Development of a Reduced Order Model Based on Aerodynamic Influence Coefficients to Simulate Aeroelastic Phenomena in Axial Compressor Blades

Abstract A reduced order model (ROM) is proposed to simulate the aeroelasticity of compressor blades in both coupled and uncoupled mode at low computational cost. The ROM describes the structural dynamics of blades using the mode superposition technique, while effective aerodynamic forces are modeled using various system identification methods. The ROM takes advantage of the aerodynamic influence coefficients (AIC) to model the aerodynamic loads of vibrating blades. System identification of unsteady aerodynamic forces is performed using ARX polynomial models, nonlinear ARX polynomials, Hammerstein-Wiener models and recurrent neural networks. Two different training signals are tested, and their performance is compared. The uncoupled results are validated against traditional Fourier method, showing good agreement in estimated aerodynamic damping. The different formulations of the coupled ROM are then compared to the coupled full order model (FOM) of reference: consistent predictions for the damping ratio are achieved, with a substantial reduction of the computational cost with respect to full order models of reference.