Nonlinear Stability Analysis of a Generic Fan With Distorted Inflow Using Passage-Spectral Method

Abstract

The dependence of the aerodynamic stability of fan blades with the nodal diameter and amplitude of the inlet perturbations is studied. The analysis is conducted using a block-wise spatial Fourier decomposition of reduced-passages to reconstruct the full-annulus solution. The method represents very efficiently the unsteady flow generated by inlet non-uniformities. The explicit spatial Fourier approximation is exploited to characterize the relevance of each nodal diameter of the inlet perturbation in the fan stall process and study the nonlinear stability in a harmonic by harmonic basis. This approximation allows studying the contribution to stall of each circumferential mode separately. The methodology has been assessed for the NASA rotor 67. The maximum amplitude of total pressure distortion at which the compressor becomes unstable and triggers a stall process has been mapped. It has been proven that despite the complexity of a screen-induced total pressure distortion the only relevant parameter for the nonlinear stability of the fan is the most unstable nodal diameter. The equivalence in terms of stability between realistic distortion screens and single harmonic distortions has been assessed. Full-annulus simulations have been conducted to assess the accuracy of the simplified nonlinear stability limit. It is concluded that performing a nonlinear simulation with the proper single harmonic perturbation is enough to assess fan stability. The total pressure error with respect the full annulus simulation including a screen-induced pressure deficit at the intake is below 10%. It is shown that for the NASA rotor 67 running at the nominal speed the most unstable nodal diameter is the first. This study not only shows a reduction in computational time to assess nonlinear fan stability by a factor of seven but also creates an efficient methodology for understanding the nonlinear instability of fans due to inlet distortion profiles.