System Identification Methods for Dynamic Testing of Fluid-Film Bearings

There are various system identification approaches typically used to extract the rotordynamic coefficients from simultaneously measured dynamic force and motion signals. Since the coefficient values extracted can vary significantly as a function of the system identification approach used, more attention is needed to treat this issue than is typically included in the rotor dynamics literature. This paper describes system identification and data reduction methods used for extracting rotordynamic coefficients of fluid-film journal bearings. Data is used from a test apparatus incorporating a double-spoolshaft spindle which permits independent control over the journal spin speed and the frequency of an adjustable-magnitude circular orbit, for both forward and backward whirling. For example, a least squares linear regression on the force-displacement equations of the experiment provides only one of the rational approaches to extract the anisotropic rotordynamic coefficients (stiffness, damping and fluid inertia effects). Rotordynamic coefficients are also extracted with both first and second order orbital frequency dependencies. To assess the quality of the measured signals, coherence functions are calculated to relate the time-averaged input motion signals and the time-averaged output force signals.