You are on page 1of 8

Scaling Laws for Radial

Clearance and Support Structure


Srikanth Honavara Prasad1
Mem. ASME
Department of Mechanical and Aerospace
Stiffness of Radial Foil Bearings
Engineering,
The University of Texas at Arlington, Design and analysis of foil bearings involve consideration to various physical aspects
500 W. 1st Street, such as fluid pressure, structural deformation, and heat generation due to viscous effects
Arlington, TX 76019 within the bearing. These complex physical interactions are mathematically governed by
e-mail: srikanth.honavaraprasad@mavs.uta.edu highly nonlinear partial differential equations. Therefore, foil bearing design involves
detailed calculations of flow fields (velocities, pressures), support structure deflections
(structural compliance), and heat transfer phenomena (viscous dissipation in the fluid,
Daejong Kim frictional heating, temperature profile, etc.). The computational effort in terms of time
Department of Mechanical and Aerospace
and hardware requirements make high level engineering analyses tedious which presents
Engineering,
an opportunity for development of rule of thumb laws for design guidelines. Scaling laws
The University of Texas at Arlington,
for bearing clearance and support structure stiffness of radial foil bearings of various
500 W. 1st Street,
sizes are presented in this paper. The scaling laws are developed from first principles
Arlington, TX 76019
using the scale invariant Reynolds equation and support structure deflection equation.
e-mail: daejongkim@uta.edu
Power law relationships are established between the (1) radial clearance and bearing
radius and (2) support structure stiffness and bearing radius. Simulation results of static
and dynamic performance of various bearing sizes following the proposed scaling laws
are presented. [DOI: 10.1115/1.4034648]

Introduction bearings, the numerical value of the load capacity coefficients


should be interpreted with caution. The ROT guideline allows
Gas foil bearings support rotating components in turbo-
direct estimation of load capacity of a specific foil bearing config-
machinery through hydrodynamic pressure generation in the fluid
uration for different sizes and speeds. Subsequently, DellaCorte
film between the bearing and the shaft. The foil bearings can be
[2] presents similar tools for simple estimation of direct stiffness
used with a variety of process gases resulting in increased flexibil-
and damping coefficients of foil bearings. However, these guide-
ity and reliability over conventional bearings. Bulky oil lubrica-
lines were developed using experimental data from open literature
tion systems can be eliminated in aerospace applications by the
and therefore represent approximate relationships only for a cer-
adoption of foil bearings which reduces the payload and the corre-
tain range of bearing sizes. These relationships may not hold for
sponding cost. These benefits can be fully exploited only when
very large bearings beyond currently available bearings in indus-
the foil bearing design is deemed feasible for a given application.
trial applications or for bearings operating in extreme conditions
Feasibility assessments of the foil bearings can be performed by
(extreme temperatures, very high speeds, dynamic loading, very
construction of hardware and rigorous experimental testing or by
high altitudes, very high ambient pressure or dense fluid, nonideal
carefully planned numerical parametric studies. The former
gases, etc.). For example, the load capacity of hydrodynamic foil
approach is capital intensive, and the latter is challenging due to
bearings is proportional to the second power of rotor characteristic
the computational complexities of nonlinear systems. Therefore, it
dimension, whereas the rotor weight is proportional to the third
is desirable to have reliable but simple design guidelines for bear-
power of rotor characteristic dimension [3]. Similarly, load
ing load capacity, stiffness of the structural support, radial clear-
capacity is proportional to speed only within a certain range of
ance, etc., and their interdependencies when candidate shaft size
shaft speeds or DN number (product of diameter in mm and speed
for the foil bearing is given.
in revolutions per minute (RPM)). Peng and Khonsari show that
DellaCorte and Valco [1] present “rule of thumb” (ROT) empir-
the limiting load capacity for foil bearings occurs for extreme DN
ical design guideline to aid feasibility assessments of foil bearings
number (>50  106) [4], while the load capacity is proportional to
in terms of load capacity. The empirical ROT of the foil bearing
speed for practical range of DN number (3  106) [5]. Therefore,
load capacity is represented by
the operating speed and size of the bearing are limited by physical
saturation limits and the ROTs must be used with caution. Further,
F ¼ nLD2 N (1)
the coefficients for load capacity, stiffness, and damping may
have other dependencies.
where, n is the bearing load capacity coefficient, which is a con-
Radil et al. investigate the effect of radial clearance on the load
stant with the units lbf=ðin3  krpmÞ, L and D are the bearing
capacity coefficient by performing load capacity tests on the foil
length and nominal diameter in inch, and N is the shaft speed in
bearings [6]. Their load capacity tests show that the maximum
krpm. The formula is based on curve fitting of load capacities of
load capacity coefficient is obtained at a certain optimum radial
various foil bearings available to the authors at the time of investi-
clearance. Operation below the optimum radial clearance
gation. The coefficient n depends on various design parameters of
increases the possibility of thermal runaway while operation
the bearings such as number of top foils, type of structural sup-
above the optimum clearance reduces the load capacity of the
ports, spatial variation of support stiffness, etc. Because there are
bearing. Their study did not include the effect of radial clearance
so many variations in the physical configuration of the foil
on dynamic characteristics such as stiffness and damping of the
bearing.
1
Corresponding author. Spakovszky and Liu [7] present scaling laws for the dynamic
Contributed by the Structures and Dynamics Committee of ASME for publication
in the JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received
behavior of ultra-short hydrostatic gas bearings. This is accom-
July 5, 2016; final manuscript received August 12, 2016; published online October plished by first finding analytical expressions for hydrostatic stiff-
26, 2016. Editor: David Wisler. ness, hydrodynamic stiffness, and damping, and then, deriving

Journal of Engineering for Gas Turbines and Power APRIL 2017, Vol. 139 / 042502-1
C 2017 by ASME
Copyright V

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a


expressions for natural frequency and damping ratio. They also where kf 0 is a proportionality constant which is a function of spe-
investigate the onset of whirl instability and provide a simple cific bearing design. The term kf 0 ðpl=10pa Þ consists of constants,
geometry dependence criterion for whirl ratio. Further, they and it can be lumped under a new constant kf and the load
develop design charts for linking fabrication tolerances, bearing capacity of the bearing can be expressed as
performance, and rotor unbalance for minimum required whirl
 
ratio. 2 R
While these papers provide useful guidelines to the bearing F ¼ kf LD N 2 (6)
C
designers when they want to estimate the load capacity and bear-
ing coefficients of certain gas bearings, there are no simple guide- For Eq. (6) to match to the empirical ROT represented by Eq.
lines on how to choose the bearing clearance and stiffness of the (1), the following relation should hold:
structural support elements of foil bearings when scale-down or
scale-up design of existing foil bearings is required. This paper  
R
aims to address these gaps and presents simple scaling laws for n ¼ kf (7)
bearing clearance and stiffness of underlying structural elements C2
for radial foil bearings.
Because both n and kf are considered to be constant for specific
type of foil bearings, the following scaling law is established for
Scaling Laws for Bearing Clearance the nominal clearance of the foil bearing for different bearing
The load carrying capacity of a foil bearing is the integration of sizes
film pressure over the entire bearing area at a particular speed of
operation R
¼ constant (8)
ð C2
F ¼ pdA ¼ pavg LD (2) From Eq. (8), it is concluded that the nominal bearing clearance
should be proportional to the square root of the bearing radius.
Consider the nondimensional compressible Reynolds equation The result can be used for the selection of initial nominal clear-
applicable to the foil bearing configuration shown in Fig. 1, ance of foil bearings with different sizes if a good reference
design is available or known in prior. For example, if C1 and R1
are clearance and radius of certain well-designed foil bearing,
      clearance of a foil bearing with R2 (but the same type as the refer-
@ 3 @P @ 3 @P @ @
PH þ PH ¼K ðPHÞ þ 2 ðPHÞ enced first bearing) can be chosen as
@h @h @Z @Z @h @s
rffiffiffiffiffi
(3) R2
C2 ¼ C1 (9)
where P ¼ p=pa , H ¼ h=C, and t ¼ s=x have been used for non- R1
dimensionalization. The pressure terms on the left side of Eq. (3)
are of order 2, while the pressure terms on the right side are of
order 1. Therefore, the average bearing pressure is related to the Scaling Laws for Support Structure Stiffness
bearing number defined as K ¼ 6lxR2 =pa C2 according to Eq. (4) Structural support structure of the foil bearings takes various
  2   2 forms such as corrugated bump foils, multileafs, elastomeric
6l 2pN R pl R material, metal meshes, etc. Because of highly nonlinear and com-
pavg / K ¼ ¼N (4)
pa 60 C 5pa C plicated nature of those support structures, they are often modeled
as a series of inertia-less mechanical elements consisting of linear
within practical range of DN number (3.5  106) applicable to spring and viscous damper for simplicity of analysis
modern turbomachinery. Therefore, Eq. (2), can be rewritten as
dd
   fb ¼ pA0 ¼ kb d þ cb (10)
pl R dt
F ¼ LDpavg ¼ kf 0 LD2 N (5)
10pa C2
where A0 ¼ 2pRL=Ngrid is the bearing surface area covered by
one support element (Ngrid is the total number of the inertia-less
linear spring/damper in computational domain), cb is the equiva-
lent viscous damping, and d is the deflection of the structure.
Using nondimensional terms, S ¼ d=C, Eq. (10) becomes
 
dS
P ¼ Kb S þ g (11)
ds

where Kb represents the nondimensional stiffness, and it is found


to be
   
kb C Ngrid C2 kb Ngrid C2 kb
Kb ¼ ¼ ¼ (12)
pa A0 2ppa R CL 4ppa R CR

In Eq. (12), L ¼ D ¼ 2R has been assumed. It can be further


assumed that the same numerical value of Ngrid can be used for
computational modeling of the foil bearing’s performance regard-
less of the physical size of the bearing as long as the bearing type
is identical. Because Kb is the scale-invariant stiffness (constant)
of the support structure, and C2 =R is also constant from Eq. (8), it
Fig. 1 Schematic description of three-pad bump foil bearing is concluded that

042502-2 / Vol. 139, APRIL 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a


kb hoffset . The ratio of these two angles is called the offset ratio
¼ constant (13) c ¼ hoffset =hpad .
CR
The scaling laws are applied to the three-pad foil bearing with
Equation (8) can be used for eliminating the nominal operating various diameters up to 300 mm to investigate whether such a
clearance from Eq. (13) large foil bearing can be designed and adopted to practical indus-
trial applications. The first step is to devise a turbomachinery con-
 
kb kb C figuration that may represent typical single stage gas turbine
¼ constant ) pffiffiffi ¼ constant ) kb / R1:5 rotors as shown in Fig. 2, where turbine impeller is made of Ni
CR CR R
alloy and rest of the system are made of stainless steel. Assuming
(14) L/D ¼ 1, for the nominal bearing size (¼ journal diameter) of D,
the assumptions in Table 1 were applied to estimate the total rotor
From Eq. (14), the support stiffness is related to bearing size as weight of the turbomachinery components. The turbomachinery
 32 system shown in Fig. 2 was created using the assumptions in
kb;1 kb;2 R2 Table 1. By suitable scaling of the solid model, the dimensions of
1:5
¼ 1:5 ) kb;2 ¼ kb;1 (15)
R1 R2 R1 all turbomachinery components were obtained for various bearing
sizes, and the results of thrust runner outer diameter (OD), impel-
The above scaling law for the support structure’s stiffness is ler diameter, and bearing shaft length are tabulated in Table 2.
based on the assumption that Ngrid is constant. Therefore, kb repre- The mass of individual components and the total load was com-
sents stiffness of entire physical structure underneath A0 . Another puted based on the material density of each component. For sim-
scaling law in terms of areal stiffness, i.e., stiffness per unit area plicity, it is assumed that two radial bearings support the half of
ðkb =Ao Þ can also be found. Using the definition of areal stiffness, entire rotor weight equally, and the average bearing pressure was
Eqs. (8) and (13) calculated by dividing the half of the total rotor weight by the
bearing area (D2 ). From Table 2, the average bearing pressure is
  2  
kb Ngrid C kb R proportional to the bearing diameter while the rotor mass
kb areal ¼ ¼ (16) increases with the third power of journal diameter. The solid
2pRL=Ngrid 2p R CR LC
model is given in a form of single-stage gas turbine but the rotor
configuration can represent various other oil-free turbomachinery.
Because the first, second, and third terms in the p
farffiffiffi right expres-
Typical turbo-air blowers use aluminum impellers instead of steel
sion of above equation are all constant, and C / R from Eq. (8)
or Ni alloy impellers, and the motor is located between the two
and L  2R, it is found that
bearings, compensating the weight reduction at the impellers. For
R 1 applications using process gas lubricants, Organic Rankine Cycle
kb areal / / pffiffiffi (17) turbo expanders and refrigerant compressors also use aluminum
LC R impellers instead of steel impellers, and the motor/generator is
located between the two bearings, leading to rather similar rotor
From above finding, p areal
ffiffiffi stiffness should be chosen to be weight as the air blowers. Therefore, it is appropriate to use the
inversely proportional to R. Areal stiffness is more convenient if rotor model in Fig. 2 as a representative system for the applicabil-
continuous structure (metal mesh or elastomeric material) is used. ity study of the scaling laws.
The proposed scaling laws are only for estimating the support The scaling law of support stiffness, Eq. (15), is applicable to
structure stiffness and radial clearance, and do not provide infor- any form of support structure. In this paper, corrugated bump foils
mation about the theoretical limiting load capacity proposed by were chosen as a support structure. For simplicity, it is assumed
Peng and Khonsari [4] of specific bearing, because the theory is
valid only for isothermal assumption. Reynolds equation based on
the continuum theory breaks down once the film thickness
approaches the scale of molecular mean free path (MMFP). The
increase in static loading is accompanied by decrease in film
thickness and corresponding increase in film temperature due to
viscous dissipation. The MMFP increases with increasing temper-
ature. Therefore, the thermal effects cause the Reynolds equation
to break down at a higher film thickness.
In summary, the scaling laws are based on matching the analyti-
cal load capacity with the experimental guideline from Eq. (1).
Therefore, it is important to note that the limitations of the load
capacity guideline also apply to the developed scaling laws.

System Configuration of Interest


The schematic of a three-pad foil bearing is shown in Fig. 1.
Dynamic instability issues arise in small foil bearings at high
speeds when uniform clearance is used. To mitigate instability
issues, nonuniform clearance is incorporated into the foil bearing
design by using the three-pad configuration. A three-pad foil bear-
ing comprises three sets of bump foils and corresponding top foils
assembled within the bearing sleeve such that the journal center
and pad centers do not coincide (nonuniform clearance). The dis-
tance between the journal center and the pad center is defined as
the pad preload offset rp . The set bore clearance can be defined by
considering a circle (set bore circle) tangent to the top foils at the
pivot location. The angular width of the pad is hpad , and the mini- Fig. 2 Turbomachinery configuration of interest: (a) three-
mum set bore film thickness location within the pad is defined as dimensional view and (b) sectional view

Journal of Engineering for Gas Turbines and Power APRIL 2017, Vol. 139 / 042502-3

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a


Table 1 Assumptions for estimating turbomachinery geometry

Shaft Thrust runner Thrust runner Thrust Compressor impeller Shaft


inner diameter inner diameter outer diameter runner thickness outer diameter length

0.66 D 0.66 D 2.2 D 0.22 D 2.75 D 3.5 D

Table 2 Dimensions and mass of turbomachinery components

Journal Thrust runner Impeller Shaft Total rotor Rotor mass Average bearing
OD (mm) OD (mm) OD (mm) length (mm) mass (kg) per bearing (kg) pressure (bar)

20 44.0 55.0 70.0 0.42 0.21 0.050


50 110.0 137.5 175.0 6.30 3.15 0.122
75 165.0 206.3 262.5 21.22 10.61 0.183
100 220.0 275.0 350.0 50.28 25.14 0.243
150 330.0 412.5 525.0 169.68 84.84 0.365
200 440.0 550.0 700.0 402.22 201.11 0.487
250 550.0 687.5 875.0 785.60 392.80 0.608
300 660.0 825.0 1050.0 1357.52 678.76 0.730

that the number of bump corrugations is maintained the same


regardless of the bearing size (i.e., bump pitch is proportional to Table 3 Clearance and bump stiffness from scaling laws
the bearing size). Under the assumption of constant number of
bumps regardless of the bearing size, the stiffness of individual Journal Clearance Bump Areal stiffness
bump for entire axial length of the bearing will be designed fol- OD (mm) (lm) stiffness (MN/m) (GN/m3)
lowing the scaling law. The assumption of constant number of
20 54 1.79 68.33
bumps (pitch scaled to the radius) is a useful tool in the computa- 50 85 7.07 43.22
tional domain which simplifies the development of the scaling 75 104 12.99 35.29
laws. However, adoption of this assumption leads to too much 100 120 20.00 30.56
local sagging of the top foil if the top foil thickness is not altered 150 147 36.74 24.95
according to bearing size. Therefore, the practical hardware 200 170 56.57 21.61
design implementing the scaling law does not have to follow the 250 190 79.06 19.33
same number of bumps. A better implementation of the scaling 300 208 103.92 17.64
law for the support structure is through the scaling law of the areal
stiffness Eq. (17).
To use the scaling laws for the clearance and bump stiffness, it
is important to choose “good” foil bearing as a reference. How- Table 4 Speed data for test cases
ever, it is not easy to choose specific foil bearing as a reference
because all the foil bearings which were successfully integrated to Journal
OD (mm) Nmin Nmax
oil-free turbomachinery were designed for certain purpose tailored
to its application. Another difficulty is that the detailed design fea- 20 35,000 63,000 91,000 119,000 147,000 175,000
ture of the foil bearings is not usually available. The authors have 50 14,000 25,200 36,400 47,600 58,800 70,000
chosen two foil bearing designs which were successfully applied 75 9333 16,800 24,267 31,733 39,200 46,667
to oil free-turbomachinery applications. The first design is 21 mm 100 7000 12,600 18,200 23,800 29,400 35,000
foil bearing developed for 12 kW turbo alternator with rated speed 150 4667 8400 12,133 15,867 19,600 23,333
of 140,000 rpm, and the second design is 102 mm foil bearing 200 3500 6300 9100 11,900 14,700 17,500
applied to 1000–5000 lbf thrust class oil-free turbine simulator 250 2800 5040 7280 9520 11,760 14,000
rotor. Both bearings were developed in the authors’ laboratory 300 2333 4200 6067 7934 9800 11,667
and applied to the intended applications successfully. However,
these two bearings are not exactly the same configuration, i.e., For all the bearing sizes (journal OD), lift off speeds are
number of top foil pads and nondimensional preload are not iden- expected to be lower than the corresponding Nmin because steady-
tical and number of bump foils are not the same either. To apply state solution for pressure and film thickness could be found for
the scaling laws for the clearance and bump stiffness, both foil all the cases. As for reference, the lift off speed of aforementioned
bearings had to be slightly adjusted to use them as common refer- actual 21 mm foil bearing developed for 12 kW turbo alternator
ences. Table 3 was created using the scaling laws with the was measured at around 14,000 rpm for actual rotor mass of
adjusted two bearings as references. 0.22 kg per bearing.
Static and dynamic performances were evaluated for each
design in Table 3 for various speeds. Because they are all different
in size, the maximum DN number of 3.5  106 was applied to find
the upper limits of the speed ðNmax Þ for the various bearing Results and Discussion
diameters. Simulation results for bearings following the proposed scaling
The chosen range of DN number ensures almost linear variation laws are presented in this section for various shaft sizes. The time
of load capacity with speed [5]. The lower bound speeds ðNmin Þ domain analysis involves simultaneous solutions for the unsteady
were assumed to be 20% of the respective Nmax . Therefore, simu- Reynolds equation, the journal orbit, and the foil structural deflec-
lations were conducted for all the cases shown in Table 4. The tions. The mathematical details of the solution procedure follow
simulations yielded stiffness and damping coefficients, eccentric- Ref. [8]. The nondimensional preload of the pad is 0.6 and the
ity, pressures, and film thickness. With the listed clearance and pivot offset is assumed to be 0.5. The model for bump dynamics
radius combinations, maximum bearing number at Nmax is 0.728. uses a structural loss factor of 0.20. All the simulations were

042502-4 / Vol. 139, APRIL 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a


performed at isothermal condition with temperature 50  C and air
viscosity calculated at the same temperature.
The variation of eccentricity with normalized speed (nondimen-
sionalized by Nmax ) for various bearing sizes is shown in Fig. 3.
The pressure and film thickness profiles are shown in Figs. 4–7.
The film thickness normalized by corresponding clearance of
each bearing is shown in Figs. 8 and 9. The peak pressure
increases with bearing size while the minimum film thickness val-
ues decrease with increasing bearing size as tabulated in Table 5.
The synchronous stiffness and damping coefficients are shown in
Figs. 10–13, in which the speeds are normalized with respect to
Nmax for each bearing size. The direct stiffness increases with the
size of the bearing as expected in Figs. 10 and 11. However, the
variations of kxx with normalized speed for the various bearing
sizes show different trends as in Fig. 10. For small bearings, both Fig. 6 Film thickness at the bearing edge for Nmin
kxx and kyy increase with shaft speed. However, for large bearings,
kxx slightly decreases with speed. This phenomenon can be
explained by considering that the total stiffness of the bearing is a
net stiffness of the gas film and the bump structural stiffness in
series. While the total bump structural stiffness remains constant
over a range of speeds for the given bearing size, the film pressure
changes with shaft speed. For 300 mm diameter bearing, for

Fig. 7 Film thickness at the bearing edge for Nmax

Fig. 3 Eccentricity versus normalized speed

Fig. 8 Nondimensional film thickness at bearing edge for Nmin

Fig. 4 Pressure profiles for Nmin

Fig. 9 Nondimensional film thickness at bearing edge for Nmax

example, increase in shaft speed results in increased film thickness


(204% increase from Nmin to Nmax Þ and corresponding reduction
in eccentricity shown in Fig. 3. The increased film thickness
accompanies reduction in the maximum peak film pressure as
shown in Table 5, thereby reducing the film stiffness.
For the smallest bearing, however, gas film thickness increases
Fig. 5 Pressure profiles for Nmax with speed too but the minimum film thickness at Nmax is only

Journal of Engineering for Gas Turbines and Power APRIL 2017, Vol. 139 / 042502-5

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a


Table 5 Maximum pressure and minimum film thickness

Journal OD (mm) Speed (rpm) H_min (lm) P_max (bar) Journal OD (mm) Speed (rpm) H_min (lm) P_max (bar)

35,000 12.9 1.19 4667 9.4 2.24


63,000 15.9 1.21 8400 14.4 2.15
20 91,000 17.4 1.24 150 12,133 18.4 2.11
119,000 18.2 1.27 15,867 21.5 2.08
147,000 18.8 1.29 19,600 24.2 2.07
175,000 19.1 1.32 23,333 26.4 2.06
14,000 12.7 1.43 3500 8.2 2.62
25,200 17.7 1.41 6300 12.8 2.51
50 36,400 20.8 1.42 200 9100 16.5 2.46
47,600 22.9 1.44 11,900 19.6 2.42
58,800 24.4 1.45 14,700 22.3 2.39
70,000 25.4 1.47 17,500 24.5 2.37
9333 11.8 1.64 2800 7.4 2.98
16,800 17.2 1.60 5040 11.4 2.86
75 24,267 20.9 1.59 250 7280 14.9 2.79
31,733 23.6 1.59 9520 17.8 2.75
39,200 25.6 1.6 11,760 20.4 2.71
46,667 27.1 1.61 14,000 22.6 2.69
7000 10.9 1.84 2333 6.8 3.34
12,600 16.3 1.78 4200 10.4 3.2
100 18,200 20.3 1.76 300 6067 13.5 3.13
23,800 23.3 1.75 7934 16.2 3.07
29,400 25.6 1.75 9800 18.6 3.03
35,000 27.5 1.76 11,667 20.7 3.00

Fig. 10 kxx versus normalized speed Fig. 12 dxx versus normalized speed

Fig. 11 kyy versus normalized speed Fig. 13 dyy versus normalized speed

48% more than that at Nmin . Thus, the maximum peak pressure noncircular orbits of the shaft and is beneficial in terms of stability
increases with speed. The bearing stiffness is consequently domi- [9]. The kyy value increases substantially with speed (470%
nated by the gas film resulting in increasing stiffness with speed. increases from Nmin Þ for small sized bearings, whereas the corre-
The kyy values, though smaller, are of the same order of magnitude sponding increase in large sized bearings is much smaller (30%
as kxx for small sized bearings. However, for larger bearings, the increase from Nmin Þ. The stiffness characteristics once again
kyy values are an order of magnitude smaller than the kxx due to depend on speed and loading which affect the film thickness and
higher vertical loading. This anisotropic stiffness results in pressure profiles.

042502-6 / Vol. 139, APRIL 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a


Fig. 14 Stiffness components for 20 mm OD bearing Fig. 17 Damping components for 20 mm OD bearing

Fig. 15 Stiffness components for 100 mm OD bearing Fig. 18 Damping components for 100 mm OD bearing

Fig. 16 Stiffness components for 300 mm OD bearing Fig. 19 Damping components for 300 mm OD bearing

The direct damping coefficients, dxx and dyy , increase with size individual thermal growths are similar. This could be done by
as expected (Figs. 12 and 13). From Fig. 12, it is observed that dxx appropriate selection of materials and fine tuning the design of
decreases with increasing the shaft speed due to stiffening effect bearing sleeve and surrounding structures from a thermal stand-
of gas film. However, as the shaft speed is increased, the direct point. Once the thermal effects are addressed by proper thermal
damping dyy slightly decreases except for the large bearings (over management, the local film thickness becomes a function of only
100 mm), where, dyy shows the maximum values in intermediate the nominal assembly clearance, centrifugal growth, shaft posi-
speeds as shown in Fig. 13. tion, and the bump deflection.
The relative magnitude of cross-coupled stiffness compared to The centrifugal expansion of the hollow shaft under plane stress
the direct stiffness and direct damping is a measure of instability model [11] is given by
of rotor-bearing systems in general [10]. As shown in Figs. 14–19,
2
the direct stiffnesses are of the same order of magnitude as the p2 qðDN Þ
rcg ¼ D ½2:3068  0:5644  (18)
cross-coupled stiffnesses in small sized bearings. However, 28800E
the relative magnitude of cross-coupled stiffness decreases with
the bearing size, which is accompanied by stronger anisotropic when assumptions in Table 1 is applied for the relation between
behavior of the direct stiffnesses as the bearing size increases. inner radius of the shaft and shaft diameter. Assuming the normal
Centrifugal expansion of the shaft and relative thermal expan- operating condition is at constant DN number for all the differ-
sion of the shaft with respect to the bearing decrease the local film ently sized bearings
thickness (and effective clearance) thereby affecting the static and
dynamic performance. In foil bearings, thermal expansion of foil rcg / D (19)
structures is negligible compared to those of bearing journal and
bearing sleeve. The relative thermal growth of the shaft with Therefore, the centrifugal growth of the shaft depends directly on
respect to the bearing sleeve can be maintained very small if their the shaft size. This aspect must be considered in the interpretation

Journal of Engineering for Gas Turbines and Power APRIL 2017, Vol. 139 / 042502-7

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a


of clearance in the scaling law; the clearance in Table 3 should be N¼ shaft speed
interpreted as the clearance at normal operating condition, which OBrg ¼ bearing center
is the difference between the clearance during the assembly at P¼ nondimensional pressure
cold condition and rcg . The smallest value of minimum film thick- pa ¼ atmospheric pressure
ness from Table 5 is 6.8 lm which is small enough to be a chal- pavg ¼ average bearing pressure
lenging issue for frequent start-stop operations and required R¼ bearing radius
tolerances of large bearing components. However, modern rcg ¼ centrifugal growth
machining operations can easily achieve the required precision of rp ¼ preload
the rotor and bearing components. Further, the average bearing z¼ coordinate in axial direction
pressures for the largest bearing is still less than 1 bar as shown in Z¼ nondimensional coordinate in axial direction
Table 2. Therefore, it is relatively easy to achieve full hydrody- c¼ offset ratio ðc ¼ hoffset =hpad Þ
namic operation. In addition, the concept of hybrid foil bearings d¼ deflection of the structure
(hydrostatic injection) would be appropriate for large foil bearings e¼ nondimensional eccentricity ðe ¼ e=CÞ
(beyond 100 mm) to avoid dry rubbing and to increase the film g¼ structural loss factor
thickness during low speeds [12,13]. h¼ angular coordinate
hpad ¼ angular width of the pad
Conclusions and Future Work K¼ bearing number
l¼ dynamic viscosity of fluid
Scaling laws for radial foil bearings were developed using the ¼ Poisson’s ratio
scale invariant Reynolds equation and the empirical NASA guide- n¼ bearing load capacity coefficient
line for load capacity estimation. The first-scaling law of the q¼ density of shaft
clearance shows that the nominal operating clearance has to be s¼ nondimensional time
proportional to the square root of the bearing radius. Similarly, a x¼ angular velocity of shaft
power law relationship between the support structure stiffness and
the bearing radius was derived, and it was found that the support
structure stiffness has to be chosen to be proportional to the bear-
References
ing radius to the power of 1.5.
[1] DellaCorte, C., and Valco, M. J., 2000, “Load Capacity Estimation of Foil Air
The implication of these scaling laws is that the preliminary Journal Bearings for Oil-Free Turbomachinery Applications,” Tribol. Trans.,
support structure stiffness and nominal clearance for radial foil 43(4), pp. 795–801.
bearings can be designed without resorting to detailed calcula- [2] DellaCorte, C., 2011, “Stiffness and Damping Coefficient Estimation of Com-
tions. The results from this work serve as a guideline for foil bear- pliant Surface Gas Bearings for Oil-Free Turbomachinery,” Tribol. Trans.,
54(4), pp. 674–684.
ing designers in quick choice of initial design parameters. [3] Kim, D., and Zimbru, G., 2012, “Start-Stop Characteristics and Thermal Behav-
Detailed design of the clearance and support structure stiffness for ior of a Large Hybrid Airfoil Bearing for Aero-Propulsion Applications,”
specific applications requires high level analyses of the bearing ASME J. Eng. Gas Turbines Power, 134(3), p. 032502.
characteristics in actual operating conditions, and it is beyond the [4] Peng, Z. C., and Khonsari, M. M., 2004, “On the Limiting Load-Carrying
Capacity of Foil Bearings,” ASME J. Tribol., 126(4), pp. 817–818.
scope of this study. In addition, rotordynamic stability of the foil [5] Peng, Z. C., and Khonsari, M. M., 2004, “Hydrodynamic Analysis of Compliant
bearing-rotor systems depends not only on the bearing itself but Foil Bearings With Compressible Air Flow,” ASME J. Tribol., 126(3),
also on detailed inertia distribution of the rotor. Investigation of pp. 542–546.
rotordynamic characteristics of the rotor models following the [6] Radil, K. C., Howard, S. A., and Dykas, B., 2002, “The Role of Radial Clear-
ance on the Performance of Foil Air Bearings,” Tribol. Trans., 45(4),
geometry in Table 2 and Fig. 2 is the scope of future study. pp. 485–490.
[7] Spakovszky, Z. S., and Liu, L. X., 2005, “Scaling Laws for Ultra-Short Hydro-
static Gas Journal Bearings,” ASME J. Vib. Acoust., 127(3), pp. 254–261.
Nomenclature [8] Kim, D., 2007, “Parametric Studies on Static and Dynamic Performance of Air
A0 ¼ bearing surface area per support structure Foil Bearings With Different Top Foil Geometries and Bump Stiffness Distribu-
C¼ nominal operating clearance tions,” ASME J. Tribol., 129(2), pp. 354–364.
[9] Gunter, E. J., and Trumpler, P. R., 1969, “The Influence of Internal Friction on
cb ¼ equivalent viscous damping coefficient the Stability of High Speed Rotors With Anisotropic Supports,” ASME J. Eng.
D¼ diameter of bearing Ind., 91(4), pp. 1105–1113.
e¼ eccentricity [10] Childs, D., 1993, Turbomachinery Rotordynamics: Phenomena, Modeling, and
Analysis, Wiley, New York.
E¼ Young’s modulus [11] Timoshenko, S. P., and Goodier, J. N., 1970, Theory of Elasticity, McGraw-
F¼ load capacity of bearing Hill, New York, pp. 80–83.
fb ¼ pressure force on structure [12] Wang, Y. P., and Kim, D., 2013, “Experimental Identification of Force Coeffi-
H¼ nondimensional film thickness cients of Large Hybrid Air Foil Bearings,” ASME Paper No. GT2013-95765.
[13] Kim, D., and Lee, D., 2010, “Design of Three-Pad Hybrid Air Foil Bearing and
kb ¼ stiffness coefficient of support structure Experimental Investigation on Static Performance at Zero Running Speed,”
L¼ length of the bearing ASME J. Eng. Gas Turbines Power, 132(12), p. 122504.

042502-8 / Vol. 139, APRIL 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/935804/ on 02/20/2017 Terms of Use: http://www.asme.org/a

You might also like