Prediction of the Broadband Noise of a Low-Speed Axial Fan by CFD Simulations and an Empirical Wall-Pressure Spectral Model

G2 - Prediction of Axial Fan Noise by Hybrid Methods (ii)

Since several years CETIAT has started a research program with scientific and industrial partners to predict the broadband noise of axial fans, especially the blade trailing-edge noise which is the major source of broadband noise when the turbulence intensity of the inlet flow is moderate, i.e. less than about 2%. The test fan used all along the study is a propeller fan of 800-mm diameter running at 600 rpm, with 2 or 4 blades with an adjustable blade pitch angle. Two of the blades are equipped with wall-pressure transducers to obtain the input data of an analytical model of blade trailing-edge noise based on Amiet's formulation.
The first part of the paper presents an overview of the experimental and prediction results obtained from the beginning which shows that the prediction deduced from the analytical model of trailing edge-noise compares fairly well with the experiment but another important source that contributes to the overall sound power level of the fan is the tip clearance noise. The contribution of this second source besides trailing-edge noise is highlighted by tests with and without casing around the impeller as well as by fitting winglets on the blade tips to modify the flow of the tip vortex at the origin of this noise mechanism. Further work is still needed to clearly understand and assess the actual contribution of tip vortex noise to the broadband noise generated by the impeller.
In the second part of the paper preliminary results of a recent research program are presented, the goal of which is to try to predict the wall-pressure spectra close to the trailing edge of the blades by an empirical model proposed by Rozenberg in which the input data are deduced from CFD simulations (RANS model). This new work is motivated by the fact that the measurement of the wall-pressure spectra on rotating blades is tricky and costly. A replacement of this experimental procedure by a prediction using non-expensive CFD calculations could be valuable if this approach provides reasonably accurate results. Comparisons of measured and predicted wall-pressure spectra are presented and discussed.