|
[Sponsors] |
December 9, 2020, 06:02 |
Improve transition and orthogonality
|
#1 |
Member
Cpt. Convergence
Join Date: Feb 2020
Posts: 98
Rep Power: 8 |
Dear community,
I am generating a multiblock mesh and the transition and mesh orthogonality has to be perfect in order to minimize the missallignment between the neighbourgh cells and ghost cells between blocks that are causing flow discontinuities and thus divergence when I switch to unsteady mode. You can observe these discontinuities between blocks in the image that I have attached. At first glance, they do not seem significant but they amplify significantly later in the transient simulations. Which is the best way to improve the orthogonality between blocks? I have attached 2 images about other sizing issues that I have in my domain. Since with ICEM I have to set a dimensional sizing to one edge, this will make sense to the edge where it is applied, but when I copy this sizing to the parallel edges, the proportions are not the right ones. Is there any command to make a proportional sizing according to the edge length and apply it to the other edges? Thank you in advance. |
|
January 31, 2021, 12:46 |
|
#2 |
Senior Member
Kira
Join Date: Nov 2020
Location: Canada
Posts: 435
Rep Power: 8 |
Hello,
What mesh law are you using? As well, are you using bunching or not? |
|
January 31, 2021, 16:49 |
|
#3 |
Member
Cpt. Convergence
Join Date: Feb 2020
Posts: 98
Rep Power: 8 |
Hi!
Thanks for your answer. I did some improvements and I guess that the sudden changes in the elements size at the farfield won't have a big impact in the solutions. However it doesn't seem right to me and I am a big disapointed with the fact that I cannot achieve smooth transitions between blocks. I am mostly using geometric1/2 and bigeometric bunching laws. However since I have to do everything by hand, I can't achieve a good quality because when I refine one region, the other looks terrible. |
|
January 31, 2021, 19:01 |
|
#4 |
Senior Member
Kira
Join Date: Nov 2020
Location: Canada
Posts: 435
Rep Power: 8 |
Hello,
Your newer images are for sure an improvement, nicely done, a step in the right direction for sure. I feel your frustration about having a nice mesh near the surface but a less than optimal one at the farfield. Three things come into mind: First; Rather than using Bigeometric for your mesh law (when I said bunching previously I meant linked bunching), try using Biexponential instead. Second, have you tried changing your growth rate in addition to spacing? Lastly, have you tried playing with the number of nodes in each section, or has that been set by something else (literature, constraints, etc)? Meshing is a very delicate balance of all three things; number of nodes, growth rate and spacing, in my opinion. Meshing well is an art. I would say I spend ~35% of my time creating the initial geometry and the remaining ~65% making my mesh the best as I can. Problem is that each mesh is different, so you can't really gain a sense of how to create it off the bat unless you have a TON of experience. I think this is why a lot of people use structured meshing or hybrid meshing (or like with unstructured patches) for airfoils and the like. |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
[ANSYS Meshing] Bad Mesh Orthogonality at Diameter change | pamstad | ANSYS Meshing & Geometry | 2 | October 31, 2019 11:36 |