One more thing is that if I change nearest-neighbor to rbf-thin-plate-splines. I get bad results as the pink curve in the following figure (the pink curve here is added into the right figure in my above post to generate the following figure).
the diff between the 2 attached .xml files are:
precice-config-rbf.xml (2.6 KB)
precice-config-nearestNeighbor.xml (2.6 KB)
Some correction:
For a simple constant viscosity fluid,
where D is the rate of deformation tensor. One direction velocity gradient might cause other direction shear stress.
Note that the solid-openfoam is expected to give different results than the other solid models, as it uses a linear material model. Putting them on the same plot is probably a bit confusing.
Are you comparing the pink with the blue, or the pink with the red?
In any case, I wouldnât expect such a big difference just by changing the mapping. Something in the setup is probably wrong here.
I could only imagine that this comes from a wrong conservative mapping (scaling down / âlosingâ the force contributions for some points). With the nearest-neighbor, only one point is taken into account. With RBF, multiple. Does this give any ideas on where to look next?
what does âlinearâ mean? Does it mean the C in the following for the small displacement elastic material law constant?
I am comparing the pink with the blue. That is my focus. 
For nearest-neighbor, because the meshes from both the fluid and the solid are almost the same, it might be easy for your algorithm to transfer the physical fields, i.e. the nodal forces. This is my rough feeling. (I understand that it seems that you can also transfer the traction on a surface, i.e. the force per unit area)
For RBF, I donât know details of this mapping. Roughly speaking, is it similar to a method to do the interpolation from one mesh to the nodes on the other mesh, for example the meshless method?
