Experimental validation of a model of fan trailing-edge noise

Abstract

Broadband noise is a major part of the noise radiated by industrial and domestic fans. For low-speed fans broadband noise contribution to the overall A-weighted noise level is often much higher than that of the tonal noise at the blade passage frequency and its harmonics. One of the main mechanisms of broadband noise is trailing-edge noise, which is due to the convection of the turbulent flow past the blade trailing edge. This paper presents results of a research program in progress, which aims at validating a model that predicts broadband noise levels of rotating blades. Trailing-edge noise prediction is made using an analytical model deduced from an extension of Amiet's formulation. The input data to the model are the frequency spectrum and the spanwise correlation length scale of the wall pressure fluctuations on the blade suction side close to the trailing edge. This model was successfully validated on single fixed airfoils at different angles of attack in an anechoic wind tunnel. The input data were measured with wall pressure transducers on the airfoil suction side. In the present study, the validity of the trailing-edge noise model is assessed for a 800-mm axial flow fan which is used without casing in order to avoid the tip clearance noise contribution. Wall pressure fluctuations are measured with small pressure transducers flush mounted on two rotating blades, using a slip ring to transmit the pressure signals to the frequency analyser. A comparison of measured and predicted far-field sound pressure spectra at different observation angles is presented for two blade pitch angles 20° and 30°. The prediction proves to be quite good at 30° but strongly underestimates fan noise levels over a large frequency range at the lower blade angle. In this case another sound source related to the blade tip vortex is detected.