CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Meshing sharp edge (no airfoil) (http://www.cfd-online.com/Forums/ansys-meshing/99929-meshing-sharp-edge-no-airfoil.html)

 jrunsten April 16, 2012 08:37

Meshing sharp edge (no airfoil)

1 Attachment(s)
Hi,

I am looking for some advice regarding meshing of really sharp geometry. Lets concider a geometry like the one in the image below. Is there a way to produce a nice quality mesh in sharp points like these? Which ever way I think about it it always ends up with crappy mesh in that area, and there's no point in starting the meshing before I have a clear idea of how to do it. Any advise?

 Far April 17, 2012 05:53

Divide into two blocks with vertical split and then merge vertices on sharp side and then use Y-blocking.

 Far April 17, 2012 08:04

Quote:
 I am looking for some advice regarding meshing of really sharp geometry
The real question is not how to mesh the really sharp corner, but the question is how to get the angle greater than 18 in such sharp corners.

 jrunsten April 17, 2012 08:17

1 Attachment(s)
Quote:
 Originally Posted by Far (Post 355132) The real question is not how to mesh the really sharp corner, but the question is how to get the angle greater than 18 in such sharp corners.
Thanks. Well yes, those two questions do come hand in hand I would say, and your formulation of the problem is really what I am interested in. I'm guessing some kind of modification to the geometry is needed.

The entire model consists of a tube splitting into two smaller ones (left out if the image), similar to the attached image. So its not as simple as I hinted in the first post. As you can see there are two such sharp corners.

 PSYMN April 17, 2012 08:56

If your geometry has a tight angle, you should assume that your mesh, which is fit to your geometry, will also have a sharp angle. It will even be sharper than that if you want multiple elements across the gap.

If I were to capture this geometry as it is, I would use wedge blocks in those corners.

Another option is to not do that exact geometry shape. Many analysts would argue that the flow thru to the very tip of that cusp is very slight and that leaving it off would not change your results very much... You could trim off the cusp somewhere near the tip. this would change your topology to something with a square end...

But Far just assumed you were asking about Hexa mesh... Is that your goal or are you looking for tetra solutions?

 Far April 17, 2012 08:59

Quote:
 If I were to capture this geometry as it is, I would use wedge blocks in those corners.

Will Wedge block be more efficient than Y-block for the Fluent and CFX?

I just want to confirm that wedge block is formed when block is collapsed at one edge or we need to go to transform block >>>>> wedge block.

 jrunsten April 17, 2012 09:05

Thanks Simon!
Yes I am looking for all hex. The pipes are very long so to save on elements in the streamwise direction I would like to avoid tetras. Slicing the domain in complicated areas like this and mesh with tetra and then sweep the pipes, is what I was thinking at first (in Workbench). But I was curious about doing it all in hexa

 PSYMN April 17, 2012 12:36

Quote:
 Originally Posted by Far (Post 355146) Will Wedge block be more efficient than Y-block for the Fluent and CFX? I just want to confirm that wedge block is formed when block is collapsed at one edge or we need to go to transform block >>>>> wedge block.
Yes, when you have a tight angle like that, a wedge fits more naturally than trying to squish a hex element (even as part of a Ygrid) into that same space.

And yes, you can get a wedge block just by collapsing one face of a block... Hexa is smart enough to convert collapsed end elements to Penta-6 instead of collapsed hexas when you output to unstructured mesh.

 Far April 18, 2012 04:06

Right now I am also working on geometry which has the real sharp corner(water turbine to be precise).

Initially I was working with Y-block in the sharp corner and quality comes out to be > 0.4. When checked in CFX (reconfirmed in ICEM as well) I got the warning about the minimum angle with value of 10.3 deg.

So after the yesterday's discussion I have converted that block into wedge block and now minimum angle comes out to be > 23 but the quality is 0.1 for the cells in that corner. Rest of the mesh has quality greater than 0.4.

Now I have no solution for this situation. Any suggestion?

 PSYMN April 18, 2012 21:42

Yea, just don't worry about it...

The problem is that the quality metric for prisms (penta-6) is really harsh. It takes the element and divides it into 3 tetras and then computes the quality of each and reports the worst.

As a result of the totally different computation, you can't really compare prism quality with other elements.

I have pushed several times to have this metric changed to not appear so negative for our prisms, but some older customers have setup all sorts of scripts based on this current formulation and don't want us to change anything there...

Try it in your solver, it won't mind those prisms a bit.

 dnsantosh April 21, 2012 02:14

NACA66012 mesh problem

hi
am trying to do tri mesh on naca66012 symmetry airfoil wing surface, because of its sharp trailing edge its not capturing mesh correctly, if i make it blunt or give some radius by making sharp to curve,if make it is there any changes will occur in results.?
thank you

 dnsantosh April 24, 2012 03:06

is it possible..?

is it possible to generate only prism on UAV by using ICEMCFD and generate tera on other software and merge them ,,,,if possible how to generate ony prisms in ICEM

thank you

 Far April 24, 2012 04:21

shall meshing and then prism

This is the bottom up approach. First make the shell meshing and then set the parameters for prism and compute mesh. you are done.

 dnsantosh April 24, 2012 04:35

generating pyramids with prisms