Comparison of perturbed Reynolds equation and CFD models for the prediction of dynamic Coefficients of sliding bearings

Abstract

The accuracy and utility of rotordynamic models for machinery systems are greatly affected by the accuracy of the constituent dynamic bearing models. Primarily, the dynamic behavior of bearings is modeled as linear combination of mass, damping, and stiffness coefficients that are predicted from a perturbed Reynolds equation. In the present paper, an alternative method using Computational Fluid Dynamics (CFD) with a moving boundary is used to predict the dynamic coefficients of slider bearings and the results are compared with the more commonly employed perturbed Reynolds equation model. A linear slider bearing geometry is investigated and the results serve as precursors to similar investigations involving the more complex journal bearing geometries. Time and frequency domain methods for the estimation of dynamic coefficients are shown to give comparable results. For CFD with a moving boundary, temporal inertia is found to have a significant effect for a reduced, squeeze Reynolds number less than one. The temporal inertia effect is captured through an added mass coefficient within the dynamic model of the bearing.