Design of Three-Pad Hybrid Air

Foil Bearing and Experimental

Investigation on Static
Performance at Zero Running
Daejong Kim
Speed
Donghyun Lee Air foil bearings (AFBs) have been explored for various micro- to midsized turbomachin-
ery for decades, and many successful applications of the AFBs to small turbomachinery
Department of Mechanical and Aerospace were also reported. As machine size increases, however, one of the critical technical
Engineering, challenges of AFBs is a wear on the top foil and rotor during starts/stops due to relatively
University of Texas at Arlington, heavy rotor weight compared with the size of the bearing. The wear on the foil increases
500 West 1st Street, Woolf Hall, with greater loading during starts/stops as a function of the coating performance. The
Arlington, TX 76019 hybrid air foil bearing (HAFB), which combines hydrodynamic pressure with hydrostatic
lift, can help to minimize/eliminate the wear problem during the start/stops. This paper
reports design and preliminary test results of hydrodynamically preloaded three-pad
HAFB aimed for midsized airborne turbomachinery applications. Designed HAFB was
manufactured and comprehensive parametric design simulations were performed using
time-domain orbit simulations and frequency-domain linear perturbation analyses to
predict performances of manufactured bearing. Static stiffness was measured at zero
running speed to investigate the load capacity of hydrostatic operation when rotor is at
stationary. The measured static stiffness showed good agreement with predictions.
关DOI: 10.1115/1.4001066兴

Keywords: hybrid air foil bearing, gas lubrication, orbit simulation, linear analysis

1 Introduction gas bearings were reported since then; Curwen et al. 关10兴 reported
tilting pad gas bearing with hydrostatic orifice formed on the
Air/gas foil bearings 共AFBs兲 have shown a promise in high-
spherical pivot for Brayton cycle turbo generator. Han et al. 关11兴
speed micro- to midsized turbomachinery. Compared with the
investigated dynamic performance of cylindrical hydrostatic air
rolling element bearings, AFBs do not require oil lubrication cir-
bearing with multiple inherent restrictors along the circumferen-
cuits, seals, and oil cooling system, allowing the system to be less
tial and axial directions for precision machine tool applications.
complicated and more environment friendly.
More recently, hydrostatic operation was applied to three-lobed
In small machines, the rotor weight does not impose significant
gas bearing 关12兴 and flexure pivot tilting pad gas bearing 关13,14兴.
load to the AFB during start/stops, and relevant dry rubbing is not The concept of the hydrostatic lift can be found in a patent on foil
a critical issue compared with the rotor-bearing instability at high thrust bearing 关15兴. However, in the patent 关15兴, compressor bleed
speeds. However, as machine size increases, the rotor weight in- air is discharged to the backside of the top foil 共i.e., space taken
creases in proportion to the third power of rotor characteristic by bump foils兲 and then guided to the air film through multiple
dimension, while the AFB’s load capacity increases in proportion holes formed on the top foil. The concept is very questionable in
to the second power of rotor characteristic dimension. Therefore, terms of effectiveness because the pressure on the backside of the
hydrodynamic AFBs have a definitive load capacity limit as the top foil is higher than hydrostatic pressure generated in the air
machine size increases. Furthermore, the hydrodynamic AFBs film, resulting in no hydrostatic lift. Application of hydrostatic
with adequate load capacity at machine operating speed has to bearing concept can be also found in a sheet metal forming and
rely on boundary lubrication with surface coatings during start/ tape recorders 关16兴, where the sheet metal or tape under tension is
stops, yielding inevitable surface wear and limited reliability. called a foil.
Much progress was reported on the surface coatings on the AFBs Active magnetic bearings also enable rubbing-free start/stops of
关1–4兴, and successful application of one of these coatings to small the rotor, and the technology has been developed since the early
power generation turbines is reported in Ref. 关5兴. However, in 1990s as a magnetic-foil hybrid 关17–19兴 or just magnetic bearing
midsized turbomachinery where journal diameter is bigger than 关20兴. However, active magnetic bearings require complicated con-
100 mm 共⬃4 in.兲, the reliability issue associated with dry rubbing trollers, and full active control of a rotor with complicated dy-
during start/stop seems unavoidable as presented in Ref. 关6兴. namic motions is a significant challenge and real field applications
Hydrostatic gas bearings with cylindrical solid wall began to are still very limited.
appear for military and space applications in the 1960s 关7–9兴 to Kim and Park 关21兴 presented the first exploratory experimental
avoid the dry rubbing during start/stop. Other forms of hydrostatic work on hydrostatic air foil bearing 共HAFB兲, combining hydrody-
namic action with a hydrostatic lift through inherent restrictor
formed on the top foil. Kumar and Kim 关22兴 presented a compu-
Contributed by the IC Engine Division of ASME for publication in the JOURNAL OF
ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received September 11, 2009;
tational model to predict stiffness and damping coefficients of the
final manuscript received December 7, 2009; published online August 27, 2010. HAFB. The study shows that HAFB has much smaller cross-
Assoc. Editor: Patrick S. Keogh. coupled stiffness than hydrodynamic AFB in general, and these

Journal of Engineering for Gas Turbines and Power DECEMBER 2010, Vol. 132 / 122504-1
Copyright © 2010 by ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 RP
RPad
RRotor

Rear duplex Magnetic Housing for HAFB

50 hp motor Coupling
bearing radial bearing
CSB
Fig. 1 Section view of the test rig with high-speed motor
„30,000 rpm… Fig. 2 Schematic description of HAFB for AFRL rotor, drawn
not in scale

coefficients are strong function of orifice sizes, supply pressure,

etc. Kumar and Kim 关23兴 further developed the concept of the lations for stability analyses, and perturbation analysis for calcu-
HAFB by applying the principle of HAFB to bump foil bearings. lation of bearing coefficients and estimation of rotor-bearing natu-
They presented experimental work on load capacity of second ral frequencies.
generation HAFB with bump foils with higher support stiffness Table 1 summarizes detailed bearing specifications, and Fig. 2
than the first design presented in Ref. 关21兴. Low speed tests at shows overall schematic description of the three-pad HAFB.
10,000 rpm for HAFB with 38.1 mm diameter resulted in more
than 150 N load capacity with a supply pressure of 4 bars. Both 3 Mathematical Model
theoretical work in Ref. 关22兴 and experimental works on HAFB Following typical assumptions and nomenclature used for gas
关21兴 were on circular HAFB with continuous single top foil with lubrications, nondimensionalized Reynolds equation for com-
four inherent restrictors. pressible fluids with hydrostatic gas supply is given by
General interest in large AFBs 共diameter⬎ 100 mm兲 has
evolved from many gas-processing and military applications. The
Mechanical Systems Branch within the Turbine Engine Division
⳵
⳵␪
冉⌳PH − PH3
⳵P
⳵␪
+
⳵
⳵Z
− PH3 冊 冉
⳵P
⳵␪
⳵
+ 2⌳␯ 共PH兲 =
⳵␶
Ṁ s
⌬␪⌬Z
冊
at Propulsion Directorate of U.S. Air Force Research Laboratory
共AFRL兲 has performed extensive research on AFB technology and 共1兲
its integration to high-speed engines for future aircraft and where P = p / Pa, ␪ = x / R, Z = z / R, ␶ = ␻st, H = h / C, ⌳
weapon systems 关24兴. The AFRL test rig shown in Fig. 1 was = 6␮␻ / pa共R / C兲2, ␯ = ␻s / ␻, and Ṁ s = 12␮RgTṁs / p2aC3. ␻s is the
designed 共by AFRL researchers兲 with a rotor supported by AFB
excitation frequency, x is the local coordinate attached on the
with 101.6 mm diameter 共4 in.兲 at the front end and by duplex ball
bearing surface along the circumferential direction, z is the axial
bearing at the rear end, and the rotor is driven by 50 hp electric
coordinate, h is the film thickness, p is the pressure, ␮ is the
motor through a flexible coupling attached to the rear end. Mag-
netic bearing located in the half way between the two bearing viscosity of air, Rg is the gas constant of air, ṁs is the air mass
supports applies various time-varying loads to the test foil bear- flow rate through the orifice, and T is the temperature of supplied
ing. Equivalent rotor weight to the foil bearing is about 25 kg 共55 air. Mathematical model of bump foils follows independent elastic
lb兲, and design speed of the test bearing is 25,000 rpm. foundation model used in Ref. 关25兴.
Current work in this paper began with designing HAFB for the Assuming the flow through the orifice is an isentropic process,
AFRL test rig. Due to very high-speed requirements of AFRL test the mass flow rates of the compressible fluid for the choked and
rotor 共DN⬎ 2.5 M兲 and large rotor mass, rotor-bearing stability unchoked conditions are given by
was a major concern, and thus preloaded three-pad HAFB was
selected as a design, and its details are presented in Table 1.
This paper presents both comprehensive parametric simulation
unchoked:
P
Ps
冉 冊
⬎
2
k+1
k/共k−1兲
= 0.5283

studies and preliminary experimental work characterizing stiffness

at zero running speed. The results from this design study may be
also applicable to midsized turbomachinery where journal diam-
Ṁ S = ⌫s PsH 冋 冉冉 冊 冉 冊 冊册
2k
k−1
P
Ps
2/k
−
P
Ps
共k+1兲/k 1/2
共2兲

冉 冊
eter is greater than 100 mm. k/共k−1兲
P 2
choked: ⬍ = 0.5283
2 Design Ps k+1
Design simulation works include imbalance response of a point
mass rigid rotor supported by HAFB via time-domain orbit simu- Ṁ S = ⌫s PsH 2 冉 冊冉 冊k
k+1
1/2
2
k+1
1/共k−1兲
共3兲

where ⌫s = 12␮CdA0冑RgT / paC3 is the feed parameter, Ps is the

Table 1 Brief design specifications of three-pad HAFB nondimensional supply pressure, k = c p / cv is the specific heat ratio
of air, and Cd is the discharge coefficient 共=0.5 in this work兲. In
Bearing nominal diameter 共2R兲 101.6 mm 共4 in.兲 the feed parameter, A0 is the reference orifice curtain area defined
Bearing length 82.55 mm 共3.25 in.兲 as A0 = ␲doC with d0 as the orifice diameter. Combined effect of
Pad preload offset 共R P兲 30 ␮m共1.18 mils兲 discharge coefficient and orifice diameter defines the feed param-
Set bore radial clearance 共CSB兲 70 ␮m共2.76 mils兲 eter if other parameters are constant. For the selected orifice di-
Stiffness per unit area of the bump foil 20 GN/ m3
ameter of 1 mm with discharge coefficient of 0.5, ⌫S = 0.1 for
Total number of orifices on the bearing 3
Orifice diameter on the top foil for 1 mm supply air with ambient temperature.
hybrid operation Rotor centrifugal growth is considered using the plane stress
Static load per bearing projected area 29.2 kN/ m2 elastic model, which agrees very well with finite element results
关26兴. The centrifugal growth of a solid rotor may be expressed by

122504-2 / Vol. 132, DECEMBER 2010 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

  g


Y HAFB

mr g

Fig. 4 Description of load vector to HAFB with respect to rotor

orientation: „a… X-direction „loaded direction… and „b…
Y-direction
X
Fig. 4. The worst rotor-orientation for AFBs is a vertical position
Fig. 3 Coordinate system for orbit simulations 共␾ = 0 deg兲, where bearing is not loaded at all. Orbit simulations
were performed for load angle of 90 deg with load on pad 共LOP兲
configuration as a reference. Other reference simulation condi-
R3␳␻2 tions are summarized in Table 2.
rg = 共1 − ␯r兲 共4兲 Imbalance responses were simulated from 1000 rpm to 30,000
4E rpm with 1000 rpm increment for 750 cycles to attain steady state
for given material density ␳, Young’s modulus E, and Poisson’s response at each simulate speed. Last 4096 data points were taken
ratio ␯. for discrete Fourier transform 共DFT兲 analyses for each speed and
Time-domain orbit simulations assumed a point mass rigid ro- all the DFT signals were combined to a waterfall plot.
tor supported by single HAFB. The rotor equation of motions Figure 5 shows the waterfall plots of HAFB characteristics in
along the X- and Y-directions 共see Fig. 3兲, and bump equation of both X- and Y-directions. No subsynchronous vibration is ob-
motions are solved in time domain using time marching tech- served for entire speed range. Because the bearing is loaded along
niques as presented in Refs. 关25,27兴. the X-direction 共by rotor weight兲, critical speed in the X-direction
Linear perturbation analysis was performed for HAFB using the is higher than that in the Y-direction. Figure 6 shows peak-to peak
method presented in Refs. 关22,25兴 to find frequency-dependent synchronous vibrations taken from time-domain speed-up simula-
bearing impedances and rotor-bearing natural frequency. After tions, which shows exact critical speeds in the X-direction of 7128
frequency-dependent stiffness and damping coefficients were cal- rpm and Y-direction of 2564 rpm.
culated, modal stiffness and modal damping along forward whirl For the comparison, imbalance responses of the hydrodynamic
coordinate were calculated to predict rotor-natural frequency, fol- AFB were also predicted. Figure 7 shows the waterfall plots of
lowing the procedure presented in Refs. 关28,29兴. The procedure is hydrodynamic AFB. Subsynchronous limit cycle begins to appear
briefly reviewed here. at around 25,000 rpm but they are bounded within a few microns.
The Laplace transform of a homogeneous equation of rotor mo- The subsynchronous vibration increases slightly from 3295 rpm to

册再 冎 冋 册再 冎
tion in cylindrical coordinate is represented as 3430 rpm as the rotor running speed increases from 25,000 to

冋 m rs 2
0
0
m rs 2
X共s兲
Y共s兲
+
zXX zXY
zYX zYY
X共s兲
Y共s兲
=0 共5兲
30,000 rpm due to the typical gas film stiffening effect. These
subsynchronous vibrations correspond to natural frequencies in
modal coordinate, which are different in general from critical
From Ref. 关29兴, it can be found that the modal natural fre- speeds observed in synchronous vibrations along X- and
quency at a certain rotor speed, ␻, can be estimated by finding a Y-directions in waterfall plot. However, the gravitational loading
modal stiffness 共kmodal兲 contour intersected by mr␻2S when plotted along the X-direction results in smaller stiffness along the
against excitation frequency ratio ␯ = ␻S / ␻. In mathematical Y-direction, resulting in similar modal natural frequency to the
terms, the modal natural frequency for forward whirling motion is critical speed in Y-direction, as presented in Fig. 5共b兲.
expressed by 3.2 Linear Analyses for Natural Frequency Identification.
␻nat = ␻␯0 for ␯0 = 兵␯兩kmodal = mr␻2S = m r␻ ␯ 其
2 2
共6兲 Figures 8 and 9 show the bearing stiffness and damping coeffi-
cients of the HAFB at 24,000 rpm at various excitation frequen-
where kmodal is the real part of the eigenvalue of the frequency- cies. Figure 10 represents bearing modal impedances along the
dependent bearing impedance matrix in Eq. 共5兲. forward whirling modal coordinate. The real part corresponds to
the modal stiffness and the imaginary part is the modal damping.
3.1 Imbalance Response of Point Mass Rigid Rotor. The
As stated in Eq. 共6兲, a crossing point of the modal stiffness and
test rig shown in Fig. 1 is consisted of a speed-controlled 50 hp
induction motor, a flexible coupling, a duplex ball bearing at the mr共␻␯兲2 versus ␯ curve represents the modal natural frequency of
rear end of the rotor, and a HAFB support at the front end. The
total rotor mass is about 40 kg, and the weight distributions be- Table 2 Simulation conditions
tween the ball bearing and HAFB are 15 kg and 25 kg,
respectively.
Feed parameter 共⌫S兲 0.1
For preliminary evaluation of rotor-bearing stability, a virtual
Supply pressure 共Ps兲 5
rotor-bearing system with 25 kg point mass rigid rotor supported Structural loss factor of elastic foundation 0.2
by one HAFB was used for imbalance response simulations. Con- Load 共rotor weight, LOP兲 245 N
sidering potential airborne military applications of the HAFB, Load angle, ␾共see Fig. 4兲 90 deg
simulations were performed for various rotor orientations with Imbalance 45 g mm
different load vectors with respect to the bearing axis, as shown in

Journal of Engineering for Gas Turbines and Power DECEMBER 2010, Vol. 132 / 122504-3

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 7,128 rpm

Peak to peak (m)

RPM

(a) X-direction (loaded direction)

(a) X-direction (loaded direction)

2,564 rpm

Peak to peak (m)

RPM

(b) Y-direction

Fig. 6 Synchronous response, peak-to-peak is twice the mag-

nitude in waterfall plot: „a… X-direction „loaded direction… and
„b… Y-direction

(b) Y-direction
bearing nominal clearance, a feed parameter of 0.2 corresponds to
Fig. 5 Waterfall plot of HAFB: „a… X-direction „loaded direc- the orifice diameter of 2 mm, which is quite large for HAFB
tion… and „b… Y-direction applications.
3.3 Effect of Load Angle. Orbit simulations were performed
for load angles of 60 deg, 45 deg, and 30 deg to investigate the
the rotor-bearing system for the forward whirl. Figure 10 shows potential of HAFB for various airborne applications. Figures 14,
the modal impedance plots at various speeds. For example, ␯n 15, and 17 show the simulated waterfall plots for load angles of
= 0.268 corresponds to natural frequency of 2412 rpm at 9000 rpm 60 deg, 45 deg, and 30 deg, respectively. For a load angle of 60
共i.e., 9000⫻ 0.268= 2412兲. Modal natural frequencies at different deg, the bearing is stable for entire speed range without any sub-
speeds can be predicted in a similar way, and they are summarized synchronous vibration. However, as the load angle is decreased to
in Table 3. The modal natural frequencies in Table 3 increase with 45 deg, subsynchronous bounded vibrations begin to appear at
the running speed due to the nonlinear stiffening effect of gas film around 27,000 rpm and increase very slowly up to 27,400 rpm, as
with speed and rotor centrifugal growth. At low speeds, the modal shown in Fig. 15. These subsynchronous vibrations are very
natural frequencies are very close to the critical speeds observed small, bounded, and almost negligible in magnitude compared
in Figs. 5 and 6. with the synchronous components. However, interestingly
Figures 11 and 12 show the synchronous bearing stiffness and enough, time response at 27,500 rpm is completely unstable and
damping coefficients with different feed parameters, ⌫S. Unlike subsynchronous vibration grows rapidly, as shown in Fig. 16,
circular single pad HAFB presented in Ref. 关22兴, cross-coupled where the dotted circle represents the bearing’s set bore circle.
stiffness kXY does not decrease much with ⌫S. For the present Similar tendency is observed for load angle of 30 deg 共Fig. 17兲.
three-pad HAFB, direct stiffness kXX decreases with ⌫S in general The bearing is stable up to 20,900 rpm with limit cycles and then
because the rotor static eccentricity decreases with ⌫S. However, the orbit blows up at 21,000 rpm, as shown in Fig. 18. The last
direct damping coefficient dXX increases with ⌫S. Combined effect several cycles in Figs. 16 and 18 are the triangular shapes bounc-
of the decreased direct stiffness and cross-coupled stiffness, and ing back from the pads.
increased direct damping leads to increased modal damping and The peculiar behavior of sudden appearance of unbounded sub-
decreased modal stiffness with ⌫S, as shown in Fig. 13. It is ex- synchronous vibrations for load angles ␾ = 45 deg and 30 deg
pected that the rotor-natural frequency would slightly decrease does not happen in hydrodynamic AFB, as shown in Fig. 19. For
with ⌫S but rotor-bearing stability would increase with ⌫S. How- ␾ = 45 deg, bounded limit cycles 共with modal natural frequencies,
ever, maximum limit of practically implementable feed parameter ⬃3000 rpm兲 initiated at ⬃20,000 rpm increase gradually until it
is around 0.2 for the currently designed HAFB. For the given is several times larger than synchronous vibrations, and they per-

122504-4 / Vol. 132, DECEMBER 2010 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Damping (kNs/m)


Fig. 9 Damping coefficient of HAFB, 24,000 rpm: „a… 9,000

rpm, „b… 15,000 rpm, „c… 24,000 rpm, and „d… 30,000 rpm

(a) X-direction (loaded direction) system becomes unstable.

Hybrid operation is more stable when full rotor weight is taken
by the bearing 共␾ = 90 deg兲, as shown in Figs. 5 and 7. However,
when the bearing takes only partial rotor weight, the hydrody-
namic operation allows slightly higher operating speeds with
bounded limit cycles. For the HAFB, the subsynchronous vibra-
tion first appears at about 30% and 20% higher speed than hydro-
dynamic bearing for the load angles of 45 deg and 30 deg, respec-
3,295~3,430 rpm tively. However, “bounded” limit cycle behavior, which is one of
the notable features of AFBs, does not appear at partially loaded
HAFB. As shown in Fig. 13, as the hydrostatic effect increases
共⌫S increases兲, the modal stiffness decreases due to the decrease in
eccentricity. Therefore, as the hydrostatic effect increases, the dy-
namic behavior of rotor depends more dominantly on the air film
共subsynchronous vibration is hydrodynamic in nature兲 than foil
structure which provides damping for the bounded subsynchro-
nous vibrations. These results indicates that the feed parameter,
⌫S, and supply pressure should be determined considering loading
condition and operating speed range.

(b) Y-direction 4 Experimental Study

Designed HAFB was manufactured with an orifice diameter of
Fig. 7 Waterfall plot of hydrodynamic AFB 1 mm 共⌫S = 0.1兲 on the top foil, as shown in Fig. 22. L-shaped
stainless steel orifice tube with an inner diameter of 1.7 mm was
laser welded on the backside of each top foil. Bump foils were
sist until somewhere between 29,000 rpm and 30,000 rpm where made of stainless steel 316 using a precision forming jig made of
the subsynchronous vibration becomes unbounded. Load angle Inconel 718.
␾ = 30 deg for hydrodynamic AFB results in similar results. Limit One of the benefits of HAFB is nonzero load capacity at zero
cycles begin to appear at ⬃16,000 rpm and they are stable up to running speed. As a preliminary experimental evaluation of the
somewhere between 22,000 rpm and 23,000 rpm 共Fig. 20兲 where designed HAFB, stiffness at zero running speed was measured
the subsynchronous vibration becomes unbounded. Figure 21 using a simple test rig, as shown in Fig. 23. A journal shaft with
shows the limit cycles of hydrodynamic AFB at 29,000 rpm 共␾ 101.6 mm 共4 in.兲 was fixed on a pedestal, and various gravita-
tional loads were applied to the test bearing through a pulley
= 45 deg兲 and 22,000 rpm 共␾ = 30 deg兲 before the rotor-bearing
mechanism along the X-direction 共see Fig. 3兲. A dial indicator
measured static deflection of the bearing to calculate stiffness of
the bearing. Supply pressure was measured using a digital pres-
16.0 sure gauge at upstream of each orifice tube, and airflow rate to the
14.0
kxx loaded pad was measured using a digital air flow meter.
kyx Bearing stiffness was measured at 5.4 bars 共78 psi兲 with airflow
12.0 rate to the loaded pad of ⬃6 SLPM at no load. Figure 24 shows
Stiffness (MN/m)

kxy
10.0 the measured load-deflection curve and predictions 共solid line兲.
kyy
Load was applied up to the point where the air flow meter reading
8.0
becomes ⬍2 SLPM. Tests were repeated seven times, and the test
6.0 results are repeatable, as shown in Fig. 24. Dial indicator reading
4.0
showed very small1 amount of residual deflection of bump foils,
which was believed to be hysteresis type unloading behavior.
2.0 Upon gently tapping the bearing sleeve, the dial indicator reading
0.0 always came back to original zero. From the seven consecutive

‐2.0 0.03 0.3 3
1
Each test shows slightly different residual deflections and the variation was
Fig. 8 Stiffness coefficient of HAFB, 24,000 rpm larger than indicator’s precision.

Journal of Engineering for Gas Turbines and Power DECEMBER 2010, Vol. 132 / 122504-5

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 mr ( ) 2 Table 3 Summary of modal natural frequencies predicted from
different speeds

Impedance (MN/m) Real part Modal natural frequency

Running speed ␯n 共rpm兲

9000 0.268 2412

Imaginary part 15,000 0.187 2802
n=0.268  24,000 0.130 3113
30,000 0.109 3279

measured stiffness is about 9% higher than prediction at low load

region but 12% smaller than prediction at high loads. Predicted
(a) 9,000 rpm
stiffness agrees well with the measured results.

4.0 5 Conclusion
mr ( ) 2
A three-pad hybrid air foil bearing was designed and con-
3.0 structed for AFRL test rig. Time-domain orbit simulations and
Impedance (MN/m)

Real part
perturbation analyses were performed to investigate the character-
2.0 istics of the bearing. To investigate the load carrying capacity of
the hydrostatic operation at zero running speed, static stiffness
1.0 was estimated by measuring deflections of the bearing under vari-
Imaginary part ous external loads.
0.0 Due to the nonlinear stiffening effect of the air film, the modal
n=0.187 
0.03 0.3 3 natural frequency increases with running speed and predicted
‐1.0 modal natural frequencies are very close to the critical speeds
along the Y-direction predicted by time-domain orbit simulations.
‐2.0 As the feed parameter is increased, the modal stiffness decreases
(b) 15,000 rpm and the modal damping increases, indicating smaller natural fre-
quency and better rotor-bearing stability. These results are consis-
tent with time-domain orbit simulations; the subsynchronous vi-
mr ( ) 2 brations of HAFB appear at higher speed than hydrodynamic
Real part
Impedance (MN/m)

14.0

12.0
Stiffness (MN/m)

Imaginary part 10.0 kxx

kyx
n=0.13  8.0
kxy
6.0
kyy
4.0
(c) 24,000 rpm
(c)
2.0

0.0
mr ( ) 2
0 0.05 0.1 0.15 0.2  S 0.25 0.3
Real part
Impedance (MN/m)

Fig. 11 Synchronous stiffness coefficient for various ⌫S,

25,000 rpm

Imaginary 4.0

n=0.109  3.5
3.0 dxx
Damping (kNs/m)

2.5 dyx
2.0 dxy
1.5 dyy

(d) 30,000 rpm

(d) 1.0
0.5
Fig. 10 Modal impedances of HAFB at various speeds 0.0
‐0.5 0 0.1 0.2 S 0.3
‐1.0
measurements up to 400 N load, bump foils made of stainless 316
seemed to have no permanent deformation. Static stiffnesses were
estimated as 2.08 MN/m for load range of 0–175 N, 6.70 MN/m Fig. 12 Synchronous damping coefficient for various ⌫S,
between 175 N and 310 N, and 15.5 MN/m at loads ⬎310 N. The 25,000 rpm

122504-6 / Vol. 132, DECEMBER 2010 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 3.5
Y

Impedance (MN/m)
3.0
Imaginary part
2.5
Real part
2.0

1.5
Load
1.0 direction X
0.5

0.0
0 0.1 0.2 S 0.3

Fig. 13 Synchronous modal impedances for various ⌫S, Film

25,000 rpm: „a… X-direction and „b… Y-direction
break

Fig. 16 Simulated journal center orbit at 27,500 rpm, ␾

= 45 deg „equivalent load= 173.2 N…

(a) X-direction (b) Y-direction

Fig. 14 Waterfall plot HAFB, ␾ = 60 deg „equivalent load= 212.2 N…. The bearing is stable for entire speed range without
any subsynchronous vibration: „a… 1000–27,400 rpm and „b… 27,000–27,400 rpm

(a) 1,000~27,400 rpm (b) 27,000~27,400rpm

Fig. 15 Waterfall plot HAFB in X-direction, ␾ = 45 deg „equivalent load= 173.2 N…. Subsynchronous bounded vibrations
begin to appear at around 27,000 rpm and increase very slowly up to 27,400 rpm.

Journal of Engineering for Gas Turbines and Power DECEMBER 2010, Vol. 132 / 122504-7

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 bearing. However, once subsynchronous vibration appears, it be-
comes unbounded very quickly as speed is increased a little be-
yond the initial onset speed of subsynchronous vibration. This
peculiar subsynchronous vibration of HAFB, prominent especially
when the bearing is lightly loaded, is believed to be hydrodynamic
instability because dynamic characteristics of HAFB depends
more on air film rather than foil structure. In addition, as load
angle is decreased, subsynchronous vibrations were predicted at
lower speeds. These results indicate that feed parameter and sup-
ply pressure should be determined based on the loading condition
and operating speed. Measured static stiffness at zero running
speed increases with applied load and shows good agreement with
prediction.

Acknowledgments
This work was supported by U.S. Air Force Research Labora-
tory through Contract No. 09-S590-0009-27-C1 and 2009 AFOSR
Summer Faculty Fellow Program. The authors also recognize Mr.
Fig. 17 Waterfall plot up to 20,000 rpm HAFB in X-direction, Garry Givan and other colleagues at Air Force Research Labora-
␾ = 30 deg „equivalent load= 122.5 N… tory for helping to set up the experimental facility for static stiff-
ness measurements.

Nomenclature
Ao ⫽ orifice curtain area
Y C ⫽ nominal bearing clearance
Cd ⫽ discharge coefficient of orifice
CSB ⫽ set bore radial clearance
do ⫽ orifice diameter
E ⫽ elastic modulus of rotor
H ⫽ nondimensional film thickness
Load h ⫽ film thickness
direction X k ⫽ ratio of specific heats for air
mr ⫽ rotor mass
Ṁ s ⫽ nondimensional mass flow rate
ṁs ⫽ mass flow rate
P ⫽ nondimensional pressure
Ps ⫽ nondimensional supply pressure
p ⫽ film pressure
Film
pa ⫽ ambient pressure
break ps ⫽ supply pressure
R ⫽ bearing radius
Rg ⫽ gas constant of air
Fig. 18 Simulated journal center orbit at 21,000 rpm, ␾ Rp ⫽ pad preload offset
= 30 deg „equivalent load= 122.5 N…: „a… X-direction and „b… Rpad ⫽ pad radius
Y-direction

(a) X-direction (b) Y-direction

Fig. 19 Waterfall plot of hydrodynamic AFB at 20,000–29,000 rpm, ␾ = 45 deg: „a… X-direction and „b… Y-direction

122504-8 / Vol. 132, DECEMBER 2010 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 (a) X-direction (b) Y-direction
Fig. 20 Waterfall plot of hydrodynamic AFB at 15,000–22,000 rpm, ␾ = 30 deg: „a… 29,000 rpm for ␾ = 45 deg and „b… 22,000
rpm for ␾ = 30 deg

0.6
0.6 Y Y

0.4
0.4

0.2 0.2

Load Load
direction direction
X X
0.0 0.0
0 0.2 0.4 0.6 0.0 0.2 0.4 0.6

(a) 29,000 rpm for =45 o

(b) 22,000 rpm for =30 o

Fig. 21 Limit cycle orbit „50 cycles… of hydrodynamic AFB

Fig. 22 Photo of manufactured HAFB

Fig. 23 Photo of the test rig for static stiffness measurements

Rrotor ⫽ rotor radius

rg ⫽ centrifugal growth of rotor
␪ ⫽ circumferential coordinate
T ⫽ temperature of air
⌳ ⫽ bearing number
t ⫽ time
␮ ⫽ viscosity of air
X,Y ⫽ Cartesian coordinate for rotor motion
␯ ⫽ excitation frequency ratio
Z ⫽ nondimensional axial coordinate for the
␯r ⫽ Poisson’s ratio of rotor
bearing
␳r ⫽ density of rotor
Greek ␶ ⫽ nondimensional time
⌫S ⫽ feed parameter ␾ ⫽ load angle

Journal of Engineering for Gas Turbines and Power DECEMBER 2010, Vol. 132 / 122504-9

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 140.0 boalternator and Associated Hardware. 2—Design of Gas Bearings,” NASA
Report No. NASA-CR-1291.
15.5 MN/m
120.0 关11兴 Han, D. C., Park, S. S., and Kim, W. J., 1994, “A Study on the Characteristics

Displacement (m)
of Externally Pressurized Gas Bearings,” Precis. Eng., 16共3兲, pp. 164–173.
100.0 关12兴 Osborne, D. A., and San Andrés, L., 2006, “Experimental Response of Simple
6.70 MN/m Gas Hybrid Bearings for Oil-Free Turbomachinery,” ASME J. Eng. Gas Tur-
80.0
bines Power, 128, pp. 626–633.
2.08 MN/m Prediction 关13兴 Zhu, X., and San Andrés, L., 2004, “Rotordynamic Performance of Flexure
60.0
17.32 MN/m Pivot Hydrostatic Gas Bearings for Oil-Free Turbomachinery,” ASME Paper
No. GT2004-53621.
40.0
关14兴 San Andrés, L., 2006, “Hybrid Flexure Pivot-Tilting Pad Gas Bearings: Analy-
20.0 sis and Experimental Validation,” ASME J. Tribol., 128, pp. 551–558.
1.91 MN/m 关15兴 Bosley, R. W., and Miller, R. F., 1998, “Hydrostatic Augmentation of a Com-
0.0 pliant Foil Hydrodynamic Fluid Film Thrust Bearing,” U.S. Patent No.
5,827,040.
0 100 200 300 400 500
Load (N) 关16兴 Eshel, A., 1979, “Numerical Solution of the Planar Hydrostatic Foil Bearing,”
ASME J. Lubr. Technol., 101, pp. 86–91.
关17兴 Heshmat, H., Chen, H. M., and Walton, J. F., II, 2000, “On the Performance of
Fig. 24 Load-deflection curve from static stiffness tests com- Hybrid Foil-Magnetic Bearings,” ASME J. Eng. Gas Turbines Power, 122共1兲,
pared with prediction; supply pressure= 5.4 bars. Prediction pp. 73–81.
under 40 N is represented as dotted curve with a stiffness of 关18兴 Swanson, E. E., and Heshmat, H., 2002, “Oil-Free Foil Bearings as a Reliable,
1.91 MN/m. High Performance Backup Bearing for Active Magnetic Bearings,” ASME
Paper No. GT2002-30291.
关19兴 Swanson, E. E., Heshmat, H., and Walton, J., 2002, “Performance of a Foil-
␻ ⫽ rotor spin speed Magnetic Hybrid Bearing,” ASME J. Eng. Gas Turbines Power, 124共2兲, pp.
375–382.
␻s ⫽ excitation frequency 关20兴 Palazzolo, A., Tucker, R., Kenny, A., Kang, K. D., Ghandi, V., Liu, J., Choi,
H., and Provenza, A., 2008, “High Temperature, Permanent Magnet Biased,
References Fault Tolerant, Homopolar Magnetic Bearing Development,” ASME Paper No.
GT2008-50917.
关1兴 DellaCorte, C., and Edmonds, B. J., 1995, “Preliminary Evaluation of PS300: 关21兴 Kim, D., and Park, S., 2009, “Hydrostatic Air Foil Bearings: Analytical and
A New Self-Lubricating High Temperature Composite Coating for use to Experimental Investigations,” Tribol. Int., 42, pp. 413–425.
800°C,” NASA Technical Report No. TM-107056. 关22兴 Kumar, M., and Kim, D., 2008, “Parametric Studies on Dynamic Performance
关2兴 DellaCorte, C., Lukaszewicz, V., Valco, M. J., Radil, K. C., and Heshmat, H., of Hybrid Air Foil Bearings,” ASME J. Eng. Gas Turbines Power, 130, p.
2000, “Performance and Durability of High Temperature Foil Air Bearings for 062501.
Oil-Free Turbomachinery,” NASA Technical Report No. TM-2000-209187/
关23兴 Kumar, M., and Kim, D., 2010, “Static Performance of Hydrostatic Air Bump
ARL-TR-2202.
Foil Bearing,” Tribol. Int., 43共4兲, pp. 752–758.
关3兴 Stanford, M. K., Yanke, A. M., and DellaCorte, C., 2004, “Thermal Effects on
关24兴 Bliss, E. J., and Brown, J. R., 2007, “Experimental Investigation of Foil Bear-
a Low Cr Modification of PS304 Solid Lubricant Coating,” NASA Technical
Report No. TM-2003-213111. ing Stiffness and Damping Characteristics,” Proceedings of the 32nd Dayton-
关4兴 Mohawk Innovative Technology, 2005, “Low-Friction, Wear-Resistant Koro- Cincinnati Aerospace Science Symposium.
lon™ C36 Coatings for High-Temperature, High-Speed, Air Foil Bearings,” 关25兴 Kim, D., 2007, “Parametric Studies on Static and Dynamic Performance of Air
Mohawk Innovative Technology Internal Newsletter 23. Foil Bearings With Different Top Foil Geometries and Bump Stiffness Distri-
关5兴 Lubell, D., DellaCorte, C., and Stanford, M., 2006, “Test Evolution and Oil- butions,” ASME J. Tribol., 129, pp. 354–364.
Free Engine Experience of a High Temperature Foil Air Bearing Coating,” 关26兴 Sim, K., and Kim, D., 2007, “Design of Flexure Pivot Tilting Pads Gas Bear-
ASME Paper No GT2006-90572. ings for High-Speed Oil-Free Microturbomachinery,” ASME J. Tribol., 129,
关6兴 Heshmat, H., 2006, “Evaluation of Coatings for a Large Hybrid Foil/Magnetic pp. 112–119.
Bearing,” ASME Paper No. IJTC2006-12328. 关27兴 Song, J., and Kim, D., 2007, “Foil Gas Bearing With Compression Springs:
关7兴 Lund, J. W., 1964, “The Hydrostatic Gas Journal Bearing With Journal Rota- Analyses and Experiments,” ASME J. Tribol., 129, pp. 628–639.
tion and Vibration,” ASME J. Basic Eng., 86, pp. 328–336. 关28兴 Kim, D., Rimpel, A. M., Chang, S. S., and Kim, J. H., 2009, “Design and
关8兴 Lund, J. W., 1967, “A Theoretical Analysis of Whirl Instability and Pneumatic Manufacturing of Meso Scale Tilting Pad Gas Bearings for 100–200 W Class
Hammer for a Rigid Rotor in Pressurized Gas Journal Bearings,” ASME J. PowerMEMS Applications,” ASME J. Eng. Gas Turbines Power, 131共3兲, p.
Lubr. Technol., 89, pp. 154–166. 042503.
关9兴 Mori, H., and Miyamatsu, Y., 1969, “Theoretical Flow-Models for Externally 关29兴 Rimpel, A. M., and Kim, D., 2009, “Rotordynamic Performance of Flexure
Pressurized Gas Bearings,” ASME J. Lubr. Technol., 91, pp. 183–193. Pivot Tilting Pad Gas Bearings With Vibration Damper,” ASME J. Tribol.,
关10兴 Curwen, P. W., Frost, A., and Lund, J. W., 1969, “High-Performance Tur- 131共2兲, p. 021101.

122504-10 / Vol. 132, DECEMBER 2010 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use