| dc.description.abstract | Modern micro-turbomachines (MTM) often employ light weight rotors operating at speeds in excess of 50 krpm. These turbomachines require radial and thrust bearings capable of handling high speeds, while also often operating in extreme environments (high and low temperature). Flexible structure air bearings, or foil bearings, provide an economical and clean (no oil or contamination) means of rotor support in this niche. To date, the most commonly utilized foil bearing is the bump-type foil bearing, which utilizes thin corrugated foil strips to support a smooth top foil on which hydrodynamic pressure builds. Despite its widespread use, each bump-foil is complex to build and model, requiring of extensive engineering knowledge and experience. As such, researchers continue to search for cheaper and less complex alternatives to the traditional bump-foil structure. Metal mesh, readily available and relatively cheap, is a viable option for radial foil bearings, but has yet to be investigated for use in foil thrust bearings. The dissertation presents the design and manufacture of a novel Rayleigh-step thrust foil bearing, whose top foil is supported by a circular layer (or layers) of compliant metal mesh screen. Static and dynamic load excitation tests (no rotor speed) with the prototype bearing reveal a structural stiffness which increases with the mass ratio of the mesh screen and decreases with the number of mesh layers used to support the top foil. Dynamic load excitation tests, up to 300 Hz., give a material loss factor γ ~ 0.2 for the mesh structure which is relatively unaffected by frequency. Foil bearings rely on structural damping from material hysteresis, and bump-type foil bearings exhibit a material loss factor which decreases quickly with an increasing excitation frequency, thus lending credence to metal mesh screen as a bearing support structure. Despite the promise of the novel Rayleigh-step foil thrust bearing, experiments with rotor speed failed several of the prototype top foils, the root-cause of which is attributed to the waviness of the metal mesh layers. This waviness unevenly bulges the thin top foil towards the spinning collar, causing solid contact between the metal top foil and collar before hydrodynamic pressure builds over the pads to separate the surfaces. The failure of several prototype foils points to the need of a robust coating for foil bearings, which can not be ignored by designers. Tests with a redesigned prototype, incorporating a circumferential taper and segmented pads, proved the bearing concept. A single pad bearing with a 55° arc extent and three layers of 40 OPI (openings per inch) mesh achieved a modest specific load of W/A = 35 kPa (per pad) at a rotor speed Ω = 40 krpm (ω·Rmid ~ 160 m/s). Further tests with a six-pad circumferentially tapered metal mesh foil thrust bearing (MMFTB) determined an ultimate load capacity of W/A ~ 25 kPa for Ω = 40 krpm. At this load capacity, the bearing temperature rise (measured via a thermocouple on the top foil backside) exceeded 110 °C. The decreased load capacity for the six-pad bearing, as compared to the single-pad bearing, results from an uneven distribution of the thrust load between the six pads, as confirmed by temperature measurements on three of the top foil undersides. Similar to the Rayleigh-step top foil, the circumferentially tapered MMFTB suffered from taper height disparities between the pads, attributed to the waviness of the mesh, which is exacerbated by stacking multiple layers. In addition to the experimental work with compliant surface thrust bearings supported by metal mesh screen, the dissertation provides a thermo-elastohydrodynamic model, validated (to some degree) with cases from the literature. However, bearing drag torque measurements and predictions for a six-pad MMFTB do not agree well. A simple Couette-flow approximation for the bearing drag torque shows that the bearing would need to operate with a uniform film thickness (hconst = 2.7 μm) to produce the measured bearing drag torque (Texp = 180 N.mm). This minute film thickness is within typical combined roughness (rotor collar + pad roughness) values, and as such likely does not produce a full fluid film during operation. The simple analysis, along with posttest photographs of the thrust collar surface, corroborates the notion that the test bearing operated in the mixed-lubrication regime, with continuous sliding contact between asperities on the rotor collar and the pads coated with a sacrificial lubricant (MoS2). A further analysis with the current model compares the steady-state performance of a MMFTB with 3 sheets of 40 OPI mesh to that of a well-known BFTB geometry from the literature. Predictions show that the MMFTB has a nearly identical drag toque to that of the BFTB for loads 10 ≤ W/A ≤ 50 kPa and a large rotor speed Ω = 70 krpm (ω·Rmid ~ 279 m/s), although it has a slightly smaller film thickness (and likely a lower ultimate load capacity). Predictions with Ω = 70 krpm (ω·Rmid ~ 279 m/s) and a specific load W/A ≤ 30 kPa show that that the MMFTB operates slightly cooler than the BFTB (ΔTmax ~ 105 vs 115 K) when no cooling flow applies to the bearing. Applying cooling flow (0 ≤ Qcf ≤ 900 LPM) through the bearing center can decrease the peak film and top-foil temperatures by ~ 40 °C, while decreasing the collar temperature by up to 10 °C. For the bearing geometry utilized in the current predictions (see Table 14), a cooling flow rate Qcf ≥ 100 LPM fully supplies the pad leading edge, such that the bearing draws no fluid from the ambient air surrounding the bearing. Increasing the cooling flow beyond this point cools the pads via forced convection in the areas beneath the pad tapers (albeit little). For the MMFTB, the mesh screens under the pad and in the land section obstruct cooling flow, such that the cooling is less effective (when compared to the BFTB). Importantly, utilizing an aluminum collar (with a high thermal conductivity, κ = 130 W/m2K) limits the thermal gradient through the collar thickness to 5 °C, reducing the possibility of thermal bending and a corresponding thermal runaway event. The dissertation adds to the archival literature on gas foil thrust bearings and provides a model for prediction of their performance. The failures detailed herein provide important lessons for foil bearing designers and researchers. Further research is needed to either qualify or disqualify metal mesh sheets as a viable underspring structure for gas foil thrust bearings, although the following document provides several cautions against their use. | en |
