Hélène Posson

Several technological effects on tonal fan noise prediction

This work aims at evaluating some technological effects usually not included in analytical or semi-analytical tonal fan noise prediction models. The methodology consists in comparing unsteady Computational Fluid Dynamics data, taking successively into account some realistic effects, with the results from a cascade response based model for rotor/stator interaction. The simulations are performed with the turbomachinery flow solver Turb’Flow on a simplified stator vane cascade. They allow evaluating and discussing the effects of both model hypotheses of no vane thickness and inviscid flow on the acoustic sources predictions. The effects of these assumptions are then assessed in terms of modal acoustic powers radiated within the duct. A similar study is performed on a realistic low pressure compressor stage. The effects of the stator vane geometry as well as the real flow configuration on the acoustic sources, extracted during the unsteady simulation, are quantified relatively to the analytical prediction of the vane loading. Finally a comparison of the acoustic powers given by both methods results in an estimation of the technological effects on the propagated noise.

Wake model effects on the prediction of turbulence-interaction broadband noise in a realistic compressor stage

The effects of the wake modelling on the prediction of the broadband noise generated by the impingement of the turbulent wakes on a stationary blade row are studied. The analysis focuses on the description of the wake shape that is usually approximated by a Gaussian curve for acoustical purposes. The prediction of the broadband noise is achieved using an analytical acoustic model dedicated to turbomachinery configurations. The characteristics of the model regarding the wake treatment are described. In order to provide a realistic reference for the wake shape, a low-speed axial compressor stage is analyzed using numerical RANS simulation. The unsteady flow-field structure is presented. The simulation is validated using a comparison with experimental data, with special attention paid for the wake numerical prediction. The shape of the rotor wakes is thoroughly described. The comparison between the simulated wake and the corresponding Gaussian approximation is detailed, showing differences between the two approaches. The broadband predictions yielded by the two wake models are finally compared. The use of the Gaussian approximation of the wakes shape for broadband noise prediction purposes is validated in the studied realistic configuration.

Effect of Rotor Shielding on Fan-Outlet Guide Vanes Broadband Noise Prediction

An analytical model for the rotor–stator broadband noise is improved by accounting for the rotor acoustic shielding. Both the analytical models for the stator broadband noise and for the rotor scattering are strip-theory approaches based on previously published formulations of the three-dimensional unsteady blade loading for a rectilinear cascade. Specific treatment is introduced to address the behavior around the cut on frequency of the duct modes both in the noise generation and in the noise scattering. A simple effect of the swirl between the rotor and the stator is considered using a Doppler shift in frequency. The power spectral density of the acoustic power and the sound pressure level at the duct wall are studied for the NASA source diagnostic test fan rig. The separate effects of the shielding by a cascade of the rotation of the rotor and of the swirl are investigated. Corotating modes dominate in the inlet. The rotor shielding decreases the acoustic power at intermediate frequencies but increases...

Experimental Validation of a Cascade Response Function For Fan Broadband Noise Predictions

B = number of vanes c = vane chord length Ci = cascade i, i 1 (B 49) or i 2 (B 98) c0 = speed of sound Em; = duct eigenfunction of the mode m; exd = unit vector in the axial direction of the duct f = frequency H i m x = Hankel function of kind i of order m Kd = vector of wave numbers in the duct reference frame Kx = streamwise aerodynamic wave number, !=Uxd kxd0 ; kzd0 = axial and radial wave numbers of the excitation in the duct reference frame, Kd0 Kd kzc0 = spanwise wave number in the cascade reference frame, Q Kd0j3 k0 = acoustic wave number, !=c0 k x;m ;d = axial wave number of the duct mode m; k x;m ;cd r = axial wave number of the duct mode m; in the cascade frame before the rotation of sweep angle Li = measured acoustic power with a turbulence grid installed (i mes) and with no grid (i 0) lr ! = spanwise correlation length Mxd = axial Mach number, Uxd=c0 m; = azimuthal and radial orders of a duct mode mg = azimuthal order of an incident turbulent component nc = unit vector normal to the blade Q = transformation matrix from duct to cascade reference frame, Qij Rc = reference frame attached to the rectilinear cascade at the radius r after rotation of stagger, lean, and sweep angle Rcd = reference frame attached to the rectilinear cascade at the radius r after rotation of stagger and lean angle Rd = stationary duct reference frame RH = hub radius Rm = mean radius at midspan, RT RH =2 RT = tip radius r = current radius St = Strouhal number, fRT=Uxd Ti = turbulence grid i, i 1 (Tu 3%) or i 2 (Tu 5:5%) Tu r = local turbulent intensity at radius r, urms r =Ux r t = time Ux r = measured axial mean flow velocity at radius r Uxd = nominal value of the axial mean flow velocity at midspan u = fluctuating velocity vector in the duct stationary reference frame u = streamwise velocity fluctuation urms = root-mean-square value of the streamwise velocity fluctuation, u p