Abstract
This study combines harmonic-balance simulations, modal response analysis, strain-gauge measurements, and machine-learning regression to assess the forced response of a three-row axial fan. At 10,500 rpm, Campbell-diagram analysis identifies a near-resonant third-mode rotor response to fourth-engine-order (4EO) excitation and a second-mode stator response to 13EO rotor-passing excitation. Separate fixed-blade and prescribed-motion calculations reveal the aerodynamic excitation and motion-induced aerodynamic contributions. The predicted peak alternating-stress amplitudes are 14.76 MPa for the rotor and 9.30 MPa for the stator. For the baseline stator, the prediction of 5.50 MPa in the gauge region differs from the two measured amplitudes by 11.0% and 5.3%. Following this baseline assessment, independently computed one-factor-at-a-time rotor cases are used to examine five geometric variables. The stagger angle, maximum thickness, and tip clearance produce the largest stress variations over the investigated ranges. The magnitudes of complex aerodynamic forces and the alternating stress values calculated by five regression methods are also compared. The combined workflow supports local design screening along the sampled parameter paths.