I’m trying to map forces in a conservative fashion from a coarse fluid mesh to a much finer structural mesh. Due to the non-conforming meshes severe unphysical spatial oscillations occur.
Now my question is, if there is any best practice to overcome this issue. In the moment I use nearest neighbor mapping and tried out some RBF methods, which lead to even more instable and unphysical solutions. I actually believe, that RBF methods could resolve my issue but I have problems setting up the right shape parameters.
As this issue is probably not totally uncommon, I wonder if there is any best practise solution for this.
At least any hint how to choose a good RBF method and parameters would be very helpful.
Thanks in advance!
Your case is uncommon in the sense that the solid mesh is much finer than the fluid mesh. In this case, conservative mappings are known to give oscillations.
For example: https://doi.org/10.1016/j.cma.2008.05.001
I guess you have different options:
- Use stresses instead of forces. Then, you could use a consistent mapping. Of course, this needs some work, but could pay off in the long run.
- RBF will be tough since the interpolant needs to be computed on the fine solid mesh. This will always be tricky. Definitely use local basis functions. Mapping configuration | preCICE - The Coupling Library. The support radius should cover maybe 3-5 neighboring vertices in every direction, but try different values here (for Gaussians see also RBF shape calculator | preCICE - The Coupling Library). What you could also try is to compute the mapping on the solid which I guess is CalculiX and therefore runs in serial from preCICE perspective. You could then also get better results with
use-qr-decomposition="yes"(but it could also be expensive).
Let us know which RBF settings your try.
This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.