No abstract

The increase of the thrust/weight ratio of aircraft engines is extremely restricted by different 3-D flow loss mechanisms. One of them is the corner separation that can form at the junction between a blade suction side and a hub or shroud. In this paper, in order to further investigate the turbulent characteristics of corner separation, large-eddy simulation (LES) is conducted on a compressor cascade configuration using NACA65 blade profiles (chord based Reynolds number: 3.82 × 10 5), in comparison with the previous obtained experimental data. Using the shear-improved Smagorinsky model as subgrid-scale model, the LES gives a good description of the mean aerodynamics of the corner separation, especially for the blade surface static pressure coefficient and the total pressure losses. The turbulent dynamics is then analyzed in detail, in consideration of the turbulent structures, the one-point velocity spectra, and the turbulence anisotropy. Within the recirculation region, the energy appears to concentrate around the largest turbulent eddies, with fairly isotropic characteristics. Concerning the dynamics, an aperiodic shedding of hairpin vortices seems to induce an unsteadiness of the separation envelope.

A large-eddy simulation is carried out on a rod-airfoil configuration and compared to an accompanying experiment as well as to a RANS computation. A NACA0012 airfoil (chord c = 0.1 m) is located one chord downstream of a circular rod (diameter d = c/10, Red = 48 000). The computed interaction of the resulting sub-critical vortex street with the airfoil is assessed using averaged quantities, aerodynamic spectra and proper orthogonal decomposition (POD) of the instantaneous flow fields. Snapshots of the flow field are compared to particle image velocimetry (PIV) data. The acoustic far field is predicted using the Ffowcs Williams & Hawkings acoustic analogy, and compared to the experimental far field spectra. The large-eddy simulation is shown to accurately represent the deterministic pattern of the vortex shedding that is described by POD modes 1 & 2 and the resulting tonal noise also compares favourably to measurements. Furthermore higher order POD modes that are found in the PIV data are well predicted by the computation. The broadband content of the aerodynamic and the acoustic fields is consequently well predicted over a large range of frequencies ([0 kHz; 10 kHz]).