| description abstract | The stiffness and damping coefficients of fluid film bearings play a key role in predicting levels of vibration and stability margins in high-performance industrial rotating machinery. However, variability in the coefficients calculated by numerical bearing codes creates inaccuracies in the rotodynamic predictions. Therefore, there is a strong need for accurate experimental measurement of bearing coefficients for validation purposes. This work examines new propagation uncertainty strategies in bearing coefficients estimation, and for the first time examines the effect of the nonlinearity of the dynamic coefficients on the experimental uncertainty estimated by the Taylor Series Method. The Montecarlo method is presented as a more accurate approach to estimating experimental uncertainty. The results of the analyses are compared to published values from previously reported studies. This paper also proposes a novel method to convert the random behavior in the output of a sensor in the time domain to the frequency domain. It is found this conversion is quite beneficial for the accuracy of the coefficients; in one example, uncertainty estimations of ±84% are reduced to just ±6%, when using the proposed method. This work also reveals that the uncertainty estimations from the Taylor Series Method are not entirely reliable, without additional checks of nonlinearity, and that converting data to the frequency domain, by using the novel method here, is useful for achieving smaller uncertainty estimations than with traditional methodology. | |
