Professional Documents
Culture Documents
General Editors:
PROFESSOR SIR NEVILL MOTT, F.R.S.
Emeritus Cavendish Professor of Physics
University of Cambridge
G. R. NOAKES
Formerly Senior Physics Master
Uppingham School
Series Editor:
DR. A. T.FULLER
Assistant Director of Research
Engineering Department
University of Cambridge
J. Halling
University of Salford
TRIBOLOGY is the new name for any problem concerned with the
carrying of load across interfaces in relative motion. Thus, although
the word is new, the subject concerns itself with such well known
topics as friction, wear and lubrication. Although this subject is
important in all industrial machinery, it also has a much wider signi-
ficance. The cleaning of our teeth or the operation of human joints
are obvious examples of tribological phenomena. It is an inter-
disciplinary subject that concerns itself with physics, chemistry,
materials science, engineering and mathematics, often all at the same
time.
This book outlines the basic principles upon which the subject
rests and shows how the various branches of science and engineering
have contributed to our understanding of these principles. The main
theme concentrates on physical principles. The mathematics
included is only given in outline and it is hoped that students will
understand the basic physical arguments even though they may not
wish to become involved in the mathematical detail.
At the end of each chapter I have indicated the types of project
activity which could enlarge our appreciation of the subject matter.
The level is generally that of a first-year undergraduate in science or
technology or a science sixth former embarking on specialization in
science or technology at a university. As with most books, the reward
which any individual will receive from this book will depend on the
time he is prepared to spend not only on understanding what is
written but also on obtaining a deeper feeling for the subject by
carrying out the projects. The book is not intended as a superficial
introduction to the subject.
I must acknowledge that this book depends considerably on the
contents of an earlier book, Principles of Tribology, written by my
colleagues and myself. The project activities have also been inspired
by the booklet Tribology Projects for Schools written by a Committee
on Tribology Panel under my chairmanship. I am very grateful to
Mr. W. E. W. Smith, of Calday Grange Grammar School, for his
many helpful suggestions during the preparation of this book, which
I am sure will enhance its value to schools. Finally, I acknowledge
the enormous help in producing the manuscript by Mrs. L. M.
Chadderton who probably deserves as much credit as I for the final
achievement.
V
CONTENTS
Preface v
Chapter l INTRODUCTION
1. 1 Tribology 1
1.2 The occurrence of tribology 1
1.3 Historical 3
1.4 Tribological solutions 7
1.5 Economic considerations 11
1.6 Conclusion 13
1. 7 Project suggestions 13
Chapter 3 FRICTION
3.1 Introduction 41
3.2 Possible causes of friction 42
3.3 The simple adhesion theory 43
3.4 Extensions to the simple theory 46
3.5 Friction between elastic bodies 49
3.6 Rolling friction 50
3.7 The experimental measurement of friction 57
3.8 Project suggestions 60
vu
Chapter 4 WEAR AND THE PROPERTIES OF
MATERIALS
4.1 Introduction 63
4.2 Types of wear 64
4.3 Mechanisms of wear 65
4.4 Factors affecting wear behaviour 72
4.5 Tribological properties of plastics 75
4.6 The measurement of wear 77
4. 7 Project suggestions 81
IX
CHAPTER 1
introduction
1.1. Tribology
THE name ' tribology ' is derived from the Greek word tribos meaning
' rubbing ', so that a literal translation would be ' the science of
rubbing '. This interpretation is a little too narrow, so that the word
is actually defined as: ' The science and technology of interacting
surfaces in relative motion and of related subjects and practices ',
and this definition, whilst embracing the literal translation, also
includes many other aspects of solids in contact. In fact the word
was specifically created to bring together the interests of chemists,
engineers, materials scientists, physicists and others in the many
aspects of such subjects as contact mechanics, friction, lubrication,
wear, etc. This wide-ranging interest in the subject underlines its
most significant academic feature; it is an interdisciplinary subject.
Although the name is a new one, the subject deals with some of the
oldest problems with which mankind has been concerned. The
wheel, an obvious tribological device, was invented to reduce the
resistance to translatory motion and its invention certainly predates
recorded history. In almost every aspect of our daily lives we are
meeting some manifestation of tribology and not only in the context
of the machinery which is such an essential part of an industrial
society.
•Ml
r
/El
10-1
I /
0
,/E2
..... ---....-E3
o' E 1 First reptiles
-0
E~------ E4 E2 Crawlers
~,£
e
Cl
.Q'
;
10- 2
I
•M3 E3 Apes
E4 Man
•M4 E5 Athlete
,< 10- 3 ./ Ml Sliding
M2 Lubricant
• - - - - M5
M6
M3 Early wheel
M4 Spoked wheel
10-4
M5 Railway
M6 Modern railway
Fig. 1.1. The reduction in the drag/weight ratio over the timescale of the
Earth's history.
2
figure we see a thirty-fold decrease in the drag /weight ratio during the
past thousand million years-the E line-arising from the evolutionary
development of mankind.
About 10 000 years ago modern man could move large bodies
only by sliding them around. Since then, by the development of
primitive lubricants based on animal fat and later a whole series of
wheel systems, he has vastly increased his ability to translate objects.
Thus in fig. 1.1 we note a reduction of some thousand-fold in the
drag /weight ratio during this 10 000 year period-the M line.
1.3. Historical
The invention of the wheel probably took place at least 6000 years
ago, evidence of its use being found in the earliest known historical
records. It is interesting to note that the highly developed Inca
civilization did not discover or exploit the principle of the wheel.
Other tribological applications stemming from the earliest days of
man's history were the use of friction to create fire, the development
of drills fitted with bearings made from bones and antlers, and the
use of elementary grinding wheels for grinding cereals and the like.
Fig. 1.2 .. The bow drill and the grinding stones used by man for the last
4000 years.
3
Fig. 1.2 shows a simple bow drill, a type which is still used in India,
and a very simple cereal-grinding stone. Evidence of the use of
lubricants has also been recorded by recent archaeological excavations.
A chariot taken from an Egyptian tomb still had wheel bearings
containing some animal fat lubricant contaminated with debris in the
form of quartz sand, compounds of aluminium, iron and lime,
presumably picked up during its service life.
The monumental task of building the Pyramids involved the
Egyptians in many tribological developments. Surviving illustrations
in the form of bas-reliefs show the use of rollers and sledges for the
transportation of large stone blocks and statues. Fig. 1.3 shows one
example of such a process, where 172 slaves are seen to be dragging
a large statue of mass about 60 Mg and weight about 6 x 10 5 N along a
wooden track, the statue being supported on a wooden sled. Closer
examination of this illustration shows one man standing at the front
of the sled pouring some form of lubricant into the path of the sled;
surely one of our earliest tribologists. It is interesting to estimate the
coefficient of friction which applied in this situation. If we consider
that each slave pulls with a force of about 800 N, the coefficient of
friction is seen to be
= no. of slaves x force per slave= 172 x 800 = _ .
0 23
µ, weight of statue 6 x 10 5
This is just about the value we would expect for a poorly lubricated
wooden slide and it leads to an interesting conclusion. This picture
must be a true record of what happened, although the artist of this
time clearly had no ability to produce a drawing according to perspec-
tive. VVe see that in a sense this picture is as faithful a contemporary
record as the photographs used by today's media.
Many other examples of tribological practice in the ancient world
could be cited but the science of tribology, as opposed to its practice,
stems from the 15th century. Leonardo da Vinci (1452-1519)
clearly recognized that the friction force is proportional to the load and
independent of the nominal area of contact. He also made the very
perspicacious statement about rolling motion that ' This kind of
friction is caused not by rubbing, but by contact in what might be
described as progress by infinitely small steps '.
The first two laws of friction are always credited to Amontons,
who stated them in 1699 and was undoubtedly unaware of Leonardo
da Vinci's work, and in 1780 the third law of friction was provided
by Coulomb. He stated that friction between two surfaces is indepen-
dent of their relative velocity, although we now know that this law is
less universally true than the first two statements, viz. that the friction
force is proportional to load and that the friction force is independent
of the nominal area of contact. About this time many other tribo-
4
V1
Fig. 1.3. Transporting an Egyptian colossus ca 1880 Il.C. From a drawing in a grotto at El Bersheh.
(Reproduced from Sir A.H. Layard, Discoveries in Nineveh and Babylon.)
logical developments occurred because of the stimulus of growing
industrialization, and this progress has continued up to the present.
The use of lubricants stems from very early times, but Newton
was the first person to study the laws of viscous flow. Theoretical
understanding of lubrication in bearings came towards the end of the
last century, and the work of Beauchamp Tower (1899), of Osborne
Reynolds (1886), and of Stokes and Petroff about the same time, is
particularly noteworthy. Since then there has been an enormous
amount of work in both the application and understanding of lubri-
cated systems.
Fig. 1.4. Detail of a fragment of the Lake Nemi Bearing (reproduced from
Vrelli Le Novi di Nemi, Libreria Delio Statto, Rome).
Table 1.1. Friction and wear results for various material combinations.
B 7
(a) Choice of materials-fig. 1.5 a
Bearing in mind the results quoted in table 1.1 we might hrst
explore the possibility of meeting our requirements by the correct
choice of our materials. Thus we note in table 1.1 that PTFE
(polytetrafluoroethylene) gives low values ofµ, and modest wear when
rubbing on hardened tool steel. For this reason such materials are
often used in so-called ' dry bearings ' and various designs which
embody them are now commercially available. Unfortunately these
materials suffer from two disadvantages. First, they are relatively
soft and we must therefore limit the size of load applied to such
contacts. Second, they are subject to failure as the temperature
rises, and since all friction generates heat, we cannot use them at
high speeds where temperature rises may become too large (see
Chapters 2 and 4). These disadvantages may in part be overcome by
using metals rather than plastics, and recent work suggests that some
forms of crystal structure in metals are particularly useful in meeting
our requirements. Alloys based on metals such as cobalt appear to
offer possible solutions in this context.
CJ~
__ ==---_-~Pressure P
1////,//1///,//,////7/1///;;////,////////
(a) (d)
(b)
1-- W///~!Wi ~ (e)
Elastomer
w
iw
-- D::::,ga,t;o
1111111111111111111111111
1;,ld
(c) (f)
t
µ
Many layers
Number of traverses ~
E
:::,
C
C
0
515
505
495
473
- Manpower savings
Lubricant savings
Investment savings
Less friction losses
'--
<l) 445
Cl.
V)
C
.'2 - - Longer machine I ife
.E
""
C 34,5
>-
a,
0
0 - - Fewer breakdowns
..0
E
'--
l':'
a:; 230
..0
E
.....2
V)
a, Less maintenance
s>
0
and replacement
V)
~
..0
·;,;
V)
0
Q
0
Fig. 1.7. The saYings achiernb]e in Britain each year due to better tribological
practice.
1.6. Conclusion
In view of the industrial importance of tribology, this aspect of the
subject usually receives the most consideration. Nonetheless, the
contribution of tribology to other aspects of our life can be very
profound, for example, in our understanding of human joints and,
in particular, to the alleviation of suffering from diseases such as
arthritis.
I hope that the reader has by now been stimulated to find out more
about the details of the various topics which have been mentioned.
It has so far been tacitly assumed that surfaces are simply planes
separating solids from their environments. This is not the case.
All surfaces are rough rather than smooth and (although this roughness
is on a microscopic scale) it adds to the complexity and interest of the
subject of tribology.
1. 7. Project suggestions
1.7.1. Historical projects
Throughout history mankind has been concerned with movmg
objects relative to each other. One can therefore devise many
historical projects in which developments in tribology may be related
to the development of man towards an industrialized society. A few
examples are:
(1) Consider tribology in the context of the earliest civilizations
leading to the achievements of the Egyptian civilization.
Examine the records of the time and establish the scientific
validity of such records, e.g. in bas-reliefs showing slaves
moving large masses; do the number of slaves and the estimated
weight of the body being moved give reasonable values of the
coefficient of friction for the tribological method being
employed?
(2) It is reported that the Inca civilization did not discover the
principles of the wheel. How did they overcome their tribo-
logical problems and to what extent were they inhibited by not
having wheels? How is the lack of wheeled transport reflected
in their civil engineering and architecture?
13
(3) Rolling motion is a well established method of solving tribo-
logical problems and has led to the development of a wide range
of ball and roller bearings. Study the history and development
of this type of bearing and evaluate its importance to society at
various stages of its development.
15
CHAPTER 2
the nature of surfaces and their contact
2.1. Introduction
TRADITIONALLY we might think of the surface of a solid as the
geometrical boundary between the solid and its environment. For
our purposes such a definition is too limited and we must probe a
good deal into the body of the solid if we are to understand how its
surface behaves. We shall therefore consider our surfaces in depth
including the nature of the surface layers and the sub-surface behaviour
of the material. We shall be interested in the physical properties of
the surface layers as well as in the geometry of the actual surface
profile. Not surprisingly these characteristics depend on the bulk
properties of the solid material, the methods by which the surface
has been produced, and the nature of the environment around the
surface. Thus the optical behaviour of a metal surface depends on
the smoothness of the surface, the character of the surface layers and
the extent to which the air has reacted with the metal to form an
oxide film.
By virtue of their high strength metals are extensively used in
tribological applications. This book will therefore chiefly be con-
cerned with metallic solids, although many of the arguments apply
equally well to other materials such as plastics.
Oxide
layer
~Quenched
layer
Base material
Waviness
Roughness
Fig. 2.2. The constituent geometric components of a solid surface.
A
I
I
I
I
~ I I
I I
1 / Eyepiece
I I
I I
I/
11
(al Micro-interferogrom
1,ii
I~
+ B .
1 1 eom splitter ~
Light source ---~- -nt-}L
*========-=----li.:=hl- ----VJ---u-
Reference
surface
! \
~~ Identical
I
~ objectives
I I
I I
~
'M.~..:..:Z~~'<...«:,,.."-7<;!
Test surface
Fig. 2.3. A schematic layout of a surface interferometer.
(e) Profilometry
The most usual method for the study of surface geometry
is profilometry. In the profilometer a very fine diamond
stylus (tip radius 2 µm or less) is drawn over the surface
21
Fig. 2.5. · A scanning electron micrograph of the surface shown in fig. 2.4,
2500 X.
Specimen
Vect;calhe;ght~
of aspent1es
e
Magnification =
cot 8
e = 5° this gives 11 x magnification
(a)
(b)
Fig. 2.6. The method of taper sectioning and a typical example of the method.
C 23
Stylus
Traversing
motor
Amplifier
Surface
trace
(al
5000
L100
(b)
~ I
/
', ___ .,,.,,.,,
\
1µmL ........ 1µml__,
lµm IOOµm
~
Actual profile Distorted record
Fig. 2.8. The actual profile of a surface compared with the surface record
obtained from a profilometer.
Fig. 2.9. A typical electron microprobe record. (a) Electron image, (b) S,
Ka: X-ray image, (c) Mn, Ka: X-ray image, (d) Pb, Ka: X-ray image.
---tS~-JS
Horizontal intercepts
----- - --------
A-LLLr~+ - l - A
Sample
intervals
-------------- - --- ---
Port of profile considered Distribution Bearing area
of ordinates curve
26
break down this complicated shape into a long series of sinusoidal
waveforms which when superposed would produce the same shape.
Unfortunately this leads to an expression with a large number of
coefficients and is therefore of little practical use in defining this
profile reasonably simply. For this reason attempts are made to
represent it by utilizing its statistical properties, as outlined below.
Assume some datum line such as AA in fig. 2.10 and measure the
vertical ordinates of the profile from this datum along a representative
length of the profile, which is assumed to contain all the essential
features of the whole surface, remembering that the profile is only a
sample taken over a small length of the total surface. From such
measurements we can state how many heights of a given ordinate
value z occur at any given horizontal level in the profile. We can then
construct a distribution curve for the profile height over the given
length of the profile.
The horizontal line passing through the centre of area of the
distribution curve is defined as the Centre Line of the profile. The
areas generated by the surface profile above and below this centre
line are equal. The departure of the profile from this centre line
(i.e. from a smooth surface) may be identified by the parameters r.m.s.
(root mean square) and c.l.a. (centre line average) which are defined
as
1 n
r.m.s. = [ - _L (z;) 2
]l/2
n,=1
1 n
c.l.a. = -L jz; I•
ni=l
For most engineering surfaces the c.l.a. and r.m.s. values are very
similar and the r.m.s. value is simply the standard deviation a of
the ordinate distribution curve shown in fig. 2.10. Both these para-
meters are readily obtained by applying the voltage signal from the
profilometer to a meter; directly for the r.m.s. value and after
full-wave rectification for the c.l.a. value. For this reason these
parameters are extensively used in engineering practice. Their
limitations are perhaps best illustrated by the idealized profiles shown
in fig. 2.11, each of which although markedly different would give the
same value of c.l.a. For this reason the tribologist seeks surface
parameters which are rather more informative.
Consider the ordinate distribution curve shown in fig. 2.10. If we
adjust the scale of this diagram so that the area enclosed by the distri-
bution curve is unity we have converted the curve into a probability
density curve, i.e. the size of the horizontal at any given value of z may
be interpreted as the probability of an ordinate of this value of z
occurring. Before leaving fig. 2.10 one other distribution is of
27
interest. By summing the horizontal intercepts produced by the
profile at any given value of z we effectively obtain the area of solid at
this level. Thus repeating this procedure for all values of z produces
a curve which is known as the Bearing Area Curve. Mathematically
it is in fact the cumulative distribution curve of the ordinates, i.e. the
integral giving the area under the ordinate distribution curve, but its
physical interpretation as the amount of area at each depth level is
more informative to the tribologist. The use of a line tracing to
predict area is of course only valid where the surface characteristics
are similar for all orientations of such a trace.
i--
(a) Peak distribution
1---
~
-......
-_____ ::..+- I
✓-,
\------ ---
i
Max.
- ---
height -4
------
\
Valley distribution Part of profile considered Ordinate
distribution
(bl
,,,- Slope_ "-0
dz
profile
Mean ordinate Mean slope
I I
Fig. 2.12. The peak, valley, ordinate and slope distributions of a typical
surface profile.
Equivalent
>, Gaussian distribution
u
C
(l)
:::,
CT
(l)
it
0·25 0
a1= load
. . axia
= stress m . l d'irect10n
.
cross-sect10na1 area
E1 = change
. . in length = stram
. . . d' .
1n axia 1 irect10n
ongmal 1ength
and
E= a 1 /E1 ,
E being the elastic constant, called the Young Modulus. Although no
stress acts transversely to the axial direction there will nevertheless he
dimensional changes in the transverse direction, for as a bar extends
axially it contracts transversely. The transverse strains Ez are related
to the axial strains E1 by the Poisson ratio v such that
where the negative sign simply means that the transverse deformation
will he in the opposite sense to the axial deformation. Under more
30
complicated conditions of loading and geometry the equations
defining stress and strain are inevitably more complex and their
treatment is beyond the scope of this book.
As the load increases, elastic behaviour is replaced by plastic
behaviour in which the material is permanently deformed; after
removal of the load the material does not return to its original shape.
The stress state at which the transition from elastic behaviour to
plastic behaviour occurs is known as the yield stress and has a definite
value for a given material at a given temperature. In tribology we are
interested in tv,o bodies which are in contact because of the applied
loading, and here the nature of the transition from elastic to plastic
behaviour is of considerable importance.
~~
H t HH
,//
(//
~
i111Tu
~
't
~zb- 1/ " "~
'~
Mean pressure = .J:!_ 23
2bl / 1 Max
stress
'
r
Mean pressure= 1r 02
w
Fig. 2.14. The contact of a flat, a cylinder and a sphere and the resulting
pressure and stress distributions.
Consider the three problems shown in fig. 2.14 a, assuming that the
materials behave elastically. The nature of the contact pressure
distribution and contact geometry is as shown in fig. 2.14 b. From
elasticity theory we can show that the state of stress in the lower body
would be as shown in fig. 2.14 c for each case, each stress line being
drawn such that the stress state along it has a constant value; the
increasing numbers indicate increasing stress values. For the contact
of the cylinder and the sphere, as the load increases the actual size
of the contact zone increases due to increased flattening of the curved
31
surfaces. The actual relationships for such contact sizes, which are
usually referred to as the Hertzian values, are:
For a cylinder on a plane
b= I 8WR(l -v2) 11;2
TTLE '
pressure distribution
p= !!~ ( -~r12_
l
For a sphere on a plane
a= 13 WR~ --v2) I1/3,
pressure distribution
- 3w
P- -TT2., 1
(. - R2)
2
l /2
'
a" a
where vis the Poisson ratio and Eis the Young modulus, and W, R and
L are as shown in fig. 2.14.
The nature of the stress distribution is of considerable interest in
these three cases. For the uniform pressure contact it is seen that
the stress increases as the surface is approached and is a maximum at
the surface. For the cylindrical and spherical contact the maximum
stress, rather surprisingly, occurs at a small distance below the surface;
see fig. 2.14 c. Thus as the load W increases and the elastic behaviour
changes to plastic behaviour we can see that for both the cylindrical
and spherical contacts plasticity first occurs below the surface rather
than at the surface as in the first example.
This is a very important result since it means that, even though the
load has created a plastic zone, catastrophic failure does not occur
since the plastic material is totally enclosed by elastic material. As
loads are increased further the volume of the sub-surface plastic zone
increases until it spreads to the surface. At this load the cylinder
or ball will begin to penetrate the surface creating a noticeable
permanent indentation.
It is conventional to define the hardness of a material by the ratio
of the load to the surface area of the permanent indentation, i.e. by the
mean contact pressure between say a ball and a plane, where hardness
H = W / TTa 2 • Because of the constraining effects of the elastic material,
plastic indentation does not materialize until the mean contact pressure
has a value of three times the yield stress Y of the material. Thus,
32
From this result we may conclude that no significant permanent
indentation of the surface will occur when spherical surfaces are
loaded against a plane until the mean contact pressure is at least three
times the yield strength of the material. This allows us to apply
much greater loads to such contacts than we might otherwise expect,
and is one of the reasons why rolling-contact bearings continue to
operate satisfactorily even when slightly overloaded.
Fig. 2.15. The contact of a smooth surface and a rough surface having
asperities of the same height.
Since a; rx JV/1 3
A1 = 1Ta12rx lV1213
or
Thus
From these two results we note that for elastic deformation the total
real area of contact is proportional to W 2 l3 , whereas for plastic
deformation it is proportional to W.
(a)
W increased
(bl
Fig. 2.16. The contact of a smooth surface and a rough surface having
asperities of varying heights.
(a) As the load increases and the smooth surface compresses the
rough surface the mean area of a contact spot remains constant,
i.e. the total real area of contact A divided by the number of
contacts is constant, although it will be appreciated that both
A and n increase with increasing load.
,...,__ _ _ _ _ _ _ _ _ A
0
--------►,
. 1 A
This material becomes plastic when A= 3 0
Fig. 2.17. The eon tact of a smooth surface and a rough surface at the point
of macroscopic plastic deformation,
w
G)
Heat- source
----u
el =µ,WU
Sao,
=0·125 (µ,W) ( u)
a ex
from the stationary heat source theory applied to the moving body 1,
and
8=0·159 (µ,W)
2
a
(£)
aexpc
1 2
'
from the moving heat source theory applied to the stationary body 2,
where a defines the area of contact; ex is the thermal conductivity;
pis the density, and c is the specific heat capacity. From these results
it is seen that the moving body temperature is the one that is more
dependent on the sliding velocity.
t Melting point
---
·-·- - · - · - - - - - -·-.:..:·-;;.-·
0 u-
Fig. 2.19. The variation of surface temperature with sliding velocity.
<-~ 35~7n~
A block of wood (a) having a smooth lower surface forms the basis
of the arrangement. Two needles (b) of identical diameter act as
low friction bearings for the pointer spindle (c) and a spacer (d) cut
from a needle of the same diameter. Needle (e), acting as the stylus,
is loaded against (c) and (d) by elastic bands so that vertical displace-
ment of (e) causes the pointer spindle to rotate and gives rise to a
proportional displacement at the tip of the counterbalanced pointer
38
(g). A lightly smoked glass screen gently touching the tip (lz)
provides a permanent record of its movement. The profilometer
is drawn across the surface to be examined at slow speed by a weight
and pulley arrangement.
Assuming that the vertical movement of (e) is small and that no
sliding occurs between (e) and (c), the height of an asperity, y, can
be deduced from the height of the trace mark, Y, from the equation
R
y=IY
where R = radius of the needle (c)
L = length of the pointer (g).
The surface profile on the smoked glass plate can be further
magnified by projecting an image using a standard slide projector.
The image on the screen may then be recorded to provide a permanent
large magnification of the surface profile.
This equipment, or any alternative design using other magnifica-
tion principles, may then be used to explore the geometrical character
of a wide variety of surfaces. These may be produced by a wide
range of machining processes, varying from abrasive processes using
emery paper to turning in a lathe. It is interesting then to consider
ways in which one may use the profiles of the surface in order to give
numerical assessment of the quality. One simple method ,vould be,
for instance, to determine the distribution of the ordinates and then
obtain the standard deviation of such a distribution. This is usually
referred to as the root mean square value of the profile. Other
measurements of this kind will no doubt suggest themselves or may be
obtained by reading the literature in the field of surface measurement.
Taper sectioning
Another method of examining surfaces is to use a taper section.
Ideally the specimen surface is plated and then sectioned at a small
angle (fig. 2.6). The sectioning must be carried out carefully to
avoid distortion of the sectioned surface. The plated layer helps to
avoid such surface distortion during the sectioning and also provides
a contrasting line of demarcation to reveal the surface profile when
viewed through a microscope.
Hardness testers
The most common way of measuring the hardness of a surface is by
indentation-pressing a hard ball or diamond pyramid into the surface
with a known force and measuring the size of the impression made.
In many projects an absolute knowledge of hardness is not required;
comparative measurements will suffice.
D 39
Two ' home-made' hardness testers are described in School Science
and Technology edited by Dr. T. Kelly, and published by Schools
Council Project Technology on behalf on the Schools Science and
Technology Committee, 1969. One of them is based on a commercial
Rockwell hardness tester with a diamond penetrator, the other is a
portable tester using precompressed springs to apply known forces
and a drill bit as indentor. By suitable choice of spring, measurements
can be made on a wide variety of materials. This type of equipment
is also useful in the study of the nature of the contact between balls,
cylinders and planes using the methods described by K. L. Johnson
(Bulletin of Mechanical Engineering Education, 6, 245, 1967).
40
CHAPTER 3
friction
3 .1. Introduction
FRICTION, as the resistance to motion when bodies slide along each
other, is part of everyday experience. The force system at the contact
can be resolved into a normal ( or load) force together with a tangential
friction force (fig. 3 .1 ). It is also usual to define two states of friction,
namely, the static friction force, which is the force to initiate sliding,
and kinetic or dynamic friction force, the force to maintain steady
state slip.
Although Leonardo da Vinci and Newton had indicated the nature
of the 'Laws of friction', it was not until some time later, in 1699,
that the first two were expressly stated by the French engineer
Amontons, whose name they now bear. The third law is due to
Coulomb.
µ=F/W
These laws are entirely empirical and, although they are, surpri-
singly, valid for most situations, exceptions exist. Of the three laws,
the first two are the most important in terms of their range of applica-
bility, the third being most frequently contravened. These three
laws of friction are:
(i) That friction is independent of the apparent area of contact.
Thus the friction between a brick and a plane does not depend
on whether it stands on its small end face or its larger faces.
The reasons for our interest in real areas of contact discussed in
Chapter 2 are now apparent. Here we saw that the real area
of contact depends only on the load, no matter what the overall
size of the geometric contact. Since friction must occur at
41
the real areas of contact, we can now see that this law has a
sound physical basis. The proof of this proposition, as out-
lined in Chapter 2, came some 260 years after the original
empirical statement.
(ii) That the friction force is directly proportional to the normal load.
The ratio of the friction force to the normal load is called the
' coefficient of friction ', µ. Again, it is only in recent years
that the scientific explanation of this law has become at all
understood. A material does not on its own have an intrinsic
coefficient of friction; ,ve must always consider the value ofµ
between two solids. Also the environment within ·which the
bodies slide upon each other can markedly affect the value ofµ.
Thus clean dry steel sliding on steel in air gives µ = O· 5,
whereas the same materials in ultra-high vacuum give values
of µ some ten times greater; note that this means F = 5 TV!
Graphite sliding on graphite in the ordinary atmosphere gives
µ = O· l but with very dry air as the environment we findµ= 0· 5,
an increase of five times in the absence of water vapour.
(iii) That kinetic friction is independent of the velocity of sliding.
Although this ' law ' is reasonably obeyed by most common
solids at modest sliding speeds, its validity as a scientific
statement is lower than that of the other two laws. Indeed
our current understanding of the effects of speed on kinetic
friction is much less developed than our understanding of the
basis of the first two laws.
Body 1
W = IW1 F= IF1
Fig. 3.2. The local asperity contacts between surfaces.
~ Soft
Xr< Hard Soft
\;Ve now see that when c approaches 1 (i.e. complete adhesion of the
junctions) µ, reaches very high values, whereas when c is small the
values are more typical of those observed.
Finally we may extend these arguments by considering a more
general asperity interaction of the type shown in fig. 3.6. If we define
the angle of the initial contact of such asperities as 0 we can carry out
detailed analysis which ultimately yields values of µ, for each 0 and
each value of c defining the interfacial shear strength. The curve for
0 = 0 is of course the solution given by equation (3.5). Typical results
are shown in fig. 3.7, while fig. 3.8 shows schematically the progress of
the interaction between the asperities when c is small and when c is
large. It is interesting to note that when high adhesion occurs,
i.e. c is large, the junctions finally break by being pulled in a tensile
fashion as one surface moves relative to the other. Evidence of such
behaviour is found in experiments carried out in ultra-high vacuum
on clean surfaces where adhesive effects would be expected to be
strongest. This theory also predicts that when 0 = 0, as in the
conditions considered earlier, one obtains the solution represented by
equation (3 .5).
47
8=20°
µ.
c---+- 10
Fig. 3.7. The variation ofµ, with c for different asperity slopes.
~ interface - (a)
) --(
>==< >
--+- -
(
------l>--
------ -+-
~-c
(b)
·c_ ~"''"'""
~ ____ __.._
~ -► --+
Fig. 3.8. The deformation of a single asperity during sliding (a) when c is
small and (b) when c is large.
48
Although the developments to the basic theory have only been given
in outline, it should be clear how they have produced a more useful
approach and how they have overcome the objections to the simple
theory. In particular we are now able to appreciate the great effect
of surface films on frictional behaviour. Thus we note how much
oxide films on metal surfaces, which inhibit the adhesion at the asperity
junctions arc responsible for the low values of µ, usually observed.
Indeed in mechanisms for space vehicles which have to operate in a
non-oxidizing environment many tribological difficulties arise due
to the absence of such oxide films, as c tends to approach unity.
L ~surep
~.x
~ 2b i - ~zb__::;:--dx
Fig. 3.9.
fl
The contact of a cylinder and a plane and the resulting pressure
distribution.
Consider only the forward part of the contact zone. The contact
pressure acting over a small element width dx and length L gives rise
to a force pLdx. The resisting moment due to this force about the
centre line of the contact is pLdx x. Thus the total resisting moment
Ai due to the forward compression is
/)
M= f p . L . x . dx.
0
1~0 M0.
Suppose the friction force moving the cylinder the distance S is F
and note that S = 0R.
51
Work done by F =Work done against hysteresis
E E S
F S=-M0=-M-
. 100 100 R.
Thus
Elli/
F= lOOR' (3.9)
This result is interesting in that it shows that the free rolling resistance
of a wheel will be less if W is small and R is large. This clearly
explains the advantages of a spoked wheel in reducing friction.
-v
l•2b+j
t
L
Non -slip area
t
Slip
area .j 2 .BI+
Fig. 3.10. The microslip in the contact zone of a cylinder rolling along a plane.
52
when T = µ, W slip occurs throughout the contact area. The interes-
ting question is what happens when O < T < µ, W. The answer may
at first sight seem somewhat strange but it will eventually be seen to be
entirely reasonable. What actually happens is shown in the details
of the contact patch shown in fig. 3 .10. A part of the contact patch
is subjected to slipping whilst in the remainder at the front of the
contact zone no relative motion occurs. As T increases the size of
the slip area increases and the whole area is slipping when T = µ, W.
Theoretically it can in fact be shown that the degree of slipping is
related to T by the formula,
.I_ = [
µ,W l
-(@)2]1/2
b '
(3.11)
8R
bµ
t rm
I
1 T l·O
µW
Fig. 3 .11. The variation of the creep with the tangential traction for a rolling
cylinder.
53
geometric considerations, i.e. S #- RB, the condition which was used
in the previous section. Thus a driving wheel rolls slightly further
than RB whilst a braking wheel rolls slightly less. These so-called
' Creep effects ' have been measured and shown to agree with this
type of theoretical treatment, fig. 3 .11, which yields
(3.12)
w w
I
' I Contact
~ffl
' I
' / - width - -
----~--r'•~-___.,. __ -·-·-A- - - A
j 2bL
1+ +120
Non slip
2b a~a
_t_ 1
Slip_,,f
areas
Fig. 3.12. The microslip in the contact zone for a ball rolling along a grooved
track.
54
3.6.4. Rolling around a curl'e
A rolling ball has to be forced to roll around a curved path by the
application of a torque 1112 , fig. 3.13. This torque then creates
microslip patterns of the type shown in fig. 3 .13, where once again
the elastic straining of the surface can accommodate some of the slip
requirement.
'-.
'·
Plan view of ball rolling around a curved path
Fig. 3.13. The microslip in the contact zone for a ball rolling around a
curved path.
I y I F
Contact
zone ~ x
~
MF~
Fig. 3.14.
0· bF·
The friction forces in the contact between a tyre and the road.
tI D
LL'."
Q)
;:
-2
0, C
'=
~
Q)
C
0
u
Fig. 3.15. The Gough plot relating cornering force and self-aligning torque.
56
but we hardly notice any change at the steering wheel due to Mz.
From C to D is the most remarkable situation where we find reversed
feel at the wheel as Fy increases. Regions B to C and C to D are
strictly for the professional driver. As we move from A to B the
slip means that the centreline of the wheels is a few degrees out of line
from the direction of motion, but from C to D one has much greater
misalignment as the car drifts around the corner. The ultimate value
of F y is defined by µ W, the point at which pure sliding occurs and
control is lost.
µ=ton 8
Fig. 3.16. The value ofµ defined by an inclined plane.
Load
Pin
Friction
force
Fig. 3.18. The pin and disc friction and wear machine.
Flexible metal
Disc bellows
Vertical force
...- transducer
~-~--=L"'-'----L..J-----~----__.,__.-To friction
'----~~----~~------r force
transducer
Base
Vacuum pumps
Fig. 3.19. A pin and disc machine for use in a U.H.V. chamber.
Driven
disc
Test surface
w w
Shim steel
\
Frictionµ.= "#loge(;~)
Spring balances
61
in fig. 3.24. A slider is pushed along a flat surface by means of a
straight edge. The new position of the slider can be shown to be
defined by the friction between it and the straight edge.
In all friction experiments such effects as surface finish, surface
coatings and lubricants may easily be studied.
F µ. 13 W sin (8 +1/t)
µ. ---
12- N - fl-13 W cos (8+1/r)
= cot (8+1/rl
I /
\ I
\-:i;7
I . Flat surface ( 3)
. I
I• I
I
\ !
I I
I
8
~ Sliders
Straight
edge
(2)
Initial position
•
Fig. 3.24. The lloothroyd friction tester.
62
CHAPTER 4
wear and the properties of materials
4.1. Introduction
AT the most elementary level wear is manifested by a loss of surface
material from one or both surfaces when they are subjected to relative
motion. Sometimes the wear is clearly visible to the naked eye; in
other cases it requires the most elaborate measurement techniques
for its detection. The relative motion producing the wear may be
macroscopically observable as in many machine elements, whereas in
other cases the motion may be a small microscopic vibration, when we
speak of fretting. An example of fretting is the wear and subsequent
corrosion we can see between a bolt or rivet head and the surface in
systems subject to severe vibration.
At the outset we might also note some other characteristics of all
wear phenomena. Wear takes place in real time and we cannot affect
the current rate of wear of materials unless we alter the load, speed,
lubrication or environmental conditions. Nonetheless as time
elapses the rate of wear may either increase or decrease due to the
complexity of the mechanics involved. Although wear clearly occurs
at the surface contact which is also the seat of the friction mechanism,
the two are not simply related one to the other. Cases have already
been shown in table 1.1 where low friction does not necessarily mean
low wear and vice versa. Indeed even with a given combination of
materials one sometimes meets situations where as time proceeds the
wear rate increases and the friction decreases.
Two other general observations about wear can be made. In
Chapter 1 it was pointed out that wear may often prove beneficial as
in ' running in ' or when employing the principle of ' planned
obsolescence '. In a more obvious way we use wear in the workshop
in filing, grinding, and in a whole range of abrasive processes.
Secondly, we must be careful about the units in which we define and
measure wear. Suppose we define it as the mass or volume of material
removed per unit distance of sliding. Now consider the wear of a
shaft rotating in a bearing. If the shaft has diameter d and the bearing
length is L, the change or in radius r per revolution due to a volume
wear V per unit distance of sliding is obtained from
LTrdor = V Trd,
63
so that
V
8r=]_;
10-7
!':! 10-S
l?
L..
0
~
,o-9 ~ /
Transition
10-,o
Mild
10 10 2 103 10 4 10 5
Lood
Fig. 4.1. The transition phenomena in wear.
Vcx:AxLcx;xL
where L 1s the distance of sliding. Thus the adhesive wear law
becomes
V=KWL
H'
or in words the ' laws ' of adhesive wear are:
(a) The volume of wear is proportional to the distance of sliding.
This relationship has been justified by experience for a wide
range of conditions.
(b) The 'l'olume of wear is proportional to the applied load. This has
also been shown to be true in many tests for limited ranges of
load, although as wear mechanisms change with increasing
load some abrupt transitions have been observed; see fig. 4.1.
(c) The z•olume of wear is in'l•ersely proportional to the hardness of
the softer material. This has also been shown to be valid,
particularly for pure metals.
Recalling the physical nature of the wear process arising from
adhesion we can give a physical meaning to the constant of propor-
tionalitv K, often called the adhesive wear coefficient or the Archard
constari't. Surface films together with the height distribution of the
asperities sec to it that adhesive effects are only significant at a certain
proportion of the asperity contacts. Also the mere fact of adhesion
does not of itself detach ,vear particles, since shearing clue to sliding
could still occur at the junction interface. Wear particles are
probably formed where the junctions are stronger than the underlying
material, so that some material from the surfaces is torn away and
eventually released by the continued sliding. So we can see that K
may be interpreted as a probability factor, that is, the factor which
indicates the probability of wear particles being created by the
adhesive effects between the populations of asperities on the two
rubbing surfaces.
"'c,,
0
..J
I
69
Rolling-contact bearings show life characteristics significantly
different from those which apply to rubbing contacts. In the latter,
any wear process is initiated at the start of sliding and the wear is
progressive. The life of such a contact is therefore defined by the
allowable wear which can occur before the the component ceases to
function adequately. In rolling contacts the fatigue failure only
occurs after many cycles of operation, but when it does occur it
produces pitting which brings the bearing to the end of its useful life.
Indeed it is standard practice to define the life of rolling-contact
bearings in terms of their fatigue resistance. If we consider that a
bearing will fail by fatigue after N revolutions of the bearing under
an applied load W it has been found that:
For roller bearings W 2 N = Constant.
For ball bearings W 3N = Constant.
From Chapter 2 we already know that the stresses produced by
spheres are greater than those produced by cylinders so that this result
is not surprising. It is also apparent from the above that the lower
the applied load ( and hence stress) the longer is the life, which is
just the sort of behaviour \Ve ,votild expect from the S /N curve shown
in fig. 4.3.
There is one final point about wear due to fatigue as opposed to
other wear mechanisms, particularly adhesive wear. To produce
fatigue failure all that is required is that the surface material should be
loaded whereas the other mechanisms require not only loading but
also actual physical contact. If the surfaces are separated by a
lubricant film, adhesive and abrasive wear are virtually eliminated.
But the applied load is still transmitted to the solid surfaces through a
lubricant film, and can still cause stresses in the surface, so that
fatigue-type failures arc not excluded. This is particularly true in the
almost perfect fatigue environment of rolling contacts. It is found
that the propensity to fatigue-type failure decreases as the oil film
thickness increases. This is not surprising since as the oil film
becomes thicker it tends to smooth out the local surface stresses.
4.3.5. Fretting
This is not really a separate mechanism of wear but it is treated
separately because it arises in rather special circumstances. It shows
how one particular wear process may be a complicated combination of
several mechanisms of wear, and also demonstrates the deleterious
effects of any wear debris which may become trapped in the contact
system.
Fretting effects are associated with the contact of surfaces in which
the sliding motion is an oscillation of relatively small amplitude, often
only a few micrometres. Since vibrations occur in virtually all
machines we find fretting occurring between surfaces in contact such
as bolted components, splines and components located by friction
such as flanges shrunk onto shafts. In such joints the vibration gives
rise to small amplitude oscillatory displacements between the surfaces
in contact.
Such contacts can exhibit a form of adhesive wear. The debris
is trapped between the surfaces and can cause a second stage of
abrasive wear. In certain cases this effect leads to the interesting
combination of a reduction in friction, but an increase in wear. The
increase in wear arises from the second stage abrasive effects while
the reduction in friction occurs because the oscillatory motion tends
to produce rough spherical wear debris particles which then act rather
like a series of small balls separating the surfaces.
F 71
Fretting often occurs in a corrosive environment such as air where
the wear processes tend to be accelerated by the rubbing off of loosely
adhering corrosion films and the increased abrasive action of the
harder oxidized wear debris. In such cases the process is known as
fretting corrosion and with steel in air is characterized by a fine reddish-
brown powder known as ' cocoa '. Fretting effects often occur in
machines during transit due to the vibrational shocks incurred. For
this reason great care is usually taken to unload all critical contacts
in the machine before its transit. The repeated ' hammering '
suffered by the stationary ball bearings of a machine during a long
journey leads to a form of fretting called ' false brinelling ', and
consequent unsatisfactory running ·when the machine is subsequently
operated.
...
0
~
H/3
I
Contact pressure
Fig. 4.5. The effect of contact pressure on wear rate.
72
12% chrome steel
I Ni-Cr-Mo
(/)
(/)
steel
Q)
C:
~
0
I
should always try to keep the applied normal load below that value
which gives rise to nominal contact pressures approaching H/3, where
H is the hardness of the materials at the operating contact temperature,
which equals the ambient temperature plus the temperature rise due
to frictional heating.
Note that K varies by a factor of 3000 for the metal pairs in the above table.
b.c.c. f.c.c.
h.c.p.
-----v
Area A
Fig. 4.8. The change in dimension due to wear.
Consider a block of this material sliding over a metal surface, fig. 4.8.
The rate of energy dissipation against friction isµ, Wv. It is reasonable
to assume that the volume wear rate of the block, V = - d V /dt, is
proportional to this rate of energy dissipation, so
Vocµ,Wv.
The block will therefore wear to a depth d such that the volume
wear V is given by
V=Ad.
Hence its rate
V=Ad
where d is the rate of increase of d with time. Thus
. Ji µ,Wv
d= A oc -r ocpv.
This shows that the rate of change of dimension of the block is
proportional to the product pv. This is the rate of change of bearing
clearance with time in any practical application and as such is a more
useful parameter than the actual volume of material removed.
76
For any material the allowable value of the pv product may be
defined and fig. 4.9 shows a typical result of PTFE if the wear rate is
to be a dimensional change of 25 µmin 100 hours.
Plastic bearings and particularly PTFE bearings must be operated
within their approved pv ratings. These ratings are associated with
wear and are particularly subject to thermal effects due to the de-
composition of the surface. Thus at higher ambient temperatures the
pv factor for such materials is considerably reduced. Such materials
when used in practical bearing designs are often associated with a
metallic matrix which provides additional strength and improves the
thermal conductivity thus allowing the easier escape of the heat
generated in rubbing. When plastics slide, problems can arise from
the generation of electrostatic charges. In such situations, designers
must incorporate earth paths to minimize the build-up of charges.
N
I
107
E
z
0
a..
'- 106
...:,
Q)
en
en
Q)
a: 105
10- 1 10
Speed / m s- 1
. ·A
I
Retaining cup
Fig. 4.11. The four ball wear machine.
78
an indication of the volume wear. This arrangement is extremely
popular for industrial testing since the test specimens, the balls, are
readily available at low cost from ball bearing manufacturers. Fig.
4.11 b shows the so-called 'disc machine' which is very useful for the
study of wear under combinations of rolling and sliding. When the
peripheral speed of both discs is the same one has pure rolling whilst
a difference in speeds implies some additional sliding. This rather
complex roll/slide process often occurs in machinery, perhaps the
best example being the contact between meshing gear teeth depicted
in fig. 4.12. At the initial contact we have rolling and sliding between
the teeth which becomes pure rolling at the pitch point B followed by
rolling with sliding in the opposite sense during the arc of disengage-
ment.
After
sliding
behaviour. Fig. 4.14 shows two profiles of the same surface, one
before use and the other after wear.
Dial
gauge .____.--"9
Cantilever ---- .,,,- .-~
fr1ct1on
indicating()=
r c,'"
0
oe\ ~o 0\e
Weights
Fig. 4.15. A simple crossed cylinder wear machine for use in a lathe.
82
CHAPTERS
properties and testing of lubricants
5.1. Introduction
ALTHOUGH oil is a lubricant, it is not true that lubricants must be oil.
In fact almost any fluid may act as a lubricant, and its most important
single property is its viscosity. The term fluid is also to be interpreted
in its widest sense as including all liquids, gases and vapours. Thus
in many applications it is possible to use that very cheap fluid, air,
as the lubricant in high speed bearings.
It is interesting to ask why oils should have become so extensively
used as lubricants during the past two hundred years. Their
effectiveness resides in their viscous properties, their non-corrosive
properties and their price and availability. After air and water, oil is
the cheapest and most plentiful fluid available on this planet.
Although air is increasingly used as a lubricant it is really only
effective for certain load and speed conditions, and whilst water is
also used as a lubricant its corrosive effect on metals makes it unattrac-
tive as a general lubricant.
5 .2. Viscosity
This is most readily recognized as that physical property which
defines the resistance to flow so that we often associate high viscosity
with such adjectives as ' thick '. The scientific definition of viscosity
is due to Newton, who in the seventeenth century introduced the
idea of layers of fluid sliding over one another, as shown in fig. 5.1.
This sliding induces interfacial shear stresses between the layers and
N e,vton postulated that these shear stresses were proportional to the
rate at which the shear strain was occurring. Thus in fig. 5.1 if we
assume that each layer has a velocity which increases linearly by an
amount ou across a vertical height oy,we may write
OU
TOC-
oy
or in the more conventional form, the tangential stress T is given by
du
T=Y)-
dy
where the constant of proportionality is the dynamic viscosity YJ and
du/ dy is the velocity gradient across the film thickness. The units
83
of viscosity are clearly those of a stress multiplied by a distance and
divided by a velocity. Thus the SI unit is N s m 2 , but the most
commonly used unit is a c.g.s. unit the poise (P) dyn s cm- 2 • For
many lubricants the poise is too large a unit so that the centipoise
(cP) which is one hundredth of a poise is more commonly employed.
Since one poise is 0· 1 N s m- 2 , 1 cP = 10- 3 N s m- 2 • Water has a
dynamic viscosity of 1 cP, air a value 0·02 cP, and lubricating oils
have values which may be as high as 400 cP or as low as 2 cP.
t
t
8y-------~----.-' +-u + 8u
,-;'--'-------,---'- u
y
t
-...-u
Fig. 5.1. Newtonian flow.
Z' VI =O
'iii L
0
u
VI
·;; u
.S!
0 H
.,
E
C:
:.::
Temp./°C
~
·.;;
0
-~ I ·5 X 10- 2
>
30 60 90 120·
Temp.l°C
In general instruments of types (a) and (b) are only suitable for
measuring the viscosity of Newtonian fluids, whilst type (c) may be
used for a wide range of fluid characteristics. Since the effects of
temperature and pressure have already been established as of para-
mount importance they must be controlled in all measurements of
viscosity. \Vhilst the effects of temperature on viscosity may be
studied in almost all types of viscometer the effects of pressure may
only be studied in a very few instruments.
_ 7Td4t/j.p
Y)- 128lq
/j.pbeing the pressure drop along the capillary. In practice the above
equation has to be slightly modified due to the end effects where the
fluid is suddenly accelerated at entry to and exit from the capillary.
Provided l / d is large, say greater than 500, these end effects are fairly
small corrections.
G 87
This type of viscometer may be used to make ' absolute ' measure-
ments of the viscosity using the Poiseuille formula together with any
' end effect' correction. More often, however, these viscometers
are used as comparative instruments. The instrument is calibrated
using fluids of known viscosity to establish its constants whence it
may then be used to determine the viscosity of any other fluid. A
commonly used viscometer of this type is shown in fig. 5.4. In use
the fluid level is adjusted to mark A and is then drawn up to a point
just above B. The time taken for the fluid level to fall from B to C
is noted and used to determine its viscosity.
-8
-c
Capillary
A- tube
2 1-14 31 32
6 1·48 41 46
10 1·84 52 59
14 2·22 65 74
18 2·65 78 90
25 3·46 105 120
35 4·71 144 164
45 5·99 184 209
60 7·92 245 279
80 10·60 326 371
100 13-2 406 463
150 19·9 620 700
200 26·8 820 940
300 40·0 1230 1410
500 66·0 2040 2320
700 93·0 2820 3250
1000 133 4100 4750
2000 260 8100 9200
3000 400 12 300 14000
5000 660 20 000 23 000
89
5.5.3. Falling-body viscometer
In the simplest type of falling-body viscometers one finds the
viscosity from the terminal velocity of a steel sphere falling under
gravity (fig. 5.6). When the sphere has reached its constant terminal
velocity v it is in equilibrium under the action of gravity, buoyancy
and drag forces, since the buoyancy and drag forces must equal its
weight.
Now weight force is J4 7TY 3 psg,
4
the buoyancy force is 7TY 3p1g
3
and the drag force is 6mirv (Stokes's formula),
where r is the radius of the sphere and p 8 and p1 are the densities of
the steel and the liquid respectively.
From weight= drag+ buoyancy we obtain
2 r 2(ps - p1)g
1)=9 V •
Guide tube
Gloss tube
Sphere----+-~
Timing
marks
Test oil---l--
Clearance
for test oil
Fig. 5.9. The principle of the disc machine high pressure viscometer.
92
pressures of the fluid in the contact region of the order of 108 Pa (N m- 2 ).
From a measurement of the torque transmitted across the fluid film
and the rate of shearing one may thus readily obtain the viscosity at
very high pressures. Such information is essential where lubricants
are used between gear teeth or between balls and rollers in rolling
bearings. Indeed it will be readily appreciated that the disc machine
itself is useful simply because it represents these types of high-
pressure contact.
5.7. Greases
Greases usually consist of a thickening agent mixed with a standard
mineral oil, often with other special additives. For certain special
greases the mineral oil may be replaced by other fluids such as silicone
fluids where a high temperature grease is required. Metallic soaps
such as calcium, lithium and sodium soaps are commonly used as
thickeners, whilst clays and silicates are also used for this purpose.
The grease has the characteristic appearance under the microscope
of a tangled web of fibres of the thickener suspended in the base oil
as seen in fig. 5 .10. The inherent stiffness of greases enables them
to be used in a wide variety of practical situations. In ball bearings
they both fulfil the role of lubricant and act as a very effective seal
against the ingress of dirt and moisture. Their stiffness also allows
them to be used in situations where a liquid lubricant would readily
drain away.
As with other lubricants several tests are carried out to define the
characteristics of greases. Perhaps the most important aspect of
greases is that they tend to be altered during operation rather more
than simpler lubricants. In high pressure contacts such as ball
bearings the fibrous structure of the grease is broken down and this
leads to a reduction in lubricating efficiency.
94
Calcium base.
Sodium base.
96
CHAPTER6
externally pressurized bearings
6.1. Introduction
Where a fluid film separates two solid surfaces it must be under
pressure. Indeed, in the absence of such a pressure any load applied
to the solids would simply squeeze the fluid out of the gap. Consider
a block of area A subjected to a load W and separated from a plane
by a film of fluid held at a constant pressure p, as in fig. 6.1. The
film can only be sustained provided that
W=pA.
If p is not constant at all points, so that each element SA carries a
load SW,
SW=p SA.
For the total effect
W= Jp dA (6· 1)
where the integral is taken over the total area. For simplicity we
shall restrict our examples to two-dimensional cases where the above
equation would produce the results shown in fig. 6.2.
,,-, Pressure p
Area A
~
I ! ! : I i :
'
tt H
! : I
rt t +~ l ~ :!+i
I
I ◄--· L--~:
!
i-.------·-· L----·--+-i
-q
-,~i,------,:h
p +3p
3x
ii'
t
p
',
Fluid flow theory justifies these arguments and yields the following
two results for the cases which are of most interest.
For flow along a uniform pipe of diameter d (fig. 6.4) one obtains
the Poiseuille formula:
q= _ d (dp) = ~
77 4
12817 dx 12817 l
(P2-P1). 6 .2 ) (
For flow between substantially parallel surfaces (fig. 6.4) one obtains
3 3
q= _h dz (dp) = h dz (p 2 -p 1 ) ( 6 _3 )
1217 dx 1217 l '
the negative sign simply indicating that the flow must be in the
direction of decreasing pressure, i.e. if dp /dx is positive the flow
Pz
.e- - - - - - ~P1
q-
( t
d
t I --q
Pz P1
-.e
82
q-
t
h
Fig. 6.4. The flow down a circular tube and between parallel plates.
98
would be in the negative x direction. The similarity of these two
equations and their consistency with our physical arguments is
evident. Equation (6.3) is important in fluid film lubrication and we
shall return to it when we deal with hydrodynamic lubrication.
This type of fluid flow, accompanying a uniform pressure gradient,
is often referred to as Poiseuille flow.
Applying equation (6.3) to the simple example shown in fig. 6.5
we find that the pressure variation is of the form shown since the flow
of fluid in each section must be the same.
. dp 1217q h
1 e -=
. · dx
---=-
h 3 dz h3
(6.4)
where h is a constant.
We immediately see that the pressure drop along the large section
is very small in relation to the other section, and for such large sections
we can therefore assume a constant pressure.
t
q- -q
~ Pressure distribution
Fig. 6.5. The pressures due to flow from a large to a narrow gap.
~~~~~~~~1!~~~~
-=-=-=-=~-=-=-=-=-=-=-=-=-=-=-
J_ h
Pressure
/ distribution
I-+-----
Fig. 6.6. The pressure distribution in a simple hydrostatic pad.
Thus
S= 3qY]l
P ha
and
W =Psl = 3qYJz2 (6.5)
2 2h 3
100
In this arrangement since q is constant we see that any change in W
is accommodated by a change in the film thickness h, and the supply
pressure Ps now varies to ensure a constant flow rate q. The
stiffness of this bearing is defined as the rate of change of W with
film thickness h.
dW 9(/YJl 2
Thus stiffness - dh
- =+---- 2h 4 •
(6.6)
W=
3
f;;t
(b+l/2)
and the stiffness is
dW 9q,YJl(l + 2b)
s= - dh =+ 2h 4 • (6.7)
I ~
~~~
1
J. .- -
-I ½1~~-b------4 ½1--
Pressure
distribution
t t
Fig. 6.7. The modification of the pressure distribution due to a recess in a
hydrostatic pad.
101
This is an increased load capacity as can be seen by comparing
figs. 6.6 and 6.7 and this type of geometry is that most commonly
employed in externally pressurized bearing designs.
We appear to have satisfied the design requirements for this type
of beating, so why should we consider any alternative? The answer
is that the use of a constant flow type of pump implies that every
bearing would require its own separate pump, a very expensive
arrangement. The advantage of the other alternative, a constant
pressure pump, is that the same pump may supply a whole series of
bearings provided that its flow capacity can meet their total require-
ments.
l the z direction
~
/. ~,%·,.,. ,. ,.-~1/~~~~.~
----------------·
t
le
t
102
The linear pressure drop along each half of the bearing ( through
which flows half of the fluid) is governed by:
<j _ h 3w Pr
(6.9)
2- 1271 l/2'
where w is the width of the bearing m the z direction. Thus
eliminating q one obtains
_ Ps
Pr-1 +ha/k
where
31rd4l
k= 128/cw ·
The load carried is then given by
W=Prlw
2
Pslw
=~--- 3
2(1 +h /k)
and the stiffness is
dW 3 (
s= - dh = 2k 1 +h 3 /k
h )2 p slw. (6.10)
This equation relates the load to the film thickness for a given
geometry of bearing. The flow through such a bearing is then found
from equations (6.9) and (6.10),
. k(pslw-2W) (6.ll)
q= 371[2 •
Bearing width w in
the z direction
• _ 1T C d2 ( Z(Ps -Pr))
q- - cl (6.15)
4 p
where p is the fluid density, d is the orifice diameter and Cc1 is the
discharge coefficient of the orifice. When such a compensator is used
the preceding analysis must be reworked starting from equation (6.15)
instead of equation (6.8), and this leads to rather more complicated
expressions for the load capacity, stiffness and oil flow.
104
6.3. The characteristics of capillary-compensated bearings
An examination of the physical meaning of equations (6.12),
(6.13) and (6.14) is rewarding since they indicate the many interesting
features of this type of bearing.
(a) For a given geometry of bearing we note from equation (6.12)
that the load capacity depends on the operational film thickness
and the form of the relationship is shown in fig. 6.13 a. We
see that the load capacity rapidly increases with reducing film
thickness.
(a)
w
(bl
s
I
~
h
(cl
q
h
Fig. 6.10. The load, stiffness and flow rate in hydrostatic pads related to the
film thickness.
(b) For the same bearing we find that equation (6.13) for the
bearing stiffness has the form shown in fig. 6.10 b. Here we
notice that there is an optimum film thickness which provides
the maximum stiffness for the bearing, although the load
capacity is smaller than it would be for thinner films. This is
important since one of the main attractions of this type of
bearing is its high stiffness and the ability of the designer to
optimize this characteristic.
105
(c) Substituting for Win the flow equation shows that the oil flow
increases with increasing film thickness in the manner shown
in fig. 6.10 c.
(d) We note in passing that the load capacity, the stiffness and the
flow rate are all increased by increased supply pressure.
(e) Also, the load capacity depends very little on YJ, whereas the
oil flow depends on 1 / YJ· So this type of bearing, being less
dependent on viscosity than hydrodynamic bearings, may
operate using a wide range of fluids.
(f) For a bearing of the same overall dimensions changing the
geometry to make b large and l smaller increases both the load
capacity and the stiffness and slightly reduces the oil flow.
w
(a)
t
(b)
I
--------~
J _ _ _ _ _ _ _ ..J
I
/ ..... - .....
I
I ' \
- ___J . ..!.... - 0 - · I-, I_,
I I / I
! \....... _ ..,/ !
I
I
Drain
channel Supply
t
Fig. 6.11. Single and multi-pad hydrostatic journal bearings.
106
6.4. Hydrostatic journal bearings
So far we have considered the hydrostatic bearing in the form of
plane pads, but the same system may be applied to journal bearings.
Fig. 6.11 shows a single-pad and a multi-pad arrangement for such a
bearing. Each pad is supplied separately from a constant pressure
supply via appropriate compensating elements. The oil then flov,·:;:
over the lands into the drainage channels and the calculations of the
preceding section apply. In fig. 6.11 a the load is carried by the
single pad, and this must therefore be aligned in the direction of the
load. In fig. 6.11 b the sequence of pads allows such a bearing to
accommodate a load in almost any direction and is useful when the
direction of the load may change during operation.
t
R
____ Velocity
distribution
1
h' not to scale
The frictional force F is then the integral of these shear stresses taken
over the operational area. Thus for a unit length of shaft,
F=TlX 1
wR
=YJhl (6.17)
The frictional torque is FR for this simple case. For the plane pad
type of bearing a relative sliding velocity U gives rise to a frictional
force YJUl/h for each unit width of the pad in the z direction. For
such a system the coefficient of friction would then be F divided by W.
We can see the major advantage of these bearings. The friction is
directly proportional to the sliding velocity, being zero when the
bodies are at rest, and still relatively small at fairly high velocities.
Drain
channel
Pressure
di sir i bution
Fig. 6.13. The multi-pad and the multi-recess hydrostatic journal bearing.
(a)
!Applied load
Supply
pressure
/Tapered
f i I rn
109
if Uh /2 = q/2 no flow occurs out of the trailing land and if Uh /2 > q/2
the fluid tends to flow into rather than out of the trailing land.
We should mention one final point which arises from motion.
Under certain circumstances pressures may be generated in the fluid
films by hydrodynamic effects. Such pressures will be additional to
those due to the external pressurizing system so that the load capacity
for bearings in motion may be greater than that predicted above.
Examination of the pressure distribution around a multi-pad
bearing in fig. 6.13 a shows that in the drainage channels the pressure
is zero, which means ambient atmospheric pressure, so if one
eliminates the drainage channels one increases the load capacity.
This leads on to the so-called multi-recess design of externally
pressurized journal bearings shown in fig. 6.13 b. The fluid from
one recess now flows across the land into the adjacent recess and in
this way increases the load capacity.
110
CHAPTER 7
hydrodynamic lubrication
7 .1. Introduction
WE have already established that if two loaded solids are to be
separated by a continuous fluid film such a film must be pressurized.
The preceding chapter gave examples of externally pressurized
bearings, hydrostatic bearings as they are often called. Now we shall
consider situations where the pressure is produced by virtue of the
viscosity of the fluid and the relative motion of the two solids. Such
bearings are often called self-acting bearings, and the lubrication
method is referred to as hydrodynamic lubrication.
.
t St t nary
ed pad
Direction of
pressure build-up~- _____ _
_-:.._-_-_-_-____ -- ---
_-_-_ - - - - - - - - - -_-__
-----u
Fig. 7.1. The convergent oil film.
Fig. 7.2. Convergent oil films in slider and partial journal bearings.
(a)
~ h
T X y 0
~L U t
(cl
=O dp.
- 1s- ve
dx
------ Pressure
I distribution
x=O x=L
(d)
up distribution up is - ve up is 0
I
~
(el
1/ .,/-'
==
Velocity
distribution
hm ~
U = Us+ Up
u u u
Fig. 7.5. The Yelocity and pressure distribution in an inclined pad bearing.
115
The form of Us is shown in fig. 7.5 b; Us is + U at y = 0 and zero at
y = h. This is known as Couette flow.
With converging walls the relationship between the fluid flow and
the pressure build-up (Poiseuille flow) is
U p =1- ( - dp)
- y(h-y). (7.2)
271 dx
We must know how the values of dp/dx vary with x, and shall assume
that this variation is as shown in fig. 7.5 c. The form of up is then
as in fig. 7.5 d and Up is zero at y = 0 and y = h. Furthermore, at the
entry section dp /dx will be positive, near the centre it will be zero,
and after this point it will be negative. This gives rise to the patterns
of up shown in fig. 7.5 d.
Since the pressure gradient and the shear flow are the only two
causes of fluid flow, the resultant velocity of the fluid follows the
patterns shown in fig. 7.5 e and is given by
_
ll-Us+Up-
_ GT(h-y)
-- + -1 ( - - dp) y(h-y). (7.3)
h 271 dx
As we are neglecting any side flow, the area of an element of film is
1 x dy per unit width of the film. The quantity flowing per unit
time is thus 11 <ly for each such elemental area and the total flow q is
given by:
+h
q= J 11dy.
0
. (2U) + 1271
q= h
h
3
(
- dp)
dx • (7.4)
But the flow through the film must be the same at all values of h so
that q111 = q and thus
116
This equation was first derived by Osborne Reynolds and is known
as Reynolds' equation. It shows why geometric convergence is
essential. If h did not vary with x, then h would equal hrn and dp /dx
would be zero at all values of x so that no hydrodynamic pressure
would ever be produced. In the case considered ,vc can also sec that
h at any value of x is defined by the equation,
Il= Il j - -hi - ho
L--X.
dp = l Z ( U a+ U
dx Y/ 2
b) h -h hm
3 •
(7.10)
~ _____._LJb
~
(a)
~
• hf - +
X ~-+-Ua X -----.- Ua - Ub ---.- Ub
h invariant
(b)
+ +
U0/
/
h invariant
(cl
~ + +-+-Ub
Fig. 7.7. Some examples of the resolution of the velocity components for the
application of Reynolds' equation.
118
7.2.3. Practical examples
We are now in a position to identify a number of practically
important situations in which pressures may be developed by hydro-
dynamic action. The first simple example is a fixed tapered pad
(fig. 7.2) which can be used in a simple linear bearing, in a thrust
bearing, or in a journal bearing. Such pads, instead of being fixed,
may be mounted on pivots, springs or elastic supports, as shown in
fig. 7.8. These flexible mountings are used to enable the pads to
take up that inclination which produces the best solution to the
particular operational conditions of load, speed and viscosity. This
is a classical example of the principle that generally ' Nature knows
best ', so that if systems are flexible they tend to take up the optimum
operating position.
~ ~ ~
u u u
Pivoted pad Elastic support Spring mounted
With journal bearings we have already seen that if the shaft runs
about a centre which is eccentric to the bearing centre the necessary
convergent film is obtained. Thus in fig. 7.9 a a partial journal
bearing is essentially the same as a curved pad bearing whilst the full
journal bearing is shown in fig. 7.9 b together with the associated
hydrodynamic pressure distributions.
w w
Pressure Pressure
distribution distribution
(a) (b)
Fig. 7.9. The pressure distribution in a partial and a full hydrodynamic
journal bearing.
119
For balls and cylinders rolling around the races of rolling-contact
bearings we are dealing with hydrodynamic films generated as shown
in fig. 7.7 c. A rather interesting example of the same kind of hydro-
dynamic film generation occurs between mating gear teeth, which as
shown in fig. 7.10) may be considered to be instantaneously rolling
and sliding one upon the other. Thus at any instant they are entirely
equivalent to two cylinders rolling and sliding one upon the other,
that is, to two cylinders rolling together with different peripheral
speeds U1 and U 2 . Such a system is then able to generate hydro-
dynamic pressure, and can operate with a fluid film under load,
according to the arguments discussed in fig. 7.7 busing equation (7.10).
Fig. 7.10. The hydrodynamic film between gear teeth and the equivalent
system.
E
-- a::
0
Material --+-
dp = 12 ( U) h - h111
dx YJ 2 1z3
and
where
n=---
hi l
ho
provided x is measured to the left from the ongm O; sec fig. 7 .12.
This choice of origin simplifies the ultimate solution. We obtain by
integration
_ _ ____.._u
-+--- L ------
x....__ __..,0
~--
.,",,I.,..,. ']\ Pressure
/I .
,/r t l
1--------distribution
\
x=L x~O
hm=2h 0 ( 21 +n)
+n
and (7.12)
A= L
n(2+n)
Substituting these back into equation (7 .11) gives the form of the
pressure distribution for this case as
122
The form of this pressure distribution is shown in fig. 7.12 and we
then define the load W per unit width of the bearing by
x=L
W= J p dx
x=O
where
K=loge(l+n)_ 2 (7.15)
n2 n(2+n)
Before we consider the physical meaning of these results let us
recapitulate the steps ,,·hich must always be taken in dealing with
any hydrodynamic lubrication problem.
Step 1. State the correct form of Reynolds' equation for the case
being considered.
Step 2. Define the ,vay in which the film geometry changes with x
for the particular contact and substitute for h in the
Reynolds' equation.
Step 3. Integrate the Reynolds' equation to obtain the pressure
distribution as a function of x. Since we shall always have
two unknowns such as h111 and the constant of integration,
we introduce two appropriate boundary conditions.
Step 4. Decide which are the most appropriate boundary conditions
and use these to find the unknowns h111 and A, and so
completely define the pressure distribution in terms of x
and the constants of the system.
Step 5. Obtain the load appropriate to the derived pressure distri-
bution by integrating the pressure over the area over which
it acts.
\Ve now return to the important physical aspects of the results
obtained for our simple slider bearing, since these will have a general
validity in hydrodynamic bearings of many types.
The effect of the degree of inclination of the slider is totally defined
by the value of K, since this itself depends on n = lzi(h 0 - 1. The
values of K for various values of n are given below:
123
Changes of more than ten times in the inclination factor n cause
very little change in K, and therefore the load which is carried for
given operating conditions. This confirms the earlier statement that
the main requirement is that a convergent film shall exist rather than
that the degree of convergence has to be very large.
Secondly we note that by rearranging equation (7.14) the minimum
film thickness h 0 is defined by
(7.17)
Fig. 7.13. The scale of the roughness to the mm1mum film thickness 111
hydrodynamic lubrication.
u u u
----u
du) < U
(dy du)
(dy U (du) > U
y=O hm y=O = hm dy y=O hm
~-------- L -----------~
Fig. 7.14. The Yelocity distribution in an inclined pad bearing.
125
Putting these equations together therefore defines (du/dy)v,~o at all
values of x along the lower surface. The frictional force at the lower
surface is then easily obtained, since it is simply given by
We may now substitute all the appropriate values from the preceding
equations into equation (7 .20), carry out the integration and obtain
(7.21)
where
4 loge (1 +n) __6_)
(
Kr= n 2+n
[
6loge(1+n)_ 12 ]1;2
n2 n(2+n)
The coefficient of friction µ, = F / W for this bearing then becomes
(7.23)
(a)
t
2
(b) '/////////##///////##/&
Direction of flow -
t h _____._ Direction of flow
o--~t____ x
(cl
Fig. 7.15. The hydrodynamic pressure arising from the squeeze film effect.
The fluid flow will be to the left when x < Xrn, where dp /dx will be
positive; and to the right when x > Xrn, where dp /dx will be negative.
Thus at x = Xrn we are not surprised to find dp / dx = 0, so that Xrn is
the point of maximum pressure.
By considering a small reduction in film thickness dh in a time dt it
is easily seen that the outward flow rate at a given value of xis given by
127
This means that the flow rate increases from zero at Xrn to a
maximum value at each end of the pad. Combining equations (7.23)
and (7.24) produces the expression for the pressure development:
dp_12 (V: V)X-Xm (7.25)
dx - Y/ z- i ~ -
The term V2 - V1 occurs since dp /dx must be positive when x < Xm.
Integration of this equation using the appropriate boundary
conditions of p = 0 at the edges of the pad then produces the pressure
distribution of the type shown in fig. 7.15 c. The summation of this
pressure over the area over which it acts then gives the additional
load capacity due to squeeze film effects.
Examination of equation (7.25) shows a remarkable similarity of
form to our original Reynolds' equation (7.8) and ,ve may combine
these two equations to give
dp =l 2
dx Y/
(U +U
2
1 )h-hm 12 (V-V)X-·Xm
2
Jz3 + r1 Ji3 . z i
(7.2 6)
vbk tvb
~ --ub
+
0 X 0 0
Ua~va --ua It Va
I t
-ub
~
No pressure No pressure
effect effect
0 0
--ub t Vb
+ +
~di::~ L»',w'~~
0-----
-ua-Ub
0------'-
tVa-Vb
dp =l 2
dx YJ
(!J +U2) lph-pmhml·
1
2 ph 3
(7.27)
(7.29)
(7 .31)
2
/
0-'l.-
r (al
/
:::----.,._ U2
h Ye c,_,--:: lsoviscous pressure
0
1/U1
Fig. 7.17. The hydrodynamic pressure distribution between loaded discs for
the isoviscous and the pressure viscosity cases.
~ Basic geometry
low loads
-,
/ , /Pressure
I I
/ I
~ I I
\
,,, Initial deformation
due to high
~
loco I pressure
/i
/ \/Pressure
,.--✓ \
/ \
Fig. 7.18. The nature of the deformation of the contacting discs clue to the
hydrodynamic pressures generated.
1. Rigid-isoviscous
These are situations where the elastic deformations may be
neglected and the effects of pressure on the viscosity of the fluid may
be negligible. This is the situation discussed at the beginning of
this section, so that the minimum film thickness h 0 is given by
t~ = (it). 4 ·9 (7.33)
2. Elastic-isoviscous
Herc we incorporate the effects of elastic deformation but again
assume that the viscosity is constant. For such situations we find
that
Ji 0 = 2 .35 (
R rv
0 0 4
·o ( ~ ) ·
RE'
riU)
(7.34)
133
In this solution we note two groups, one as it were defining the
hydrodynamic effects, and the other the elastic deformation effects.
The latter group contains the elastic constants of the materials since
(7.35)
01
(a)
Disc
W/2 w 0 W/2
Oil
' --- -- - - -
(bl
Pressure
distribution
Oil
138
CHAPTERS
the selection of tribological solutions
8.1. Introduction
THE foregoing chapters have identified the scientific principles on
which most tribological solutions depend. Although some examples
of these principles have been discussed they have so far only been
included as illustrations of specific methods of solution to tribological
problems. As yet no attempt has been made to demonstrate the
rules which are used in deciding on, say, a rolling-contact bearing
rather than a hydrodynamic bearing, in a small electric motor. In
most practical situations one method of solution will have advantages
over the others. Also not surprisingly the ' rules of the game '
follow logically from the knowledge of the scientific principles
involved in each method of solution. Since industrial situations offer
the widest scope for the application of tribological knowledge we shall
concentrate on problems taken from this area.
Normal load
~
~
- Friction
wear
Environment
The ' black box ' concept of the tribological solution is of two solid
bodies subjected to a normal load, having relative motion to each other
and the whole operating within a defined environment as in figure 8.1.
The effects from such a system which concern the tribologist are
simply friction and wear.
139
8.2. Environment
The environment may be atmospheric, vacuum, chemical vapours,
water, etc., e.g. the bearings supporting a ship's propeller operate in
sea water whilst some of the bearings of a space vehicle must operate
in the very high vacuum of outer space. The majority of industrial
contacts operate in a normal atmospheric environment, although even
here the ambient temperature may vary from very low values in
certain refrigeration machinery to very high values in such industrial
plants as steelworks.
The complete range of environmental problems cannot be dealt
with in this book, but it will be obvious that in what follows the
designer must always consider the particular environmental problems
when choosing a tribological solution. Thus mineral oils cannot be
used in hydrodynamic bearings operating at very low temperatures
since the oil would solidify, whilst the use of dry plastic bearing
materials at very high temperatures is impossible due to the thermal
degradation of such materials. Over the years engineers have shown
considerable enterprise in overcoming the problems of the environ-
ment as the following examples indicate.
In the liquid-metal-cooled nuclear reactor certain bearings in the
pumping circuit have to be capable of prolonged operation within
the liquid metal environment. This is an essential safety feature
since any atmospheric contamination of a material such as liquid
sodium would be clearly disastrous, so that such bearings have to be
' sealed ' into the system. In these situations the use of conventional
bearings appears impossible until we remember that we already have
a liquid available, liquid sodium! The design which is employed
therefore uses pressurized liquid sodium as the lubricant in a logical
design extension of the conventional hydrostatic bearing.
In bearings which are to be used in ultra-high vacuum applications
such as certain X-ray equipment and in space vehicles, the metallic
parts of bearing components would rapidly weld together. This
problem was discussed in Chapter 3 where it was shown that in the
absence of protective surface films, such as oxides in the atmosphere,
the surfaces weld together to give very high values of friction. The
use of liquid lubricants in such situations is unacceptable since in
such a vacuum the lubricant vapour pressure is still high enough for
the lubricant to evaporate. A well tried solution to this problem is
to use ball bearings and to coat the balls with a solid lubricant. The
' lubricant ' which is often used is a soft metal such as lead, coated
to a thickness of a few micrometres. Bearings treated in this way
can operate satisfactorily for several years in such applications as
communication satellites.
An alternative solution to this problem uses rather more conven-
tional solid lubricants such as molybdenum disulphide. With such
140
lubricants the cage of the ball bearing is made from the solid lubricant
and on each revolution of the balls in the bearing a little lubricant is
transferred to the balls as they slide against the cage. This small
quantity of lubricant is sufficient to lubricate the contacts between the
ball and the races of the bearing.
8.3. Load
The single resultant force acting between bodies in contact is
conveniently considered in terms of two components at right angles
to one another. The normal component of this force is the load
which is being applied across the contact, although it may not be the
externally applied load. Thus if an applied load W is carried on an
inclined slider, the normal load at the contact is W cos ex; see fig. 8.2.
Likewise the contact of a dry rubbing journal bearing would be as
shown in fig. 8.3, where an applied load W gives rise to a normal load
of value W cos ex at the contact. (In these diagrams the clearances
between the shaft and the bearing housing are grossly exaggerated
for ease of understanding.) If, and only if, the two bodies in fig. 8.2
and 8.3 had a constant relative velocity would the system be in
equilibrium, when the components W sin ex would be the friction force
at the contact and we should have tan ex=µ. In the case of the dry
journal bearing the driving torque would then be WR sin ex.
In what follows we shall use the term load to define the normal
load carried across the tribological contact. Thus each particular
practical application will have to be analysed to determine the relation-
ship between the load as defined .here and the actual applied load
carried by the particular device. In most cases these two values of
load will be very nearly the same and in many cases are actually
identical.
Load, defined as a force, is a vector quantity, with direction as well
as magnitude. The magnitude of the load may be constant, as for
example in the main bearings of a large electric alternator. It may
vary in magnitude as happens in the main bearings of a car engine
141
throughout the inlet, compression, power and exhaust strokes of the
cylinders. The actual nature of the variation in magnitude of course
depends on the particular application considered. Fortunately the
changes in magnitude of the load do not normally cause much of a
problem since, although an increase in load produces an increase in
the frictional force, the coefficient of friction is not materially affected.
The main difficulties arising from varying magnitude of the load are
unwanted vibration and/or instability and, if the magnitude becomes
very large, breakdown of the bearing. On the other hand, a rapidly
increasing load may even have beneficial effects. For example, in a
lubricated bearing the higher the pressure of the lubricant film the
greater the load-carrying capacity. Consequently a rapid increase
in load will tend to produce higher pressures in the lubricant film by a
squeeze film effect (Chapter 7, section 7.2.6).
~Piston
w
/
l\Jagnitude Dircct£on
Constant Constant
Varying Varying
Applying the same format to the speed conditions we find there arc
eight possible speed conditions:
Varying
~ Varying
X Varying
(a) (b)
logW logW
~
Static strength
Inertia
force
limit
log U log U
(cl (d)
log W
logU
V = ~ WU. (8.2)
Thus a specified wear rate for given materials will always be defined
by a straight line of slope - 1 when log Wis plotted against log U as
shown in fig. 8.8 b.
log W
Safe Safe
operating operating
regime regime
Fig. 8.9. The definition of the safe operating regimes for three different types
of bearing.
z - - Hydrodynamic bearing
C C
- · - Dry rubbing bearing
- - - - Ball bearing
10
10 2
3
---~--
'\.
"'--
. ............ ,
·"'· -----
10
1
0·25 in } .
_ m dta shaft
0 0064
10
10
R.P.M.
8.7. Conclusions
Our discussion has shown how the scientific principles of tribology
may be used to help us in the choice of the best tribological solution
for a particular application. In practice, however, life tends to be
rather more complicated than these simple ideas might suggest.
For instance, having decided to use a ball bearing for a particular
situation we must now examine the wide range of different types of
ball bearing which are available. A glance at any ball bearing
manufacturer's catalogue will indicate the extent of this range and
further engineering expertise is required if the correct choice is to
be made. Such knowledge is beyond the scope of this book, but the
foregoing should have illustrated the value of tribological knowledge
in arriving at the optimum solution.
log U
Fig. 8.11. The matching of bearing characteristics to the operational loads
and speeds.
E
'-
@))
9
Ll-
(\J ~
:,
E I 1000 U>
'-
z I
Elastomer
U>
~
a.
~ c,,
:, C
U>
U>
.§
Q)
Q)
d:: Ll
10 6 E
:,
100 E
"><'
0
:z
10 5 -
10
0 10 20 30
Maximum angular deflection of oscil lotion degrees
154
INDEX
157
sen