6

Long Hydrodynamic Journal

Bearings

6.1 INTRODUCTION
A hydrodynamic journal bearing is shown in Fig. 6-1. The journal is rotating
inside the bore of a sleeve with a thin clearance. Fluid lubricant is continuously
supplied into the clearance. If the journal speed is sufficiently high, pressure
builds up in the fluid film that completely separates the rubbing surfaces.
Hydrodynamic journal bearings are widely used in machinery, particularly
in motor vehicle engines and high-speed turbines. The sleeve is mounted in a
housing that can be a part of the frame of a machine. For successful operation, the
bearing requires high-precision machining. For most applications, the mating
surfaces of the journal and the internal bore of the bearing are carefully made with
precise dimensions and a good surface finish.
The radial clearance, C, between the bearing bore of radius R1 and of the
journal radius, R, is C ¼ R1
R (see Fig. 6-1). In most practical cases, the
clearance, C, is very small relative to the journal radius, R. The following order of
magnitude is applicable for most design purposes:
C
 10
3 ð6-1Þ
R
A long hydrodynamic bearing is where the bearing length, L, is long in
comparison to the journal radius ðL  RÞ. Under load, the center of the journal,

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 F IG. 6-1 Hydrodynamic journal bearing (clearance exaggerated).

O1 , is displaced in the radial direction relative to the bearing center, O, as

indicated in Fig. 6-1. The distance O
O1 is the eccentricity, e, and the
dimensionless eccentricity ratio, e, is defined as
e
E¼ ð6-2Þ
C
If the journal is concentric to the bearing bore, there is a uniform clearance
around the bearing. But due to the eccentricity, there is a variable-thickness
clearance around the bearing. The variable clearance, h, is equal to the lubricant
film thickness. The clearance is converging (h decreases) along the region from
y ¼ 0 to y ¼ p, where the clearance has its minimum thickness, hn . After that, the
clearance is diverging (h increases) along the region from y ¼ p to y ¼ 2p, where
the clearance has a maximum thickness, hm . The clearance, h, is a function of the
coordinate y in the direction of the rotation of the journal. The coordinate y is
measured from the point of maximum clearance on the centerline O O. An
approximate equation for the thin clearance in a journal bearing is
hðyÞ ¼ Cð1 þ e cos yÞ ð6-3Þ
The minimum and maximum clearances are hn and hm, respectively, along the
symmetry line O O1 . They are derived from Eq. (6-3) as follows:
hn ¼ Cð1
eÞ and hm ¼ Cð1 þ eÞ ð6-4Þ
Whenever e approaches 1 (one), hn approaches zero, and there is an undesirable
contact between the journal and bearing bore surfaces, resulting in severe wear,

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 particularly in a high-speed journal bearing. Therefore, for proper design, the
bearing must operate at eccentricity ratios well below 1, to allow adequate
minimum film thickness to separate the sliding surfaces. Most journal bearings
operate under steady conditions in the range from e ¼ 0:6 to e ¼ 0:8. However,
the most important design consideration is to make sure the minimum film
thickness, hn , will be much higher than the size of surface asperities or the level
of journal vibrations during operation.

6.2 REYNOLDS EQUATION FOR A JOURNAL

BEARING
In Chapter 4, the hydrodynamic equations of a long journal bearing were derived
from first principles. In this chapter, the hydrodynamic equations are derived from
the Reynolds equation. The advantage of the present approach is that it can apply
to a wider range of problems, such as bearings under dynamic conditions.
Let us recall that the general Reynolds equation for a Newtonian incom-
pressible thin fluid film is



@ h3 @p @ h3 @p @h
þ ¼ 6ðU1
U2 Þ þ 12ðV2
V1 Þ ð6-5Þ
@x m @x @z m @z @x

Here, U1 and U2 are velocity components, in the x direction, of the lower and
upper sliding surfaces, respectively (fluid film boundaries), while the velocities
components V1 and V2 are of the lower and upper boundaries, respectively, in the
y direction (see Fig. 5-2). The difference in normal velocity ðV1
V2 Þ is of
relative motion (squeeze-film action) of the surfaces toward each other.
For most journal bearings, only the journal is rotating and the sleeve is
stationary, U1 ¼ 0; V1 ¼ 0 (as shown in Fig. 6-1). The second fluid film boundary
is at the journal surface that has a velocity U ¼ oR. However, the velocity U is
not parallel to the x direction (the x direction is along the bearing surface).

F IG. 6-2 Velocity components of the fluid film boundaries.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 Therefore, it has two components (as shown in Fig. 6-2), U2 and V2, in the x and y
directions, respectively:
U2 ¼ U cos a
ð6-6Þ
V2 ¼
U sin a
Here, the slope a is between the bearing and journal surfaces. In a journal
bearing, the slope a is very small; therefore, the following approximations can be
applied:
U2 ¼ U cos a  U
ð6-7Þ
V2 ¼
U sin a 
U tan a
The slope a can be expressed in terms of the function of the clearance, h:
@h
tan a ¼
ð6-8Þ
@x
The normal component V2 becomes
@h
V2  U ð6-9Þ
@x
After substituting Eqs. (6.7) and (6.9) into the right-hand side of the Reynolds
equation, it becomes
@h @h @h @h
6ðU1
U2 Þ þ 12ðV2
V1 Þ ¼ 6ð0
U Þ þ 12U ¼ 6U ð6-10Þ
@x @x @x @x
The Reynolds equation for a Newtonian incompressible fluid reduces to the
following final equation:



@ h3 @p @ h3 @p @h
þ ¼ 6U ð6-11Þ
@x m @x @z m @z @x
For an infinitely long bearing, @p=@z ffi 0; therefore, the second term on the left-
hand side of Eq. (6-11) can be omitted, and the Reynolds equation reduces to the
following simplified one-dimensional equation:


@ h3 @p @h
¼ 6U ð6-12Þ
@x m @x @x

6.3 JOURNAL BEARING WITH ROTATING

SLEEVE
There are many practical applications where the sleeve is rotating as well as the
journal. In that case, the right-hand side of the Reynolds equation is not the same
as for a common stationary bearing. Let us consider an example, as shown in Fig.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 6-3, where the sleeve and journal are rolling together like internal friction pulleys.
The rolling is similar to that in a cylindrical rolling bearing. The internal sleeve
and journal surface are rolling together without slip, and both have the same
tangential velocity oj R ¼ ob R1 ¼ U.
The tangential velocities of the film boundaries, in the x direction, of the
two surfaces are
U1 ¼ U
ð6-13Þ
U2 ¼ U cos a  U
The normal velocity components, in the y direction, of the film boundaries (the
journal and sleeve surfaces) are
V1 ¼ 0
@h ð6-14Þ
V2  U
@x
The right-hand side of the Reynolds equation becomes
@h @h @h @h
6ðU1
U2 Þ þ 12ðV2
V1 Þ ¼ 6ðU
U Þ þ 12ðU
0Þ ¼ 12U
@x @x @x @x
ð6-15Þ
For the rolling action, the Reynolds equation for a journal bearing with a rolling
sleeve is given by



@ h3 @p @ h3 @p @h
þ ¼ 12U ð6-16Þ
@x m @x @z m @z @x
The right-hand side of Eq. (6-16) indicates that the rolling action will result in a
doubling of the pressure wave of a common journal bearing of identical geometry
as well as a doubling its load capacity. The physical explanation is that the fluid is
squeezed faster by the rolling action.

6.4 COMBINED ROLLING AND SLIDING

In many important applications, such as gears, there is a combination of rolling
and sliding of two cylindrical surfaces. Also, there are several unique applications
where a hydrodynamic journal bearing operates in a combined rolling-and-sliding
mode. A combined bearing is shown in Fig. 6-3, where the journal surface
velocity is Roj while the sleeve inner surface velocity is R1 ob .
The coefficient x is the ratio of the rolling and sliding velocity. In terms of
the velocities of the two surfaces, the ratio is,
R1 ob
x¼ ð6-17Þ
Roj

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 F IG. 6.3 Journal bearing with a rotating sleeve.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 The journal surface velocity is U ¼ Roj , and the sleeve surface velocity is the
product xU , where x is the rolling-to-sliding ratio. The common journal bearing
has a pure sliding and x ¼ 0, while in pure rolling, x ¼ 1. For all other
combinations, 0 < x < 1.
The tangential velocities (in the x direction) of the fluid–film boundaries of
the two surfaces are

U1 ¼ R1 ob ¼ xRoj
ð6-18Þ
U2 ¼ Roj cos a  Roj

The normal components, in the y direction, of the velocity of the fluid film
boundaries (journal and sleeve surfaces) are

V1 ¼ 0
@h ð6-19Þ
V2 ¼ Roj
@x
For the general case of a combined rolling and sliding, the expression on the
right-hand side of Reynolds equation is obtained by substituting the preceding
components of the surface velocity:

@h @h @h
6ðU1
U2 Þ þ 12ðV2
V1 Þ ¼ 6ðxU
U Þ þ 12ðU
0Þ
@x @x @x
@h
¼ 6U ð1 þ xÞ ð6-20Þ
@x
Here, U ¼ Roj is the journal surface velocity. Finally, the Reynolds equation for
a combined rolling and sliding of a journal bearing is as follows:



@ h3 @p @ h3 @p @h
þ ¼ 6Roj ð1 þ xÞ ð6-21aÞ
@x m @x @z m @z @x

For x ¼ 0 and x ¼ 1, the right-hand-side of the Reynolds equation is in

agreement with the previous derivations for pure sliding and pure rolling,
respectively. The right-hand side of Eq. (6-21a) indicates that pure rolling
action doubles the pressure wave in comparison to a pure sliding.
Equation (6-21a) is often written in the basic form:



@ h3 @p @ h3 @p @h
þ ¼ 6Rðoj þ ob Þ ð6-21bÞ
@x m @x @z m @z @x

In journal bearings, the difference between the journal radius and the bearing
radius is small and we can assume that R1  R.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 6.5 PRESSURE WAVE IN A LONG JOURNAL
BEARING
For a common long journal bearing with a stationary sleeve, the pressure wave is
derived by a double integration of Eq. (6.12). After the first integration, the
following explicit expression for the pressure gradient is obtained:

dp h þ C1
¼ 6U m ð6-22Þ
dx h3

Here, C1 is a constant of integration. In this equation, a regular derivative replaces

the partial one, because in a long bearing, the pressure is a function of one
variable, x, only. The constant, C1 , can be replaced by h0 , which is the film
thickness at the point of peak pressure. At the point of peak pressure,

dp
¼0 at h ¼ h0 ð6-23Þ
dx

Substituting condition (6-23), in Eq. (6-22) results in C1 ¼
h0 , and Eq. (6-22)
becomes

dp h
h0
¼ 6U m ð6-24Þ
dx h3

Equation (6-24) has one unknown, h0 , which is determined later from additional
information about the pressure wave.
The expression for the pressure distribution (pressure wave) around a
journal bearing, along the x direction, is derived by integration of Eq. (6-24),
and there will be an additional unknown—the constant of integration. By using
the two boundary conditions of the pressure wave, we solve the two unknowns,
h0 , and the second integration constant.
The pressures at the start and at the end of the pressure wave are usually
used as boundary conditions. However, in certain cases the locations of the start
and the end of the pressure wave are not obvious. For example, the fluid film of a
practical journal bearing involves a fluid cavitation, and other boundary condi-
tions of the pressure wave are used for solving the two unknowns. These
boundary conditions are discussed in this chapter. The solution method of two
unknowns for these boundary conditions is more complex and requires computer
iterations.
Replacing x by an angular coordinate y, we get

x ¼ Ry ð6-25Þ

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 Equation (6-24) takes the form

dp h
h0
¼ 6URm ð6-26Þ
dy h3

For the integration of Eq. (6-26), the boundary condition at the start of the
pressure wave is required. The pressure wave starts at y ¼ 0, and the magnitude
of pressure at y ¼ 0 is p0 . The pressure, p0 , can be very close to atmospheric
pressure, or much higher if the oil is fed into the journal bearing by an external
pump.
The film thickness, h, as a function of y for a journal bearing is given in Eq.
(6-3), hðyÞ ¼ Cð1 þ e cos yÞ. After substitution of this expression into Eq. (6-26),
the pressure wave is given by
ðy ðy
C2 dy h dy
ðp
p0 Þ ¼ 2

0 ð6-27Þ
6mUR 0 ð1 þ e cos yÞ C 0 ð1 þ e cos yÞ3

The pressure, p0 , is determined by the oil supply pressure (inlet pressure). In a

common journal bearing, the oil is supplied through a hole in the sleeve, at y ¼ 0.
The oil can be supplied by gravitation from an oil container or by a high-pressure
pump. In the first case, p0 is only slightly above atmospheric pressure and can be
approximated as p0 ¼ 0. However, if an external pump supplies the oil, the pump
pressure (at the bearing inlet point) determines the value of p0 . In industry, there
are often central oil circulation systems that provide oil under pressure for the
lubrication of many bearings.
The two integrals on the right-hand side of Eq. (6-27) are functions of the
eccentricity ratio, e. These integrals can be solved by numerical or analytical
integration. Sommerfeld (1904) analytically solved these integrals for a full
(360 ) journal bearing. Analytical solutions of these two integrals are included in
integral tables of most calculus textbooks. For a full bearing, it is possible to solve
for the load capacity even in cases where p0 can’t be determined. This is because
p0 is an extra constant hydrostatic pressure around the bearing, and, similar to
atmospheric pressure, it does not contribute to the load capacity.
In a journal bearing, the pressure that is predicted by integrating Eq. (6-27)
increases (above the inlet pressure, p0 ) in the region of converging clearance,
0 < y < p. However, in the region of a diverging clearance, p < y < 2p, the
pressure wave reduces below p0 . If the oil is fed at atmospheric pressure, the
pressure wave that is predicted by integration of Eq. (6-27) is negative in the
divergent region, p < y < 2p. This analytical solution is not always valid,
because a negative pressure would result in a ‘‘fluid cavitation.’’ The analysis
may predict negative pressures below the absolute zero pressure, and that, of
course, is physically impossible.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 A continuous fluid film cannot be maintained at low negative pressure
(relative to the atmospheric pressure) due to fluid cavitation. This effect occurs
whenever the pressure reduces below the vapor pressure of the oil, resulting in an
oil film rupture. At low pressures, the boiling process can take place at room
temperature; this phenomenon is referred to as fluid cavitation. Moreover, at low
pressure, the oil releases its dissolved air, and the oil foams in many tiny air
bubbles. Antifoaming agents are usually added into the oil to minimize this effect.
As a result of cavitation and foaming, the fluid is not continuous in the
divergent clearance region, and the actual pressure wave cannot be predicted
anymore by Eq. (6-27). In fact, the pressure wave is maintained only in the
converging region, while the pressure in most of the diverging region is close to
ambient pressure. Cole and Hughes (1956) conducted experiments using a
transparent sleeve. Their photographs show clearly that in the divergent region,
the fluid film ruptures into filaments separated by air and lubricant vapor.
Under light loads or if the supply pressure, p0 , is high, the minimum
predicted pressure in the divergent clearance region is not low enough to generate
a fluid cavitation. In such cases, Eq. (6-27) can be applied around the complete
journal bearing. The following solution referred to as the Sommerfeld solution, is
limited to cases where there is a full fluid film around the bearing.

6.6 SOMMERFELD SOLUTION OF THE

PRESSURE WAVE
Sommerfeld (1904) solved Eq. (6-27) for the pressure wave and load capacity of a
full hydrodynamic journal bearing (360 ) where a fluid film is maintained around
the bearing without any cavitation. This example is of a special interest because
this was the first analytical solution of a hydrodynamic journal bearing based on
the Reynolds equation. In practice, a full hydrodynamic lubrication around the
bearing is maintained whenever at least one of the following two conditions are
met:
a. The feed pressure, p0 , (from an external oil pump), into the bearing is
quite high in order to maintain positive pressures around the bearing
and thus prevent cavitation.
b. The journal bearing is lightly loaded. In this case, the minimum
pressure is above the critical value of cavitation.
Sommerfeld assumed a periodic pressure wave around the bearing; namely, the
pressure is the same at y ¼ 0 and y ¼ 2p:

pðy¼0Þ ¼ pðy¼2pÞ ð6-28Þ

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 The unknown, h0 , that represents the film thickness at the point of a peak pressure
can be solved from Eq. (6-27) and the Sommerfeld boundary condition in Eq.
(6-28). After substituting p
p0 ¼ 0, at y ¼ 2p, Eq. (6-27) yields
ð 2p ð
dy h0 2p dy
2

¼0 ð6-29Þ
0 ð1 þ e cos yÞ C 0 ð1 þ e cos yÞ3
This equation can be solved for the unknown, h0 . The following substitutions for
the values of the integrals can simplify the analysis of hydrodynamic journal
bearings:
ð 2p
dy
Jn ¼ n ð6-30Þ
0 ð1 þ e cos yÞ
ð 2p
cos y dy
In ¼ n ð6-31Þ
0 ð1 þ e cos yÞ

Equation (6-29) is solved for the unknown, h0 , in terms of the integrals Jn :

h0 J 2
¼ ð6-32Þ
C J3
Here, the solutions for the integrals Jn are:
2p
J1 ¼ ð6-33Þ
ð1
e2 Þ1=2
2p
J2 ¼ ð6-34Þ
ð1
e2 Þ3=2


1 2p
J3 ¼ 1 þ e2 ð6-35Þ
2 ð1
e2 Þ5=2


3 2p
J4 ¼ 1 þ e2 ð6-36Þ
2 ð1
e2 Þ7=2
The solutions of the integrals In are required later for the derivation of the
expression for the load capacity. The integrals In can be obtained from Jn by the
following equation:
Jn
1
Jn
In ¼ ð6-37Þ
e
Sommerfeld solved for the integrals in Eq. (6-27), and obtained the following
equation for the pressure wave around an infinitely long journal bearing with a
full film around the journal bearing:
6mUR eð2 þ e cos yÞ sin y
p
p0 ¼ ð6-38Þ
C 2 ð2 þ e2 Þð1 þ e cos yÞ2

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 The curves in Fig. 6-4 are dimensionless pressure waves, relative to the
inlet pressure, for various eccentricity ratios, e. The pressure wave is an
antisymmetrical function on both sides of y ¼ p. The curves indicate that the
peak pressure considerably increases with the eccentricity ratio, e. According to
Eq. (6-38), the peak pressure approaches infinity when e approaches 1. However,
this is not possible in practice because the surface asperities prevent a complete
contact between the sliding surfaces.

6.7 JOURNAL BEARING LOAD CAPACITY

Figure 6-5 shows the load capacity, W , of a journal bearing and its two
components, Wx and Wy . The direction of Wx is along the bearing symmetry
line O O1 . This direction is inclined at an attitude angle, f, from the direction of
the external force and load capacity, W . In Fig. 6-5 the external force is in a
vertical direction. The direction of the second component, Wy is normal to the Wx
direction.
The elementary load capacity, dW , acts in the direction normal to the
journal surface. It is the product of the fluid pressure, p, and an elementary area,

F IG. 6-4 Pressure waves in an infinitely long, full bearing according to the Sommerfeld
solution.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 F IG. 6-5 Hydrodynamic bearing force components.

dA ¼ LR dy, of the journal surface bounded by a small journal angle, dy. An

elementary fluid force (dW ¼ p dA) is given by,

dW ¼ pLR dy ð6-39Þ

The pressure is acting in the direction normal to the journal surface, and dW is a
radial elementary force vector directed toward the journal center, as shown in Fig.
6-5.
In a plane-slider, the pressure is acting in one direction and the load
capacity has been derived by integration of the pressure wave. However, in a
journal bearing, the direction of the pressure varies around the bearing, and
simple summation of the elementary forces, dW , will not yield the resultant force.
In order to allow summation, the elementary force, dW , is divided into two
components, dWx and dWy, in the directions of Wx and Wy, respectively. By using
force components, it is possible to have a summation (by integration) of each
component around the bearing. The elementary force components are:

dWx ¼
pLR cos y dy ð6-40Þ

dWy ¼ pLR sin y dy ð6-41Þ

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 The two components of the load capacity, Wx and Wy, are in the X and
Y directions, respectively, as indicated in Fig. 6-5. Note the negative sign in
Eq. (6-40), since dWx is opposite to the Wx direction. The load components are:
ð 2p
Wx ¼
LR p cos y dy ð6-42Þ
0
ð 2p
Wy ¼ LR p sin y dy ð6-43Þ
0

The attitude angle f in Fig. 6-5 is determined by the ratio of the force
components:
Wy
tan f ¼ ð6-44Þ
Wx

6.8 LOAD CAPACITY BASED ON SOMMERFELD

CONDITIONS
The load capacity components can be solved by integration of Eqs. (6-42 and
(6-43), where p is substituted from Eq. (6-38). However, the derivation can be
simplified if the load capacity components are derived directly from the basic Eq.
(6-24) of the pressure gradient. In this way, there is no need to integrate the
complex Eq. (6-38) of the pressure wave. This can be accomplished by employing
the following identity for product derivation:
ðuvÞ0 ¼ uv0 þ vu0 ð6-45Þ
Integrating and rearranging Eq. (6-45) results in
ð ð
uv0 ¼ uv
u0 v ð6-46Þ

In order to simplify the integration of Eq. (6-42) for the load capacity component
Wx , the substitutions u ¼ p and v0 ¼ cos y are made. This substitution allows the
use of the product rule in Eq. (6-46), and the integral in Eq. (6-42) results in the
following terms:
ð ð
dp
p cos y dy ¼ p sin y
sin y dy ð6-47Þ
dy
In a similar way, for the load capacity component, Wy , in Eq. (6-43), the
substitutions u ¼ p and v0 ¼ sin y result in
ð ð
dp
p sin y dy ¼
p cos y þ cos y dy ð6-48Þ
dy

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 Equations (6-47) and (6-48) indicate that the load capacity components in Eqs.
(6-42) and (6-43) can be solved directly from the pressure gradient. By using this
method, it is not necessary to solve for the pressure wave in order to find the load
capacity components (it offers the considerable simplification of one simple
integration instead of a complex double integration). The first term, on the right-
hand side in Eqs. (6-47) and (6-48) is zero, when integrated around a full bearing,
because the pressure, p, is the same at y ¼ 0 and y ¼ 2p.
Integration of the last term in Eq. (6-47), in the boundaries y ¼ 0 and
y ¼ 2p, indicates that the load component, Wx , is zero. This is because it is an
integration of the antisymmetrical function around the bearing (the function is
antisymmetric on the two sides of the centerline O O1 , which cancel each other).
Therefore:

Wx ¼ 0 ð6-49Þ

Integration of Eq. (6-43) with the aid of identity (6-48), and using the value of h0
in Eq. 6-32 results in


6mUR2 L J2
Wy ¼ I2
I3 ð6-50Þ
C2 J3

Substituting for the values of In and Jn as a function of e yields the following

expression for the load capacity component, Wy . The other component is Wx ¼ 0;
therefore, for the Sommerfeld conditions, Wy is equal to the total load capacity,
W ¼ Wy :

12pmUR2 L e
W ¼ ð6-51Þ
C 2
ð2 þ e Þð1
e2 Þ1=2
2

The attitude angle, f, is derived from Eq. (6-44). For Sommerfeld’s conditions,
Wx ¼ 0 and tan f ! 1; therefore,
p
f¼ ð6-52Þ
2
Equation (6-52) indicates that in this case, the symmetry line O O1 is normal to
the direction of the load capacity W.

6.9 FRICTION IN A LONG JOURNAL BEARING

The bearing friction force, Ff , is the viscous resistance force to the rotation of the
journal due to high shear rates in the fluid film. This force is acting in the
tangential direction of the journal surface and results in a resistance torque to the

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 rotation of the journal. The friction force is defined as the ratio of the friction
torque, Tf , to the journal radius, R:
Tf
Ff ¼ ð6-53Þ
R
The force is derived by integration of the shear stresses over the area of the
journal surface, at y ¼ h, around the bearing. The shear stress distribution at the
journal surface (shear at the wall, tw ) around the bearing, is derived from the
velocity gradient, as follows:

du
tw ¼ m  ð6-54Þ
dy ðy¼hÞ

The friction force is obtained by integration:

ð
Ff ¼ tðy¼hÞ dA ð6-55Þ
A

After substituting dA ¼ RL dy in Eq. (6-55), the friction force becomes

ð 2p
Ff ¼ mRL tðy¼hÞ dy ð6-56Þ
0

Substitution of the value of the shear stress, Eq. (6-56) becomes

ð 2p

4 3h0
Ff ¼ mURL
2 dy ð6-57Þ
0 h h

If we apply the integral definitions in Eq. (6-30), the expression for the friction
force becomes


mRL J2
Ff ¼ 4 J1
3 2 ð6-58Þ
C J3

The integrals Jn are functions of the eccentricity ratio. Substituting the solution of
the integrals, Jn , in Eqs. (6.33) to (6.37) results in the following expression for the
friction force:
mURL 4pð1 þ 2e2 Þ
Ff ¼ ð6-59Þ
C ð2 þ e2 Þð1
e2 Þ1=2

Let us recall that the bearing friction coefficient, f , is defined as

Ff
f ¼ ð6-60Þ
W

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 Substitution of the values of the friction force and load in Eq. (6-60) results in a
relatively simple expression for the coefficient of friction of a long hydrodynamic
journal bearing:
C 1 þ 2e2
f ¼ ð6-61Þ
R 3e
Comment: The preceding equation for the friction force in the fluid film is based
on the shear at y ¼ h. The viscous friction force around the bearing bore surface,
at y ¼ 0, is not equal to that around the journal surface, at y ¼ h. The viscous
friction torque on the journal surface is unequal to that on the bore surface
because the external load is eccentric to the bore center, and it is an additional
torque. However, the friction torque on the journal surface is the actual total
resistance to the journal rotation, and it is used for calculating the friction energy
losses in the bearing.

6.10 POWER LOSS ON VISCOUS FRICTION

The energy loss, per unit of time (power loss) E_ f , is determined from the friction
torque, or friction force, by the following equations:
E_ f ¼ Tf o ¼ Ff U ð6-62Þ
where o ðrad=sÞ is the angular velocity of the journal. Substituting Eq. (6-59) into
Eq. (6-62) yields
mU 2 RL 4pð1 þ 2e2 Þ
E_ f ¼ ð6-63Þ
C ð2 þ e2 Þð1
e2 Þ1=2
The friction energy losses are dissipated in the lubricant as heat. Knowledge of
the amount of friction energy that is dissipated in the bearing is very important for
ensuring that the lubricant does not overheat. The heat must be transferred from
the bearing by adequate circulation of lubricant through the bearing as well as by
conduction of heat from the fluid film through the sleeve and journal.

6.11 SOMMERFELD NUMBER

Equation (6-51) is the expression for the bearing load capacity in a long bearing
operating at steady conditions with the Sommerfeld boundary conditions for the
pressure wave. This result was obtained for a full film bearing without any
cavitation around the bearing. In most practical cases, this is not a realistic
expression for the load capacity, since there is fluid cavitation in the diverging
clearance region of negative pressure. The Sommerfeld solution for the load
capacity has been improved by applying a more accurate analysis with realistic

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 boundary conditions of the pressure wave. This solution requires iterations
performed with the aid of a computer.
In order to simplify the design of hydrodynamic journal bearings, the
realistic results for the load capacity are provided in dimensionless form in tables
or graphs. For this purpose, a widely used dimensionless number is the
Sommerfeld number, S.
Equation (6-51) can be converted to dimensionless form if all the variables
with dimensions are placed on the left-hand side of the equations and the
dimensionless function of e on the right-hand side. Also, it is the tradition in
this discipline to have the Sommerfeld dimensionless group as a function of
journal speed, n, in revolutions per second, and the average bearing pressure, P,
according to the following substitutions:
U ¼ 2pRn ð6-64Þ
W ¼ 2RLP ð6-65Þ
By substituting Eqs. (6.64) and (6.65) into Eq. (6.51), we obtain the following
dimensionless form of the Sommerfeld number for an infinitely long journal
bearing where cavitation is disregarded:

2
mn R ð2 þ e2 Þð1
e2 Þ1=2
S¼ ¼ ð6-66Þ
P C 12p2 e
The dimensionless Sommerfeld number is a function of e only. For an infinitely
long bearing and the Sommerfeld boundary conditions, the number S can be
calculated via Eq. (6-66). However, for design purposes, we use the realistic
conditions for the pressure wave. The values of S for various ratios of length and
diameter, L=D, have been computed; the results are available in Chapter 8 in the
form of graphs and tables for design purposes.

6.12 PRACTICAL PRESSURE BOUNDARY

CONDITIONS
The previous discussion indicates that for most practical applications in machin-
ery the pressures are very high and cavitation occurs in the diverging region of the
clearance. For such cases, the Sommerfeld boundary conditions do not apply and
the pressure distribution can be solved for more realistic conditions. In an actual
journal bearing, there is no full pressure wave around the complete bearing, and
there is a positive pressure wave between y1 and y2 . Outside this region, there is
cavitation and low negative pressure that can be ignored for the purpose of
calculating the load capacity. The value of y2 is unknown, and an additional
condition must be applied for the solution. The boundary condition commonly
used is that the pressure gradient is zero at the end of the pressure wave, at y2 .

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 This is equivalent to the assumption that there is no flow, u, in the x direction at
y2 . When the lubricant is supplied at y1 at a pressure p0 , the following boundary
conditions can be applied:

p ¼ p0 at y ¼ y1
dp
¼ 0 at y ¼ y2 ð6-67Þ
dy
p ¼ 0 at y ¼ y2

In most cases, the feed pressure is close to ambient pressure at y ¼ 0, and the first
boundary condition is p ¼ 0 at y ¼ 0. The pressure wave with the foregoing
boundary conditions is shown in Fig. 6-6.
The location of the end of the pressure wave, y2 , is solved by iterations. The
solution is performed by guessing a value for y2 that is larger then 180 , then
integrating Eq. (6-27) at the boundaries from 0 to y2 . The solution is obtained
when the pressure at y2 is very close to zero.
Figure 6-6 indicates that the angle y2 and the angle of the peak pressure are
symmetrical on both sides of y ¼ p; therefore both have an equal film thickness h
(or clearance thickness). The constant, h0 , representing the film thickness at the
maximum pressure is

h0 ¼ Cð1 þ cos y2 Þ ð6-68Þ

After the integration is completed, the previous guess for y2 is corrected until a
satisfactory solution is obtained (the pressure at y2 is very close to zero).
Preparation of a small computer program to solve for y2 by iterations is
recommended as a beneficial exercise for the reader. After h0 is solved the
pressure wave can be plotted.

F IG. 6-6 Pressure wave plot under realistic boundary conditions.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 Example Problem 6-1
Ice Sled
Two smooth cylindrical sections support a sled as shown in Fig. 6-7. The sled is
running over ice on a thin layer of water film. The total load (weight of the sled
and the person) is 1000 N. This load is acting at equal distances between the two
blades. The sled velocity is 15 km=h, the radius of the blade is 30 cm, and its
width is L ¼ 90 cm. The viscosity of water m ¼ 1:792  10
3 N-s=m2.
a. Find the pressure distribution at the entrance region of the fluid film
under the ski blade as it runs over the ice. Derive and plot the pressure
wave in dimensionless form.
b. Find the expression for the load capacity of one blade.
c. Find the minimum film thickness ðhn ¼ hmin Þ of the thin water layer.

Solution
a. Pressure Distribution and Pressure Wave
For hn =R  1, the pressure wave is generated only near the minimum-film
region, where x  R, or x=R  1.
For a small value of x=R, the equation for the clearance, between the
quarter-cylinder and the ice, h ¼ hðxÞ, can be approximated by a parabolic wedge.
The following expression is obtained by expanding the equation for the variable
clearance, hðxÞ, which is equal to the fluid-film thickness, into a Taylor series and
truncating powers higher than x2 (see Chapter 4).
If the minimum thickness of the film is hn ¼ hmin , the equation for the
variable clearance becomes

x2
hðxÞ ¼ hn þ
2R

F IG. 6-7 Sled made of two quarter-cylinder.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 For dimensionless analysis, the clearance function is written as


x2
hðxÞ ¼ hn 1 þ
2Rhn
The width, L (in the direction normal to the sled speed), is very large in
comparison to the film length in the x direction. Therefore, it can be considered
an infinitely long bearing, and the pressure gradient is
dp h
h0
¼ 6mU
dx h3
The unknown, h0 , is the fluid film thickness at the point of peak pressure, where
dp=dx ¼ 0.
Conversion to Dimensionless Terms. The conversion to dimensionless
terms
p ffiffiffiffiffiffiffiffiffiffiis similar to that presented in Chapter 4. The length, x, is normalized by
2Rhn , and the dimensionless terms are defined as
x h
x
¼ pffiffiffiffiffiffiffiffiffiffi and h
¼
2Rhn hn
The dimensionless clearance gets a simple form:
h
ð
xÞ ¼ 1 þ x
2
Using the preceding substitutions, the pressure gradient equation takes the form
dp 6mU h

h
0
¼ 2
dx hn ð1 þ x
2 Þ3
Here the unknown, h0 , is replaced by
h
0 ¼ 1 þ x
20
The unknown, h0 , is replaced by unknown x0 , which describes the location of the
peak pressure. Using the foregoing substitutions, the pressure gradient equation is
reduced to
dp 6mU x
2
x
20
¼ 2
dx hn ð1 þ x
2 Þ3
Converting dx into dimensionless form produces
pffiffiffiffiffiffiffiffiffiffi
dx ¼ 2Rhn d x

The dimensionless differential equation for the pressure takes the form
h2 x
2
x
20
pffiffiffiffiffiffiffiffiffiffin dp ¼ d x

2Rhn 6mU ð1 þ x
2 Þ3

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 Here, the dimensionless pressure is
h2
p
¼ pnffiffiffiffiffiffiffiffiffiffi p
6U m 2Rhn
The final equation for integration of the dimensionless pressure is
ð ðx 2
h2n 1 p x

x
20
p
¼ pffiffiffiffiffiffiffiffiffiffi dp ¼ d x
þ p0
2Rhn 6mU 0 1 ð1 þ x
2 Þ3
Here p0 is a constant of integration, which is atmospheric pressure far from the
minimum clearance. In this equation, p0 and x0 are two unknowns that can be
solved for by the following boundary conditions of the pressure wave:
at x ¼ 1; p¼0
at x ¼ 0; p¼0
The first boundary condition, p ¼ 0 at x ¼ 1, yields p0 ¼ 0. The second
unknown, x0 , is solved for by iterations (trial and error). The value of x0 is
varied until the second boundary condition of the pressure wave, p ¼ 0 at x ¼ 0,
is satisfied.
Numerical Solution by Iterations. For numerical integration, the boundary
x
¼ 1 is replaced by a relatively large finite dimensionless value where pressure
is small and can be disregarded, such as x
¼ 4.
The numerical solution involves iterations for solving for x0. Each one of
the iterations involves integration in the boundaries from 4 to 0. The solution for
x0 is obtained when the pressure at x ¼ 0 is sufficiently close to zero. Using trial
and error, we select each time a value for x0 and integrate. This is repeated until
the value of the dimensionless pressure is nearly zero at x ¼ 0. The solution of the
dimensionless pressure wave is plotted in Fig. 6-8. The maximum dimensionless
pressure occurs at a dimensionless distance x
¼ 0:55, and the maximum
dimensionless pressure is p
¼ 0:109.
h2n 1
Dimensionless Pressure; p
¼ pffiffiffiffiffiffiffiffiffiffi p
2Rhn 6mU
Discussion of Numerical Iterations. The method of iteration is often
referred to as the shooting method. In order to run iteration, we guess a certain
value of x
0 , and, using this value, we integrate and attempt to hit the target point
x
¼ 0, p
¼ 0.
For the purpose of illustrating the shooting method, three iterations are
shown in Fig. 6-9. The iteration for x
0 ¼ 0:25 results in a pressure that is too high
at x
¼ 0; for x
0 ¼ 0:85, the pressure is too low at x
¼ 0. When the final iteration
of x
0 ¼ 0:55 is made, the target point x
¼ 0, p
¼ 0 is reached with sufficient
accuracy. The solution requires a small computer program.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 F IG. 6-8 Dimensionless pressure wave.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 F IG. 6-9 Solution by iterations.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 b. Load Capacity
The load capacity for one cylindrical section is solved by the equation
ð
W ¼L p dx
ðAÞ

Converting to dimensionless terms, the equation for the load capacity becomes
pffiffiffiffiffiffiffiffiffiffi
2Rhn pffiffiffiffiffiffiffiffiffiffi ð
W ¼L 6mU 2Rhn p
d x

h2n ðAÞ

For the practical boundaries of the pressure wave, the equation takes the form
ð0
2R
W ¼ L 6mU p
d x

hn 4

c. Minimum Film Thickness

Solving for hn, the minimum thickness between the ice and sled blade is
ð0
12LRmU
hn ¼ p
d x

W 4

Numerical integration is performed based on the previous results of the

dimensionless pressure in Fig. 6-8. The result for the load capacity is obtained
by numerical integration, in the boundaries 4 to 0:
ð0
0
p
d x
 S p
i Dxi ¼ 0:147
4 4

The minimum film thickness is

12  0:9 m  0:3 m  1:792  10
3 ðN-s=m2 Þ  4:2 m=s

hn ¼ ð0:147Þ
500 N

hn ¼ 7:27  10
6 m ¼ 7:27  10
3 mm

The result is: hn ¼ 7:27 mm.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 Example Problem 6-2
Cylinder on a Flat Plate
A combination of a long cylinder of radius R and a flat plate surface are shown in
Fig. 5-4. The cylinder rotates and slides on a plane (there is a combination of
rolling and sliding), such as in the case of gears where there is a theoretical line
contact with a combination of rolling and sliding. However, due to the hydro-
dynamic action, there is a small minimum clearance, hn . The viscosity, m, of the
lubricant is constant, and the surfaces of the cylinder and flat plate are rigid.
Assume practical pressure boundary conditions [Eqs. (6-67)] and solve the
pressure wave (use numerical iterations). Plot the dimensionless pressure-wave
for various rolling and sliding ratios x.

Solution
In Chapter 4, the equation for the pressure gradient was derived. The following is
the integration for the pressure wave. The clearance between a cylinder of radius
R and a flat plate is discussed in Chapter 4; see Eq. (4-33). For a fluid film near
the minimum clearance, the approximation for the clearance is

x2
hðxÞ ¼ hmin þ
2R

The case of rolling and sliding is similar to that of Eq. (6-20), (see Section 6.4).
The Reynolds equation is in the form



@ h3 @p @ h3 @p @h
þ ¼ 6U ð1 þ xÞ
@x m @x @z m @z @x

Here, the coefficient x is the ratio of rolling and sliding. In terms of the velocities
of the two surfaces, the ratio is

oR
x¼
U

The fluid film is much wider in the z direction in comparison to the length in the x
direction. Therefore, the pressure gradient in the axial direction can be neglected
in comparison to that in the x direction. The Reynolds equation is simplified to
the form


@ h3 @p @h
¼ 6U ð1 þ xÞ
@x m @x @x

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 The preceding equation is converted into dimensionless terms (see Example
Problem 6-1):

x h h0
x
¼ pffiffiffiffiffiffiffiffiffiffi ; h
¼ ; and h0 ¼
2Rhn hn hn
dp 6mU ð1 þ xÞ x
2
x
20
¼
dx h2n ð1 þ x
2 Þ3

Converting the pressure gradient to dimensionless form yields

h2 x
2
x
20
pffiffiffiffiffiffiffiffiffiffin dp ¼ ð1 þ xÞ d x

2Rhn 6mU ð1 þ x
2 Þ3

The left hand side of the equation is the dimensionless pressure:

ð ðx
h2n 1 p x
2
x
20
p
¼ ð1 þ xÞ pffiffiffiffiffiffiffiffiffiffiffi dp ¼ ð1 þ xÞ d x
þ p0
2RHn 6mU 0
1 ð1 þ x
2 Þ3

Here, p0 is a constant of integration, which is atmospheric pressure far from the

minimum clearance. In this equation, p0 and x0 are two unknowns that can be
solved for by the practical boundary conditions of the pressure wave; compare to
Eqs. (6-67):

p ¼ p0 at x ¼ x1
dp
¼ 0 at x ¼ x2
dx
p ¼ 0 at x ¼ x2

Atmospheric pressure is zero, and the first boundary condition results in

p0 ¼ 0. The location of the end of the pressure wave, x2 , is solved by iterations.
The solution is performed by guessing a value for x ¼ x2 ; then x0 is taken as x2 ,
because at that point the pressure gradient is zero.
The solution requires iterations in order to find x0 ¼ x2, which satisfies the
boundary conditions. For each iteration, integration is performed in the bound-
aries from 0 to x2 , and the solution is obtained when the pressure at x2 is very
close to zero.
For numerical integration, the boundary x
1, where the pressure is zero, is
taken as a small value, such as x
1 ¼
4. The solution is presented in Fig. 6-10.
The curves indicate that the pressure wave is higher for higher rolling ratios. This
means that the rolling plays a stronger role in hydrodynamic pressure generation
in comparison to sliding.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 F IG. 6-10 Pressure wave along a fluid film between a cylinder and a flat plate for
various rolling-to-sliding ratios.

Problems
6-1 A journal bearing is fed a by high-pressure external pump. The pump
pressure is sufficient to avoid cavitation. The bearing length L ¼ 2D.
The diameter D ¼ 100 mm, the shaft speed is 6000 RPM, and the
clearance ratio is C=R ¼ 0:001.
Assume that the infinitely-long-bearing analysis can be
approximated for this bearing, and find the maximum load capacity
for lubricant SAE 10 at average fluid film temperature of 80 C, if the
maximum allowed eccentricity ratio e ¼ 0:7.

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.

 6-2 A flat plate slides on a lubricated cylinder as shown in Fig. 4-7. The
cylinder radius is R, the lubricant viscosity is m, and the minimum
clearance between the stationary cylinder and plate is hn . The elastic
deformation of the cylinder and plate is negligible.
1. Apply numerical iterations, and plot the dimensionless
pressure wave. Assume practical boundary conditions of
the pressure, according to Eq. 6.67.
2. Find the expression for the load capacity by numerical
integration.

6-3 In problem 6-2, the cylinder diameter is 250 mm, the plate slides at
U ¼ 0:5 m=s, and the minimum clearance is 1 mm (0.001 mm). The
lubricant viscosity is constant, m ¼ 10
4 N-s=m2 . Find the hydro-
dynamic load capacity.

6-4 Oil is fed into a journal bearing by a pump. The supply pressure is
sufficiently high to avoid cavitation. The bearing operates at an
eccentricity ratio of e ¼ 0:85, and the shaft speed is 60 RPM. The
bearing length is L ¼ 3D, the journal diameter is D ¼ 80 mm, and
the clearance ratio is C=R ¼ 0:002. Assume that the pressure is
constant along the bearing axis and there is no axial flow (long-
bearing theory).

a. Find the maximum load capacity for a lubricant SAE 20

operating at an average fluid film temperature of 60 C.
b. Find the bearing angle y where there is a peak pressure.
c. What is the minimum supply pressure from the pump in
order to avoid cavitation and to have only positive pressure
around the bearing?

6-5 An air bearing operates inside a pressure vessel that has sufficiently
high pressure to avoid cavitation in the bearing. The average viscosity
of the air inside the bearing is m ¼ 2  10
4 N-s=m2 . The bearing
operates at an eccentricity ratio of e ¼ 0:85. The bearing length is
L ¼ 2D, the journal diameter is D ¼ 30 mm, and the clearance ratio
is C=R ¼ 8  10
4 . Assume that the pressure is constant along the
bearing axis and there is no axial flow (long-bearing theory).
a. Find the journal speed in RPM that is required for a bearing
load capacity of 200 N. Find the bearing angle y where
there is a peak pressure.
b. What is the minimum ambient pressure around the bearing
(inside the pressure vessel) in order to avoid cavitation and
to have only positive pressure around the bearing?

Copyright 2003 by Marcel Dekker, Inc. All Rights Reserved.