Your blocking stratégy is wrong ... way too complex.
See attached what I would have done.

Just think that O-Grid is good only with circle shape ... that's all, it's not a good idea with rectangular shape.

Thanks BrolY, the strange thing is when I reduced the bunchings in the x directions I had no negative volumes. I don't see why that happened.

In the blade problem, is block orientation a possible cause for negative volumes? I attach a picture of blocks colored by orientation.

In fact, it makes sense ... the blocks were highly distorted. If you add more or less nodes, ICEM would sometimes manage to handle this distortion, sometimes not. It would depend on the number of nodes.

As I said before, I can't help you without .tin and .blk files, but try to move vertices in order to correct block distortion.

By the way, you said you can't share the .tin file because someone else created the geometry. But you can share the .blk file, it's yours ! Maybe we could help you a little bit more with it ...

Blocking files
 

Indeed! I upload the blocking I had when I opened the thread (Blade 22 02 13.blk) and a simplified one, based on the first one but with many blocks mergec (Blade 25 02 13.blk).

Thanks a lot!

I had a quick look at your blocking file.
There are a lot of troubles : splits which are not propagated as they should (that is not an issue, unless you are really confotable with blocking), vertices which are not aligned correctly (so blocs are distorted) etc ...
Far could help you more than me here, but I think you should redo from the beginning your blocking and watch some videos on youtube ...

Thanks BrolY,

I'm comfortable with blocks and index cause I've redone this and other projects many times :D. However, I'll try to propagate splits whenever it's possible.

So that's a reason for distorsion? Vertices always have to be alligned??

I started following Ralen's approach for the rounded tip of the blade (as suggested by Far) adapted to my case, an entire blade.

http://www.cfd-online.com/Forums/ans...ca0009-3d.html

I was trying to be systematic with the blocking but I had to move many vertex to get blocks simpler and closer to the blade surface.
But yes, I'm thinking on redoing all the blocking from the beginning. I'm also also checking the forum for similar cases and making trials with a simpler geometry too.

Any useful videos? Besides turboengineer channel ;)
http://www.youtube.com/user/turboengineern

Vertices need to be as much as possible aligned. Because a block is defined by 4 vertices, so if they are not well aligned, your block would be distorted.
When the block is rectangular, you have a perfect mesh with a determinant of 1 and a min angle of 90°. But the more you move vertices, the lower the quality is reduced. That's why it's recommanded to have a determinant > 0.2 and a min angle > 18° to have bad elements.

Besides reducing the elements quality, what seems pretty obvious, can it cause negative volumes too?

Yes, if elements are penetrating geometry or making very high angle.

Negative volumes in a simpler case
 

Hello everybody,

I'm making some tests to understand how the negative volumes appear so I'm using a simpler geometry of a NACA cylinder without twist nor chord length change along the height as the blade is. I'm trying to use a very simple blocking, including an o-grid, with very small element height near the surface in order to avoid wrong y+ values later on. Even so, negative volumes still appear all over the surface.
How can I avoid that?
Should I make the o-grid length bigger?
Should I use shorter blocks along the cylinder height?
Can it be due to lack of orthogonality next to surface elements?
Could anybody take a look at my geometry and blocking and give me any advice?

Thanks a lot, guys.

blocking is not correct.

So what could be a more appropiate blocking?

I've tried with a more adapted to rounded shapes one, I've moved some vertex but same results. (see attached file)

This may be one of the solutions...

Pics
 

http://imageshack.us/a/img35/4967/a001st.jpg

http://imageshack.us/a/img841/8404/a002m.jpg

http://imageshack.us/a/img534/2787/a006d.jpg

http://imageshack.us/a/img202/7589/a004hg.jpg

http://img443.imageshack.us/img443/7917/a003.jpg

http://img823.imageshack.us/img823/1303/a005qu.jpg

Wow! Thanks a lot Far!

I was watching your video:
http://www.youtube.com/watch?v=mgjRT4WY_iA

Some questions about your blocking:
Would the inner o-grid be necessary if the cylinder went from side to side and there were no need to do the internal section mesh?
Putting and edge along the trailing edge is not recommended?
Is an o-grid necessary in the outside of the wing or should I avoid it?
Can I use this approach for the twisting blade?
Should I divide the blocks along the blade height not to get very twisted ones?

Now I have to do that with the rounded tip so I have to "merge" your methodology for the Naca wing and Ralen's for the rounded tip. And eventally apply all that to the blade...

Another mesh I gotta do is a T-intersection of NACA wings like this one. I'll upload here the geometry.

Thanks for the help. I really appreciate it!

Negative volumes again
 

I've done the blocking again, except that inside the wing, it's not complicated but in the end I get many negative volumes in the upwind area! Why?? I've run the fix check/fix blocks tool and fix inverted blocks with no luck.

I upload pictures and files.

collapse trailing edge. Trailing edge is very sharp and this blocking strategy wont work. Try the one I posted earlier.

Edit : Min quality is now 0.06 after adjusting the vertices and increasing no of nodes from 15 to 40.

Did you avoid negative volumes by just adjusting the vertices and increasing the no of nodes?

yes. but it did not increase quality much, though it avoids the negative volume.

I've tried the previous blocking with a collpased trailing edge but I still get negative volumes (see picture 2) :confused:. I attach the files.

I'm starting to have nightmares with negative volumes