Reduced-Order Modeling in Rotordynamics and Its Robustness to Random Matrix Perturbation

Abstract

Abstract The aim of this paper is to apply and compare four reduced-order modeling strategies to compute the unbalanced vibration response of a rotating machine and evaluate its robustness to random matrix perturbation. The full finite element model of the rotor is built using a rotordynamic open source software (ROSS), and is reduced through different methods, namely: (1) modal reduction, (2) Krylov subspace, (3) Guyan reduction, and (4) system equivalent reduction–expansion process (SEREP). To evaluate the robustness of the obtained results, this paper proposes to perturb the stiffness matrix obtained using the reduced-order models applying the random matrix theory. A simple rotor (three discs) and a more complex gas turbine model (21 discs) are analyzed. Results show that Guyan is not the most appropriate reduction technique for the systems analyzed, but the other three strategies yield good results. In addition, the same random perturbation in the reduced-order stiffness matrices produces a similar level of uncertainty on the stochastic unbalanced responses.