|
[Sponsors] |
[ICEM] trying to understand multiblock hexa,structured,unstructured |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
August 23, 2016, 13:18 |
trying to understand multiblock hexa,structured,unstructured
|
#1 |
New Member
Join Date: Aug 2016
Posts: 9
Rep Power: 9 |
Hi everyone,
Beginner here. This is the procedure I mechanically follow while meshing anything: create block, split blocks, make associations, move vertices, de-activate solid regions by adding to VORFN. And finally convert to 'unstructured mesh' by right clicking pre-mesh under the model tree. I understand that is this method is referred to as 'multi-block structured hexa meshing.' From another thread, I gather that structured and unstructured refer to the manner in which data regarding the cells and their location are stored, and that structured gives the advantage of computational ease while solving. The thing that bothers me is the last part of my meshing procedure, where I 'convert to unstructured'. Isn't that unnecessarily destroying the advantage of a structured mesh? The solver I use is FLUENT. I know that it is an 'unstructured' solver. Does that mean it cannot read a structured mesh? If so, what does read a structured mesh? Somebody sort me out please thanks |
|
August 23, 2016, 13:40 |
|
#2 |
Super Moderator
|
In old times, FDM (finite difference method) was used to solve the CFD problems. By design it can only work on hexa/quad meshes in computational space. I dont want to go into details of jacobian etc and the level of complexity for the real geometries.
And the mesh output was used to arranged through IJK index. Hence the name structured. Then control volume method come into play and they tend to take advantage the this method for complex geometries without going into physical and computational domain transformation. But it was felt that it is very difficult or cumbersome to work only with hexa/structured meshes. Infact fluent 4 (or may be 5, some body correct me) was only structured solver. Then they started to work on the unstructured meshes (tetra most) and idea was borrowered from finite element method. and it had no connection among neighbouring nodes. and therefore it required another table (matrix) to keep record of it. Hence the name unstructured. In short now Hexa mesh output is in unstructured manner , so it loses its that advantage. but still it has the advantage of clustering easily in the regions boundary layer, shear layers etc. |
|
August 23, 2016, 13:54 |
|
#3 |
New Member
Join Date: Aug 2016
Posts: 9
Rep Power: 9 |
Thank you for replying,
So here is what i've understood: 'structured' format is no longer in use since we have moved from FDM to FVM, and the developers decided to add a 'convert to unstructured' option rather than start fresh and have the data stored in the unstructured format right from the beginning. am I right? Also, the create/split block method is 'multi block structured' right? I have come across 'multi block unstructured', is that a typo? or is the difference irrelevant since we anyway convert to unstructured? |
|
August 23, 2016, 14:08 |
|
#4 | |
Super Moderator
|
you know current flows from -ve to +ve. But still you find out that current flows from +ve to -ve in books. Reason is simple, they dont want to change terminology and every thing
Quote:
When you right click on pre-mesh, it will ask you to convert mesh. There is one option "multi block" which is in true sense a structured mesh. Which is still used by many NASA solvers. Even now nasa shifted its most advanced CFD solver to both mesh types... https://www.grc.nasa.gov/WWW/winddocs/user/intro.html But for fluent or CFD or any other commerical CFD codes, you have to convert it to unstructured mesh. So now we have unstructured-Hexa, unstructured-tetra, unstructured-prism, unstructured-hexa+tetra+prymids etc ... Just drop the first word, we have mesh types : hexa, tetra, prism, prymids etc... Hope it is clear... |
||
August 23, 2016, 14:16 |
|
#5 |
New Member
Join Date: Aug 2016
Posts: 9
Rep Power: 9 |
Very clear.
Thanks!! |
|
May 1, 2021, 03:42 |
|
#6 | |
New Member
Mohammad Fathi
Join Date: Mar 2020
Posts: 4
Rep Power: 6 |
Quote:
|
||
|
|