Spectral decomposition of the aerodynamic noise generated by rotating sources

A method is posed for separating the noise emitted by an aerodynamic source from propagation effects using spectral decomposition. This technique is applied to the power spectra of a fan measured at several rotational speeds. Although it has been conceived for rotating sources as turbomachinery rotors, the method may be easily applied to low speed stationary sources such as jets and flows in stators and about isolated airfoils. Based on the similarity theory, a clear description of the structure of the power spectrum of the received noise is given and the effect of rotational speed variations is considered as a means to obtain a data set suitable to perform the spectral decomposition. The problem is analyzed in order to clarify possibilities and limitations of the method and then an algorithm is presented which is based on the solution of the derived equations. Particular care is devoted to both the numerical details and the operative aspects. The validation of the algorithm is performed by means of numerically generated input data. Next, in order to verify the ability of the method in separating scattered from emitted sound, an automotive cooling fan has been tested in the DIMSET hemi-anechoic room in a free-field configuration and with a shielded microphone. These two apparently distinct spectra collapse to within less than 2 dB after the spectral decomposition has been performed. The tests prove the ability of the method despite the modest quantity of input data.