Abstract
The linearized stiffness and damping coefficients of fluid film journal bearings play a key role in predicting vibration levels and stability margins in high-performance rotating machinery. The need for accurate values of these coefficients grows in importance as higher speeds, higher loads, and new operation conditions demand more exact rotordynamic predictions. The subject of this study aims to increase the accuracy and relevance of experimentally derived dynamic coefficients by covering the three following objectives: 1) propose a new method for improving uncertainty estimations for experimentally derived dynamic coefficients, 2) propose a new method for a more accurate identification of the dynamic coefficients, and 3) develop a set of design guidelines for a test rig capable of meeting the demanding dynamic test conditions relevant to modern industrial machinery.
This dissertation first presents an analysis of uncertainty estimations applicable to dynamic coefficients obtained by single-sample single-frequency dynamic tests. The effect of the non-linearity of the dynamic coefficients on the uncertainty estimated by the Taylor Series Method is analyzed, for the first time, and the Monte Carlo Method is presented as a more accurate approach to estimating these uncertainties. Results of analyses for these two methods are presented and compared to published values from previously reported studies. This dissertation also proposes a novel method for converting the random uncertainty in the output of a sensor from the time domain to the frequency domain. It is found that this conversion is quite beneficial for improving the accuracy of the identified coefficients. In one example, uncertainty estimations of ±84% are reduced to just ±6% by using the proposed method. These analyses reveal that uncertainty estimations from the Taylor Series Method are not entirely reliable without additional checks of non-linearity; and that converting the random uncertainty from sensors to the frequency domain, by using the novel method here, is useful for achieving smaller uncertainty estimations than with traditional methodology.
Classical techniques to experimentally identify the dynamic coefficients of a fluid film journal bearing assume that the dynamic forces and displacements are measured exactly at the midplane of the bearing. However, the actual measurements are usually taken at some distance away from the bearing (miscollocation). In addition, the flexible behavior of the rotor may be important. As a result, a significant error could be included in the identified coefficients if these conditions are not considered. Therefore, this dissertation proposes a new method to accurately identify the eight dynamic coefficients of fluid film journal bearings by accounting for miscollocation of sensors and rotor flexibility. Numerical validation shows the error in the identified coefficients to be less than 0.001% when the proposed identification method is applied to three different rotor-bearing configurations. These configurations included a test bearing floating around a rotor, a rotor floating within the clearance of two identical journal bearings, and a rotor floating within two different journal bearings.
Finally, despite many years of measurement and testing, the dynamic characterization of fluid film bearings is still a field in progress; and the need for more experimental testing and validation under more demanding conditions has produced a demand for new test rigs, particularly those with a capacity to reach speeds on the order of 20,000 rpm and test bearings with diameters of 150 mm and larger. Therefore, this dissertation also develops and presents, for the first time, a guideline for researchers and engineers involved in the design of a test rig for the dynamic characterization of radial fluid film bearings. This guideline covers the design steps, starting with an analysis of the input information provided as industry needs, translating those needs into design requirements, next designing and/or selecting the critical components of the rig in order to meet the design requirements, and finally verifying the suitability of the design process. The purpose of this guideline is to inform the test rig design process, such that the designer has a directed focus and can make accurate and fast decisions. Recommendations are provided along with specific background information for the designer to consider.
