Dynamic response of a rotor system under parametric and nonparametric uncertainties on bearings

Abstract

Journal bearings play a key role in the dynamics of a rotating machine. However, the characterization of this element, via sti ness and damping coe cients, is a challenging task due to the presence of uncertainties, either by the mathematical modeling or obtaining calculation parameters. In this dissertation, a probabilistic model is proposed to model the uncertainties in the bearings in two di erent ways. First, using the parametric strategy, which each parameter is modeled using probability theory. Subsequently, proposing a new strategy, combining a nonparametric approach to matrix perturbation with decomposition of singular values (SVD), to include the uncertainties related to the mathematical model. In parallel, a computational tool (ROSS - Rotordynamic Open-Source Software) developed by the Laboratory of Acoustics and Vibrations in partnership with Petrobrás is used for rotordynamic analysis, capable of performing this modeling. The results are obtained through this software and show how these uncertainties a ect the variation of the coe cients over a frequency range and their consequent impact on the dynamic behavior of a compressor. Finally, a post-processing of the results is done to ensure that the compressor complies with speci c international standards.