Inherently transient cases will rarely converge in steady-state, so taking into account most flows in real life are transient, you have quite a slim chance of getting anything to converge neatly... You puck most probably drags huge separated vortexes behind it so steady-state will behave exactly like you described