So the missing velocity component in the fluid solver information output seems to be related to the “empty” boundary condition from OpenFOAM. I switched to wall boundary conditions for front and back and now it works for different fluid densities without the floating point exception.
I guess the error is really related to the quasi-2D calculation of OpenFoam, if boundary conditions of front and back are set to "empty and the full 3D calculation of my solid solver, where the z-components should really be non existent or manually set to 0. If OpenFoam does a “full” 3D simulation the 3D solid response seems to work as intended.
I’m pretty sure that’s the reason for the behavior of the simulation, so I’ll mark this as solution
and try to carry on with different boundary conditions or at least make sure that no quasi-2D
calculation is happening.
Thanks again for the help so far.