The accuracy of rotordynamic analysis of aeroengine structures is typically limited by a trade-off between the capabilities and the computational cost of the squeeze-film damper (SFD) bearing model used. Identification techniques provide a means of efficiently implementing complex nonlinear bearing models in practical rotordynamic analysis; thus facilitating design optimisation of the SFD and the engine structure. This thesis considers both identification from advanced numerical models and identification from experimental tests. Identification from numerical models is essential at the design stage, where rapid simulation of the dynamic performance of a variety of designs is required. Experimental identification is useful to capture effects that are difficult to model (e.g. geometric imperfections). The main contributions of this thesis are: • The development of an identification technique using Chebyshev polynomial fits to identify the numerical solution of the incompressible Reynolds equation. The proposed method manipulates the Reynolds equation to allow efficient and accurate identification in the presence of cavitation, the feed-groove, feed-ports, end-plate seals and supply pressure. • The first-ever nonlinear dynamic analysis on a realistically sized twin-spool aeroengine model that fulfills the aim of taking into account the complexities of both structure and bearing model while allowing the analysis to be performed, in reasonable time frames, on a standard desktop computer. • The introduction and validation of a nonlinear SFD identification technique that uses neural networks trained from experimental data to reproduce the input-output function governing a real SFD. Numerical solution of the Reynolds equation, using a finite difference (FD) formulation with appropriate boundary conditions, is presented. This provides the base data for the identification of the SFD via Chebyshev interpolation. The identified 'FD-Chebyshev' model is initially validated against the base (FD) model by application to a simple rotor-bearing system. The superiority of vibration prediction using the FD-Chebyshev model over simplified analytical SFD models is demonstrated by comparison with published experimental results. An enhanced FD-Chebyshev scheme is then implemented within the whole-engine analysis of a realistically sized representative twin-spool aeroengine model provided by a leading manufacturer. Use of the novel Chebyshev polynomial technique is repeatedly demonstrated to reduce computation times by a factor of 10 or more when compared to the basis (FD) model, with virtually no effect on the accuracy. Focus is then shifted to an empirical identification technique. Details of the commissioning of an identification test rig and its associated data acquisition system are presented. Finally, the empirical neural networks identification process for the force function of an SFD is presented and thoroughly validated. When used within the rotordynamic analysis of the test rig, the trained neural networks is shown to be capable of predicting complex nonlinear phenomena with remarkable accuracy. The results show that the neural networks are able to capture the effects of features that are difficult to model or peculiar to a given SFD.
展开▼
