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/)
-   -   [ICEM] Dealing with Singularity..! (http://www.cfd-online.com/Forums/ansys-meshing/81675-dealing-singularity.html)

JOKER November 3, 2010 10:48

Dealing with Singularity..!
 
Hy Everyone !

I have gain good experience of ICEM for both 2D and 3D geometries. But I m still finding trouble with singularities.:(
In ICEM, I resolve singularities by merging vortices but in outer domain edges remain apart, making confusions. I don't understand if collapsing outer blocks or merging them with neighbors will work.

i have littile knowledge of gridgen where we just make an eage or vortix singular and complete blocks. I dont know if ICEM works that way.

Little explanation will be appreciated!:)

PSYMN November 3, 2010 12:14

Top down vs. Bottom up.
 
I suspect you are not using Hexa correctly...

Other tools, like Gridgen, follow a bottom up approach where you create verts, edges, faces and eventually blocks.

ICEM CFD starts with a top down approach where you start with a single block and subdivide to get your domain. ICEM CFD is more like a sculptor than a brick layer.

If you try to use ICEM CFD the same way that you would use Gridgen, you will not be successful.

Ideally, a single training day would give you a great head start into the process. If that is not an option, you should certainly go thru some tutorials.

If you have done the tutorials, and you think you are using Hexa correctly, then perhaps you could use some images to better illustrate your issue with "singularities". Your problem description is so vague that I am not even certain you are using Hexa...

JOKER November 3, 2010 15:58

5 Attachment(s)
Hello,

May be i m confusing the things as i always does..:D


Here I generate a quick problem ...!

It is a wing-fuselage attachment (1,2), i put blocking as shown (3,4) ; split the block at wing fuselage intersection and associated the edges to fuselage, but now don't understand how to fit blocking to wing part and also manage sharp TE.

I did it by merging vortices but it generate edges in outer fields (5,6,7), although i did not check the pre-mesh I know this is not a good blocking practice...!

JOKER November 3, 2010 16:03

2 Attachment(s)
Attachment to previous post

dst November 4, 2010 09:32

Hi,

general approaches of airfoil (as a specific case of a wing) blocking can be found here:
http://www.youtube.com/watch?v=tYrbScUH9RE

The exact blocking scheme will depend on your real problem (geometry). Normally, for wing-bodies O- C- E- block structures can be used. Unfortunately, not for all of the cases...

JOKER November 4, 2010 14:15

I have done 2D AIRFOIL tutorial..very helpful.
I have done almost all manual tutorials..:o
Even done Boeing model included in training.. much similar like wingbody tutorial..!
.
.
Point is I don't understand what is going on when creating Triangular and other complex faces in 3D ..similarly in Above model i found difficulty because fuselage and wing root share same edges, now split at TE...

someone give a solution to above problem please..! may be it help me better understand ICEM

E- blocking..? i dont know that, anyone please suggest some good reference book on blocked meshing..:confused:

dst November 5, 2010 03:41

Can you post your geometry (project) directly from ICEM here?

JOKER November 5, 2010 08:57

File size is larger then .zip option..is there some other way..?

dst November 5, 2010 10:11

Save just a geometry *.tin file from ICEM. If it is still bigger than 97Kb, you could put it to some file-sharing system.

PSYMN November 6, 2010 16:04

Generic suggestion...
 
For a typical sharp trailing edge wing in a box or extending from a fuselage I would say...

1) You should extend all your splits thru the blocking... It may be a little harder on your eyes, but it makes the next step easier... (and you can always merge blocks again later if you want).

2) You need to collapse the thin blocks behind your wing... But the thing about collapse is that you must do it out to the extents... That means you must collapse from the trailing edge back to the outlet and sideways into the fuselage and out to the far field... This means your "wedge" block will extend from the fuselage to the far field. Part will be in the wing and the rest will be past the wing tip.

3) Once you collapse that block, further steps like creating a CGrid will be much harder, so look after that sort of prep before the collapse (in the same way that I did it for the online demo).

PSYMN November 6, 2010 16:21

Specific Suggestion
 
1 Attachment(s)
But your model is not typical because of the shape of your "fuselage". It starts out sharing the index with the collapsed block, but then doesn't collapse. This makes it trickier...

So, you have a choices...

Do you really need that fuselage to be that shape or did you just think it would help with the modeling? It doesn't help. It makes things more difficult by creating additional triangular regions, so unless you need it that way... fix the geometry and go with the top down approach.

If you do actually need that step down shape... It can be done. Your model still has an element of simplicity... Topologically speaking, it is just a sweep from the symmetry plane to the far field. Look at it in 2D... You could even project all the wing root curves to the symmetry plane so you can see it as a 2D geometry... The blocking is relatively simple then. It is basically a box, with a CGrid in it (arround the wing shape and extending backward). Within the CGrid you have a box representing the round half of the airfoil before it separates into the wedge. Behind that you would have two boxes, each split diagonally for the wing shape and ending at the trailing edge. Behind those, you have two more boxes representing the trailing edge blocks still on the fuselage and behind that, two more long blocks back to the far field. Then go back and put an Ogrid inside the first block at the front of the wing to improve the quality past the wing tip (if you have flow in that region).

Here is a rough 2 minute sketch of the 2D blocking at the symmetry plane...
Attachment 5301


Then with the 2D blocking done (create these blocks manually from verts and points or create big elements and load blocking from unstructured mesh)), you can extrude these toward the far field. THen just add two splits (wing root, wing tip). Then delete blocks that are inside your wing and fuselage. Then associate blocking with geometry, setup sizes, etc.

JOKER November 8, 2010 13:30

Thanks for a brief reply..!

so it is no longer top to bottom approach..
what if I want to starts with 3D blocking and then do some portions using 2D blocking options..
is it possible.?

PSYMN November 8, 2010 13:47

No, it doesn't allow 2D and 3D blocking in the same file. You can start with 2D and convert to 3D or vice versa.

It can all be done with 3D blocking (collapsing blocks into those wedges is possible). I just think it is more work than creating the 2D blocking and extruding it.

When creating 2D blocks, you can change the type of block you create to "degenerate" and then just pick 3 verts to get your wedge...

JOKER November 8, 2010 17:13

2 Attachment(s)
Hello, Everyone..!

I just made a quick mesh using above procedure by PSYMN, here are pics. I have not set mesh parameters yet.

PSYMN November 9, 2010 12:59

Great.
 
Good job with the blocking topology :cool:

JOKER November 10, 2010 08:57

Next step
 
Hy Everyone..!

So bad if we can't do 3D blocking and then 2D blocking simultaneously..

Now how to start complex geometry..If start with 3D blocking and splitting and down the way realized it is bit difficult to pursue with splitting or impossible in some cases..then what are the options..

Otherwise starts with 2D blocking every time..if so then which would be easier

1) either starts 2D surface blocking (If geometry has hole or discontinuity then blocking will not be smooth..)

2) or starts a planner 2D block and fit it to one surface, and then made rest of the blocking using vertices and points

I think I have done with automated 3D blocking and this chapter is closed..:D


All times are GMT -4. The time now is 21:35.