Fault Simulation in a Gearbox Using Finite Element Model Reduction Techniques L. Deshpande, N. Sawalhi, R.B. Randall School of Mechanical and Manufacturing Engineering, University of New South Wales, Sydney 2052 ABSTRACT A lumped parameter model (LPM) has previously been used to model gear and bearing faults in a gearbox. It was found that simulated signals for localized bearing faults had good correspondence with measured ones in a narrow high frequency band demodulated for envelope analysis. However, for extended faults, there is more interaction with the gearbox structure, as the fault modulates the gear meshing, and the correspondence was poorer. Forces at the bearings from the LPM model were applied to a finite element (FE) model of the casing, and the results improved but were still deficient. This paper has benefited from CMS (component mode synthesis) based FE model reduction techniques to reduce the FE model of a gearbox casing into manageable and well representative degrees of freedom of the casing. The reduced model of the casing was embedded with the LPM of the internals, which was previously obtained with the aid of Simulink® and has the capability of capturing time-varying stiffness nonlinearities arising from gears and bearings and has also the capability of simulating geometrical faults (spalls) for both gears and bearings. In order to extend the validity of the combined /reduced model, the forces are extracted from combined/reduced model and convolved with the impulse responses corresponding to the FRFs of the whole gearbox. The results show the improvements obtained through combining the reduced model of the casing with the LPM model giving a much better correspondence with measured signals. This has been verified for both the fault free and faulty cases. 1. Introduction The application of Machine Condition Monitoring (MCM) techniques for the detection, diagnosis and prognosis of incipient faults; has traditionally been an integral part of the maintenance strategy which is aimed at improving the productivity. However, with the advances in the computational techniques, it is now possible to simulate the faults and gain an insight into the complex interactions between the various components. Fault simulation allows the analyst to vary the location and dimension of the fault, control the operating conditions and gather data without experiencing the costly failures, thus eliminating the need for large numbers of experiments. The gathered data which is cheaply available can be used to train neural networks in the diagnostic and prognostic process. This paper describes fault simulation in a geared transmission, one of the critical equipments found virtually in all industries. A typical gearbox is characterised by the complex dynamic interaction among the individual components such as gears, shafts, bearings and the casing. Lumped parameter models (LPM) are widely utilized to simulate the dynamic behavior of mechanical systems such as gearboxes. LPMs give reasonable representation of the dynamics of the system if masses can be lumped at certain locations, such as gears, shafts, bearings etc. LPMs have the advantage T. Proulx (ed.), Linking Models and Experiments, Volume 2, Conference Proceedings of the Society for Experimental Mechanics Series 5, 399 DOI 10.1007/978-1-4419-9305-2_30, © The Society for Experimental Mechanics, Inc. 2011
RkJQdWJsaXNoZXIy MTMzNzEzMQ==