CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Meshing strategy for 3D sharp point (http://www.cfd-online.com/Forums/ansys-meshing/107986-meshing-strategy-3d-sharp-point.html)

D-willism October 10, 2012 22:55

Meshing strategy for 3D sharp point
 
1 Attachment(s)
Do you guys think this sharp point can be meshed well with hexa grids? any suggestions on the meshing strategy?

Looking forward to your replies, thank you!

diamondx October 10, 2012 23:03

hmmm very difficult...
I once face a sharp geometry but not like this one. i had to use y-grid.
take a look at this video, there is no sound but i added annotations:
http://www.youtube.com/watch?v=nuiRkZqaEz4
try it with this method , come back to us here let us know how was it

diamondx October 10, 2012 23:25

wait a second; y-grid needs to be applied to triangular shapes... yours is not...
just realized that while laying in my bed :D
May i ask what are you trying to model ?

D-willism October 11, 2012 00:01

Modelling? you mean CFD analysis? I am modelling the air flowing across this cone shape, so I have a cylinder enclosing the cone shape. but around the sharp point, the aspect ratio and skewness of the grids are kind of poor. Does this answer your doubt?

diamondx October 11, 2012 14:40

what software did you use to generate your mesh, can you post some screenshots ?

davidwilcox October 12, 2012 01:04

I once meshed a geometry with a sharp end. I used the quarter O grid blocking method for the sharp end. Maybe you can give that a go. :)

D-willism October 12, 2012 07:44

3 Attachment(s)
Quote:

Originally Posted by diamondx (Post 386193)
what software did you use to generate your mesh, can you post some screenshots ?

I am using ICEM CFD. Three pisc show the geometry, blocking and poor quality grids. all the poor quality grids gather around and are due to the wedge-like blocks at the right bottom of the rectangle geometry. If I can get rid of these two wedge-shape blocks which means only four blocks left around the half-cone shape, the quality of the grids should be very high. Any ideas or suggestions how to achieve this?

Best Regards

D-willism:)

D-willism October 12, 2012 07:45

I will try that once I get to university tomorrow. thx

Far October 12, 2012 11:56

share geometry

D-willism October 12, 2012 20:42

1 Attachment(s)
Quote:

Originally Posted by Far (Post 386310)
share geometry

zip file includes 2 geometries I am trying now.

Best Regards

D-willism:)

Far October 13, 2012 03:43

For rectangular enclosure blocking is simple. Min quality is 0.9, however min angle is 16.9. Try solving this mesh in Fluent.

http://imageshack.us/a/img268/559/rectangle1.png

http://imageshack.us/a/img189/5875/rectangle2.png

http://imageshack.us/a/img843/7373/rectangle3.png

http://imageshack.us/a/img132/5555/rectangle4.png

http://imageshack.us/a/img89/1381/rectangle5.png


Circular Enclosure: The blocking of this geometry with circular enclosure is felt difficult to me always. Can someone guide us how to handle the circular domain in efficient way?

Far October 13, 2012 03:56

1 Attachment(s)
Please find enclosed files (.prj, .tin, .blk). After some adjustments in blocking min angle is greater than 18.

Far October 13, 2012 04:02

Quote:

Originally Posted by D-willism (Post 386261)
I am using ICEM CFD. Three pisc show the geometry, blocking and poor quality grids. all the poor quality grids gather around and are due to the wedge-like blocks at the right bottom of the rectangle geometry. If I can get rid of these two wedge-shape blocks which means only four blocks left around the half-cone shape, the quality of the grids should be very high. Any ideas or suggestions how to achieve this?

Best Regards

D-willism:)

The elements are prism and ICEM is very harsh on prisms quality (Simon told it in many posts, search you can find one). So give it try with your mesh also in solver and see what happens in terms of convergence with both meshes

D-willism October 13, 2012 04:27

Thank you SO SO SO much! You've been very kind

D-willism October 13, 2012 05:43

When I opened the project file using ICEM CFD 14.0, it required a attribute file? But obviously it is not in the file you posted, does it really matter? And hmm... because you adjusted the angle from 16 to 18, the min quality degrades and not the same as what you show us in the picture. Could you please post the files before you made angle adjustments please if you can? Thank you so much!!

Best Regards

D-willism

Far October 13, 2012 06:15

1 Attachment(s)
Quote:

When I opened the project file using ICEM CFD 14.0, it required a attribute file? But obviously it is not in the file you posted, does it really matter?
This file does not matter for the blocking. It has other functions.



In this mesh min quality is 0.8 and min angle is 18. Please uncheck the solid part if it is turned-on...


http://imageshack.us/a/img818/3897/rectangle6.png

http://imageshack.us/a/img713/7498/rectangle7.png

D-willism October 13, 2012 23:05

Thank you Far! I am now trying to block and mesh the geometry without having those wedge-shape blocks at the bottom of the geometry, which means only four blocks left enclosing the semi-cone (solid, as you named in the file). Do you think it is feasible?

Best Regards
D-willism :)

Far October 14, 2012 01:45

I am sending you two project files (both include .prj, .blk and .tin). That is before merging the vertices and then after merging the appropriate vertices. Now you can apply the Y-block option on two wedge shapes blocks to get what I posted earlier.

PS. I have also used the build topology command to create the points at the required locations.
I have also used the ungroup curves command, available in blocking association panel

Far October 14, 2012 02:07

1 Attachment(s)
please see these five files, containing five steps from the creating the initial block to the final mesh. Quality is around 0.8 and min angle is 18

D-willism October 14, 2012 02:50

Thank you Far! what I was trying to say is: do you reckon if it is feasible or possible to block the domain into four sub blocks, instead of including small wedge-like blocks as well underneath the sub blocks (big/major ones)?

Best Regards
D-willism


All times are GMT -4. The time now is 16:40.