Stability analysis of hydrodynamic bearing with herringbone grooved sleeve

Shih-Kang Chen, Hsien-Chin Chou, Yuan Kang

n
Department of Mechanical Engineering, Chung Yuan Christian University, Chung-Li 320, Taiwan, ROC
a r t i c l e i n f o
Article history:
Received 12 July 2011
Received in revised form
18 May 2012
Accepted 21 May 2012
Available online 5 June 2012
Keywords:
Hydrodynamic journal micro-bearing
Reynolds equation
Finite difference method
Stability analysis
a b s t r a c t
This study presents a numerical method to investigate the stability analysis of a hydrodynamic journal
micro-bearing. The governing dimensionless Reynolds equation is solved by using the nite difference
method. The Greens theorem is utilized to deal with the discontinuity of the oil-lm thickness, and the
Gaussian elimination method is used to solve the simultaneous discrete equations. Optimal values for
various design parameters are obtained to maximize the load capacity and to improve the stability
characteristics. The accomplishments of this study will help the designers who deal with the grooved
micro-bearing to select the design parameters to satisfy the required characteristics.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction
The herringbone grooved journal bearing (HGJB) is designed
for micro-spindle in order to improve the lubrication character-
istics and it gradually replaces the early design of plain journal
bearing (PJB) used in the storage media, such as CD, VCD, and
DVD. However the bearing characteristics and spindle stability
are always worse than those of a PJB except the HGJB operates
under low eccentricity conditions. This deterioration is caused by
the saw-toothed pressure distribution generated by the grooves
located on the internal surfaces of a HGJB.
The narrow groove theory (NGT) was used to analyze the HGJB
by early researchers; it is assumed that the number of grooves in
the bearing is innite and the pressure distribution induced by
the grooves is simplied as the mean value of the saw-toothed
pressure distribution, which is higher than that in a PJB. Hirs [1]
applied the NGT to obtain the pressure distribution, friction
moment, and stability of a HGJB. Bootsma [2,3] investigated load
capacity and stability of a HGJB for different shapes of spiral
grooves including spherical and conical proles by the NGT. They
both compared the analysis results with experimental data.
Kang et al. [4] applied the nite difference method with the
help of conservation of mass, Gr umbel boundary conditions, and
coordinate transformation to analyze the static and dynamic
characteristics of a HGJB. The results showed that the stability of
a PJB at low eccentricity can be improved by applying either
rectangular or circular grooves. Jang and Yoon [5,6] analyzed the
difference of static and dynamic characteristics between a grooved
bearing with plain journal (GBPJ) and a plain bearing with grooved
journal (PBGJ); the results revealed that the whirl vibration can be
induced in PBGJ because the groove number multiplying the
rotating speed can be an excitation frequency to disturb the journal
leaving its equilibrium position. Lee and his colleagues [79]
modied Elrods method [10] and applied coordinate transforma-
tion to analyze three different groove designs in the investigation
of the footprints in cavitation distributions within a HGJB. In
addition, they investigated the differences between HGJB and PJB
to conclude that HGJB is better than PJB on the operating stability
and the region of cavitation distributions is also decreased. Rao and
Sawicki [11] used the perturbation method and coordinate trans-
formation to solve Elrods cavitation equation to study the static
load capacity, dynamic stiffness and damping coefcients, stability
threshold, and critical whirl ratio to yield that HGJB is better than
PJB at a low eccentricity ratio.
This study applies the nite difference method associated with
Gaussian elimination method to solve the lubricant lm Reynolds
equation with the Gr umbel boundary conditions. Except Kang
et al., the aforementioned researchers all use other various
numerical methods. The Gaussian elimination method is faster
than an iterative method in solving a set of simultaneous equa-
tions. This study will also pay attention to the characteristics of
the lubricant ow both in the regular and inverse directions. The
aim of this study is to choose and optimize the design parameters
of a HGJB to establish the design rules and analysis techniques for
the miniature spindle motors with high rotating speed.
2. Finite difference method for solving perturbation
of oil lm
For a hydrodynamic HGJB as shown in Fig. 1, the herringbone
grooves with the rectangular cross section having depth h
g
and
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/triboint
Tribology International
0301-679X/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.triboint.2012.05.015
n
Corresponding author. Tel.: 886 3 2654315; fax: 886 3 2654351.
E-mail address: yk@cycu.edu.tw (Y. Kang).
Tribology International 55 (2012) 1528
width l
g
are arranged with bilateral symmetry along the z-axis.
The groove angle b is positive when the direction of circumfer-
ential ow Q
c
is the same as shown in this gure, and b is negative
when the direction of Q
c
is inverse. The Reynolds number for this
HGJB at the maximum rotational speed is
Re
rVD
m

886 0:002 7500 0:004
0:036
1476:67,
which is less than the upper limit of the value: Re 2300 for
laminar ow. With the assumptions of incompressible and
laminar ow for lubricant, which is fed into the clearance
between the plain journal and the grooved sleeve, and using the
following dimensionless parameters: h h=c, P P=P
a
, y x=R,
z z=L=2, and t ot, the governing dimensionless Reynolds
equation can be expressed in the yz plane as
@
@y
h
3 @P
@y
_ _

D
L
_ _
2
@
@z
h
3 @P
@z
_ _
L
@h
@y
2l
@h
@t
_ _
1
where L6mo=P
a
c=R
2
is the bearing parameter, l o
p
=o is
the whirl frequency ratio, o
p
is the journal whirl frequency, o is
the journal rotating speed, and L and D are the length and
diameter of the bearing, respectively.
Because of the bilateral symmetry of the herringbone grooves,
the calculating domain of Eq. (1), which is only the lower half of
the lubricant lm as shown in Fig. 1(b), can be numerically solved
to obtain its load capacity, side leakage, total friction force, and
dynamic characteristics. As the results are multiplied by two the
exact complete characteristics of this lm are obtained. It is
divided into mn grids as shown in Fig. 2, with m nodes along
the lateral axis and n nodes along the vertical axis for numerical
computations. The node spans are Dy 2p=m in the y-axis and
Dz 1=n in the z-axis. Three representative grids A, B and C,
located on the interior, the symmetric axis and the periphery of
the calculating domain respectively, and the details of their grid
structure are shown in Fig. 3, where refers to a grid node and
refers to the eight points of neighborhood in the eight-point
discrete grid [12]. As shown in Fig. 3, O
ij
contains node G
ij
that is
Nomenclature
A
ij,k
applied by eight point discrete grid method; the eight
points are separated equally on the closed border G
ij,l
surrounding the chosen grid node G
ij
, where k18
and l 14
B
ij
,B
ij
the normal and cross damping coefcients and their
dimensionless forms B
ij
B
ij
co=P
a
LD (i, j x, y) in the
Cartesian coordinate, and i,j e,f in the perturba-
tion coordinate
c radial clearance (m)
c
d
decaying exponent of perturbed whirl
D, R diameter and radius of bearing (m)
e
o
,e
p
static eccentricity and radial displacement of per-
turbed whirl (m)
F
f
,F
f
friction force due to shear stress within the lm (N),
and its dimensionless form, F
f
F
f
=P
a
LD
F
x
,F
y
force components due to lubricant lm pressure in
the x and y directions (N)
G
ij
grid node with the position coordinate (i, j)
h,h local lm thickness (m), and its dimensionless form,
h h=c
h
o
,h
o
static lm thickness (m), and its dimensionless form,
h
o
h
o
=c
h
g
groove depth (m)
I
c
the threshold of instability, I
c
m
c
l
2
c
K
ij
,K
ij
the normal and cross stiffness coefcients and their
dimensionless forms K
ij
K
ij
c=P
a
LD (i, j x, y) in the
Cartesian coordinate, and i,j e,f in the perturba-
tion coordinate
L axial length of bearing (m)
l
g
, l
r
groove and ridge width (m)
m,m
c
dimensionless mass parameter, mmco
2
=2P
a
LD;
and critical value for m
N number of herringbone grooves
O
b
,O
j
bearing center, and journal (spindle) center
P,P the pressure distribution of lubricant lm (N/m
2
),
P P=P
a
P
a
atmospheric pressure (10
5
N/m
2
)
P
l
k
,P
l
k
lubricant lm pressure distributions (N/m
2
);
P
l
k
P
l
k
=P
a
; subscript k denoted for o (static), e and
f (for perturbation); superscript l denoted for real
part by R, and for imaginary part by I
Q
Z
,Q
Z
volumetric ow rate (m
3
/s) of lubricant, and its
dimensionless form Q
Z
Q
Z
P
a
c
3
=12mR
_ _
1
, Z c,s
for the circumferential ow and the side leakage
s characteristic frequency ratio of the journal motion,
s c
d
io
p
=o
t time (s)
W
k
,W
k
unidirectional load capacity (N), and its dimension-
less form, W
k
W
k
=P
a
LD, k x,y in the Cartesian
coordinate, and k e,f in the perturbation coordinate
X,Y dimensionless amplitudes of whirl motion in x and y
directions, X X=c, Y Y=c
x, y, z Cartesian coordinates of lubricant lm
Greek symbols
y,y,z dimensionless coordinates, y x=R, y y=c,
z z=L=2
~
y relative angle measured from the line connecting
equilibrium positions of journal center and bearing
center,
~
y yf
0
G
ij
,G
ij,k
the virtual closed border and its four segments
surrounding the grid node G
ij
, k14
L bearing parameter, L6mo=P
a
c=R
2
O
ij
the computational mesh of the node G
ij
surrounded
by G
ij,k
a groove width ratio, a l
g
=l
r
l
g

b groove angle
g groove depth ratio, g h
g
=c
d
k
criterion values of convergence, k13
e
0
,e
p
eccentricity ratios, e
0
e
0
=c for static equilibrium,
and e
p
e
p
=c for perturbation
f
o
,f
p
static attitude angle, and angular displacement of
perturbed whirl
l,l
c
whirl frequency ratio, l o
p
=o; and the critical one,
l
c
o
p
=o
c
m dynamic viscosity of lubricant (N s/m
2
)
m
f
the friction coefcient, m
f
F
f
=W
t dimensionless time, t ot
o rotating speed of spindle (rad/s)
o
p
whirl frequency of perturbed motion (rad/s)
S.-K. Chen et al. / Tribology International 55 (2012) 1528 16
located at the center, which is situated at the coordinates of
i1 2pR=mj1 L=2ncot b in the lateral direction and
j1 L=2n in the vertical direction measured from the lower
left-hand corner as the origin point, which is denoted for each
grid with the width being Dy and the length being Dz=sin b.
For a typical point within the interior as shown in Fig. 3(a),
the approximate values of lm thickness for the four terminal
points A
ij,1
, A
ij,4
, A
ij,5
, and A
ij,8
, are the average of those for their
adjacent four grid nodes, and the values for the four inner points
A
ij,2
, A
ij,3
, A
ij,6
, and A
ij,7
, are the average of three times of the
lm thickness for the adjacent terminal point and one time of that
for the far terminal point. For grid nodes located at the cross
point of symmetric axis of grooved sleeve and the edge of one
groove as shown in Fig. 3(b), the boundary condition is @P
in
=@z 0
so that P
i,n1
P
i,n1
and their coefcients for the equation
of P
i,n
are also equal. When a grid node is located at the cross
point of the bearings periphery and the edge of one groove
as shown in Fig. 3(c), its boundary conditions will be P
i,1
1,
i 12m.
The dimensionless rate of net ow passing through this
grid can be determined by surface integration of Eq. (1) as follows:
__
O
ij
@
@y
h
3 @P
@y
_ _

D
L
_ _
2
@
@z
h
3 @P
@z
_ _ _ _
dy dz
__
O
ij
L
@h
@y
2l
@h
@t
_ _
dy dz
2a
where the surface integral along O
ij
on the left-hand-side of Eq. (2a)
can be transformed into a closed line integral on the basis of
Greens theorem as follows:
_
G
ij

D
L
_ _
2
h
3 @P
@z
_ _
dy h
3 @P
@y
_ _
dz
_ _

__
O
ij
L
@h
@y
2l
@h
@t
_
dy dz
_
2b
where G
ij

4
k 1
G
ij,k
is the perimeter of O
ij
and consists of
G
ij,1
: A
ij,1
A
ij,4
, G
ij,2
: A
ij,4
A
ij,5
, G
ij,3
: A
ij,5
A
ij,8
, and G
ij,4
: A
ij,8
A
ij,1
, and
the direction along the integrating path is counterclockwise.
The discontinuity of lubricant lm at the edge between each
groove and ridge makes the partial differential term @h=@y
i,j
in
Eq. (2b) meaningless. The method of eight-point discrete grid is
applied to determine the approximation of @h=@y
i,j
by the
numerical integration of lm thickness along the surrounding
border to deal with the discontinuity problem of lm thickness in
differential determination. According to this method, the rst
term in the left-hand-side of Eq. (2b) can be divided as the
summation of four line integrals along the segments from G
ij,1
to
G
ij,4
. However, the projection of G
ij,2
on the y-axis is Dz cos b
journal
grooved
sleeve
z

c
g
h
journal
bearing
g
l
g
l
r
l

Q
Q
s
c
ridge groove
Fig. 1. Hydrodynamic herringbone grooved journal bearing. (a) Cross-section of spindle- bearing system. (b) Expansion graph of herringbone grooves.
z

groove ridge
x R =
A
B
symmetric axis
( ) interior
C periphery
one of
edges
Fig. 2. Grid system of calculating domain.
S.-K. Chen et al. / Tribology International 55 (2012) 1528 17
and that of G
ij,4
will be Dz cos b; then, the summation of the line
integrals along G
ij,2
and G
ij,4
equals to zero. Thus, the integration
yields
_
G
ij
h
3 @P
@z
_ _
dy
_
G
ij,1
h
3 @P
@z
_ _
dy
_
G
ij,3
h
3 @P
@z
_ _
dy 3
The trapezoidal rule is applied to approximate the line inte-
grals of Eq. (3) as
_
G
ij,1
h
3 @P
@z
_ _
dy P
i,j 1
P
i,j
h
3

i1=4, j 1=2
h
3

i 1=4, j 1=2
_ _
Dy
2Dz
3a
_
G
ij,3
h
3 @P
@z
_ _
dy P
i,j
P
i,j1
h
3

i1=4, j1=2
h
3

i 1=4, j1=2
_ _
Dy
2Dz
3b
Consequently
_
G
ij

D
L
_ _
2
h
3 @P
@z
_ _
dy
D
L
_ _
2
P
i,j 1
P
i,j
h
3

i1=4, j 1=2
_
h
3

i 1=4, j 1=2
_
Dy
2Dz

D
L
_ _
2
P
i,j
P
i,j1
h
3

i1=4, j1=2
_
h
3

i 1=4, j1=2
_
Dy
2Dz
4
_
G
ij
h
3 @P
@y
_ _
dz P
i 1,j
P
i,j
h
3

i 1=2, j 1=2
_
h
3

i 1=2, j1=2
_
Dz
2Dy
P
i,j
P
i1,j
h
3

i1=2, j 1=2
_
h
3

i1=2, j1=2
_
Dz
2Dy
5
__
O
ij
@h
@y
dy dz
_
j 1=2
j1=2
_
i 1=2
i1=2
@h
@y
dy dz
_
j 1=2
j1=2
h9
i 1=2
i1=2
dz
h
i 1=2, j 1=2
h
i1=2, j 1=2
_
h
i 1=2, j1=2
h
i1=2, j1=2
_
Dz
2
6
The eight points from A
ij,1
to A
ij,8
of grid node G
ij
do only for the
determinations of approximate lm thickness as mentioned
above and do nothing for the determinations of lm pressure.
Due to a small perturbation, the journal center leaves its
equilibrium position e
o
,f
o
as shown in Fig. 4 and whirls with
radial and tangential components Ree
p
e
st
and Ree
o
f
p
e
st
, which
induce pressure and thickness perturbations on the lm. In the
rst-order approximation, the dimensionless pressure perturba-
tion can be expressed as
P P
0
e
p
e
st
P
e
e
0
f
p
e
st
P
f
7
where P
0
P
0
=P
a
is the dimensionless static pressure, P
e
P
e
=P
a
and P
f
P
f
=P
a
are the radial and tangential dimensionless
perturbation pressures respectively, f
p
is the angular
displacement of perturbation, and s c
d
io
p
=o is the charac-
teristic frequency ratio of journal whirl. In a similar manner,
the dimensionless thickness of lm perturbation can be expressed
as
h h
0
e
p
e
st
cos
~
ye
0
f
p
e
st
sin
~
y 8
at ridges, and
h h
0
ge
p
e
st
cos
~
ye
0
f
p
e
st
sin
~
y 9
j
O
O

W
0
e
x
F
y
F
x
y

c
c

O
O
de
dt
d
e
dt

Fig. 4. Whirling motion of lubricant lm within a hydrodynamic bearing: (a) the

static equilibrium position and (b) the perturbation coordinates of journal center.
ij,1

ij,2

ij,4

ij,3

(i - 1, j)
(i+1, j)
(i, j-1)
(i, j+1)
(i, j)
ij

ij,2
A
ij,1
A
ij,3
A
ij,4
A
ij,8
A
ij,7
A
ij,6
A
ij,5
A

i
i +
j +
j
1
2
1
2
1
2
1
2
i i
1
4
1
4

z
z

G
i, j
i, j 1
G
i+1, j
G
i+1, j+1
G
i, j+1
G
(i+1, j+1)
(i+1, j 1) -
i+1, j 1
G
- -
(i , j+1) - 1
(i , j 1) - 1 -
i , j+1
G
-1
i , j
G
-1
i , j 1
G
-1 -
edge
edge
z

sin
in,2

in,4

in,3

(i - 1, n) (i+1, n)
(i, n-1)
(i , n)
in

in,8
A
in,7
A
in,6
A
in,5
A

i
i +
n?
1
2
1
2
1
2
i +
i
1
4
1
4

z
z

G
i,
i,n 1
G
i+1, n
G
(i+1, n 1) -
i+1, n 1
G
- -
(i , n 1) - 1 -
i , n
G
-1
i , n 1
G
-1 -
edge
2
) (
2
1
(
2
1
)
n
2
1
symmetric axis
i,n 1
G
-
i , n 1
G
-1 -
i+1, n 1
G
-
in,5
A
in,6
A
in,7
A
in,8
A
edge

sin
z

2
z

2
i1,2

i1,4

i1, 3

(i - 1, 2) (i+1, 2)
(i, 1)
(i, 2)
i1
?
i1,1
A
i1,2
A
i1, 3
A
i1, 4
A

i
i +
1+
1
2
1
2
1
2
i + i
1
4
1
4

z
z

G
i,
i, 1
G
i+1, 2
G
(i+1, 1)
i+1, 1
G
(i , 1) - 1
i , 2
G
-1
i , 1
G
-1
edge
2
) (
2
1 (
2
1
)
2
2
1
periphery
z

2sin
Fig. 3. Details of computational grid structure. (a) Details of A. (b) Details of B.
(c) Details of C.
S.-K. Chen et al. / Tribology International 55 (2012) 1528 18
at grooves, where
~
y yf
0
is the relative angle measured from
the line connecting equilibrium positions of journal center and
bearing center, h
0
h
0
=c is the dimensionless static lm thickness
with h
0
h
w
for ridges and h
0
g h
w
for grooves, and g h
g
=c is
the groove depth ratio.
Substituting Eqs. (7)(9) into Eq. (2b) and neglecting perturba-
tion terms higher than the rst order gives
_
G
ij

D
L
_ _
2
h
3
w
@P
0
@z
_ _
dy h
3
w
@P
0
@y
_ _
dz
_ _

__
O
ij
L
@h
w
@y
dy dz 10
of the static part that is obtained from the constant terms, and
_
G
ij

D
L
_ _
2
h
3
w
@P
e
@z
_ _
3
D
L
_ _
2
h
2
w
cos
~
y
@P
0
@z
_ _ _ _
dy

_
G
ij
h
3
w
@P
e
@y
3h
2
w
cos
~
y
@P
0
@y
_ _
dz

__
O
ij
Lsin
~
y2
c
d
o
cos
~
ydy dzi
__
O
ij
L 2
o
p
o
cos
~
y
_ _
dy dz
11
of a dynamic part that is obtained from perturbation terms with
e
p
e
st
, and
_
G
ij

D
L
_ _
2
h
3
w
@P
f
@z
_ _
3
D
L
_ _
2
h
2
w
sin
~
y
@P
0
@z
_ _ _ _
dy

_
G
ij
h
3
w
@P
f
@y
3h
2
w
sin
~
y
@P
0
@y
_ _
dz

__
O
ij
L cos
~
y2
c
d
o
sin
~
y
_ _
dy dz
i
__
O
ij
L 2
o
p
o
sin
~
y
_ _
dy dz 12
of another dynamic part that is obtained from perturbation terms
with e
0
f
p
e
st
, on both sides of this equation.
The central difference method can be used for numerical
integration of Eqs. (10)(12), which are discretized in the same
form as follows:
A
k,ij
P
k

i1,j
B
k,ij
P
k

i,j
C
k,ij
P
k

i 1,j
D
k,ij
P
k

i,j1
E
k,ij
P
k

i,j 1
F
k,ij
13
where P
k
represents P
0
and the complex forms: P
e
ReP
e

i ImP
e
P
R
e
iP
I
e
and P
f
ReP
f
i ImP
f
P
R
f
iP
I
f
; F
k,ij
is the
complex form of the non-homogeneous terms of Eqs. (10)(12),
and k0, e, or f. Further, Eq. (13) can be expanded into real and
imaginary parts and ve equations for P
0
, P
R
e
, P
I
e
, P
R
f
, and P
I
f
are
obtained. The details about coefcients A
k,ij
to F
k,ij
of ve equa-
tions, which are expressed by Eq. (13), are shown in Appendix A.
Eq. (13) represents mn coupled equations that describe the ow
rate passing through the mn grids.
3. Static analyses
For Eq. (13) of P
0

ij
, the coefcients and non-homogeneous
term F
0,ij
are functions only of dimensionless static lm thickness.
The components of the dimensionless load capacity can be
determined from P
0

ij
by
W
x

W
x
P
a
LD
2
_
1
0
_
2p
0
P
0
cos
~
yR dy dz 2

n
j 1

m
i 1
P
0
cos
~
y
ij
RDy Dz
14a
in the lateral direction of static load, which should be zero, and
W
y

W
y
P
a
LD
2
_
1
0
_
2p
0
P
0
sin
~
yR dy dz 2

n
j 1

m
i 1
P
0
sin
~
y
ij
RDy Dz
14b
in the negative direction of static load. For solving Eq. (13)
iteratively, an initial guess includes both distributions of static
pressure P
0

ij
and attitude angle f
o
. Iterations are terminated
when three conditions as described below are satised:
Max9P
k

new
ij
P
k

old
ij
9rd
1
,
W
W

rd
2
, and Df
o
9f
new
o
f
old
o
9rd
3
15
where superscripts new and old denote the newest result and the
last one of iteration, and d
1
, d
2
, and d
3
are obtained by a
negotiation between accuracy and computing time. The solution
process expressed in ow chart is shown in Fig. 5.
In Eq. (10), the rst term in the left-hand-side is the volumetric
ow rate of the grid perpendicular to dy. For the value of (D/L)
being unit, the dimensionless side leakage Q
s
in the z-axis is
obtained by
Q
s

12mR Q
s
P
a
c
3

12mR
P
a
c
3
2
_
1
0
_
2p
0
h
3
0
@P
0
@z
R dy dz

24mR
P
a
c
3

n
j 1

m
i 1
h
3
0

ij
1
Dz
P
0

i,j 1=2
P
0

i,j1=2
_ _
R Dy Dz
16
The second term in the left-hand-side of Eq. (10) is the
volumetric ow rate of the grid perpendicular to dz, and the
right-hand-side is that referring to the rotating shaft and the oil
viscosity along the y-axis. So, the dimensionless circumferential
ow rate Q
c
in the y-axis of the lubricant lm is obtained by
Q
c

12mR Q
c
P
a
c
3

12mR
P
a
c
3
2
_
1
0
_
2p
0
Lh
0
h
3
0
@P
0
@y
_ _
R dy dz

24mR
P
a
c
3

n
j 1

m
i 1
Lh
0

ij
h
3
0

ij
1
Dy
P
0

i 1=2, j
P
0

i1=2,j
_ _
_ _
R Dy Dz
17
Because the total friction force of the lm is the surface integral of
the shear stress along O
ij
due to the rotation of journal, the
dimensionless total friction force F
f
of the lubricant lm in the
y-axis is obtained by
F
f

F
f
2cLP
a
2
_
1
0
_
2p
0
h
0
4
@P
0
@y

L
12h
0
_ _
R dy dz m
f
W
2

n
j 1

m
i 1
h
0

ij
4
1
Dy
P
0

i 1=2,j
P
0

i1=2,j
_ _

L
12h
0

ij
_ _
R Dy Dz
18
where m
f
F=W is the friction coefcient.
4. Dynamic stiffness and damping coefcients
The perturbation load capacity W
e
induced by the innitesimal
perturbation term e
p
e
st
P
e
, is the surface integral along the
S.-K. Chen et al. / Tribology International 55 (2012) 1528 19
lubricant lm of this perturbation and can be subdivided into
radial and tangential parts as
W
e

R
2
_
1
0
_
2p
0
P
e
R cos
~
y dy dz 19a
W
e

T
2
_
1
0
_
2p
0
P
e
R sin
~
y dy dz 19b
However, this perturbation load capacity is also a linear combina-
tion of stiffness coefcients and damping coefcients and can be
non-dimensionalized as
W
e

R

W
R
P
a
LDe
p

K
ee
c
P
a
LD
i
B
ee
co
p
P
a
LD
K
ee
ilB
ee
20a
W
e

T

W
T
P
a
LDe
p

K
fe
c
P
a
LD
i
B
fe
co
p
P
a
LD
K
fe
ilB
fe
20b
Similarly, the perturbation load capacity W
f
induced by the
innitesimal perturbation term e
0
f
p
e
st
P
f
is the surface integral
along the lubricant lm of this perturbation and can be subdi-
vided as
W
f

R
2
_
1
0
_
2p
0
P
f
R cos
~
y dy dz 21a
W
f

T
2
_
1
0
_
2p
0
P
f
R sin
~
y dy dz 21b
This perturbation load capacity is also a linear combination of
stiffness coefcients and damping coefcients and can be non-
dimensionalized as
W
f

R

W
R
P
a
LDe
o
f
p

K
ef
c
P
a
LD
i
B
ef
co
p
P
a
LD
K
ef
ilB
ef
22a
( )
( )
( )
( )
( ) ( )
( )
new
P

( )
old
P

( )
new
P
( )
old
P

( ) ( )
( )
new
P

( )
old
P

( )
new
P

( )
old
P

START
L R c N
P
initial guess of attitude angle
and initial guess of static pressure
Solve static part of finite difference
Reynolds equation using Gauss
-Siedel method to determine
stati
=

+
c pressures
convergence on static
Solve dynamic part of finite difference
Reynolds equation using Gauss-Siedel
method to determine nodal dynamic
pressures
Compute dynamic
coefficients
END
No
Yes
No Yes
No
Yes
Compute
] [
Output static pressures
a
x
W
y
W
and
new
0
new
0
new
0
P
whirl ratio
grids relaxation
Calculate film thickness and
Reynolds equation (18)
o
o
P
nodal
old
0
P
Check
pressure
new
0
P
old
0
P
Check
convergence
on dynamic
pressure
R
P
R
P
I
P
I
P
ij
K
ij
B
Compute critical mass
parameter and critical
c
m
c
coefficients of finite difference
new
0
P
old
0
P
,
, , , , , ,
,
, , ,
,
Input geometric parameters:
Input physical parameters:
Input convergence parameters:
factor, iteration threshold of iteration error threshold, critical whirl ratio
Output maximum pressure,
load capacity, attitude angle,
side leakage, stability
,
,
, , ,
5
o
10

5
10

W
/
W
=
=

+
+
]
] [
[
Fig. 5. Solution scheme.
B
earin
g
c
en
ter
w
h
irl
o
rb
it
o
f
jo
u
rn
al
ce
n
te
r
0
e
m
0
y
0
( , ) e
p
0 p
0
( , )
p
e
b
O
b
O
j
O
j
O
x
0
ep
Fig. 6. Model of rotorhydrodynamic-bearing system.
Table 1
Dimensionless design parameters of HGJB.
Bearing length to diameter ratio, L/D 1
Groove angle, b 20701, 20 to 701
Groove number, N 8
Groove width ratio, a 1/4, 1/3, 1/2
Groove depth ratio, g 0.5, 1.0, 1.5, 2.0
Eccentricity ratio, e 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7
Bearing parameter, L 188.4956, 75.39822
S.-K. Chen et al. / Tribology International 55 (2012) 1528 20
W
f

T

W
T
P
a
LDe
o
f
p

K
ff
c
P
a
LD
i
B
ff
co
p
P
a
LD
K
ff
ilB
ff
22b
where B
ij
B
ij
co=P
a
LD and K
ij
K
ij
c=P
a
LD (i,j e,f). Hence,
Eqs. (20) and (22) are the dimensionless stiffness and damping
coefcients under the condition of any whirl frequency ratio l,
and they can be obtained directly by integrating the dimension-
less real and imaginary parts of perturbation pressures P
e
and P
f
of Eqs. (19) and (21) as
K
ee
2
_
1
0
_
2p
0
P
R
e
y,zR cos
~
y dy dz 2

n
j 1

m
i 1
P
R
e
cos
~
y
ij
R Dy Dz
0.0 0.2 0.4 0.6 0.8
0
2
4
6

W
numerical results of Lee et al [9]
present results
experimental results of Hirs [1]
100
1000
2
5
6

2
5
6
1
2
8

2
5
6
1
2
8

1
2
8
6
4

1
2
8
6
4

6
4
T

(
s
e
c
)
grid number (mn)
Fig. 7. Evaluation of the validity and time spent for present method. (a) Result comparison for load capacity. (b) Computing time for static analyses and the determination
of dynamic coefcients.
0
2
4
6
8
10
12
W
30
50
70
90
o
o
o
0
2
4
6
8
10
12
W
30
50
70
90
0
2
4
6
8
10
12
W
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
30
50
70
90
= 0.5
= 1.0
= 2.0
c=6 m :
3 m :
= 20
= 40
= 60
= 70
c=6 m :
3 m :
=1/2
=1/3
=1/4
c=6 m :
3 m :
PJB3, PJB6
=1/2
=1/3
=1/4
c=6 m :
3 m :
PJB3, PJB6
= 20
= 40
= 60
= 70
c=6 m :
3 m :
PJB3, PJB6
PJB3, PJB6
PJB3, PJB6
= 0.5
= 1.0
= 2.0
c=6 m :
3 m :
PJB3, PJB6
Fig. 8. Static dimensionless load capacity (W) and attitude angle (f
o
) versus eccentricity ratio (e) for PJB and various design parameters of HGJB with various clearances.
(a) b301, g0.5, with various a. (b) a1/4, g0.5, with various b. (c) a1/4, b301 with various g.
S.-K. Chen et al. / Tribology International 55 (2012) 1528 21
K
fe
2
_
1
0
_
2p
0
P
R
e
y,zR sin
~
y dy dz 2

n
j 1

m
i 1
P
R
e
sin
~
y
ij
R Dy Dz
K
ef
2
_
1
0
_
2p
0
P
R
f
y,zR cos
~
y dy dz 2

n
j 1

m
i 1
P
R
f
cos
~
y
ij
R Dy Dz
K
ff
2
_
1
0
_
2p
0
P
R
f
y,zR sin
~
y dy dz 2

n
j 1

m
i 1
P
R
f
sin
~
y
ij
R Dy Dz
lB
ee
2
_
1
0
_
2p
0
P
I
e
y,zR cos
~
y dy dz 2

n
j 1

m
i 1
P
I
e
cos
~
y
ij
R Dy Dz
lB
fe
2
_
1
0
_
2p
0
P
I
e
y,zR sin
~
y dy dz 2

n
j 1

m
i 1
P
I
e
sin
~
y
ij
R Dy Dz
lB
ef
2
_
1
0
_
2p
0
P
I
f
y,zR cos
~
y dy dz 2

n
j 1

m
i 1
P
I
f
cos
~
y
ij
R Dy Dz
lB
ff
2
_
1
0
_
2p
0
P
I
f
y,zR sin
~
y dy dz 2

n
j 1

m
i 1
P
I
f
sin
~
y
ij
R Dy Dz
23
The stiffness and damping coefcients obtained from Eq. (23)
are expressed in ef coordinate system. To make the follow-up
comparison more clear, we need to transform the coordinate
system of the results from ef to xy. The transformation
equations are
K
xx
K
xy
K
yx
K
yy
_ _

cos f
o
sin f
o
sin f
o
cos f
o
_ _
K
ee
K
ef
W
y
K
fe
K
ff
W
x
_
_
_
_
cos f
o
sin f
o
sin f
o
cos f
o
_ _
T
24
B
xx
B
xy
B
yx
B
yy
_ _

cos f
o
sin f
o
sin f
o
cos f
o
_ _
B
ee
B
ef
B
fe
B
ff
_
_
_
_
cos f
o
sin f
o
sin f
o
cos f
o
_ _
T
25
5. Stability analysis
For a balanced rigid rotor, supported horizontally by two
identical HGJB on both ends, the shaft with a concentrated mass
m as the rotor at its center is shown in Fig. 6. When the rotor is
disturbed, the center of rotor will deviate from its equilibrium
position. This rotor center will whirl with a complex frequency
c
d
io
p
, and the whirl motion is dependent on the dynamic
characteristics of the rotor-bearing system. If c
d
o0, the rotor
center will eventually whirl back to its equilibrium position; if
c
d
40, the rotor center whirls farther away from its equilibrium
position as time increases and will cause damage to bearing or
journal when contact occurs. When c
d
0, which is the critical
condition between stability and instability, the track of rotor
center will be a circle, so called limit circle. With a small
disturbance, the dimensionless perturbation equation of the
journal with whirl center at O
j
in a referenced xy coordinate
0
200
400
600
800
1000
1200
s
Q
s
Q
s
Q
=1/2
=1/3
=1/4
=1/2
=1/3
=1/4
c=6 m :
3 m :
c=6 m :
3 m :
1x10
4
2x10
4
3x10
4
4x10
4
5x10
4
6x10
4
7x10
4
c
Q
0
500
1000
1500
= 20
= 40
= 60
= 70
= 20
= 40
= 60
= 70
c=6 m :
3 m :
0.0
2.0x10
4
4.0x10
4
6.0x10
4
8.0x10
4
c
Q
( 10 )
1.0
2.0
2.5
1.5
0.5
0.0
= 0.5
= 1.0
= 2.0
= 0.5
= 1.0
= 2.0
c=6 m :
3 m :
c=6 m :
3 m :
c=6 m :
3 m :
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0
2.0x10
4
4.0x10
4
6.0x10
4
8.0x10
4
1.0x10
5
c
Q
PJB3
PJB6
PJB3
PJB3
PJB3
PJB3
PJB6
PJB6
PJB6 PJB6
PJB3
PJB6
Fig. 9. Dimensionless side leakage (Q
s
) and total circumferential ow rate (Q
c
) versus eccentricity ratio (e) for PJB and various design parameters of HGJB with various
clearances. (a) b301, g0.5, with various a. (b) a1/4, g0.5, with various b. (c) a1/4, b301 with various g.
S.-K. Chen et al. / Tribology International 55 (2012) 1528 22
system can be described by
m 0
0 m
_ _

x

y
_ _

B
xx
B
xy
B
yx
B
yy
_ _
_
x
_
y
_ _

K
xx
K
xy
K
yx
K
yy
_ _
x
y
_ _

0
0
_ _
26a
where mmco
2
=2P
a
LD is the dimensionless mass parameter.
For the special case of c
d
0, we have s io
p
=o
c
il
c
, x Xe
ilc t
and y Ye
ilc t
. Substituting x and y into Eq. (26a) gives
K
xx
il
c
B
xx
I
c
K
xy
il
c
B
xy
K
yx
il
c
B
yx
K
yy
il
c
B
yy
I
c
_ _
X
Y
_ _

0
0
_ _
26b
where X and Y are the whirl amplitudes in x and y directions,
respectively, I
c
m
c
l
2
c
is the threshold of instability and repre-
sents the impedance of the self-excited vibration.
The determinant of Eq. (26b) must be zero for non-trivial
solutions of X and Y. Thus, setting the determinant of Eq. (26b)
equal to zero both in imaginary and real parts, the critical
dynamic stabilities I
c
and l
c
2
of this rotor-bearing system can be
obtained as follows:
I
c

K
xx
B
yy
K
yy
B
xx
K
xy
B
yx
K
yx
B
xy
B
xx
B
yy
27
l
2
c

I
c
K
xx
I
c
K
yy
K
xy
K
yx
B
xx
B
yy
B
xy
B
yx
28
where I
c
can be directly determined by Eq. (27) after K
ij
and B
ij
have been calculated. Then, l
c
can be obtained by substituting I
c
into Eq. (28). Finally, the dimensionless parameter of critical mass
can be obtained by dividing I
c
by l
c
2
, written as m
c
I
c
=l
2
c
. The
stability condition for whether rotor whirls or not is also depen-
dent on the dimensionless mass parameter, m. The rotor is stable
when m is less than m
c
; on the other hand, the rotor is unstable if
m is larger than m
c
.
6. Results and discussions
Based on the above analyses, a computational program is
built up to calculate the static and dynamic characteristics
of a HGJB with the dimensionless design parameters as shown
in Table 1. These data were chosen to represent a miniature
hydrodynamic bearing usually used in a hard disk drive system.
A detailed comparison among the present results and the experi-
mental data of Hirs [1] and the numerical results of Lee et al. [9] is
shown in Fig. 7(a). The present results match very well with their
results.
The major difference between HGJB and PJB is the inuence of
grooves. In order to master the design techniques of a HGJB,
the inuences of design parameters on the static and dynamic
characteristics are investigated in great detail. According to the
0
200
400
600
800


f
F


f
F


f
F
=1/2
=1/3
=1/4
c=6 m :
3 m :
10
2
10
3
10
1
f
f

=1/2
=1/3
=1/4
c=6 m :
3 m :
0
200
400
600
800
= 20
= 40
= 60
= 70
c=6 m :
3 m :
10
2
10
3
1
10
= 20
= 40
= 60
= 70
c=6 m :
3 m :
0
200
400
600
800
= 0.5
= 1.0
= 2.0
c=6 m :
3 m :
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
10
2
10
3
10
1
= 0.5
= 1.0
= 2.0
c=6 m :
3 m :
PJB3
PJB3, PJB6
PJB3
PJB3, PJB6
PJB3, PJB6
PJB3
PJB6
PJB6
PJB6
Fig. 10. Dimensionless friction force (F
f
) and friction coefcient (m
f
) versus eccentricity ratio (e) for PJB and various design parameters of HGJB with various clearances.
(a) b301, g0.5, with various a. (b) a1/4, g0.5, with various b. (c) a1/4, b301 with various g.
S.-K. Chen et al. / Tribology International 55 (2012) 1528 23
detailed comparison among the computed results, the optimum
design of a HGJB for the most stable and maximum load capacity
can be obtained.
As shown in Table 1, the complete design parameters are taken
into account for the analyses of both static and dynamic char-
acteristics of a HGJB. The minus sign of groove angle b signies
the ow at inverse direction. The time spent for various grid
meshes is calculated as shown in Fig. 7(b). Because of the
application of matrix operating techniques, the time spent in
calculating is very little. For the need of accuracy, the grid mesh of
128256 is applied in both static and dynamic cases and the
time spent is 380 s on average.
The clearance has no effect on both the dimensionless load
capacity W and the attitude angle f
o
. W increases with the
increase of e, and that of PJB is a little higher than those of HGJB at
the beginning, then the difference between them increases
rapidly with the increase of e as shown in Fig. 8. However, f
o
decreases with the increase in e. In order to nd out the inuence
of groove width ratio a on the W and f
o
of a HGJB, groove angle
b301 and groove depth ratio g0.5 were chosen in the present
study. The simulation results indicate that the W of PJB is greater
than all of HGJB after e0.2 as shown in Fig. 8(a). The f
o
of HGJB
with various a are very close and slightly less than that of PJB.
Choosing a1/4 and g0.5, the simulation results indicate that
the W of HGJB with various b are very near and all less than that
of PJB as shown in Fig. 8(b). The f
o
of PJB and HGJB with various b
are all very close for various e. Choosing a1/4 and b301,
the simulation results indicate that the sequence of W is g
0:54g 1:04g 2:0 for HGJB as shown in Fig. 8(c). The sequence
of f
o
is also the same.
Both the dimensionless side leakage Q
s
and circumferential
ow rate Q
c
increase with the increase in e, and those of PJB are
always less than those of HGJB as shown in Fig. 9. The results
show very clear that both Q
s
and Q
c
are inversely proportional to
0
1x10
8
2x10
8
3x10
8
4x10
8
HGJB3
HGJB6
x
x
K
0
1x10
6
2x10
6
3x10
6
x
x
B
HGJB3
HGJB6
0.0
2.0x10
8
4.0x10
8
6.0x10
8
8.0x10
8
1.0x10
9
HGJB3
HGJB6
x
y
K
0.0
5.0x10
5
1.0x10
6
1.5x10
6
2.0x10
6
HGJB3
HGJB6
x
y
B
-1.0x10
9
-8.0x10
8
-6.0x10
8
-4.0x10
8
-2.0x10
8
0.0
HGJB3
HGJB6
y
x
K
0.0
5.0x10
5
1.0x10
6
1.5x10
6
2.0x10
6
HGJB3
HGJB6
y
x
B
0.0
2.0x10
8
4.0x10
8
6.0x10
8
HGJB3
HGJB6
y
y
K
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0
1x10
6
2x10
6
3x10
6
HGJB3
HGJB6
y
y
B
0.0
2.0x10
4.0x10
6.0x10
8.0x10
-5.0x10
0.0
5.0x10
1.0x10
1.5x10
2.0x10
2.5x10
0
1x10
2x10
3x10
4x10
5x10
0.1 0.2 0.3 0.4
0.1 0.2 0.3
0.1 0.2 0.3
0.05 0.10 0.15 0.20
0.0
5.0x10
1.0x10
1.5x10
2.0x10
2.5x10
PJB3
PJB6
PJB3
PJB6
PJB3
PJB6
PJB3
PJB6
PJB6
PJB3
PJB3
PJB3
PJB3
PJB3
PJB3
PJB3
PJB3
PJB6
PJB6
PJB6
PJB6
PJB6
Fig. 11. Dynamic stiffness coefcients (K
ij
) and damping coefcients (B
ij
) versus eccentricity ratio (e) for PJB and HGJB with various clearances, and a1/2, b301, g0.5.
S.-K. Chen et al. / Tribology International 55 (2012) 1528 24
the cube of clearance. The results of Q
s
are very close for HGJB
with various a as shown in Fig. 9(a). However, the sequence of Q
c
is a 1=44a 1=34a 1=2. The sequence of Q
s
is b 7014
b 6014b 4014b 201 for HGJB with various b as shown in
Fig. 9(b). However, the results of Q
c
are very close for various e.
The sequence of Q
s
is g 2:04g 1:04g 0:5 for HGJB as shown
in Fig. 9(c); the sequence of Q
c
is the same.
The dimensionless total friction force F
f
increases with the
increase in e, and that of PJB is always greater than those of HGJB
as shown in Fig. 10. The values of F
f
with clearance c3 mm are
four times that of F
f
with clearance c6 mm but the friction
coefcient m
f
has the same value. However, m
f
has an abrupt drop
from a high value near 15020 when e goes from 0.01 to 0.1, and
goes down to near unit when e0.8. The sequence of F
f
is
a 1=24a 1=34a 1=4 for HGJB as shown in Fig. 10(a). The
results of F
f
are very close for HGJB with various b as shown in
Fig. 10(b). The sequence of F
f
is g 2:04g 1:04g 0:5 for HGJB
as shown in Fig. 10(c).
Both the dynamic stiffness coefcients K
ij
excluding K
yx
and
damping coefcients B
ij
increase with the increase in e, and those
of PJB are greater than those of HGJB when e40.4 as shown in
Fig. 11. PJB3 and PJB6 are denoted for PJB with clearances 3 mm
and 6 mm. HGJB3 and HGJB6 are denoted similarly and both with
a1/2, b301, and g0.5. The sequences of K
xy
, B
xx
, B
xy
and B
yy
are very clear, which are PJB34HGJB34PJB64HGJB6. In the case
of K
yx
, their values are all negative and they decrease with the
increase in e; the sequence is HGJB64PJB64HGJB34PJB3.
The clearance also has no effect on both the stability threshold
m
c
and the critical whirl frequency ratio l
c
. m
c
of HGJB decreases
from a large value near 90 down to about 13, however that of PJB
is nearly unchanged at about 12 in comparison with HGJB before
e 0:2, and then they go inversely at near 13 all the way to e 0:8
as shown in Fig. 12. Except two in HGJB with various b and two in
HGJB with various g, l
c
of HGJB and that of PJB decrease at rst and
then increase rapidly with the increase of e. The l
c
of those four in
HGJB go down for large eccentricity.
The minus groove angle has great inuence on the perfor-
mances of the stability threshold m
c
and the friction coefcient m
f
for HGJB with a 1=2 and g 0:5 as shown in Fig. 13. The friction
coefcient for HGJB with b 201 is about ve times greater
than that with b 401, which is over double of the rest when
e0.01, and then they all come close to each other and near unit.
The stability thresholds for HGJB with minus groove angle,
excluding b 701, are all negative which means the systems
are unstable. The performance of PJB keeps unchanged regardless
whether the spindle rotates in regular or inverse direction. In
the other conditions, the minus groove angle for HGJB has little
effect.
10
30
50
70
90
c
m
c
m
c
m

=1/2
=1/3
=1/4
c=6m :
3m :
0.45
0.50
0.55
0.60
c

=1/2
=1/3
=1/4
c=6m :
3m :
10
30
50
70
90

= 20
= 40
= 60
= 70
c=6m :

3m :
0.45
0.50
0.55
0.60

= 20
= 40
= 60
= 70
c=6m :
3m :
0
50
100
150
200

= 0.5
= 1.0
= 2.0
c=6m :
3m :
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.45
0.50
0.55
0.60
0.65
0.70

= 0.5
= 1.0
= 2.0
c=6m :
3m :
instability
instability
instability
instability
instability
instability
stability
stability
stability
stability
stability
stability
PJB3, PJB6
PJB3, PJB6
PJB3, PJB6
PJB3, PJB6
PJB3, PJB6
PJB3, PJB6
Fig. 12. Stability threshold (mc) and critical whirl speed ratio (l
c
) versus eccentricity ratio (e) for PJB and various design parameters of HGJB with various clearances.
(a) b301, g0.5, with various a. (b) a1/4, g0.5, with various b. (c) a1/4, b301 with various g.
S.-K. Chen et al. / Tribology International 55 (2012) 1528 25
7. Conclusions
The conclusions of this study are as follows.
(1) The static load capacity W, the side leakage Q
s
, the circum-
ferential ow rate Q
c
, and the total friction force F
f
for all
kinds of bearings increase with the increase in eccentricity
ratio e.
(2) For the stability threshold m
c
, those of HGJB are much larger
than PJB at low eccentricity ratio, but they are nearly the same
beyond the threshold.
(3) For a HGJB with journal rotating along the inverse direction,
the stability threshold m
c
is negative, which means that the
rotor-bearing system is unstable.
(4) The bearing clearance c has no effect on the static load
capacity W, the attitude angle f
o
, the stability threshold m
c
,
and the critical whirl ratio l
c
.
(5) From the above analyses, the optimum design parameters can
be obtained. The hydrodynamic bearing with herringbone
grooved sleeve having a1/2, b401, and g1.0 will be the
best choice for L/D1 and light loading applications irrespec-
tively of the value of clearance.
Acknowledgment
The authors greatly appreciate the support of this work by
Grants NSC 92-2212-E-033-016 from the National Science Council
0
5
10
15
W
= -20
= -40
= -60
= -70
30
50
70
90
0
= -20
= -40
= -60
= -70
0
100
200
300
s
Q
= -20
= -40
= -60
= -70
4
(10 )
0.8
1.0
1.4
c
Q
= -20
= -40
= -60
= -70
50
100
150
200
f
F
= -20
= -40
= -60
= -70
10
2
10
3
10
4
1
10
f
= -20
= -40
= -60
= -70
-125
-10
0
10
20
c
m
= -20
= -40
= -60
= -70
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.45
0.50
0.55
0.60
c
= -20
= -40
= -60
= -70
stability
instability
instability
stability
PJB
PJB
PJB
PJB
PJB
PJB
PJB
PJB
Fig. 13. Characteristics of various bearing design versus eccentricity ratio (e) with clearance 6 mm for PJB and HGJB with a1/2, g0.5 and minus groove angle b. (a) Static
load capacity (W ) and attitude angle (f
0
). (b) Dimensionless side leakage (Q
s
) and circumferential ow (Q
c
). (c) Dimensionless friction force (F
f
) and friction coefcient (m
f
).
(d) Dynamic stability threshold (m c
) and whirl frequency ratio (l
c
).
S.-K. Chen et al. / Tribology International 55 (2012) 1528 26
and MEA 98-EC-17-A-01-02-0704 from the Ministry of Economic
Affairs, Taiwan, ROC.
Appendix A. The coefcients of Eq. (13) for P
0
, P
e
, and P
f
The details about coefcients A
k,ij
to F
k,ij
of ve equations for
P
0
, P
R
e
, P
I
e
, P
R
f
, and P
I
f
expressed by Eq. (13) are given as follows.
1. Coefcients of P
0
:
_
G
ij

D
L
_ _
2
h
3
w
@P
0
@z
_ _
dy
D
L
_ _
2
P
0

i,j
P
0

i,j1
_ _
1
Dz
h
3
0

i1=4, j1=2
h
3
0

i 1=4, j1=2
_ _
Dy
2

D
L
_ _
2
P
0

i,j 1
P
0

i,j
_ _

1
Dz
h
3
0

i 1=4, j 1=2
h
3
0

i 1=4, j 1=2
_ _
Dy
2
A1a
_
G
ij
h
3
w
@P
0
@y
_ _
dz P
0

i,j
P
0

i1,j
_ _
1
Dy
h
3
0

i1=2,j1=2
h
3
0

i1=2,j 1=2
_ _
Dz
2
P
0

i 1,j
P
0

i,j
_ _
1
Dy
h
3
0

i 1=2,j1=2
_
h
3
0

i 1=2,j 1=2
_
Dz
2
A1b
__
O
ij
L
@h
w
@y
dy dz L
_
j 1=2
j1=2
_
i 1=2
i1=2
@h
0
@y
dy
_ _
dz
L
Dz
2
h
0

i 1=2, j 1=2
h
0

i1=2, j 1=2
_
h
0

i 1=2, j1=2
h
0

i1=2, j1=2
A1c
Substituting Eqs. (A1a)(A1c) into Eq. (10) and rearranging it
gives Eq. (13) with the following coefcients:
A
0,ij
h
3
0

i1=2, j 1=2
h
3
0

i1=2, j1=2
C
0,ij
h
3
0

i 1=2, j 1=2
h
3
0

i 1=2, j1=2
D
0,ij

D
L
_ _
2
Dy
Dz
_ _
2
h
3
0

i1=4, j1=2
h
3
0

i 1=4, j1=2
_ _
E
0,ij

D
L
_ _
2
Dy
Dz
_ _
2
h
3
0

i1=4, j 1=2
h
3
0

i 1=4, j 1=2
_ _
B
0,ij
A
0,ij
C
0,ij
D
0,ij
E
0,ij

F
0,ij
Lh
0

i 1=2, j 1=2
h
0

i 1=2, j1=2
h
0

i1=2, j 1=2
h
0

i1=2, j1=2
Dz
2. Coefcients of P
e
:
_
G
ij

D
L
_ _
2
h
3
w
@P
e
@z
dy
D
L
_ _
2
P
e

i,j
P
e

i,j1
_ _
1
Dz
h
3
0

i1=4, j1=2
h
3
0

i 1=4, j1=2
_ _
Dy
2

D
L
_ _
2
P
e

i,j 1
P
e

i,j
_ _
1
Dz
h
3
0

i1=4, j 1=2
h
3
0

i 1=4, j 1=2
_ _
Dy
2
A2a
_
G
ij
3
D
L
_ _
2
h
2
w
cos
~
y
@P
0
@z
dy 3
D
L
_ _
2
P
0

i,j
P
0

i,j1
_ _
1
Dz
h
2
0
cos
~
y
i1=4, j1=2
h
2
0
cos
~
y
i 1=4, j1=2
_ _
Dy
2
3
D
L
_ _
2
P
0

i,j 1
P
0

i,j
_ _
1
Dz
h
2
0
cos
~
y
i1=4, j 1=2
h
2
0
cos
~
y
i 1=4, j 1=2
_ _
Dy
2
A2b
_
G
ij
h
3
w
@P
e
@y
dz P
e

i,j
P
e

i1,j
_ _
1
Dy
h
3
0

i1=2, j1=2
_
h
3
0

i1=2, j 1=2
_
Dz
2
P
e

i 1,j
P
e

i,j
_ _
1
Dy
h
3
0

i 1=2, j1=2
h
3
0

i 1=2, j 1=2
_ _
Dz
2
A2c
_
G
ij
3h
2
w
cos
~
y
@P
0
@y
dz P
0

i,j
P
0

i1,j
_ _
1
Dy
3h
2
0
cos
~
y
i1=2, j1=2
_
3h
2
0
cos
~
y
i1=2, j 1=2
_
Dz
2
P
0

i 1,j
P
0

i,j
_ _
1
Dy
3h
2
0
cos
~
y
i 1=2, j1=2
_
3h
2
0
cos
~
y
i 1=2, j 1=2
_
Dz
2
A2d
__
O
ij
L sin
~
y dy dz L
_
j 1=2
j1=2
_
i 1=2
i1=2
sin
~
y dy
_ _
dz
L
Dz
2
cos
~
y
i 1=2, j 1=2
cos
~
y
i1=2, j 1=2
_
cos
~
y
i 1=2, j1=2
cos
~
y
i1=2, j1=2
A2e
__
O
ij
2L
c
d
o
cos
~
y dy dz 2L
c
d
o
_
j 1=2
j1=2
_
i 1=2
i1=2
cos
~
y dy
_ _
dz
L
c
d
o
Dz sin
~
y
i 1=2, j 1=2
sin
~
y
i1=2, j 1=2
_
sin
~
y
i1=2, j1=2
sin
~
y
i1=2, j1=2
A2f
i
__
O
ij
2L
o
p
o
cos
~
y dy dz i2L
o
p
o
_
j 1=2
j1=2
_
i 1=2
i1=2
cos
~
y dy
_ _
dz
iL
o
p
o
Dz sin
~
y
i 1=2, j 1=2
sin
~
y
i1=2, j 1=2
_
sin
~
y
i 1=2, j1=2
sin
~
y
i1=2, j1=2
A2g
Substituting Eqs. (A2a)(A2g) into Eq. (11) and rearranging it
gives Eq. (13) with the following coefcients for real and imagin-
ary parts of P
e
:
1. for real part P
R
e
:
A
R
e,ij
A
0,ij
, B
R
e,ij
B
0,ij
, C
R
e,ij
C
0,ij
, D
R
e,ij
D
0,ij
, E
R
e,ij
E
0,ij
F
R
e,ij
G
R
e,ij
S
R
e,ij
P
0

i1,j
T
R
e,ij
P
0

i,j
U
R
e,ij
P
0

i 1,j
V
R
e,ij
P
0

i,j1
Z
R
e,ij
P
0

i,j 1
G
R
e,ij
L cos
~
y
i 1=2, j 1=2
cos
~
y
i 1=2, j1=2
_
cos
~
y
i1=2, j 1=2
cos
~
y
i1=2, j1=2
_
Dy
L
c
d
o
sin
~
y
i 1=2, j 1=2
sin
~
y
i 1=2, j1=2
_
sin
~
y
i1=2, j 1=2
sin
~
y
i1=2, j1=2
_
Dy
S
R
e,ij
3 h
2
0
cos
~
y
i1=2, j 1=2
h
2
0
cos
~
y
i1=2, j1=2
_ _
U
R
e,ij
3 h
2
0
cos
~
y
i 1=2, j 1=2
h
2
0
cos
~
y
i 1=2, j1=2
_ _
Z
R
e,ij
3
D
L
_ _
2
Dy
Dz
_ _
2
h
2
0
cos
~
y
i1=4, j 1=2
h
2
0
cos
~
y
i 1=4, j 1=2
_ _
Z
R
e,ij
3
D
L
_ _
2
Dy
Dz
_ _
2
h
2
0
cos
~
y
i1=4, j 1=2
h
2
0
cos
~
y
i 1=4, j 1=2
_ _
T
R
e,ij
S
R
e,ij
U
R
e,ij
V
R
e,ij
Z
R
e,ij

2. for imaginary part P

I
e
:
A
I
e,ij
A
0,ij
, B
I
e,ij
B
0,ij
, C
I
e,ij
C
0,ij
, D
I
e,ij
D
0,ij
, E
I
e,ij
E
0,ij
S.-K. Chen et al. / Tribology International 55 (2012) 1528 27
F
I
e,ij
L
o
p
o
sin
~
y
i 1=2, j 1=2
sin
~
y
i 1=2, j1=2
_
sin
~
y
i1=2, j 1=2
sin
~
y
i1=2, j1=2
Dy
3. Coefcients of P
f
:
_
G
ij

D
L
_ _
2
h
3
w
@P
f
@z
dy
D
L
_ _
2
P
f

i,j
P
f

i,j1
_ _
1
Dz
h
3
0

i1=4, j1=2
h
3
0

i 1=4, j1=2
_ _
Dy
2

D
L
_ _
2
P
f

i,j 1
P
f

i,j
_ _
1
Dz
h
3
0

i1=4, j 1=2
h
3
0

i 1=4, j 1=2
_ _
Dy
2
A3a
_
G
ij
3
D
L
_ _
2
h
2
w
sin
~
y
@P
0
@z
dy 3
D
L
_ _
2
P
0

i,j
P
0

i,j1
_ _
1
Dz
h
2
0
sin
~
y
i1=4, j1=2
h
2
0
sin
~
y
i 1=4, j1=2
_ _
Dy
2
3
D
L
_ _
2
P
0

i,j 1
P
0

i,j
_ _
1
Dz
h
2
0
sin
~
y
i1=4, j 1=2
h
2
0
sin
~
y
i 1=4, j 1=2
_ _
Dy
2
A3b
_
G
ij
h
3
w
@P
f
@y
dz P
f

i,j
P
f

i1,j
_ _
1
Dy
h
3
0

i1=2, j1=2
_
h
3
0

i1=2, j 1=2
_
Dz
2
P
f

i 1,j
P
f

i,j
_ _
1
Dy
h
3
0

i 1=2, j1=2
h
3
0

i 1=2, j 1=2
_ _
Dz
2
A3c
_
Gij
3h
2
w
sin
~
y
@P
0
@y
dz P
0

i,j
P
0

i1,j
_ _
1
Dy
3h
2
0
sin
~
y
i1=2, j1=2
_
3h
2
0
sin
~
y
i1=2, j 1=2
_
Dz
2
P
0

i 1,j
P
0

i,j
_ _
1
Dy
3h
2
0
sin
~
y
i 1=2,j1=2
3h
2
0
sin
~
y
i 1=2,j 1=2
_ _
Dz
2
A3d
__
O
ij
L cos
~
y dy dz L
_
j 1=2
j1=2
_
i 1=2
i1=2
cos
~
y dy
_ _
dz
L
Dz
2
sin
~
y
i 1=2, j 1=2
sin
~
y
i1=2, j 1=2
_
sin
~
y
i 1=2, j1=2
sin
~
y
i1=2, j1=2
A3e
__
O
ij
2L
c
d
o
sin
~
y dy dz 2L
c
d
o
_
j 1=2
j1=2
_
i 1=2
i1=2
sin
~
y dy
_ _
dz
L
c
d
o
Dz cos
~
y
i 1=2, j 1=2
cos
~
y
i1=2, j 1=2
_
cos
~
y
i 1=2, j1=2
cos
~
y
i1=2, j1=2
A3f
i
__
O
ij
2L
o
p
o
sin
~
y dy dz i2L
o
p
o
_
j 1=2
j1=2
_
i 1=2
i1=2
sin
~
y dy
_ _
dz
iL
o
p
o
Dz cos
~
y
i 1=2, j 1=2
cos
~
y
i1=2, j 1=2
_
cos
~
y
i 1=2, j1=2
cos
~
y
i1=2, j1=2
A3g
Substituting Eqs. (A3a)(A3g) into Eq. (12) and rearranging it
gives Eq. (13) with the following coefcients for real and imagin-
ary parts of P
f
:
1. for real part P
R
f
:
A
R
f,ij
A
0,ij
, B
R
f,ij
B
0,ij
, C
R
f,ij
C
0,ij
, D
R
f,ij
D
0,ij
, E
R
f,ij
E
0,ij
F
R
f,ij
G
R
f,ij
S
R
f,ij
P
0

i1,j
T
R
f,ij
P
0

i,j
U
R
f,ij
P
0

i 1,j
V
R
f,ij
P
0

i,j1
Z
R
f,ij
P
0

i,j 1
G
R
f,ij
L sin
~
y
i 1=2, j 1=2
sin
~
y
i 1=2, j1=2
_
sin
~
y
i1=2, j 1=2
sin
~
y
i1=2, j1=2
_
Dy
L
c
d
o
cos
~
y
i 1=2, j 1=2
cos
~
y
i1=2, j 1=2
_
cos
~
y
i 1=2, j1=2
cos
~
y
i1=2, j1=2
_
Dy
S
R
f,ij
3 h
2
0
sin
~
y
i1=2, j 1=2
h
2
0
sin
~
y
i1=2, j1=2
_ _
U
R
f,ij
3 h
2
0
sin
~
y
i 1=2, j 1=2
h
2
0
sin
~
y
i 1=2, j1=2
_ _
V
R
f,ij
3
D
L
_ _
2
Dy
Dz
_ _
2
h
2
0
sin
~
y
i1=4, j1=2
_
h
2
0
sin
~
y
i 1=4, j1=2
_
Z
R
f,ij
3
D
L
_ _
2
Dy
Dz
_ _
2
h
2
0
sin
~
y
i1=4, j 1=2
_
h
2
0
sin
~
y
i 14, j 1=2

T
R
f,ij
S
R
f,ij
U
R
f,ij
V
R
f,ij
Z
R
f,ij

2. for imaginary part P

I
f
:
A
I
f,ij
A
0,ij
, B
I
f,ij
B
0,ij
, C
I
f,ij
C
0,ij
, D
I
f,ij
D
0,ij
, E
I
f,ij
E
0,ij
F
I
f,ij
L
o
p
o
cos
~
y
i 1=2, j 1=2
cos
~
y
i 1=2, j1=2
_
cos
~
y
i1=2, j 1=2
cos
~
y
i1=2, j1=2
Dy
References
[1] Hirs GG. The load capacity and stability characteristics of hydrodynamic
grooved journal bearings. ASME Transactions 1965;8:296305.
[2] Bootsma J. The gas-to-liquid interface of spiral groove journal bearings and
its effect on stability. ASME Transactions, Journal of Lubrication Technology
1974:33745.
[3] Bootsma J. Spherical and conical spiral groove bearingsPart II: load capacity
and stability. ASME Transactions, Journal of Lubrication Technology 1975:
2439.
[4] Kang K, Rhim Y, Sung K. A study of the oil lubricated herringbone grooved
journal bearingPart 1: numerical analysis. Journal of Tribology 1996;118:
90611.
[5] Jang GH, Yoon JW. Nonlinear dynamic analysis of a hydrodynamic journal
bearing considering the effect of rotating or stationary groove. Journal of
Tribology 2002;124:297304.
[6] Jang GH, Yoon JW. Stability analysis of a hydrodynamic journal bearing with
rotating herringbone grooves. Journal of Tribology 2003;125:291300.
[7] Wan JM, Lee TS, Shu C, Wu JKA. Numerical study of cavitation foot-prints in
liquid-lubricated asymmetrical herringbone grooved journal bearings. Inter-
national Journal of Numerical Methods for Heat & Fluid Flow 2002;12(5):
51840.
[8] Wu JK, Li AF, Lee TS, Shu C, Wan JM. Operator-splitting method for the
analysis of cavitation in liquid-lubricated herringbone grooved journal
bearings. International Journal for Numerical Methods in Fluids 2004;44:
76575.
[9] Lee TS, Liu YG, Winoto SH. Analysis of liquid-lubricated herringbone grooved
journal bearings. International Journal for Numerical Methods for Heat &
Fluid Flow 2004;14(3):34165.
[10] Elrod HG. A cavitation algorithm. ASME Transactions, Journal of Lubrication
Technology 1981;103:3504.
[11] Rao TVVLN, Sawicki JT. Stability characteristics of herringbone grooved
journal bearings incorporating cavitation effects. Journal of Tribology 2004;126:
2817.
[12] Sun GS, Feng M. Design of hydrodynamic and hydrostatic hybrid-bearing.
Beijing: World Publishing Co; 1993 [in Chinese].
S.-K. Chen et al. / Tribology International 55 (2012) 1528 28