Ansys Employee
I'd use a fixed mesh and then mess with the wall velocities to mimic the case you want. That way you may get a steady state result, with layering you need a transient solution and may find you need a very long domain to get a reasonable result.