plan of approach required
 

i have attached the jpeg of my 3D wing. it is completely asymmetric wing. I would like to know a crude plan of approach for the blocking. I personally think unstructured mesh would be better suited. please give you comments upon this idea as well..

I have pasted the pictures for quick reference..

http://img96.imageshack.us/img96/1453/topview2r.jpg

http://img688.imageshack.us/img688/6220/topviewr.jpg

Hello Ananthakrishnan,

Without being an expert in any way, I would suggest you follow a similar blocking strategy to the sphere meshing tutorial of ICEM. In that tutorial, if I remember correctly, a block is placed in the middle of the sphere with an O-grid placed around it. You could do something similar here, only with a rather irregularly shaped central block (very thin most probably). Obviously the main problem would be mesh quality in the sharp corners at the top and left sides of your geometry.

Browsing through the ICEM tutorials may give you a better idea too. :)

I agree, however, that depending on the purpose of your simulation, it may be a better idea to use tetra.

Good luck!

Little more info
 

Hi Ananthakrishnan,

Just came across your thread. Are you still interested in block-mesh for this geometry? I may be able to share some ideas here.

If you could post the following images, I shall try the blocking.

a) Sectional view of the wing [deselect everything else, choose +z axis view, take a snap shot]

b) Same view, with domain boundaries included

i have added the geometry file..

Hi,

This wing has sharp trailing edge. So, blocking may not be feasible (Senior members can comment on this). I would suggest you to go ahead with un-structured mesh approach with sufficient prism layers. You could use the mesh density as well to ensure finer tet-mesh around the wing.

P.S.
I dont have an aero back-ground. But, in real life situation, wouldn't there be a thickness in the trailing edge?

I have a few ideas. Could you attach the original geometry? I need to create a part for the surfaces of the wing: the upper and lower (it does not have a name in project).

P.S. I find it. It's PART_2 and PART_3.

did you find the solution?

While trying to do it.

I tried doing it with the usual technique employing a C grid and assosiating the edges. A y grid for trailing edge. The mesh looks pretty descent but the determinant 2*2*2 quality starts from -0.04..

I am thinking of mapped/free and swept blocks technique for the centre part containing the wing.. Will it be of any use?? If yes i need help in applying it..
Need help to either change the blocking strategy from start or improving the mesh quality with other techniques..

Ralen>where you able to find out the surfaces?? shall i upload again with proper names??

https://www.dropbox.com/sh/1n0atvmgw...r%20struct.rar

The above link contains a mesh without Y block..You can directly download the .blk and .tin files by simply clinking the link..If you zoom in on the tip or root of the wing, you will find that i have not assosiated the edge..Without associating it, the mesh is fine but when i associate it, the mesh at the leading edge of the wing gets completely twisted..

Please help in associating it without twisting the mesh

Without associating
http://img521.imageshack.us/img521/1791/mesh2r.jpg

After associating

http://img521.imageshack.us/img521/1791/mesh2r.jpg

http://img651.imageshack.us/img651/74/mesh3.jpg

Ahh, classic misdirection...

You were focused on the edge across the tip, but that was not your actual problem. You actual problem causing the mesh collapse along the front edge of the wing is that you have the edges above and below the leading edge both projected to it... Right click on edges and "show associations" to make this clear. You may have also had a 2nd problem of associating the leading root edge to only the top or bottom profile curve, and that made the first error harder to find...

Attachment 13210

So how to fix?

The best way is to create an ISO curve along the wing so that you have a curve to project to (Geometry => Create Curve => Surface Parameter => Method (Point on Edge)) And then associate those edges to the new curves...

Attachment 13211

Attachment 13212

You should also put an Ogrid thru your wing... It doesn't matter inside the wing, but it would have helped past the wing tip. Since you needed to collapse the trailing edge, this will require a wing box on the front half of the wing with the Ogrid thru that... you could create another iso line for the split down the middle of the wing, but I didn't bother because I kind of like it straight.

This shows the block selection for the Ogrid...
Attachment 13213

While doing the block selection, I noticed that you had not deleted the blocks from inside the wing (or at least put them into a solid part), so I did that also.

There was a little work adjusting edge distributions on the new edges (copied from nearby)...

Attachment 13214

And here is the blocking file...

I did plan to attach my version of the tetin file with the two iso curves, but it is too big to attach easily, so you can just create the isocurves yourself...

Would you like to sharegeometry through ansys ftp or www.4shared.com

@Simon : It is great, as great as should be :cool:.

I am interested that how much quality you get for this type of geometry after blocking.

I didn't check... But I also didn't try. I am sure there will be some need to adjust vertex locations and edge distributions.

Well, what I am trying to ask is that, geometry has very high curvature and sudden change in airfoil size. Therefore may be one need some advanced topologies as you have discussed in three part youtube tutorial on airfoil.

This blocking looks fine to me...

I guess it may be good to have the Ogrid wrap around the end of the wing, but that is much harder to do with a sharp trailing edge...

Basically, you end up creating the Ogrid so it fills the the space around the wing and back to the outlet. You split this Ogrid for the surface of the wing (some Ogrid outside and some inside) and then you collapse the inside portion behind the trailing edge. Sorry, I don't have time to do that right now.

But as I said, this blocking is pretty good and can be adjusted (vertex movement and edge distribution) to exceed the quality requirements of all the major solvers.

And what about the triangular block at the rear of the wing? Convert to Y-block? My idea was somewhat different, but the first picture is the same. :)

Similar idea is also discussed here. http://www.cfd-online.com/Forums/ans...mesh-wing.html

Thanks a lot Simon, It is great..I am just working on improving the quality a bit..

I have one doubt left...:( .. It is about the conceptual requirement of the o grid (wing box) inside the wing..I didnt understand when you said "to collapse the trialing edge, a wing box is required required inside the wing"...
Can anyone explain it a little more..

@Ralen and Far---after applying yblock for the triangular elements, worst quality by det 3*3*3 is around 0.6

I just meant that you can't put an OGrid inside a wedge... At least I don't think you can (I don't think I have tried it many years). So you must split your wedge so that you have a box and a wedge, then you can put the Ogrid inside the box...

It is still not perfect because while the leading edge side of the OGrid solves your 180 degree element problem, the other side gives you 45 degree elements (Probably not a problem for the solver, but not ideal either)... If you created the Ogrid in the wing before collapsing the trailing edge, I guess you could collapse the back side of the Ogrid and have three wedge blocks inside the wing profile, but that topology would require starting over.

This is more than better.

You mean that either put the o-block before collapse (before wedge block is formed) command or make the box (four sides), where one can put the o-block. It is very much clear to me, if I am getting you right.

@Ananthakrishnan: In wedge block, you cannot put the o-block with simple command, because it will again give you wedge block. For this you need to think the some innovative method.

It is also discussed here http://www.cfd-online.com/Forums/ans...mesh-wing.html with different methods.