mesh cylinder
 

Hey guys,
Iam still having problems with my geometry... Could someone please have a look at it and help me out? Seriously I have been struggling with this for a week now.
The Premesh Quality shows that the elements are not really regularly in my cylinder.
Ive tried meshing a single cylinder with a single surface, opensided. the same problem occurs. If i put a surface on the side where its not supposed to be, it works perfectly.
could someone please help me and tell me where the problem lies? I bet its all about the associations again, but i cant figure it out :(
Iam attaching the geometry and bulk files.

Hey dude,

at first sight, i actually thought the problems were related to the associations. But take a look at your blocking! None of the blocks were connected. So I had to merge verticies, delete and recreate the o-grid (you have to do the split up to the bottom!).

Det 3x3x3/Quality now > 0,6

you have to set edge parameters yet.

so created through bottom-up approach!!!!!

bottom-up or top-down.....what's the point?

blocking was just incorrect. Or is it a a appeal for the use of top-down approach!? :)

Both.

For this geometry one has to use both approaches to build the topology. While working on the bottom-up approach, the blocking vertices may be disconnected. So it is better to keep in mind this issue.

ok, i get your point now ;)

Hello everyone,
sorry for hijacking this post. I tried to mesh the geometry and i realize that it does same thing, so where i did is adding surfaces in blue as you can see in pictures below(first one), everytime i close with a surface, the mesh is good, if i dont close like the last picture, the mesh come out bad. my associations are goood since i tried it on that little inlet there. where does the problem comes from.

Just created one block and fit it into the square box and cylinder. Then split the main block into 9 blocks: 8 for square box and one for cylinder. Created O-block in cylinder up to bottom of square section. Then used bottom-up approach (extrude face along curve) to create the blocking for the rectangular pipes (2 on each side and 8 in total)
http://img405.imageshack.us/img405/9340/mixture1.png
http://img254.imageshack.us/img254/8950/mixture2.png

I didn't add any surface and still has no problem with meshing.

Applied link edges command: linked the top square edges to bottom square edges
http://img207.imageshack.us/img207/7660/mixture3.png
http://img542.imageshack.us/img542/7740/mixture4.png

you should resign from your job and start working as a ICEM instructor :)

Probably you'll earn even more money :D

I would definitely call on your services

Done :D. But how did you know this :rolleyes:

made edge settings and quality is also further improved >>> 0.84

http://img254.imageshack.us/img254/1...mixerfinal.png

bottom up approach works fine.
like a boss, i agree with energy382. thanks anyway

While I agree with Far that top down is easiest and fastest way to handle this model...

Bottom up would also work fine. You just made one real mistake...

When you create a block in ICEM CFD, it assumes that all faces must project to the nearest surface. If that block is attached to another block of the same material, then the projection of the face between the blocks is turned off because that is an "internal face". This happens automatically, but you can control it interactively thru the face associations.

In your case, each of the cylinders (geometry) was attached to a larger model on one end, so there was no surface there. However, you had not actually connected your blocks (by merging the verts at the corners...). So each block thought that it was free standing and needed to form a closed surface. However, without a surface to project the inner end to, the elements simply projected to wrong surface...

Energy382 solved the problem when he merged the verts (and the blocking), because as soon as the faces were merged it automatically changed the projection on those faces to "none".

I used both, first top-down for main geometry and bottom-up for 8 pipes, therefore I didn't come across issue like unconnected blocks. Hence no need to merge vertices.
PSYMN is this good strategy?


@ PSYMN : are my skills good enough (though I don't think so) to fit into this role? Any chance at ANSYS? :D

O-grid faces selection
 

I was wondering about the criteria in selection an O-grid faces?
What to account for when choosing the faces after dividing a given block with an o-grid method in ICEM.
Thanks for your answers
:)