Case 1144261305

FACTORS INFLUENCING THE PERFORMANCE OF FOIL GAS THRUST
BEARINGS FOR OIL-FREE TURBOMACHINERY APPLICATIONS
by
BRIAN DAVID DYKAS
Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy
Dissertation Advisor: Dr. Joseph M. Prahl
Department of Mechanical and Aerospace Engineering
CASE WESTERN RESERVE UNIVERSITY
May 2006
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of
Brian D. Dykas
______________________________________________________
candidate for the Ph.D. degree *.
Joseph M. Prahl
(signed)_______________________________________________
(chair of the committee)
Edward B. White
________________________________________________
J. Iwan D. Alexander
________________________________________________
Robert L. Mullen
________________________________________________
________________________________________________
________________________________________________
March 23, 2006

(date) _______________________
*We also certify that written approval has been obtained for any
proprietary material contained therein.
Table of Contents
Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
1 Introduction 1
1.1 Oil-Free Turbomachinery Applications . . . . . . . . . . . . . . . . . 1
1.2 Enabling Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Present Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Background of the Art 11

2.1 Foil Air Bearing Development . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Current State of the Art Practices . . . . . . . . . . . . . . . . . . . . 16
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3 Gas Film Characteristics 20

3.1 Hydrodynamic Features . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Thermal Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Structural Response and Effects . . . . . . . . . . . . . . . . . . . . . 24
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4 Methods 26
4.1 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.1.1 Thrust Bearing Test Rig . . . . . . . . . . . . . . . . . . . . . 26
4.1.2 Test Specimens . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.1.3 Cooling Techniques . . . . . . . . . . . . . . . . . . . . . . . . 35
4.1.4 Operating Torque and Power Loss Measurements . . . . . . . 36
4.1.5 Load Capacity Testing . . . . . . . . . . . . . . . . . . . . . . 38
4.1.6 Wear Measurements . . . . . . . . . . . . . . . . . . . . . . . 38
iii
4.1.7 Bearing Temperature Measurement . . . . . . . . . . . . . . . 39
4.2 Numerical Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.1 Hydrodynamic Modeling . . . . . . . . . . . . . . . . . . . . . 44
4.2.2 Modeling of Structural Deformation . . . . . . . . . . . . . . . 45
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5 Bearing Torque and Power Loss Results 48

5.1 Experimental Torque/Power Loss Measurements . . . . . . . . . . . . 48
5.1.1 Bearing Torque versus Speed and Load . . . . . . . . . . . . . 48
5.1.2 Effect of Runner Surface Roughness . . . . . . . . . . . . . . . 52
5.1.3 Effect of Cooling Flow . . . . . . . . . . . . . . . . . . . . . . 53
5.2 Numerical Predictions of Bearing Torque . . . . . . . . . . . . . . . . 56
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6 Load Capacity Results 61

6.1 Experimental Measurement of Load Capacity . . . . . . . . . . . . . 62
6.1.1 Load Capacity as a Function of Speed . . . . . . . . . . . . . 62
6.1.2 Effect of Cooling Air Flow . . . . . . . . . . . . . . . . . . . . 64
6.1.3 Effect of Runner Surface Finish . . . . . . . . . . . . . . . . . 66
6.1.4 Impact of Top Foil Coating . . . . . . . . . . . . . . . . . . . 67
6.2 Numerical Prediction of Load Capacity . . . . . . . . . . . . . . . . . 70
6.3 General Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
7 Compressibility Number and Film Thickness 77

7.1 Compressibility Number . . . . . . . . . . . . . . . . . . . . . . . . . 78
7.2 Film Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.3 Relative Effects of Thermal Management and Runner Surface Finish . 89
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
8 Physics of Thermal Management 92

8.1 Impacts of Runner Heat Transfer . . . . . . . . . . . . . . . . . . . . 93
8.2 Measurement of Bearing Temperature Gradients . . . . . . . . . . . . 97
8.3 Impact of Runner Convective Heat Transfer on Bearing Temperature
Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
8.4 Runner Material Effects . . . . . . . . . . . . . . . . . . . . . . . . . 104
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
9 Summary and Conclusions 108

9.1 Summary of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
9.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
9.3 Implications and Future Work . . . . . . . . . . . . . . . . . . . . . . 111
iv
A Discussion of Error 114
B Derivation of Characteristic Gas Film Parameters 117
C Runner Design Limitations 121
D Raw Experimental Data 126
Bibliography 132
v
List of Tables
4.1 Test Bearing Parameters . . . . . . . . . . . . . . . . . . . . . . . 34

4.2 Material Properties of Test Runner Materials at 20◦ C . . . . 37
8.1 Rotating Disk Convection Correlations . . . . . . . . . . . . . . 101
vi
List of Figures
1.1 Photographs of Modern Foil Bearings . . . . . . . . . . . . . . . 2

1.2 Disassembled Air Cycle Machine from a B-2 Aircraft . . . . . 3
1.3 Diagram of a Notional Closed Brayton Cycle Rotor . . . . . . 4
1.4 Cross Sectional View of PS304 Solid Lubricant Coating . . . 6
2.1 Examples of Rigid Thrust Bearing Geometries . . . . . . . . . 12

2.2 Cross Section of a Generation III Foil Journal Bearing . . . . 13
2.3 First Appearance of Bump Foil Type Thrust Bearing in Patent
Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Modern High Load Capacity Foil Thrust Bearing . . . . . . . 15
3.1 Exaggerated Cross Sectional View of Gas Film Control Volume 22

3.2 Deformed Top Foil Shape Under Hydrodynamic Pressure . . 25
4.1 Cutaway View of the Rotating Section of the High Speed Foil
Thrust Bearing Test Rig . . . . . . . . . . . . . . . . . . . . . . . 27
4.2 Thrust Bearing Test Rig Loader Section . . . . . . . . . . . . . 28
4.3 Diagram of Thrust Bearing Loading Mechanism . . . . . . . . 29
4.4 Thrust Bearing Test Rig Torque Measurement System . . . . 29
4.5 Photograph of Pneumatic Loading Arrangement . . . . . . . . 30
4.6 Test Thrust Runner . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.7 Photographs of Test Thrust Runner Surfaces . . . . . . . . . . 33
4.8 Photographs of Test Thrust Bearings . . . . . . . . . . . . . . . 35
4.9 Sketch of Runner Assembly with Aluminum Annulus . . . . . 36
4.10 Time Trace of Typical Load Capacity Test . . . . . . . . . . . . 40
4.11 Photograph of Optical Profilometer . . . . . . . . . . . . . . . . 40
4.12 Thrust Bearing Thermocouple Instrumentation . . . . . . . . 43
4.13 Foil Wear Scar Due to Tack Weld Distortions . . . . . . . . . . 43
5.1 Torque Versus Load From 25-55 krpm . . . . . . . . . . . . . . 49

5.2 Power Loss Versus Load From 25-55 krpm . . . . . . . . . . . . 50
vii
5.3 Bearing Torque vs Load with Chromium-Coated Runner and
Cooling Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4 Bearing Torque vs Load Showing Effect of Runner Surface
Roughness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.5 Effect of Cooling Flow on Bearing Torque at 55 krpm Against
PS304 Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.6 Power Loss at Load Capacity for Various Cooling Flow Rates 55
5.7 Comparison of Numerical Predictions of Bearing Torque to
Experimental Data at 25krpm . . . . . . . . . . . . . . . . . . . 58
5.8 Comparison of Numerical Predictions of Bearing Torque to
Experimental Data at 55krpm . . . . . . . . . . . . . . . . . . . 59
6.1 Thrust Bearing Load Capacity as a Function of Speed . . . . 63

6.2 Effect of Cooling Flow on Load Capacity . . . . . . . . . . . . . 65
6.3 Effect of Runner Surface Finish on Load Capacity . . . . . . . 68
6.4 Failure of Bearing with Uncoated Top Foils . . . . . . . . . . . 70
6.5 Thrust Bearing Surface Profile After High Load Operation . 71
6.6 Comparison of Numerical Predictions of Bearing Load Ca-
pacity to Experimental Data . . . . . . . . . . . . . . . . . . . . 73
7.1 Load versus Compressibility Number at Various Speeds . . . 79

7.2 Load Capacity Versus Compressibility Number . . . . . . . . . 80
7.3 High Speed Limit Behavior . . . . . . . . . . . . . . . . . . . . . 82
7.4 Load Versus Compressibility Number for Various Cooling Flow
Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.5 Film Thickness Versus Load at Various Speeds . . . . . . . . . 85
7.6 Film Thickness near Load Capacity . . . . . . . . . . . . . . . . 87
7.7 Film Thickness Versus Load at 25 krpm, 0.52 kg/min . . . . . 88
7.8 Load Versus Compressibility Number Showing the Effects of
Cooling and Runner Surface Roughness . . . . . . . . . . . . . 90
8.1 FEA Plot of Runner Axial Deflection due to Through-Thickness

Heat Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
8.2 Photograph of Wear Scar on Bearing Inner Radius . . . . . . 96
8.3 Thermocouple Locations Near Top Foil Trailing Edge . . . . . 98
8.4 Trailing Edge Temperature Gradients From 25-65 krpm . . . 99
8.5 Estimated Convection Coefficient on Runner Backside at 35
krpm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
8.6 Trailing Edge Temperature Gradients versus Scaled Power Loss103
8.7 Runner Thermal Conductivity Effects on Temperature Dis-
tributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
viii
C.1 Diagrams Showing Various Runner Design Features . . . . . . 124
C.2 FE Model of Runner Face Axial Displacement at 60 krpm . . 125
ix
ACKNOWLEDGEMENTS
The author extends his gratitude to Dr. Christopher DellaCorte of the NASA
Glenn Research Center for his support and encouragement throughout this course of
study. In addition, the author wishes to express his gratitude to Dr. Joseph Prahl
of Case Western Reserve University for his counsel over the years, and for serving as
the author’s academic advisor throughout his undergraduate and graduate studies.
The members of the author’s committee, Dr. Edward White, Dr. Iwan Alexander,
and Dr. Robert Mullen are also thanked.
The author wishes to thank Dr. Robert Bruckner of NASA for his extensive help
with numerical modeling and hydrodynamic analysis, as well as Kevin Radil of the
Army Research Laboratory at GRC for his assistance with experimental methods.
The author would also like to thank the rest of the members of the Oil-Free Turbo-
machinery team of NASA Glenn Research Center for their advice and assistance.
In particular, Brian Edmonds is thanked for his tireless effort in overseeing fabrica-
tion and modification of experimental hardware. The author would also like to extend
his gratitude to Donald Striebing of the Army Research Lab and Victor Lukaszewicz
for their help around the lab. Furthermore, the author wishes to thank John Lucero
for performing the profilometric analysis.
Additionally, the author would like to acknowledge Dr. Hooshang Heshmat of
Mohowk Innovative Technology, Inc. and Dr. Giri Agrawal of R&D Dynamics
Corporation for supplying the test bearings used in this work.
Finally, the author wishes to offer his very sincere appreciation to the National
Aeronautics and Space Administration and the Case School of Engineering for gen-
erous financial support of this research.
x
NOMENCLATURE
Symbol Description
h Heat transfer coefficient, W
m2 K
h Gas film thickness, m

h∗ Nondimensional gas film thickness, h
hi
hc Characteristic (Couette) film thickness, m

hi Inlet gas film thickness, hi
kair Thermal conductivity of air, W
mK
Nu Local Nusselt number, hr

kair
P Local pressure, P a
P̄ Average load pressure, P a
∗
P Non-dimensional pressure, P
Pa
Pa Inlet pressure, P a
r Local radius, m
R Outer radius of runner, m
r∗ Nondimensional radial coordinate, Rr
ri Inner radius of runner, m
ωr 2
Rer Rotational Reynolds number based on local radius, ν
ωR2
ReR Rotational Reynolds number based on runner outer radius, ν
Retrans Rotational Reynolds number at transition to turbulence

To Experimentally measured bearing torque, N m
u Circumferential velocity, m
s
y Film thickness coordinate, m

6μωr 2
Λ Compressibility number, Pa h2
6μωR2
Λc Characteristic compressibility number, Pa h2c
xi
μ Dynamic viscosity of air, kg
ms
μ∗ Nondimensional viscosity, μμa

μa Gas viscosity at inlet, kg
ms
m2
ν Kinematic viscosity of air, s
ξ Bearing radius ratio, rRi

ρ Air density, kg
m3
ρ∗ Nondimensional gas density, ρ

ρa
ρa Inlet gas density, kg

m3
τw Wall shear stress, P a

ω Runner rotational speed, rad
s2
xii
Factors Influencing the Performance of Foil Gas Thrust Bearings for Oil-Free
Turbomachinery Applications
Abstract
by
BRIAN DAVID DYKAS
The operating characteristics of foil gas thrust bearings are explored experi-
mentally and analytically to ascertain the physical mechanisms that limit bearing
performance. Measurements of bearing power loss and load capacity made in a
variety of configurations highlight several important factors which influence perfor-
mance. Consistent with conventional hydrodynamic theory, surface condition of the
foil and surface condition of the runner have a large influence on bearing performance.
Furthermore, active thermal management via cooling air flow and passive thermal
management via conduction through the runner have a large influence.
Thermal effects are shown to be more pronounced at higher loads where gas
film heat generation and resulting thermoelastic distortion are larger, but smooth
lubricious surfaces are needed to achieve these loads. With non-optimal surface
conditions such as high levels of roughness, it is shown that asperity contact dominates
over thermal deformation. This dissertation quantifies the effects of these non-ideal
surface conditions on the load capacity of foil thrust bearings.
xiii
It is determined that both smooth, low friction surfaces combined with adequate
thermal management are necessary to support large loads at high speeds. Further-
more, analysis and modeling suggest that enhanced thermal management is possible
by optimizing the thermal characteristics of the runner, an approach not yet exploited
by the foil bearing community.
xiv
Chapter 1
Introduction
Foil gas thrust bearings are self-acting, compliant-surface hydrodynamic bearings
which operate on a thin lubricating gas film. These bearings are in widespread use
in oil-free microturbines and aircraft air cycle machines. While the gas film prevents
rubbing contact during normal operation, solid lubricants are applied to both foils and
runners to prevent wear and galling during startup, shutdown, and overload conditions
when sliding contact occurs. Despite their use for decades, theoretical understanding
of foil thrust bearings remains incomplete. Models have been developed to predict
hydrodynamic performance, but remain largely unvalidated. A major obstacle to
improved modeling is a poor understanding of the phenomena that limit thrust foil
bearing operation, particularly at high loads and speeds.
1.1 Oil-Free Turbomachinery Applications
The National Aeronautics and Space Administration currently supports a pro-
gram to research advanced Oil-Free Turbomachinery technology in order to realize
1
improvements in the high speed rotating machinery for aircraft and space vehicles.
This program seeks to replace conventional oil-lubricated rolling element bearings
with hydrodynamic foil bearings (Figure 1.1) lubricated by a gaseous working fluid.
Foil bearings have seen service in some high speed turbomachines for several
decades. The first widespread commerical application of oil-free foil gas bearings
was in air cycle machines (ACMs) for aircraft cabin pressurization and circulation
in the 1970s1 (see Figure 1.2). Relatively low temperature and narrow operating
range of the ACMs allow simple foil bearings with limited load capacity to operate
with soft, low temperature polymeric solid lubricants. The benefit of using oil-free
foil gas bearings relative to oil-lubricated bearings was maintenance reduction and
elimination of contamination of cabin air from oil emissions. Since their introduction,
oil-free ACMs have accumulated millions of hours of reliable service, and achieved a
mean time between failures (MTBF) in excess of 100,000 hours2 .
Since the successful application of oil-free bearing to ACMs, foil bearing technol-
(a) Journal Bearing (b) Thrust Bearing
Figure 1.1: Photographs of Modern Foil Bearings
2
ogy advances have resulted in commercially available turbocompressors3 , small gas
turbine electrical generators (<100 kW)4 , and the successful demonstration of an oil-
free diesel truck turbocharger5 and small turbojet engine for a US Navy target drone6 .
Auxiliary power units (APUs), gas turbine generators (>200 KW), small turbofan
aeroengines, and closed-loop Brayton cycle generators for space nuclear power are all
near-term applications for foil bearings enabled by the advancing state of the art.
In the case of typical turbofan aviation engines, studies show that with oil-free
technology, total engine weight can be reduced by approximately 15% and engine
maintenance lowered by 50%. Bearing temperature and shaft speed limits are in-
creased, allowing for a 20% power density increase and an 8% reduction in direct
operating costs7 . A recent propulsion system study on a notional 50 passenger regional
jet concludes that an improvement of 2.9% in take-off gross weight (TOGW) and 3.4%
in block fuel can be achieved through oil-free engines8 .
Figure 1.2: Disassembled Air Cycle Machine from a B-2 Aircraft (photograph
reproduced from Agrawal2 )
3
In the case of non-terrestrial power generation, the high energy density of nuclear
power sources is attractive to spaceflight mission designers. Thermoelectric power
generation with a nuclear heat source has been used in spacecraft for decades, but does
not scale well to the power and efficiency requirements of future high-power missions9 .
A closed Brayton cycle (CBC) dynamic power conversion concept requires oil-free
rotor supports for long endurance. Foil bearings eliminate the problems associated
with liquid lubricants including radiation incompatibility and contamination of the
working fluid, and will serve as the rotor supports for Brayton cycle space power
conversion. Figure 1.3 shows a diagram of a candidate CBC rotor10 .
Figure 1.3: Diagram of a Notional Closed Brayton Cycle Rotor
4
1.2 Enabling Technologies
The achievable reductions in weight, complexity, and maintenance that accompany
the use of the process fluid as a lubricant have always been desirable for turboma-
chinery. While foil bearing development was still in its infancy, several unsuccessful
attempts were made at gas bearing supported turbomachines. These attempts failed
largely due to a few key deficiencies in gas bearing technology at the time.
The low load capacity of early thrust bearings limited their use to applications
with narrow operating ranges where thrust forces were low, and transient operation
at off-design conditions might require hydrostatic load sharing. A lack of appropriate
high temperature solid lubricants to reduce friction and wear during sliding contact
at startup and shutdown further hindered foil bearing adoption. Finally, modeling
of gas bearing behavior was mostly limited to simple cases of rigid geometry and
constant lubricant properties, requiring a rather expensive hardware-intensive testing
approach. These deficiencies in technology at the time contributed to the abandon-
ment of a United States Air Force project to develop an oil-free APU11 . Recently
however, these limitations have been partially addressed by industry, academia, and
government researchers.
Within the past decade, advances in three key technology areas have resulted
in renewed interest in oil-free rotor support systems. Foil air bearings have been
developed with much higher load capacity than the previous state-of-the-art, and
improved rotordynamic characteristics12. Analytical modeling has advanced with
increased computing power such that advanced numerical modeling can be con-
5
ducted relatively cheaply and quickly. Finally, the NASA-developed PS300 series
of plasma sprayed composite solid lubricant shaft coatings13,14 represents the most
recent generation of high temperature, wear resistant solid lubricant coatings for this
application (see Figure 1.4). The coatings have endured 100,000 start-stop cycles at
temperatures from 93◦ C up to 650◦ C, representing a lifetime of twice that required
for most turbomachinery applications15 . Additionally, foil coatings are available to
provide additional lubrication during the initial break-in period of turbomachines16,17 .
Figure 1.4: Cross Sectional View of PS304 Solid Lubricant Coating
1.3 Present Work
While the advantages of foil bearings in high speed turbomachines have been
determined and feasibility demonstrated, much work remains to characterize bearing
behavior as it relates to system integration. There is currently a wealth of data
6
on foil journal bearing load capacity and rotordynamic characteristics, with work
beginning on journal bearing thermal behavior and management. There is, however,
comparatively little published data regarding even basic performance characteristics
of foil thrust bearings.
Furthermore, the limited experimental data available for foil thrust bearings is
provided by bearing manufacturers and is geared toward maximum attainable perfor-
mance. Designers of next-generation oil-free turbomachinery require a comprehensive
characterization of the bearings over a wide range of conditions in order to achieve
workable and efficient designs.
Current independent research on foil thrust bearings focuses on quantifying basic
mechanisms that govern and limit their performance. Elucidating these phenomena
is critical to advance understanding and guide modeling efforts. The investigations
presented herein also allow more objective evaluation of bearing design by quanti-
fying the influence of various factors that are mostly independent of the foil design.
Providing this context will help turbomachinery designers understand the practices
of the bearing community and design around their strengths and weaknesses.
With nominal gas film thicknesses on the order of tens of microns, gas bearings
are very sensitive to flatness and roughness of the foil and runner surfaces. While
compliant bearings provide tolerance to non-ideal runner surfaces, quantitative mea-
surements of misalignment and runner distortion effects on bearing performance have
not been reported, and these measurements are very difficult to make. Additionally,
while surface roughness and lubricity effects have been studied in foil journal bear-
ings17 , the same has not been reported for their thrust bearing counterparts. This
7
dissertation quantifies the effect of surface condition (both foil and runner) on the
load capacity of foil thrust bearings.
A further focus of study is the characterization of the bearing thermal envi-
ronment. While bearing torque and power loss are typically less in foil bearings
than rolling element bearings, the power loss is far more load-dependent in foil
bearings. As a result, the thermal environment is more challenging to characterize
over the operating envelope. Whereas heat generated in oil-lubricated bearings can
be managed by circulating and cooling the lubricant, air bearings operate with a
lubricant that has a much lower heat capacity and is not recirculated and cooled.
Instead, heat is removed through conduction in the support structure and through
any net flow of the working fluid through the bearing cavity. With higher resistances
to heat transfer, gas bearings are prone to significant temperature gradients within
the structural components which can adversely affect performance.
This dissertation investigates and compares thermal management phenomena to
the magnitude of surface condition effects. Active flow of cooling air through the
bearing foils is compared to passive thermal management achieved through conductive
heat transfer design. Thermal effects are shown to be more pronounced at higher loads
where gas film heat generation and resulting thermoelastic distortion are larger, but
smooth lubricious surfaces are needed to achieve these loads. With non-optimal sur-
face conditions including large roughness, it is shown that asperity contact dominates
over thermal deformation. In that case, thermal management techniques are shown
to have a lesser effect on bearing performance.
In this dissertation, the limiting factors for thrust foil bearings, namely surface
8
condition and thermal management, are investigated. Experimental results are com-
pared with existing theory and analysis. It is anticipated that this characterization
effort will provide valuable guidance for the successful implementation of foil gas
bearings to future more demanding oil-free turbomachinery systems.
References
[1] Agrawal, G., 1990. “Foil Gas Bearings for Turbomachinery”. In 20th Intersociety
Conference on Environmental Systems, no. SAE Paper 901236, Society of
Automotive Engineers.
[2] Agrawal, G. L., 1997. “Foil Air/Gas Bearing Technology An Overview”. In

International Gas Turbine and Aerospace Congress and Exhibition, no. ASME
paper 97-GT-347, American Society of Mechanical Engineers.
[3] Mohawk Innovative Technology, Inc., 2005. High-Speed, Oil-Free, Motorized

Spindle. Press Release, January.
[4] Capstone MicroTurbine Corporation, 2000. Capstone Turbine Ships New

Microturbine Product; New Capstone 60 Power System Generates Twice the
Electricity of Current Models. Press Release.
[5] Heshmat, C., Heshmat, H., Valco, M. J., Radil, K., and DellaCorte, C.,
2005. “Foil Bearings Makes Oil-Free Turbocharger Possible”. In Proceedings
of WTC2005, World Tribology Congress III, no. WTC2005-63724, American
Society of Mechanical Engineers/Society of Tribologists and Lubrication
Engineers.
[6] Mohawk Innovative Technology, Inc., 2002. WJ24-8 Turbojet Engine

Demonstration Testing with Air Foil Bearing. Press Release, December.
[7] DellaCorte, C., and Pinkus, O., 2000. Tribological Limitations in Gas Turbine
Engines: A Workshop to Identify the Challenges and Set Future Directions.
NASA TM 2000-210059, National Aeronautics and Space Administration,
Cleveland, OH.
[8] Bruckner, R. J., 2004. A Propulsion System Analysis of Oil Free Turbomachinery
for Aviation Turbofan Engines. AIAA 2004-4189, American Institute of
Aeronautics and Astronautics, Reston, VA.
9
[9] Mason, L. S., 1999. Surface Nuclear Power for Human Mars Missions. NASA
TM 1999-208894, National Aeronautics and Space Administration, Cleveland,
OH.
[10] Howard, S. A., and DellaCorte, C., 2006. Gas Foil Bearings for Space
Propulsion Nuclear Electric Power Generation. NASA-TM 2006-214115,
National Aeronautics and Space Administration, Cleveland, OH.
[11] Suriano, F. J., 1981. Gas Foil Development Program; Final Report. AFWAL-TR
81-2095, Air Force Wright Aeronautical Laboratories, Dayton, Ohio.
[12] Heshmat, H., 1994. “Advancements in the Performance of Aerodynamic Foil

Journal Bearings: High Speed and Load Capacity”. ASME Journal of Tribology,
116, pp. 287–295.
[13] DellaCorte, C., and Edmonds, B. J., 1995. Preliminary Evaluation of PS300:
A New Self-Lubricating High Temperature Composite Coating for Use to 800C.
NASA TM 107056, National Aeronautics and Space Administration, Cleveland,
OH.
[14] DellaCorte, C., and Edmonds, B. J., 1999. Self-Lubricating Composite

Containing Chromium Oxide. U.S. Patent 5,866,518.
[15] DellaCorte, C., Lukaszewicz, V., Valco, M. J., Radil, K., and Heshmat, H., 2000.
“Performance and Durability of High Temperature Foil Air Bearings for Oil-Free
Turbomachinery”. Tribology Transactions, 43(4), pp. 774–780.
[16] Radil, K. C., and DellaCorte, C., 2001. The Effect of Journal Roughness and
Foil Coatings on the Performance of Heavily Loaded Foil Air Bearings. NASA
OH.
[17] DellaCorte, C., Zaldana, A., and Radil, K. C., 2002. A Systems Approach to the
Solid Lubrication of Foil Air Bearings for Oil-Free Turbomachinery. NASA TM
2002-211482, National Aeronautics and Space Administration, Cleveland, OH.
10
Chapter 2
Background of the Art
2.1 Foil Air Bearing Development
The foil gas bearings that enable oil-free turbomachinery systems are compliant,
self-acting, hydrodynamic journal and thrust bearings which use ambient air or any
other gas as a lubricant. During loaded operation, a very thin film of air on the order of
10 μm1 is developed between the bearing foil surface and shaft or thrust runner, which
results in no wear during operation2 . Compliant foil structures consisting of a smooth
top foil supported by an elastic foundation are tolerant to misalignment and distortion
as well as allowing for design-tailored stiffness and damping characteristics3 . They
offer improvements over rolling element bearings including higher shaft speed limited
only by the burst strength of the rotating hardware, the ability to run from cryogenic
temperatures to 650 ◦ C (approximately the temperature of high pressure compressor
discharge in SOA gas turbine aeroengines), and low frictional losses. These charac-
teristics enable development of advanced turbomachinery with lower maintenance,
higher efficiency and power density, and reduced weight and complexity4 .
11
Rigid geometry hydrodynamic bearings using gas as a lubricant have been around
for decades, utilizing general designs that mimic oil-lubricated bearings, but with
lower load capacity (Figure 2.1). In 1953, Blok and Van Rossum5 first coined the
term foil bearing to describe a thin cellophane foil, devoid of rigidity, wrapped around
a rotating journal, lubricated with oil. They showed that the compliance of the foil
surface allowed it to deform in the presence of hydrodynamic pressure, creating a
more favorable geometry and increasing the minimum film thickness at a given load,
reducing torque and enhancing performance.
Figure 2.1: Examples of Rigid Thrust Bearing Geometries6
In the years and decades following the introduction of the foil bearing, new designs
for bearings with compliant structures appeared, most focusing on journal bearings
which support the shaft laterally and govern the most critical rotordynamic behavior.
Early patents for gas thrust bearings with compliant geometry claim that while
foil journal bearings were advancing rapidly due to a geometry inclined to form a
lubricating gas film between the shaft and bearing, thrust bearings were progressing
12
more slowly due to a difficulty in creating an efficient convergent wedge and load
carrying surface7,8 .
Bump foil type journal bearings, whose configuration is shown in Figure 2.2,
emerged over the years to become the most advanced design available to date9,10 .
In this style bearing there are three main components - the top foil, bump foil,
and bearing housing. The top foil wraps around the journal, and can be reinforced
with stiffener foil(s), outside of which are one or more compliant bump foils, that
are contained within a solid bearing housing. The current state of the art designs
are described by DellaCorte and Valco11 as Generation III foil bearings, which have
stiffness properties that vary along the axial, radial, and circumferential directions in
the bearing. Stiffness variation allows for bearings to be made with different stiffness,
damping, and load capacity characteristics.
Figure 2.2: Cross Section of a Generation III Foil Journal Bearing
13
In response to the problems reported in foil thrust bearing development, bearings
with rigid pads supported on compliant structures were introduced, but were ineffec-
tive at light loads and high speeds7,12 . Further research on compliant thrust bearings
soon resulted in designs employing the bump foil structures that had proven successful
in journal bearings. In this first appearance of this type of foil thrust bearing, the foil
structure is comprised of a wave spring (bump foil) resting on a nonrotating backing
plate. The wave spring supports a thick foil plate, on top of which is a thinner
top foil (Figure 2.3). The thinner top foil provides an efficient wedge at low speeds
and loads, while the underlying thicker foil supports the top foil and limits foil sag
between bumps at high speed and load, where hydrodynamic pressures are higher13 .
This load-tailored stiffness is notable in that bump foil style journal bearings did not
generally include this type of advanced design feature until second generation designs
emerged after years of development11 .
Current advanced foil thrust bearings generally consist of an annular backing
plate, to which a number of sector-shaped pads are tack welded. The pads consist
of one or more underlying bump foils and a smooth top foil, between which one or
more stiffener foils can be placed9 . The foils which compose the pads are arranged
and shaped so as to form an initial converging wedge with the runner, followed by a
comparably flat load carrying section. Nickel-based superalloys such as Inconel X-750
are the current material of choice for these foils due to, for example, their weldability,
fatigue resistance, strength and modulus properties at high temperature, and creep
resistance. A patent drawing of an advanced foil thrust bearing is shown in Figure
2.4. Other foil thrust bearing arrangements exist, but are not considered in this study.
14
Figure 2.3: First Appearance of Bump Foil Type Thrust Bearing in Patent
Literature, adapted from Fortman13
Figure 2.4: Modern High Load Capacity Foil Thrust Bearing, adapted from
Heshmat9
15
2.2 Current State of the Art Practices
Until recently, neither advanced thrust bearing hydrodynamic modeling nor high
speed high load test facilities existed to foster the rapid improvement of thrust bearing
design. With the construction of advanced test rigs14,15 and improved hydrodynamic
modeling code16 , accurate determination of thrust foil bearing performance and be-
havior is now possible across a wide range of operating conditions.
Over the course of decades of development, thrust bearing designers have come
to rely on several methods to increase load capacity. Universally recognized as being
critical to successful foil bearing systems are the tribological coatings on both foil and
runner. These coatings form a system which must yield a low coefficient of friction,
yet withstand high temperatures and exhibit low wear. These systems are often
proprietary and not fully described in the open literature, which makes it difficult
to evaluate bearing foil designs relative to one another when different tribological
systems are employed.
Thrust runner coatings are typically hard, durable coatings with low surface
roughness. Thin, dense chromium coatings 5-10 μm thick are common, yielding a
surface with average roughness Ra =0.05-0.1 μm, but can only withstand on the order
of 5,000-10,000 start-stop cycles and temperatures of 25 - 500◦ C. By contrast, NASA-
developed PS300 series coatings can withstand over 100,000 start-stop cycles at tem-
peratures of 650◦ C or more. However, the plasma spray deposition of PS304 coatings
results in porosity in the coating, with typical surface roughness of Ra=0.2-0.8 μm
after grinding and polishing. Until the coating achieves a lower surface roughness
16
through a break-in wear process, bearing performance can suffer significantly17 .
Foil coatings also play a central role in the load capacity behavior of “Generation I”
foil thrust bearings, which are charaterized by simple foil structures with little spatial
tailoring of stiffness properties (see DellaCorte and Valco11 for foil journal bearing
classification). For the same foil design, tests have shown a remarkable increase in
load capacity with the addition of a soft solid lubricant coating on the top foil. When
a break-in procedure including low load start-stop cycles and near-load capacity tests
at low speed is performed, the foil coating is seen to wear significantly in areas where
the bump foil contacts the top foil. With further tests relatively little additional
wear occurs, as determined by optical profilometry18 . This indicates in addition to
its solid lubricant properties, the abradable nature of the top foil allows the surface
to compensate for foil sag16 at higher hydrodynamic pressures by allowing a more
favorable foil geometry to develop through the wear process. As such, foil coatings
can have a dramatic effect on load capacity.
Another parameter that is seldom fully reported is the mass flow rate of cooling
air forced through the bearing. Most high-load testing of thrust bearings is achieved
using some amount of airflow through the bump foil structure to remove heat from
the bearing and to reduce temperature gradients and corresponding thermal dis-
tortions within the structural elements. While effective at increasing load capacity
in a laboratory setting, this practice is undesirable in the integrated system due
to increased complexity and reduced thermodynamic efficiency. As a result of this
practice, performance data on foil thrust bearings at high speed and high load often
require contextual information about the cooling flow rates involved in the test.
17
It is desirable to lessen the need for foil coatings and cooling air as bearings
develop. To some degree, the requirement for these techniques may be highly influ-
enced by today’s rudimentary thrust foil bearing designs. It is anticipated that the
fundamental studies carried out in the present work will improve understanding and
enable future designs that lessen or eliminate the needs for cooling flow and sacrificial
tribological coatings.
References
[1] Ruscitto, D., McCormick, J., and Gray, S., 1978. Hydrodynamic Air Lubricated
Compliant Surface Bearing For an Automotive Gas Turbine Engine I - Journal
Bearing Performance. NASA CR 135368, National Aeronautics and Space
Administration, Cleveland, OH.
[2] DellaCorte, C., Lukaszewicz, V., Valco, M. J., Radil, K., and Heshmat, H., 2000.
[3] Heshmat, H., Shapiro, W., and Gray, S., 1981. “Development of Foil Journal
Bearings for High Load Capacity and High Speed Whirl Stability”. In
Proceedings of the ASLE-ASME Joint Tribology Conference, no. 81-Lub-36,
American Society of Lubrication Engineers/American Society of Mechanical
Engineers.
[4] DellaCorte, C., and Pinkus, O., 2000. Tribological Limitations in Gas Turbine
Engines: A Workshop to Identify the Challenges and Set Future Directions.
NASA TM 2000-210059, National Aeronautics and Space Administration,
Cleveland, OH.
[5] Blok, H., and vanRossum, J., 1953. “The Foil Bearing - A New Departure in
Hydrodynamic Lubrication”. Lubrication Engineering, 9(6), pp. 316–320.
[6] Anderson, W. J., and Bisson, E. E., 1964. Advanced Bearing Technology. NASA-
SP 38, National Aeronautics and Space Administration, Washington, D.C.
[7] Nemeth, Z. N., 1977. Experimental Evaluation of Foil-Supported Resilient-Pad

Gas-Lubricated Thrust Bearing. NASA TP 1030, National Aeronautics and
Space Administration, Cleveland, OH.
18
[8] Etsion, I., 1977. “A Cantilever Mounted Resilient Pad Gas Thrust Bearing”.
Journal of Lubrication Technology, 99, pp. 95–100.
[9] Heshmat, H., 1999. High Load Capacity Compliant Foil Hydrodynamic Thrust
Bearing. U.S. Patent 5,961,217.
[10] Heshmat, H., 2005. “Major Breakthrough in Load Capacity, Speed and
Operating Temperature of Foil Thrust Bearings”. In Proceedings of WTC2005,
World Tribology Congress III, no. WTC2005-63712, American Society of
Mechanical Engineers/Society of Tribologists and Lubrication Engineers.
[11] DellaCorte, C., and Valco, M. J., 2000. “Load Capacity Estimation of Foil
Air Journal Bearings for Oil-Free Turbomachinery Applications”. Tribology
Transactions, 43, pp. 795–801.
[12] Nemeth, Z. N., 1979. Operating Characteristics of a Cantilever-Mounted

Resilient-Pad Gas-Lubricated Thrust Bearing. NASA TP 1438, National
Aeronautics and Space Administration, Cleveland, OH.
[13] Fortmann, W. E., 1978. Dual Wedge Fluid Thrust Bearing Including Wave
Spring. U.S. Patent 4,082,375.
[14] Hryniewicz, P., Locke, D., and Heshmat, H., 2003. “New-Generation
Development Rigs for Testing High-Speed, Air-Lubricated Thrust Bearings”.
Tribology Transactions, 46(4), pp. 556–559.
[15] Bauman, S., 2005. An Oil-Free Thrust Foil Bearing Facility Design, Calibration,
and Operation. NASA TM 2005-213568, National Aeronautics and Space
[16] Bruckner, R. J., 2004. “Simulation and Modeling of the Hydrodynamic, Thermal,
and Structural Behavior of Foil Thrust Bearings”. PhD dissertation, Case
Western Reserve University, Cleveland, OH.
OH.
[18] Lucero, J., 2006. Measured Foil Frequencies and Mode Shapes as a Function
of Wear for a Gas Foil Thrust Bearing. NASA-TM (to be published), National
19
Chapter 3
Gas Film Characteristics
3.1 Hydrodynamic Features
Gaseous hydrodynamic lubrication is characterized by very thin films which stem
from the low viscosity of gasses relative to typical liquid lubricants. Gas lubrication
is also characterized by compressible flow, and viscosity that increases with, but is
less dependent on, temperature than liquid lubricants.
In conventional gas lubrication analysis, the continuity equation and Navier-Stokes
equations are scaled and reduced to a compressible form of the Reynolds equation.
The energy equation is often simplified by assuming an isothermal flow. Viscosity
and heat capacity are often assumed constant. Traditional assumptions are often
appropriate in bounding the behavior of the gas film, but cannot adequately predict
behavior near the load capacity of the bearing where the gas film is in distress and
near rupture.
The most advanced modeling of foil thrust bearings to date, including a rigorous
derivation of the appropriate forms of the governing fluid equations can be found
20
in a dissertation by Bruckner1 . That analysis includes the effects of compressibil-
ity, density models, viscous heat generation, compliant structure, and temperature
dependent fluid properties on the solution of the equations of the gas film.
Surfaces are assumed perfectly flat in numerical models, where real surfaces are
characterized by some level of roughness and lubricity. As the film nears load capacity,
the thinnest portion of the film may allow localized asperity contact as the bearing
transitions to mixed (gaseous and solid) lubrication. This onset of contact generates
large amounts of frictional heat due to the large relative surface velocity. This
localized severe heating causes thermal expansion distortion of the foil and/or runner
surfaces, leading to less favorable geometry (local thinning of film), which further
increases localized heating. This type of thermal runaway is suspected to be the
general mechanism of foil bearing failure.
3.2 Thermal Behavior
During steady-state operation, the heat generated in the gas film must be trans-
ferred out at the same rate that it is generated. This can be illustrated by constructing
a control volume which encloses the gas film between the top foil and runner surfaces,
and extending around the periphery of the top foil. Heat transfer occurring at the
boundaries includes conduction into the runner or bearing top foil as well as heat
advected (transport of heat energy in a vector field) into and out of the control
volume in the gas flow. In general, gas flows into the film at the leading edge and is
heated due to compression and viscous dissipation from boundary work, then flows
21
out at the trailing edge of the pad or at the sides due to leakage. Due to the thinness
of the gas film compared to radial and circumferential pad lengths, radiation heat
transfer from within the control volume and conduction of heat through the gas at
the control volume boundaries can be ignored. Therefore, heat is transported out of
the control volume either by conduction into the thrust bearing foil and runner, or
by advection of the bulk gas flow, as shown in Figure 3.1.
Figure 3.1: Exaggerated Cross Sectional View of Gas Film Control Volume
The fraction of heat transfer that occurs by either conduction or advection has
not been determined, however qualitative statements can be made about the system’s
behavior. The bump foil support structure, made of thin metallic foils, effectively
insulates the top foil from the bearing backing plate. Almost all heat that conducts
into the top foil from the gas film must be removed by the air between the foils and the
backing plate by convection. This mode of heat transfer is generally inefficient unless
22
a substantial flow of air at a lower temperature is forced through the foil structure
to enhance convective heat transfer. Since a majority of heat in foil journal bearings
is transferred out of the gas film via conduction through the metallic foil and runner
surfaces2 , this may also be the case for thrust bearings.
Heat conducted into the thrust runner must be conducted down the shaft, or
removed from the exposed rotating surfaces by convection. The high surface speeds
of the thrust runner result in large Nusselt numbers, so the forced convection on
the runner is rather good considering the range of achievable Nusselt numbers for
convection in air. Since the thermal conductivity of nickel superalloys is poor relative
to many metals, the problem of heat transfer into and through the runner must
consider both the conduction through the runner and the heat transfer coefficients at
the boundaries.
Heat transfer is not currently modeled in foil bearing analysis, but appears to
play a significant role in the bearing performance at high speeds and loads3 . It is
anticipated that future foil bearing analysis codes will begin to address this effect as
experimental efforts map and characterize the thermal behavior of foil bearings. The
present work investigates the effect of cooling on bearing performance, and begins to
examine potential alternate methods of thermal management. Continuing research
is beginning to investigate heat transfer out of the gas film and in the runner, which
will provide a better understanding of the thermal limit in foil thrust bearings.
23
3.3 Structural Response and Effects
The elastic nature of the thrust pad support structure gives the top foil some
compliance, which makes the bearing tolerant to misalignment and distortion of the
runner surface. This compliant surface also allows the bearing to run at higher loads
with thicker film thicknesses than its rigid counterpart4 .
Although compliance of the bearing surface is desirable for these various opera-
tional concerns, it complicates analysis. Since the gas film typically has a thickness on
the order of tens of microns, very small distortions of the top foil or runner surfaces can
have large effects on the gas film pressure distribution. This stiff coupling between
the hydrodynamic and structural governing equations presents a challenge for the
convergence of numerical models. Figure 3.2 shows a calculated top foil shape under
hydrodynamic pressure1 .
Additionally, heat transfer in the bearing and runner can cause thermoelastic dis-
tortions that exceed the nominal film thickness. This warping of the film boundaries
can significantly alter the film thickness and pressure distributions in the bearing
and influence load capacity. Modeling of structural heat transfer and the resulting
deflections is currently limited to qualitative reasoning and rudimentary analysis due
to large uncertainty in the thermal boundary conditions. The present work begins
to characterize the heat transfer paths in thrust foil bearings with thermocouple
instrumentation that will aid in the development and validation of numerical modeling
which includes thermal effects.
24
Figure 3.2: Deformed Top Foil Shape Under Hydrodynamic Pressure, as
calculated by Bruckner1
References
and Structural Behavior of Foil Thrust Bearings”. PhD dissertation, Case Western
Reserve University, Cleveland, OH.
[2] Salehi, M., Swanson, E., and Heshmat, H., 2000. “Thermal Features of Compliant
Foil Bearings - Theory and Experiments”. In Joint Tribology Conference,
no. 2000-TRIB-38, American Society of Mechanical Engineers/Society of
Tribologists and Lubrication Engineers.
[3] Dykas, B., DellaCorte, C., Prahl, J., and Bruckner, R., 2006. “Thermal
Management Phenomena in Foil Gas Thrust Bearings”. In Proceedings of Turbo
Expo 2006: Power for Land, Sea, and Air, no. 2006-91268, American Society of
Mechanical Engineers.
[4] Heshmat, H., Walowit, J. A., and Pinkus, O., 1983. “Analysis of Gas Lubricated
Compliant Thrust Bearings”. ASME Journal of Lubrication Technology, 105,
pp. 638–646.
25
Chapter 4
Methods
4.1 Experimental Methods
Because of the relatively complex interactions of the structural, thermal, and fluid
governing equations, modeling of foil bearings is a nontrivial task. Experimental data
is needed on many fronts to guide and validate modeling efforts in order to understand
and predict foil bearing performance. The present work relies heavily on experimental
data in the development of theories regarding the physical mechanisms that govern
bearing performance.
4.1.1 Thrust Bearing Test Rig
Testing of foil thrust bearings for the current work is done on a high speed thrust
bearing test rig at NASA Glenn Research Center which is described by Hryniewicz
et al.1 and Bauman2 . This newest generation test rig consists of a turbine-driven
rotating shaft supported radially on foil journal bearings, with a magnetic thrust
bearing to maintain axial position (Figure 4.1). A test thrust runner is bolted to the
26
shaft at the end opposite of the turbine. This oil-free shaft support system allows
the rig to operate at speeds up to 80,000 rpm while supporting an axial load of up to
3100 N (700 lbs). In one configuration, the test bearing can also be run at elevated
chamber temperatures to 540◦ C (1000◦ F).
Figure 4.1: Cutaway View of the Rotating Section of the High Speed Foil
Thrust Bearing Test Rig
A separate, non-rotating section of the rig, shown in Figs. 4.2, 4.3, was designed
to load the test thrust bearing against the runner, while measuring bearing torque to
within about 15% accuracy. As a major requirement, it is designed to allow access
to the test bearing and runner for the measurement of temperatures and runner
distortion. At the heart of the loader section, a test thrust bearing is mounted to a
16mm diameter hardened steel shaft supported by a pair of linear ball bearings to
permit rotation and axial translation. With the thrust bearing mounted to this small
diameter, linear bearing supported shaft, the resistance to rotation is small, such that
27
torque can be measured by a 250 gram force LVDT load cell attached to a torque arm
extending radially outward from the shaft (Figure 4.4). A pneumatic bellows-type
actuator loads the shaft through a ball contact on a pivoted arm (Figure 4.5). Due to
care in aligning the ball contact with the center of the shaft, tests of repeatability and
accuracy indicate that this method of measuring torque is sufficient for the required
tests.
Figure 4.2: Thrust Bearing Test Rig Loader Section
The test rig shaft position is monitored by a displacement probe and serves as
a feedback for the magnetic bearing control system. Since the test bearing load is
applied through a mechanical/pneumatic system and not through a similar actively
controlled position system, the complex dynamic interactions of the test rig shaft
and test bearing shaft result in adverse rotor dynamics (large amplitude axial vi-
brations) at some operating conditions which precludes obtaining test data in this
28
Figure 4.3: Diagram of Thrust Bearing Loading Mechanism
Figure 4.4: Thrust Bearing Test Rig Torque Measurement System
29
Figure 4.5: Photograph of Pneumatic Loading Arrangement
range of operation. Torque data is not given for conditions where this increased axial
displacement is present. The minimum speed tested is 25 krpm, below which it is
difficult to maintain a constant shaft speed and some rig dynamic complications are
present.
Shaft speed, bearing torque, and bearing load signals are acquired and recorded
throughout these tests, along with additional bearing instrumentation such as ther-
mocouples in the case of temperature testing. Various test rig operational parameters
are also recorded. Tests are generally performed at a constant speed while varying
axial load. Although speeds to 80 krpm are attainable, test speeds in most cases
are limited to 55-65 krpm to reduce the probability of severe bearing damage when
loaded to near capacity.
When using cooling air flowing through the bearing foils, the cavity pressure in
30
the center of the thrust bearing is measured, as the increased pressure has the effect
of hydrostatically unloading the test bearing. A correlation between cavity pressure
and hydrostatic unloading force is given by Bauman2 , and used in this work to correct
for this effect, where the cavity pressure is dependent on speed, cooling flow rate, and
load. Measured cavity pressures approached 35 kPa (gauge) at the highest flow rates
and loads.
4.1.2 Test Specimens
Thrust Runners
The basic thrust runner design for these tests is described as a central hub,
extending radially out from which is an annular disk. The outer diameter of the
hub section is 4.29 cm, and the outer diameter of the disk is 10.16 cm. A small
(14 mm diameter by 25 mm long) cylindrical protrusion extends from the hub and
fits into a radially centered hole in the test rig shaft to ensure radially concentric
positioning of the runner with respect to the shaft rotational axis. This is critical
to attain a well balanced rotating group as the runners are component balanced
independent of other rig hardware. The runner is bolted to the shaft through the
hub (Figure 4.1), allowing for its removal and replacement with another specimen.
This basic runner configuration is shown in Figure 4.6. The standard test runner for
this research is constructed from Inconel 718, a nickel-based superalloy chosen for its
high temperature strength. The annular disk portion of the runner has a nominal
thickness of 1.27 cm and is coated on the bearing surface with a 250 μm thick PS3043
solid lubricant coating. The coating is plasma sprayed onto the runner surface and
31
Figure 4.6: Test Thrust Runner
ground to the desired thickness.
The plasma spray process is responsible for some porosity in the coating, leading
to an as-ground surface roughness of 0.2 - 0.8 μm Ra 4 . To achieve a better running
surface, a colloidal form of molybdenum disulfide (MoS2 ) in alcohol is sprayed onto
the surface and then burnished to achieve a smooth glossy finish. This procedure is
recommended by DellaCorte et al.5 as an overlay for the break-in of PS304-coated
shafts for foil journal bearings, but its effectiveness is less apparent when used on a
thrust runner.
Since thrust bearings run against unbroken-in PS304 coatings demonstrate a
lower load capacity than when run against smoother chrome coatings, test data is
taken using a standard runner with a thin dense chrome (TDC) coating rather than
the baseline PS304. The chrome coating gives a smoother surface and results in
32
significantly higher load capacity. However, the chrome coating is not as durable
as, and is more prone to heat checking (fine cracks) than, the PS304. Because
of this, the chrome coating may be more typical of laboratory tests than high-
performance turbomachinery conditions, where high temperature capability and long
life are necessary. Photographs of both the PS304 and thin dense chrome coated
runners are given in Figure 4.7.
Surface roughness of the test thrust runners is measured using a contacting stylus-
type surface profiler. Six 4 mm long radial scans were taken at various locations on
each runner, and the Ra values for the scans were averaged. The thin dense chrome
coating roughness was measured to be Ra = 0.077 μm with a standard deviation of
σ = 0.012 μm. The “as-ground” post-test PS304 coating roughness was measured to
be Ra = 0.481 μm with a standard deviation of σ = 0.195 μm.
(a) PS304/MoS2 -Coated Runner (b) Chrome-Coated Runner
Figure 4.7: Photographs of Test Thrust Runner Surfaces after being run in
tests of bearing load capacity.
33
Parameter Bearing A Bearing B
6
U.S. Patent 4,277,111 4,462,7007
Inner Diameter 4.65 cm 4.62 cm
Outer Diameter 9.02 cm 8.97 cm
Pad Angular Extent 42◦ 41◦
Single Pad Area 5.47 cm2 5.28 cm2
Number of Pads 8 8
Total Pad Area 43.8 cm2 42.2 cm2
Top Foil Coating KorolonTM,8 800 PTFE
Table 4.1: Test Bearing Parameters
Thrust Bearings
Due to high specimen costs and limited availability, it is only possible to test a
relatively small sample of different thrust bearings, which can vary in size, foil design,
and foil coating thickness/composition. In this study, bearings are selected from two
different manufacturers in the same size class. The top foil dimensions (inner radius,
outer radius, and angular extent) are almost identical, but the design of the underlying
bump foil structure and the top foil coatings differ between the bearings.
Table 4.1 lists relevant bearing parameters for the two test bearing types on which
a bulk of the tests are performed. Figure 4.8 shows photographs of the two basic test
bearing configurations. It should be noted that in this figure, bearing B has been run
to high load while bearing A is shown in its unrun condition. The different conditions
are unintentional, a consequence of limited available photographs.
34
4.1.3 Cooling Techniques
Thermal management, suspected to have a large influence on bearing behavior,
is investigated in a rudimentary manner by selecting two methods of “cooling” for
evaluation. A PS304-coated, standard Inconel 718 runner without any forced cooling
air flow is taken as the baseline “uncooled” case.
To simulate an active cooling method consistent with industry practices, air is
forced through the bump foil structure from the inner diameter of the bearing and
exhausting from the outer diameter. Cooling flow rate is varied from 0 to 0.52 kg/min,
with the hydrostatic unloading effect of this method corrected for.
An alternative method of thermal management is a passive method relying on
increased thermal conductivity of the runner to remove heat from the gas film. In this
case, the annular disk portion of the runner is constructed from an alloyed aluminum
(7075-T6) pressed onto an Inconel 718 hub. The aluminum has an order of magnitude
(a) Bearing A - never run, Au-coated (b) Bearing B - post-test, Au-coated
Figure 4.8: Photographs of Test Thrust Bearings
35
larger thermal conductivity, and a slightly lower thickness of 0.95 cm. A sketch of
the runner assembly is given in Figure 4.9. A comparison of the material properties
of Inconel 718 and 7075-T6 aluminum9 is given in Table 4.2. This runner has a
PS304/MoS2 coating with similar surface roughness to the baseline PS304-coated
runner.
Figure 4.9: Sketch of Runner Assembly with Aluminum Annulus
4.1.4 Operating Torque and Power Loss Measurements
In order to characterize the thermal behavior of the thrust bearings and runners,
the bearing power loss (i.e. heat generation) must be determined for a wide range of
operating conditions. This is accomplished by directly measuring the torque on the
non-rotating bearing over the ranges of load and speed expected in turbomachinery
applications. Power loss is determined simply by multiplying the measured bearing
torque by the rotational speed of the shaft.
36
Property Inconel 718 Aluminum 7075-T6
Young’s Modulus (GPa) 200 71
Yield Strength (MPa) 1185 503
kg
Density ( m3 ) 8220 2800
Poisson’s Ratio 0.29 0.33
W
Thermal Conductivity ( m◦ C ) 11.4 130
−6 m
Thermal Expansion Coefficient (×10 m◦ C ) 12.1 21.6
Specific Heat Capacity ( kgJ◦ C ) 430 960
Table 4.2: Material Properties of Test Runner Materials at 20◦ C9
The baseline power loss data is collected by incrementally increasing the load at
constant speed until reaching a predefined upper limit (typically 75-100% of load
capacity, determined by a priori knowledge of the bearing limits). Load is generally
held for two minutes or more to allow transient thermal phenomena to subside in the
bearing/runner system. At low loads the torque tends to reach its steady state value
immediately, while at loads nearer to the capacity of the bearing, the torque trace
may exhibit settling times that can exceed a few minutes. These load increments are
held longer to allow the torque to settle before recording the “steady state” value.
This process is repeated at various speeds of interest to obtain a family of curves
depicting bearing power loss. Since cooling techniques can affect the operating torque
at a given set of conditions by changing the thermoelastic distortions and resulting
film thickness distribution, power loss curves must be measured with each cooling
condition and compared to the baseline case.
Although the experimentally measured bearing torque describes only the aggre-
gate gas film behavior and does not provide information on local film characteristics,
37
a great deal of insight can be gained from these data. If a constant film thickness is
assumed, bearing torque can be used to determine the characteristic film thickness and
compressibility number, which are useful in analysis of the gas film. Appendix B gives
a description of this analysis technique along with a derivation of these quantities.
4.1.5 Load Capacity Testing
Tests of load capacity are performed by incrementally increasing the bearing load
at a constant speed, allowing one to two minutes of dwell time between successive
load increments. This dwell time is consistent with the amount of time required for
foil temperature gradients to approach steady state, and is characteristic of torque
trace settling times. At loads close to the bearing capacity, the torque signal becomes
noticeably erratic and more time between load intervals is allowed. This process
attempts to capture steady state load capacity behavior.
Load capacity is reached when rapid increases in the torque signal occur with
little or no additional load. With near-capacity operation characterized by unsteady
torque and intermittent asperity contact, there is a significant degree of variability in
the measured load capacity (±5% or more). A data trace from a typical load capacity
test is given in Figure 4.10.
4.1.6 Wear Measurements
In order to analyze wear of the foil coatings resulting from operation, an optical
profilometry system is used. The NewView 5000, manufactured by Zygo Corporation
of Middlefield, CT, uses scanning white light interferometry to image and measure
38
the top foil topography (see Figure 4.11) in post-test inspections.
Light directed through the microscope is split within an interferometric objective
(lens). One portion is reflected off a highly polished internal reference surface in
the objective and the other is reflected off the part to be measured. The resulting
fringes are analyzed to determine the surface topography of the part. Depths up to
100 μm can be imaged with 0.1 nm vertical resolution and 0.4 nm root mean square
repeatability, regardless of objective magnification. Lateral resolution, which depends
on the field of view of the objective in use, ranges from 0.45 μm at 100x magnification
to 11.8 μm at 1x magnification.
Multiple scans of parts up to 13 cm in diameter can be stitched together to form
a topographic image of the whole part. This procedure is used to produce the surface
profiles of individual bearing top foils, which shows top foil coating wear from high-
load operation. A detailed description of these thrust bearing wear measurements is
given by Lucero10 .
4.1.7 Bearing Temperature Measurement
The temperature field within the gas film is of interest to researchers in order
to characterize the heat transfer where the heat is generated and temperatures are
highest. Direct temperature measurements within the gas film would be difficult, if
not impossible, due to its thinness. Therefore, representative measurements of film
temperatures must be made by instrumenting the bearing foils or runner as close to
the gas film as possible. Within the bearing, temperatures of the top foil are the most
representative of gas film temperatures as it is the immediate boundary of the film
39
100 350
Speed
90 Torque
300
Load
80
Speed (krpm), Torque (N*mm)
70 250
60
200
Load (N)
50
150
40
30 100
20
50
10
0 0
0 10 20 30 40 50
Elapsed Time (minutes)
Figure 4.10: Time Trace of Typical Load Capacity Test
Figure 4.11: Photograph of Optical Profilometer
40
and has limited thermal conduction.
Due to the complex and relatively thin nature of the bearing foil structure,
instrumentation of the foils is made difficult by the requirement that this not deform
the foils or interfere with the designed operation of the bearing. The bump foil
support structure limits access to the top foil, so an alternate approach is to attach a
thermocouple to a bump foil that is in contact with the top foil. Radil and Zeszotek11
used this method to measure foil temperatures in compliant journal bearings. They
routed a thermocouple through an access hole in the bearing shell and tack-welded
the junction to the backside of a bump at a point where it was in immediate contact
with the top foil.
Although thrust bearings have different geometry, they can be instrumented in
the same manner, as shown in Figure 4.12. A hole is made through the bearing plate
by first drilling most of the way through the plate, finishing the hole using electric
discharge machining (EDM). A thin flexible forty gauge (0.076 mm diameter), type
K, polymer-sheathed thermocouple wire is fed through the hole and the junction is
attached to the backside of the bump foil where it is in direct contact with the top
foil. Whereas the thermocouple junctions in Radil and Zeszotek’s test bearings were
tack-welded to the bump foil, for this work they will be attached with a thermally
conductive epoxy. Initial tests with tack welded thermocouples showed that highly
localized foil distortions often occurred which had an adverse effect on bearing per-
formance, as was primarily evidenced by wear scars at the thermocouple locations
(Figure 4.13). The epoxy chosen has high thermal conductivity but is only rated to
a temperature of 260◦ C (500◦ F). Although this temperature limit restricts testing to
41
low ambient temperatures, it represents a more reliable, non-intrusive instrumentation
method for foil thrust bearings.
Thermocouple-instrumented thrust bearings are first used to provide baseline
data on characteristic temperatures and temperature gradients within the bearing
at various operating conditions. Bearing size and instrumentation techniques limit
the number of thermocouples per pad to a maximum of about three to five. These
can be arranged to provide information on temperature gradients of interest, such as
radial gradients at a given angular location. The effectiveness of thermal management
techniques is evaluated based on the reduction of temperature gradients and efficiency
of heat removal. Those thermal management techniques investigated include active
cooling, such as blowing air through the bearing, and passive cooling methods, such
as runner assemblies with high thermal conductivity components.
4.2 Numerical Techniques
Because hardware-intensive approaches to foil bearing development are costly and
time consuming, analytical and numerical methods to accurately predict bearing
performance are desirable to reduce development costs. Although this study is
largely experimental, two numerical techniques are employed to corroborate observa-
tions of foil thrust bearing operational characteristics. Finite difference (FD)-based
hydrodynamic modeling and finite element (FE) structural modeling both support
experimental observations.
42
Figure 4.12: Thrust Bearing Thermocouple Instrumentation
Figure 4.13: Foil Wear Scar Due to Tack Weld Distortions
43
4.2.1 Hydrodynamic Modeling
Numerical simulations of bearing operation to determine torque and load capacity
behavior are obtained using the the equations, methods, and codes described by
Bruckner12 . His general thrust foil bearing model includes a total of nine unknowns,
which include the three scalar velocity components, pressure, temperature, density,
viscosity, specific heat capacity, and structural deflection.
A scaling analysis employing the traditional assumptions of hydrodynamic thin
films permits the elimination of the three velocity components and associated mo-
mentum equations. Furthermore, for the purpose of simplifying the effects present in
the model, an isothermal assumption is made which eliminates the need to solve the
energy equation. The heat capacity and viscosity relationships, which were assumed in
the general model to be functions of temperature only, can be set as constants (equal
to unity when expressed in nondimensional terms) with the isothermal assumption.
The three remaining variables are the gas pressure and density, and the structural
deflection. The system of equations which allows for the solution of these variables
consists of the compressible Reynolds equation, an equation of state, and a structural
model. Under the isothermal conditions assumed, the ideal gas equation of state
allows the nondimensional pressure to be set equal to the nondimensional density. As
such, the Reynolds equation and structural model may be solved for the pressure and
film thickness distribution.
The structural model used consists of a bending-dominated top foil reinforced
by a supporting bump foil structure. The bump foil structure is modeled as a set of
44
discrete bumps with an aggregate stiffness equal to that of a test bearing. This model
permits the mechanism of foil sag between bumps to be shown in the deformed top
foil shape, within the limits inherent to a bending-dominated top foil model. Further
detail of this structural model is given by Bruckner12 .
The exact foil thicknesses and other geometric details of the test bearings are not
given here, as they may be considered proprietary. However, a cursory review of the
patent literature shows that foil thicknesses on the order of 0.1mm are typical. The
numerical results presented herein are based on the actual (measured) geometry of
the test bearings in order to closely approximate the deformation behavior. They
are intended to demonstrate that this previously developed code is validated by the
experimental data now available.
4.2.2 Modeling of Structural Deformation
In order to examine the flatness of the runner surface against which the bearing
operates, mechanical and thermal stresses are modeled in a commercially available
finite element (FE) analysis software. A three dimensional model of a standard thrust
runner is created as an input to an automatic meshing routine, which generates a close
representation of the runner geometry with linear tetrahedral elements. Material
properties for Inconel 718 at 25◦ C, shown in Table 4.2, are used.
The model is constrained by fixing the cylindrical outer surface of the cylindrical
protrusion to prevent nodal displacement axially or circumferentially. Radial growth
is not restrained however, and since the portion of the runner in which deflections are
of interest is sufficiently far away from the static constraint, the difference between
45
this imposed constraint and the physical system has a vanishingly small impact on
the displacements of interest. Centrifugal loads are applied to the model and a static
analysis is run to determine resulting displacements.
Thermal modeling assumes all exposed faces of the runner are insulated unless
another boundary condition is imposed. Boundary conditions applied in this work
are specified heat flux (Neumann) and Newton’s law of cooling (Robin) conditions.
Heat flux is specified at the bearing running surface on the front of the runner and a
convection condition is applied to the rear, exposed face. For convective heat transfer
on the back face, a constant, average assumed convection coefficient is imposed,
leading to a purely axial heat flux in the case of an axisymmetric runner with an
insulated outer rim.
To determine the runner thermoelastic behavior, the temperature distribution is
used as an input to a static analysis of runner stress and displacement, with the
temperature of the backside of the runner is used as a reference for zero thermal
strain. Displacement plots for mechanical, thermoelastic, or combined stresses can
be generated in parametric studies of runner behavior at various operating conditions
References
[1] Hryniewicz, P., Locke, D., and Heshmat, H., 2003. “New-Generation
Development Rigs for Testing High-Speed, Air-Lubricated Thrust Bearings”.
Tribology Transactions, 46(4), pp. 556–559.
[2] Bauman, S., 2005. An Oil-Free Thrust Foil Bearing Facility Design, Calibration,
46
[3] DellaCorte, C., 1998. The Evaluation of a Modified Chrome Oxide Based High
Temperature Solid Lubricant Coating for Foil Gas Bearings. NASA TM 1998-
208660, National Aeronautics and Space Administration, Cleveland, OH.
OH.
[6] Gray, S., and Heshmat, H., 1981. Support Element for Compliant Hydrodynamic
Thrust Bearing. U.S. Patent 4,277,111.
[7] Agrawal, G., 1984. Hydrodynamic Fluid Film Thrust Bearing. U.S. Patent
4,462,700.
[8] Heshmat, H., Hryniewicz, P., Walton, J., Willis, J., Jahanmir, S., and
DellaCorte, C., 2005. “Low Friction Wear-Resistant Coatings for High-
Temperature Foil Bearings”. Tribology International, 38.
[9] Davis, J., ed., 1998. Metals Handbook Desk Edition. ASM International.
[10] Lucero, J., 2006. Measured Foil Frequencies and Mode Shapes as a Function
of Wear for a Gas Foil Thrust Bearing. NASA-TM (to be published), National
[11] Radil, K. C., and Zeszotek, M., 2004. “An Experimental Investigation into the
Temperature Profile of a Compliant Foil Air Bearing”. Tribology Transactions,
47(4), pp. 470–479.
and Structural Behavior of Foil Thrust Bearings”. PhD dissertation, Case
Western Reserve University, Cleveland, OH.
47
Chapter 5
Bearing Torque and Power Loss

Results
5.1 Experimental Torque/Power Loss Measurements
Bearing torque and power loss are measured over a wide range of operating con-
ditions in the test thrust bearings. A number of factors which affect bearing torque
including speed, load, cooling flow rate, and surface conditions are varied and torque
measured. The plots of torque and power loss data in this section are single measure-
ments and are plotted without error bars for clarity. Generally speaking the data are
very repeatable (±3Nmm). The raw experimental data presented in this section can
be found in tabular form in Appendix D.
5.1.1 Bearing Torque versus Speed and Load
Very little foil thrust bearing torque data is available in the open literature over a
range of applicable speeds and loads. This lack of information remains a hurdle for
turbomachinery designers, who are concerned with the power loss in the bearing and
48
the corresponding effect on thermal management requirements for stable operation.
To look at the effects of speed and load on bearing torque and power loss, bearing
B is tested against an Inconel 718 runner with a PS304/MoS2 surface as described in
Chapter 4, without any cooling air flow. At speeds from 25-55 krpm, the measured
bearing torque is given in Figure 5.1, and the corresponding power loss is given in
Figure 5.2, as reported by Dykas et al1 . The torque and power loss measurements
shown include measurements up to near the measured load capacity as detailed in
Chapter 6.
80
25 krpm
30 krpm
70 35 krpm
40 krpm
45 krpm
60 50 krpm
55 krpm
50
Torque (N mm)
40
30
20
10
0
0 10 20 30 40 50 60 70 80
Load (kPa)
Figure 5.1: Torque Versus Load From 25-55 krpm of bearing B against a PS304-
coated runner, showing that torque generally increases with larger loads
and higher speeds. It is notable that the effect of a doubling speed from
25 to 50 krpm has the effect of increasing torque only by about 35%, as
measured at approximately 65 kPa.
49
100
25 krpm
90 30 krpm
35 krpm
40 krpm
80
45 krpm
50 krpm
70 55 krpm
Power Loss (kW/m )
2
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80
Load (kPa)
Figure 5.2: Power Loss Versus Load From 25-55 krpm of bearing B against a
PS304-coated runner, showing the same data as in the previous figure,
but multiplying torque by speed to obtain power loss on the ordinate.
Power loss is shown to increase with both speed and load.
It can be seen in these plots that for the conditions tested, the bearing torque and
power loss are relatively small for low to moderate loads. Power loss on the order
of hundreds of watts is typical for this range of operation. The bearing torque and
power loss increase with both speed and load, which both tend to increase shear rate
within the gas film.
Based on tests of load capacity discussed in Chapter 6, the sustainable axial loads
using a rougher runner surface without cooling air are somewhat lower than with
smoother surfaces and active thermal management. Figure 5.3 shows the effect of
50
speed and load on bearing torque when bearing B is run under more ideal conditions
of surface roughness and thermal management. Again, the torque is given over the
full range of bearing operation, from light loads to near the load capacity for a given
speed. It is notable that with these more ideal surface and thermal conditions, the
bearing is able to support significantly more load and run at higher rates of heat
generation, due to a smaller attainable film thickness.
180
25 krpm
160
40 krpm
140
120
Power Loss (kW/m )
2
100
80
60
40
20
0
0 20 40 60 80 100 120 140 160
Load (kPa)
Figure 5.3: Bearing Power Loss vs Load with Chromium-Coated Runner

and Cooling Flow for bearing B, showing the bearing behavior when a
smoother runner and active cooling flow are used to allow it to operate
with larger axial loads.
51
5.1.2 Effect of Runner Surface Roughness
Bearing torque is measured with both the PS304 and chromium coatings to
determine the extent to which the surface roughness affects the torque over the
range of average runner surface roughnesses to be expected in a real turbomachine.
Figure 5.4 shows torque versus load for both runner coatings at 25 krpm with 0.52
kg/min of cooling flow. The lower speed and higher cooling flow rate is chosen for
this comparison in order to minimize any thermal management effect on the torque
curves.
90
PS304
80 Chromium
70
60
Torque (N mm)
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
Load (kPa)
Figure 5.4: Bearing Torque vs Load Showing Effect of Runner Surface

Roughness, using bearing B at 25 krpm and 0.52 kg/min cooling air flow.
Torque vales differ slightly, with the rougher surface showing evidence of
less torque than the smoother surface at otherwise identical operating
conditions.
52
It is notable that the smoother chromium surface results in a higher bearing
torque given the same operating conditions as compared to a PS304-coated bearing.
However, the curves are obtained from two bearings of identical design, but the
operation of each may differ slightly. As a result, any effect that roughness has on
performance may be overwhelmed by the variations of the bearings. Since the flow in
the gas film is likely laminar due to the thin gap, roughness may be expected to have
no noticeable effect on shear stress provided it is small compared to the film thickness.
Although it is difficult to draw a strong conclusion based on this experimental data,
it is presented for the sake of completeness.
5.1.3 Effect of Cooling Flow
Since cooling air flow is able to increase load capacity and tolerable rates of heat
generation, it is desirable to determine if varying the cooling flow rate at a given
speed and axial load condition affects the bearing torque. This secondary effect, the
result of cooling effects on film thickness distribution and/or viscosity variation, may
be significant if wide ranges of cooling flow rates and heat generation are considered.
Figure 5.5 shows the measured bearing torque for bearing B at 55 kprm over a
range of loads and cooling air flows. For the range of cooling flows tested, it appears
that the flow rate does have an effect on bearing torque, but at high loads where heat
generation is largest, this effect is a modest percentage of absolute torque (∼25%).
Figure 5.6 shows the maximum attainable power loss over a speed range from 25-
55 krpm with cooling flow rates up to 0.52 kg/min. A rather large amount of scatter is
present in these data due to erratic torque behavior near the bearing load capacity and
53
70
Cooling Flow Rate
0 kg/min
60 0.17 kg/min
0.35 kg/min
50
Torque (N mm)
40
30
20
10
0
0 10 20 30 40 50 60 70 80
Load (kPa)
Figure 5.5: Effect of Cooling Flow on Bearing Torque at 55 krpm Against

PS304 Surface for bearing B, showing a small effect of cooling air flow
on the resulting bearing torque. For a given load, torque appears to
decrease as cooling flow is increased.
54
corresponding difficulty getting accurate steady state torque measurement, but this
plot shows a general trend of larger attainable heat generation rates with increased
speed. The absolute power loss values are relatively small, particularly compared to
the values of heat generation in Figure 5.3, owing to unbroken-in PS304 surface used
to generate these data. For the flow rates tested, cooling flow does not appear to
have a large effect on the maximum sustainable bearing power loss.
100
Cooling Flow Rate
90 0 kg/min
0.17 kg/min
0.35 kg/min
80
70
Power Loss (kW/m2)
60
50
40
30
20
10
0
0 10 20 30 40 50 60
Speed (krpm)
Figure 5.6: Power Loss at Load Capacity for Various Cooling Flow Rates for
bearing B run against a PS304-coated runner at various speeds. Tolerable
power loss near load capacity is shown to increase with speed, but cooling
appears to have little effect.
55
5.2 Numerical Predictions of Bearing Torque
Experimental data is used to validate numerical models of the gas film. While
standard thin film assumptions are clearly applicable in foil gas bearings, the isother-
mal assumption noted in Section 4.2.1 may not be appropriate for some or all regions
of bearing operation. In physical terms, numerical results presented here represent
a highly cooled bearing where any heat generated by compression or dissipation is
immediately removed through the gas film boundaries (foil or runner surfaces).
In Figure 5.4, the effect of runner surface roughness was compared for a bearing
operating at 25 krpm and with 0.52 kg/min of cooling air flow. Since this speed
is the lowest tested in this study, the bearing heat generation is the lowest of any
speed, as shown in Figure 5.2. This in conjunction with a relatively large cooling flow
rate yields gas film conditions that more closely approximate isothermal conditions
than would higher speeds and lower cooling flow rates. Furthermore, the torque data
obtained for a runner coated with thin dense chrome more closely approximates the
perfectly smooth surfaces assumed in a numerical model.
A numerical analysis of an isothermal bearing operating at 25 krpm is performed
over a range of loads, where the deformed foil shape at the previous iteration is used as
an input to the subsequent iteration. The bearing torque, pressure distribution, and
resulting deflected foil shape are then calculated to determine the load/torque data
and a deflected foil shape. At each successive iteration the runner surface is brought
slightly closer to the bearing to increase the load. While this iteration scheme does
not provide a self-consistent solution at each iteration, sufficiently small steps reduce
56
the bias error in the solution.
Figure 5.7 is a plot of the data which appears in Figure 5.4, along with a nu-
merical calculation of a torque/load curve at 25 krpm. The calculated curve gives
torque magnitudes that closely approximate those obtained experimentally with a
smooth runner. Agreement between experimental and numerical results is better at
moderate bearing loads, where a thrust foil bearing would be designed to operate.
At lower loading, the bearing foils may not have fully seated on the bump foils and
manufacturing variations between pads and bearings may have large effects on the
measured torque. At higher loading, the effects of foil sag between bumps may not
be accurately accounted for in the numerical model. Furthermore, increasing heat
generation at higher loads may render the isothermal assumption in the numerical
model less valid.
A higher speed simulation of bearing load/torque behavior is calculated at 55
krpm, where heat generation is larger. In this case, the calculated data are plotted
against experimentally measured torque curves from a PS304-coated thrust runner
with varying amounts of cooling flow (Figure 5.8). Based on the previous figure, it is
expected that the data generated from the numerical study more closely approximates
the behavior of a bearing running with a smoother coating. However, limited exper-
imental data with the thin dense chrome-coated runner restricts plotting options.
The isothermal model can be thought of as having a large (tending toward infinite)
cooling flow rate, and so it appears that the numerical results continue the trend
of decreasing torque with increasing cooling flow rate. In general, this numerical
model is able to predict with reasonable accuracy the magnitude and character of the
57
90
Chromium
80 PS304
Numerical
70
60
Torque (N mm)
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
Load (kPa)
Figure 5.7: Comparison of Numerical Predictions of Bearing Torque to

Experimental Data at 25krpm for bearing B operating against both
smooth TDC and as-ground PS304 runner surfaces. The numerical
simulation shows generally good agreement with the torque magnitude
for a bearing operating against a smooth runner surface, particularly at
moderate loading conditions.
58
load/torque curves for a simple thrust foil bearing geometry.
70
0 kg/min (Experimental)
0.17 kg/min (Experimental)
60 0.35 kg/min (Experimental)
Isothermal (Numerical)
50
Torque (N mm)
40
30
20
10
0
0 10 20 30 40 50 60 70 80
Load (kPa)
Figure 5.8: Comparison of Numerical Predictions of Bearing Torque to

Experimental Data at 55krpm where experimental data is obtained
from bearing B running against an as-ground PS304 runner with various
cooling flow rates and over a range of loads. The isothermal case studied
numerically shows good agreement with the trending evident in the
experimental data.
In both Figures 5.7 and 5.8, the plot of torque versus load contains an inflection
point at moderate loads. This behavior is not observed in the experimental data,
and its exact cause is not fully understood. However, as previously noted, under
very lightly loaded conditions where the top foil has not been fully seated in the
experimental configuration, the numerical model overpredicts the bump foil support
stiffness and may not capture the real film thickness distribution. At higher loads
59
beyond the inflection point, both the experimental and numerical data exhibit positive
curvature. As the film thickness decreases at high loads, the amplitude of top
foil waviness due to foil sag becomes large relative to the film thickness, and any
inadequacies of the structural model can have a large effect on the calculated torque.
Irrespective of the extreme ends of bearing loading, the previously developed model2
appears to predict bearing torque with reasonable accuracy at moderate loads.
References
60
Chapter 6
Load Capacity Results
Load capacity is a performance metric used to determine the effectiveness of the foil
structure in maintaining an adequate gas film. Over the past few years, load capacity
has risen with advanced bearing designs1,2 and by employing adequate cooling air
and very smooth runner surfaces. Of interest is the effect that cooling has on the
load capacity of the bearing, which is an indicator of the importance of thermal
management design relative to the physical design of the bearing. Furthermore, since
high temperature, durable coatings such as PS304 may require special application and
finishing techniques3 or overlay coatings to achieve highly smooth surface finish at
entry into service, it is also necessary to quantify the effect of typical surface finishes
on foil thrust bearing performance.
61
6.1 Experimental Measurement of Load Capacity
6.1.1 Load Capacity as a Function of Speed
Until recently, even basic estimates for achievable load capacity in foil bearings
were not readily available to turbomachinery designers. To address this particular
need, DellaCorte and Valco1 proposed a “rule of thumb” model to predict foil journal
bearing load capacity. This simple tool matches experimental data well over a wide
range of bearing sizes and speeds. Their model predicts a linear relationship between
journal speed and load capacity for the speed range of engineering interest. They
extended the rule of thumb concept to thrust foil bearings, but their complementary
model has not been corroborated due to a lack of sufficient experimental data.
Figure 6.1 shows a plot of load capacity versus speed for bearing A running against
a PS304-coated runner. The load capacity is given in units of pressure, where the
largest supported axial load has been divided by the total top foil area to give an
average pressure at capacity. Note that this differs from the bearing swept area based
on inner and outer foil diameters by about 10%. Tests are performed from 25-55 krpm
(89-196 m/s runner surface velocity based on mean bearing diameter). Load capacity
is not measured below 25 krpm because of test rig operational complications, but it
should be noted that the load capacity tends toward zero as the speed is reduced to
the liftoff speed of the bearing.
In contrast to the behavior of foil journal bearings in DellaCorte and Valco’s study,
this thrust foil bearing does not exhibit a load capacity that increases linearly with
speed. The highest recorded load capacity is found to occur at the lowest speed
62
80
70
60
Load Capacity (kPa)
50
40
30
20
10
0
0 10 20 30 40 50 60
Speed (krpm)
Figure 6.1: Thrust Bearing Load Capacity as a Function of Speed for bearing
B running against a PS304-coated runner in the absence of cooling air
flow, showing a decreasing trend in load capacity with increased speed.
Load capacity is said to approach zero at zero speed.
63
tested, with the lowest load capacity occurring at the highest speed tested. Load
capacity therefore is shown to trend downward with speed over this range when the
thrust bearing is run against a PS304-coated runner without any cooling air flow. It
has been suspected that this behavior is related to inadequate thermal management
of the bearing-runner system.
6.1.2 Effect of Cooling Air Flow
In general, larger loads on the bearing result in a thinner gas film, which in turn
leads to increased shear rate. This leads to increased shear stress and more heat
generation. Reduced volume of gas in the thinner film results in a lower total heat
capacity of the film. The higher temperatures which result tend to increase the
viscosity of the gas, which theoretically should increase load capacity. This is in
contrast to the temperature dependence of liquid lubricants, which lose effectiveness
with increasing temperature due to reduced viscosity.
In practice however, foil bearings do not exhibit higher load capacities if tempera-
ture is allowed to increase. Heat transfer within the bearing systems has the undesir-
able effect of producing thermoelastic distortions within the structural components.
Temperature gradients within the bearing and runner can result in deformations
larger than the gas film thickness, which is typically on the order of tens of microns
or less. These deformations can then lead to local gas film rupture as described by
Dykas and Howard4 .
The determination of the effect of cooling on load capacity is a first step to
understanding the degree to which thermal phenomena influence bearing operation.
64
In this work, cooling air is used ostensibly to reduce structural temperature gradients
and the consequent thermoelastic distortion.
Figure 6.2 shows a plot of load capacity versus speed with varying amounts of
cooling flow through the bearing. Tests are performed with cooling flow rates of 0,
0.17, 0.35, and 0.52 kg/min against a PS304-coated runner.
120
Cooling Flow Rate
0 kg/min
0.17 kg/min
100 0.35 kg/min
0.52 kg/min
80
Load Capacity (kPa)
60
40
20
0
0 10 20 30 40 50 60
Speed (krpm)
Figure 6.2: Effect of Cooling Flow on Load Capacity for bearing B running
against a PS304-coated runner with various cooling flow rates. Load
capacity is seen to increase with cooling flow rate.
It is shown that at 25 krpm, load capacity is relatively insensitive to cooling
flow rate, indicating that thermal management may not be critical at speeds below
this. Differences in measured load capacity at this speed are within the estimated
65
uncertainty of approximately ±5 kPa. Above 25 krpm however, load capacity is seen
to increase with increasing cooling flow beyond the estimated uncertainty, indicating
that at speeds greater than the lowest tested, thermal management is needed to
achieve the highest possible performance.
Furthermore, it is notable that at each cooling condition, load capacity tends to
flatten or trend downward with increasing speed. This decrease in load capacity with
increasing speed is contrary to current understanding of foil gas bearings. It may be
postulated that this decrease in load capacity is due, at least partly, to the increase
in heat generation and thermoelastic deformation with increasing speed (see Chapter
5 for the effects of speed and load on heat generation). Based on this hypothesis, a
reduction of temperature gradients in the structural elements would yield increased
load capacity, a result which suggests that alternate thermal management methods
including passive structural features may be possible.
6.1.3 Effect of Runner Surface Finish
Since SOA numerical modeling assumes a perfectly smooth runner surface, the
effect of real surfaces on load capacity is of interest. The thin dense chromium coatings
used in laboratory tests typically have surface roughness of Ra = 0.05-0.10 μm, which
is small compared to estimated film thicknesses. The unbroken-in PS304 surface
by comparison has surface roughnesses of 0.3-0.8μm, closer to the estimated film
thicknesses. Quantifying the effect of surface roughness on the load capacity behavior
of the bearing is needed to determine the requisite surface condition of production
thrust runners.
66
The two runner coatings (PS304 and TDC) are compared in load capacity tests
of bearing A with 0.52 kg/min of cooling flow. A cooled configuration is chosen in
order to reduce the thermal effects and to lessen the risk of destructive failure. Load
capacity of the chromium-coated runner is tested at 25, 35, and 40 krpm, while the
PS304-coated runner is tested only at 25 krpm in this configuration. A plot of load
capacity for the two surface roughnesses is given in Figure 6.3. Although the rougher
surface is tested at only one speed, the difference in performance can be clearly seen
from these data. The load capacity against PS304 at 25 krpm, where cooling air has
been shown to have little effect, is only 64% of that when a smooth chromium-coated
runner is used. This difference in load capacity of the two runner surfaces is consistent
with the results obtained by Radil and DellaCorte5 in foil journal bearings.
6.1.4 Impact of Top Foil Coating
Almost invariably, bearing top foils are coated with a thin (on the order of 25
μm thickness) solid lubricant coating. The addition of this coating is intended to
improve tribological properties from the baseline foil material5 . The improved friction
coefficient can be verified in pad-on-disk or equivalent tests. However, it is notable
that these foil coatings are typically soft materials which do not exhibit particularly
long wear life.
In a series of checkout runs with the test rig described herein, uncoated thrust
bearings demonstrated severely limited load capacity relative to their expected perfor-
mance. Bearings with identical foil design which had a solid lubricant coating applied
to the top foil were found to have dramatically higher load capacity in comparison
67
200
Chromium
180 PS304
160
140
Load Capacity (kPa)
120
100
80
60
40
20
0
0 5 10 15 20 25 30 35 40 45
Speed (krpm)
Figure 6.3: Effect of Runner Surface Finish on Load Capacity for bearing
A running against both PS304 and chromium runner coatings with 0.52
kg/min of cooling air flow. The smoother chromium surface is able to
support significantly more load compared to the as-ground PS304 surface.
68
tests. This result underscores the importance of advanced coating systems for bearing
foils.
To explore the function of a thick solid lubricant foil coating, load capacity was
tested in a bearing with a much thinner low friction top foil coating. A bearing
identical to bearing B was subjected to a high temperature environment to burn
off the foil coating. After the coating was removed, a colloidal form of molybdenum
disulfide (MoS2 ) in alcohol was sprayed onto the surface and lightly polished. This was
done to provide a low friction coefficient in the event of asperity contact or light dry
rubbing. The replacement of a comparatively thick (25 μm) solid lubricant coating
with a thin (∼3 μm) MoS2 film is significant in that the low friction characteristics
of the foil are retained, but the coating thickness and adhesion to the surface is much
reduced.
A load capacity test was then performed on the bearing at 30 krpm with 0.69
kg/min of cooling air flow. Based on tests of a coated bearing at these conditions,
the expected load capacity was more than twice that at which the uncoated bearing
failed. Figure 6.4 shows the wear scars on the failed bearing. In this photograph the
large, shiny area on the foil in the three o’clock position is unrelated to the failure.
A conclusion drawn from these results is that the foil coating may serve a function
beyond improved lubricity. A low friction foil coating is a necessary but not sufficient
condition for maximum performance from a given bearing design. Visual evidence
and profilometric data of coating wear from previous tests suggest that an additional
function of thick (∼25μm) top foil coatings is to provide an abradable surface that,
when run-in, compensates for foil sag between bump foils. A post-test profile of a
69
Figure 6.4: Failure of Bearing with Uncoated Top Foils with wear scars visible
toward the inner diameter of the top foils. These scars are most
pronounced where the bump foils are in contact with the top foil.
bearing that has been highly loaded is given in Figure 6.5.
6.2 Numerical Prediction of Load Capacity
Experimental observations of load capacity behavior in foil thrust bearings show
a dependence on speed and cooling flow rate which can be examined further in a
numerical model. It is hypothesized that thermoelastic distortions play a role in
limiting the bearing load capacity, a theory supported by the increase in capacity
with increased cooling flow rate. The tendency for load capacity to decrease with
increasing speed (and corresponding increase in heat generation) also supports this
theory, and suggests that the bearing requires a larger average film thickness near
load capacity when thermoelastic distortions are larger.
70
Figure 6.5: Thrust Bearing Surface Profile After High Load Operation,
showing the topography of a top foil coating after being worn by high
load testing. A profile along a line extending from the leading edge to the
trailing edge as indicated is given in the lower portion of the figure. The
deepest wear troughs are located where the bump foils support, and are
in contact with, the top foil.
71
Figure 6.6 contains the load capacity data previously shown in Figure 6.2, with
the addition of some load versus speed curves generated by a numerical analysis.
The numerical curves are obtained by setting an initial bearing-to-runner spacing
at a low speed and increasing speed in small increments, calculating the resulting
pressure distribution and foil deflection at each step. The hydrodynamic pressure
increases with speed, and the foil deflects slightly with the increased pressure load.
In this iteration scheme, the film thickness is not exactly constant, but calculations
of bearing torque at each iteration enable the calculation of a Couette film thickness
that is very nearly constant. This scheme tends to produce a load-speed curve with
a slight negative curvature as the structure relaxes to allow increased film thickness
at higher speeds.
In order to demonstrate the effect of the film thickness on the hydrodynamic
bearing load, three initial gaps are set between the bearing and runner, while using
the same foil geometry. The load curve obtained using the smaller initial gap yields
higher bearing loads in response to the larger contraction ratio. This can be seen in
Figure 6.6, while the curve at lower loads is obtained from a larger initial (low speed)
film thickness. Note that the negative curvature of the curves increases as the film
thickness decreases. In addition to the contribution of foil deflection to the curvature,
this behavior is consistent with high speed behavior (large compressibility number)
in gas bearings.
The nature of the experimental data compared to the numerically-determined load
curves supports the theory that thermoelastic distortion has the effect of increasing
the required average film thickness at load capacity. A family of curves for various
72
120
Experimental (0 kg/min)
Experimental (0.17 kg/min)
100 Numerical (small gap)
Numerical (moderate gap)
Numerical (large gap)
80
Load (kPa)
60
hc = 6.3 Pm
hc = 9.4 Pm
40
hc = 12.4 Pm
20
0
0 10 20 30 40 50 60
Speed (krpm)
Figure 6.6: Comparison of Numerical Predictions of Bearing Load Capacity

to Experimental Data using experimental data from bearing B and
an isothermal numerical model of a simple thrust bearing. The load
capacity behavior described by the numerical model is based on an
initial prescribed bearing-to-runner gap with speed increased in small
increments. Curves based on three different initial gaps are given, with
the smallest gap giving rise to higher loads due to a larger contraction
ratio.
73
average film thicknesses can be drawn between the small-gap and large-gap curves
in Figure 6.6, and it is demonstrated in this figure that for a given cooling flow
rate (tracing one of the experimental curves), the maximum load supportable by
the bearing occurs at larger film thicknesses as speed (and heat generation) increase.
Alternatively, to reach the same characteristic film thickness at load capacity (tracing
one of the numerical curves with roughly constant hc ), cooling flow rate must be
increased with speed.
It should be further noted that in the numerical simulations, top foil sag between
bumps results in a wavy bearing surface. As demonstrated earlier in this chapter,
experimental observations show that in practice, the foil coating can be abraded
at high loads in order to compensate for foil sag, resulting in a bearing surface that
under hydrodynamic pressure, is flatter with respect to the runner. This phenomenon
is not accounted for in the numerical simulations, and the wavy surface can create
nonsensical artifacts at very high loads where the film thickness in portions of the
bearing tend toward zero (see the highest loads in the hc = 6.3μm curve in Figure
6.6).
6.3 General Observations
Based on this series of load capacity tests, it is shown that factors other than
the particular bearing design are critical to achieving maximum performance in a
foil thrust bearing system. These factors include thermal management and surface
condition of both the bearing and runner, previously studied to some degree in foil
74
journal bearings, but perhaps more critical in foil thrust bearings which typically
operate at a higher percentage of their rated load capacity.
Considering its demonstrated effect, very little quantitative data is reported for
the effect of thermal management in foil bearings. It is now obvious that this is an
important consideration, in particular because it is desirable to minimize parasitic
cooling flows in turbomachinery. Further, since bearing manufacturers typically
employ cooling air for foil bearing thermal management in laboratory testing, reported
data should be considered within the context of any thermal management technique
used.
The effects of runner surface condition and foil coatings, while not yet reported
in thrust foil bearings, have been examined to some degree in foil journal bearings5,6 .
When compared with a thin dense chromium coating, the reduced performance asso-
ciated with a rougher as-ground PS304 surface is consistent with reported results for
journal bearings. However, to the author’s knowledge the use of a thick abradable
foil coating as a wear surface for in situ optimization has not been reported. The
need for foil coatings to perform this function may be minimized with better a priori
design and manufacturing methods.
The result that these bearings are limited thermally and by surface condition
presents a challenge for turbomachinery designers, who must design a rotor support
system around issues such as these. In the following chapters these effects are
discussed on a more fundamental level in an attempt to understand how to overcome
their limitations.
75
References
[2] Heshmat, H., 2005. “Major Breakthrough in Load Capacity, Speed and Operating
Temperature of Foil Thrust Bearings”. In Proceedings of WTC2005, World
Tribology Congress III, no. WTC2005-63712, American Society of Mechanical
Engineers/Society of Tribologists and Lubrication Engineers.
[3] Kim, J. H., Blanchet, T., and Calabrese, S., 2004. “High Velocity Oxyfuel
Deposition for Low Surface Roughness PS304 Self-Lubricating Composite
Coatings”. Tribology Transactions, 47(1), pp. 157–169.
[4] Dykas, B., and Howard, S. A., 2004. “Journal Design Considerations for
Turbomachine Shafts Supported on Foil Air Bearings”. Tribology Transactions,
47(4), pp. 508–516.
[5] Radil, K. C., and DellaCorte, C., 2001. The Effect of Journal Roughness and Foil
Coatings on the Performance of Heavily Loaded Foil Air Bearings. NASA TM
76
Chapter 7
Compressibility Number and Film

Thickness
While load capacity measurements are effective in demonstrating bearing per-
formance, they alone are not able to describe bearing operation and limits in a
fundamental way. In order to evaluate the gas film behavior, experimentally measured
6μωR2
bearing torque is used to estimate a compressibility number (Λc = Pa h2c
) and a
characteristic film thickness (hc ). In the context of hydrodynamic theory, these
parameters are more meaningful than a simple load capacity.
Both compressibility number and characteristic film thickness (Λc and hc , respec-
tively) are determined by assuming uniform Couette flow in an annular film that is
given by the inner and outer diameters of the bearing pads. This assumed flow field
permits the calculation of the film characteristics given an experimentally measured
torque. The derivation and significance of these parameters is detailed in Appendix
B. Unless otherwise noted, all data presented in this section is from bearing B running
against a PS304-coated thrust runner.
77
7.1 Compressibility Number
Translating experimentally measured torque data into a characteristic bearing
number is a way to compare laboratory test data to first principles analysis and
modeling results. This method of data reduction is useful for conveying the basic
physics of thrust bearing operation, and are shown to agree well with theory.
Figure 7.1 is a plot of experimentally measured combinations of load and charac-
teristic compressibility number. Data are taken at several speeds, and are shown
to demonstrate similar behavior. Although some bias is visually evident in the
various speed curves, speed appears to have little effect on the curves of load versus
compressibility number.
Figure 7.2 incorporates all of the load capacity data points shown in Figure 6.2,
but are plotted versus the experimental bearing number rather than the speed. It is
shown that cooling air has the effect of increasing the load capacity of the bearing at
a given compressibility number. Since the compressibility number calculated here is
proportional to the bearing torque squared, these data indicate that increased cooling
allows the bearing to operate at a higher “hydrodynamic efficiency.” In this context,
additional load supported by the bearing at a given Λ is an indication of increased
hydrodynamic efficiency.
Furthermore, it can be seen that at 0.52 kg/min cooling flow, the bearing is
able to operate at significantly higher bearing numbers than in the cases with lesser
cooling. While this allows increased performance, it appears to be more expensive in
terms of cycle efficiency (cooling flow is typically bled from a compressor stage within
78
120
25 krpm
30 krpm
35 krpm
100 39 krpm
42 krpm
45 krpm
48 krpm
80 55 krpm
Load (kPa)
60
40
20
0
0 20 40 60 80 100 120 140 160 180 200
/c
Figure 7.1: Load versus Compressibility Number at Various Speeds showing a

general curve of load versus characteristic compressibility number in a gas
foil thrust bearing with 0.52 kg/min of cooling air flow. The load behavior
for a given compressibility number appears to be somewhat insensitive to
speed such that data from multiple speeds can be incorporated in a single
curve.
79
120
Cooling Flow Rate
0 kg/min
0.17 kg/min
100 0.35 kg/min
0.52 kg/min
80
Load Capacity (kPa)
60
40
20
0
0 20 40 60 80 100 120 140 160 180 200
/c
Figure 7.2: Load Capacity Versus Compressibility Number comparing data

taken at different cooling flow rates. Increased cooling flow is shown to
increase achievable loads and compressibility numbers.
80
the turbomachine) to implement. It can be seen that the load capacity increases
with bearing number over the range shown, but that as bearing number continues to
increase, the curve flattens so that large increases in bearing number are needed to
support additional load. This flattening of the curve is a characteristic of the “high
speed limit” in gas bearings1 , which is described in more detail in Appendix B.
In order to more closely examine the high speed limit behavior of the bearings,
the average load pressure is multiplied by the apparent film thickness, a quantity
that for isothermal rigid step bearings with fixed contraction ratio and infinite width,
becomes constant with Λ for sufficiently large values of Λ.
Figure 7.3 gives a plot of this quantity versus bearing number for various speeds
with a cooling flow rate of 0.52 kg/min. Data is taken from low loads up to near the
load capacity at each speed in order to show the behavior throughout the operational
envelope. The largest flow rate is used for this test in order to more closely approxi-
mate an isothermal case than would be achievable with an uncooled test. These data
show a rapid increase in (P̄ hc ) with increasing bearing number followed by a levelling
off and slight decrease in (P̄ hc ) with further increases in Λc .
When torque data are taken over a wide range of speeds, loads, and cooling flow
rates, it can be plotted as in Figure 6.2, but showing the behavior throughout the
operating range rather than just near the load capacity. As shown in Figure 7.4, the
result is a smoother curve that includes data at lower bearing numbers, where typical
operation might be expected to occur. Note that with increasing cooling flow rate
the bearing supports more load for a given compressibility number, except for the
largest value of cooling flow. It is unknown why this trend reverses, but the effect
81
1.2
25 krpm
30 krpm
35 krpm
1.0 39 krpm
42 krpm
45 krpm
48 krpm
0.8 55 krpm
Phc (N/m)
0.6
0.4
0.2
0.0
0 20 40 60 80 100 120 140 160 180 200
/c
Figure 7.3: High Speed Limit Behavior of bearing B against a PS304-coated

runner with 0.52 kg/min cooling flow, which shows asymptotic behavior
of the quantity P̄ hc at large values of Λc , in accordance with the so-called
high-speed limit behavior in gas bearings
82
is relatively small and is suspected to be related to the relatively high flow rate and
flow patterns within the bearing chamber. A more interesting feature of this plot is,
as stated in reference to Figure 7.2, at 0.52 kg/min of cooling flow, the bearing is able
to run at higher Λc , which increases the load capacity.
120
Cooling Flow Rate
0 kg/min
0.17 kg/min
100
0.35 kg/min
0.52 kg/min
80
Load (kPa)
60
40
20
0
0 20 40 60 80 100 120 140 160 180 200
/c
Figure 7.4: Load Versus Compressibility Number for Various Cooling Flow
Rates including data from lightly loaded conditions to near load capacity,
over a speed range from 25-55 krpm. Most significant is the increased
compressibility numbers and loads achievable at the maximum cooling
flow rate.
7.2 Film Thickness
Gas film thickness (and in particular the minimum thickness) is typically used
in numerical modeling of hydrodynamic bearings to determine load capacity. This is
83
somewhat arbitrary since corroborative experimental gas film thickness measurements
are particularly difficult to make on account of the thinness of the film, high rotational
speeds, and flexible nature of the bearing geometry. To the author’s knowledge, the
only reported experimental measurement of gas film thickness in foil bearings is given
by Ruscitto et al.2 , whose work concerns foil journal bearings.
Without experimental measurement of limiting film thickness behavior, numerical
models have little hope of accurately predicting the load capacity of foil thrust
bearings in the near future. The effort and expense of making direct measurements of
film thickness can be avoided, however, by making use of the experimentally measured
bearing torque and simple analysis to determine a characteristic film thickness that
can aid the modeling effort.
In this section, a characteristic film thickness is examined by assuming Couette
flow in a uniform annular gap between the runner and bearing over a range of
conditions, particularly near the measured load capacity. While film thickness varies
over the pad, this characteristic value is an important parameter for the corroboration
of numerical models.
Figure 7.5 gives curves of experimentally determined characteristic film thickness
versus load at several different speeds, and with 0.52 kg/min of cooling air running
against a PS304 coated runner. It is shown that as load is increased, the film thick-
ness is decreased, as expected. Furthermore, at a specified load, the film thickness
generally increases with increased speed. These results are entirely consistent with
traditional hydrodynamic theory.
When film thickness behavior near load capacity is examined however, a somewhat
84
60
25 krpm
35 krpm
45 krpm
50
55 krpm
Film Thickness (micron)
40
30
20
10
0
0 20 40 60 80 100 120
Load (kPa)
Figure 7.5: Film Thickness Versus Load at Various Speeds, verifying that the
bearing operates with a larger characteristic film thickness as speed is
increased at conditions of constant load. When load is increased at a
constant speed, the characteristic film thickness decreases. These results
are in accordance with hydrodynamic theory
85
unexpected trend appears. Measurements of the maximum sustainable (steady)
torque on the bearing before load capacity is reached are used to determine the
characteristic thickness at which film rupture is imminent.
In conventional hydrodynamic analysis, the film’s capacity is said to be reached
when the minimum film thickness approaches (or is some fixed multiple of) the
bearing/runner surface roughness3 , regardless of shaft speed. At this point, inter-
mittent asperity contact occurs at high relative speeds and the heat dissipated in
these contacts can quickly lead to thermal runaway. Contrary to this reasoning,
Figure 7.6 shows the characteristic film thickness at maximum load to increase with
increased speed, and to decrease with increased cooling.
The non-constancy of minimum achievable (characteristic) film thickness suggests
that a simple asperity contact model may not be adequate to describe the physical
mechanism responsible for film rupture. Since the characteristic film thickness near
load capacity increases with increasing heat generation (increasing speed and load
capacity), the implication may be that thermoelastic distortion of the runner face
exceeds the magnitude of the surface roughness and therefore dominates the load
capacity behavior at high speeds. Since existing hydrodynamic models do not account
for structural heat transfer and the resulting thermoelastic effects, they do not predict
this type of effect.
To compare the effects of surface roughness on the characteristic film thickness,
data from tests of bearing B at 25 krpm with 0.52 kg/min is plotted as an apparent
film thickness versus load. This is shown in Figure 7.7, where it can be seen that
the film thickness behavior is similar for both runner surfaces, with the smoother
86
18
Cooling Flow Rate
0 kg/min
16
0.17 kg/min
0.35 kg/min
14 0.52 kg/min
0.52 kg/min (Chromium)
Film thickness (micron)
12
10
0
0 10 20 30 40 50 60
Speed (krpm)
Figure 7.6: Film Thickness near Load Capacity, demonstrating that both active
thermal management and smoother runner surfaces allow the bearing to
operate at smaller characteristic film thickness.
87
60
PS304
Chromium
50
Film thickness (micron)
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
Load (kPa)
Figure 7.7: Film Thickness Versus Load at 25 krpm, 0.52 kg/min, showing
the rougher as-ground PS304 surface to result in a slightly larger
characteristic film thickness at otherwise identical operating conditions.
chromium surface showing a slight bias toward a thinner film for the same operating
conditions.
It is noted that due to the rudimentary nature of the film thickness estimation,
inferences of film behavior should be carefully scrutinized. The use of experimentally
measured film thickness is described here to suggest potential modeling efforts and to
verify assumptions about the typical film thicknesses experienced in these bearings.
88
7.3 Relative Effects of Thermal Management and
Runner Surface Finish
It has been definitively shown that smoother runner surfaces and active thermal
management can increase the load capacity of foil thrust bearings. As a demonstration
of the relative magnitudes of these effects, a plot of bearing load versus compressibility
number is shown in Figure 7.8 comparing data from an uncooled test with a “rough”
runner to data from cooled tests with both smooth and rough runners.
The baseline data is taken at multiple speeds against a PS304-coated Inconel 718
runner with no active cooling. Compared to this, the data from the same runner with
0.52 kg/min of cooling flow demonstrates higher attainable compressibility numbers
and correspondingly large attainable loads. These data points are the same as given
in Figure 7.4. However, an additional series of data points has been added, those
taken from a 40 krpm load capacity test of a bearing against a smoother, chrome-
coated runner with 0.52 kg/min of cooling flow. These data show a further increase
in attainable load and compressibility number.
This plot shows definitively that the performance potential of a given thrust
foil bearing design is highly dependent on both thermal management and surface
condition. Furthermore, as shown in Figure 7.6, the limiting Couette film thickness
is also dependent on thermal management and surface finish, where lower attainable
film thicknesses correspond with the higher observed load capacities. Beyond the
treatment of thermal and surface effects in Chapters 5 and 6, characteristic film
thickness and compressibility number better describe the implications of the observed
89
140
0 kg/min (PS304)
0.52 kg/min (PS304)
120 0.52 kg/min (Chromium)
100
Load (kPa)
80
60
40
20
0
0 50 100 150 200 250 300
/c
Figure 7.8: Load Versus Compressibility Number Showing the Effects of

Cooling and Runner Surface Roughness, with a PS304-coated
runner and no cooling flow corresponding to the lowest achievable
loads and compressibility numbers. A bearing run against a PS304-
coated runner with 0.52 kg/min of cooling air is able to run at higher
compressibility numbers than achievable with any lesser amount of cooling
flow. Finally, the smooth chromium runner coating used in conjunction
with 0.52 kg/min cooling flow allows the bearing to achieve higher loads
and compressibility numbers than with the other tested configurations.
90
raw load capacity and torque data.
References
[1] Hamrock, B. J., 1994. Fundamentals of Fluid Film Lubrication. McGraw-Hill,
Inc., New York.
[2] Ruscitto, D., McCormick, J., and Gray, S., 1978. Hydrodynamic Air Lubricated
[3] Lu, X., and Khonsari, M., 2005. “On the Lift-Off Speed in Journal Bearings”.
Tribology Letters, 20, pp. 299–305.
91
Chapter 8
Physics of Thermal Management
The data obtained from tests of bearing load capacity are consistent with the theory
that thermoelastic distortions within the bearing/runner system are responsible for
performance loss. This decreased performance is most notable with increased speed
and load, which tend to increase bearing power loss.
Work by Dykas and Howard1 highlights the need for consideration of structural
temperature distributions in foil bearing systems. The current work includes dis-
cussion of thermoelastic effects in thrust bearings. While these theories have not
been fully tested, they are presented to the bearing community as considerations
that warrant further investigation, and are the subject of ongoing research. In
understanding thermal expansion effects on geometric distortion, bearing and system
design can be advanced and parasitic cooling requirements reduced.
92
8.1 Impacts of Runner Heat Transfer
One mechanism of thermal distortion is related to the transfer of heat through
the thrust runner. Some of the heat generated in the gas film conducts through
the thrust runner, with heat flow components in both axial (through-thickness) and
radial directions. The low thermal conductivity of many high strength alloys used to
fabricate the runners renders them vulnerable to larger temperature gradients needed
to conduct the heat through the thickness.
Higher temperatures in the bearing face of the runner results in thermal expansion
greater than that of the back (exposed) face, assuming a positive coefficient of thermal
expansion. The resulting thermoelastic stresses tend to bend the runner face away
from the operating bearing, leaving a convex running surface with respect to the
bearing.
Pan and Sternlicht2 investigated this phenomenon in spiral groove gas bearings as
a load capacity-limiting factor inherent to bearing operation. They conclude that for
spiral groove bearings, the realizable load capacity may be only half the theoretical
capacity. From a material properties standpoint, runner distortion is proportional
to the coefficient of thermal expansion, and inversely proportional to the thermal
conductivity.
The compliant nature of foil bearings makes them more tolerant of, but not im-
mune to, thermal and centrifugal distortion than their rigid counterparts. Tolerance
to predictable distortions can theoretically be tailored in the bearing by spatially
varying the support stiffness3 , but the bearing must be able to accommodate a wide
93
range of operating conditions where distortions may vary significantly.
For the bearing and runner geometry in this work, an analysis is performed
to determine the peak-to-peak magnitude of the surface deflection due to an axial
heat load, as reported by Dykas et al4 . A finite element model of the runner is
constructed from the three dimensional geometry. The analysis is axisymmetric, and
room temperature properties of Inconel 718 are used.
A uniform distributed heat input of 100 W is applied to the bearing surface of the
runner. The temperature of the back face of the runner annulus is set to zero, with
the hub surfaces insulated. The heat input is forced to conduct axially through the
1.27 cm thick annulus, resulting in a temperature difference between the bearing and
exposed surfaces of 17◦ C.
This temperature distribution is input to a structural simulation of the resulting
deformation. For this case of 100 W of axial heat flux, the amplitude of the convexity
is approximately 10 μm from the inner diameter of the bearing surface where it joins
the hub, to the outer diameter of the bearing surface (Figure 8.1). This deviation
from flatness is approximately the same magnitude as a characteristic film thickness,
and indicates the potential for this mechanism to alter bearing operation. It is notable
that in some failed bearings, wear scars near the inner radius of the bearing after load
capacity tests provide anecdotal corroboration of this particular mode of failure. An
example of this is shown in Figure 8.2.
Given that power loss at low to moderate loads has been measured to be a few
hundred watts or less over the speed range tested, the 100 W assumed axial flux
would appear to be a large proportion of generated heat. This number is somewhat
94
Figure 8.1: FEA Plot of Runner Axial Deflection due to Through-Thickness
Heat Transfer, showing that the runner takes on a convex shape with
respect to the bearing. This forces a larger portion of the load to be
supported on the inboard portion of the bearing.
95
Figure 8.2: Photograph of Wear Scar on Bearing Inner Radius after a load
capacity test. The presence of wear on the inner radius with little to
no wear on the outer portion of the bearing supports the convex runner
theory.
arbitrary, but is more appropriate for higher speeds and loads than shown in Figure
5.2, with associated larger total power loss.
Actual runner temperature distributions will certainly deviate from this simplified
case, with radial variation in the annulus and heat flux through additional surfaces,
however this analysis demonstrates the potential for modest axial heat flux in the
runner to significantly affect the flatness of the runner. Furthermore, large centrifugal
stresses may also contribute to a distorted runner face. Cumulative effects of thermal
and rotational distortions are not considered here, but are noted for their importance
in runner design.
For this thermoelastic phenomenon, cooling air forced through the bearing foils
should lessen the deformation of the runner by reducing the temperatures at the bear-
96
ing face and the axial temperature gradients through the runner thickness. Because
this convexity effect is qualitatively consistent with experimental observations of load
capacity and wear scars, it is presented as a potential performance-limiting mecha-
nism. Introducing cooling air to the bearing reduces thermal distortions, leading to
observed increases in load capacity.
8.2 Measurement of Bearing Temperature Gradi-

ents
As moderate through-thickness temperature gradients in the thrust runner can be
shown to result in substantial deformation of the bearing surface, radial temperature
gradients within the bearing and runner can also exist, with similar potential for
harmful thermoelastic distortion.
An investigation by Dykas and Radil5 measured radial temperature gradients at
the trailing edge of thrust pad over a range of speeds and loads when running against
an Inconel 718 runner with a PS304 coating. Measurements of the temperature
distributions are performed in the absence of a cooling air flow, which would flow
over the thermocouple junctions and degrade accuracy of the measurement. Here
the gradient is determined by the temperature difference across the inner and outer
thermocouples (8.3), though the temperature profile is close to linear across the the
three thermocouples on the bearing. These radial gradients increase with increasing
bearing power loss, as expected (Figure 8.4). Furthermore, it is clear from this
data that at a given value of power loss, radial temperature gradients decrease with
97
increasing speed.
Figure 8.3: Thermocouple Locations Near Top Foil Trailing Edge in bearing
A. The thermocouples are secured to the underlying bump foil where it
is in contact with the top foil, as described in Chapter 4.
These measured temperature gradients are not necessarily the largest within the
bearing, but the thermocouple locations are chosen based on accessibility to instru-
mentation. The temperature gradients measured in a foil thrust bearing are similar
in magnitude to those measured by Radil and Zeszotek6 in foil journal bearings.
While it is difficult to predict the thermoelastic stresses in bearing foils without
having a more detailed understanding of the overall temperature distribution, these
data show significant temperature gradients within the bearing, which at the very
least will need to be addressed in the hydrodynamic modeling of foil thrust bearings to
pinpoint the most performance-critical thermal effects in foil thrust bearing systems.
98
3.0
25 krpm
32 krpm
2.5 35 krpm
Radial Temperature Gradient ( C/mm)
45 krpm
55 krpm
65 krpm
o
2.0
1.5
1.0
0.5
0.0
0 50 100 150 200 250 300 350 400 450 500
Heat Generation (W)
Figure 8.4: Trailing Edge Temperature Gradients From 25-65 krpm for
bearing A, showing that the temperature gradient increases with the
bearing power loss at any given speed. It also shows that for a given
amount of bearing power loss (heat generation), the temperature gradient
decreases as the shaft speed increases.
99
8.3 Impact of Runner Convective Heat Transfer
on Bearing Temperature Gradients
The decrease in temperature gradients with speed for a given rate of heat gener-
ation (noted in Figure 8.4) may be attributed to increasing convective heat transfer
at the root of the runner annulus. Increasing speed results in a larger Reynolds
number at a given radial location, and a corresponding increase in Nusselt number,
and influences the heat transfer characteristics of the thrust runner.
The test runner has more complex geometry than a plain disk/cylinder, and ro-
tates within a test rig-specific cavity geometry. Furthermore, the high runner surface
speeds result in a large Reynolds number, compressible flow. These characteristics
make a detailed analysis of runner heat transfer difficult due to uncertainty in the
flow physics and convective heat transfer.
A simplified, first order analysis is presented here to address the type of study
needed to understand heat transfer within the runner and how this might be exploited
as a thermal management technique. Convective heat transfer on the back side
(exposed to air) is assumed to behave as a simple rotating disk exposed to an infinite
quiescent medium, and heat transfer correlations for that configuration are used.
Correlations for local heat transfer in a flow induced by a rotating disk are given in
terms of the local Reynolds number, Rer , in the form, N u = cRea , where c and a are
constants specific to the flow regime and geometry. Here a constant Prandtl number
has been assumed for air, allowing the Prandtl number dependence of the Nusselt
number to be accounted for in the constant c. The generally accepted value for the
100
Flow Regime Convection Correlation Applicable Region
0.5
Laminar N u = 0.33Re Rer < 1.95 × 105
Transition N u = 10 × 10−20 Re4 1.95 × 105 < Rer < 2.5 × 105
Turbulent N u = 0.0188Re0.8 Rer > 2.5 × 105
Table 8.1: Rotating Disk Convection Correlations
exponent a is 0.5 for laminar flow and 0.8 for turbulent flow, based on an analytical
solution with experimental verification. Transition from laminar to turbulent flow on
a rotating disk usually takes place at around Rer ≈ 2 × 105 .
For laminar flow below Retrans the heat transfer coefficient h is independent of
radial location on the disk. However, for turbulent flow h is radially dependent, and
is proportional to r3/5 . As a result of the heat transfer coefficient increases with radial
location on the disk face, there is less resistance to convective transfer of heat out of
the disk backside at the outer diameter.
Based on this transition Rer ≈ 2×105 , it would appear that under some conditions
inner portions of the runner backside might experience laminar flow, with a transition
to turbulence occurring at some radial location, outboard of which the heat transfer
coefficient increases with radius. Figure 8.5 shows a plot of calculated convection
coefficient on the runner backside versus radial location based on the heat transfer
correlations given by Popiel and Boguslawski7 , with a transition region from 1.95 ×
105 < Rer < 2.5 × 105 . A summary of the correlation behavior they report for air,
ignoring Grashof number effects, is given in Table 8.1.
The observed decrease in radial temperature gradient with increased speed at a
given rate of heat generation may be considered in the context of the convective heat
101
400
350
h Wm2 °C
300
250
200
150
0.025 0.03 0.035 0.04 0.045 0.05

Radius m
Figure 8.5: Estimated Convection Coefficient on Runner Backside at 35

krpm for a disk rotating in still air, showing a constant convection
coefficient for a laminar region near the inner radius. Outboard of this
the convection coefficient increases rapidly in a transitional region of the
flow, with a fully turbulent region on the outer radial portion of the disk.
102
transfer on the exposed backside of the runner. With increasing speed the inner
radial portion of the exposed runner face experience a transition to turbulent flow
and increasing heat transfer coefficient.
If the bearing power loss is scaled with respect to the runner convection coefficient,
a revised plot of the data in Figure 8.4 appears as Figure 8.6. Here the parameter
scaling the bearing power loss is Re4/5 , a measure of turbulent heat transfer.
3.0
25 krpm
32 krpm
35 krpm
2.5
45 krpm
46 krpm
55 krpm
o
2.0
65 krpm
1.5
1.0
0.5
0.0
0.0E+00 1.0E-03 2.0E-03 3.0E-03 4.0E-03 5.0E-03 6.0E-03 7.0E-03 8.0E-03
0.8
Heat Generation/Re (W)
Figure 8.6: Trailing Edge Temperature Gradients versus Scaled Power Loss
for bearing A, where the speed is scaled by Re0.8
R , and the various speed
curves lie roughly within a single curve.
Note that the temperature gradient curves for the range of speeds tested tend
to collapse into a single curve when plotted in this manner. This result allows
103
temperature gradients at the trailing edge to be estimated for conditions of heat
generation and speed that are not tested in this study. The degree to which this
behavior applies to other aspects of the temperature distribution within the foils
has not yet been determined. However, as further research into foil temperature
distributions is undertaken, it is anticipated that the behavior shown here will become
more clear.
Nonetheless, it is demonstrated by this plot of temperature gradients versus a Re0.8

R
scaled bearing power loss that the temperature distributions in the bearing can be
expressed over the range of speeds tested with a single curve. This is useful for simple
estimations of thermal behavior based on expected operating conditions, and further
study may demonstrate that the applicable speed range can be extended to lower
and higher speeds. In the case of a thermal failure mechanism that is temperature-
gradient dependent, this curve could predict the rate of heat generation corresponding
to the onset of failure for a given bearing speed.
8.4 Runner Material Effects
The positive influence of cooling air flow on the load capacity of foil thrust bearings
demonstrates a need for thermal management to achieve maximum performance. As
previously noted, active cooling with a bleed flow is undesirable, so passive methods
of thermal management which take advantage of the rotating hardware to induce flow
or sink heat are preferable. In order to investigate the potential use of runner heat
sinking, a runner assembly was fabricated with a high thermal conductivity annulus
104
as detailed in Section 4.1.3. The aluminum portion of the runner has an order of
magnitude larger thermal conductivity, resulting in less resistance to internal heat
conduction.
Figure 8.7 shows a plot of trailing edge temperature gradients versus a scaled
speed parameter as in Figure 8.6, but here the data from a baseline Inconel 718
runner is compared to that from the aluminum runner. It is clear that increased
thermal conductivity of the runner material results in smaller temperature gradients
for the same scaled speed parameter, but the difference is within a factor or two
rather than an order of magnitude. This may stem from the relative magnitudes of
conductive and convective resistances to heat transfer or from the degree of thermal
coupling between the gas film and runner.
This result indicates that the temperature gradients within the bearing foils can
be reduced by using a runner material with higher thermal conductivity, but the
achievable reduction in temperature gradients is modest compared to the ratio of
thermal conductivities. Furthermore, freedom in design and material selection is
very limited for thrust runners, as detailed in Appendix C. This effect is nonetheless
demonstrated to be significant and should be considered in product design.
105
3.0
25 krpm (aluminum)
32 krpm (aluminum)
2.5 40 krpm (aluminum)
48 krpm (aluminum)
25 krpm (inconel)
32 krpm (inconel)
o
2.0
35 krpm (inconel)
45 krpm (inconel)
46 krpm (inconel)
1.5 55 krpm (inconel)
65 krpm (inconel)
1.0
0.5
0.0
0.0E+00 1.0E-03 2.0E-03 3.0E-03 4.0E-03 5.0E-03 6.0E-03 7.0E-03 8.0E-03
0.8
Heat Generation/Re (W)
Figure 8.7: Runner Thermal Conductivity Effects on Temperature Distri-

butions in bearing A, where temperature gradients are plotted versus
the scaled power loss parameter for several speeds. Data obtained with a
standard Inconel 718 runner is compared to that of an aluminum runner,
showing the higher thermal conductivity of the 7075-T6 results in a
decrease in bearing temperature gradients.
References
[1] Dykas, B., and Howard, S. A., 2004. “Journal Design Considerations for
47(4), pp. 508–516.
[2] Pan, C., and Sternlicht, B., 1967. “Thermal Distortion of Spiral-Grooved Gas-
Lubricated Thrust Bearing Due to Self Heating”. ASME Journal of Lubrication
Technology, 89(2), pp. 197–202.
106
[5] Dykas, B., and Radil, K., 2005. “Experimental Measurement of Thrust
Foil Bearing Temperature Profiles”. In Proceedings of WTC2005, World
[6] Radil, K. C., and Zeszotek, M., 2004. “An Experimental Investigation into the
47(4), pp. 470–479.
[7] Popiel, C. O., and Boguslawski, L., 1974. “Local Heat-Transfer Coefficients on
the Rotating Disk in Still Air”. International Journal of Heat and Mass Transfer,
18, pp. 167–170.
107
Chapter 9
Summary and Conclusions
9.1 Summary of Results
In order to address the inadequacy of the body of knowledge concerning foil gas
thrust bearings, a comprehensive experimental study is undertaken. Measurements
of bearing torque and load capacity are made over a wide range of rotational speeds,
axial loads, cooling flow rates, and surface conditions. First principles analysis and
numerical modeling, where appropriate, are used to interpret and understand the
experimental results. Generally good agreement between experimental and analyti-
cal/numerical results is obtained.
The experimental measurements serve to characterize the bearing behavior through-
out the operational envelope, and highlight critical areas that further development
work should address in order to improve foil thrust bearing performance and adoption.
All of these factors have been previously suspected as able to enhance performance,
but the level of understanding remains largely anecdotal and limited to a small
number of foil bearing specialists. Furthermore, this study shows that many factors
108
influencing bearing performance (bearing design and foil coating, surface topography
and tribological characteristics, thermal management) have similar magnitudes and
must all be considered to achieve high load capacity.
Load capacity data demonstrate that foil coatings may be needed not only to
provide a low friction foil surface, but can also serve as abradable surfaces to allow
the bearing to correct for non-optimal geometry when deformed under hydrodynamic
pressure. Predictions of foil sag between bumps correlate well with observed foil
coating wear, and suggest that improved bearing design may be needed, and is being
enabled through this more fundamental understanding.
In addition to coatings on the top foil, two runner surface coatings are investigated.
A smooth chromium coating is shown to have significantly higher load capacity than
a somewhat rougher and higher friction as-ground PS304 surface with a molybdenum
disulfide overlay. This behavior is consistent with the reported results for foil journal
bearings, but this effect had not yet been verified for thrust bearings. In service,
PS304 coatings are expected to break in and provide a surface comparable to the TDC
coatings, but the as-ground condition is shown here to be an important consideration
for entry into service.
Cooling air flow is shown to enhance load capacity at high speeds, but is shown
to have little effect at the lowest speed tested. This behavior indicates an interplay
between hydrodynamic behavior and thermal behavior where the relative magnitudes
vary over the speed range of interest. While shown to increase the capacity of the
bearing, the use of bleed flows for thermal management in turbomachines incurs a
thermodynamic penalty and should be minimized.
109
Finally, preliminary studies of thrust bearing temperature distributions show that
large radial temperature gradients, present near the trailing edge of the pad, may
correlate with a thermal limit in the bearings. Increased thermal conductivity of the
runner is shown to decrease these temperature gradients significantly in the absence
of cooling air flow.
9.2 Conclusions
Two basic mechanisms are shown to limit thrust foil bearing performance. The
absence of sufficient thermal management and non-optimum geometry of the foil and
runner surfaces both reduce achievable capacity in foil thrust bearings, and further
study is warranted on these effects to enhance the operation of the bearings.
In the case of inadequate thermal management, the physical mechanism suspected
for reduced performance is thermoelastic distortion of the runner and foil surfaces.
This distortion causes a redistribution of the gas film pressure field, causing some
portions of the film to shoulder a higher portion of the axial load than intended.
Thinning of the film occurs in certain areas at lower loads than intended, resulting in
lower load capacity and potentially higher viscous friction.
The roughness of the foil and runner surfaces also represent non-optimal geometry
since current hydrodynamic modeling assumes perfectly smooth surfaces. Surface
roughness reduces the effective gas film thickness for a given mean separation distance,
and decreases the load at which surface asperities begin to contact.
Waviness of the top foil surface as a result of foil sag between bumps is also
110
classified as a non-optimal geometry effect. The wear of the foil coating where
the bump foil supports the top foil implies that the top foil becomes wavy under
hydrodynamic pressure and the top foil wear allows the deformed top foil to attain a
flatter surface with respect to the runner plane.
At a fundamental level, these factors all entail deviation from optimal geometry.
Hydrodynamic modeling which tends to idealize the surfaces, has not included these
effects. Since they have been shown to influence bearing operation, the implication
is that current modeling methods for foil thrust bearings cannot accurately predict
their limits with appropriate physical principles.
9.3 Implications and Future Work
The results of this study show that beyond the specific foil design of a given
bearing, a number of other factors can influence bearing performance. These factors,
which include surface condition and thermal/elastic distortion, can easily erode much
of a bearing’s potential performance capability. In a general sense, these can be
categorized as system-level design issues whereas the particular foil structure can be
considered a component-level design.
It is now clear that continuing development work on foil thrust bearings must
include these effects in order to properly evaluate the adequacy of the system design.
Further study will be needed on several fronts to improve the capability of thrust foil
bearings, particularly with respect to turbomachinery integration.
Current, ongoing research on foil and runner tribological coatings is addressing
111
the need for robust, high-temperature, abradable foil coatings and smooth, lubricious
runner coatings. It is expected that with improved understanding of these factors,
optimized thrust bearing designs are likely to result. The conclusions of this disserta-
tion emphasize the importance of coating research to address composition and surface
characteristics.
Increased attention will need to be paid to thermal management issues, par-
ticularly as the applied load and associated power loss are expected to increase.
Tacit awareness of the importance of thermal management exists in the foil bearing
community, but there remains a fundamental lack of understanding of the specific
thermal mechanism (e.g. thermomechanical, thermoelastic, heat transfer) that limits
capacity. Continued experimental work in this area is necessary to develop more
efficient thermal management techniques which may involve runner heat sinking and
tailored, flow-inducing geometries.
In addition to this experimental evaluation of thermal management techniques, it
will be necessary to consider thermal effects in numerical models. It is becoming ap-
parent that modeling which ignores heat transfer inadequately describes the behavior
of bearings at high rates of heat generation. Even simple structural heat transfer
models should improve the accuracy of numerical models and aid in the design of
efficient thermal management techniques.
Research in these key areas will complement the development effort being con-
ducted by bearing manufacturers and facilitate the design of integrated systems.
Some aspects of these research topics are currently being addressed in various studies,
but this dissertation demonstrates their relative importance and begins to quantify
112
attainable improvements. Furthermore, it provides a context for turbomachinery
designers to understand the importance of factors beyond bearing design in achieving
efficient, high performance oil-free rotor support systems.
113
Appendix A
Discussion of Error
Error and uncertainty are present in both the experimental measurements and
the numerical simulations presented in this text. In the case of numerical modeling,
accuracy is limited not by the numerical method employed, but by the physical model
input to the analysis. The assumptions and simplifications made in the numerical
simulations are treated in the body of this dissertation. This appendix addresses the
largest sources of experimental error present and the relative magnitudes.
Experimental Measurements
Four primary types of measurements are made in this research study - speed, load,
torque, and temperature. These quantities are all subject to error and uncertainty,
which in some case is difficult to quantify.
Speed measurements are made by a transistor-based fiber optic sensor, and the
digital output signal is fed to a frequency-to-voltage converter. The bias error in
this signal conditioning arrangement is small compared to the degree with which the
114
speed can be controlled during a test. Generally, the speed can be controlled within
±150 rpm, which is ±0.6% at the lowest speed tested.
Axial load measurements are made from a strain gauge-based load cell that is
placed in line with the test bearing shaft. Direct calibration of this load cell demon-
strates the ability to measure load to the greater of ±10 N or ±4%. However, when
cooling air is used, the elevated pressure in the bearing cavity leads to a hydrostatic
effect on the test bearing that acts to share the axial load imparted through the shaft.
A calibration of cavity pressure versus hydrostatic unloading is used to account for
this phenomenon, but this adds an addition uncertainty of perhaps ±10 N. Note that
this largest uncertainty applies at the highest cooling flow rate of 0.52 kg/min and is
less at smaller flow rates.
Uncertainty in the torque signal is perhaps the most difficult to quantify. The
LDVT load cell and signal conditioning arrangement used to measure torque has
far less inherent error than the physical torque being transmitted. The reactionary
torque of the test thrust bearing is transmitted through the shaft to the load cell,
but the linear rotary bearings supporting this shaft have a small breakaway torque
that may support some of the torque. Furthermore, the load is applied through a
ball contact at the end of the shaft, and the small contact area can also contribute
a small amount of error to the torque measurement, also through a static friction
effect. However, repeatability tests at similar conditions show little variation in the
measured torque at identical operating conditions. In a test where load is increased
slowly and constantly, followed by a slow and constant unloading of the bearing gives
a plot of torque that shows little hysteresis. This suggests the restraining torque of
115
the linear bearings and ball contacts is small. The tests designed to determine the
error in torque suggest an uncertainty of about ±3 Nmm or ±15%.
Temperature measurements taken from the Type K thermocouples are expected
to have bias error of up to ±2.2◦ C for the relatively low temperatures tested. This is
based on the voltage-to-temperature polynomial fit, and further error is introduced
due to precision (±0.3◦ C), and also small errors due to installation (cold junction
compensation) and signal noise. Since temperature gradients rather than absolute
temperature values are given, these error values are expected to be overly conserv-
ative. Due to close positioning of thermocouple junctions and the small differences
in measured temperature (up to roughly 20◦ C), the largest contribution to error in
the temperature gradient measurements should come from a lack of precision. A
reasonable estimate for the error in these measurements based on these assumptions
◦C
is ±0.12 mm .
Aside from errors present in the process signals, it is important to note that
the bearings tested are subject to manufacturing variability which may be large due
to the very low volume production of the particular designs tested. It is difficult
to quantify the degree of expected variability in the bearings because of a lack of
sufficient operational experience, but it is expected to be in the tens of percent.
For this reason, comparative measurements are made from the same bearing when
possible. Where data from different bearings are presented together, it is noted
in the text. While the issue of large suspected variation between test specimens
is significant, it is speculated that this will decrease with increase production vol-
ume.
116
Appendix B
Derivation of Characteristic Gas

Film Parameters
Compressible Reynolds Equation
Scaling analysis of the mass and momentum conservation equations in gas bearings
produces the compressible Reynolds equation, shown below in cylindrical coordinates
(starred notation denotes a dimensionless quantity):

1 ∂ ρ∗ h∗3 ∂P ∗ ∂ ρ∗ h∗3 ∂P ∗ ∂(ρ∗ h∗ )
+ ∗ r∗ ∗ =Λ
r∗ ∂θ μ∗ ∂θ ∂r μ ∂r∗ ∂θ
In this nondimensional form of the Reynolds equation, the terms on the left hand
side represent the divergence of the radial and azimuthal flow rates, and the term
on the right hand side is an expression of the variation of height and density in the
azimuthal direction, sometimes referred to as the advection mass transport term.
It is notable that the mass transport term on the right hand side of this equation
is scaled by the bearing compressibility number, Λ. As the compressibility number
becomes very large (as in a high speed and/or highly loaded bearing), the right hand
117
side term dominates the gas film behavior since terms on the left hand side are O(1).
The resulting form of the Reynolds equation is:
∂(ρ∗ h∗ )
Λ =0
∂θ
which implies that:
(ρ∗ h∗ ) = constant
In many analyses of gas film lubrication, the gas is assumed to be isothermal,
which simplifies the analysis by allowing the energy equation to be ignored. Further
assuming an ideal gas allows the substitution of pressure for density:
(P ∗ h∗ ) = constant
which is the result for isothermal ideal gas lubrication in the high speed limit.
Characteristic Film Thickness
In addition to calculation of bearing power loss and apparent friction coefficient,
speed/torque data can be used to characterize the nature of the gas film to corroborate
and guide numerical modeling.
The torque that is measured on the bearing is the integral of the quantity shear
stress multiplied by the radius, acting over the surface of the bearing:
2π R
To = (τw r) r dr dθ
0 ri
Substituting in the expression for wall shear stress:
2π R
∂u
To = rμ r dr dθ
0 ri ∂y
118
In order to solve this integral using experimentally measured speed and torque, the
flow field is assumed to be a Couette flow with uniform film thickness extending from
the inner bearing radius to the outer bearing radius, and extending a full 360 degrees
around the bearing:

2π R
ωr
To = rμ r dr dθ
0 ri hc
Integrating this expression to get hc in terms of experimentally measurable parameters
yields:
πμω (R4 − ri4 )

hc =
2To
Note that the Couette film thickness is a simple estimate of gas film behavior re-
quiring only geometric information, the viscosity of air, and experimentally measured
speed and torque be known in order to calculate a film thickness.
Characteristic Compressibility Number
The compressibility number, Λ, appears in the Reynolds equation scaling the
mass transport term and is a measure of the compressibility of the fluid relative to
the surface speed of the runner:
6μωR2
Λ=
Pa h2c
In order to determine the characteristic compressibility number of the bearing
given experimentally measurable quantities, the Couette film thickness derived above
119
is substituted into the expression for Λ to give:
24To2
Λc =
Pa π 2 μωR6 (1 − ξ 4 )2
This expression represents the experimentally measured characteristic compressibility
number of the bearing, which allows inferences to be made about the region in which
the gas film is operating.
120
Appendix C
Runner Design Limitations
Materials
Rotating hardware in gas turbine engines can be subjected to severe conditions
of stress and temperature which in many cases require that components be made
of advanced superalloy metals. Because of the large rotational stresses and high
temperatures, components must retain strength at high temperature, have good creep
resistance, and have at least moderate toughness.
There is some freedom in the design of the thrust runner since it will not be
subjected to stress and temperature levels as severe as those experienced by the
turbine blades, and there may be some choice of where the thrust bearings are located
within a turbomachine. This may permit the thrust bearings and runner to operate
at moderate temperatures (a few hundred degrees Celsius), which would provide an
expanded envelope of potential runner materials. SOA high pressure compressor
discharge temperatures can approach 650◦ C in modern aeroengines.
Of particular interest may be maximizing the thermal conductivity of the runner
121
to increase its heat sinking capabilities and to reduce temperature gradients which
cause structural distortion. A small coefficient of thermal expansion is also desirable
to minimize thermoelastic deformation. In terms of thermophysical properties, an
important material consideration is then the distortion parameter of the runner,
αQ
Π=
kLc
where:
k = thermal conductivity of runner
Lc = characteristic length
α = coefficient of thermal expansion
Q = heat input
Although limited freedom may exist to increase the value of this nondimensional
parameter, it is nonetheless a consideration in runner design. High strength at op-
erating temperature, toughness, and compatibility with tribological materials are all
essential considerations that may limit the ability to reduce the distortion parameter.
A characteristic value of this parameter using the baseline 1.27 cm thick Inconel 718
thrust runner with 100 W of heat input is Π = 8.4 × 10−3 .
Physical Dimensions
The physical size of the thrust runner is limited by mass, stress, rotordynamic,
and power loss considerations. In general, it is desirable to have the smallest runner
possible to limit mass, stress, and power loss. However, since bearing load capacity is
a function of both surface area and speed, the minimum required thrust runner size
122
is dictated by axial load requirements.
Turbomachines currently under consideration for oil-free operation are relatively
small and operate at speeds from roughly 50-100 krpm. These machines may require
thrust bearings that have load capacities in excess of a few hundred pounds, so
bearings with outer diameters of 10 cm or more are required. Larger bearings
would result in increased power loss, as well as larger mass, stress, and rotordynamic
complications.
In addition to the outer diameter of the thrust runner, the thickness is also a
critical design parameter for the same reasons as for the diameter. A runner must be
thick enough to resist bending under large axial loads, which for loads on the order
of a few hundred pounds is no thinner than about 5 mm. However, the disk must
be as thin as possible to reduce mass and inertia. As a result of these competing
requirements, there will be a rather narrow envelope of possible runner dimensions
for any given design.
Features
In order to enhance heat transfer out of the thrust runner, a multitude of features
can be included on various parts of the runner. Surface roughness, holes, vanes,
and fins in various configurations can all be used to enhance heat transfer from the
rotating disk. The design and placement of these features can be tailored to control
temperature distributions within the bearing and runner, based on knowledge of the
thermal phenomena that govern bearing limitations.
However, particular attention must be paid to any undesirable structural defor-
123
mation caused by these features, particularly when the features are not axisymmetric.
A variety of features have been explored in a finite element (FE) numerical model of
the runner to determine the resulting deformations at various operating conditions.
Features such as radial holes and vanes were seen to cause a rippling of the running
surface which is generally viewed as detrimental to the performance of the bearing.
While numerous notional designs have been examined in FE models to determine
structural response to various combinations of centrifugal, axial load, and thermal
conditions, these design-specific results are mostly beyond the scope of the current
work. Figure C.1 shows a few examples of designs that include flow-inducing geome-
tries that may be characteristic of passive thermal management features.
Figure C.1: Diagrams Showing Various Runner Design Features where the
runner on the left incorporates radial cooling air channels, the middle
runner is designed with a weakened backside to counteract thermally-
induced convexity, and the right runner incorporates a centrifugal
pumping design.
124
It is noteworthy that the addition of non-axisymmetric features has the potential
to affect the flatness of the runner face at high rotational speeds and corresponding
centrifugal stresses. An example of this is shown in Figure C.2, where radially extend-
ing cooling holes in the runner are spaced periodically around the circumference and
extending to the center of the runner hub. The face shows a waviness caused by these
radial channels, with an amplitude of approximately 2.5 μm. This example highlights
the potentially deleterious effect of non-axisymmetric geometry, and demonstrates
another limitation on runner design freedom.
Figure C.2: FE Model of Runner Face Axial Displacement at 60 krpm,

showing the effect of radial cooling holes on the flatness of the runner face.
Peak-to-peak amplitude of the circumferential waviness is approximately
2.5 μm.
125
Appendix D
Raw Experimental Data
This appendix contains much of the raw experimental torque and power loss data
presented in Chapters 5-7 for bearing B running against an as-ground PS304-coated
Inconel 718 thrust runner. Measured bearing torque, power loss, and characteristic
compressibility number and film thickness are given for speeds from 25-55 krpm,
cooling flow rates up to 0.52 kg/min, and loads up to the bearing load capacity.
Speed Cooling Flow Load Torque Power Loss Λc hc

(krpm) (kg/min) (N) (N mm) (W) (µm)
25 0.00 44 20 53 14 22.8
25 0.00 89 26 68 23 17.9
25 0.00 133 32 83 34 14.7
25 0.00 178 38 101 50 12.1
25 0.00 222 44 115 65 10.5
25 0.00 267 49 127 79 9.6
25 0.00 289 52 136 91 8.9
25 0.00 311 55 145 103 8.4
30 0.00 44 19 59 9.7 29.9
30 0.00 89 24 75 16 23.5
30 0.00 133 31 96 26 18.3
30 0.00 178 37 117 39 14.9
30 0.00 222 43 135 52 13.0
30 0.00 267 50 156 69 11.2
30 0.00 289 52 163 76 10.7
126
35 0.00 44 19 70 8.8 33.8
35 0.00 89 25 91 15 26.1
35 0.00 133 31 112 22 21.3
35 0.00 178 37 137 33 17.4
35 0.00 222 44 161 47 14.7
35 0.00 44 24 87 14 27.4
35 0.00 89 29 108 21 22.1
35 0.00 133 35 128 29 18.6
35 0.00 178 41 149 40 16.0
35 0.00 222 49 178 57 13.4
35 0.00 245 52 190 65 12.5
35 0.00 267 55 203 74 11.7
40 0.00 44 26 109 14 28.6
40 0.00 89 33 137 23 22.7
40 0.00 133 41 170 35 18.3
40 0.00 178 47 199 47 15.7
40 0.00 222 53 222 59 14.0
40 0.00 245 58 241 70 12.9
40 0.00 267 61 256 78 12.2
40 0.00 289 66 274 90 11.3
45 0.00 44 24 112 11 35.2
45 0.00 89 31 144 17 27.4
45 0.00 111 34 160 21 24.7
45 0.00 133 36 170 24 23.1
45 0.00 156 42 197 33 20.0
45 0.00 178 44 208 36 19.0
45 0.00 200 49 229 44 17.2
45 0.00 222 52 245 50 16.1
45 0.00 245 57 266 60 14.8
45 0.00 267 61 287 69 13.7
45 0.00 289 68 319 86 12.3
50 0.00 44 26 136 11 35.7
50 0.00 89 33 172 18 28.3
50 0.00 111 37 195 23 24.9
50 0.00 133 41 213 28 22.8
50 0.00 156 43 225 31 21.6
50 0.00 178 46 243 36 20.0
50 0.00 200 51 266 43 18.3
50 0.00 222 55 290 51 16.8
50 0.00 245 61 319 63 15.2
127
50 0.00 267 66 343 72 14.2
50 0.00 289 71 373 85 13.0
55 0.00 44 29 169 13 34.8
55 0.00 89 36 208 20 18.2
55 0.00 133 43 247 28 23.8
55 0.00 178 50 286 38 20.5
55 0.00 222 59 338 53 17.4
55 0.00 245 64 371 63 15.9
25 0.17 40 12 33 5.2 37.3
25 0.17 111 21 56 15 21.6
25 0.17 156 26 68 23 17.9
25 0.17 178 29 77 29 15.8
25 0.17 200 33 86 36 14.2
25 0.17 222 36 95 44 12.8
25 0.17 245 40 104 53 11.7
25 0.17 267 45 118 69 10.3
25 0.17 289 50 130 83 9.3
35 0.17 44 18 66 7.8 35.9
35 0.17 89 24 87 14 27.4
35 0.17 133 31 112 22 21.3
35 0.17 178 36 133 31 18.0
35 0.17 222 44 161 47 14.7
35 0.17 267 51 186 62 12.8
35 0.17 289 55 203 74 11.7
35 0.17 311 60 219 86 10.9
35 0.17 333 66 240 103 9.9
45 0.17 44 19 91 6.9 43.5
45 0.17 89 26 122 13 32.2
45 0.17 133 32 149 19 26.4
45 0.17 178 38 181 28 21.8
45 0.17 222 45 213 38 18.5
45 0.17 266 53 250 53 15.7
45 0.17 288 58 272 62 14.5
45 0.17 311 62 293 72 13.4
45 0.17 333 67 314 83 12.5
55 0.17 44 24 137 8.6 43.0
55 0.17 89 31 176 14 33.5
55 0.17 133 37 215 21 27.4
55 0.17 178 44 254 30 23.2
55 0.17 222 52 299 41 19.6
128
55 0.17 266 61 351 57 16.7
25 0.35 38 11 30 4.3 41.1
25 0.35 82 17 44 9.6 27.4
25 0.35 126 23 59 17 20.5
25 0.35 170 28 74 27 16.4
25 0.35 215 35 92 41 13.3
25 0.35 259 42 109 59 11.1
25 0.35 281 47 123 74 9.9
25 0.35 303 50 132 85 9.2
35 0.35 38 13 48 4.0 50.0
35 0.35 82 17 62 6.9 38.3
35 0.35 126 23 83 12 28.8
35 0.35 170 28 104 19 23.0
35 0.35 215 35 128 29 18.6
35 0.35 258 42 153 42 15.5
35 0.35 303 47 174 54 13.7
35 0.35 325 52 190 65 12.5
35 0.35 347 55 203 74 11.7
35 0.35 369 61 224 89 10.7
35 0.35 391 67 244 107 9.7
45 0.35 39 18 85 6.1 46.2
45 0.35 83 23 106 9.5 37.0
45 0.35 127 28 133 15 29.6
45 0.35 171 34 160 21 24.7
45 0.35 215 41 192 31 20.5
45 0.35 259 47 224 42 17.6
45 0.35 304 55 261 57 15.1
45 0.35 347 63 298 75 13.2
45 0.35 370 68 319 86 12.3
55 0.35 39 21 124 7.0 47.6
55 0.35 83 27 156 11 37.7
55 0.35 127 33 189 16 31.2
55 0.35 171 41 234 25 25.1
55 0.35 215 47 273 34 21.5
55 0.35 259 55 319 47 18.4
55 0.35 304 64 368 62 16.0
25 0.52 22 9 24 2.7 51.4
25 0.52 59 14 35 6.2 34.2
25 0.52 101 19 50 12 24.2
25 0.52 143 24 62 19 19.6
129
25 0.52 187 29 77 29 15.8
25 0.52 229 36 95 44 12.8
25 0.52 274 44 115 65 10.5
25 0.52 297 49 127 79 9.6
25 0.52 319 53 139 95 8.7
30 0.52 41 11 35 3.6 49.3
30 0.52 120 23 71 14 24.7
30 0.52 187 32 99 28 17.6
30 0.52 231 38 121 41 14.5
30 0.52 276 46 146 60 12.0
30 0.52 320 55 174 86 10.1
30 0.52 364 64 202 116 8.6
30 0.52 385 69 217 133 8.1
30 0.52 406 77 241 165 7.3
30 0.52 430 81 256 185 6.8
35 0.52 38 17 62 6.9 38.3
35 0.52 78 21 79 11 30.3
35 0.52 120 26 95 16 25.0
35 0.52 162 32 116 24 20.5
35 0.52 207 38 141 35 16.9
35 0.52 250 44 161 47 14.7
35 0.52 294 52 190 65 12.5
35 0.52 339 60 219 86 10.9
35 0.52 383 69 253 114 9.4
35 0.52 411 75 273 133 8.7
35 0.52 439 81 298 159 8.0
39 0.52 27 14 55 4.0 53.4
39 0.52 84 21 88 9.9 33.7
39 0.52 165 33 134 23 22.1
39 0.52 229 43 175 40 16.9
39 0.52 273 50 203 53 14.6
39 0.52 316 58 235 71 12.6
39 0.52 360 64 263 89 11.2
39 0.52 401 76 309 123 9.6
39 0.52 426 81 332 142 8.9
42 0.52 22 17 75 5.7 46.0
42 0.52 78 25 109 12 31.4
42 0.52 163 37 164 28 20.9
42 0.52 229 47 209 45 16.4
42 0.52 272 54 239 59 14.4
130
42 0.52 315 62 273 77 12.5
42 0.52 357 69 303 95 11.3
42 0.52 401 77 338 118 10.2
42 0.52 422 84 368 140 9.3
45 0.52 34 19 91 6.9 43.5
45 0.52 125 32 149 19 26.4
45 0.52 190 41 192 31 20.5
45 0.52 232 47 224 42 17.6
45 0.52 276 54 256 55 15.4
45 0.52 319 61 287 69 13.7
45 0.52 364 68 319 86 12.3
45 0.52 390 75 351 104 11.2
45 0.52 57 26 122 13 32.2
45 0.52 143 38 181 28 21.8
45 0.52 185 44 208 36 19.0
45 0.52 228 51 240 48 16.4
45 0.52 272 58 272 62 14.5
45 0.52 314 64 303 77 13.0
45 0.52 357 73 346 101 11.4
48 0.52 28 20 102 7.2 43.8
48 0.52 100 31 153 16 29.2
48 0.52 161 37 187 24 23.9
48 0.52 226 47 239 39 18.8
48 0.52 271 54 273 51 16.4
48 0.52 314 62 312 68 14.3
48 0.52 358 71 358 89 12.5
48 0.52 381 76 380 100 11.8
48 0.52 403 80 403 113 11.1
48 0.52 422 84 420 122 10.7
55 0.52 58 29 169 13 34.8
55 0.52 143 42 241 27 24.4
55 0.52 184 47 273 34 21.5
55 0.52 227 57 325 49 18.1
55 0.52 41 25 143 9.4 41.1
55 0.52 103 33 189 16 31.2
55 0.52 145 40 228 24 25.8
55 0.52 187 47 273 34 21.5
55 0.52 229 55 319 47 18.4
55 0.52 273 63 364 61 16.1
131
Bibliography
Agrawal, G., 1984. Hydrodynamic Fluid Film Thrust Bearing. U.S. Patent 4,462,700.
Agrawal, G., 1990. “Foil Gas Bearings for Turbomachinery”. In 20th Intersociety
Conference on Environmental Systems, no. SAE Paper 901236, Society of
Automotive Engineers.
Agrawal, G. L., 1997. “Foil Air/Gas Bearing Technology An Overview”. In

International Gas Turbine and Aerospace Congress and Exhibition, no. ASME paper
97-GT-347, American Society of Mechanical Engineers.
Anderson, W. J., and Bisson, E. E., 1964. Advanced Bearing Technology. NASA-
SP 38, National Aeronautics and Space Administration, Washington, D.C.
Bauman, S., 2005. An Oil-Free Thrust Foil Bearing Facility Design, Calibration,
Blok, H., and vanRossum, J., 1953. “The Foil Bearing - A New Departure in
Hydrodynamic Lubrication”. Lubrication Engineering, 9(6), pp. 316–320.
Bruckner, R. J., 2004. A Propulsion System Analysis of Oil Free Turbomachinery

for Aviation Turbofan Engines. AIAA 2004-4189, American Institute of Aeronautics
and Astronautics, Reston, VA.
Bruckner, R. J., 2004. “Simulation and Modeling of the Hydrodynamic, Thermal,

Capstone MicroTurbine Corporation, 2000. Capstone Turbine Ships New

Microturbine Product; New Capstone 60 Power System Generates Twice the
Electricity of Current Models. Press Release.
Davis, J., ed., 1998. Metals Handbook Desk Edition. ASM International.
132
DellaCorte, C., 1998. The Evaluation of a Modified Chrome Oxide Based High
Temperature Solid Lubricant Coating for Foil Gas Bearings. NASA TM 1998-
DellaCorte, C., and Edmonds, B. J., 1995. Preliminary Evaluation of PS300: A New
Self-Lubricating High Temperature Composite Coating for Use to 800C. NASA TM
DellaCorte, C., and Edmonds, B. J., 1999. Self-Lubricating Composite Containing

Chromium Oxide. U.S. Patent 5,866,518.
DellaCorte, C., Lukaszewicz, V., Valco, M. J., Radil, K., and Heshmat, H., 2000.
DellaCorte, C., and Pinkus, O., 2000. Tribological Limitations in Gas Turbine
Engines: A Workshop to Identify the Challenges and Set Future Directions. NASA
TM 2000-210059, National Aeronautics and Space Administration, Cleveland, OH.
DellaCorte, C., and Valco, M. J., 2000. “Load Capacity Estimation of Foil
DellaCorte, C., Zaldana, A., and Radil, K. C., 2002. A Systems Approach to the
Dykas, B., DellaCorte, C., Prahl, J., and Bruckner, R., 2006. “Thermal Management
Phenomena in Foil Gas Thrust Bearings”. In Proceedings of Turbo Expo 2006:
Power for Land, Sea, and Air, no. 2006-91268, American Society of Mechanical
Engineers.
Dykas, B., and Howard, S. A., 2004. “Journal Design Considerations for
47(4), pp. 508–516.
Dykas, B., and Radil, K., 2005. “Experimental Measurement of Thrust Foil Bearing
Temperature Profiles”. In Proceedings of WTC2005, World Tribology Congress
III, no. WTC2005-63564, American Society of Mechanical Engineers/Society of
Tribologists and Lubrication Engineers.
Etsion, I., 1977. “A Cantilever Mounted Resilient Pad Gas Thrust Bearing”. Journal
of Lubrication Technology, 99, pp. 95–100.
133
Fortmann, W. E., 1978. Dual Wedge Fluid Thrust Bearing Including Wave Spring.
U.S. Patent 4,082,375.
Gray, S., and Heshmat, H., 1981. Support Element for Compliant Hydrodynamic
Thrust Bearing. U.S. Patent 4,277,111.
Hamrock, B. J., 1994. Fundamentals of Fluid Film Lubrication. McGraw-Hill, Inc.,

New York.
Heshmat, C., Heshmat, H., Valco, M. J., Radil, K., and DellaCorte, C., 2005. “Foil
Bearings Makes Oil-Free Turbocharger Possible”. In Proceedings of WTC2005,
World Tribology Congress III, no. WTC2005-63724, American Society of Mechanical
Heshmat, H., 1994. “Advancements in the Performance of Aerodynamic Foil Journal

Bearings: High Speed and Load Capacity”. ASME Journal of Tribology, 116,
pp. 287–295.
Heshmat, H., 1999. High Load Capacity Compliant Foil Hydrodynamic Thrust
Bearing. U.S. Patent 5,961,217.
Heshmat, H., 2005. “Major Breakthrough in Load Capacity, Speed and Operating
Temperature of Foil Thrust Bearings”. In Proceedings of WTC2005, World
Heshmat, H., Hryniewicz, P., Walton, J., Willis, J., Jahanmir, S., and DellaCorte,
C., 2005. “Low Friction Wear-Resistant Coatings for High-Temperature Foil
Bearings”. Tribology International, 38.
Heshmat, H., Shapiro, W., and Gray, S., 1981. “Development of Foil Journal
Bearings for High Load Capacity and High Speed Whirl Stability”. In Proceedings
of the ASLE-ASME Joint Tribology Conference, no. 81-Lub-36, American Society
of Lubrication Engineers/American Society of Mechanical Engineers.
Heshmat, H., Walowit, J. A., and Pinkus, O., 1983. “Analysis of Gas Lubricated
Compliant Thrust Bearings”. ASME Journal of Lubrication Technology, 105,
pp. 638–646.
Howard, S. A., and DellaCorte, C., 2006. Gas Foil Bearings for Space Propulsion
Nuclear Electric Power Generation. NASA-TM 2006-214115, National Aeronautics
and Space Administration, Cleveland, OH.
134
Hryniewicz, P., Locke, D., and Heshmat, H., 2003. “New-Generation Development
Rigs for Testing High-Speed, Air-Lubricated Thrust Bearings”. Tribology
Transactions, 46(4), pp. 556–559.
Kim, J. H., Blanchet, T., and Calabrese, S., 2004. “High Velocity Oxyfuel Deposition
for Low Surface Roughness PS304 Self-Lubricating Composite Coatings”. Tribology
Transactions, 47(1), pp. 157–169.
Lu, X., and Khonsari, M., 2005. “On the Lift-Off Speed in Journal Bearings”.
Tribology Letters, 20, pp. 299–305.
Lucero, J., 2006. Measured Foil Frequencies and Mode Shapes as a Function of Wear
for a Gas Foil Thrust Bearing. NASA-TM (to be published), National Aeronautics
and Space Administration, Cleveland, OH.
Mason, L. S., 1999. Surface Nuclear Power for Human Mars Missions. NASA TM
Mohawk Innovative Technology, Inc., 2002. WJ24-8 Turbojet Engine Demonstration

Testing with Air Foil Bearing. Press Release, December.
Mohawk Innovative Technology, Inc., 2005. High-Speed, Oil-Free, Motorized

Spindle. Press Release, January.
Nemeth, Z. N., 1977. Experimental Evaluation of Foil-Supported Resilient-Pad

Gas-Lubricated Thrust Bearing. NASA TP 1030, National Aeronautics and Space
Nemeth, Z. N., 1979. Operating Characteristics of a Cantilever-Mounted Resilient-

Pad Gas-Lubricated Thrust Bearing. NASA TP 1438, National Aeronautics and
Space Administration, Cleveland, OH.
Pan, C., and Sternlicht, B., 1967. “Thermal Distortion of Spiral-Grooved Gas-
Lubricated Thrust Bearing Due to Self Heating”. ASME Journal of Lubrication
Technology, 89(2), pp. 197–202.
Popiel, C. O., and Boguslawski, L., 1974. “Local Heat-Transfer Coefficients on the
Rotating Disk in Still Air”. International Journal of Heat and Mass Transfer, 18,
pp. 167–170.
Radil, K. C., and DellaCorte, C., 2001. The Effect of Journal Roughness and Foil
Coatings on the Performance of Heavily Loaded Foil Air Bearings. NASA TM
135
Radil, K. C., and Zeszotek, M., 2004. “An Experimental Investigation into the
47(4), pp. 470–479.
Ruscitto, D., McCormick, J., and Gray, S., 1978. Hydrodynamic Air Lubricated
Salehi, M., Swanson, E., and Heshmat, H., 2000. “Thermal Features of Compliant
Foil Bearings - Theory and Experiments”. In Joint Tribology Conference, no. 2000-
TRIB-38, American Society of Mechanical Engineers/Society of Tribologists and
Lubrication Engineers.
Suriano, F. J., 1981. Gas Foil Development Program; Final Report. AFWAL-TR
81-2095, Air Force Wright Aeronautical Laboratories, Dayton, Ohio.
136

Case 1144261305

Uploaded by

Copyright:

Available Formats

Case 1144261305

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Case 1144261305

Uploaded by

Copyright:

Available Formats

FACTORS INFLUENCING THE PERFORMANCE OF FOIL GAS THRUST

BEARINGS FOR OIL-FREE TURBOMACHINERY APPLICATIONS

BRIAN DAVID DYKAS

Submitted in partial fulﬁllment of the requirements

Dissertation Advisor: Dr. Joseph M. Prahl

Department of Mechanical and Aerospace Engineering

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

candidate for the Ph.D. degree *.

March 23, 2006

Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

2 Background of the Art 11

3 Gas Film Characteristics 20

5 Bearing Torque and Power Loss Results 48

6 Load Capacity Results 61

7 Compressibility Number and Film Thickness 77

8 Physics of Thermal Management 92

9 Summary and Conclusions 108

B Derivation of Characteristic Gas Film Parameters 117

C Runner Design Limitations 121

D Raw Experimental Data 126

4.1 Test Bearing Parameters . . . . . . . . . . . . . . . . . . . . . . . 34

8.1 Rotating Disk Convection Correlations . . . . . . . . . . . . . . 101

1.1 Photographs of Modern Foil Bearings . . . . . . . . . . . . . . . 2

2.1 Examples of Rigid Thrust Bearing Geometries . . . . . . . . . 12

3.1 Exaggerated Cross Sectional View of Gas Film Control Volume 22

5.1 Torque Versus Load From 25-55 krpm . . . . . . . . . . . . . . 49

6.1 Thrust Bearing Load Capacity as a Function of Speed . . . . 63

7.1 Load versus Compressibility Number at Various Speeds . . . 79

8.1 FEA Plot of Runner Axial Deﬂection due to Through-Thickness

h Gas ﬁlm thickness, m

hc Characteristic (Couette) ﬁlm thickness, m

Nu Local Nusselt number, hr

Retrans Rotational Reynolds number at transition to turbulence

y Film thickness coordinate, m

μ∗ Nondimensional viscosity, μμa

ξ Bearing radius ratio, rRi

ρ∗ Nondimensional gas density, ρ

ρa Inlet gas density, kg

τw Wall shear stress, P a

BRIAN DAVID DYKAS

performance. Measurements of bearing power loss and load capacity made in a

variety of conﬁgurations highlight several important factors which inﬂuence perfor-

mance. Consistent with conventional hydrodynamic theory, surface condition of the

management via conduction through the runner have a large inﬂuence.

surface conditions on the load capacity of foil thrust bearings.

by the foil bearing community.

Foil gas thrust bearings are self-acting, compliant-surface hydrodynamic bearings

hydrodynamic performance, but remain largely unvalidated. A major obstacle to

bearing operation, particularly at high loads and speeds.

1.1 Oil-Free Turbomachinery Applications

The National Aeronautics and Space Administration currently supports a pro-

gram to research advanced Oil-Free Turbomachinery technology in order to realize

This program seeks to replace conventional oil-lubricated rolling element bearings

mean time between failures (MTBF) in excess of 100,000 hours2 .

(a) Journal Bearing (b) Thrust Bearing

Figure 1.1: Photographs of Modern Foil Bearings

operating costs7 . A recent propulsion system study on a notional 50 passenger regional

in block fuel can be achieved through oil-free engines8 .