I am testing the mapping method in 3D coupling. Three methods (N-N, N-P and rbf-thin-plate-splines) are tested, but all methods perform worse. The exported results are shown as following:
config.yml (241 Bytes)
precice-config.xml (3.9 KB)
preciceDict.txt (668 Bytes)
For N-P mapping, the magnitude of force on solid mesh is so small.
For N-N mapping, the range of force magnitude can be received. But the mapping position can’t match well.
For rbf mapping, the value of force is strange.
Firstly, I want to know the reason why the mapping performs worse. Then I want obtain some advices to get the better mapping.
I am not sure I understand where each domain is and where are the interfaces. I assume this is some kind of membrane between two fluid domains, is that correct? Are these the inner and outer walls of the membrane, or really a Fluid and a Solid mesh? Or is it a completely artificial case to test the mapping?
Could you please also specify where you are splitting the domain, in case you are running in parallel?
You did very well exporting the data already. You also have correctly used consistent mappings for displacement, and conservative mappings for forces.
One thing that may be wrong in the nearest projection mapping is that the
preciceDict does not specify to provide mesh connectivity. This would mean that NP becomes essentially a NN mapping.
You can read more about the different mapping methods in the preCICE v2 paper: https://open-research-europe.ec.europa.eu/articles/2-51/v2
You can also use ASTE to study the different mapping methods, excluding any issues with OpenFOAM, CalculiX, and their adapters: Artificial Solver Testing Environment (ASTE) | preCICE - The Coupling Library