Abstract
Efficient and accurate parameter identification methods are urgently needed to develop high-performance sealing devices. Although the existing prediction methods have been widely validated, researchers are still grappling with the tradeoff between frequency resolution and computational cost. This paper proposed a Hilbert-transform-based time-frequency analysis method to identify the dynamic characteristics of annular gas seals. The frequency-sweep whirling orbit model (FSM) was adopted as the excitation signal of rotor whirling motion. The transient flow field within the seal clearance was numerically simulated by the computational fluid dynamic (CFD) solution combined with the mesh deformation technique. The empirical amplitude-modulated (AM) and frequency-modulated (FM) decomposition and the direct quadrature method were utilized to approximate the analytic signals of monitored response forces. High-resolution and frequency-dependent dynamic coefficients were determined by the Hilbert transform (HT) of the linear response-force/rotor-motion model. To validate the present method and determine the frequency-sweep mode, the dynamic coefficients of three typical annular gas seals, including a labyrinth seal (LABY), a fully partitioned pocket damper seal (FPDS), and a hole-pattern seal (HPS), were evaluated by the linear and quadratic frequency-sweep modes, respectively. The results indicate that the dynamic coefficients obtained by the quadratic FSM are all in good agreement with the experimental data, and have almost the same accuracy as the well-known multifrequency whirling orbit model (MFM). Combined with the proposed adaptive time-step scheme, the present method can obtain high-frequency-resolution dynamic coefficients of annular gas seals with a 48.6% reduction in computational time compared to MFM. Furthermore, the application of the present method in wide-bandwidth cases was also illustrated.