My notional problem (https://www.simscale.com/projects/mshorter/uniform_relative_velocity_inlet-_straight_pipe-_high_re-_rossby_number_unity-_mesh_study/ ) is a circular pipe along the z axis, with the inlet centered at the origin, diameter 0.1m, length 1m with uniform rotation. The body rotates about the x axis, with the axis located at the origin (which is at the inlet of the duct). For now, the solid walls are slip, the turbulence is set to laminar, and the viscosity set very small (1e-9) to model an approximation of an inviscid flow. Inlet velocity is 10 m/s.
I have tried MRF but that does not give sensible results or good convergence: even if I set rotation to zero, the residuals do not converge and the result looks dramatically different from a stationary case. AMI and solid body rotation are both time marching which increases computation cost significantly of course and do not behave well either. I know that MRF and AMI are really made for multiple bodies/domains (e.g. rotor and stator) so I think having one domain that rotates could be the problem here.