| Large-eddy simulations of round forced impinging jets at laboratory scale are performed, as a simplified model of the helicopter rotor wake; the goals were to understand the dynamics of the vortices, and to help develop engineering correlation and models. First, I studied the flow over smooth surfaces, then the Reynolds number effects and the performance of lower-level models in this flow were explored. Finally, I investigated surface roughness effects.;The interaction between the vortices and the wall resulted in the formation of secondary vorticity that was lifted up and rolled around the primary vortex. These large-scale vortices play an important role in determining the flow. The dynamics of the vortex evolution are initially dominated by an inviscid mechanism. As the interaction with the wall begins, however, the contribution of turbulent diffusion to the vorticity transport (TVD) becomes the leading term weakening the primary vortex. This phenomenon is extremely robust, leading to similar vortex decay rates regardless of the level of turbulence in the jet and the nature of the near-wall flow. The mean-flow three-dimensionality due to a short-wavelength azimuthal instability also plays a role in the evolution of the vortices.;Surface roughness significantly modified the near-wall flow, including displacement of the velocity profile, increase of wall stress, and amplification of turbulence levels. The near-wall flow changes due to the roughness are advected to the outer layer during the separation of the secondary vorticity, rather than being confined in the vicinity of the roughness elements. The robust TVD mechanism, however, prevent the vortices from being strongly affected by these changes.;This flow is particularly challenging for turbulence models. The tested RANS models exhibited poor performance. The error sources were first identified as their incapability in predicting both the shear-layer instability, and the local unsteady separation. Secondly, the well-known shortcomings of the eddy-viscosity assumption in wall jets were amplified here by the counter-gradient transport during the vortex evolution. Finally, since the TVD term depends on the second derivative of the shear stress, the RANS model errors were naturally amplified. The WMLES predicted vortex decay in good agreement with the resolved LES data. |
