Computational Modeling of Pad Surface Irregularities in Fluid Film Bearings

Branagan, Michael, Mechanical and Aerospace Engineering - School of Engineering and Applied Science, University of Virginia
Goyne, Chris, EN-Mech/Aero Engr Dept, University of Virginia
Fittro, Roger, EN-Mech/Aero Engr Dept, University of Virginia

Bearings serve a critical role in rotordynamic systems by providing support for the rotating components. Fluid-film bearings work by supporting the rotor on a thin film of fluid. This type of bearing can be responsible for providing the majority of damping in a system, which reduces rotor vibrations. The working surfaces of these bearings are often made of a softer, sacrificial layer that is used to protect the rotor surface in the case of metal-to-metal contact and to absorb any hard particulates in the fluid that could score the journal surface. Under heavy loads, this layer can be heavily damaged resulting in bearing failure when the journal speed is too slow to support the load on a hydrodynamic film. In these cases, high pressure oil is often supplied to the working surface of the bearing via ports to lift the rotor hydrostatically. These ports are feed into machined jacking pockets or grooves, which serve to distribute the oil underneath the rotor. The bearing load is then held hydrostatically until the rotor speed is adequate to support the load on a hydrodynamic film. Often the high pressure oil supply is shut off during normal operation to reduce power loss. It is generally assumed that these pockets and grooves do not affect the hydrodynamic performance of the bearing but data on the accuracy of this assumption is limited.

This dissertation presents a comprehensive, multi-part study on the influence of hydrostatic lift features on the performance of fluid-film, journal bearings and an additional study on the applicability of different turbulence models in thin-film applications. The first study presents an examination of a two-pad, fixed-geometry bearing with a stadium-shaped/rectangular jacking pocket using CFD simulations. As the depth of the feature increases, the pressure profile is found to shift through two different regimes. The first is characterized by an increase in the load capacity of the bearing and occurs with depths shallower than 0.28x the bearing radial clearance (C_b). The second regime is characterized by a loss of load capacity and an equalizing of the pressure throughout the pocket. This regime occurs for pockets with depth up to 6.6x C_b. Finally, increasing the pocket depth beyond this depth does not change the pressure profile or further reduce load capacity. Next design-of-experiment and regression models were utilized to examine the influence of the jacking pocket geometry on the power loss, journal position, and stiffness of the film. The bearing stiffnesses varied by up to 25% from the nominal, smooth bearing case for the direct stiffnesses and 12% for the cross coupled stiffnesses. Current literature on fluid-film bearings with hydrostatic lift features has been limited to multi-recess bearings with applied high pressure oil or to thrust bearings. This paper is the first to examine the influence of the geometry of one such feature on the operational and dynamic characteristics of the bearings.

The next study expanded upon the first study by examining the influence of a pair of double diamond, jacking pockets and an hourglass-shaped, jacking groove on the same two-pad bearing. The same two regimes were found as the depth of the jacking feature was increased. The first regime occurs with depths shallower than 0.28x C_b to 0.60x C_b. For all three jacking features, the second regime occurred for pocket depths up to 7x C_b. Increasing the feature depth beyond this depth ceased to influence the pressure profile further. Next design-of-experiment and regression models were used to examine the influence of the the different jacking feature geometries on the power loss, journal position, and stiffness of the film. The presence of all three jacking feature had a minimal influence on the bearing power loss and the journal position. The power loss varied between 3% to 1% of the nominal, smooth bearing case for the different designs. The journal position varied up to 6% of the nominal case. The bearing stiffnesses varied by up to 40% from the nominal case for the direct stiffnesses and 104% for the cross coupled stiffnesses. The jacking grooves had significantly less influence than the pockets. This study expanded the applicability of the first novel study by examining two additional jacking features, which included multiple pockets and a system of grooves.

The third study was an examination of the applicability of the Reynolds equation in analyzing bearings with jacking grooves. Several methods were examined to allow the use of Reynolds equation. Biasing the element distribution towards the jacking features was crucial for keeping the analysis runtime within a reasonable limit while still achieving accurate results. The Reynolds equation was compared with CFD results for several different pocket geometries. The Reynolds equation accurately captured the pressure profile, despite the lack of fluid inertia, when an optimization was performed on the journal position. Furthermore, the increase in rotor eccentricity compared with the CFD equilibrium position was limited to 4\% of the radial clearance. Reynolds equation is thus an excellent tool for efficiently analyzing bearings with jacking features and will result in a slightly more conservative bearing design due to the lack of inertia. Multiple studies have been performed on hydrostatic lift features which utilized the Reynolds equation for capturing the hydrodynamic behavior in the bearing. This has been done without any justification for its use. This study was the first to examine the applicability of the Reynolds equation and its assumptions across a hydrostatic lift feature. This study will increase the credibility of these studies by providing justification for its use.

Computational fluid dynamics (CFD) is a powerful tool for examining the behavior in a fluid-film bearing. Modeling turbulence in a bearing is challenging due to the wide range of Reynolds number that can occur in a single bearing. The last study examines three different methods for modeling turbulence, along with the laminar assumption, in a four-pad journal bearing using CFD. The predominant model that is used with CFD in prior literature is a two-equation turbulence model. This is often done regardless of the Reynolds number or presented with inadequate data for calculating the Reynolds number. A pad model was developed for a four-pad, tilting-pad, journal bearing. Each model is simulated across a broad range of Reynolds numbers. The two-equation is not always the best choice and justification should be presented for the choice of turbulence model. The one-equation turbulence model has the advantage of accurately predicting the behavior of laminar flow and providing a better prediction near the onset of turbulence. This is highly advantageous in bearings where portions of the film can be turbulent while other portions are laminar. A single turbulence model can be provided for the whole bearing which will greatly reduce simulation runtimes.

PHD (Doctor of Philosophy)
Fluid-film bearings, Journal Bearings, Turbulence, Hydrostatic lift features
All rights reserved (no additional license for public reuse)
Issued Date: