Modal Strain Energy Approach Applied in an Uncertainty Propagation Analysis Dedicated to Rotating Machines

Abstract

The vibration responses of rotating machines can be affected by the inherent uncertainties of their parameters. Therefore, the study of uncertainty quantification in rotating machines is relevant aiming at both increasing the performance of the machine and preventing failures. Among the various stochastic approaches used to model the uncertainties affecting the system, the stochastic finite element method received attention in the last few years. Uncertain parameters are commonly discretized by using Karhunen-Loève expansion together with Latin Hypercube and Polynomial Chaos. In the present contribution, the uncertain information is treated by using the Latin Hypercube approach. In this context, uncertainty analysis based on stochastic methods is an expensive task when applied to rotating machines of industrial interest. Thus, reduced models become an interesting alternative. The Modal Strain Energy (MSE) approach is commonly used due to the representativeness of the obtained reduced model and computational time savings. In this context, this paper is dedicated to the analysis of the uncertainties that affect the dynamic behavior of a horizontal rotating machine, composed by a flexible shaft containing two rigid discs and supported by two ball bearings. The finite element model of the considered rotor system was reduced by using the MSE approach. The obtained results demonstrated the efficiency of the methodology conveyed.