I am trying to make a 2D model of the Magnus effect of a rotating cylinder in Fluent5. The results Fluent is giving me are 15% to 50% lower than the experimental data given from Rouse and Prandtl. My entire domain model is large enough to prevent wall effects. I used grid packing next to the cylinder with the heigth of the first row along the wall of the cylinder equal to 1e-5m. I am using the k-epsilon RNG viscous model with 2-layer zonal near-wall treatment. My solutions converged at e-6 residuals. I am sure that my results are not grid dependent, and I am using the default values for atmospheric conditions of air. I am unable to find a better method to bring my results closer to experimental answers and would appreciate any suggestions to assist me.

Hi Dahvid Brown! I think that Realizable k-e model will be serve more accuracy. (Fluent 5.0 User Guide vol.2 9.2.7) Realizable k-e model provides good performance for flows involving rotation, seperation, recirculation... Good luck to you!