Forums/ansys-meshing/87095-gambit-problem-size-function.html)

El Wastl April 11, 2011 10:56

Gambit Problem Size Function
 
Hello

Im working on a numerical analisys of a turbine blade (2D). I have already done some simulations in fluent with my gambit-made 20.000 cells grid. In Fluent I adapted this grid with the dynamic refine function, so I had a almost perfectly refined grid apart from the trailing edge, where you could see 20 edges put together forming the trailing edge instead of the semicircle as expected.

Therefor I tried to change my grid in Gambit meshing the turbine blade with an interval size of 0.001 and creating a size funcion (meshed type / 1.1 growth factor / 3 max.). Instead of showing me cells starting with the interval size of 0.001, Gambit shows me a grid in the trailing edge that doesn't fulfil this criterion. (Only in a really unimportant part of the profile this criterion is fulfilled)

It would be very nice if somebody can help me with my problem. Thank you in advance.

El Wastl

-mAx- April 12, 2011 02:43

if you attached a picture I cannot see anything

El Wastl April 12, 2011 04:39

1 Attachment(s)
Of course... here it is ;)
In the image you can see the meshed trailing edge, but the grid, generated with the meshed-type size function, doesn't fulfil the given distances of the first row of cells....

-mAx- April 12, 2011 07:44

delete your mesh, also your size function.
Create a new size function (type fixed)
Select your edge as source, give 0.001 as start size, 1.1 as growth, and 3 as max size.

El Wastl April 13, 2011 10:16

1 Attachment(s)
Thank you for your help.
Now I have two size function:

sf1:
source all edges of the profil
attachment: face (flow field)

sf2:
source: trailing edge semi circle
attachment: face (flow field)

With these functions I would like to have a grid with small cells next two the edges of my profil. But why Gambit makes me a non symetric grid like you can see it in the image. Info: The edges of the profil are NURBS! Are there problems with meshing NuRBS?





But there's n other problem:

-mAx- April 14, 2011 07:23

do you have only one face?
If you want a full symetric mesh around your trailing edge you can enforce gambit to link surface mesh together.
But you will need to split your surface into smaller surfaces

El Wastl April 14, 2011 09:13

Thank you very much, at last I can really improve my grid!!! I tried it with an extra face around my trailing edge and it works ;)

Can you recommend me adding a certain number of faces (and their form!!) all round my profile?

The face linking function, you mentioned, is it called "LInk face meshes" in Gambit? Why does Gambit want to know sth about vertices?

-mAx- April 14, 2011 09:41

1 Attachment(s)
you are free to decompose your surface as you want.(for unstructured mesh)
But it has to be symmetric.
For linking the surface mesh, you need to give one vertice for the first surface, and another one for the linked surface.
In your case you will give the same vertice.
http://my.fit.edu/itresources/manual...mg0303.htm#689
Attachment 7265

El Wastl April 14, 2011 09:59

perfect...thank you for your helful explaination

El Wastl April 15, 2011 05:51

1 Attachment(s)
Hello Max
Thank you for your tips. Already I have obtained some results with my old "bad" grid (only one face) in FLuent, but now I want to improve my grid. In my case I cannot create symetric subfaces because of the "flow channel"??!! You can see my geometry in the attachement
Thank you in advance!
Basti

-mAx- April 15, 2011 05:55

what is your question?

El Wastl April 15, 2011 06:18

Quote:

Originally Posted by -mAx- (Post 303623)
you are free to decompose your surface as you want.(for unstructured mesh)
But it has to be symmetric.
Attachment 7265

In my case I can't create symmetric subfaces because of the channel geometry (periodic boundaries). But nevertheless I would like to have a fine and symmetric grid all round the trailing edge! Do you have a recommendation? Thanks

-mAx- April 15, 2011 07:17

1 Attachment(s)
Then create a subdomain in the wake of trailing edge which will be symmetric in respect of your yellow edges-chain.
In my example, I meshed the subdomains very quick with pave, but you can more decompose those sufaces for having a nice quad mesh
I also attached a BL at the trailing edge, showing you that you can also include one.
Attachment 7282


All times are GMT -4. The time now is 07:42.