I am not happy with mesh. I dont know why mesh goes circular to square shape and then again square to circular. In this type of geometry where internal and external parts have similar shape, we should have the very good looking circular mesh. Any idea why this is happing?

http://img337.imageshack.us/img337/3862/51976330.png

http://img684.imageshack.us/img684/8881/58104809.png

http://img72.imageshack.us/img72/7494/21404523.png

http://img215.imageshack.us/img215/2884/68130736.png

I Followed it but the main problem is the bottom circle of the hemisphere. It itself need a 2D O-grid but in 3D blocking it is not possible as I know.
here is a 2D o-grid for a circle.
http://www.cfd-online.com/Forums/att...1&d=1339053729

it is impossible because Left Far-field and Right Far-field are Inlet and Outlet or vice versa.

as you mentioned it is not a good one, look at the min angle: we have 0-9 degree cells.http://www.cfd-online.com/Forums/att...1&d=1339053637

No, It is not true. You must have activated solid part. I was talking about the general mesh shape otherwise quality is good and angle is higher than 40.

I guess we are on different track. I guess circle should be hollow. Can you describe your problem again with geometry pics to be modelled and some additional description?

Here is a sample picture: the hemisphere is solid not hollow
http://www.cfd-online.com/Forums/att...1&d=1339058837
and we need farfield as I created in the geometry file.

Ok , got it. Nevertheless, we still make the hemishpere as hollow since fluid is not going inside.

In this case the far feild on the other side should be rectangular in shape to ease the meshing.

In the mean time for hemisphere section, use strategy we have just discussed. And in the other part with flat end (cylindrical enclosure), use simple topology and then create o-grid inside circular base all the way to end of cylindrical domain. Got it?

I think it has no difference. The block associates with the far field whether in sphere shape or rectangular.

Sorry I didn't:confused:.
I think my problem is the mesh on the bottom of hemisphere (as I showed in previous comments) that we should make a 2d o-grid on that circle but I couldn't do it.
If you have time to make a correct blocking, I will appreciate;).

I have already started to work on this. Going to post it shortly

Done
 

Done. It has some tricky part, when I was trying to snap vertices to hemisphere. Unlike the space cube tutorial where all things were done nicely with associate vertex to point on the inner cube, but this is not the case with this problem.

Enjoy and let me know if you get same problem as discussed above.

pics
 

Pics:

http://img819.imageshack.us/img819/9939/88103190.png
http://img816.imageshack.us/img816/6307/16606127.png
http://img215.imageshack.us/img215/4139/22079432.png
http://img407.imageshack.us/img407/710/80107674.png
http://img838.imageshack.us/img838/2852/82813442.png
http://img692.imageshack.us/img692/1319/35161749.png

Fantastic, I also did it and you can see. Thank you for your attention.You create a hemisphere part including the large circle. I create the hemisphere part exactly the hemisphere so in association process first I associate the bottom of block to bottom of hemisphere (using "associate edge to curve") and then associate the faces of my inner block to hemisphere part(using "associate face to surface") and finally using "snap Project vertices" vertices go to the right position and there is no need to move vertices manually.

Thanks again.

I learn to feel the ICEM
 

I guess, now I am feeling the ICEM from heart. For example in this case if you snap the outer block first and then inner block, you will get the mesh problems as I have stated previously. You have to snap vertices at once. So make all oblocks, change material and snap the vertices.

Mine vertices for inner block were placed at Y= 1.9, while Reza's vertices (same vertices) were found to be at Y = 2.9. When used above trick or method, I have also got the vertices at Y = 2.9.

Have fun.

Dont use this method, this method is used for making the internal wall. You just need to change the material and you will get the mesh on hemisphere.

I didn't exactly understand Why not to use it?
How do you associate the inner block to hemisphere?

When you delete the internal block or change its material, icem automatically makes the associations.

I found the reason for this. This is because I have forgot to switch back transfinite degree = Quadratic to transfinite degree = linear (linear is default option), so ICEM made it my default method and I was wondering for weeks that this is bug in ICEM. But now things are good :D.

Mid and outlet plane with quadratic option:
http://img33.imageshack.us/img33/6284/84012066.png
http://img837.imageshack.us/img837/4931/98393900.png

Mid and outlet plane with linear option
http://img406.imageshack.us/img406/5254/92553335.png
http://img812.imageshack.us/img812/4431/39806727.png