Accuracy Evaluation of Experimentally Derived Dynamic Coefficients of Fluid Film Bearings Part II: Case Studies

Abstract

Six case studies are presented on the predicted uncertainties of derived bearing coefficients resulting from uncertainties in experimentally measured displacements, velocities, accelerations, and forces. In a previous paper the full set of equations, including eight linearized bearing coefficients, were derived from which bearing coefficients can be calculated based on measured quantities. Different algorithms for obtaining and reducing data are developed and presented for the cases. To demonstrate the method, bearings with known coefficients are chosen and uncertainties are theoretically applied to the measurements. The “measured” coefficients are then predicted and compared to the known values. In the first two cases the full set of equations is used for a three-lobe bearing with different excitation orbits. For case 3 the full set of equations is used for a tilt pad bearing with no cross-coupled terms. In case 4 the equations are rewritten with no cross-coupled terms for the same tilt pad bearing; for case 5 further simplifications in the velocity and acceleration terms are made and related to sinusoidal displacement. Lastly, orbits of linear motion are considered which utilize points at which two or more of the measured quantities are zero instantaneously. The six cases studied do not indicate any generalized rules. The cases do however indicate that uncertainties in the derived coefficients can vary by factors of 3 or more simply by applying different algorithms. The analysis indicates that because of the large number of measured quantities and limiting assumptions on some of the algorithms, an experimentor must prudently choose the optimum algorithm to minimize the resulting uncertainties.

Presented at the 44th Annual Meeting in Atlanta, Georgia May 1–4, 1988

Notes

Presented at the 44th Annual Meeting in Atlanta, Georgia May 1–4, 1988