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] Blocking topology for cruciform fins (http://www.cfd-online.com/Forums/ansys-meshing/106583-blocking-topology-cruciform-fins.html)

siw September 2, 2012 10:55

Blocking topology for cruciform fins
 
1 Attachment(s)
Hi,

Please can somebody explain the blocking topology for a body with cruciform fins (i.e. fins in an X orientation)? I cannot use 1/2 or 1/4 symmetry and some fins have blunt and some have sharp trailing edges so it'll be a mixture of C-grids and O-grids. I've tried but I cannot get anything to work.

I've attached a sketch (simple and not to scale) of one attempt (I hope it is clear). I've also tried rotating the block 45 degs before splitting and associating etc to align with the fins but I still cannot get the blocking right.

I've read the topic http://www.cfd-online.com/Forums/ans...lex-shape.html but that did not help.

Thanks

diamondx September 2, 2012 11:19

Good morning Stuart,
I never had the chance to mesh something like that. After giving it a tought , i think your blocking is complicated. Yet i need to try mine but how about seeing thing this way:
after looking at geometry i rotated it a little to make the fins perpendicular, and to make more easy i replaced the circle with shape. It gave me this
https://dl.dropbox.com/u/35161486/square.PNG

I concluded that i can block this geometry without any problem, hence it is very similar to your... Do you agree ?
If you have a geometry we can try together...

Far September 2, 2012 14:04

This is the basic idea of blocking this geometry regardless of meshing software. But this is handled in better way with ICEM CFD

http://img255.imageshack.us/img255/2...methoughts.png

stuart23 September 2, 2012 20:51

2 Attachment(s)
C'mon guys! This was too easy! Hows about a bit of a challenge next time ;););)

Far September 2, 2012 21:40

well he is facing problem when fins are on airplane ;)

stuart23 September 2, 2012 22:15

@far: challenge NOT accepted!

Far September 2, 2012 23:35

accepted ;) .............

siw September 3, 2012 02:52

3 Attachment(s)
Thanks all for your replies - I hope you'll stick at it with me as this is the most difficult mesh I've ever needed to make and accurate results are very important to me here. Which is why I'm going down the Hexa route rather than with Tetra/Prism which I have tried but as always the Prisms muck things up (Yes - I've read the various ANSYS posts/presentatiations on making good prisms) as there is some other complex geometry.

I've continued and my current attempt has reached a dead-end. Attached are some sketchs (poorly drawn) of some other things I've tried - including first rotating the block by 45 deg at align with the fins.

diamondx: I don't follow your picture. How can that actaully be made?

Far: your image does not show - only a grey and white striped box is shown.

stuart23: if you find this one easy please can you say the steps that need to be done?

Where to I nee to make splits and assocations etc?

Regards

stuart23 September 3, 2012 03:22

3 Attachment(s)
Stuart, (this feels like I'm talking to myself)

I have reposted Fars pic in a jpg format because you could not see the original.

What I did was first create a circle that circumscribed the outside of the fins. if you don't want the ends of the fins to be curved: split the circle, delete the end curves and replace them with lines.

I then initalised the blocking and created the O-grid block. After that, split the block on one of the radial edges to create a concentric "doughnut" block. I then split the circumferential edges near the sides of each fin to create a "union jack" topology. After all the associations (they should be quite obvious), you can then just delete the blocks on the inside of the object.


I hope this helps, post any more queries you have.

Stu

siw September 3, 2012 03:59

1 Attachment(s)
stuart23, I think I understand what you did (the pictures make it so much easier to follow).

How would you further extend that to a fin which has a rounded leading edge and either a blunt or sharp trailing edge?

I have modified one of the images from my previous post. Is it possible to merge the 3 blocks (in red) so the green and purple edges would be opposite and thus make a single block between a set of fins? NB: I project a vertex the fuselage surface.

Thanks

stuart23 September 3, 2012 04:12

I believe that you should be able to merge the blocks. They are parallel o-grid blocks, so it should work.

As far as converting the model to 3d, you might want to do the other splits (ie Leading Edge C-Grid and blunt Trailing Edge) before you create the splits as above. This way you should be only performing the splits on the blocks at the same x location (lengthwise) as the fins.

I am not sure exactly what to do before and after the body (I assume this is a missile). You will probably end up with wedges along the line of axis-symetry unless you do something different for the nose. I'm sure there are other ways to avoid wedges, I just can't think of them right now.

Stu

siw September 3, 2012 04:58

stuart23,

My first series of splits was length-wise along the body. That way I have single block containing a piece of the fuselage and the fins.

Thanks

diamondx September 3, 2012 11:35

A small video about it
http://www.youtube.com/watch?v=QdJHD9DQyak

It's important to notice how edges take different directions when o-grid are created...

siw September 4, 2012 02:04

diamondx, thanks for the video that has helped considerably:).

I am never able to plan successful blocking, as you can see from my sketches. How didi you go about seeing that blocking?

Far September 4, 2012 07:12

Quote:

Originally Posted by siw (Post 380089)
diamondx, thanks for the video that has helped considerably:).

I am never able to plan successful blocking, as you can see from my sketches. How didi you go about seeing that blocking?



He is our heavy weight champ

stuart23 September 4, 2012 07:47

I've heard that diamondx actually dreams in structured mesh!

diamondx September 4, 2012 10:04

Quote:

He is our heavy weight champ
All the credit goes to stuart23 this time ! i just re-did his blocking.
One useful tip: In unstructured mesh, it's useful to filter the curves and the points that you don't need. But when using blocking, i found it useful to keep them, or more , to add them where you don"t have so you can know where your going. it's a common practice i used to do when i just started blocking. Because when you perform association, you get the chance to have a good preview on your mesh by hitting pre-mesh.
Blocking is about adapting the mesh to your geometry, so where you have a curve and a point somewhere, think about creating an edge and a vertices there, and sometimes slicing edge is not the only option, i go through all the buttons in the blocking tab and i try them even not sure what the result will be... until i find something that fits. the tip above about the edge that take different direction with the o-grid was a big thing when i learned blocking.

Quote:

I've heard that diamondx actually dreams in structured mesh!
Funny but it happens to me, sometime i'm so stuck with a mesh that i dream about it, i wake up in the night with new ideas, but i stay in my bed til the morning before i try them :p...
Yesterday night i was reading stuffs in the link monash.edu.au . You did all the cluster thing, pretty amazing...
i also have the chance to work on a cluster here, here is the link http://www.clumeq.ca/index.php/en/ab...uters/colossus
But i don't use all the nodes because of license limitation :mad: other departments use that for biology and DNA stuff which are very demanding in computer resources.

stuart23 September 4, 2012 20:01

Quote:

Funny but it happens to me, sometime i'm so stuck with a mesh that i dream about it, i wake up in the night with new ideas, but i stay in my bed til the morning before i try them :p...
Hahaha diamondx, I'm no psychologist, but this sounds like you have a huge obsession!!!

I am very jealous of your cluster. Your 1000 node, IB QDR interconnected Lustre Filesystem monster outweighs my 100 node, GigE interconnected RAID5 Filesystem machine!!!! I really like the layout of your cluster as well, it reminds me of the old Cray Machines (X1 etc) but on a much larger scale.

I am however lucky to have access to a pool of 240 Ansys HPC licenses though, the licenses alone are probably worth more than the entire cluster.

stuart23 September 4, 2012 20:10

Quote:

I am never able to plan successful blocking, as you can see from my sketches. How didi you go about seeing that blocking?
Stuart,

Blocking takes practise to master. Whenever someone posts something on this forum, think about how you could block it, maybe even have a go if you have time. By trying other problems, and seeing other peoples solutions, you will build up a skill-base that will be invaluable for your own projects. I have defiantly learnt far more blocking skills from watching the "cfd-online sensais" than from doing it myself.

If your stuck on a problem, print out a picture of your geometry, and use a pen (very old fashioned!) to draw on how you want the final mesh to look. Then it should be easier to pick out where to put verticies and edges, and from there you can devise a blocking strategy. From the scans you made, it looked like you were doing this, just keep practising and practising and practising!!!

siw September 5, 2012 06:18

Yep, I spend a fair amount of time at the start doing blocking sketches before I even start up ICEM. However, as you haven seen attached to the topic in this instance but it did not help much.

One can either 'see' topologies or not; no matter how much practice is tried (e.g. me:mad:)


All times are GMT -4. The time now is 19:53.