![]() |
T Junction Meshing using ICEM CFD
3 Attachment(s)
Hello,
I am new to ANSYS and I am currently learning meshing in ICEM CFD for my work on T Junction based flows. I have a few questions 1. What is the purpose of an O Grid? Is it only used at the boundaries? where can I learn more about the application of an O Grid? 2. The quality of mesh in the T Junction (where the two pipes intersect) where the important physics needs to be captured is average (It's related to my lack of understanding and using the blocking function properly). I have also posted some pictures which display the mesh along with the quality histogram. Any suggestion on how to improve the meshing functionality and reducing errors would be welcome. Thanks in advance. |
4 Attachment(s)
|
thank you
Hello Mr. Far,
Thank you very much for your help. I think I understand the blocking strategy that you have implemented. I will try to do it and see what happens. Regards, Karthick |
Just brief steps :
1. Create main block 2. split at the junction and then two splits on the two sides of small pipe and perpendicular to bigger pipe 3. delete two blocks on right and left of the small pipe 4. Make associations 5. snap vertices. 6. make o-grid inside the small small upto the large pipe's bottom. Select two extreme faces. 7. Make the O-grid for the larger pipe and again select the respective end faces. 8. go to edge mesh parameters and update all sizes and go to second tab and give size 2 and apply. Keep on applying until mesh size is at-least 20000 hexas 9.Check premesh. 10. One horizontal split in the large pipe. 11. Play with vertices location panel to adjust the blocking to higher quality. |
Smooth mesh
3 Attachment(s)
Smooth mesh after applying the elliptical solver.
|
Hello Mr. Far,
Could you please explain me as to what is the function of an O Grid? Also, when checking premesh quality histogram, the angles are not good? How to improve the angles in the mesh? Also, for this T junction geometry, what are the other Premesh quality criterion that should be checked? Regards Karthick |
Min angle is 32 and min quality is 0.6. which is good enough.
|
The O-Grid is very powerful because with this circular shape, O-Grid helps to have orthogonal cells around the circular shape.
Try to mesh a tube with, and without O-Grid, and it will be obvious ;) |
Mesh quality is 0.6 in ICEM and min orthogonal quality is 0.5 in Fluent and after smoothing (orthogonality) min mesh quality is 0.28 in ICEM and min orthogonal quality is 0.6 in Fluent.
Why smoothness decreases quality in ICEM and enhances the quality in Fluent? What is the relationship between ICEM quality and Fluent orthogonal quality? |
ICEM help manual about the quality :
Hexa The quality is a weighted diagnostic between Determinant (between -1 and 1), Max Orthogls (normalized between -1 and 1; if deviation from orthogonality is greater than 90 degrees, then the normalized value will be smaller than 0) and Max Warpgls (normalized between 0 and 1; warpage of 0 degrees is 1, warpage of 180 degrees is 0). The minimum of the 3 normalized diagnostics will be used. When you do orthogonal smoothing, you increase the orthogonal quality, but maybe, it also decreases the quality of your determinant. I guess Simon could help us on that. |
Check this Figure.
http://afinemesh.files.wordpress.com...el-529x359.png Can you tell me which quality he is talking about? |
5 Attachment(s)
Another idea.
1. Split selected block. 2. Split selected block. 3. Delete default associations of six edges and associate them to surface. 4. Snap Project Vertices. |
2 Attachment(s)
5. Create O-Grid with all blocks and faces of three ends.
|
can you share the files?
By the way excellent comeback. |
good
3 Attachment(s)
Ralen method is simple and somehow giving better quality and good looking mesh with few splits as you can see in attached pics(Mine in this post and Ralen' idea in next post.
My method: |
Ralen's Method
3 Attachment(s)
Here are the images from Ralen's Method:
|
1 Attachment(s)
|
Hi Ralen,
The link is died, could you please upload again? Thank you very much! |
| All times are GMT -4. The time now is 08:21. |