CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (https://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Blocking strategy for two hemispheres connected by a pipe(negative volume) (https://www.cfd-online.com/Forums/ansys-meshing/136383-blocking-strategy-two-hemispheres-connected-pipe-negative-volume.html)

Polarbear May 27, 2014 12:30

Blocking strategy for two hemispheres connected by a pipe(negative volume)
 
5 Attachment(s)
Hi all,
I have a problem that I've been banging my head against the wall for the whole day.
I have a hemisphere that is connected to a cylinder which in turn is connected to a second hemisphere. I want to do a structured mesh and approach the problem like this.

1. I start with making a block and split it up into three blocks.
2. I make an O-grid split and remove the other blocks from the cylinders.
3. I make a second O-grid split since I want boundary layers in the cylinder.
4. I then associate faces with surfaces and edges with curves
5. I snap the vertices into place and start working on the distribution of nodes. I can also use Edit edge-automatic linear to achieve a blocking that follows the surface.
6. I look at the 2x2x2 determinants, loads of them are negative and this becomes a problem when I load it into an unstrucutred mesh and check for errors, I get tons of errors(regarding negative volume).

Anyone having any ideas?

Polarbear May 27, 2014 12:32

3 Attachment(s)
More pics and files

kad May 27, 2014 15:28

Your blocking strategy is not correct. Do not select the faces on the round side of the hemispheres. Check the tutorial cube in a hemisphere. It comes with ICEM. Blocking strategy should be easy to derive from this.

Polarbear May 29, 2014 02:26

Thanks kad! I devised a new blocking strategy by using the tutorial, that is to make box-cylinder-box block mesh. And then connect it to two hemispheres with a hole in them(as shown in the sphere cube ICEM tutorial) using a interface.

Rep to you!

Cheers

Far August 21, 2014 09:02

1 Attachment(s)
it is almost two month old thread, but i feel it is good idea to post solution to this problem. Steps are:

1. Create one block with name "fluid" and also create three materials points "fluid" in both hemispheres and connecting pipe.

2. Split block flat sides of both hemispheres. And delete center block (corresponding to center pipe) permanently. When blocks are deleted permanently, edges on either sides are no more parallel and therefore you can specify different no of nodes on both sides.

Associate edges to curves and snap vertices to the surfaces of both hemisphere.

3. Apply sphere-cube strategy, but select face(s) on flat side for both blocks. Adjust ogrid using either move vertices or rescale o-block command.

4. Now make four splits for the pipe connecting both sides. You need to apply split operation on both sides as geometry is no parallel and hence split will not propagate to other side.

5. Now create block for the pipe using "block from vertices" and select eight vertices on both ends of pipe in particular order. See help for this command.

5. Now create ogrid around centre block passing from both HS (hemisphere) and pipe.

6. Again use O-grid command and again select same blocks and this time create ogrid inside by unchecking option "around". Adjust ogrid edges on the surfaces of HS

7. Adjust pre edge mesh parameters and see the premesh. If satisfied with mesh quality and no of nodes on each edge convert this mesh to unstructured mesh and set boundary conditions, solver and output mesh.

8. Import mesh in fluent and check the mesh quality there as well. Min orthogonal quality should be atleast 0.1 (0.01 is minimum requirement through)

http://s25.postimg.org/728gibk3z/Untitled.png

Polarbear August 21, 2014 15:57

Thanks far for the extensive answer! I'm finished with this problem but it's always good to learn something new. It was way above me block meshing skills at the time!

Cheers, thank you both(kad) for taking your time to help people and spread knowledge!

Far August 22, 2014 08:10

just a side question: Was this problem related to multiphase simulation?

Polarbear August 23, 2014 07:38

Nope. Single phase compressible flow:)


All times are GMT -4. The time now is 23:54.