Proper Orthogonal Decomposition Method for the Prediction of Fan Broadband Interaction Noise

Abstract

The turbulent wake generated by a rotor interacting with outlet guide vanes (OGVs) is one of the dominant broadband noise sources in a turbofan engine. An accurate representation of the rotor wake turbulence is therefore important for the reliable prediction of the rotor–OGV interaction noise. This paper presents a turbulence synthesis method based on a spectral proper orthogonal decomposition (POD) representation of the turbulent wake that aims to reproduce the desired velocity cross-spectrum along the OGV leading edges where noise is emitted due to turbulence–OGV interaction. The method is first developed in the frequency domain based on a superposition of vortical modes with the appropriate amplitudes. Fourier modes and POD modes are proposed to represent the two-point velocity spectrum. The POD modes will be shown to be highly efficient in reconstructing the flowfield near the tip region where a large-scale coherent structure is present. An extension of the POD synthetic turbulence method to the time domain is also presented by means of a white noise filtering technique to allow the generation of time-varying velocity signals with the desired cross-spectral characteristics. The results show that both the one- and two-point statistics can be closely reproduced. The proposed frequency-domain POD synthetic turbulence method for fan broadband noise prediction for a realistic fan–OGV configuration is illustrated by the use of a frequency-domain linearized Navier–Stokes solver to predict the sound power radiation due to each vortical mode. Overall sound power levels at a number of discrete frequencies are predicted and compared against measured noise data. Agreement is found to be within the uncertainty of the noise sound power measurement.