Stability of rotors supported by tilting pad journal bearings.

摘要

Rotor instability, a self-excited vibration phenomenon, remains a serious and costly problem with turbomachinery. Avoidance is attempted through rotordynamic modeling as well as testing. However, large modeling uncertainties exist, and common shop testing methods provide no correlation with predictions. For tilting pad journal bearings, the central uncertainty lies in whether or not they should be represented with frequency dependent dynamics. The two existing representations, one being a simplification of the other, can produce very different stability predictions. Such predicted differences suggest that stability testing should be an effective platform for distinguishing between the two representations. Yet there exists no clear understanding of what techniques are appropriate for measuring rotor stability. A method for stability measurement of rotor systems is developed in this investigation. Applying this method, experimental evidence is provided to help reduce the uncertainties surrounding the stability of rotors supported by tilting pad journal bearings.; Using simulated rotor systems with known damped eigenvalues, the challenges facing the fundamental system measurement problem, damping ratio estimation, are examined. The close proximity of forward and backward rotor modes is shown to diminish the accuracy of two popular single degree of freedom damping estimators, amplification factor and logarithmic decrement. Two widely used multiple degree of freedom techniques, backward autoregression (AR) and the prediction error method (PEM), demonstrate very reliable accuracy. Based in the time domain and requiring only response output measurements, backward AR in conjunction with blocking excitation is concluded to be an effective method for rotor stability measurement.; A rotor supported by two tilting pad journal bearings on flexible supports serves as the test apparatus. Two magnetic actuators are also incorporated, one providing external forcing and the other applying destabilizing cross-coupled stiffness. The rotor model is reconciled using experimental modal analysis. Polynomial transfer function models for the flexible supports are developed from frequency response function measurements. Control and estimation of the actuators' forces is done using calibrated circuit models.; For two bearing preload designs, measurements of pad temperatures, shaft centerlines, unbalance response and stability are compared with modeling predictions. Base and threshold stability measurements are performed at speeds above and below the first critical speed. The first forward and backward modes' stability are also measured as a function of cross-coupled stiffness. Predictions based on frequency dependent tilting pad bearing dynamics exhibited significantly better correlation with the stability measurements than those assuming frequency independent dynamics. However, unbalance response and stability correlations indicated that bearing damping and support anisotropy are overpredicted. These deviations emphasize the value of conducting such measurements to avoid rotor instability problems.