CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] o grid blocking strategy? (

icemaniac178 February 11, 2011 01:02

o grid blocking strategy?
2 Attachment(s)
i have some problem regarding about using the automatic o grid blocking option technique in ICEM CFD.
i have a very thin 1 mm geometry (pictures attached) and i try to mesh it using the hexa blocking method.
my question is

1) is it the mesh generation will be satisfy if i just use the ordinary rectangular block around the geometry and associate the edge to the curve or

2) use the o grid by doing a rectangular block around the geometry then o grid the block and then associate the edge to surface and curve accordingly.

FYI this geometry is attached to other geometries. so the blocking could be more complicated when is used the o grid generation.

PSYMN February 11, 2011 12:32

Ogrid or HGrid?
If you map an HGRID block to this shape, you will have "corner" hexa elements with 6 nodes projecting to the curved walls. (3 nodes for quads on the corners of the flat face) This opens up the angle (angle approaches 180 degrees as the mesh gets finer and finer).

If you plan on having a course mesh, the quality may be OK (175 degrees), also, some solvers are more tolerant of wide angle elements.

If you are meshing around this shape, then the OGrid is not as important.

But generally speaking, if you are meshing inside this sort of oblong shape, an OGrid should be used to prevent bad elements in the corners, give you some independence from mesh refinement and to better capture viscous flow along the walls.

icemaniac178 February 17, 2011 04:06

H grid? sorry i only understand when u said about o grid or c grid. i already tried the o grid inside this geometry but the pre mesh does not give me any better mesh distribution. is it means that i have to play around with the edge parameters?
i have no experience in setting up the edge parameters nicely. do you have some tips how to start using the edge parameters correctly?

PSYMN February 17, 2011 12:31

An H grid is just a regular mapped grid, squares in a box.

An ogrid inside this would certainly give a good mesh, so you probably just need to work on how you placed it, etc. If you had posted a pic of your OGrid, I could comment...

Try some tutorials such as the 2D Car, the 3D Pipe junction to get started.

Most hexa tutorials will give you practice with Ogrids and Edge Distribution. These can be found in the customer portal, but I think I uploaded the 11.0 set and put the link on CFD-Online several times if you wanted to find it.

icemaniac178 February 17, 2011 13:34

thank you mr simon,
i think i`m little bit confuse about about the o grid inside and the out side the geometry. actually the o grid that i make is outside the geometry. so is it possible when i am going to do the second o grid on the center block of the first o grid.
in other words o grid block at center block of the first o grid. phew! it was so challenging but it is worth it to get the experience in hexa mesh blocking.

i have tutorial ICEM CFD for ansys 12.1 version. it concentrate more on blocking rather than the edge parameter. so i will look for 11.0 version.

PSYMN February 17, 2011 22:04

it is very flexible...
Yes, you can put Ogrids inside Ogrids, etc. No problem...

maalan August 19, 2012 15:49


Originally Posted by PSYMN (Post 295822)
Yes, you can put Ogrids inside Ogrids, etc. No problem...

Hi Simon!! Hope you're fine!!

I want to do a mesh with an ahmed body inside the tunnel and then generating an o-grid around the car and then another over the first one and if posible a 3rd one which ends in the tunnel edges. I am able to do just the first thing, could you tell me briefly how to do the other ones? I saw this idea in a document you posted hear and it was applied to a f1 car.

Once the mesh is done, I'd like to mesh just the block after the body with an unstructured mesh, i.e., I want an hybrid mesh. I have investigated on it a lot but I don't get it yet.

thank you so much for your help and hope to hear from you soon!!

PSYMN August 19, 2012 20:50

I am not sure what you are asking about the Ogrids.

As for the Hybrid mesh, there are lots of ways to get that done. If you are starting from a blocked mesh like this and just want to remesh a volume, you can just delete all the mesh in an area and then run delaunay tetra with the option to fill holes in mesh. Instead of replacing the hexa mesh, it will only look for holes and fill those...

maalan August 19, 2012 21:11

thank you so much for your reply, simon!! i will try the unstructured mesh issue as you told me!

about ogrids i would like to get something like this:

specifically the right picture since the left one i got it.

im sure you will know how to do it!!

thanks a lot!

maalan August 20, 2012 06:28

And this is the document you uploaded:

I would like to get a block structure like that of the page 1-59.

Finally, how I can delete the mesh inside a block?? I try edit mesh-->delete elements, but i don't find the option to select blocks...

I hope with a little more of help I could get my mesh! Thank you!

rsaha March 19, 2013 18:56

1 Attachment(s)

Please take a look at the picture below. How can I create O-grid around the tube without the o-grid propagating to both sides?

Thank you.

Far March 19, 2013 19:28

Both sides? Can you be more specific? To answer your question in general, yes you can control behaviour of O-grid as required.

diamondx March 19, 2013 21:02

it's impossible... this is how structured mesh works

PSYMN March 19, 2013 21:10

diamondx is right in that it needs to end some how... But it is possible to have the Ogrid not extend all the way to walls. Instead, it would wrap around the ends of the pipe.

The process is to slice for the ends of the segment and then also slice one more time part way between the end of the segment and the wall... when you place your Ogrid, select the block the pipe is in, plus the blocks on either side. Do not select any "faces". Apply.

rsaha March 20, 2013 00:18

1 Attachment(s)
Thank you very much for your replies.

@Simon, your instructions are very helpful. Here is a snapshot of what I come up with.

All times are GMT -4. The time now is 04:29.