630 Mechanical Engineering Design

W W W W
1
Nf 2
Nj

Nb

Nj Nj Nj Nj

Nj
Nb = 0, Nf = 0 Nb = 0, Nf = Nj Nb = 0, Nf = Nb = Nj , Nf = 0
2
N = Nj + 0 – 2(0) = Nj N = Nj + 0 – 2Nj = Nj N = Nj + 0 – 2Nj 2 = 0 N = Nj + Nj – 2(0) = 2Nj
(a) (b) (c) (d)

Figure 12–11
How the significant speed varies. (a) Common bearing case. (b) Load vector moves at the same speed as the journal. (c) Load
vector moves at half journal speed, no load can be carried. (d) Journal and bushing move at same speed, load vector stationary,
capacity halved.

This group of variables tells us how well the bearing is performing, and hence we may
regard them as performance factors. Certain limitations on their values must be imposed by
the designer to ensure satisfactory performance. These limitations are specified by the char-
acteristics of the bearing materials and of the lubricant. The fundamental problem in bear-
ing design, therefore, is to define satisfactory limits for the second group of variables and
then to decide upon values for the first group such that these limitations are not exceeded.

Significant Angular Speed

In the next section we will examine several important charts relating key variables to
the Sommerfeld number. To this point we have assumed that only the journal rotates
and it is the journal rotational speed that is used in the Sommerfeld number. It has been
discovered that the angular speed N that is significant to hydrodynamic film bearing
performance is6

N = |N j + Nb − 2N f | (12–13)

where Nj = journal angular speed, rev/s

Nb = bearing angular speed, rev/s
Nf = load vector angular speed, rev/s
When determining the Sommerfeld number for a general bearing, use Eq. (12–13)
when entering N. Figure 12–11 shows several situations for determining N.

Trumpler’s Design Criteria for Journal Bearings

Because the bearing assembly creates the lubricant pressure to carry a load, it reacts to
loading by changing its eccentricity, which reduces the minimum film thickness h 0 until
the load is carried. What is the limit of smallness of h 0 ? Close examination reveals that
the moving adjacent surfaces of the journal and bushing are not smooth but consist of
a series of asperities that pass one another, separated by a lubricant film. In starting a

6
Paul Robert Trumpler, Design of Film Bearings, Macmillan, New York, 1966, pp. 103–119.
 Lubrication and Journal Bearings 631

bearing under load from rest there is metal-to-metal contact and surface asperities are
broken off, free to move and circulate with the oil. Unless a filter is provided, this debris
accumulates. Such particles have to be free to tumble at the section containing the min-
imum film thickness without snagging in a togglelike configuration, creating additional
damage and debris. Trumpler, an accomplished bearing designer, provides a throat of at
least 200 μ in to pass particles from ground surfaces.7 He also provides for the influence
of size (tolerances tend to increase with size) by stipulating
h 0 ≥ 0.0002 + 0.000 04d in (a)
where d is the journal diameter in inches.
A lubricant is a mixture of hydrocarbons that reacts to increasing temperature by
vaporizing the lighter components, leaving behind the heavier. This process (bearings
have lots of time) slowly increases the viscosity of the remaining lubricant, which
increases heat generation rate and elevates lubricant temperatures. This sets the stage
for future failure. For light oils, Trumpler limits the maximum film temperature Tmax to
Tmax ≤ 250◦ F (b)

Some oils can operate at slightly higher temperatures. Always check with the lubricant
manufacturer.
A journal bearing often consists of a ground steel journal working against a softer,
usually nonferrous, bushing. In starting under load there is metal-to-metal contact,
abrasion, and the generation of wear particles, which, over time, can change the geo-
metry of the bushing. The starting load divided by the projected area is limited to
Wst
≤ 300 psi (c)
lD
If the load on a journal bearing is suddenly increased, the increase in film temper-
ature in the annulus is immediate. Since ground vibration due to passing trucks, trains,
and earth tremors is often present, Trumpler used a design factor of 2 or more on the
running load, but not on the starting load of Eq. (c):
nd ≥ 2 (d)

Many of Trumpler’s designs are operating today, long after his consulting career is
over; clearly they constitute good advice to the beginning designer.

12–8 The Relations of the Variables

Before proceeding to the problem of design, it is necessary to establish the relationships
between the variables. Albert A. Raimondi and John Boyd, of Westinghouse Research
Laboratories, used an iteration technique to solve Reynolds’ equation on the digital
computer.8 This is the first time such extensive data have been available for use by
designers, and consequently we shall employ them in this book.9

7
Op. cit., pp. 192–194.
8
A. A. Raimondi and John Boyd, “A Solution for the Finite Journal Bearing and Its Application to Analysis
and Design, Parts I, II, and III,” Trans. ASLE, vol. 1, no. 1, in Lubrication Science and Technology,
Pergamon, New York, 1958, pp. 159–209.
9
See also the earlier companion paper, John Boyd and Albert A. Raimondi, “Applying Bearing Theory to the
Analysis and Design of Journal Bearings, Part I and II,” J. Appl. Mechanics, vol. 73, 1951, pp. 298–316.
632 Mechanical Engineering Design

The Raimondi and Boyd papers were published in three parts and contain 45
detailed charts and 6 tables of numerical information. In all three parts, charts are used
to define the variables for length-diameter (l/d) ratios of 1:4, 1:2, and 1 and for beta
angles of 60 to 360◦ . Under certain conditions the solution to the Reynolds equation
gives negative pressures in the diverging portion of the oil film. Since a lubricant can-
not usually support a tensile stress, Part III of the Raimondi-Boyd papers assumes that
the oil film is ruptured when the film pressure becomes zero. Part III also contains data
for the infinitely long bearing; since it has no ends, this means that there is no side leak-
age. The charts appearing in this book are from Part III of the papers, and are for full
journal bearings (β = 360◦ ) only. Space does not permit the inclusion of charts for par-
tial bearings. This means that you must refer to the charts in the original papers when
beta angles of less than 360◦ are desired. The notation is very nearly the same as in this
book, and so no problems should arise.

Viscosity Charts (Figs. 12–12 to 12–14)

One of the most important assumptions made in the Raimondi-Boyd analysis is that
viscosity of the lubricant is constant as it passes through the bearing. But since work is
done on the lubricant during this flow, the temperature of the oil is higher when it leaves
the loading zone than it was on entry. And the viscosity charts clearly indicate that the
viscosity drops off significantly with a rise in temperature. Since the analysis is based
on a constant viscosity, our problem now is to determine the value of viscosity to be
used in the analysis.
Some of the lubricant that enters the bearing emerges as a side flow, which carries
away some of the heat. The balance of the lubricant flows through the load-bearing zone
and carries away the balance of the heat generated. In determining the viscosity to be
used we shall employ a temperature that is the average of the inlet and outlet tempera-
tures, or

T
Tav = T1 + (12–14)
2

where T1 is the inlet temperature and T is the temperature rise of the lubricant
from inlet to outlet. Of course, the viscosity used in the analysis must correspond
to Tav .
Viscosity varies considerably with temperature in a nonlinear fashion. The ordi-
nates in Figs. 12–12 to 12–14 are not logarithmic, as the decades are of differing vertical
length. These graphs represent the temperature versus viscosity functions for common
grades of lubricating oils in both customary engineering and SI units. We have the
temperature versus viscosity function only in graphical form, unless curve fits are devel-
oped. See Table 12–1.
One of the objectives of lubrication analysis is to determine the oil outlet temper-
ature when the oil and its inlet temperature are specified. This is a trial-and-error type
of problem. In an analysis, the temperature rise will first be estimated. This allows
for the viscosity to be determined from the chart. With the value of the viscosity, the
analysis is performed where the temperature rise is then computed. With this, a new
estimate of the temperature rise is established. This process is continued until the
estimated and computed temperatures agree.
To illustrate, suppose we have decided to use SAE 30 oil in an application in which
the oil inlet temperature is T1 = 180◦ F. We begin by estimating that the temperature rise
 Lubrication and Journal Bearings 633

Figure 12–12 10 4
5
Viscosity–temperature chart 3
in U.S. customary units. 2
(Raimondi and Boyd.) 10 3
5
3
2

10 2

5
4
3
2

SA
10 E
70
60
50
Absolute viscosity (␮reyn)

5 40
30
4
20
3 10

1
B

0.5

0.4

0.3

0.2
30 50 100 150 200 250 300
Temperature (°F)

will be T = 30◦ F. Then, from Eq. (12–14),

T 30
Tav = T1 + = 180 + = 195◦ F
2 2
From Fig. 12–12 we follow the SAE 30 line and find that μ = 1.40 μreyn at 195◦ F. So
we use this viscosity (in an analysis to be explained in detail later) and find that the
temperature rise is actually T = 54◦ F. Thus Eq. (12–14) gives
54
Tav = 180 + = 207◦ F
2
634 Mechanical Engineering Design

Figure 12–13 10 4
5
Viscosity–temperature chart
3
in SI units. (Adapted from 2
Fig. 12–12.)
103

3
2

102

SA
5 E
70
Absolute viscosity (mPa·s)

4
60
3 50

40
2
30
20
10

10

2
10 20 30 40 50 60 70 80 90 100 110 120 130 140
Temperature (°C)

This corresponds to point A on Fig. 12–12, which is above the SAE 30 line and indi-
cates that the viscosity used in the analysis was too high.
For a second guess, try μ = 1.00 μreyn. Again we run through an analysis and this
time find that T = 30◦ F. This gives an average temperature of

30
Tav = 180 + = 195◦ F
2

and locates point B on Fig. 12–12.

If points A and B are fairly close to each other and on opposite sides of the SAE 30
line, a straight line can be drawn between them with the intersection locating the cor-
rect values of viscosity and average temperature to be used in the analysis. For this illus-
tration, we see from the viscosity chart that they are Tav = 203◦ F and μ = 1.20 μreyn.
 Lubrication and Journal Bearings 635

Figure 12–14 103

5
Chart for multiviscosity 4
lubricants. This chart was 3
2
derived from known viscosities
at two points, 100 and 210°F, 102
and the results are believed
to be correct for other 5
4
temperatures. 3

10

20
W
5 –5
0
4 20
W
Absolute viscosity (␮reyn)

–4
3 0

10
2 W
–3
0

1
5W
–3
0

10 20
W W

0.5

0.4

0.3

0.2
50 100 150 200 250 300
Temperature (°F)

Table 12–1 Viscosity Constant

Oil Grade, SAE ␮0, reyn b, °F
Curve Fits* to Approxi-
mate the Viscosity versus 10 0.0158(10−6) 1157.5
−6
Temperature Functions 20 0.0136(10 ) 1271.6
for SAE Grades 10 to 60 30 0.0141(10−6) 1360.0
−6
Source: A. S. Seireg and 40 0.0121(10 ) 1474.4
S. Dandage, “Empirical Design −6
50 0.0170(10 ) 1509.6
Procedure for the Thermody- −6
namic Behavior of Journal 60 0.0187(10 ) 1564.0
Bearings,” J. Lubrication
Technology, vol. 104, *␮  ␮0 exp [b/(T  95)], T in °F.
April 1982, pp. 135–148.