Prediction of Gas Thrust Foil Bearing Performance For Oil-Free Automotive Turbochargers

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Prediction of Gas Thrust

Foil Bearing Performance


for Oil-Free Automotive
Turbochargers
Luis San Andr
es Green technologies are a mandate in a world concerned with saving resources and pro-
Mast-Childs Professor tecting the environment. Oil-free turbocharger (TC) systems for passenger and commer-
Fellow ASME cial vehicles dispense with the lubricant in the internal combustion engine (ICE), hence
Department of Mechanical Engineering, eliminating not just oil coking, but also suppressing nonlinear behavior, instability and
Texas A&M University, excessive noise; all factors to poor reliability and premature mechanical failure. The
College Station, TX 77843 work hereby presented is a stepping stone in a concerted effort toward developing a com-
e-mail: LSanAndres@tamu.edu putational design tool integrating both radial and thrust foil gas bearings for oil-free
automotive TCs. The paper presents the physical analysis and numerical model for pre-
Keun Ryu1 diction of the static and dynamic forced performance of gas thrust foil bearings (GTFBs).
Assistant Professor A laminar flow, thin film flow model governs the generation of hydrodynamic pressure
Department of Mechanical Engineering, and a finite element plate model determines the elastic deformation of a top foil and its
Hanyang University, support bump strip layers. For a specified load, the analysis predicts the minimum gas
Ansan, Gyeonggi-do 426-791, South Korea film thickness, deformation and pressure fields, the drag torque and power loss, and the
e-mail: kryu@hanyang.ac.kr axial stiffness and damping force coefficients, respectively. Open source archival test
data on load capacity and drag torque serves to benchmark some of the model predic-
Paul Diemer tions. Next, predictions are obtained for a GTFB configuration designed for an oil-free
Director of Engineering TC operating at increasing gas temperatures, axial loads, and shaft rotational speeds.
BorgWarner Turbo Systems, The largest drag torque occurs at the highest temperature since the gas viscosity is also
Arden, NC 28704 highest, whereas the largest load determines operation with a minute film thickness that
e-mail: pdiemer@borgwarner.com sets a limit for the manufacturing tolerance. While airborne, the drag friction factor for
the bearing is small, ranging from 0.009 to 0.015, thus demonstrating the advantage of
an air bearing technology over engine oil-lubricated bearings. The synchronous speed
axial stiffness increases with operating speed (and load), whereas the axial damping
coefficient remains nearly invariant. The operating gas temperature plays an insignificant
role on the variation of the force coefficients with frequency, whereas the operating speed
and the ensuing applied thrust load determine the largest changes. The model predicts,
as an excitation frequency (x) increases, a GTFB axial stiffness (Kz) that hardens and a
damping coefficient (Cz) that quickly vanishes. The most important finding is that CzX/Kz
 c ¼ the material loss factor for the bearing. Hence, the success of foil bearing technol-
ogy relies on the selection of a metal underspring structure that offers the largest
mechanical energy dissipation characteristics. [DOI: 10.1115/1.4028389]

1 Introduction time lag response due to lesser drag. Eliminating dependence on


the lubricant of the ICE is a major development that will reduce
Gas foil bearings (GFBs) are compliant surface hydrodynamic
issues of reliability including oil coking and suppression of multi-
bearings using ambient air as the working fluid media. GFBs ena-
ple subsynchronous rotor whirl motions responsible for both
ble microturbomachinery (MTM < 400 kW) to operate at high
excessive noise and premature mechanical failure [5,6]. With an
rotating speed and high temperature with significant reduction in
adequate thermal management, oil-free TCs could achieve higher
drag power losses and increases in system thermal efficiency. Oil-
rotor speeds than units using oil-lubricated (semi) floating ring
free systems have lesser part count, footprint and weight and are
bearing systems; and hence, would lead to smaller power pack-
environmentally friendly with demonstrated savings in long-
ages. Other benefits include higher ICE efficiency and lesser emis-
interval maintenance expenses.
sions to satisfy stricter environmental constraints. These are
GFBs offer distinct advantages over rolling elements bearings
forthright steps toward a green technology.
including no DN value limit, reliable high temperature operation,
Alas, GFBs have demerits of excessive drag and wear of pro-
and large tolerance to debris and rotor motions, including tempo-
tective coatings during rotor startup and shutdown events [7]. In
rary rubbing and misalignment. Current commercial applications
addition, expensive developmental costs and, until recently, inad-
include cryogenic turbo expanders, blowers, and micro gas tur-
equate predictive tools limited the widespread deployment of
bines [1,2]. Applications under development include automotive
GFBs. In particular, at high temperature conditions, reliable oper-
TCs and midsize aircraft gas turbine engines [3,4]. Oil-free TCs
ation of GFB supported rotor systems depends on adequate engi-
for passenger and commercial vehicles will rely on GFBs to pro-
neered thermal management and proven solid lubricants
vide rotor support, radial and axial, with significant reductions in
(coatings) [8].
1
Corresponding author.
Contributed by the Structures and Dynamics Committee of ASME for publication 2 A Brief Review of the Prior Literature
in the JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received
July 10, 2014; final manuscript received July 26, 2014; published online September Bump-type foil bearings, radial and thrust, in use since the
30, 2014. Editor: David Wisler. 1960 s have been applied to a variety of small (oil-free) rotating

Journal of Engineering for Gas Turbines and Power MARCH 2015, Vol. 137 / 032502-1
C 2015 by ASME
Copyright V

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


machinery operating at high speeds. For decades, GFBs remained configuration and materials of tested generation I TFBs and report
as a niche application where oil contamination was not permitted experimental results for the drag torque versus operating speed for
and also for operating conditions at very low/high temperatures increasing applied loads until an ultimate limit denoted by a sud-
where mineral oil lubricants (or liquids) could not survive. Do den increase in drag torque (and dissipated power). There are no
recall that gases have notoriously low viscosities; and hence, a gas experimental data on measured film thickness or foil deformation,
bearing cannot generate a substantial load capacity under hydro- etc. The experimental data reported, see Refs. [16] and [17] for
dynamic operation. In addition, their damping capability is example, evidence a significant degree of variability and lack of
extremely low, vanishing quickly as the surface speed increases or repeatability for the same test TFB configuration. The differences
the excitation frequency increases. are even more significant amongst different builds of the same
Conventional (rigid surface) gas bearings demand of very tight configuration.
clearances and their cost of construction and assembly is exces- There are useful design guidelines for radial foil bearings
sive. The success of compliant surface gas bearings as reliable (bump type) based on an accumulated wealth of empirical evi-
support elements in high speed machinery, being the foil bearing dence [13]. However, similar empirically based guidelines for
one type, relies precisely on their resilience (their surface flexibil- TFBs are yet to appear. In particular, the characterization of TFB
ity) to accommodate shaft excursions (displacements) and their dynamic force coefficients (stiffness and damping) is absent in the
ability to provide a significant measure of mechanical energy dis- literature.
sipation through other mechanisms than viscous damping. Inci- This paper presents a predictive tool to perform the engineering
dentally, tolerances of construction with foil bearings are lesser design of thrust foil bearings, in particular for automotive TC
than those for rigid surface bearings albeit polished surface condi- applications. The paper details the mathematical analysis for the
tions, for the shaft in particular, remain as gas film clearances are flow field in a gas bearing and its integration with a finite element
quite small. structural model for the deformation of the top foil and its under-
From the early 1980 s, the development of bump-type foil bear- spring support due to the acting pressure field. Sections 4 and 5
ings took off with concerted efforts to advance predictive design present comparisons of the model predictions against a few instan-
models to widen the range of applications. Heshmat et al. [9,10] ces of experimental data, and provide a thorough discussion of
published frequently on pushing the limit of load capacity for predictions for a thrust foil bearing designed to enable an oil-free
GFBs-bump type and analyses coupling the gas film pressure field TC. The selected metal underspring structure must offer adequate
to the elastic surface deformation field via a simplified uniform mechanical energy dissipation characteristics.
(underspring) stiffness model. Efforts of Heshmat et al. and domi-
nance in the field remained unchallenged until the mid 1990s
when Iordanoff [11,12] made systematic analysis of the support
bump foils and established true limits of operation at high speeds. 3 A Model for Foil Thrust Bearings
At the time, many designs of GFBs were patented, and hence their Figure 1 depicts a typical TFB with a number of circular sector
wide spread application necessarily restricted. pads facing the rotor collar spinning with angular speed X. As
In the late 1990s, NASA Glenn Research Center established an shown schematically in Fig. 2, each bearing pad comprises of an
oil-free turbomachinery program that focused first on the develop- engineered underspring structure and a top foil. The elastic sub-
ment of coatings and processes for high temperature environments structure typically integrates a single bump-foil type strip layer or
with envisioned applications in TCs and (midsize) gas turbines. a preformed metal structure with specific elastic properties.
Throughout the 2000s, DellaCorte et al. championed the wide- In operation, gas is drawn into the thin film gap between the
spread usage of GFBs through the dissemination of procedures for rotor collar and top foils, and a self-generated hydrodynamic pres-
construction of generation I and II journal foil bearings [13] and sure field reacts to the axial load acting on the rotor. The bearing
first generation thrust foil bearings (TFBs) [14] along with experi- geometry and operating conditions, undersprings’ mechanical
mental verification in crafted test rigs [15]. Only recently, Dick- properties, and the top foils material and coating determine the
man [16] and Stahl [17] report experimental results on the static TFB forced performance, static and dynamic. In turbomachinery,
load and drag torque of a generation I GTFB, as per the design axial loads arise from the uneven pressure distribution on impeller
and construction method in Ref. [14]. The NASA program also wheels, for example. The axial force is thus load dependent,
supported the development of computational engineering predic- increasing dramatically with rotor speed and changes in fluid den-
tive tools for radial foil bearings [18,19]. Dedicated test rigs at sity of the compressed or pumped medium, for example.
Texas A&M University served to benchmark model predictions Figure 2 shows a cross sectional view of a TFB pad, with polar
against measurements. coordinates (r, h, z) to describe a flow region with film thickness
TFBs have not enjoyed the same interest and development as h. Reynolds equation describes the generation of hydrodynamic
radial foil bearings; in particular, with regard to benchmarking
results of predictive tools against test data, in scope very limited
to this date. Current research by Lee et al. at KIST (South Korea)
also encompasses system integration into turbo compressors and
automotive TCs, see Refs. [20–24]. The analyses and experimen-
tal data by Lee et al. evidence an urgency toward developing com-
mercial products for the enormous automotive market in Asia.
For completeness, other current developments in thrust foil
bearings include those of Somaya et al. [25] (Japan) on thrust
bearings with viscoelastic supports, Lee and Kim [26] (U.S.) with
TFBs enhanced by hydrostatic pressurization, Zhou et al. [27]
(China) with a novel TFB punching dimples on the top foil to act
as the underspring element, and Conboy [28] (U.S.) applying
GTFBs to MTM for CO2 compression.
In spite of the numerous predictive models advanced, for exam-
ple, Refs. [9–12,20,21,25–28], there is still little experimental evi-
dence for the static load performance of TFBs that can either
benchmark fully model predictions or lead to optimized bearing
configurations. The open-source NASA publications [14,15] and
other academic theses [16,17] offer some details (not all) on the Fig. 1 Schematic view of a three pad thrust foil bearing

032502-2 / Vol. 137, MARCH 2015 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


Fig. 2 Schematic view of a bump strip layer and top foil in a
TFB. Rotor surface speed 5 Xr.

Fig. 3 Schematic view of top foil supported on linear under-


springs and depiction of a finite element for structural elasticity
pressure (P) in a gas film and under laminar flow and isothermal analysis
conditions
   
@ qh3 @P 1 @ qh3 @P @ @Qx @Qy
 þ Xrh þ r þ ðqhÞ ¼ 0 þ  ðP  Pa Þ ¼ 0
r@h 12l r@h 2 r@r 12l @r @t @x @y
(5)
(1) @Mx @Myx @Myx @My
þ  Qx ¼ 0; þ  Qy ¼ 0
@x @y @x @y
where q ¼ P=<g T and l are the gas density and viscosity, respec-
tively. Note that the model is isothermal in both the gas film and
the surrounding solids. The shear forces (Q) and bending moments (M) are
The film thickness (h) separating the thrust collar from the top    
foil is @w @w
Qx ¼ A55 /x þ ; Qy ¼ A44 /y þ
  @x @y
h
0  h  HT ! hðhÞ ¼ he þ Dh 1  þ wðh;rÞ @/x @/y @/x @/y
HT (2) Mx ¼ D11 þ D12 ; My ¼ D12 þ D22 (6)
@x @y @x @y
HT < h  HP ! hðhÞ ¼ he þ wðh;rÞ  
@/x @/y
Mxy ¼ D66 þ
which superimposes the deformation (w) to the film thickness @y @x
from a rigid surface condition. Above Dh is the taper height.
In general, the top foil deformation (w) is proportional to the Above, the parameters Dij represent the plate rigidity coefficients
pressure difference, and depends on the underspring stiffness (KB) and A44 and A55 are the coefficients for the transverse shear strain
and the top foil rigidity (DTF) [29], i.e., terms [29].
Following Reddy [29], Ref. [30] details the finite element for-
 
wðh;rÞ ¼ f ðPðh;rÞ  Pa Þ; KB ; DTF (3) mulation to reach the structural formulation:
e
kue ¼ f e (7)
with
n oT  e
where u e ¼ w  e /ex /ey ; f e ¼ Q  Me M  e T are vec-
ETF t3TF 1 2
DTF ¼  2
 (4) tors of nodal generalized displacements and forces, respectively,
12 1  TF and ek is the shell stiffness matrix for one element.
The stiffness matrix ks for one structural bump supporting a top
where (ETF,  TF) are the Young’s modulus and Poisson ratio for foil at one of its edges (side with nodes
 2 and 3 in the bottom of
the top foil and tTF is the foil thickness. Note that small changes in Fig. 3) is derived from kl ¼ KB  s0  ler where s0 is the bump
foil thickness (including its coating thickness) can produce sub- pitch, ler is the radial length of the element boundary, and KB is the
stantial changes in the elastic characteristics of the top foil. bump stiffness per unit area, estimated using Iordanoff’s [12] ana-
Figure 3 depicts the circular sector shape top foil and under- lytical expressions for a bump with both sides free, or one end
spring support modeled as a two-dimensional flat shell supported free and the other fixed; both accounting for sliding effects with a
on a distributed set of springs located at every bump pitch. The dry-friction coefficient. When considering the dynamic behavior
figure also shows a four-node, shell finite element (Xe) supported of0 a bump-foil underspring support, a complex stiffness,
on an axially distributed linear spring on one edge. The element is KB ¼ KB ð1 þ icÞ, incorporates an empirical structural loss factor
trapezoidal with radial and circumferential lengths equal to lr and (c) representing the ability of the underspring structure to dissipate
lc ¼ rdh, respectively. The shell (top foil) thickness is ht and the mechanical energy.
underspring at one edge of the element has stiffness per unit radial Adding the bump stiffness matrix ks at the appropriate locations
length kl. At each node, there are three degrees of freedom, a while assembling the shell element stiffness matrices ek leads to
transverse deflection (w) and two rotations, /x and /y. the global system of equations
According to the first-order shear deformation theory for an iso-
tropic elastic (top foil) material [29], the elasticity equations for [
Nem
the top foil element relating the bending moments (Mx, My, Myx) K G UG ¼ F G KG ¼ ½e k þ ks  (8)
and shear forces (Qx, Qy) to the plate elastic deflections are e¼1

Journal of Engineering for Gas Turbines and Power MARCH 2015, Vol. 137 / 032502-3

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


Further processing assigns the known boundary conditions for Table 1 Geometry and physical parameters for thrust foil bear-
the top foil at its fixed edges, w ¼ /x ¼ /y ¼ 0 and reduces the ing in Ref. [16]
degrees of freedom for the problem to
Parameters Magnitude
K G
G ¼ F
 GU (9)
Number of pads, NPAD 6
Outer diameter, Do 0.1016 m
The reduced stiffness matrix K  G is symmetric and positive defi- Inner diameter, Di 0.0510 m
nite, U G is the vector of generalized deflections (transverse dis- Pad arc extent, HP 45 deg
placements and rotations), and F  G is the vector of generalized Pad taper extent, HT 15 deg
forces, mostly pressures acting on the top foil. Pad taper, Dh 0.050 ma
Prior to the analysis coupling the structural deflection (w) to the Top foil material Inconel X-750
thin film gas pressure (P) governed by Eq. (1), the stiffness matrix Top foil thickness, tTF 0.150 mm
Bump-foil material (eight bumps) Inconel X-750
is decomposed using Cholesky’s procedure [31], i.e.,
Bump foil thickness, tBF 0.102 mm
Bump pitch, s0 5.00 mm (average)a
 G ¼ LG LT
K (10) Bump half length, l0 1.60 mma
G
Bump height, hBF 0.500 mma
Numerical solution of Eq. (9), using the decomposition, is quite Friction coefficient (bump and bearing surface), lf 0.10a
fast since it involves two procedures, first a forward substitution Bump stiffness/area, KB 6.44 N/mm3
solving LG x ¼ FG, and next a backward substitution solving Structural loss factor, c 0.20a
LTUG ¼ x. The transverse deflection field (w) is extracted from a
Assumed value based on the authors’ practical experience.
UG and used to update the film thickness for solution of Reynolds
equation within the framework of an iterative scheme.
The pad equilibrium (minimum) film thickness he is determined
from the balance of static forces; that is, the bearing reaction force
(FZo ) opposing the imposed (external) axial load (W). Rotor axial
motions (Dz) at whirl frequency (x) produce a TFB reaction force
(DFZ) equal to the integration of the pressure field on the pad sur-
face, i.e.,
N
Xpads þ

FZ ¼ ðP  Pa Þk rdrdh ¼ FZO  ZZ Dzeixt (11)


k

pffiffiffiffiffiffiffi
Above, i ¼ 1 is the imaginary unit and ZZ is a mechanical im-
pedance; its real part revealing the TFB stiffness, and its imagi-
nary part the TFB damping coefficient,2 i.e.,
Fig. 4 Measurements and current predictions. Test data taken
Zz ¼ Kz þ ixCz (12)
from Ref. [16]. Drag torque versus shaft speed for applied
load 5 40 N. TFB geometry in Table 1.
The force coefficients are frequency dependent due to the gas
compressibility and the flexibility of the bearing underspring
structure. magnitudes. The bearing configuration and materials, as well as
Presently, the exact advection model in Ref. [32] is adopted to the experimental results, are open source. Hence, the test data
solve the partial differential equations for the pressure fields have become the de facto standard for benchmarking model pre-
(P0, Pz) in the gas film. The control volume method ensures dictions. Note that Dickman does not release all information to
numerically stable and accurate solutions at arbitrary operating manufacture the bearing components nor the bearings’ assembly
conditions, including those with large speeds and minute film and tolerances nor states the experimental uncertainty except for
thicknesses (K ! 1). casual observations on repeatability.
The numerical procedure is two-fold. First find, at steady-state, The experimental procedure called for increasing the axial
the operating minimum film thickness that generates the hydrody- applied load on the bearing until a sudden raise in the drag torque
namic pressure field reacting to an imposed axial thrust load; pre- becomes apparent. Dickman names this load as the ultimate load
dictions of drag power loss and mechanical deformation of the top capacity and reports a 30% variability amongst the seemingly
foils and underspring structure follow. Second, find for a multi- identical TFBs facing the same shaft collar (runner) and a 25%
tude of frequencies the perturbed pressure field (Pz) that generates variability with one bearing running against different shaft collars,
the bearing force coefficients, namely, stiffness and damping, coated and uncoated.
which largely determine the thrust bearing resilience and dynamic Figure 4 depicts predicted and measured drag torque versus
stability. shaft speed to 40 krpm for an applied load equaling 40 N (W/Area-
TB ¼ 0.06 bar (0.95 psi)). For this small load condition, the drag
torque predictions are in agreement with the test data for increas-
ing shaft speeds in the region where the TFB has already lifted to
4 Model Validation: Predictions Versus Test Data establish a gas film. The graph also includes a measured large tor-
in Ref. [16] que due to dry-friction prior to the bearing lifting off at 5 krpm.
Dickman [16] presents experimental results for three identical Note that the bearing speed number,3 K ¼ ð6lX=Pa ÞðRO =he Þ2 ,
TFBs operating with shaft speeds as high as 40 krpm and under ranges from 29 to 17; thus denoting operation with moderate gas
increasing loads. Table 1 lists the geometry and known material compressibility. The shear flow Reynolds number, Re
properties of the TFB with assumed dimensions and physical ¼ ðqXRO he =lÞ, ranges from 12 to 359 as the shaft speed increases

2
The description is brief and incomplete. A perturbation analysis is conducted to
3
find Reynolds equations for the equilibrium pressure field (P0) and a complex K  1 denotes a regime of operation dominated by fluid compressibility effects,
perturbed pressure field (Pz). These PDES are coupled to the structure FE Eq. (9). while ac > 1 signifies a bearing with a stiff under-spring layer (hard bumps).

032502-4 / Vol. 137, MARCH 2015 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


from 5 krpm to 40 krpm, hence denoting operation in the laminar
flow regime.
Figure 5 shows comparisons of predicted and measured drag
torque versus applied load operating at a constant shaft speed of
21 krpm. The graph shows good agreement between the predicted
and measured torques for small to moderate loads 120 N (W/
Area ¼ 0.18 bar (2.85 psi)). As the load increases the calculated
torque is lesser than the measured drag torque. The largest differ-
ence is at the highest applied load; which in the tests caused a sud-
den increase in torque due to rubbing contact between the top foil
and the shaft collar.
To provide rationale for the divergence between the model
results and the test data, Fig. 6 shows the predicted minimum film
thickness and maximum top foil deformation versus applied load,
Fig. 5 Measurements and current model predictions. Test data and Fig. 7 presents contour surfaces for the film thickness, gas
taken from Ref. [16]. Drag torque versus applied load at 21 film pressure and top foil deformation for two loads, 50 N and
krpm. TFB geometry in Table 1. 181 N (W/AreaTB ¼ 0.29 bar (4.33 psi)). As the load increases,
the minimum film thickness decreases exponentially, whereas
the (max.) top foil deformation increases linearly. For the range
of loads noted (50 N–180 N), the bearing speed number,
K ¼ ð6lX=Pa ÞðRO =he Þ2 , increases from 24 to 696, while the bear-
ing compliance factor aC ¼ Pa/(heKB) increases from 0.98 to 5.25.
The large K and ac at the highest load denote an operating regime
with large (gas) compressibility effects and with a very compliant
underspring structure. Incidentally, the shear flow Reynolds num-
ber Re ¼ ðqXRO he =lÞ decreases from 114 toward 21 as the load
increases from 51 N toward 180 N, thereby evidencing character-
istic operation in the laminar flow regime.
Note in Fig. 7 the pressure fields are very different; the low
load condition (50 N) generates a profile with its peak magnitude
at the location where the top foil taper ends and then steadily
drops toward ambient. This pressure field is typical of a bearing
operating with a low speed number, K ¼ 24, i.e., with small fluid
compressibility effects. On the other hand, at the large load of
Fig. 6 Predicted minimum film thickness and maximum top 180 N, the minimum film thickness equals  3 lm and K ¼ 696 (a
foil deformation versus applied load for operation at 21 krpm.
TFB geometry in Table 1.
large magnitude); and hence, the pressure profile is uniform
downstream of the taper section with an abrupt drop toward

Fig. 7 Contours of predicted film thickness, gas film pressure, and top foil deformation for
operation at 21 krpm and two applied loads: (a) 50 N and (b) 180 N. TFB geometry in Table 1.

Journal of Engineering for Gas Turbines and Power MARCH 2015, Vol. 137 / 032502-5

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


ambient at the pad trailing edge. The top foil deformation is also
large in the downstream section of the bearing pad and shows dips
or sags in between the location of the bump strip layers.
Recall that the ultimate load in the experiments is Wult  181 N
(Wult/AreaTB ¼ 0.29 bar (4.33 psi)) that caused a sudden increase
in drag torque (see Fig. 5). It is not surprising there should be a
large difference between the measured and predicted torque at the
largest loads since the (predicted) film thickness is so small that
surface roughness effects must play an important role in the gen-
eration of drag torque (and dissipation power).
For completeness, Fig. 8 depicts, as load increases, the pre-
dicted synchronous speed stiffness and damping force coeffi-
cients, and the ratio of force coefficients (CzX/Kz) to realize the
character of the damping coefficient is of structural type. Recall
that the loss factor used to obtain the predictions is c ¼ 0.2. Hence,
the TFB damping is proportional to its stiffness and inversely pro- Fig. 9 Axial load (normalized with respect to maximum load)
portional to frequency. Both the stiffness and damping coefficients acting on GTFB versus shaft speed (normalized with respect to
(Kz, Cz) increase with load since the film thickness decreases. It is maximum shaft speed)
interesting to note that the ratio (CzX/Kz) decreases steadily with
applied load toward c ¼ 0.2. Thus, it becomes apparent that at
operating with either a large load or at a large operating speed
(not shown for brevity), the ratio Cz X=Kz ! c.

5 Prediction of Performance for a Thrust Foil


Bearing in an Oil-Free Turbocharger
Predictions are obtained for the static and dynamic forced per-
formance of an envisioned TFB for small size oil-free TCs. The
bearing, made of Stainless Steel, has inner and outer diameters
equal to 26 mm and 54 mm, respectively. The underspring struc-
tural design follows current practice, see Ref. [14] for example.
The rotor speed is as large as 210 krpm and the TFB must operate
with gas temperatures up to 250 C. The TFB configuration is

Fig. 10 GTFB for oil-free TC: minimum film thickness and max-
imum foil deformation versus axial load (and speed) for opera-
tion at 21 C, 140 C, and 250 C. (a) Minimum film thickness
normalized with respect to minimum film thickness at maximum
shaft speed. (b) Maximum foil deformation normalized with
respect to minimum film thickens at maximum shaft speed.

proprietary; hence, complete information on its geometry and ma-


terial properties cannot be listed. Note that the loss factor c ¼ 0.32
is used in all predictions below. The chosen magnitude for c is
based on extensive empirical evidence gathered from radial foil
bearings; see Ref. [33] for example.
Figure 9 depicts the thrust load acting on the TFB versus shaft
speed. The loads and speeds shown are normalized with respect to
Fig. 8 Predictions versus applied load for speed 21 krpm: (a) the largest acting axial load and highest shaft speed, respectively.
synchronous speed stiffness and damping coefficients and (b) The load shown is determined from the balance of thrust loads
ratio of force coefficients. TFB geometry in Table 1. generated in the back of the turbine and compressor wheels.

032502-6 / Vol. 137, MARCH 2015 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


Fig. 11 GTFB for oil-free TC: drag power loss and friction fac-
tor versus shaft speed (and load) for operation at gas tempera-
tures of 21 C, 140 C, and 250 C. (a) Drag power loss
normalized with respect to maximum power loss. (b) Friction
factor.

Predictions of TFB performance follow for operation with air at


three temperatures: 21 C, 140 C, and 250 C, that span condi-
tions of operation with a cold and hot ICE. As the shaft speed and
thrust load increase, Fig. 10 shows that the (normalized) minimum
film thickness decreases while the maximum elastic deformation
of the top foil increases linearly.
Temperature has a negligible on the deformation and a minor
effect on the film thickness. Operation at a higher temperature
leads to a larger film thickness since the gas viscosity also
increases with temperature. However, note that at the highest
speed and load condition, the minimum film thickness is very
small and hence, surface roughness of the shaft and top foil must
play a dominant role on the performance of the TFB. Note that for Fig. 12 GTFB for oil-free TC: synchronous speed stiffness and
the range of shaft speeds, thrust loads, and gas temperatures con- damping coefficients and (CzX/Kz) versus axial load (and
sidered, the bearing speed number (K) and the shear flow Reyn- speed) for operation at gas temperatures of 21 C, 140 C, and
250 C. (a) Stiffness at synchronous speed. (b) Damping at syn-
olds number (Re) range from 44 to 1251 and from 233 to 66,
chronous speed. (c) CzX/Kz.
respectively.
Needless to say, the high load and high speed operation could
lead to thermal runaway, sudden intermittent contact with the Figure 12 depicts, as the applied load increases with shaft
thrust collar, and likely imminent failure. In addition, for such speed, the normalized synchronous speed stiffness and damping
small film thickness, an adequate installation and alignment of the coefficients and the ratio (CzX/Kz) for operation at three gas tem-
TB faces (top foil) with the thrust collar is crucial. peratures. Note that the stiffness increases linearly with applied
Figure 11(a) shows the drag power (To  X) increasing with load, a typical condition for a soft TFB4 (ac < 1.0).
shaft speed (and load) and with gas temperature (higher viscosity). In Fig. 12(b), the bearing damping coefficient shows a peculiar
A friction factor f ¼ To =WRm equals the drag torque (To) divided behavior, most affected by temperature at the lowest load (and
by the product of the applied load (W) times the mean radius
(Rm ¼ 1=2 (Ri þ Ro)). Figure 11(b) depicts a rather small friction
factor f 2 ½0:009; 0:014 that makes the gas TFB an extremely 4
The bearing compliance factor (ac) increases from 0.12 (at the highest
attractive support for an oil-free TC; i.e., nearly frictionless. temperature and lowest load) to 0.50 (at the lowest temperature and highest load).

Journal of Engineering for Gas Turbines and Power MARCH 2015, Vol. 137 / 032502-7

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


Fig. 13 GTFB for oil-free TC: axial stiffness versus frequency
Fig. 15 GTFB for oil-free TC: ratio of force coefficients (Czx/
ratio for operation at three rotor speeds and gas temperature of
Kz) versus frequency ratio (x/X) for operation at three rotor
250 C
speeds and gas temperature of 250 C

modeled simply with a (single) material hysteresis coefficient,


largely determines the magnitude of the damping coefficient.
Well designed foil bearings must enable an effective dissipation
of mechanical energy dissipation. However, practice shows the
loss coefficient (c) is most difficult to characterize empirically and
(probably) degrades with environmental conditions; temperature,
humidity and atmospheric composition, in particular.
Figure 15 depicts the ratio (Czx/Kz) as frequency increases for
operation at three increasing shaft speeds. For all conditions, the
ratio of force coefficients is lesser than c ¼ 0.32; showing peak
magnitudes for excitation frequencies coinciding with the shaft
rotational speed, (x=X  1). The ratio (Czx/Kz) is lowest for the
highest shaft speed (and applied load) since the speed number (K)
is quite large. Recall K  1 denotes an operating region where
gas compressibility is paramount. Incidentally, note that (Czx/Kz)
for the condition with c ¼ 0 (no structure material hysteresis) is
Fig. 14 GTFB for oil-free TC: axial damping versus frequency the lowest amongst all the predictions, 50% of the predicted
ratio for operation at three rotor speeds and gas temperature of
250 C
magnitudes when using c ¼ 0.32.

shaft speed) condition, while at high speed the damping coeffi- 6 Conclusions
cient is the largest. The ratio (CzX/Kz) shows a near constant The paper presents a physical model for prediction of the static
trend, independent of gas temperature or load condition. Of and dynamic forced response of GTFBs. A laminar flow, thin film
course, this means that the structural damping commands the dis- flow model governs the generation of hydrodynamic pressure and
sipation characteristics of the TFB. a FE plate model determines the elastic deformation of a top foil
Figures 13 and 14 show the variation of the normalized TFB and the under spring support, namely, bump strip layers. A pertur-
stiffness and damping coefficients versus excitation frequency bation analysis produces zeroth and first-order equations for pre-
ratio (x=X) for operation at three shaft speeds (45%, 70%, and diction of the GTFB static load and drag torque, and the axial
90% of top speed) and at gas temperature of 250 C. Predictions stiffness and damping force coefficients, respectively. The force
with similar trends are obtained for operation at lower tempera- coefficients are frequency dependent due to the fluid compressibil-
tures and not shown for brevity. Recall the applied load increases ity, the flexibility of the top foil, and the resilience of the bump
with shaft speed, and hence, the stiffness coefficients correspond- strips support structure.
ingly increase. As frequency increases, the TFB stiffness hardens Predictions from the model are in very good agreement with
(increases) by 50% for the low speed (low load) condition, and the test data for an open-source TFB [16]. Other predictions, not
by just 16% for the high speed (high load). In general, the hot shown for simplicity, correlate well with numerical results from
TFB has a lesser stiffness than the cold bearing since it operates other models in Refs. [20] and [25]. Even though claiming to be
with a smaller film thickness; see Fig. 10(a). open source, Refs. [16] and [17] do not deliver enough informa-
As depicted in Fig. 14, the variation of the TFB axial damping tion to fully model the tested foil bearings. The experimental data
coefficient with frequency is more complicated. The predictions are limited to load capacity and drag torque.
are obtained with a material loss factor c ¼ 0.32. The graphs show Nonetheless, predictions show other results including the evolu-
Cz in logarithmic scale to make less dramatic their drop as fre- tion of the pressure and deformation fields as rotor speed and/or
quency increases. The damping coefficient is largest for the low load increase, for example. In addition, the model predicts, as an
speed (low load) condition. The figures include a prediction of excitation frequency increases, a TFB axial stiffness (Kz) that
damping coefficients without accounting for the structural damp- hardens and an axial damping coefficient (Cz) that decreases rap-
ing effect, i.e., c ¼ 0. The results shown correspond to 90% of idly. The most important finding is related to ascertaining that
max. shaft speed (normalized axial load ¼ 0.76). Note how small Cz x=Kz  c.
is Cz, just a fraction at a frequency ratio (x=X) ¼ 1 and vanishing Predictions for a GTFB designed for use in an oil-free TC
quickly as the frequency increases. Hence, as is well known with application account for the operating temperature range and
foil bearings, the dry-friction between the bumps and the bearing increasing axial loads in the operating speed range. The largest
support and between the top foil and the bumps’ crests, both load determines operation with a very small film thickness,

032502-8 / Vol. 137, MARCH 2015 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


probably in the limit of operation for a well designed thrust bear- T¼ temperature (K)
ing. The GTFB friction factor (f) is approximately constant, To ¼ drag torque (N
m)
f  0.009 to 0.015, over the whole speed range. The largest drag tTF, tBF ¼ thickness of top foil and bump foil (m)
occurs at the highest temperature since the gas viscosity is also w¼ top foil axial deformation (m)
highest. The (synchronous speed: x ¼ X) axial stiffness (Kz) W¼ applied load (N)
increases with operating speed (and load), whereas the axial x, y, z ¼ coordinate system on plane of bearing (m)
damping coefficient (Cz) changes little. The TFB stiffness and X,Y ¼ inertial coordinate system (m)
damping coefficients vary in their expected form as the excitation Zz ¼ Kz þ i xCz, bearing impedance (N/m)
frequency increases. The operating (gas) temperature plays an in- ac ¼ ððPa =he Þð1=KB ÞÞ, bearing compliance coefficient
significant role on the variation of the force coefficients with exci- c¼ material structural loss factor
tation frequency. On the other hand, the operating speed and the Dh ¼ taper height (m)
ensuing applied thrust load determine the largest change in the h¼ circumferential coordinate (rad)
TFB force coefficients. HP ¼ top foil (pad) arc extent (rad)
As is well known, the appropriate characterization of the me- HT ¼ top foil (pad) tapered arc extent (rad)
chanical energy dissipation in the sliding dry-friction between the K¼ ð6lX=Pa ÞððRo =he ÞÞ2 , bearing speed number
bump foils and its foundation structure remains as a major l¼ gas viscosity (Pa
s)
unknown. Material or hysteretic damping is known to represent lf ¼ dry-friction coefficient (bump and its support)
best measured results. However, this parameter (c) depends not ¼ (material) Poisson ratio
only on the material properties and geometry but also on the oper- q¼ P=<g T, gas density (kg/m3)
ating conditions, including atmospheric conditions, as well as the /x, /y ¼ top foil angular deflections (rad)
current state of the surfaces, inevitable degrading after extended x¼ excitation frequency (rad/s)
periods of operation. Reliable loss factors are a must to ensure the X¼ shaft collar rotational speed (rad/s)
adequate performance of GTFBs. These coefficients can only be
determined experimentally.
Presently, the computational model for TFBs is used in
Matrices and Vectors
e
conjunction with the thermo hydrodynamic model for radial foil k¼ FE top foil element stiffness matrix
bearings detailed in Refs. [18] and [30]. Hence, the work accom- G ¼
F G, F global force vector, full and reduced
plished is a concerted effort toward developing a tool integrating G ¼
KG , K FE global stiffness matrix; full and reduced
both radial and thrust foil bearings for the design, prototyping and ks ¼ underspring element stiffness matrix
troubleshooting of oil-free automotive TCs. LG ¼ lower triangular stiffness matrix
G ¼
UG , U generalized displacement vector
Acknowledgment
The research is sponsored by BorgWarner Turbo Systems.
Acronyms
Thanks to Mr. Paul Anschel at BorgWarner Turbo Systems for his FE ¼ finite element
interest and support. GFB ¼ gas foil bearing
GTFB ¼ gas thrust foil bearing
ICE ¼ internal combustion engine
KIST ¼ Korea Institute of Science and Technology
Nomenclature MTM ¼ microturbomachinery
  NASA ¼ National Aeronautics and Space Administration
AreaTB ¼ ð1=8ÞNpad HP D2o  D2i , area of thrust bearing
2
(m ) TC ¼ turbocharger
Cz ¼ TFB axial damping coefficient (N
s/m) References
Do, Di ¼ top foil outer and inner diameters (m) D ¼ 2  R [1] DellaCorte, C., and Pinkus, O., 2000, “Tribological Limitations in Gas Turbine
DTF ¼ ETF t3TF =12ð1  TF 2
Þ, top foil rigidity (N m) Engines: A Workshop to Identify the Challenges and Set Future Directions,”
E ¼ Young’s modulus (Pa) NASA Glenn Research Center, Cleveland, OH, Report No. NASA/TM-2000-
f ¼ ðTo =WRm Þ, drag friction coefficient 210059/REV1.
[2] Chen, H. M., Howarth, R., Geren, B., Theilacker, J. C., and Soyars, W. M.,
Fz ¼ bearing axial reaction force (N) 2000, “Application of Foil Bearings to Helium Turbocompressor,” 30th Turbo-
h, he ¼ film thickness, minimum film thickness (m) machinery Symposium, Turbomachinery Laboratory, Texas A&M University,
hBF ¼ p bump
ffiffiffiffiffiffiffi height (m) Houston, TX, pp. 103–113.
i ¼ 1, imaginary unit [3] Heshmat, H., and Walton, J. F., 2000, “Oil-Free Turbocharger Demonstration
Paves Way to Gas Turbine Engine Applications,” ASME Paper No. 2000-GT-
KB ¼ underspring stiffness per unit area (N/m3) 0620.
KB0 ¼ KB ð1 þ icÞ, complex stiffness per unit area (N/m3) [4] Walton, J. F., Heshmat, H., and Tomaszewki, M. J., 2004, “Testing of a Small
Kz ¼ TFB axial  stiffness
 coefficient (N/m) Turbocharger/Turbojet Sized Simulator Rotor Supported on Foil Bearings,”
kl ¼ KB  s0  ler (N/m) ASME Paper No. GT2004-53647.
[5] Klaus Wolff, K., Steffens, C., Aymanns, R., Stohr, R., and Pischinger, S., 2008,
lc ¼ top foil FE circumferential length (m) “Turbo Charger Noise—Development of a Methodology for the Acoustic Turbo
lr ¼ top foil FE radial length (m) Charger Layout,” 2008 JSAE Annual Congress, Yokohama, Japan, May 21–23,
l0 ¼ bump half length (m) Paper No. 20085246.
Mx, My, Mxy ¼ bending moments in top foil per unit length (N) [6] Polichronis, D., Evaggelos, R., Alcibiades, G., Elias, G., and Apostolos, P.,
2013, “Turbocharger Lubrication—Lubricant Behavior and Factors That Cause
NPAD ¼ number of pads in bearing Turbocharger Failure,” Int. J. Autom. Eng. Technol., 2(1), pp. 40–54, available
P ¼ gas pressure in film (Pa) at: http://www.academicpaper.org/index.php/IJAET/article/view/33/pdf
Pa ¼ ambient pressure (Pa) [7] Dellacorte, C., Lukaszewicz, V., Valco, M. J., Radil, K. C., and Heshmat, H.,
Pz ¼ first-order (perturbed) pressure (Pa/m) 2000, “Performance and Durability of High Temperature Foil Air Bearings for
Oil-Free Turbomachinery,” STLE Tribol. Trans., 43(4), pp. 774–780.
P0 ¼ equilibrium pressure (Pa) [8] Heshmat, H., Tomaszewski, M. J., and Walton, J. F., 2006, “Small Gas Turbine
Qx, Qy ¼ shear forces in top foil per unit length (N/m) Engine Operating With High-Temperature Foil Bearings,” ASME Paper No.
r ¼ radial coordinate (m) GT2006-90791.
<g ¼ gas constant (J/kg C) [9] Heshmat, H., Walowit, J., and Pinkus, O., 1983, “Analysis of Gas-Lubricated
Compliant Thrust Bearings,” ASME J. Lubr. Technol., 105(4), pp. 638–646.
Ro, Ri ¼ top foil outer and inner radius (m) [10] Heshmat, C. A., Xu, D. S., and Heshmat, H., 2000, “Analysis of Gas Lubricated
s0 ¼ bump pitch (m) Foil Thrust Bearings Using Coupled Finite Element and Finite Difference
t ¼ time (s) Methods,” ASME J. Tribol., 122(1), pp. 199–204.

Journal of Engineering for Gas Turbines and Power MARCH 2015, Vol. 137 / 032502-9

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms


[11] Iordanoff, I., 1998, “Maximum Load Capacity Profiles for Gas Thrust Bearings [22] Kim, T. H., Lee, Y.-B., Kim, T. Y., and Jeong, K. H., 2011, “Rotordynamic
Working Under High Compressibility Number Conditions,” ASME J. Tribol., Performance of an Oil-Free Turbo Blower Focusing on Load Capacity of Gas
120(3), pp. 571–576. Foil Thrust Bearings,” ASME J. Gas Turbines Power, 134(2), p. 022501.
[12] Iordanoff, I., 1999, “Analysis of an Aerodynamic Compliant Foil Thrust Bear- [23] Lee, Y.-B., Park, D.-J., Kim, T. H., and Sim, K., 2012, “Development and Per-
ing: Method for a Rapid Design,” ASME J. Tribol., 121(4), pp. 816–822. formance Measurement of Oil-Free Turbocharger Supported on Gas Foil
[13] DellaCorte, C., Radil, K. C., Bruckner, R. J., and Howard, S. A., 2008, “Design, Bearings,” ASME J. Gas Turbines Power, 134(3), p. 032506.
Fabrication, and Performance of Open Source Generation I and II Compliant [24] Sim, K., Kwon, S. B., Kim, T. H., and Lee, Y.-B., 2013, “Feasibility Study of
Hydrodynamic Gas Foil Bearings,” STLE Tribol. Trans., 51(3), pp. 254–264. an Oil-Free Turbocharger Supported on Gas Foil Bearings Via On-Road Tests
[14] Dykas, B., Bruckner, R., DellaCorte, C., Edmonds, B., and Prahl, J., 2009, of a 2-Liter Class Diesel Vehicle,” ASME J. Gas Turbines Power, 135(5), pp.
“Design, Fabrication, and Performance of Foil Gas Thrust Bearings for Micro- 052701.
turbomachinery Applications,” ASME J. Gas Turbines Power, 131(1), [25] Somaya, K., Yoshimoto, S., and Miyatake, M., 2009, “Load Capacity of Aero-
p. 012301. dynamic Foil Thrust Bearings Supported by Viscoelastic Material,” Proc. Inst.
[15] Dykas, B. D., and Tellier, D. W., 2008, “A Foil Thrust Bearing Test Rig for Mech. Eng., Part J, 223(4), pp. 645–652.
Evaluation of High Temperature Performance and Durability,” Army Research [26] Lee, D., and Kim, D., 2011, “Design and Performance of Hybrid Air Foil Thrust
Laboratory, Adelphi, MD, Report No. ARL-MR-0692. Bearing,” ASME J. Gas Turbines Power, 133(4), p. 042501.
[16] Dickman, J. R., 2010, “An Investigation of Gas Foil Thrust Bearing Perform- [27] Zhou, Q., You, H., and Chen, R., 2012, “Development of Foil Thrust Bearings
ance and Its Influencing Factors,” M.S. thesis, Case Western Reserve Univer- With Simple Structure for Micro Turbines,” Adv. Mater. Res., 368–373, pp.
sity, Cleveland, OH. 1392–1395.
[17] Stahl, B. J., 2012, “Thermal Stability and Performance of Foil Thrust [28] Conboy, T. M., 2013, “Real-Gas Effects in Foil Thrust Bearings Operating in
Bearings,” M.S. thesis, Case Western Reserve University, Cleveland, OH. the Turbulent Regime,” ASME J. Tribol., 135(3), p. 031703.
[18] San Andres, L., and Kim, T. H., 2010, “Thermohydrodynamic Analysis of [29] Reddy, J. N., 1993, An Introduction to the Finite Element Method, McGraw-
Bump Type Gas Foil Bearings: A Model Anchored to Test Data,” ASME J. Gas Hill, Inc., New York, Chap. 12.
Turbines Power, 132(4), p. 042504. [30] San Andres, L., and Kim, T. H., 2009, “Analysis of Gas Foil Bearings Integrat-
[19] San Andres, L., Ryu, K., and Kim, T. H., 2011, “Thermal Management and ing FE Top Foil Models,” Tribol. Int., 42(1), pp. 111–120.
Rotordynamic Performance of a Hot Rotor-Gas Foil Bearings System—Part II: [31] Press, W. H., Flannery, B. P., Teukolsky, S. A., and Vetterling, W. T., 1992,
Predictions Versus Test Data,” ASME J. Gas Turbines Power, 133(6), Numerical Recipes in FORTRAN 77: The Art of Scientific Computing, Cam-
p. 062502. bridge University Press, Cambridge, UK, pp. 89–91.
[20] Park, D.-J., Kim, C.-H., Jang, G.-H., and Lee, Y.-B., 2008, “Theoretical Con- [32] Faria, M., and San Andres, L., 2000, “On the Numerical Modeling of High
siderations of Static and Dynamic Characteristics of Air Foil Thrust Bearing Speed Hydrodynamic Gas Bearings,” ASME J. Tribol., 122(1), pp. 124–130.
With Tilt and Slip Flow,” Tribol. Int., 41(4), pp. 282–295. [33] San Andres, L., Ryu, K., and Kim, T. H., 2011, “Identification of Structural
[21] Lee, Y.-B., Kim, T. Y., Kim, C. H., and Kim, T. H., 2011, “Thrust Bump Air Stiffness and Energy Dissipation Parameters in a Second Generation Foil Bear-
Foil Bearings With Variable Axial Load: Theoretical Predictions and ing: Effect of Shaft Temperature,” ASME J. Gas Turbines Power, 133(3),
Experiments,” STLE Tribol. Trans., 54(6), pp. 902–910. p. 032501.

032502-10 / Vol. 137, MARCH 2015 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 04/29/2015 Terms of Use: http://asme.org/terms

You might also like