A mode-based fast reconstruction method for the generation of 'buzz-saw' noise sources in transonic fans

摘要

Small but inevitable non-uniformities always exist in real fans. They are responsible for the non-uniform signatures observed in the up-propagating rotor-alone pressure field. In the transonic regime, shock trains emerge in this pressure field to form the source of "buzz-saw" noise. Blade non-uniformities destroy the periodicity of blade passages. Thus, full-annulus three-dimensional simulations are needed to numerically study the buzz-saw noise generated by real rotors. Moreover, due to the random property of blade non-uniformities, statistical analysis is needed to reveal the general characteristics of buzz-saw noise generated by the same series of rotors. This leads to huge and unnecessary computational costs. A duct-mode-based method is proposed to reduce the computational costs. Given a fan rotor with arbitrary stagger variation on each blade, the rotor-alone pressure field is reconstructed by three CFD simulations of the "basic rotors". This method is applied to the modified NASA Rotor 67 with random stagger variations. Two sections at different axial stations are tested. The rotor-alone pressure field at each station is validated by three-dimensional RANS simulation. In case the blade variations are small, reconstruction error of this method is found to be less than 0.5 dB when compared with full-annulus RANS simulations. Then, a number of random staggered rotors are created by adding stagger variations to Rotor 67. Rotor-alone pressure profiles of the these rotors are generated. General patterns of the buzz-saw noise generated by them are investigated by statistically analyzing their pressure profiles. Nonlinear resonance is observed during the propagation of buzz-saw noise. This explains the mechanism underlying the generation of buzz-saw noise. Additionally, it is found that stagger variation leads to the increase of buzz-saw noise by both increasing the total sound power and decreasing the decaying ratio of the noise. (C) 2020 Elsevier Ltd. All rights reserved.