Unsteady simulations of the Wellborn diffusing S-Duct

Diffusing S-shaped ducts can be important components in the integration of aero-engines and airframe design. Wellborn et al., at the Internal Fluid Mechanics Facility at the NASA Lewis Research Centre, conducted a comprehensive experimental study of a Diffusing S-duct design. Here, a range of computational methods are applied to the geometry that was used in this experiment, along with two more on geometrically similar experiments which were conducted at ONERA. Some RANS turbulence closure models are found to be effective in a few cases for predicting the behaviour of some of the parameters in which a designer may be interested—but there are some flow features which these approaches are simply not able to capture. Full direct numerical simulation is ruled out as too computationally demanding, and the high Reynolds number means that this is also true for wall resolved LES. Instead, a zonal RANS/LES approach is used to cover over the finest scales within the boundary layer. This method is shown to be more effective, without tuning, than the RANS techniques on the duct designs. The prediction of separation, however, is still dependent on RANS models, and remains difficult to accurately capture. The use of unsteady methods also allows the exploration of time dependent flow structures reaching the fan face, which are completely missed by the steady RANS. Proper Orthogonal Decompositon has been used to extract these modes from noisy unsteady probe data, and a parallel project is exploring the effects that these may have on fan performance.