(Tribology Series 42) A. Sethuramiah (Eds.) - Lubricated Wear - Science and Technology-Academic Press

TRIBOLOGY SERIES
Editor
D. Dowson (Gt. Britain)
Advisory Board
W.J. Bartz (Germany) N. Gane (Australia)
R. Bassani (Italy) W.A. Glaeser (U.S.A.)
B. Briscoe (Gt. Britain) H.E. Hintermann (Switzerland)
H. Czichos (Germany) K.C. Ludema (U.S.A.)
K. Fdedrich (Germany) W.O. Winer (U.S.A.)
Vol. 6 Friction and Wear of Polymers (Bartenev and Lavrentev)

VoI. 10 Microstructure and Wear of Materials (Zum Gahr)
Vol. 11 Fluid Film Lubrication - Osborne Reynolds Centenary (Dowson et al., Editors)
Vol. 12 Interface Dynamics (Dowson et al., Editors)
VoI. 13 Tribology of Miniature Systems (Rymuza)
VoI. 14 Tribological Design of Machine Elements (Dowson et al., Editors)
Vol. 15 Encyclopedia of Tribology (Kajdas et al.)
Vol. 16 Tribology of Plastic Materials (Yamaguchi)
Vol. 17 Mechanics of Coatings (Dowson et al., Editors)
Vol. 18 Vehicle Tribology (Dowson et al., Editors)
Vol. 19 Rheology and Elastohydrodynamic Lubrication (Jacobson)
Vol. 20 Materials for Tribology (Glaeser)
Vol. 21 Wear Particles: From the Cradle to the Grave (Dowson et al., Editors)
Vol. 22 Hydrostatic Lubrication (Bassani and Piccigallo)
Vol. 23 Lubricants and Special Fluids (Stepina and Vesely)
Vol. 24 Engineering Tribology (Stachowiak and Batchelor)
Vol. 25 Thin Films in Tribology (Dowson et al., Editors)
Vol. 26 Engine Tribology (Taylor, Editor)
Vol. 27 Dissipative Processes in Tribology (Dowson et al., Editors)
Vol. 28 Coatings Tribology- Properties, Techniques and Applications in Surface
Engineering (Holmberg and Matthews)
Vol. 29 Friction Surface Phenomena (Shpenkov)
Vol. 30 Lubricants and Lubrication (Dowson et al., Editors)
Vol. 31 The Third Body Concept: Interpretation of Tribological Phenomena
(Dowson et al., Editors)
Vol. 32 Elastohydrodynamics- '96: Fundamentals and Applications in Lubrication
and Traction (Dowson et al., Editors)
Vol. 33 Hydrodynamic Lubrication - Bearings and Thrust Bearings (Fr~ne et al.)
Vol. 34 Tribology for Energy Conservation (Dowson et al., Editors)
Vol. 35 Molybdenum Disulphide Lubrication (Lansdown)
Vol. 36 Lubrication at the Frontier- The Role of the Interface and Surface Layers
in the Thin Film and Boundary Regime (Dowson et al., Editors)
Vol. 37 Multilevel Methods in Lubrication (Venner and Lubrecht)
Vol. 38 Thinning Films and Tribological Interfaces (Dowson et al., Editors)
Vnl 0.(3 T r i h n l n n i t - , ~ l I~.,'.ea~rl",h. I ~ r n m I~Ant4al I~vn,,~rirn,--nf f n Inrll=c=fri,~i I::)rnhlm
Preface
Wear in lubricated contacts and the associated problems of running-in and scuffing
are of increasing importance in modem machinery. This is a consequence of the
modem compact designs that increase the operating severity. Though lubricated
contacts are more important industrially, emphasis so far has been on dry wear and
several books have been written on this topic. The information related to lubricated
wear is scattered and is mainly available in journals. The reported investigations in
turn seem to emphasise one aspect or the other of the wear process. The author felt
the need to treat lubricated wear in a consolidated manner. The present book is the
outcome of the effort in this direction.
The book is aimed at all those-interested in lubricated wear including the lubricant
formulators, user industries, and basic researchers in tribology. Besides
consolidation of the available information the major purpose of the book is to
strengthen the theory-practice interface. Such a focus identifies the gaps in
knowledge that need to be filled and will be of interest to tribologists. Development
of knowledge is a long-term goal and the industry will be interested in the more
immediate improvements to practice. This is addressed in the book and well-argued
methodologies are suggested that can ameliorate the present situation.
The developments in practice will depend on the effective dialogue between

engineers and scientists from various disciplines. This dialogue is only possible
when all aspects related to lubricated wear are well appreciated. The first chapter
gives a perspective of tribology to those who have major interest in the lubricant
chemistry with secondary interest in tribology. The second chapter provides an
overview of lubricant technology and will be useful to tribologists who need an
appreciation of the complexity of a formulated lubricant, and the limits within
which it has to function.
The subsequent chapters deal with wear and the associated problems of scuffing,
running-in and fatigue. The starting point is the third chapter devoted to dry wear.
The fourth and fifth chapters deal with mechanisms of boundary lubrication and
wear modelling of metallic materials. The sixth chapter considers the dry and
lubricated wear of polymers and ceramics, which is of growing importance. The
vi Preface
seventh chapter deals with laboratory evaluations. Fatigue and wear in mixed
lubrication is covered in the eighth chapter. The final chapter deals with the
important issue of relevance of laboratory tests to real situations.
The coverage in each chapter has to be necessarily limited and it was a difficult
task to decide on the extent of coverage. The main approach adopted is to clarify
the central concepts involved and then critically examine their utility. The
necessary background needed with regard to temperature rise, contact mechanics,
adsorption, and chemical reactions is pr6vided where appropriate. The coverage in
these areas is not extensive but adequate to serve the main objective of the book.
Detailed theoretical coverage of hydrodynamic and mixed lubrication is not within
the scope of this book. A reasonable attempt has however been made to explain the
central ideas and how boundary contact is affected in mixed lubrication. Adequate
references are provided so that the interested reader can pursue any topic in more
detail. References and nomenclature have been given chapter-wise.
Some of the new aspects considered in the book may interest the reader. These are
given below.
* The role of chemical reactions in wear and scuffing is examined from a

different perspective. This leads to a new proposal for antiwear mechanism
of chemical additives.
9 The low wear rates encountered in lubricated conditions can only be
judged by effective quantification of running-in and steady state wear. A
methodology based on experimental research is presented and its utility
discussed.
9 A new approach to use polymer composites for hydrogenerator thrust pads
is presented based on experimental research. This extends the potential use
of polymeric materials for heavy-duty applications under lubricated
conditions.
, Asperity level conformity can sometimes occur in the contact.
Experimental research conducted on this aspect brings out the importance
of such conformity in reducing metal contact in mixed lubrication.
9 Boundary lubrication has an important role to play in metal forming
operations and is difficult to simulate. A new experimental approach based
on oblique plastic impact is presented.
9 Measurement of low levels of wear in industrial components is of
importance. An experimental method based on bearing length curves is
presented to measure liner wear.
Preface vii
The wear map approach has been emphasised and its utility in improving
wear evaluation has been given a detailed consideration.
In the final analysis a book of this type has to judged on the basis of its impact on
practice. Any improvements to practice in the near future will be a source of
satisfaction to the author.
Acknowledgments
I wish to express my thanks to all those who have contributed to the successful
completion of the book. First and foremost I wish to thank Professor Duncan
Dowson who in the capacity of editor of the Tribology series, made valuable
comments with regard to all the chapters. His contribution has been of significant
importance in shaping the book. I am also thankful to Mr. Dean Eastbury who has
promptly responded to all my questions regarding the publication aspect.
Several examples of the work elaborated in the book are based on my research and
industrial experience spread over 30 years. Many colleagues and students have
contributed to this effort. During my stay at IIT, Delhi I had the good fortune of
interacting with Professor U. R. K. Rao, Dr. C. R. Jagga, Dr. Braham Prakash, and
Dr. Om Prakash who collaborated with me in academic and industrial research.
Their interaction helped me to sharpen my focus on the theory-practice interface
for which I am thankful to them. I am also thankful to Professor B. C. Nakra who
inspired me to join IIT, Delhi. The stay at IIT has helped me to develop ideas at a
deeper level.
My thanks are also due to Dr. Rajesh Kumar who has recently completed his PhD
and actively participated in all aspects of the book writing. His research
contributions on rtmning-in have an important place in the book. Many of my
former students have contributed to lubricated wear with their academic work.
Their contributions have an important place in the present volume. The interaction
with these students, and shaping of ideas with them has been the most satisfying
part of my career. I take this opportunity to thank Doctors Mange Ram Tyagi, V.
R. K. Sastry, Rathin Kumar Banerjee,. S. K. Karmakar, R. Prakash, and T. R.
Choudhary for their direct and indirect contribution to this volume.
Preface vii
The wear map approach has been emphasised and its utility in improving
wear evaluation has been given a detailed consideration.
In the final analysis a book of this type has to judged on the basis of its impact on
practice. Any improvements to practice in the near future will be a source of
satisfaction to the author.
Acknowledgments
I wish to express my thanks to all those who have contributed to the successful
completion of the book. First and foremost I wish to thank Professor Duncan
Dowson who in the capacity of editor of the Tribology series, made valuable
comments with regard to all the chapters. His contribution has been of significant
importance in shaping the book. I am also thankful to Mr. Dean Eastbury who has
promptly responded to all my questions regarding the publication aspect.
Several examples of the work elaborated in the book are based on my research and
industrial experience spread over 30 years. Many colleagues and students have
contributed to this effort. During my stay at IIT, Delhi I had the good fortune of
interacting with Professor U. R. K. Rao, Dr. C. R. Jagga, Dr. Braham Prakash, and
Dr. Om Prakash who collaborated with me in academic and industrial research.
Their interaction helped me to sharpen my focus on the theory-practice interface
for which I am thankful to them. I am also thankful to Professor B. C. Nakra who
inspired me to join IIT, Delhi. The stay at IIT has helped me to develop ideas at a
deeper level.
My thanks are also due to Dr. Rajesh Kumar who has recently completed his PhD
and actively participated in all aspects of the book writing. His research
contributions on rtmning-in have an important place in the book. Many of my
former students have contributed to lubricated wear with their academic work.
Their contributions have an important place in the present volume. The interaction
with these students, and shaping of ideas with them has been the most satisfying
part of my career. I take this opportunity to thank Doctors Mange Ram Tyagi, V.
R. K. Sastry, Rathin Kumar Banerjee,. S. K. Karmakar, R. Prakash, and T. R.
Choudhary for their direct and indirect contribution to this volume.
viii Acknowledgements
I am also thankful to several other tribologists who have interacted with me at
conferences and meetings. These interactions have been useful in exchanging
ideas. The interesting discussions I had with Professor K. C. Ludema on several
occasions have been of particular relevance to this book.
Finalisation of the book and printing are time consuming tasks. In this connection I
wish to express my sincere thanks to Mr. M. Satyanarayana, CFO, of Nipuna
Services Ltd for providing me with state-of-the-art facilities to complete the job
quickly.
Turning to the more distant past, my own doctoral research was conducted at the
Tokyo Institute of Technology during 1969-73 under the able guidance of late
Professor T. Sakurai. He has been instrumental in shaping me into a full-fledged
researcher. My first exposure to lubrication research was at the French Petroleum
Institute where I was a trainee during 1962-63. This training played an important
role in my later development.
I also like to express my thanks to my son, Ravindra who helped me with the
Autocad drawings. My daughter, Sushma helped me with issues related to
permissions and the latest technical information. Last but not the least I wish to
express my warm thanks to my wife, Jayashree who did her best to let me work
without hindrance. Without her continued support this book would not have been
possible.
1. Tribology in perspective
1.1 Introduction
This chapter is intended for those readers who have expertise in lubrication science
and technology with secondary interest in tribology. Though many specialised
books are available in the area, it is considered appropriate to introduce the main
ideas in a consolidated manner. This will equip the reader for the specialised
chapters on boundary lubrication and wear.
Tribology may be defined as the science and technology of surfaces in relative

motion. It encompasses the commonly known aspects of friction, wear and
lubrication. Such a definition emphasises the fact that friction, wear, and
lubrication are interconnected and need to be studied together. Tribology can also
be considered as an enabling technology as technological development depends on
the satisfactory resolution of tribological problems. In the present chapter, the laws
of friction are first stated and the conceptual explanation of these laws is given.
The limitations of the simple model are considered followed by model that takes
into account the growth in contact area. Surface roughness and contact models are
considered in the next section and their implications to tribology are discussed. The
issue of surface temperature rise is considered next. This is followed by a
consideration of friction of non-metallic materials. The final section deals with
lubrication. Boundary lubrication and wear are only mentioned where appropriate
as separate chapters are devoted to these subjects.
1.2 Laws of friction and their explanation
The origin of friction has interested several investigators since 14th century.
Historical development of the ideas involved is well documented by Dowson [1 ].
The major earlier investigators included Amontons, Coulomb, and Leonardo de
Vinci. The laws of friction are generally associated with the name of Amontons,
which state that the friction force in a sliding contact is proportional to the normal
load and is independent of the geometric area of contact. Friction force is the
tangential force necessary to overcome friction. Experimental studies by several
researchers have shown that these laws are approximately valid. Many attempts
CHAPTER 1. TRIBOLOGY IN PERSPECTIVE
have been made in the past to explain these laws without much success. The first
satisfactory explanation was due to Bowden and Tabor [2]. Their model provides a
conceptual explanation of the laws of friction. They considered that when two
rough surfaces are brought into contact under normal load the real contact is made
at some asperity peaks only. The pressure at these contacts is very high and plastic
flow at these asperities is assumed. The flow pressure is taken as the hardness of
the metal. They further assumed that the shear strength of the real junctions is
equal to the shear strength of the material. The contact model for two identical
metals is given in Fig. 1.1. On this basis the laws of friction can be explained as
follows:
L
A r ----- (1.1)
H
F=SmA ~ (1.2)
f = __.-~W
Sm (1.3)
L H
where
Ar = real contact area, m 2
L = load, N
H = hardness, N/m 2
F = friction force, N
sm = shear strength of metal, N/m 2
f = friction coefficient
From Eq. (1.3) it can be seen that f is a constant since s m and H are material
properties. It follows that friction force is proportional to load. As real area is
governed only by hardness and load, it is independent of geometric area.
The above explanation is based on contact of metallic materials. When the two
contacting surfaces are not identical s m and H are taken to be those of the softer
material. Since Sm is nearly equal to 1/6 of the hardness for a metallic material it
follows that f is about 0.16 and is a constant irrespective of the material
combination. Both these observations are not validated in practice. Surfaces in
normal atmosphere can have f values ranging from 0.3 to 1.2 and depend on the
material combination. One aspect of practical interest is that friction coefficient for
a given material pair is reasonably constant over a range of operating conditions.
1.2. LAWS OF FRICTION AND THEIR EXPLANATION
~ t
Fig. 1.1. Real area spots within circular geometric area of diameter d.
The value of f however depends on the material pair, surface condition, and
environment.
The problem was recognised by Tabor [3] and he provided a plausible explanation
for the phenomenon. The earlier model proposed considered yielding is only due to
normal force. In reality, yielding occurs under combined normal and tangential
stresses. Assuming plane strain conditions, the relation proposed for plastic
deformation is
2 2
p 2 + as i = as m (1.4)
where p is the normal pressure, a is a constant, s i is the shear strength of the

interface, and s m is the shear strength of the metal.
From this equation, two important observations can be made. For a given load L,
the normal pressure p decreases and tends to zero as s i approaches s m . This
happens because the real area increases and hence the normal pressure decreases. If
s---L= ~: and a = 9 the following equation can be derived for friction coefficientfas
Sm
given below:
I K2 (1.5)
f = 9(1- K2)
The above equation shows that for very clean surfaces with K approaching 1.0, f
can reach very high values. However, even for a minor contamination with say, ~c
= 0.95 f reaches a modest value of 1.0. Experimental verification of these ideas is
quite extensive. With very clean surfaces generated under high vacuum in the
range of 10.6 to 10.8 torrfvalues greater than 10 can be obtained for several metals.
It is usual to refer to the asperity contacts as adhesive junctions and the theory
involved as adhesion theory of friction.
The above theory cannot be quantified as ~: cannot be defined for a tribological

contact. While high friction can be reconciled with junction growth, plastic
deformation of the junctions additionally contributes to friction. Hence, the values
differ from metal to metal depending on issues like ductility, work hardening, and
atomic arrangement on the surface. The atomic arrangement is of importance since
shear strength depends on the crystallographic plane on which sliding occurs and
can be well demonstrated for single crystals [4]. Thus prediction o f f from the
above equation is not possible even for a given ~. Yet another problem with the
theory is that metals with hexagonal close packed (hcp) structure do not show
strong junction growth and slide easily even with very clean surfaces as shown for
cobalt alloys [5]. Despite these limitations, the qualitative aspect of the theory is of
importance. In space applications tribological contacts operate under very high
vacuum. Under such condition surfaces will no longer have protective oxides and
are prone to high friction and seizure. These contacts can be operated only with
thin lead or solid lubricant films interposed between the surfaces that eliminate
direct asperity contact.
The normal engineering surfaces are generally covered with oxides that reduce the
direct metallic contact. This is an advantage from a tribological point of view. The
protection from oxide films depends on several factors like composition, adherence
to surface, and the balance between removal and formation rates. The asperity
contacts in real systems can range from elastic to plastic contacts as will be seen in
the later section. Even when the contacts are predominantly elastic and covered
with protective oxides the f values can range between 0.3 and 1.2 for most of the
metals.
1.2. LAWS OF FRICTION AND THEIR EXPLANATION
1.2.1 Other aspects of friction
Besides adhesion and plastic deformation, it is recognised that ploughing which

results in plastic grooving can also contribute to friction. Ploughing can occur
when hard asperities of one surface groove the softer surface or when abrasive
particles come between the surfaces. The ploughing action due to asperities
depends on their shape and hardness. Models are available to estimate the
deformation losses when a cone or a sphere sinks into a surface and displaces the
material ahead by plastic deformation while sliding [6,7]. Models for ploughing are
based on a sinking depth needed to support the load and the tangential force needed
to displace the material ahead. The load support area is based on half the vertical
projected area while the tangential force is based on the horizontal projected area
of the groove. For engineering surfaces, the ploughing component of friction is
generally small. This is because the hemispherical asperities with radii in the range
of 100-1000 ktm have negligible ploughing action. Ploughing is important in
operations like grinding and cutting. Hard wear particles or abrasive dust particles
can also contribute to ploughing.
The explanation of friction so far has not distinguished between static and dynamic
friction. Static friction refers to the value at the onset of sliding while kinetic
friction refers to the value during gross sliding. Kinetic friction is usually smaller
than static friction. When kinetic friction is lower than the momentary static
friction oscillations are possible which is referred to as stick-slip. Stick-slip
depends on the elastic response of the system, and the variation of kinetic friction
with velocity. Stick-slip problems can arise in components like brakes leading to
vibration and noise. Careful control of materials in such cases is important to
overcome the problem.
In the case of pure rolling between hard metals the energy losses involved are
mainly due to elastic hysteresis. The energy losses for such cases are very small
and when expressed in terms of a friction coefficient are < 0.01. In the case of
elastic materials like rubber tires, the energy losses are substantial and fvalues for
automobile tires range from 0.2 to 0.8.
1.3 Contact of surfaces
On a microscopic scale, the surfaces are not atomically smooth and have
undulations. These undulations are referred to as roughness. The roughness can be
traced by a fine diamond stylus that follows the surface profile. A typical
roughness trace is illustrated in Fig. 1.2. In this profile, the high points, called
CHAPTER 1. TRJBOLOGY IN PERSPECTIVE
asperities, appear sharp because the vertical magnification is much higher than the
horizontal magnification. When the scaling in both directions is the same,
asperities will have gentle slopes and this should be kept in mind. Surfaces can
have waviness due to manufacturing processes and this waviness is filtered by cut-
off filter. This ensures that the parameters calculated from the profile are
representative of the roughness. Several roughness parameters are then obtained by
processing the digitised data with microprocessor. Centre line average (R~) and
rms value ( R q ) are based on a mean line that divides the roughness profiles into
equal areas. The measured heights at different points of the profile, below and
above the mean line, are processed to obtain these values in the given sampling
length. In common engineering practice R a is usually specified for the surfaces
while the rms value is common for statistical analysis of the surfaces. R q is
usually referred to as o in contact analysis. The R a values depend on the nature of
application and typically range from 0.1 to 2.01am. The other parameters that can
be routinely obtained by the instrument include asperities per unit length, average
asperity slope and curvature. The values as obtained are two dimensional along a
chosen line.
o.amm~
Fig. 1.2. Roughness profile of cast iron surface with R~= 0.88 lam.
The three dimensional parameters are necessary to characterise the contact. For
isotropic surfaces, designating the high spots as summits and, those in the profile
trace as peaks the following assumptions may be made for a surface with Gaussian
height distribution [8]:
The average curvature of summits is nearly same as that obtained for

peaks.
1.3. CONTACT OF SURtVACES
2. The density of summits (asperities) per unit area rl is approximately

equal to 1.811pz where rIp is the number of peaks per unit length.
3. The distribution of summits is Gaussian with rms value same as that for
general height distribution.
Two rough surfaces, which are nominally fiat, may now be brought into contact. It
is easier to visualise the problem in terms of a rough surface contacting a rigid fiat
surface. As asperity contacts depend on the nature of height distribution, it is
necessary to specify the distribution function. The present treatment is based on a
Gaussian distribution that reasonably approximates many engineering surfaces.
The concept is not restricted and can be applied to other kinds of distribution as
well provided the distribution functions are known. Fig. 1.3a shows the nature of
general height distribution and the cumulative height distribution. Fig.l.3b shows
the contact situation. The shaded area indicates the contact zone. When a rough
surface is brought into contact with a smooth rigid fiat under load highest asperities
first come in contact. An asperity with height z i will be compressed by (z i - d)
where d is the distance from-mean line and the rigid flat. As the load increases
asperities with lower heights come into contact to support the load. The problem is
to find the real contact area as a function of load assuming elastic contact.
Following Greenwood and Williamson [9] the following integrals can be obtained:
oo
n = rlA n I ~0(z)dz (1.6)

d
oo
A, = rt[3rlA. I (z - d)q0(zldz (1.7)

d
L = 47 E ~0.5 rlA. (z - d) 3/2q~(z)dz (1.8)

d
where
n = number of contacts
A~ = total real area of contacts, rn 2
L = total load, N
1] = asperity density = number of asperities per m 2
An = nominal contact area, m2
= mean asperity radius, m
d = separation between rigid plane and mean surface height, m
z = asperity height over mean surface height line, m
,p(z) = Gaussian height distribution function =

1 (z2)
o , f ~ exp - 2--~-
-1
_/1-vl 2
+ , N/m 2 where v~,2 and El,2
1 -- V 2 2 refer to
E1 E2
Poisson ratio and elasticity modulus of the two surfaces.
Profile 0 0.5 1
Height Cumulotive
d l s t rl but ion d i stri bution
(o)
Rigid flat f A 9
(b)
Fig. 1.3. (a) Roughness trace and Gaussian and cumulative height distribution; (b) Loaded
contact between rigid flat and rough surface with separation d.
For the case of rigid flat contacting a rough surface E will be governed by Poisson
ratio and elastic modulus of the rough surface only. The Eq. (1.6) expresses the
number of asperities in contact on the basis that all those asperities exceeding d
will be in contact. The expressions for real area and load are obtained from basic
contact mechanics relationships for spherical contact. They relate the deformation
depth to load and contact area for a known 13. For a given asperity 'i' the Eqs.
(1.9a) and (1.9b) give these relationships as follows:
A i = 7r6il3 (1.9a)
Li =T E 4 I~I/2~
i3/2 (1.9b)
1.3. CONTACT OF SURFACES
where A i is the area, L i is the load on asperity with height z i , and ~ii is the
deformation equal to (z i - d).
t0(z) is known for a Gaussian distribution and the above equations can be
numerically integrated. The major conclusion from this modelling is that the real
area is nearly proportional to the load even if the contacts are elastic. Also real area
is nearly independent of geometric area.
The actual contact is between two rough surfaces. The usual approach is to treat the
contact as an equivalent surface against a rigid fiat. The effective modulus E is as
defined under Eq. (1.8). Equivalent radius 13", and equivalent roughness a* are
defined in Eqs. (1.10a) and (1.10b) as
[~*: 13'131 (1.10a)

13, + 13:
!
9 / 2 2
a =~/(Yl + a 2 (1 10b)
where 131 and 132 are the asperity radii of the two surfaces and a~ and a a are the
rms roughness of the two surfaces. In the case of above contact model with rigid
smooth surface [3" and a* will be governed by [3 and a of the rough surface only.
1.3.1 Implications of the contact model
The laws of friction derived for plastic contact postulate that real area is
proportional to load and is independent of the geometric area. This is also valid for
elastic contacts. Hence, it is considered that laws of friction are also applicable to
elastic contacts. It is however difficult to define the adhesive forces for elastic
contacts. Their values are expected to be lower than in the case of plastically
deforming asperities. On the other hand, for a given load, the real area is higher
compared to the plastic case and the overall fvalue may not be very different. The
argument provided here is only directional. The growth in contact area is difficult
to envisage for elastic contacts and the real area may be governed only by the
normal load. The issue of adhesion is important with regard to wear also and is
elaborated in chapter 3.
10 CHAPIER 1. I'RIBOLOGY IN PERSPECTIVE
The deformation of an asperity contact will become plastic beyond a critical

deformation depth. On this basis, plasticity index has been proposed in [9] to
characterise the deformation as
E
(1.11)
where H is the hardness of the softer material. It has been proposed that for ~g <
0.6 the overall contact is essentially elastic while for values of ~g > 1.0 the contact
is predominantly plastic.
As an example let two identical mild steel surfaces with E~.2=203 GPa, H=l.8 GPa,
G 1,2=0.2 ~m, and 131,2=500 ~m be brought into contact. Based on earlier equations
for equivalent values V is calculated as 2.08. This value indicates the surfaces
predominantly yield plastically. Similar surfaces made of hard steel with
H=7.0GPa will have a V value of 0.57 and the contact will be essentially elastic.
Thus, plasticity index can be used for an approximate assessment of the contact
situation.
The above model is for static contact. During sliding, the load progressively shifts
from one set of asperities to another. It is considered, statistically speaking, that the
contact situation repeats itself. This may not be true particularly for cases with
significant plastic deformation. Strong junctions with contact growth may upset the
load sharing and some of them may contribute to friction due to residual adhesion
even when the load is released in the contact. It may also be observed that the
model depends on average parameters. Some individual contacts may have values
far different from average and these may be critical in the initial adjustment of
surfaces normally referred to as running-in. It is also known that the roughness
parameters as obtained are a function of sampling interval, the cut-off selected, and
the stylus dimensions. Current research based on detailed formulation of random
surfaces and fractals [10,11] may eventually provide better contact models.
Another issue is the scale of observation. For example, if the surface roughness is
obtained by atomic force microscopy many smooth areas observed with stylus
meter will show undulations. The fundamental question then is what is roughness.
From a practical point of view, one has to be content with the explanation of a
phenomenon based on observation at a relevant scale. Awareness of this problem is
essential since explanations of mechanisms in tribology are based on observations
at different scales.
1.3. CONTACT OF SURFACES 11
The contact model has important implications to wear. The removal process is
governed by the nature of contact and is far more sensitive to the surface condition
than friction. Thefvalue for metals ranges from 0.3 to 1.2 in normal atmosphere as
stated earlier. On the other hand, the wear rates can vary by several orders of
magnitude. Boundary lubrication can have a dramatic influence on wear. A drop of
mineral oil introduced between two rubbing surfaces can bring down friction by a
factor of 10 while the wear may reduce by a factor of 1000 or more.
1.4 Surface temperatures in sliding contact
The frictional heating in tribological contacts has a significant influence on the

performance. The heat generated due to friction is conducted away into the solids
and there is a rise in temperature at the interface over and above the bulk
temperature of the solids. This temperature can affect the surface oxidation,
lubricant failure, chemical reaction with additives, and a host of related problems.
The temperature rise is calculated theoretically as it is very difficult to measure
these temperatures.
1.4.1 Theoretical calculation
Jaegar [12] presented an analysis of moving and stationary heat sources that forms
the basis for most of the work in this area. When steady state is assumed, which is
normally the case, simple final equations are available to analyse the temperature
problem. In the present analysis, these equations are used and they provide
adequate background to appreciate the problem. When steady state conditions do
not occur the original equations of Jaegar are to be applied.
The temperature rise can be analysed based on a square projection of one body
sliding on the other as shown in Fig. 1.4. This projection is making contact over the
whole geometric contact area of 4l 2 . Both the bodies are considered semi-infinite
for this analysis. Body II is moving over body I with a velocity v. Body I, which is
stationary receives heat from a moving heat source while body I that is moving,
receives heat from a stationary heat source. It is now required to estimate the
surface temperature rise in the contact. The following nomenclature is adopted:
Cp = specific heat (J/Kg~

k = thermal conductivity (W/m~
l = half contact dimension of heat source (m)
q = heat flux (W/m2)
12 CHAPTER 1. TRIBOLOGY IN PERSPECTIVE
v = sliding velocity (m/s)

AO = average temperature rise (~
Z = thermal diffusivity ( m 2/S )
vl
Pe = ~ , Peclet number, non-dimensional
2X
L = load (Kg)
Two equations form the basis for calculating average temperature rise in contact
and are based on simplification of the original theory. For a stationary heat source
average temperature rise is given by
! l l l l l l l l l l / l / / / . ,[/l l l l l l l l h
\'~\\\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
1
1
Fig. 1.4. Square projection of body II moving over stationary surface I with velocity v.
This equation is also applicable for slow speed contact with Pe < 0.5. For a moving
heat source with P~ > 5.0 the mean temperature rise is
A0 =0.75 Pe
The case for Peclet number ranging from 0.5 to 5.0 shall be treated later.
The distribution of temperature in the square contact is not uniform and varies from
-I to +l. The position of the maximum temperature varies with P~. The simplified
equations refer to an integrated average value over the contact. For stationary
contact, the maximum temperature is _= 1.15 times the average value while for
moving source with P~ > 5.0 the maximum value is nearly 1.5 times the average
value. The basic equations above refer to heat flow into one body or the other. In
sliding contact, the heat generated is distributed into both the bodies II and I. The
common interface is considered to have the same temperature. This requirement
1.4. SURFACE TEMPERA TURES IN SLIDING CONTACT 13
means that the heat flux q generated shall be so partitioned that A0 is the same for
both the bodies. The heat flowing into the two bodies is ql and q2 respectively.
Define a partition coefficient qo which is the fraction of total heat q entering the
body I. Then (1-q~) is the fraction of heat going into body II. On this basis, surface
temperature rise for slow and fast moving contacts can be estimated as given
below.
For slow moving contacts with Pc < 0.5, stationary contact theory is applicable to
both the bodies. Since A0 is the same for both the bodies it follows
0946i )= 0946/q,lk,,) (1.14)
where k I and k H refer to the thermal conductivities of the two bodies I and II.
Taking k ~ = k l it can be shown

k ii
A0= 0.946( q l / 0 + k*)-' (1.15)

t,k,)
Similarly average temperature rise for cases with Pe > 5.0 is
0.75
(ql / Pc
AO= ~k:) (1.16)
* -1/2
1 + 0.793k P~
Another approach frequently used to estimate temperature rise is to assume the

total heat flux q enters one body or the other and calculate temperature rise for
body I and body II as A0~ and A0 Ii from the corresponding equations for moving
and stationary sources. The average surface temperature rise A0 is then given by
1 1 1
= ~ + ~ (1.17)
A0 A0 x A0.
Both the approaches above are equivalent.

Theoretically for semi-infinite bodies with 0~ bulk temperature the surface

temperature is equal to the temperature rise in contact. With finite sized contacts
the surface temperature, also called contact temperature is
Contact temperature ( Tc ) = Bulk temperature ( Tb ) + A0 (1.18)
The above approach is applicable to lubricated contacts where bulk temperatures of

both surfaces tend to be equal. Unequal bulk temperatures are a complex problem
as the heat flow into the two bodies is modified which in turn affects the
temperature rise [13]. The other complications that arise include variations in
thermal properties due to temperature as well as influence of oxide and other films
on heat transfer [14].
The above equations are for square contacts that are also used for circular contacts
replacing I by the radius of contact. Also, the square geometric area is assumed to
make full contact over the geometric area. In a tribological contact, the real area is
much smaller as discussed earlier. The temperature at asperities needs to be
estimated besides the overall temperature rise in contact. This problem is addressed
in chapter 3 considering that the asperity temperature rise is superimposed on the
temperature rise over the geometric area. The asperity temperatures can be
significantly higher than the temperatures based on the geometric area. The
duration of asperity contacts on the other hand is short and it is not easy to decide
which temperature is governing the tribological mechanism involved.
The case for Peclet numbers in the range of 0.5 to 5.0 is treated by a graphical
procedure [ 12]. The temperature rise for the stationary source as applicable to body
II is used as earlier to obtain A0 n assuming all heat is flowing to body II. For body
I receiving heat from a moving source A0~ is given by
AO~ = 0.946 e (1.19a)
This equation is the same as for stationary source multiplied by a coefficient E. E

in turn is obtained from
e = 0.398yPe-' (1.19b)
where y is the ordinate value from Fig. 1.5 corresponding to the calculated Pe.
1.4. SURFACE TEMPERATURES IN SLIDING CONTACT 15
Another appro0ch suggested by Greenwood [ 15] involves an interpolation method

according to which the temperature rise is estimated by the following equation:
1 1 1
2
A02 A01 A01I2
It may be noted that A0~ in this case is directly obtained from Eq. (1.13) at the
corresponding P~. As this calculation does not involve the graphical procedure, it
is more convenient.
The above procedure can be illustrated by the following example. Consider a

stationary square pin of area 1.8 l xl 0 -s m z . This pin is in sliding contact with a
disk rotating with a velocity of 5.11 m/s. The load applied is 100 N. Assume f is
0.4. The pin and the disc are made of the same steel with the following thermal
properties:
Thermal conductivity (k) -- 43 W/m~

Thermal diffusivity (X) = 1.172xl 0 -5 m 2/S
The problem is to find the surface temperature rise based on geometric contact
area.
0 1 2 3 4 S
Pe
Fig. 1.5. 'y' as a function of Peclet number.

16 ('HAPTER 1. TRIBOLOGY IN PERSPECTIVE
For the square contact l = [,]1"81x10-5 = 2.127x10-3m

V 4
k and Z are same for both the bodies. The heat flux is taken as the frictional work
done per unit time and unit area. Also since Nm/s - Watt (W)
q = ~ _ f L v_ 0.4x100x5.11 =1.12x107 W/m 2

area 1.8 lx10-5
vl
PC = ~ = 463.7
2Z
As Pe > 5.0, the high speed equation (1.13) is applicable. Assuming all heat is
flowing to the surface I
A01 =0.75 Pe =18.71
For body II receiving heat from stationary source Eq.(1.12) is applicable. So
In
A0 ii = 0.946 1-- = 525 ~
k
Now temperature rise A0 is obtained from
1 1 1
A0 18.71 525
So, A0 = 18.07 ~
It is of interest to note that when Peclet number is high the interface temperature
rise is governed by the moving source theory.
Thermal properties affect the temperature significantly. If for example the steel pin
is replaced by a glass pin with k = 1.25 W/m~ and Z = 0 . 0 8 x l 0 5 m2/s the
temperature rise will be as high as 130~
1.5. FRICTION OF NON-METALS 17
1.5 Friction of non-metals
This short section deals with the issues of friction related to ceramics and
polymers. Ceramics are considered first followed by polymers. This coverage also
includes an overview of the different categories of the materials used.
1.5.1 Ceramics
Engineering ceramics are being increasingly used for metal cutting tools, metal
forming dies, rolling element bearings, and other applications. Ceramic coatings
are also used for wear resistance. Their main advantages include low density, high
strength and corrosion resistance. They also retain their strength to very high
temperatures unlike metals and this makes them particularly suited for high
temperature applications. The major disadvantage of ceramics is their low fracture
toughness. This can be improved to some extent by controlling the grain size, and
incorporating suitable sintering aids. Fabrication of ceramics involves compacting
and sintering processes and careful control is necessary to obtain uniform
structures. Machining of ceramics also presents problems due to their high
strength.
The oxide ceramics normally used include alumina (A12 03) and zirconia (ZrO2)
while the non-oxide ceramics include silicon carbide (SIC), silicon nitride (Si3N4),
tungsten carbide (WC), titanium carbide (TIC), and titanium nitride (TIN). Sialons
are a special class of ceramics in which nitrogen in silicon nitride is partially
replaced by A1 and oxygen. The basic contact mechanics developed for metals can
be used to study ceramic contact as well. Two different kind of pairing can be
envisaged. The first is ceramic Vs ceramic. When the contacts are elastic ceramics
have f values in the range of 0.1 to 0.3 in normal atmosphere. When contact
conditions are severe, leading to fracture of ceramics at micro or macro level the
hard debris increases the f values to a higher range of 0.5 to 0.8. At high
temperatures, the friction of ceramics is higher in comparison to room temperature
as the adsorbed species are desorbed. The second pairing is ceramic Vs metal. In
this case, metal can be transferred to the ceramic and the friction changes to the
metal-metal pair. The metals can also undergo oxidation by interacting with
oxygen. A detailed consideration of ceramic tribology is available in an edited
book [ 16].
Friction of some ceramics is sensitive to the environment. Silicon nitride is an

interesting example. This ceramic is known to react strongly with water forming
18 ('HAPTER 1. I'RIBOLOGY IN PERSPECTII~T
oxides and hydroxides that reduce friction significantly. A recent example where
friction of the nitride is studied directly in water may be cited here [17]. The
friction reduces due to surface interaction and then reaches negligible levels as the
highly polished surface promotes hydrodynamic effects. Friction of alumina is also
sensitive to water vapour and friction reduction occurs because of hydroxide
formation [ 18]
For many ceramic combinations, wear tends to be high in sliding. The major cause
is the fracture at micro or macro levels governed by the tensile stresses at the
surface. Under moderate conditions wear can be controlled by lubricants with
approaches available for metals. For high temperature applications novel ideas are
necessary as liquid lubricants cannot function.
1.5.2 Polymer friction
Polymers are being used for bearing applications in automobile, aircraft, and in
food and textile industries. Special applications for polymers include space and
artificial human joints. Many polymers can be easily shaped and this is an
advantage. Polymers have the disadvantage of low strength and limited
temperature capability. These limitations restrict their use to milder operations in
comparison to metals.
Thermoplastics form the major class for tribological applications. The commonly
used materials include high density polyethylene (HDPE), polyamide (Nylon 6-6),
polyoxymethylene (Acetal), and polytetrafluoroethylene (PTFE). Frictional
behaviour of these polymers has been studied extensively. The studies have been
conducted with metallic counter faces as well as glass. Two important
characteristics govern the polymer friction. One is the polymer transfer to the
counter face while the other is the viscoelastic behaviour during sliding. Hence,
unlike metals friction can vary significantly with operating conditions. With
polymers that have highly linear configuration the friction is lower. This is the case
with all the above materials except nylon, which has bulky side groups.
PTFE is one of the most widely used low friction materials. The friction coefficient
for PTFE varies between 0.06 and 0.2 depending on the operating conditions.
Higher stress in contact tends to reduce the friction coefficient of this material.
PTFE friction is governed by transfer of a thin layer on the counter face that gets
oriented in the direction of sliding. Once oriented the friction becomes low with
PTFE sliding against itself. The transferred layers are weakly attached to the
counter face and are easily worn out. Thus, wear rates of PTFE are high and are not
1.5. FRICTION OF NON-METALS 19
acceptable. Wear rates can be effectively reduced by fillers like brass, copper oxide
and graphite. It is common to use two or more fillers together for effective
performance. Such polymer composites can reduce wear by 2 to 3 orders of
magnitude. Some of the fillers also strengthen the polymer. The friction of the
composite materials is only marginally higher than pure PTFE. PTFE based
materials can be used up to 250~ as its melting point is 327~
Nylon friction is higher than PTFE and f varies in the range of 0.4 to 0.6. The
transfer to the counter face is lumpy unlike the thin oriented layers of PTFE. The
transferred material is more strongly attached to the counter face and wear of this
polymer is lower and acceptable in many applications. Nylon can also be
impregnated with fillers to control its wear further. The operating temperature limit
for nylons is around 200~
Normal low density polyethylene has lumpy transfer and has high friction. It also
has high wear in comparison to nylon. HDPE on the other hand has lower friction
with f ranging from 0.2 to 0.3. Layered transfer occurs in this case like PTFE
though the layers are relatively thicker than for PTFE. The wear of HDPE is
significantly lower than PTFE. The low melting temperature of 120~ for
polyethylenes restricts their application significantly. The most important polymer
used for biological applications in this class is the ultra high molecular weight
polyethylene (UHMWP) with typical molecular weight of 4x10 6, which is about
ten times the normal molecular weight of HDPE. UHMWPE is biocompatible with
good strength and low wear rate. It also has low f in the range of 0.1 to 0.2.
UHMWPE also has better abrasion resistance than HDPE.
Acetal also exhibits low f values in the range of 0.15 to 0.3, with wear rates
comparable to those of nylon, and can be used to temperatures up to 140~
Several other thermoplastics like polymethylmethacrylate (PMMA), and
polyphenylene sulphide (PPS) are also used for tribological applications and the
above discussion was limited to the more commonly used materials. Two
specialised books available cover the tribology of polymers in detail [19,20].
Development of composites with different short and long fibres as well as granular
materials is receiving increased attention. Besides polymers, metal matrix
composites are also gaining in importance. An edited book on the tribolgy of
composites is available dealing with both polymer and metal matrix composites
[21].
Thermosetting polymers are also used for bearing applications with or without
fillers. Thermosetting bearings are more difficult to fabricate. These are cross-
linked polymers, which decompose at high temperatures unlike thermoplastics that

melt at a specific temperature. The major classes used include phenolics, epoxy
resins, and polyimides. These bearing materials with various fillers are used for
several applications like rolling mill bearings where they are successful in
replacing metallic bearings. Phenolics are also used in brake materials as binder
along with several fillers. The operating limits of polymeric materials are getting
higher both in terms of strength as well as temperature by developing fibre based
composites and use of high temperature polymers like polyimides and
polyetherether ketone (PEEK). It is likely that in future many applications
involving metallic bearings may be replaced by polymer composites, which can run
dry or lubricated. The bearings are of different types starting from simple bushes to
porous materials with impregnated polymers. While large body of available
literature can give rough criteria for selection, it is imperative that effective
simulation studies be carried out for selecting the materials. For example,
development of polymer pads for hydrogenerator thrust pads conducted by
Choudhary et al [22] involved simulation in a thrust beating rig to ensure
performance. Control of friction in brakes, tires, clutches and other components is
of practical importance and in all such cases simulation is the only way to ensure
performance. The technological aspects of friction are considered in detail by Blau
[23]. The major role of the available knowledge base in tribology is the insight it
provides in understanding system specific behaviour.
Polymer operating limits are normally specified in terms of pv factor where p is the
pressure and v is the velocity. Limits are specified for continuous operation. This
value is related to the temperature limit for the material as pv is roughly related to
temperature rise in the contact. As discussed in the section on surface temperature,
the temperature rise is related to PC and for high values of Peclet number, the
temperature rise should be related to v 1/2 . A better approach to the thermal
problem has been presented by Ettles et al [24].
1.6 Lubrication
In a broader sense, lubrication may be defined as the complete or partial separation

of surfaces by interposed films. The present section is confined to liquid lubricants
only. Complete separation can be affected through fluid films whose thickness is
higher than the asperity dimensions. Such separation can be affected by
hydrodynamic or elastohydrodynamic lubrication (EHL). Pressure generation and
load support in these cases is due to hydrodynamic action where viscosity plays a
major role. At the other extreme load may be supported by molecular films with no
1.6. L UBRICA TION 21
hydrodynamic load support. This regime of lubrication is referred to as boundary

lubrication. In boundary lubrication, lubricant-solid interactions play the major
role. The lubrication can also involve both hydrodynamic and boundary effects.
Such a regime is referred to as mixed lubrication regime. This section first deals
with viscosity followed by hydrodynamic lubrication and EHL. The final part of
the section is limited to few observations on mixed and boundary lubrication.
Many specialised books are available dealing with hydrodynamic lubrication and
EHL. Treatment of this vast area in a section can only provide an appreciation of
the principles involved. EHL is also referred to as EHD lubrication.
1.6.1 Viscosity
Lubricants offer resistance to shear that can be characterised by viscosity. Viscous
flow is analogous to movement in a pack of cards. The lowest card is stationary
while the top card at height z moves with a velocity u. Viscosity 11 may be
expressed as
q=- (1.21)
s
where
= shear stress, N/m 2
= shear strain rate du/dz, s-
In this equation rl is constant provided shear stress is proportional to shear strain.

Such behaviour is termed Newtonian. Most of the lubricating oils behave as
Newtonian fluids at modest pressures and shear rates commonly encountered in
hydrodynamic lubrication.
Viscosity is expressed in absolute units (N.s/m 2) in the above equation. Many

viscometers measure kinematic viscosity which is defined as viscosity /force
density and has units of m2/s. When expressed as cm2/s the kinematic viscosity
is called stoke. Centistoke (cSt) is 0.01stoke and is commonly used in
characterising viscosity in industry and has units of mm2/s.
Viscosity varies significantly with temperature. The temperature dependence can

be approximated by the following relation
q = Ae b'rr (1.22)
22 CHAPTER 1. TRIBOLOGY IN PERSPECTII/E
where, A and b ~ are constants for the given fluid
The important equation used in terms of cSt (Walther equation) is expressed as
loglog(v + c ' ) : a ' - b' (logT) (1.23)
where v is kinematic viscosity in cSt, and a', b' and c' are constants for a given
fluid. When fluid viscosity is lower than 2.0 cSt the value of c' varies. Normally
one deals with fluid viscosities greater than 2.0 cSt.
This equation is considered accurate and widely used industrially. This relation
forms the basis for ASTM D341. ASTM refers to American Society for Testing
and Materials. In this method, a graph is scaled on the X and Y axes as per the
above equation. If viscosity is known at any two temperatures, a line can be drawn
passing through these points and viscosity at any other temperature can be obtained
by interpolating or extrapolating from this line. The slope of the line is indicative
of the variation of viscosity with temperature.
1.6.1.1 Viscosity index
Viscosity index (VI) is a comparative number that characterises the rate of change
of viscosity with temperature. The kinematic viscosity in cSt of the test lubricant is
compared with two reference oils, which are arbitrarily designated as 0, and 100 VI
fluids. For oils with VI up to and including 100, VI is defined as
w
VI = _--=----=-_L
- U t x 100 (1.24a)
L-H
where
Ut = viscosity of the test fluid at 40~
L = viscosity at 40~ of the reference fluid with 0 VI and having the
same viscosity as test fluid at 100~
H = viscosity at 40~ of the reference fluid with 100 VI and having
the same viscosity as test fluid at 100~
Higher the VI lower the change in viscosity with temperature. 100 VI oils are
Pennsylvania oils while 0 VI oils are Texas naphthenic oils. To calculate the VI the
viscosity of the test fluid is first measured at 40~ and 100~ Then the required 0
and 100 VI fluids that have the same viscosity as the test fluid are selected from
available tables. The same tables provide also the corresponding viscosities at 40~
1.6. LUBRICATION 23
m m
for the reference fluids that are the L and the H values needed. Detailed
methodology is available in ASTM D2270.
Viscosity index is widely used to characterise lubricants as will be seen in chapter

2. Petroleum lubricants manufactured until eighties had VI values in the range of
90-100. Modem lubricants can have VI values in the range of 100-150. The
equation used to find VI for oils with VI greater than 100 is
V I : [{(antilogN)-1}/ 0.00715] + 100 (1.24b)

where
N = 0ogH - logUt)/logYt
In the above U, and Yt are the kinematic viscosities of the test fluid at 40~ and
100~ respectively. H is the same as defined in Eq. (1.24a).
1.6.1.2 Pressure effect
Viscosity also changes with pressure. Barus equation is a good approximation for
the pressure influence and is expressed as
q=rl0 e~p (1.25)

where
110 = absolute viscosity at the given temperature and p = 0
= pressure coefficient of viscosity, m 2/N at the given temperature
11 = viscosity at pressure p
More accurate expressions for temperature and pressure effects are based on
Roelands equations [25]. From a practical point of view, the pressure effect on
viscosity is significant only at high pressures as in the case of line and point
contacts. This is because of the value of ~ which typically ranges from 1.0-
2.0x10 -8 m2/N. Also, at high pressures, the fluid behaves in a non-Newtonian
manner. Its behaviour may be approximated by a glassy solid with a limiting shear
stress.
1.6.2 Hydrodynamic lubrication
The flow of a lubricant through a convergent wedge generates pressure that can
support a load. The mechanism involved is hydrodynamic in nature and this type of
24 CHAPTER 1. TRIBOLOGY IN PEtL~PECTIVE
lubrication is referred to as hydrodynamic lubrication. This lubrication is

responsible for load support in thrust and journal bearings. The principle may be
illustrated by a slider bearing as shown in Fig.l.6. When the lubricant flows
through the convergent wedge, the mass flow rate should be the same. This is
possible only through pressure generation that in turn modifies the velocity
distribution across the film in such a way that the overall flow is the same. The
velocity profile at the inlet and outlet are shown in Fig. 1.6. Thrust bearings which
support axial load utilise this principle. The rotating runner is supported on a set of
thrust pads which are tapered. The thrust pads can be fixed or pivoted. In pivoted
pads the gap changes as a function of operating conditions.
In the case of journal bearing which supports the radial load the lubricant flows
through a convergent-divergent wedge. During operation, the shaft is eccentric
with reference to the stationary journal. A typical full journal bearing is shown in
Fig. 1.7. Eccentricity e is shown in the figure. Radial clearance c is defined as the
difference in the radii of the journal and the bearing.
The flow of fluid in the bearing is governed by the well-known Reynolds equation
that forms the basis for design. The normally used equation for two-dimensional
flow is
0 h3 + =12url0 (1 26)
where
h - film thickness at (x, y)
-u = ~ u ~+ u 2 where u~ and u 2 are the velocities of the two surfaces
2
This equation is based on incompressible flow and the assumption that flow is in X
and Y directions only. The flow in the Y direction (into the paper) is also called
side leakage. In some cases, flow in the Z direction also is to be considered in
which case the three-dimensional equation is necessary. Such situations arise in
dynamically loaded bearings and porous bearings. If the equation is treated in one
dimension with no flow in Y direction (infinite dimension in Y direction) and
applied to the slider bearing in Fig. 1.6 the equation simplifies to
dp= 6~rlo(h - h ) (1.27)

dx -ha
Eccentricity 9
Bearing
Journal
O' and 0 Journal

a n d s h a f t centres
'b
Pressur(
distribu
Fig.l.6. Slider bearing showing velocity and pressure distribution.
Pmax z
Pressu
W 0 x
A Stationary
hl ~'r ha
ng
~ . . , . _ _ U _ . . b _ _ _._,,.
Fig. 1.7. Journal bearing showing eccentricity and pressure distribution.

where h is the film thickness at d__pp= 0

dx
The Eq. (1.27) can be easily solved analytically with the known boundary
conditions at the inlet and outlet.
Side leakage cannot be neglected for finite sized bearings and the two dimensional
equation has to be solved by finite difference methods. The Reynolds equation may
be expressed in cylindrical coordinates in the case of journal bearings. With journal
bearings, it is difficult to specify the boundary conditions in the divergent zone and
some assumptions are necessary. Hamrock [26] has provided a detailed
consideration of these methods. Graphical procedures are also available due to
Raimondi and Boyd [27,28] that are commonly used. These procedures are also
well discussed in [26]. The charts relate the bearing characteristic number to the
minimum film thickness for different length/width ratios of the bearing. In the case
of journal bearing the bearing characteristic number is defined as rl
where N is in revolutions per second and p is the pressure, N/m 2. Here p refers to
the pressure based on the projected area. In the case of thrust bearing it is defined
as where m - ~ and u is sliding velocity, m/s. The film

b
thickness increases with the beating characteristic number. The temperature
increases in the bearing due to friction and the effective film thickness has to be
arrived at by an iterative procedure. Simplified graphical procedures are available
to account for temperature rise and given in the earlier references [27,28]. The
pressure effects on viscosity are not significant in hydrodynamic bearings and are
neglected.
Bearings are designed so that the operating minimum film thickness is adequate for
complete separation of the surfaces. The film thickness builds up from zero at the
start to the designed value at the operating speed. Some boundary contact and wear
are inevitable during start and stop of a bearing. The surfaces can be separated in
the initial zone by an external pressure. Separation by external pressure is also
possible between parallel surfaces. This is referred to as hydrostatic lubrication. As
external pumping to high pressures is necessary with the associated hardware, this
approach is not common.
1.6.3 E l a s t o h y d r o d y n a m i c l u b r i c a t i o n
Concentrated stresses occur in line and point contacts. Such contacts occur in
components like gears and rolling element bearings. Stresses and their distributions
in line and point contacts are treated elsewhere in the book. For the present, only
the available equations to calculate EHL film thickness are considered. At this
stage, a brief consideration of the origin of EHL as a separate regime is in order. A
rolling line contact involves equal velocities of the two rollers. A rolling/sliding
contact involves unequal rolling velocities. In this case, the relative sliding velocity
is the difference between the two rolling velocities. When the film thickness
between two rolling/sliding cylinders was calculated by normal hydrodynamic
theory the film thickness was found to be negligible. On the other hand, many
gears that simulate such contacts were found to operate with negligible wear
suggesting separating films. When pressure effect on viscosity and elastic
deformation were taken into account, the film thickness calculated was found to be
adequate to justify the observed operation in the gears. The final equations now
commonly used are based on the complete numerical solutions as formulated by
Dowson and Higginson [29] and Hamrock and Dowson [30]. The nature of the film
and pressure distribution is shown in Fig. 1.8 for a line contact. It may be seen that
the film is nearly parallel except for a small portion where there is a decrease in
thickness. The point of minimum thickness corresponds to the pressure spike in the
pressure distribution. The overall pressure distribution is similar to the normal
pressure distribution based on Hertzian theory except for the pressure spike. The
formula for minimum film thickness is given by
hmin - 2.65 G~176

- (1.28)
R W~
where
G = ~E ~ dimensionless material parameter
U u~
. . . .+. .u 2 , dimensionless speed parameter
= qo 2E'R
w
W - ~ , dimensionless load parameter
E*RLr
R =
1
1-11
+ ,m
E* 1-V12
= El + ,N/m 2
E~ ))
w = total load in the contact, N
L~ = length of the contact, m
u~, u 2 = velocities of the two surfaces
Equations are also available for point contacts. The relationships are similar except
for some change in exponents.
_o- /
..
__ Inlet region _1_ Hertzion region _1

- " " - I - - ' . . . . . . I irl
Fig. 1.8. Line contact illustrating film thickness and pressure distribution.
The above equation is for contacts in which pressure effect on viscosity and elastic
deformation are significant as in the case of steels. Such a contact is designated as
elastic-variable viscosity (EV) regime. Several other situations can occur in which
these conditions are not applicable. For example in the case of a contact between
steel and rubber the elastic deformation is dominant while the pressure effect on
viscosity may be negligible. At the other extreme with two very rigid surfaces, the
deformation is small while the pressure effect on viscosity is significant. Modified

equations are available to treat such problems [30,31 ].
Film thicknesses in EHL range from 0.1 to 1.0 pm and are an order of magnitude
lower than observed in hydrodynamic lubrication. Also, as can be seen from the
equations load influence on film thickness is small. Rolling/sliding velocity, elastic
modulus, and pressure coefficient of viscosity have a dominant influence on the
film thickness.
1.6.4 Thinning films and boundary lubrication
It is desirable to operate tribological contacts with fluid films completely

separating the surfaces. It is normally considered that complete separation occurs
when the minimum film thickness is 3a ~ where a" is the composite roughness of
the surfaces as defined in Eq. (1.10b). The ratio of film thickness to roughness is
expressed as
A =
h min (1.29)
~J
Many real components operate with A values less than 3.0 leading to asperity
contacts. This can be easily appreciated with regard to EHL in which film
thicknesses are of the same order as the roughness. Even with hydrodynamic
conditions, the film thickness may reach very low values. Ring-liner contact in IC
engine is a good example where the film thickness at the top and bottom dead
centres is very low due to reversal of velocity at these points. At the middle of the
stroke where maximum velocity occurs the film thickness is higher and complete
separation of surfaces is possible. Even at the middle of the stroke the film
thickness may be about 1.0 ~m due to the complex nature of lubrication and high
temperatures. As a rough guide, it is considered that when A is less than 0.5 the
load is entirely supported by boundary layers. This regime is referred to as
boundary lubrication. Boundary lubrication is mainly governed by the lubricant-
solid interaction. Load is supported by both hydrodynamic and boundary films for
A in the range of 0.5 to 3.0. This regime is referred to as mixed or partial
hydrodynamic regime.
Film thickness is a function of operating conditions. With rapid technological

changes, the equipment sizes are reducing and tribological components are
operating under severe conditions. This leads to thinning films with increased
boundary contact. Better methods of controlling wear and friction in boundary

contacts are being increasingly sought. Another aspect of thinning films is the
calculation of film thickness. Both EHL and hydrodynamic theories are based on
smooth surfaces. When thin films are involved with A ratios of 6.0 or less the
roughness influence on film thickness has to be taken into account. This is a
difficult problem to be resolved. The problem is analogous to the flow of water in a
channel. Flow of water can be easily modelled when the flow is normal. When the
flow is reduced to a trickle one can observe meandering flow that is affected by
boulders and the sand. In a similar fashion the flow, and hence the film thickness,
are affected by roughness for low A values. It is also postulated theoretically that
when the roughness has preferred orientation the film thickness is influenced
positively or negatively by the lay direction. This idea opens up the possibility of
"gaining" film thickness by manipulating the surface roughness. These issues will
be considered in more detail in chapter 8 dealing with mixed lubrication.
References
1. D. Dowson, History of Tribology, Longman, New York, 1979.

2. F. P. Bowden and D. Tabor, The Friction and Lubrication of Solids, Part II, Oxford
University Press, London, 1964, Chapter IV.
3. D. Tabor, Junction growth in metallic friction: the role of combined stresses and surface
contamination, Proc. Roy. Soc. London, A 251 (1959) 378.
4. D. H. Buckley, Surface Effects in Adhesion, Friction, and Lubrication, Tribology series
5, Elsevier, Amsterdam, 1981.
5. P. M. Vedamanikam and D. V. Keller, A correlation between static adhesion data and the
dynamic friction coefficients for two cobalt alloys and iron under vacuum conditions,
ASLE Trans. 16 (1973) 73.
6. E. Rabinowicz, Friction and Wear of Materials, Wiley, New York, 1995.
7. N. P. Suh, Tribophysics, Prentice-Hall, New Jersey, 1986.
8. K. L. Johnson, Contact Mechanics, Cambridge University Press, London, 1985, 410.
9. J. A. Greenwood and J. P. B. Williamson, Contact of nominally fiat surfaces, Proc. Roy.
Soc. London, A 295, (1966), 300.
10. S. Ganti and B. Bhushan, Generalised fractal analysis and its applications to
engineering surfaces, Wear, 180 (1995), 17.
11. D. J. Malvaney, D. E. Newland, and K. F. Gill, A complete description of surface
texture profiles, Wear, 132 (1989) 173.
12. J. C. Jeagar, Moving sources of heat and the temperature at sliding contacts, Proc. Roy.
Soc. NSW, 76 (1942) 203.
13. X. Tian and F. E. Kennedy, Contact surface temperature models for finite bodies in dry
and lubricated sliding, J. Trib., ASME, 115 (1993) 411.
REFERENCES 31
14. G. A. Berry and J. R. Barber, The division of frictional heat- A guide to the nature of
sliding contact, J. Trib., ASME, 106 (1984) 405.
15. J. A. Greenwood, An interpolation formula for flash temperatures, Wear, 150 (1991)
153.
16. S. Jahanmir (ed.), Friction and Wear of Ceramics, Marcel Dekker, New York, 1992.
17. T. Saito, Y. lmada, and F. Honda, Chemical influence on wear of Si3N4 and hBN in
water, Wear, 236 (1999) 153.
18. R. S. Gates, S. M. Hsu, and E. E. Klaus, Tribochemical mechanism of alumina with
water, Trib. Trans STLE, 32 (1989) 357.
19. Y.Yamaguchi, Tribology of Plastic Materials, Tribology series, 16, Elsevier,
Amsterdam, 1990.
20. G. M. Bartenev and V. V. Lavrentiev, Friction and Wear of Polymers, Elsevier,
Amsterdam, 1981.
21. K. Friedrich (ed.), Advances in Composite Tribology, Composite Materials Series, 8,
Elsevier, Amsterdam, 1993.
22. T. R. Choudhary, A. Sethuramiah, O. Prakash, and G. V. Rao, Development of
polytetrafluoroethylene composite lining for a hydrogenerator thrust pad application,
Proc. Instn. Mech. Engrs. 214, Part J, (2000) 375.
23. P. J. Blau, Friction Science and Technology, Marcel Dekker, New York, 1996.
24. C. M. Ettles and C. E. Hardie, The friction of some polymers and elastomers at high
values of pressure X velocity, J. Trib., ASME, l l0 (1988) 678.
25.C.J.A. Roelands, Correlation aspects of the viscosity-temperature-pressure relationship
of lubricating oils, Druk, V. R. B. Groingen, Netherlands (1966).
26. B. J. Hamrock, Fundamentals of Fluid Film Lubrication, McGraw-Hill, New York,
1994.
27. A. A. Raimondi and J. Boyd, Applying bearing theory to the analysis and design of
pad-type bearings, ASME Trans. 77 (3), 1955, 287.
28. A. A. Raimondi and J. Boyd, A solution for the finite journal bearing and its application
to analysis and design (in three parts), ASLE Trans. 1 (1958) 159.
29. D. Dowson and G. R. Higginson, Elastohydrodynamic Lubrication, Pergamon, Oxford,
1977.
30. B. J. Hamrock and D. Dowson, Ball Bearing Lubrication-The Elastohydrodynamics of
Elliptical Contacts, Wiley, New York, 1981.
31. R. D. Arnell, P. B. Davies, J. Halling, and T. L. Whomes, Tribology Principles and
Design Applications, Springer Verlag, New York, 199 I.
Nomenclature
a' constant
32 f'HAPTER 1. TRIBOLOGY IN PERSPECHVE
A dimensional constant
Ai area of contact at asperity i
A, nominal contact area
A, total real contact area
b slider bearing length
b constant
b* dimensional constant
c radial clearance
c constant
Cp specific heat
d separation between rigid plane and mean surface height
E effective modulus
E* effective modulus defined'as E/2
El, E2 elastic modulus of the two bodies in contact
f coefficient of friction
F friction force
G dimensionless material parameter
H hardness
H viscosity of the reference fluid with 100 VI in relation to test fluid
h film thickness at dp/dx = 0
hi, h2 inlet and outlet film thickness in slider beating
hmin minimum film thickness
k thermal conductivity
k l , klI thermal conductivities of the bodies I and II
llt
kI
k
kli
l half contact dimension of square heat source
L load in dry contact
Lc length of EHL line contact
Li load on asperity i
L viscosity of the reference fluid with 0 VI in relation to test fluid
hi - h 2
m
b
n number of asperity contacts
N revolutions per second
m
N = OogH - logU,)/logYt
P pressure
NOMENCLA T~/RE 33
vl
Pe Peclet n u m b e r -
2Z
q heat flux
r journal radius
R equivalent radius of cylinders
R1, R2 radii of cylinders 1 and 2
Ra centre line average surface roughness
root-mean-square (rms) surface roughness
s shear strain rate du/dz
si shear strength of the interface
sm shear strength of the metal
T temperature
bulk temperature
T~ contact temperature
u sliding velocity
Ul ~ U 2 velocities of the two surfaces 1,2 in lubricated contacts
U mean surface velocity defined as (u~ + u2)/2
U dimensionless speed parameter
Ut viscosity of the test fluid at 40~
v sliding velocity, m/s in dry contact
w total load in EHL line contact
W dimensionless load parameter
y ordinate value in Fig. 1.5
Yt viscosity of the test fluid at 100~
z asperity height over mean surface line
Greek Letters
(I constant
mean asperity radius
equivalent asperity radius
~1~ [~2 mean asperity radii of surfaces 1 and 2
Z thermal diffusivity
6i contact deformation of asperity i - ( z~ - d)
A0 temperature rise in the contact
A0 I surface temperature rise assuming all heat is flowing to body I
AOII surface temperature rise assuming all heat is flowing to body II
34 CHAPTER 1. I'RIBOLOGY IN PERSPECTIVE
8 coefficient = 0.398yP~ -~
q number of asperities per unit area
1] absolute viscosity
rip viscosity at pressure p
rip number of peaks per meter length
rl0 absolute viscosity at the given temperature and p = 0
tp partition coefficient
o(z) height distribution function
K ratio of shear strength of the interface to the shear strength of the
metal, s---L
Sm
A ratio of film thickness to roughness
v kinematic viscosity
v 1) V2 Poisson ratios of the bodies 1 and 2
rms roughness of the surface
(3"I) (~2 rms roughness of surfaces 1 and 2
G* equivalent roughness
shear stress in the fluid
pressure coefficient of viscosity at the given temperature
plasticity index
35
2. Lubricant Technology
A Survey
2.1 Introduction
A lubricant may be defined as a solid or fluid film interposed between surfaces in

relative motion to reduce friction and/or wear as defined in the previous chapter.
The present chapter deals mainly with petroleum based lubricants used extensively
in industry. Conventional and modem technologies involved in the manufacture of
petroleum base stocks are considered first. Common additives and the formulation
technology form the next section. Major lubricant specifications and test methods
are then covered followed by a discussion of the issues related to performance.
Tribological implications are also briefly considered and serve as a prelude to the
main theme of this book. The next section deals with synthetic lubricants. The final
section deals with the environmental issues involved in the use and disposal of
lubricants. The literature available in lubricant technology is vast and the scope of
the present chapter is limited to an appreciation of the modem technologies and
their impact on lubricant performance.
2.2 Petroleum base stock manufacture
2.2.1 Crude distillation
Crude petroleum as received is first distilled in an atmospheric distillation column.

Several side cuts are drawn from this column with varied boiling ranges
representing gasoline, kerosene/jet fuel, diesel, and gas oil. Gaseous products are
recovered from the top of the column. Gas oil may be subjected to other processes
like fluidised catalytic cracking to obtain more valuable lighter products. In some
cases crude may first be desalted as necessary. The product range suggested is
typical but varies depending on the crude and market needs. The bottom product
from this column, called reduced crude, is then subjected to vacuum distillation to
obtain the base oils of different boiling ranges. While the term, 'base stock' is
appropriate they are also called 'lubes' or 'lubricating oils' though they are not
formulated lubricating oils. They are also referred to as mineral oils. The context in
which these words are used is normally clear.
36 CHAPTER 2. LUBRICANT TECHNOLOGY- A SURVEY
The basic scheme involved in distillation is shown in Fig. 2.1. Distillation is a

complex process involving several side streams, several reflux streams, and steam
stripping. In vacuum column steam is also used to reduce partial pressure. Crude
consists of a large number of varied hydrocarbons and empirical procedures were
used in the design [1]. The more recent approaches consist of dividing the crude
into several close boiling components, called pseudo components, and then using
elaborate computer programs for design [2]. Such procedures, which approach
rigorous multi-component distillation column design, provide better control of the
distillation process. Referring to Fig. 2.1 the crude is first heated in a furnace and
then fed to the atmospheric column where it is distilled. As shown, several side
streams are withdrawn after steam stripping. Reflux streams can also be seen
including the top reflux. Uncondensed gases are recovered from the top reflux
drum and used for further processing. Vacuum distillation can yield several cuts
called neutrals and in the present scheme three side draws are shown. The top
product can be a light viscosity cut termed as spindle oil. The bottom product from
vacuum column is called vacuum residue and in some cases can yield very heavy
oils called bright stocks by deasphalting. It may be noted here that all reduced
crudes are not suitable for lube manufacture and must have high molecular weight
components of desirable structure. Vacuum distillation can also be practiced on the
reduced crude, which is not lube bearing. The distilled high boiling fractions,
called vacuum gas oils, can be cracked to obtain desirable fuels. The viscosities of
base oils produced in a lube distillation unit can range from 10 cSt at 40~ for
spindle oils to as high as 600 cSt at 40~ for bright stock with intermediate values
for the neutrals. Neutral is normally preceded by a number, which refers to the
older Saybolt Universal Seconds (SUS) viscosity based on time to flow through an
orifice. With regard to heavy bright stocks, the convention is to precede the SSU
value at 210~ The equivalent cSt values at 40~ for neutrals approximately range
0.19 to 0.21 times the SSU value at 100~ The fuels as well as lubes undergo
further refining of varied levels. This is necessary to meet the specifications of
these products.
Though the main interest here is with regard to lubes it is of interest to consider
briefly the tribological problems related to fuels. In modem refineries aviation
turbine fuel (ATF), also called jet fuel, is subjected to hydroprocessing to improve
quality. The processing removes most of the polar impurities. As the fuel itself
lubricates piston pumps that deliver fuel to the combustion chamber, the better
fuels are found to be deficient in boundary lubrication leading to wear and seizure
problems. The associated problems are well documented [3,4]. A recent test
method, ASTM D5001, is aimed at assessing the lubricity of ATF by a ball-on-disc
2.2. PETROLEUM BASE STOCK MANUFACTURE 37
laboratory rig. Another example is the diesel fuel now being manufactured with
very low sulphur, which is causing wear problems that are currently under
investigation [5,6]. Boundary lubrication problems with fuels are thus also
considered as lubricated wear problems.
2.2.2 R e f i n i n g of b a s e stocks
The base oils obtained by distillation need further refining to improve their
characteristics. The selected processing depends on the nature of the base stock and
the desired quality of the refined base oils. It is hence necessary to appreciate the
nature of base oils as obtained by distillation and then go on to a consideration of
refining processes.
CW
~----~ ] [-~ Gale, to processing
/ 6o"x.~ C 9
I ........ ~ 1 7 6 1 7 6 1 7 6
PRL__.._J_ _ ,0 I CGosoline) / ~Steam

oil
I 3 ~ C,~L] __ L-r':#-----Light Neutrat
N e,
100N
7 L.~_ ~ Kerosene/Jet *uel / I ~---,-Mediurn
350N
700N
C ~ Gas oil
Furnoce
. V a c u u m residue
- ne~,,c'-$d ~r,,de
PR - Pump around reflux

TR.- Top reflux
C - Cooler
CW - Condenser
Fig. 2.1. Crude distillation process. Plate numbers are indicative only.
2.2.3 Base oils
Crude oils are classified as paraffinic, napthenic, and aromatic depending on the
predominant structures in the complex molecules. Base oils obtained by distillation
are similarly termed and correspond to the original crude. Aromatic crudes are not
used for the manufacture of base oils as aromatics have poor viscosity-temperature
relations and decrease rapidly in viscosity with temperature. They also have
thermal and oxidative stability problems. Paraffinic structures have very good
viscosity- temperature relationship besides stability. Thus, most of the base oils
manufactured in modem refineries are paraffinic in nature and the focus is on these
oils. Naphthenic oils, which are intermediate in nature, are manufactured in limited
quantities, mainly for low temperature applications because of their good low
temperature fluidity.
The paraffinic base oils range in molecular weight from C25 to C40. Most of the
molecules in the oils will have mixed structures. For example a long paraff'mic side
chain may be attached to a benzene ring. The structure of this molecule is mainly
paraffinic in the sense that most of the carbon atoms in the molecule are of
paraffinic nature. To appreciate the structures better Fig. 2.2 illustrates three typical
molecules in which the normal and isoparaffin chains are linked to a saturated ring,
condensed naphtheno-aromatic rings, and to a 3-ring aromatic structure. The
structure (a) will have high paraffinic nature followed by (b) and (c). The structure
(c) is useful only if the aromatics are saturated. It may be appreciated that several
other combinations are possible. It is apparent that paraffinicity is relative. There
are two approaches available to characterise the nature of hydrocarbons involved.
One is the difficult process of isolating the molecules to the extent possible by a
combination of gas chromatography and mass spectrometry and determining the
structures. Nuclear magnetic resonance spectroscopy is also used to understand the
structures in more detail. Dorinson and Ludema [7] have reported a good survey of
these methods. The other approach is to estimate the paraffinic, napthenic, and
aromatic carbon content by indirect methods like n-d-M (n is refractive index, d is
density and M is molecular weight), characterisation factor K, and other techniques
which are well documented by American Petroleum Institute [8]. These methods
are based on the fact that density, boiling range, and refractive index vary with the
nature of hydrocarbons and hence can be correlated with the overall nature of the
fractions involved. These approaches, which are indirect, have stood the test of
time and are extensively used. These techniques may not be adequate to
characterise very heavy crudes, which the refiners are increasingly obliged to
process. Also more advanced methods are needed to characterise the polynuclear
aromatics. These issues are well covered in a recent book [9]. The other structures
in the base oils are those containing hetero-atoms, which are mainly sulphur and
nitrogen. Some oils may have significant sulphur and nitrogen, which need to be
removed by hydrogenation.
(o|
CH3
(a) long isoparaffin chain linked to
naphthene ring.
~ I
(CH2) 8 -- CH --CH --C~ -- CH3
I I
CH3 CH3
(b) (b) two naphtheno-aromatic rings

bridged by normal paraffin chain.
(c)
(c) 3-ring condensed aromatic
R I ~ R2
linked to two small alkyl groups.
Fig. 2.2. Typical structures in base oil.
2.2.4 Technology of refining base oils
The distilled paraffinic oils need upgrading in three areas. One is the reduction in
aromatic content to improve VI and oxidation stability, second is the removal of
long chain normal paraffinic structures which form wax and create low temperature
flow problems, and the third is the removal of sulphur, nitrogen and other polar
impurities to acceptable levels. The third is also referred to as a finishing process. It
is also termed hydrofinishing when it involves hydrogen treatment. Aromatic
removal can be affected by solvent treatment. The main solvents used in the
industry for this purpose are furfural, N-Methyl-2-Pyrrolidone (MP), and phenol.
These solvents preferentially dissolve the aromatic constituents thereby enriching
the raffinate with desirable constituents. The extract is stripped of the solvent and
the aromatics are utilised as feed stocks for petrochemical industries or fluidised
catalytic cracking. Instead of removing, the aromatics can be converted to saturated
cyclic structures by hydrogenation. Hydrogenation also simultaneously removes
sulphur and nitrogen. Solvent extraction involves loss of product while there is no
loss in hydrogenation as aromatics are converted and retained. Solvent extraction
can also be done on the vacuum residue to obtain very heavy oil called bright
40 CHAPTER 2. LUBRICANT TECHNOLOGY A SURVEY
stock. The solvent used for this purpose is propane, which preferentially dissolves
the oil fraction and rejects the asphaltenes. This process, called propane
deasphalting, is done only if the bright stock content is high enough to
economically process the residue. The conventional dewaxing process is again
solvent based. The major solvent used is methyl ethyl ketone (MEK). The wax is
precipitated from the solution by chilling to low temperature and then filtering. The
hydrofinishing is meant to remove polar impurities by a hydrogenation process and
is the final step to improve colour and stability. This brief discussion suggests that
in a refinery several combinations of treatment are possible which indeed is the
case. The flow diagram given in Fig. 2.3 illustrates the various combinations
possible. Chemical processes consisting of acid and alkali treatments are normally
not used in modem refineries though mentioned here. Catalytic dewaxing
mentioned is a modem process which is considered next.
R~']N~ D~-W~U<ED nNISHED

DISTII.J_ATES OILS OILS PRODUCTS
' k '- '

Of~ LOBE
WAXES
* VACUUM RESIDUUM
Fig. 2.3. Refining of base oils. (From Ref. [11] p2, by courtesy of Marcel Dekker Inc.)
The modem refining processes include catalytic dewaxing, catalytic isomerisation,

and hydrocracking. Catalytic dewaxing and isomerisation processes involve
selective zeolites with specific pore sizes that preferentially allow long chain
paraffins into the pores where they undergo transformations. These transformations
can involve cracking leading to lighter ends, which can be distilled off to obtain
fuels, or moderate cracking coupled with isomerisation resulting in isoparaffins
that have high VI. These processes are conducted under hydrogen pressure leading
to saturation of the molecules. These processes are very specialised and are made
possible by the major developments in zeolite manufacture leading to controlled
pore sizes. Different metals are incorporated to influence the hydrogenation
function.
Hydrocracking is a severe cracking and hydrogenation process originally

developed to manufacture high quality fuels. The process is flexible and, for
example, typical heavy gas oil can be cracked to diesel or gasoline preferentially by
changing the operating conditions in a hydrocracker. With developments in
catalysts including higher tolerance to impurities, the process in the recent decade
is being adapted for lube manufacture.
The flexibility is such that vacuum gas oils which are not lube beating in the
conventional sense can be converted to base oils of high quality. Hydrocracking
also removes sulphur and nitrogen during processing. The reactions involved in
hydrogenation are shown in Fig. 2.4. Besides saturation, hydrocracking also
involves isomerization and the opening up of naphthenic tings resulting in
paraffinic structures. The catalyst has cracking and hydrogenation functions, which
can be selectively controlled, by the extent of zeolite in the silica-alumina base and
the type and concentrations of metals used. The metals used include cobalt,
molybdenum, nickel, and tungsten. Noble metals like platinum are also
incorporated in some cases. A general coverage of hydrogen processing and
catalysts is available in a recent handbook [10] while specific consideration with
regard to lubricants is available in the book by Sequeira [11]. An example can
explain the flexibility involved. Consider a highly aromatic high boiling
component unsuitable for lubricant manufacture in the conventional sense. This
component can undergo significant saturation via hydrogenation. The naphthenic
structures can further undergo ring splitting leading to desirable paraffinic
structures resulting in good base oil. In the case of long chain waxy feed, waxy
structures can be partially cracked and hydrogenated leading to less viscous, but
more acceptable, base oils. Thus, there are several combinations, which can be
adopted and integrated into overall lube manufacturing schemes. It needs to be
pointed out that all treatments involving hydrogen may be categorised under
hydroprocessing. The difference between the various categories is the relative
emphasis on cracking and hydrogenation. The operating conditions, catalysts, and
hydrogen consumption vary with the nature of treatment. Thus hydrogenation used
in place of solvent treatment to saturate aromatics involves negligible cracking
while hydrofinishing involves final removal of polar impurities and involves a mild
treatment. The wax cracking mentioned above is also a hydrocracking process, but
it is selective to waxes.
It is here necessary to briefly consider other issues involved in hydroprocessing.

The poisoning of catalyst limits the efficiency and cycle life of a catalyst and in
some cases pretreatment of feed is necessary to reduce impurities. Reaction rates
for various components can be vastly different. As an example, desulphurisation of
42 CHAPTER 2. LUBRICANT TECHNOLOGY- A SURI/TY
thiophene can occur 27 times faster than dibenzothiophene [12]. Hence major
structures involved in the feed and their reaction kinetics are to be understood in
selecting the nature of the process and the operating conditions. The costs involved
in severe processes like hydrocracking can be significant in terms of investment
and operating expenses. Hydrogen availability is another major factor. These
aspects deter some of the refiners from adopting large scale hydroprocessing.
PURIFICATION
RCHzSH+Hz ~dLT._ALY_~.T_~ RCH3+HzSI
IR
~'--S"S~ +H2 CMALYST:._ CHsCHzCHCHs+H2SI
- ..r
~ +H2~ 0
SATURATION
=,
~,.. Rz
+H, ~
~
L=
=[ v
, R
L=A=j
v v
=
CRACKING
~ .,.Rz R,..~..Rz ~Rs
Fig. 2.4. Chemical reactions in hydroprocessing. 'S' refers to sulphur in the first reaction
and to saturated rings in others. (From Ref. [ 11] p120, by courtesy of Marcel Dekker Inc.)
2.3 Additives and formulation technology
Finished base oils as obtained from refinery are normally blended with additives to
obtain lubricating oils used in industry. Additives are added to enhance the existing
properties of the base oil and to impart new specific properties.
2.3.1 T h e n a t u r e o f additives
It is convenient first to appreciate the overall physicochemical properties of

lubricants as given in Table 2.1. The test methods are given in ASTM standards for
2.3. ADDITIVES AND FORMULATION TECHNOLOGY 43
petroleum products [13]. While these are American standards, many other
standards are available from organisations of different countries, which in some
cases differ from ASTM standards. These standards detail the procedure required
to conduct the tests and include information on repeatability and reproducibility of
each test. The table also gives the central purpose of each test. The lubricant
specifications prescribe the acceptable values of the various properties and are
drawn up by different organisations in various countries. There are thus a large
number of specifications for lubricants. The original properties of the base oil
strongly influence the additive level to meet the required enhancement. Another
important aspect of additives is that more than one additive is routinely added to
improve several properties together. It is important that such systems do not
interact with each other in an antagonistic manner. Sometimes the additives may
interact synergetically enhancing the effect of each, which is desirable. When the
additive system does not interact negatively it is said to be compatible. Yet another
aspect is the ability of different commercial oils for the same end use to be
compatible. As an example, if a customer is obliged to mix different brands of
engine oils they should not adversely affect the performance. The major categories
of chemical additives used for enhancement of properties are described below. The
detailed treatment of chemical structures is available in the literature [14,15].
Lubricants and their applications are covered in several books [ 16,17,18] and the
following is a limited treatment. Detergent additives consist mainly of oil soluble
sulphonates and phenates of calcium, barium, and magnesium. These additives
have a major role in engine oils to prevent deposition of carbon and varnish in the
piston-ring zone. Modem detergents are overbased to provide a reserve of alkali to
neutralise the acidic components formed during combustion. Dispersants again are
mainly used in engine oils to prevent low temperature sludge deposition and
include succinimides, polyamides, and copolymers containing polar groups. The
antioxidants used include hindered phenols like ditertiarybutyl paracresol and
amines like 13-naphthyl amine. These additives find application in turbine and
hydraulic oils. Zinc dialkyl dithiophosphates are a very special class of additives
extensively used in engine oils. These additives are multifunctional and can
simultaneously function as antiwear, antioxidant, and anticorrosive additives.
Silicones are the usual antifoam agents used at ppm level to control foaming in
turbine oil, hydraulic oils, and gear oils when necessary. The VI improvers mainly
consist of long chain polymers, which have higher solubility at higher temperatures
thereby compensating for the loss of viscosity of the base oil. This results in higher
VI for the system. The major categories used are polymethacrylates,
polyisobutenes, and olefin copolymers. The molecular weight of the polymers used
is above 50,000. VI improvers are mainly used in engine oils and also in some
other applications like gear oils. Pour point depressants are polymeric molecules
Table 2.1: Physicochemical properties of lubricants
Property Test Method (ASTM) ~ Comments

1. Flash and fire D92, D93, D56, DI30 Approximate characterisation of flammability.
point Specification control and for detection of mix-
up with higher/lower boiling components.
2. Aniline point D611 Indicates aromaticity and useful to check seal
compatibility.
3. Neutralization D974, D664 Specification control on base stocks and
number finished products. Increase in acidity is
Total acid number generally an indication of oxidative
Total base number degradation oil. Base number indicates alkali
reserve in engine oils due to detergents.
4. Water content D95, D 1744 Water deleterious with regard to oxidation and
corrosion. Control and estimation critical in
turbine oils, hydraulic oils, and transformer
oils.
5. Carbon residue D189, D524 Indicates tendency for carbon deposits. Less
widely used now.
6. Pour point and D97, D2500 Relevant with regard to low temperature flow
cloud point properties within limits. More representative
methods utilising rotary viscometers are now
being used to characterise flow.
7. Ash content, D482, D874 Indicates overall metallic content due to
sulphated ash additives. Increase in use indicative of wear
content debris accumulation.
8. Sulphur content D129 Sulphur in base oil limited by specification.
Also estimation gives sulphur content of
additives when present.
9. Copper strip D1275 Indicative of 'active' sulphur content.
corrosion Important with regard to copper containing
metals like bronzes. Also indicative of refining
level of base stocks.
10. Viscosity and D445, D2270 Major factor in lubricant selection and
viscosity index invariably specified. Increase in viscosity in
use is indicative of oxidation.
11. Oxidation D943 and other tests Usually long duration tests. Critical for
stability applications like turbine and hydraulic oils.
12. Demulisibility D1401 Indicates the ability to separate water and is
characteristics important for turbine oil, hydraulic oil.
13. Foaming D892 Foaming means ineffective lubrication and
characteristics should be controlled.
14. Additive Modem methods like Very important to know the level of additives
elements emission spectroscopy and how they are depleting.
or wet methods listed in
ASTM.
15. Tribological D2783, D4172 and Estimation of wear and load bearing capacity at
tests others a ~iven set of conditions.
"ASTM methods are representative only. A large number of methods are available in some cases like
oxidation tests.
2.3. ADDITIVES AND FORMULA 110N TECHNOLOGY 45
with a polar group. These molecules prevent agglomeration of wax crystals and
improve the low temperature flow properties of the lubricant. These additives again
have major application for engine oils. Extreme pressure (EP), and antiwear
additives are sulphur-phosphorous compounds used to control wear and scuffing
and are dealt with in the chapters on boundary lubrication and lubricated wear.
2.3.2 Formulation methodology
Base oils with the required characteristics are blended with the selected additive
package. The modem blending plants are automated with precise control of the
additive dosage and blending parameters. The product is then evaluated as per
specifications set for the finished lubricant. If the lubricant does not pass the
specifications, the formulations are re-worked until the specifications are met. For
formulations already set, the procedure involves routine manufacture. The
additives may be available as a package from additive manufacturer or may be
individually purchased, blended, and optimised by the formulator. The lubricant
formulation is no longer a simple blending of known packages. It is now a very
dynamic and demanding technology. The first factor is the continuing demand
from equipment manufacturers for better products with longer life under
increasingly severe operating conditions. These demands have their effect on
specifications, which get more stringent. In some cases the equipment
manufacturer may demand qualification against some in-house tests in addition to
the specifications. The other factor is the increased environmental protection
demanded of the lubricants, which translates into regulations on toxicity and bio-
degradability. In addition, the formulation will have to be cost effective to compete
in the market. These demands also have their influence on the quality of base oils.
This is indeed the reason for increasing levels of hydroprocessing resulting in more
stable and very high VI base oils. Earlier base stocks were considered adequate
with 95 to 100 VI. Today oils with VI above 120 have become common. These
hydroprocessed oils have also some deficiencies like poorer solubility for
additives. These are taken care of by modification of additives. The demands also
have an impact on additive technology, resulting in better additives. The
methodology available for performance testing is a major problem. Laboratory
tests in many cases cannot correlate well with practice, necessitating expensive
field trials. With regard to engine oils, several engine tests may be needed for
qualification which is an expensive and time consuming process. This important
issue of correlation between laboratory and field performance shall be considered
later.
2.4 Lubricant specifications
Lubricant specifications define firstly the limits on physicochemical characteristics,

which are appropriate for the particular lubricant. They also specify limits on
performance in laboratory rigs as applicable. The term laboratory rig is used here
with reference to mechanical test rigs. It is common to divide lubricants into
automotive and industrial lubricants. Automotive lubricants consist of engine and
gear oils. Engine oils form about 50% of the lubricants manufactured and have
extensive engine tests incorporated in their specifications. Thus they need to be
treated as a separate class.
2.4.1 Automotive lubricants
Engine oils are the major category of automotive lubricants. The first classification
to be considered for engine oils is the well-known SAE viscosity classification.
This classification is based on the viscosity ranges of the lubricants in cSt at 100~
In all six winter grades and six summer grades are specified. The winter grades
range from 0W to 25W while the summer grades range from 20 to 60. The recent
SAE specification SAE J300-1999 specifies important additional requirements. It
specifies low temperature cranking and pumping requirements for the winter
grades which must be additionally met. The low temperature rheological behaviour
of lubricants is complex and the prescribed tests conducted with rotary viscometers
provides reasonable assurance of field performance. Another important
development is the high temperature high shear (HTHS) viscosity at 150~
specified for the summer grades. This is because the recent research has shown this
viscosity is more relevant with regard to beating performance.
It is necessary here to consider the concept of multigrade oil. This refers to those
oils which can meet the winter and the summer grade simultaneously. Thus 20W-
40 means oil that meets 20W as well as 40 grade. This is made possible by adding
VI improvers to the relatively low viscosity winter grade oil. While the original
purpose was mainly to avoid two separate grades, the multigrade oils are capable of
reducing energy losses due to lower friction. The stability of VI improvers at high
shear rate in the ring-liner zone and their rheological behaviour are now areas of
major investigation. Several laboratory tests are available to study the shear
stability of VI improvers.
The performance with regard to detergency, sludge, corrosion, and wear is

necessarily determined by special engine tests. This is because effective laboratory
2. 4. LUBRICANT SPECIFICATIONS 47
small scale tests are at present not available for such evaluations, though several
methods have been attempted. Detailed consideration of the engine tests is not
within the scope of the present chapter and only the major approaches involved are
indicated. Diesel and gasoline engine oils are classified into several categories
depending on the severity of operation. For each category, several engine tests are
specified. The main aspects studied are oil oxidation and beating corrosion, rust,
high temperature deposits valve train wear, and low temperature sludge. Some
engine tests are also conducted to study ring-liner wear and fuel economy. Several
organisations are involved in drawing up classification systems and engine tests.
These include the American Petroleum Institute, US Military, and European
Commission. The engine tests are .changing continuously and it is common to
observe that several tests become obsolete and are replaced by new ones. At
present there is no other approach available to ensure lubricant performance. The
severe performance requirements are being met by substantial improvements in
additive technology and base stocks. The base oils now available are categorised
into five types as per API 1509. Types I, II, and III are petroleum based while IV
and V are synthetic based. Type III is a very high VI base stock with a minimum
VI of 120 and is now replacing other types, mainly in engine oils. High to very
high VI base oils are now available because of the modem hydroprocessing
technologies discussed earlier.
The detailed engine test procedures are available through the standards
organisations. A list of engine tests currently used by API to qualify diesel engine
oils is given in Table 2.2. This is to illustrate the complex testing requirements.
Each lubricant category will have its specific engine test requirements. Thus API
CF-4 will have to pass Caterpillar 1K, Mack T-6, and Mack T-7 tests to qualify.
Two-stroke engine oils have special requirements and are designated by API into
five categories TA to TE with increasing severity. The engine tests used evaluate
lubricity, ring sticking, detergency, and the tendency for pre-ignition. The engine
tests involved are jointly developed by CEC, ASTM, API, and SAE. 2-Stroke
engines are extensively used in India and China and there is special interest in
developing specifications in this area. The Indian standard IS: 14234-1996 covers
engine tests based on indigenous 2-Stroke engines. Japan also has detailed engine
tests based on their JASO standards. Rigorous specifications are also to be met by
the automatic transmission fluids and gear and axle oils.
The viscosity grades for gear oils are governed by SAE J306 recommended
practice. These cover nine grades ranging from 70W to 250 with corresponding
48 CHAPTER 2. LUBRICANT TECHNOLOGY A SURVEY
Table 2.2" Current diesel engine tests for API category lubricants.
(Source: lubrizol.com/reference library)
i , ,, , , ,
Caterpi- Roller-
Parameter Caterpi- liar 1M- Caterpi- DDC Mack Mack Mack Follower
liar lk llar IN 6V92TA T-6 T-7 T-8
PC Wear
Engine
Displacement 2400 2000 2400 9046 11,000 11,000 11,930 6200
(cu cm)
Test Duration (h) 252 120 252 100 600 150 250 50
Test Phases 1 1 1 2 3 1 1 1
Oil Change None None None None None None None None
Period (h)
Engine Speed 2100 1800 2100 1200 & 1400- 1200 1800 1000
(rpm) 2300 21 O0
653 &
Fuel Rate (kW) 140 103 140 1130 523-697 514 795 113
Overall Load 945 & 1262-

1240 972 1240 1655 1458 636
(kPa) 1075 1531
Temperature 127 124 127 103 & 66-99 57.2 43 Not
(~ 143 applicable
Absolute 248 & Not
240 179 179 98-155 101 203
Pressure (kPa) 352 applicable
Water Outlet
Temperature 93.3 85 93.3 84 87.8 85 85 118
(oc)
Oil-to-Bearing
102.2 &
Temperature 107.2 96.1 96.1 112.8 112.8 104.4 118
111.1
(~
Fuel Sulphur 0.38- 0.40 0.03-
(% mass) 0.42" 0.35 0.05 0.1-0.4 0.1-0.3 max 0.05 0.03-0.05
Test included in API CF- API CF, API CG- API CF-2, API CF- API CF- API API CG-4,
specifications 4 CF-2, 4, MIL- MIL- 4 4 CG-4, MIL-PRF-
MIL- PRF- PRF- MIL- 2104G
PRF- 2104G 2104G PRF-
2104G 2104G
Test evaluates
O O
9 - . . . . o
2.4. LUBRICANT SPECIFICATIONS 49
viscosities of 4.1 to 41.0 cSt at 40~ The conditions are particularly severe for
hypoid axles and tests are required on specified axles. Two important tests are CRC
L-37, which is a low speed high torque test, and CRC L-42, which is a high speed
shock loading test. These tests can be met only with a high level of extreme
pressure additives containing organic sulphur and phosphorous compounds. This is
a classic example of successful use of additives. The operation of hypoid axles
without scuffing is not possible without these additives. The other tests include
oxidation stability, rust, and foaming charecteristics. The API designation GL-5
covers these requirements for severe applications. The manual gearboxes may be
lubricated by lower grades like GL-4 or GL-3. Automatic transmission fluids are
now highly specialised and have to meet various requirements. Besides
physicochemical tests, wear tests on laboratory rigs are specified. Friction
durability tests and shudder tests are also carried out. The shudder test is intended
to ensure freedom from stick-slip behaviour. The specifications currently used are
from Ford and General Motors of US and are designated by MECRON-V and
DEXRON-III. These specifications are also followed in Europe and Japan with
some modifications.
2.4.2 Industrial lubricants
The major classes of industrial lubricants may be divided into metal working
fluids, hydraulic oils, turbine oils, and gear oils. Many applications for industrial
lubricants are less severe than for engine oils, but laying down specifications for
industrial oils has its difficulties of a different kind. These difficulties arise from
very diverse applications of these products in a given category and it is impossible
to prescribe tribological rig tests in each case. Thus, metal removal processes can
involve cutting, drilling, tapping, honing, grinding, and in each operation, again
there can be wide variations. The specifications have to necessarily end up with a
compromise, which is a minimum requirement. Such compromises are inadequate
to predict performance effectively and reliance has to be placed on the knowledge
of the lubricant supplier and the equipment manufacturer, particularly for severe
operations. For many new applications assured performance can be based on field
trials only. This is unlike the case for engine oils where more reliance can be
placed on the engine tests for selecting formulations. The specifications considered
below should be viewed within these limitations.
2.4.2.1 Metal working fluids
Metal working fluids are divided into water based emulsions and neat fluids. The
water based fluids consist of oil in water emulsions and are widely used in metal
cutting operations. Besides emulsifiers several other additives are added depending
on the conditions of operation. Antifungal and antimicrobial additives are common
to prevent bacterial growth in aqueous systems that can cause skin dermatitis. Mild
EP additives are also added for more severe operating conditions. Neat fluids, also
called oil-based fluids, have requirements for EP additives, and demulsifiers to
readily separate water. Oil mist is a problem in these operations, which is taken
care of by antimisting agents. The specifications closely control the corrosion,
rusting, and oxidation by laboratory tests. The tribological tests prescribed include
the well-known 4-Ball EP and antiwear tests, Timken OK load test, and Falex EP
test. These tests are governed by ASTM D2783, D2266, and D3233 respectively.
Depending on the application, some of these tests and the limits are specified.
However in view of their inadequacy many organisations rely on in-house tests.
The Falex torque test involving measurement of torque during tapping is an
additional test used by some organisations. Metal forming lubricants are a special
class and dealt quite extensively in two books [ 19,20]. The major problem in metal
forming is to ensure tribolgical performance. Normal tribological tests mentioned
above are used to screen products. These tests do not involve gross plastic
deformation and several tests involving plastic deformation have been developed.
These methods have limited applicability, as they cannot simulate varied metal
forming operations.
2.4.2.2 Hydraulic oils
Most of the modem hydraulic oils are antiwear oils containing mild EP additives.
They also have rust and oxidation inhibitors incorporated into the formulation.
Hydraulic oils are expected to last for a long duration and so an oxidation test,
ASTM D943 is incorporated. Oxygen is bubbled at controlled temperature in the
presence of water and catalysts in this long duration test. Acidity level after 1500
hours operation forms the test criterion. In the case of turbine oils, which shall be
discussed next, the acidity is tested after 2000 hours. The other tests include
demulsibility to ensure effective separation of water. The performance test is
normally conducted in piston or vane pumps and limits are laid down. The major
specifications are due to Denison, Vickers, and Cincinnati Milacron. It is of interest
to note that wear testing is conducted directly on the pumps used in the hydraulic
systems and the evaluations are expected to be more reliable. The oils are also
evaluated for air release property by ASTM D3427. The ability to quickly release
air from the system is important to operate the pumps without encountering the
cavitation problem.
2. 4. LUBRICANT SPECIFICATIONS 51
2.4.2.3 Turbine oils
Turbine oils are unique for their long life with about 10% topping per year. Many
steam turbines can operate for ten years without any oil change. Even gas turbines
can operate for three years without oil change. This is achieved by special base
stocks that have excellent water separability and good oxidation stability. Turbine
oils are divided into R&O and EP categories. The R&O category refers to the rust
and oxidation inhibited oil. It is fortified mainly with antioxidant and rust inhibitors
along with a foam inhibitor. Such oils are used for systems that do not involve
reduction gears. In cases where reduction gears are involved, turbine oils with mild
EP additives are used. When EP additives are involved a tribological test in the
FZG gear test as per DIN 5134 is specified. Another oxidation test, called rotary
bomb oxidation test (RBOT) as per ASTM D2272 is also sometimes specified.
RBOT is more often used to study residual life of turbine oils. It may be noted that
several viscosity grades are available for turbine and hydraulic oils and are
normally specified as per the ISO classification. Air release is also specified, as in
the case of hydraulic oils.
2.4.2.4 Industrial gear oils
The widely used specifications for gear oils are those specified by the American
Gear Manufacturers Association (AGMA). The gear oils are divided into nine
categories ranging from 5 to 50 cSt in viscosity at 40~ The corresponding ISO
grades range from 48 to 1000. The other known specifications are due to US Steel
and David Brown. The tests include oxidation stability, rust, foaming, and
demulsibility. Tribological tests are included for EP grade oils. The AGMA
specifies FZG and Timken OK load tests. US Steel includes also the 4-Ball EP and
antiwear tests. In view of the large varieties of gears used field performance cannot
be correlated with these rig tests. The tests are, however, useful in ensuring that the
lubricants have adequate EP and antiwear levels at least in the prescribed figs. The
FZG test is considered by industry as more reliable than the 4-Ball EP test.
2.5 Performance issues
Lubricant technology has advanced to a high level of sophistication as already

noted. High quality base oils obtained via hydrogenation and corresponding
developments in additives are helping to meet stringent specifications. Synthetic
fluids are competing for a place but they still form a minor share of the market. The
present section concerns itself with petroleum based lubricants. Despite careful
formulation, problems do arise in use. Bore polishing problems a decade back and
the more recent problem of black sludge in engines are examples of such problems.
Such problems are solved as they arise with cooperative effort between equipment
manufacturers, oil companies, and additive manufacturers. One serious gap that
remains is the inability to predict performance based on laboratory tests except in
some cases. This has implications for the quicker development of products, as at
present candidate products cannot be screened effectively in a shorter time. In
addition, it has implications with regard to customers consisting of both equipment
manufacturers and end users. Customers often want to select amongst competing
products all of which meet a given specification. Independent laboratories can do
this provided there are methodologies available. Another area of customer interest
is to know the rejection limits in service. While great strides have been made in
additive development and selection of base oils scientific understanding of
performance has lagged behind. The present section deals with the broader issues
of methodology. The problems involved are first illustrated with the example of
turbine oil which is a relatively simple formulation. The issues related to engine oil
performance are then considered. This is followed by a consideration of the
tribological problems related to lubricants.
2.5.1 Oxidation of turbine oils
Turbine oils are evaluated to ensure oxidation stability by ASTM D943, in which
oxidation is conducted at 95~ in the presence of copper and iron catalysts and
water. Oxygen is bubbled through the oil at a prescribed flow rate. The acidity
developed over a long duration of 2000 hours forms the criterion and is normally
limited to 2.0 mg equivalent of K(OH). Excellent water separabilty and good air
release properties are also required. Even for this simple system, the oxidation
stability cannot be predicted based on the composition of the base oil. The main
problem is in understanding the influence of different constituents in the base oil. It
is known that polynuclear aromatic structures, as well as complex sulphur and
nitrogen compounds, have a significant influence and these compounds cannot be
characterised in detail. Also the combined effect of these compounds is difficult to
predict. The complex problem has recently been studied by neural network
simulations, which related oxidation to seven likely constituents of the base oil
with a given additive system and different base stocks [21 ]. The purpose in citing
this example is to show that knowledge levels are not adequate for the prediction of
oxidation even for simple systems. The 'vagaries' of the base stock are likely to be
less for the hydroprocessed base stocks of type III described earlier which contain
very low aromatics, sulphur, and nitrogen. It is now demonstrated that such oils,
2. 5. PERFORMANCE ISSUES 53
with the usual levels of additives, can have oxidation life 2 to 3 times higher than
the conventional oil formulations [22].
Fundamental oxidation studies on hydrocarbons are well documented and Igarashi

et al [23,24] have used the available information on rate constants to develop an
effective computer simulation model to predict oxidation behaviour of hexadecane
in the presence of di-tertiarybutyl paracresol (DBPC), which is a commonly used
antioxidant in turbine oils. Their model also took into account the direct oxidation
of DBPC which is an important consideration. They then tried to treat turbine oil
oxidation in terms of a mixture of hexadecane and tetralin, which are well-defined
hydrocarbons amenable to kinetic treatment. Such extension to turbine oils is open
to debate, but the oxidation modelling for pure hydrocarbons may be adaptable to
pure synthetics like poly alpha olefins.
Another interesting aspect of turbine oils is the application of the concept of

residual life applied successfully by industry. ASTM D4378-84a governs this
practice. The RBOT test that determines oxidation life based on pressure drop in an
oxygen bomb is used for this purpose. The procedure is as per ASTM D2272. The
life is measured in minutes rather than hours and hence is quick to carry out. The
field samples are taken at fixed intervals, say, every 5000 hours and then subjected
to this test. The life is compared to that of the fresh oil and oil rejection is fixed at
30 to 40% of the original oil. At any stage, the remaining life can also be assessed
assuming an exponential change with time. The methodology followed taking into
account the topping rate has been described [25]. Exponential decay implies
pseudo first order kinetics, which is a reasonable assumption. The test proce,.d.ddureis
important from the methodology point of view. This is an example in which a real
system is evaluated periodically by means of a laboratory technique and correlated
with practice. Besides oxidation, air release, foaming, and other tests are also done
and should be within acceptable limits as laid down. The oxidation test is selected
because of its predictive capability of the real system.
Steam and gas turbine oils are undergoing changes and more effective additives
like amines are being used increasingly with or without DBPC. Base stocks are
improving significantly and these systems, along with rust inhibitors and other
additives may have to be reexamined for interactions, if any. Also the existing
methods to evaluate oxidation stability may or may not be adequate. Thus the
evaluation methods need continuous updating. The difficulties involved in
fundamental modelling make this inevitable at least in the near future.
2.5.2 Evaluation of engine oils
Simulation has been successful in one area of engine oils. This relates to the
performance aspects related to viscosity. Cold cranking as well as pumping
behaviour can be well assessed in the laboratory using rotary viscometers. Such
simulations involved development over a long period of time. Similarly, the high
temperature, high shear, viscosity specifications are based on the realisation that
viscosity requirements in bearings are governed by effective viscosity at the
relevant shear rates rather than kinematic viscosity. If the viscosity changes during
operation are not large, the information obtained with fresh lubricants is adequate
for prediction. Deeper understanding of rheology is also of relevance to the energy
losses and it is now possible to at least predict viscous losses from fundamental
considerations [26]. Boundary friction losses are more difficult to predict though
some attempts are being made with laboratory friction machines. Simulations are
successful in this area because the physical property involved, and its variations,
can be predicted on a fundamental basis.
Performance characteristics that depend on chemical changes like sludge, carbon

deposits, and wear are now being assessed through engine tests due to the inability
to simulate this complex situation in the laboratory. Different additive-additive
interactions have been summarized by Spikes [27]. When the interaction with the
base oils and the environment are also taken into account, the mechanisms become
far more involved. Major issues related to the state of art in engine oils and other
automotive lubricants are well covered in a book edited by Bartz [28]. Attempts are
continuing to develop better laboratory techniques for simulation. One recent
example is the evaluation by D4742 for gasoline engine oils. The oxidation is
conducted under oxygen pressure in a bomb in the presence of water, catalyst, and
oxidised fuel. The induction time related to a sharp pressure drop forms the test
criterion. It is of interest to observe that 'partial simulation' is attempted with fuel
preoxidised in the presence of NO2 as these components come in contact with the
oil in engines via blow-by. The method states that the technique is able to correlate
with viscosity increase in engine sequence tests but cautions that the test is not a
substitute for engine tests. Such tests may be able to screen formulations to a
limited extent. Laboratory test methods of various kinds have been developed over
several decades and will not be gone into here. Most of the tests are not successful
and have limited possibilities.
It is no exaggeration to state that the effective simulation for engine oils is the
engine itself. With this real situation, can there be any other approach to screening?
One possibility is the development of laboratory tests centred around the engine
2.5. PERFORMANCE ISSUES 55
tests. During engine tests, samples can be withdrawn at different time intervals and
tested in a known or developed laboratory test. The trend observed should be
capable of predicting the final test result in the engine. This is based on the
reasonable assumption that precursors to deterioration, as well as deteriorated
products, accumulate with time. Thus in an engine test run for 200 hours to rate
deposits, a laboratory test with samples drawn say every 2 hours may adequately
predict trends in 20 hours. The test methods may have to be different and should
not be restricted to 'standard' tests. This approach coupled with detailed chemical
analysis, can provide better fundamental understanding since real changes are
being monitored. If successful, they provide opportunity to screen formulations
with shorter duration engine tests. In a simpler system like turbine oils trends in
oxidation are predictable by evaluating field samples periodically as stated earlier.
It is considered that this approach is worth considering for complex systems as
well. One recent paper by Kollmann et al [29] is of relevance here. In this paper the
authors emphasise the fact that oil drain interval is a function of operating
conditions and proposed a monitoring system based on electrical conductivity. The
monitoring can be done on-line and provides information to the user regarding oil
rejection. The methodology is of importance as this is based on the recognition that
oil deterioration can be predicted only on the basis of the changes in the engine
itself. Several investigations are reported in literature in which used oils at different
intervals are analysed. However, the suggestion here is to use samples for trend
analysis via laboratory tests.
2.5.3 Tribological implications

Many tribological contacts operate in the mixed and boundary regimes leading to
friction and wear in the contacts. Friction can be reduced to some extent by special
additives called friction modifiers. The aspect of fluid film friction is not being
considered here. While boundary friction control is useful the major problem of
boundary contacts is wear which needs to be controlled. The concepts involved
will be discussed in chapters 4 and 5. Wear control in lubricated contacts is an
integral part of the overall lubricant performance. In many engineering situations,
wear control is accomplished by antiwear additives. These additives are effective
only if their interaction with the surfaces leads to wear resistant films. In complex
systems, surface interactions can occur with other additives, as well as oxidised
products, interfering with the antiwear performance. The other problem is that wear
testing is in many cases carried out in standard test machines, which simulate
neither the contact conditions nor the materials. Thus, more often than not there is
no correlation between laboratory test and the real wear in a component. One
example is a study of engine oil oxidised to different levels and tested for wear in a
cam-and-tappet rig and a 4-Ball wear tester [30]. No correlation was observed
between the two. In another example the influence of dithiophosphtes containing
different metals and side chains has been evaluated by four different tribological
tests [31 ]. The products were rated for performance in the 4-Ball EP and antiwear
testers as well as in an FZG rig for wear and EP properties. The authors found
different ratings with different machines. They also found that the ratings with a
given machine itself can change depending on the criterion used. These examples
show that characterizing wear and the associated problem of load carrying capacity
even for fresh lubricants is ill defined. This raises questions with regard to practical
use of large amount of work on EP and antiwear additives simply rated by 4-Ball
machines. Fundamental studies done on the reaction mechanisms concentrate on
composition of chemical films and their thickness. The link between film
composition and wear is not yet clear. Understanding of additive-additive and
additive-base oil interactions with regard to these additives in complex systems is
not yet available. The mechanisms involved in running-in are yet poorly defined.
Under these circumstances, it is necessary to look comprehensively into wear and
related problems with an aim to improve practice. Besides lubricants many other
parameters like materials, surface roughness, and oxide films affect wear. It is
hence important to address all these problems and bring theory closer to practice.
This is the main objective of the present book.
2.6 Synthetic lubricants
Synthetic lubricants may be defined as those fluids, which are chemically

synthesised unlike mineral oils obtained from crude. Synthetic lubricants account
for about 2% of the global lubricant market which is growing. They are used in
many advanced applications like aviation and space where mineral oils cannot
function. Synthetics are also being used in the more traditional applications like
engine oils. This section considers the categories of lubricants and their
application.
2.6.1 Types of synthetic lubricants

A state of the art consideration of synthetic lubricants has been presented in a
recent book [32], which considers both applications and chemistry. Many types of
synthetic lubricants are now available and each type has advantages and
disadvantages in comparison to a mineral oil. A good comparison of synthetic
fluids is given by Bartz [33] with a 1 to 5 scale of decreasing performance. This
information is reproduced in Table 2.3. The comparison is directional and provides
2.5. PERFORMANCE ISSUES 57
an approximate rating. It shows that all synthetics have some desirable and
undesirable properties and the selection is based on the specific need and the
expected properties. It can be seen that perfluoroalkyl ethers are outstanding with
regard to oxidation and thermal stability, but deficient with regard to corrosion,
compatibility with mineral oils, and biodegradability. In addition, the cost of the
fluid can be 500 times that of mineral oil. Less expensive fluids like diesters, and
polyalpha olefins (PAO) cost 3-5 times the mineral oil and are increasingly being
used for automotive and industrial applications. The structures of commonly used
fluids are given in Fig. 2.5.
2.6.2 Applications
Formulated lubricants based on PAO and diesters are finding increasing use in
engine oils, particularly in Europe. The formulations may be based fully on the
synthetic fluid or partially mixed with the mineral base fluid. PAO, as can be seen
from Table 2.3, has very good VI that normally exceeds 130, excellent low
temperature properties, and very good thermal and oxidation stability. The
requirements for VI, HTHS viscosity, and low temperature properties for engine
oils are already considered. PAO is superior in all these properties and can reduce
or even eliminate VI improvers. Another aspect is the significantly lower volatility
of PAO in comparison to mineral oils of similar viscosity. Volatility is of
increasing concern in Europe to reduce pollution and oil consumption. Some
engine specifications include the volatility levels, as per the DIN 5158 NOACK
method. PAO is thus well suited for this purpose. It also has superior oxidation
stability and this contributes to the longer life of the oil. Another advantage is that
PAO, being a hydrocarbon is completely miscible with mineral oil and can be
partially blended with it. Diesters, termed as dicarboxylic acid esters in the table,
are also used and are comparable to PAO. They are deficient in hydrolytic stability
and seal compatibility and these properties have to be taken into account during
use. Biodegradability of diesters is superior to PAO and mineral oils and hence
they are finding increasing use in 2-Stroke engines where oil is premixed with the
fuel. PAO and diesters are also making in-roads into industrial lubricants. A recent
unique example is the successful replacement of water based cutting fluids by neat
diester fluids by Dailmer-Benz [34]. Despite the high initial cost, the overall
benefit in terms of pollution control and work quality result in a better cost to
benefit ratio. Several known applications and new possibilities exist for various
synthetics. These include fire resistant hydraulic fluids, special compressor oils,
and gear oils. Polyalkylene glycols, polyisobutylenes and esters are being more
widely used in metal cutting and forming operations because of their low staining
58 CHAPTER 2. L U B R I C A N T T E C H N O L O G Y - A SURVEY
Rapseed oils
PAMA/PAO-
Cooligomers
Alkylated cy-
clopentanes
Dialkylcarb-
G onatcs
Cyclophosp-
I...,
hazene fluids . . . . . . I ~~
Chlorofluor-
C3 ocarbons r~
Silahydroc-
arbons
0 Silicate
>., esters
e.)
Silicone oils
::3
0
o
p.,,
phosphate
u'3 esters
Triaryi o
f~ phosphate 9- -- v~
g2x.
esters
f~ N~lyl ._ ,---
L__J ~o~
Dic~oxflk
acid ~ler~
o ~o~~
f.r..
9
P~fluon~l-
v~
kyledz~rs
Po~~e
s~.ots.
~rorrm~
o
Polyalpha-
eq
okf'ms
q.4 Polyisob~
o tc~cs
o Mineral oils . . . . . ,, . . . . . I . . . . I
L', ~"
-_ "=
o 4J
'i
o
(0
f'4
el
2.6. SYNTHETIC LUBRICANTS 59
Class TYl;~II Ilrudur=l Iormula

S ~ hydrocarbon (POlyt~tene) {--CH =--CHt--CH= --CH =--).
I I
Chlorolluorocadxm --C.-C--
I I
F F ,,
Diester CIH ,;,- O --CO--Call ,I---CO--O--~.H ,r

CH zOOC--CIH it
I
Neopentyt polyol ester CH =--CHz --C -- OOC--G, H,
I
CH~OOC~C=IH 1~
0
I!
Fstly acid ester C 13Hz?--OC--C I=H=7
CH,
i I
Polyglycd e4her HO( CH2 ....CH-O--)~H
Fluo~oester F(CF2),CEI2OOC(CF~).F
Phosphate ~ter (CH=---CIH 4--O) 3P==O
Silicate e=ler Si~O-OeH,~,),
04119 C4Hit
1 1
O O
I I
DLsiloxane C, H ~--O~S~--O--S,--O--C 4H ~
I I
O O
I I
C, H0 C4H9
CH.,r CH3"I CHj
I / I / I
Silicono CH,~--Si--I O - - S i ~ I O--Si- CH
9 I / I / I
CH, L CH~J,, CH~
Silane (C ,,~H,-s)S;i(C4H 1:=)3
Poiypl~enyl elner
Perl'iuorOalkyl {X;~ether F-- (?---?--)O--?-Cr,

\CIF 3 F / / F
Fig. 2.5. Typical structures of major synthetic fluids.

(From Ref. [ 18] p 169 )
60 CHAPTER 2. LUBRICANT 1ECHNOLOGY- A SURI/TY
and clean burn-off characteristics. Another important application is for aviation gas
turbines. Polyol esters are the main lubricants meeting the stringent DEngRD 2497
specification. These lubricants are manufactured only by a few companies in the
world. The specification is rigorous with regard to oxidation stability, deposits and
performance in gear rigs. Other fluids mentioned in the table are less used due to
cost as well as application problems. Silicones are very stable but have poor
lubrication properties. Silicone based greases are used to a limited extent in space
and related applications. Silahydrocarbons are a relatively new class in which
carbon is replaced by silicon and attached to alkyl groups. Polyphenyl ethers, as
well as perfluoroalkyl ethers have potential application for very high temperatures
exceeding 300~ One recent application of perfluoro ether is its use in the
computer disc-head interface to prevent stiction [35]. The film applied is only 2-3
nanometers thick and is effective.
Tribological implications are directionally similar to those raised with regard to

mineral based lubricants. Synthetics that are polar generally have better tribological
properties. When additives are added, their compatibility with the base fluid is
important and in each case, this has to be ensured. One extreme example is silicone
fluid whose tribological properties cannot be improved with known additives.
Research is being conducted to develop special additives for silicone fluids.
2.7 Environmental issues
The lubricants used are disposed of in various ways. A survey made of the
European scene [36] shows that out of 2.6 million tons of reusable used lubricants,
1.1 million tons is unaccounted for, and may have been indiscriminately disposed
of f'mding its way into the environment. Similar problems exist in the US and other
countries. There are also situations in which loss into the environment is
unavoidable as in mobile out-board equipment, metalworking, and other
operations. The lubricants, with different degrees of deterioration, eventually end
up as pollutants in the ecosystem. The hazards involved have to be assessed. Such
pollution can be particularly harmful to aquatic life. The ecological problems are
very involved and many answers are partial though in the desirable direction.
Methods to evaluate hazardous chemicals are available. These methods are adapted
for lubricant evaluations with modifications. Three aspects to be considered are
toxicity, bioaccumulation, and biodegradability. Toxicity is evaluated on the basis
of lethal effect on selected organisms at a particular dosage within a prescribed
time period. The product is considered lethal if 50% of the population is killed in
the specified time period. The German DIN specifications are well known in this
area. These specifications describe procedures for mammalian, fish, and bacterial
2. 7. ENVIRONMENTAL ISSUES 61
toxicity. Fish toxicity is of particular importance to aquatic life and is determined

by DIN 38 412-15. Water hazard or WGK is based on the water endangering
number (WEN) that is referred to as WKZ in German specifications. The procedure
depends on finding an average WEN based on the individual WEN values for the
three toxicities referred to above. For fish and bacterial toxicity it is found as
follows:
WEN = (-log NEC in ppm/106 ppm)
where NEC refers to 'no effect concentration'. The WEN for mammalian toxicity
on the other hand is obtained by assigning a number based on lethal toxicity. The
hazard levels are then classified on the basis of the ranges of WEN values.
It may be noted here that these specifications are constantly evolving and the above
discussion only presents the central approach to quantify toxicity.
Bioaccumulation is the concentration of the product from water or food. This is

related to the metabolism and is assessed by bioconcentration factor.
Biodegradability refers to the ability of naturally occuring microorganisms to
degrade a product. This is the main aspect being studied with regard to lubricants.
Biodegradability is assessed by the CEC-L-33-A-93 test procedure. In this test a
defined amount of fluid is added to mineral media inoculated with microorganisms
from natural sources. The mixture is shaken in the dark for 21 days and degradation
measured by the loss of infrared bands. ASTM D5864 is based on a similar
approach though the criterion used to measure degradation is based on the
evolution of CO2. A value of 70% degradation in CEC test rates the material as
readily biodegradable.
Mineral oils are not considered toxic. But some chemical additives like zinc
dithiophosphtes and calcium alkaryl sulphonates are considered toxic and
substitutes for these additives are being developed. The issue of bioconcentration
has not been well studied with regard to lubricants. Biodegradability of mineral oils
is not high and many fluids fall below 70% limit. This has led to the development
of easily degradable vegetable oil based lubricants for some applications. Some
synthetic fluids like diesters also are easily biodegradable. The tribological
performance of these lubricants is also being studied [37,38]. Many countries are
encouraging ecofriendly lubricants through ecomarks like 'Blue Angel' in
Germany. Such marking demands very low toxicity and biodegradability of greater
than 80%. The market for eco-friendly lubricants is growing and several
formulations based on rapeseed oil, ester, and polyglycols are now available. The
major interest in these lubricants is at present in Germany, Austria, and

Switzerland. There is a significant effort in various countries with regard to testing
environmental effects and it is likely that global standards may emerge in the near
future. The discussion of environmental issues here has been limited to lubricants
and the issues of air pollution from engine exhaust emissions is not considered.
Lubricants also contribute to exhaust emissions due to evaporation. There is also
evidence that lubricants contribute to particulate emissions in diesel engines [39].
Another aspect of importance is the carcinogenicity of used engine oils. It is known
that polynuclear aromatics produced in the combustion process accumulate in the
engine oils [40]. These products are known carcinogens and can pose disposal
problems.
Centralised collection and disposal in an organised manner is encouraged in many

countries through legislation and awareness programmes. But still a large amount
of oil is disposed off indiscriminately. The oils collected in the organised sector are
utilised for heating and other purposes as well as re-refining. Nearly 20-30% of
used engine oils are re-refined and the base stocks obtained are blended with
additives and sold. Such re-refining is very desirable as the lubricant is effectively
recycled. The re-refining technology poses problems. The acid and clay treating
process, which is easy to implement for small-scale operations, has environmental
problems due to acid sludge and used clay disposal. In addition, modem lubricants
are drained after long usage and need stronger treatments resulting in lower yield of
base oil. Processes, which use solvent extraction and/or hydrotreatment, are
economical only for large-scale plants. It is hence not surprising that the acid clay
route is still an important process for re-refining in many countries. With ecological
concerns building up there is incentive to develop cost effective technology for
small-scale operations. In the past, several re-refining methods have been tried and
some of the ideas involved may be worth reconsidering. One can also speculate on
the possibility of the used oils being pre-treated and directly used in lube refineries
in deasphalting or other units. Used oil can form a small percentage of the existing
feed. Such possibilities need detailed assessment.
Biodegradable lubricants do have their limitations. Synthetic fluids are expensive

as discussed in the previous section. Oxidation stability of fatty oils is poor and
they can be used only at moderate temperature. In some countries, use of vegetable
oil is mandatory for some applications like chain saw lubrication. But large number
of industrial applications will continue to be based on mineral oils and the parallel
developments in collection and re-use technologies is essential.
63
References
1. R. N. Watkins, Petroleum Refinery Distillation, Gulf Publishing, Houston, 1973.

2. J. D. Seader, J. J. Siirola, and S. D. Barinicki, Petroleum and complex mixture
distillation, in Perry's Chemical Engineers' Handbook, 7th edition, McGraw-Hill, New
York, 1997, 13-85.
3. R. T. Aird and S. L. Forgham, The lubricating quality of aviation fuels, Wear, 18 (1971)
361.
4. R. A. Vere, Lubricity of aviation turbine fuel, SAE Paper No. 690667 (1969) 2237.
5. D. Cooper, Laboratory screening tests for low sulphur diesel fuel lubricity, Lub. Sci., 7
(1995) 133.
6. S. Ikejima, H. Sugi, T. Ichikawa, K. Taniguchi, and K. Saito, Study of diesel fuel
lubricity, JSAE Review, 18 (1997) 203.
7. A. Dorinson and K. C. Ludema, Mechanics and Chemistry in Lubrication, Tribology
series 9, Elsevier, 1985, Chapter 16.
8. American Petroleum Institute Technical Data Book- Petroleum Refining, API,
Washington, 1983.
9. M. R. Gray, Upgrading Petroleum Residues and Heavy Oils, Marcel Dekker, New York,
1994.
10. A. G. Bridge, Hydrogen processing, in R. A. Meyers (ed.), Handbook of Modern
Petroleum Technology, 2"d edition, McGraw-Hill, New York, 1997, Chapter 14.1.
11. A. Sequeira. Jr, Lubricant Base Oil and Wax processing, Marcel Dekker, New York,
1994.
12. F. Buyan, Mobil hydrodesulphurisation process for distillates, Proc. Mobil Technology
Seminar, Tata McGraw-Hill, New Delhi, 1995, 173.
13. Methods of Test for Petroleum Products and Lubricants, 5.01 to 5.04, American Society
for testing Materials, 1998.
14. C. V. Smalher and R. K. Smith, Lubricant Additives, The Leizus-Hiles Co, Cleveland,
1967.
15. M. W. Ranney, Lubricant Additives- Recent Developments, Noyes Data Corporation,
New Jersey, 1978.
16. E. R. Booser (ed.), CRC Handbook of Lubrication, Vols. I and II, CRC Press, 1983.
17. R. M. Mortier and S. T. Orszulk (eds.), Chemistry and Technology of Lubricants,
Blackie, Glasgow, 1992.
18. E. R. Braithwaite (ed.), Lubrication and Lubricants, Elsevier, Amsterdam, 1967.
19. E. S. Nachtman and S. Kalipakjian (eds.), Lubricants and Lubrication in Metalworking
Operations, Marcel Dekker, New York, 1985.
20. J. P. Byers (ed.), Metal Working Fluids, Marcel Dekker, New York, 1994.
21. B. Basu, D. Saxena, V. Kaul, M.I.S. Sastry, and R. T. Mokken, Prediction of oxidation
stability of inhibited base oils using an artificial neural network (ANN), Lub. Sci., 10
(1998) 121.
22. D. C. Kramer, B. K. Lok, and R. R. Krug, The evolution of base oil technology, in W.
R. Herguth and T. H. Wame (eds.), Turbine Lubrication in the 21 st Century, ASTM
STP 1407, ASTM, 2001.
23. J. Igarashi and T. Yoshida, Computer simulation of turbine oil oxidation 1"
consumption of a hindered phenol antioxidant in model hydrocarbon systems at
115~ Lub. Sci., 7 (1994) 3.
24. J. Igarashi and T. Yoshida, Computer simulation of turbine oil oxidation 2:
consumption of a hindered phenol antioxidant at 115~ Lub. Sci. 7 (1995) 107.
25. M. J. Denherder and P. C. Vienna, Control of turbine oil degradation during use, Lub.
Eng., (1981) 67.
26. C. Bovington, V. Anghel, and, H. A. Spikes, Predicting sequence VI and sequence VIA
fuel economy from laboratory bench tests, SAE Paper No. 972925 (1997).
27. H. A. Spikes, Additive- additive and additive - surface interactions in lubrication, Lub.
Sci., 2 (1989) 3.
28. W. J. Bartz (ed.), Engine Oils and Automotive Lubrication, Expert Verlag GmbH,
Berlin, 1993.
29. K. Kollman, T. GiJrtier, K. Land, W. Warnecke, and H. D. Mtieller, Extended oil drain
intervals- conservation of resources or reduction of engine life (Part II), SAE Paper
No. 981443 (1998) 728.
30. G. Monteil, A. M. Merillon, J. Lonchampt, and C. Roques-Carmes, Evaluation of the
antiwear performance of the aged oils through tribological and physicochemical tests,
Lub. Sci., 4 (1992) 155.
31. M. Born, J. C. Hipeaux, P. Marchand, and G. Parc, The relationship between chemical
structure and effectiveness of some metallic dialkyl and diaryl dithiophosphates in
different lubricated mechanisms, Lub. Sci., 4 (1992) 93.
32. R. L. Shubkin (ed.), Synthetic Lubricants and High-Performance Functional Fluids,
Marcel Dekker, New York, 1993.
33. W. Bartz, Comparison of synthetic fluids, in R. L. Shubkin (ed.), Synthetic Lubricants
and High-Performance Functional Fluids, Marcel Dekker, New York, 1993, Chapter
14.
34. Theo Mang, Future trends in machine tool lubrication, in D. Basu and S. P. Srivastava
(eds.), Proc. Intl. Sym. on Fuels and Lubricants, Tata McGraw-Hill, New Delhi, 1997,
49.
35. B. Bhushan and Z. Zhao, Macroscale and microscale tribological studies of molecularly
thick boundary layers of perfluoropolyether lubricants for magnetic thin film rigid
disks, J. Info. Storage Proc. Syst., 1(1999) 1.
36. CONCAWE, Report No. 5, Collection and disposal of used lubricating oil, 1996.
REFERENCES 65
37. B.- R. H6hn, K. Michaelis, and R. D. DiSbereiner, Load carrying capacity properties of
fast biodegradable gear lubricants, Lub. Eng., 55 (1999) 15.
38. W. Belluco and L. De Chiffre, Testing of vegetable-based cutting fluids by hole making
operations, Lub. Eng., 57 (2001) 12.
39. P. Tritthart, F. Ruhri, and W. Cartellieri, The contribution of the lube oil to particulate
emission of heavy duty diesel engines, in W. J. Bartz (ed.), Engine Oils and
Automotive Lubrication, Expert Verlag GmbH, Berlin, 1993, Section 4.8.
40. P. Van Donkelaar, Environmental effects of crankcase and mixed lubrication, in W. J.
Bartz (ed.), Engine Oils and Automotive Lubrication, Expert Verlag GmbH, Berlin,
1993, Section 4.11.
67
3. Dry wear mechanisms and modelling
3.1 Introduction
Wear occurs in tribological components due to several processes that result in the
loss of material. The progressive loss of material leads to dimensional changes and
eventual loss of performance. Replacement of worn components involves the cost
of the new components as well as expenses involved in maintenance and down
time. There is an obvious economic advantage in reducing wear and prolonging the
life of components. Understanding of wear mechanisms and the ability to predict
wear by appropriate models can lead to better control of wear. With this aim in the
background a large amount of fundamental and applied research is being conducted
in this area. The emphasis has been on dry wear of materials though lubricated
contacts are industrially far more common. Several specialized textbooks are
available dealing exclusively with wear of materials some of which may be cited
[1,2,3].
The generation and removal of wear particles is a complex process. The

mechanisms involved are conventionally classified into adhesive, abrasive, fatigue,
and oxidative wear. In many engineering situations two or more mechanisms may
be operative and it is difficult to model such situations. The separation into
categories is essential for understanding individual mechanisms. Controlled
laboratory studies can be conducted where one mechanism dominates leading to an
understanding of that mechanism. Here again the presumed mechanism cannot be
assured leading to uncertainty in the understanding. Despite this scenario a broad
understanding of mechanisms i.s possible. The purpose of this chapter is to provide
the necessary background on dry wear of materials. Dry wear in the present context
refers to situations where no external lubricant is applied. The coverage is limited
to an appreciation of the main concepts involved. Some of the mechanisms will be
elaborated in later chapters.
The first two sections deal with contact stresses and asperity temperatures that are
of importance both for dry and lubricated wear. Adhesive wear phenomena are
then considered followed by a consideration of the available models. Abrasive
wear is then covered in a similar manner. As fatigue wear will be covered later in a
68 CHAPTER 3. DRY WEAR MECHANISMS AND MODELLING
separate chapter the coverage is limited to a few general observations. This is

followed by a consideration of oxidative wear. The final section deals with wear
maps. All these sections will deal only with metallic materials. Dry and lubricated
wear of non-metallic materials is covered in a separate chapter.
3.2 Contact stresses
Contact stresses and their distribution is of importance in tribology. This section

provides basic formulae for stress distributions in concentrated point and line
contacts. Concentrated contacts are also referred to as Hertzian contacts. Detailed
consideration of contact mechanics is not within the scope of the book and the
interested reader can consult specialised textbooks [4,5] in the area.
3.2.1 Surface Stresses
Concentrated contacts that are non-conformal are characterised by high stresses.

Point contact and line contact are the basic contacts considered here. The results
presented here are for elastic contacts and smooth surfaces. The point and line
contacts are illustrated in Fig. 3.1.
toad P
Ip
(Q) I (c)
%
Po
I_ * _l_ * _1
<b~ I (d)
Fig. 3.1. (a) Spheres in elastic contact and (b) semi-elliptical pressure distribution; (c)
cylinders in parallel contact and (d) semi-elliptical pressure distribution.
3.2. CONTACT STRESSES 69
Under load the surfaces deform elastically with typical pressure distributions as
shown in this diagram. The loading in both cases is static. For two spheres of radii
R l and R 2 in contact under a load P
'/3
contact radius, a = 4E
I3PRI (3.1)
where v I and V 2 are the Poisson ratios of the two materials and E t and E 2 are
the elastic modulus of the two materials.
2 l_V2 2 1 1 1
Also, --1 _ 1 - V l +- and . . . . + ....
E E 1 E 2 R Rl R 2
The elliptical pressure distribution is
p(r) = po 1 - (3.2)
where r is any radius within the contact circle
The maximum pressure p o occurs on the axis of symmetry and is

2P
P0 - - - m
7[a
2
The mean pressure p m -- ~ P0
3
2
The contact deformation 8 = a

R
With these formulae the equations (1.9a) and (1.9b) in chapter 1 for asperity
deformation can be easily derived.
For two parallel cylinders in line contact with load P now expressed per unit length
I 2 P R ] /2
half contact width b = rtE I (3.3)
I r~211/2
pressure distribution p(x) = po 1 - (3.4)
3P
also peak pressure p o - 2 rt a 2
rtp0
and mean pressure p m =
4
central deformation 15=

E1 E2
Formulae are available in standard texts to calculate stress distributions with

arbitrary contact geometries and are not given here. Stress distributions at any point
in the sub-surface can also be obtained by the available methods [4,5]. With real
surfaces the contacts are at the asperities and this will induce changes in stress
distribution as compared to smooth surfaces [6,7]. The overall stress distribution at
the surface is not significantly affected for roughness levels commonly used in
practice though stresses at contact spots will be higher. Sub-surface stresses in the
zone close to the surface are more strongly influenced by local asperity stresses.
3.2.2 Sub-surface stresses
The point of maximum shear stress and its magnitude are of particular importance.
For point contact rmax-- 0.31 PO and occurs at a depth of 0.48a for steel surfaces on
the axis of symmetry. In the case of line contact rmax = 0.30P0 and occurs at a
depth of 0.78b for steel surfaces on the axis of symmetry. These values refer to
situations where only normal stresses are acting. In many contacts the maximum
Hertzian stress is higher than the yield stress and the plastic deformation is initiated
at the point of maximum shear stress. This stress is of importance with regard to
the initiation of sub-surface cracks.
3.2.3 Frictional traction
When frictional (tangential) force is applied the stress distributions in the contact
change. The stresses at any point are obtained by superposing the stresses due to
normal and tangential forces provided the Poisson ratios of the two surfaces do not
differ significantly. This is normally the case for metallic contacts. Tangential
stresses influence the maximum shear stress and its location. The point of
3.2. CONTACT STt~SSES 71
maximum shear stress moves towards the surface as frictional force increases and
can reach the surface when the friction coefficientfexceeds 0.3. Another important
influence of tangential force is the generation of large tensile stresses on the
surface. In the case of line contact the maximum tensile stress occurs at the trailing
edge of the contact. The trailing edge is with reference to the direction of tangential
force. The value is equal to 2 f P0 and is hence substantial. Such stresses can
induce surface cracks in the material and contribute to fatigue wear. For point
contacts the maximum tensile stress occurs over the trailing edge of the contact
periphery. This stress is given by k/p 0 where
k/ = n f ( 4 + v ) ( 1+- 2 v ) (3.6)
8 3
where v is Poisson ratio
Here again the tensile stresses can be substantial.
The above treatment is limited and several aspects of contact mechanics are left
out. It is hoped this limited treatment provides an adequate background to
appreciate stress related problems in tribology. Some aspects will be elaborated
upon where necessary in the text. The SI units are commonly used in calculations.
3.3 Asperity temperatures

The problem of surface temperature rise has been considered in section 1.4. The
temperature rise was treated on the basis of heat distributed over the geometric
contact area. This temperature rise is now designated as A0g. Generation of heat is
actually at the asperities. It is of interest to estimate the asperity temperature rise
A0~. The asperity temperature rise can be significantly higher than A0g. The high
localised temperature rise at asperities affects chemical reactions, material
transformations, and several related phenomena. The average asperity temperature
rise is high because the heat flux generated over the real area is much higher in
comparison to that based on the geometric area. Several approaches with different
levels of sophistication are available to deal with the problem. The commonly used
simpler approach is adopted here to calculate these temperatures. The first step is to
treat the temperature rise at the asperities neglecting the rise of temperature over
the geometric area. This is defined as a case with no thermal interaction. The next
step is to deal with the situation of thermal interaction leading to more realistic
estimation of temperatures. These two cases are considered below.
3.3.1 Asperity temperatures without thermal interaction
The model for contact considered here is based on plastic flow at the asperities.
The real contacts can be entirely elastic in which case the areas and heat
distribution get modified. The real area will have contact spots of varying
diameters and it is usual to assume an average asperity contact diameter. Firstly
since real area is known on the basis of load and hardness heat flux is calculated for
the real area. This heat flux is flowing through the asperities. Since the average
diameter of the asperity is known (or assumed) the Peclet number for the asperity
can be calculated. Then the same equations that govern the temperature rise as
discussed in chapter 1 are used to estimate asperity temperature rise. Since the
thermal interaction is neglected the asperity temperature is obtained by adding the
bulk temperature to the estimated temperature. The estimation is illustrated by an
example. It may be noted that surfaces I and II in section 1.4 will now be referred
to as 1 and 2 wherever applicable. For convenience the case of steel on steel for
which overall temperature rise was calculated in section 1.4 is reconsidered here.
The contact is a square contact with a geometric area of 1.81x10 5 m 2 with half
contact width equal to 2.127x10-3m. The load is 100 N and the sliding speed is
5.11 m/s. For consistency with the previous example calculation is done with a
square asperity. In the present example the contact occurs at real area spots
distributed within the geometric square area. The parameters related to the
asperities carry the subscript 'a' in the following text.
Assume hardness of 3.26xl 09 (N/m 2)
For the load of l00 N the real area A r = Load = 3.067x10 -8 (m 2)

Hardness
The calculated heat flux on the basis of real area, qa = 6.6xl 09 (W/m 2 )
This value may be compared with q value of 1.12xl 07 based on the geometric
area.
Assume an average square asperity area of 100 lam 2. The half contact dimension la
for the asperity is 5x10 -6 m. The Peclet number Pea for asperity contact is now
calculated to be 1.09. This may be compared with the large Peclet number of 463.7
on the basis of the geometric contact dimension. This is because the asperity
dimension is much smaller.
3.3. ASPERITYTEMPERATURES 73
Since qa and Pea are known the temperature rise at the asperity may be obtained as
was done for geometric contact area. In this case the graphical procedure based on
Fig. 1.5 is used since Pr is between 0.5 and 5.0. The value obtained for the
asperity temperature rise A0~ is 256~ Heat is generated at all the asperities in
contact and the temperature rise is same for all of them provided they all have the
same dimensions. This temperature rise is much higher than the overall surface
temperature rise calculated earlier as 18.1~ This calculation shows that high
asperity temperatures can easily arise in the contact. It may also be noted that for a
given heat flux q~ temperature rise increases with the asperity diameter. For
example if the average asperity l,, is 20 /xm instead of 10 /zm the asperity
temperature rise is calculated as 436~ In the process of running-in where some
misalignment is possible at the micro level, some asperity contacts may be
particularly large. Such asperities may be critical as failure at these contacts can
trigger a run away situation that can lead to problems like scuffing. Such problems
may be analysed better in terms of critical asperity events rather than 'average'
behaviour in the contact. While this simple analysis.neglects thermal interaction the
temperature may be considered as a lower bound temperature.
In some cases the overall heat generation may be small enough to justify using this
approach. For example if the above contact is lubricated the temperature rise at
asperities will be much smaller. I f f in this boundary lubricated contact is 0.1, the
temperature rise for a 10 /zm asperity will be around 64~ and the overall
temperature rise will also be low. For practical purposes the asperity temperature in
this case may be obtained by adding the bulk temperature to A0 a . The asperity
contacts are circular and an approximate estimation may be made substituting
contact radius for 1,. Tian and Kennedy [8] proposed useful equations for circular
contacts both for maximum and average temperatures. The advantage of these
equations is that they are generalised for any Peclet number and have a sound
theoretical basis. Their general equation for average temperature rise at one
surface, A0 is
0.61
A01,2 = 2qa x n ~ ~0/-.6575 + Pel )~~ (3.7)
k 1,2 ,2
where q is the heat flux and 'a' is contact radius and applicable for any Peclet
number P~. Subscripts 1, 2 refer to one surface or the other. Adapting this equation
for the present case of asperity contact and utilising Eq. (1.17)
A0 =2q~a~ i kn0_s~0.6575 +06;0

P~ + I0.65750.5 il (3.8)
It may be noted kt = k 2 in this example and is replaced by the general term k.

Calculation on this basis gives an average temperature rise of 254~ and is very
close to the value of 256~ estimated earlier for a square contact.
The above equation is for a plastic contact area at the asperity. Asperity contacts
can be elastic in many cases. For elastic contact the pressure distribution is
elliptical and not uniform as in the case of plastic contact. The approach in this case
is to obtain q~ on the basis of mean asperity pressure and not the hardness. For
example if mean asperity pressure P m is 0.4H then q~ will be 2.64 x 1 0 9 . The
general equation for elastic contact for the present case from the same reference
[8] is
A0,,2 = 2 q a [ 0.732 )0.5] (3.9)

kl,2n ~ (0.874 + Pel,z
Adapting this equation for the present case
A0 = 2qaaa
[krto.5{(0.874 +0.732
P~a )o.s + 0.874~
}1 (3.10)
The value calculated on this basis is 108~ and is substantially lower than 256~
for the plastic case. Well-run engineering surfaces will have mainly elastic contacts
and this is advantageous from temperature point of view. For elastic contact the
number of contacts and their area can be estimated from the contact theory given in
chapter 1. From this information average size and mean pressure at the asperities
can be obtained.
It is difficult to decide whether average rise or maximum rise governs the

tribological behaviour. It is also difficult to understand the relative importance of
overall temperature rise and asperity temperature rise. The temperature rise at the
contact is also referred to as flash temperature.
3.3. ASPERITY TEMPEt~ TURES 75
3.3.2 Thermal interaction
While the rise in asperity temperature can be estimated by the above procedure it is
necessary to find the actual asperity temperature. If the thermal interaction is
neglected this temperature is simply obtained by adding the bulk temperature to the
asperity rise. In many lubricated contacts with low heat generation this procedure is
adequate. In the case of dry contacts with high heat flux and no cooling by the
fluid, thermal interaction becomes more important. Thermal interaction refers to
the overall rise in the surface temperature due to the contribution from all the
asperities distributed over "the geometric contact area. This rise has to be
superimposed additionally to obtain the asperity temperature. The asperity
temperature 0a is then
0 a = A0 a + A0g + 0 b (3.11)
where
A0a = temperature rise due to asperity interaction alone
A0 g = temperature rise over the geometric area
0b = bulk temperature assumed here to be the same for both the
bodies
The main approach is to consider that the asperity temperature in the contact is the
same for the contacting asperities. The heat is partitioned between the two surfaces
in such a way that the following equality results
(A0 a -b A0g) body l = (A0a q- A0g) body2 (3.12)
The asperity temperature rise that now includes thermal interaction is designated as
A0,t and is equal to the sum expressed above. In this particular case the heat
flowing to the two bodies will be based on the partition of the heat flux q into the
two bodies. The problem is to find A0at which will be referred to as asperity
temperature rise with thermal interaction.
The approximate method is based on the previous approach. First it is considered

that all heat is flowing into body 1. The temperature rise at the asperity and
geometric area at the respective Peclet numbers is calculated and summed. This
gives A0m. Similarly A0at2 is obtained for the surface 2. The average temperature
rise A0atis then obtained using Eq. 1.17. In the following equations it is convenient
to express temperature rise for asperity as well as well as geometric area in terms
76 CHAPTER 3. DRY WF~4RMECHANISMS AND MODELLING
of heat flux based on real area. This is done by defining two non-dimensional
parameters as
l ~ - ~l~
- -- half width of asperity contact/half width of geometric contact
lg
A ~ - ~Ar -- real area/geometric area

Ag
This is now applied to the multiple contact situation as in the previous case but
with thermal interaction. Consider the whole heat flowing into body 2. The
temperature rise at the asperity contact will be
0.946qalgl* 0.946q lgA*

A0,t2 = + ~ (3.13)
k2 k2
This equation adds the temperature rise over geometric area to include the
interaction effect. The term qaA* represents the heat flux over the geometric area.
For body I assuming high speed equations are applicable with Peclet numbers for
asperity and geometric area greater than 5.0
0.75qalgl * Pea/2
I
0.75qalgA* p~l/2
m0at 1 = + (3.14)
kl kl
where PC refers to the Peclet number based on geometric area. Peclet numbers are
not distinguished by subscripts here as in the present case they are applicable to
one surface only.
Similar relation was proposed by Suh [9] for the high speed case with a correction.
This correction is to account for the fact that the overall heat generation should
exclude the heat generation at the asperity under consideration which is already
accounted for by the first term. Usually a large number of asperities are in contact
and the effect due to a single asperity is neglected in the above treatment. The
asperity temperature rise mOat that includes thermal interaction can be obtained
from
3.3. ASPERITY TEMPEtL4 TURES 77
1 1 1
A0at A0atl A0,,t2
The high speed equation above is of limited utility. Usually Pea will be much
smaller than Pe and so will not be governed by the high speed equation. Also Pe
can have any value depending on speed and contact dimension. If it is assumed
that the heat flow at the asperity and geometric area are governed by the respective
Peclet numbers, the temperature for a moving source may be expressed as
m0atl = asperity temperature rise at its Pea + geometric temperature rise at its Pe
The rest of the procedure is the same to estimate A0at. As an example A0 a t Was
estimated for square asperity with plastic contact dimension of ! = 5 /zm for the
same conditions as in previous example. The value estimated is 313~ that may be
compared with a value of 256~ obtained without thermal interaction. Thus
thermal interaction increased the temperature rise by 57~ The equations can be
adapted approximately for circular contacts replacing l by the radius a. From the
general equations cited earlier for circular contacts it is possible to adopt them for
the temperature evaluation and obtain better estimates. The asperity temperature is
obtained by adding the bulk temperature to A0at. Note that the treatment was
confined to a sliding situation in which one surface is stationary and only the Peclet
numbers for one moving source were applied. If both surfaces are moving Peclet
numbers are to be considered for both the surfaces.
In the present analysis the asperity temperature is equalised. Hence the geometric
area temperature rise for the two surfaces will not be the same. With regard to bulk
temperature it may be considered as same for both the surfaces for lubricated
contacts. Experimental evidence shows that in lubricated contacts the bulk
temperature may be nearly equal to the oil bath temperature [ 10]. The main reason
for this is the convective cooling due to the oil that tends to equalise the
temperatures. But the temperature rise in the contact itself can be found by the
same procedure using the f v a l u e for the lubricated case. The dry contact situation
is more complex and the bulk temperatures can be different for both surfaces and
the problem needs a complete heat transfer analysis [11]. Another factor is the
increase in the temperature of the sliding track over the bulk temperature which
also is to be taken into account. These details are not considered here. The
temperature estimates on the basis of the present analysis are to be considered
approximate. Complete thermal analysis is necessary to predict temperatures more
accurately.
78 ('HAPIER 3. DRY WEAR MECHANISMS AND MODELLING
3.4 A d h e s i v e w e a r
3.4.1 Phenomena
Adhesive wear occurs when there is mutual transfer of materials in a moving

contact. The process involves firstly transfer of the material from one surface to the
other at the interacting asperities. The particles can grow by further transfer in
repetitive contacts and eventually get removed as loose debris. There can also be
back transfer of material from one surface to the other. These phenomena were
well demonstrated in the earlier literature [12]. There is also evidence of
intermetallic diffusion based on the analysis of wear particles [13]. When transfer
occurs on a large scale leading to performance failure of the components the
situation is normally referred to as scuffing. Scuffing is due to large scale transfer
and its control is of importance in tribological contacts. The term 'adhesive wear'
is meaningful only for situations where progressive wear occurs. The transfer and
removal processes can reach a quasi equilibrium state. In such cases a steady wear
rate can be observed for the interacting surfaces for a given set of operating
conditions.
The transfer at the asperity junctions is considered to be due to intermetallic

bonding of the metals. The junction so formed can break in the softer material
leading to a transferred fragment. Some junctions can break in the harder material
as well leading to transfer of the harder material onto the softer material.
Transferred particles can also spread due to shearing resulting in patchy films.
When the interface of the junction is weaker than either of the metals shearing
occurs at the interface resulting in lower adhesive wear. Adhesion in a fundamental
sense occurs between any two contacting asperities. Adhesion need not necessarily
involve material transfer. Adhesive wear is observed only in those situations where
material transfer occurs.
Real surfaces are covered with oxides and any metal-metal bonding can only occur
through the oxides. Thus the nature of oxides and their adherence to surfaces
strongly influences adhesive wear. The steady wear process may involve removal
of transferred material as well as oxides and it is difficult to separate their relative
contributions. Thus when we refer to adhesive wear we are assuming it is the rate
controlling mechanism. Fig. 3.2 illustrates schematically adhesive wear mechanism
between surfaces covered with oxide films. When asperities deform plastically
larger adhesion is expected due to increased disruption of the oxide layer.
3. 4. A D H E S I V E WEAR 79
.,...........,~ ..,.,.,...~..........,.
odhesive
contacts
(a) (b)

p~rtic[es
Fig. 3.2. Illustration of adhesive transfer through oxide layers at asperity level" (a) approach
(b) contact and (c) disengagement.
3.4.2 Adhesive wear modelling
The complexity of adhesive wear is evident from the previous section. It is difficult
to model such a process and to predict wear quantitatively. An early model of
adhesive wear is due to Archard [14]. He considered that the wear originated at the
real contact spots. Assuming that the real area is plastic, and a hemispherical
particle is involved in the wear event, the equation may be derived as follows:
W
Real area A r = ~ where W is the load.
H
Assume average asperity contact diameter d. If a hemispherical wear particle is

produced at a contact over a sliding distance d its volume will be
1 /l:d3 (3.15a)
vd 12
Since the total number of asperities are equal to real area/individual asperity area it
can be shown that the total volume generated at all the asperity contacts over a
sliding distance d is
80 CHAPTER 3. DRY WFAR MECHANISMS AND MODELLING
1 Wd (3.15b)
Vd=3 H
If we now define wear rate V r as wear volume per unit sliding distance it follows
V =--Vd =---lW (m3/m) (3.15c)

r d 3H
The wear volume V is normally known over a sliding distance l. In these terms
ZVd V 1W
(3.15d)
V,- Zd -T=3-H--
The actual wear rates observed are far lower than those obtained from this
equation. To obtain realistic values Archard introduced a wear coefficient K and
his final equation is
W
V~ = K ~ (3.16)
H
K is a non-dimensional wear coefficient and can vary over a wide range of 10 -2 to

10 -5 depending on materials and operating conditions. Archard proposed that K
represents the probability of forming a wear particle in a given encounter. The
constant 1/3 in Eq.(3.15d) now gets included in K. The problem with this concept
is that there is no valid reason why only one event in a large number of encounters
should be successful in removing a particle. Also the wear coefficients can only be
obtained on the basis of experiments and cannot be obtained from any physical
model. On the basis of wear coefficients the wear may be considered mild or
severe. When wear coefficients are less than 10 -4 the wear may be called mild
while it may be referred to as severe when the coefficient is greater than 10 -3 .
Real surfaces are covered with oxide films. Adhesive wear in such situations can
occur only through oxide films and the actual metal contact will be a fraction c~ of
the real area. It is reasonable to assume the average diameter of the transfer particle
at an asperity contact is that of a circular patch of metal contact area. On this basis
diameter of the transfer particle is d,f~. Consider that all events result in a transfer
3.4. A DHESI VE WEAR 81
particle. As the wear particle is generated per event the sliding distance involved
remains d as in Archard's model. It can be shown that the wear rate is
--=5 (3.17)
l
where c~is fractional metal contact area within the real area
51.5
In other words may be considered the wear coefficient that results from the
3
extent of adhesive contact through oxides assuming all metallic contacts lead to a
transferred particle. The coefficient now has a physical basis. If for example, ot is
0.1, the wear rate in the present model is 1.05x10 -2 xA~ and so the wear coefficient
is 1.05x10 -2. Depending on ct it can have a similar range to that in Archard's
equation but has a different meaning. It may be noted that in this model every
metal contact leads to a particle and it is assumed that wear is governed by a
transfer mechanism alone. Here again the coefficient can only be determined
experimentally. This model is only a plausible suggestion to provide a physical
basis to adhesive wear coefficient.
Other models have looked at the wear coefficient in terms of contact mechanics.
Greenwood's model discussed in chapter 1 can be used to obtain the extent of
plastic and elastic contact based on roughness. If it is assumed that only plastic
contacts contribute to adhesive wear, wear coefficients can be estimated [ 15]. Kato
et al [ 16] also considered that adhesive wear occurs when the deformation is plastic
and modelled the transfer process. The elastic contacts are considered to wear by a
fatigue wear mechanism. In general terms wear due to asperity interaction may be
referred to as sliding wear. Sliding wear can be due to several mechanisms
including adhesion. The suggestion of the author is to consider adhesive wear as
due to overall fractional adhesive metal contact through oxide films. When asperity
deformation is plastic larger extent of metallic contact occurs as opposed to elastic
contact. Oxides also can wear by a chemical wear mechanism that will be
discussed later. Modelling on the basis of one mechanism or the other amounts to
an assumption that the presumed mechanism is rate determining. The problem is
difficult to resolve on the basis of available literature. Further complications that
can arise in adhesive wear situations include large-scale modifications in near
surface structures [17] debris compaction in the contact region [18] and work
hardening of the asperities that can lead to abrasive wear [19].
82 CHAPTER 3. DRY WEAR MECHANISMS" AND MODELLING
From the discussion in the previous paragraph it is clear that definitive modelling
of the wear process is not possible. The model due to Archard has practical utility
in terms of experimental observations. Many experiments do show that wear rates
are proportional to real area within some limits of operating conditions though the
detailed mechanisms involved are complex. It is appropriate to consider Archard's
model as an equation that describes sliding wear process in general with an
experimentally determined coefficient. The wear coefficient will have a specific
value for a given range of operating conditions. In another range of operating
conditions a different wear coefficient may characterise the wear behaviour. Such
variations arise due to several reasons that include near surface material
transformations and oxidation. The temperature rise in contact has an important
role to play in these modifications. Transitions of a sudden nature can occur when
operating conditions exceed certain limits. A typical example of transitions as
observed by Welsh [20] that is often quoted in the literature is shown schematically
in Fig. 3.3. This diagram illustrates load dependent transitions and recovery due to
changes in surface films and changes in hardness of the steel. Wear coefficients can
provide limited predictive capability at least in laboratory machines. Wear
coefficients are also useful in comparing wear behaviour of different material
combinations for screening purposes.
o1
o
._J
Load
Fig. 3.3 Load wear transitions observed with steel from Ref [20].
As stated earlier sliding wear can have many mechanisms and adhesive wear is one
of them. It is possible to study the mechanisms of sliding wear over a wide range of
operating conditions and to map the wear behaviour as discussed in the final
section.
3.5. ABRASIVE WEAR 83
3.5 A b r a s i v e w e a r
3.5.1 Phenomena
Abrasive wear is normally associated with material removal when a hard sharp
particle slides against a softer material. Grinding of material is a good example of
the process in which abrasive grains remove material from the softer surface by a
cutting action. In this mode the grain penetrates the surface and the grooved
volume is removed during sliding. Abrasive wear may be classified as two-body or
three-body. The two-body mode refers to a situation where one of the surfaces is
abrasive in relation to the other. Grinding with an abrasive paper is a typical
example of this. The three-body mode, as the name suggests, involves abrasion of
two surfaces in relative motion due to abrasive third body particles that are drawn
between surfaces. Dust particles that get into a lubricating oil circuit can act as
third bodies. Wear particles that work harden and circulate between the surfaces
can also act as third bodies.
The major portion of research on abrasive wear is with two bodies. This is because
the system is better defined in comparison to a three-body system in which the hard
particles are ill defined. Early research by Khruschev and Babichev [21]
established that for pure metals abrasive wear decreases as hardness increases.
Sundararajan [22] found that for pure metals the relative abrasive wear resistance
(RAWR) may be expressed as
RA~VR oc H ~ where H is the bulk hardness of the metal and RAWR is the
relative wear resistance compared to pure iron.
With different steels the change in RAWR with hardness is lower with the
exponent varying from 0.2 to 0.55 and is related to the type of steel and its
microstructure. This was explained by Sundararajan [23] on the basis of critical
strain needed for chip formation.
Cutting action should depend on the relative hardness of the abrasive to that of the
metal. Earlier work [3] has shown that abrasive wear becomes negligible when the
hardness ratio of the abrasive to the metal is less than the critical range of 1.6 to
1.0.
The above observations refer to situations where cutting wear is predominant. To

generalise abrasive action one has to consider particles of varying hardness and
shapes moving against a relatively softer surface. If a hard asperity slides over a
softer material it can remove material by cutting action or it may only plough a
groove without material removal depending on the operating conditions. There can
also be wedge formation in the contact region for operating conditions between
ploughing and cutting modes. In wedge formation a built up wedge is formed from
the softer material. In principle wear mechanisms involved in all these modes may
be classified under abrasive wear. The three modes are illustrated in Fig. 3.4 which
is based on [24]. It may be noted that plastic flow results in ridges. The ratio 3
defined as shown in the figure is indicative of the cutting action of the abrasive. In
cutting the volume of ridge material is much smaller than the groove volume
resulting in [3 = 1 for ideal cutting. On the other hand for ideal ploughing/3 = 0.
3.5.2 Abrasive wear modelling
The elementary model considers a conical indentor grooving through a softer

surface. This is illustrated in Fig. 3.5. A triangular cross section is involved and for
ideal cutting the wear volume V will be based on the grooved volume generated
over distance l and is given by
V = d21 tan0 (3.18a)

where
- depth of penetration
- distance travelled
- semi-included angle of the cone
Assuming the real area is governed by the hardness and noting that the real area
will be half the projected area during motion real contact area is
(d tan0)2 _- ___W (3.18b)

2 H
Combining the above two equations (a), and (b)
n ~n0 ~
All the grooved material need not be removed as wear volume. This can happen
due to several reasons, like repetitive contact in the same groove that changes the
contact conditions and the hardness, and abrasive particle size distribution. Ridges
3.5. ABRASIVE WEAR 85
as discussed earlier also influence the wear process. Once again there is a need to
introduce a wear coefficient Kab as has been done by Archard [25] and the
abrasive wear rate may now be expressed as
V W
- - = Kab ~ (3.19)
l H
(a)
(Io) (c)
/XA
~= AA~-AA
/kA~
Fig. 3.4. Abrasive wear modes: (a) ploughing (b) wedge action (c) cutting and (d) definition
of ]3 based on cross-sectional area of ridge and groove. Based on Ref. [24].
Fig. 3.5. Abrasive cutting by conical indentor.

Abrasive wear coefficients vary from 10 -~ to 10 -3 which means the wear rates can
be significantly higher than in the case of mild adhesive wear. This is of
importance in many lubricated contacts where the deleterious effect of few
abrasive particles can outweigh the effective wear protection of lubricants. There
has been reasonable success in estimating K~b in a simple scratching mode. Zum
Gahr [26] conducted scratching experiments with a diamond pyramid. He
estimated/3 theoretically on the basis of hardness and strain rates and found good
correlation between theory and experiment.
The above modelling is valid for ductile materials. Brittle materials in abrasion can
undergo lateral cracking due to tensile stresses at the surface leading to brittle
fracture. In such cases the problem has to be treated in terms of fracture toughness
of the material.
The problem of three-body wear, which is industrially important, has not been
modelled adequately. Some qualitative observations can be made on the influence
of abrasive particles in a lubricated system. Firstly the particles that induce
abrasive wear are those that can enter the contact zone. Thus particles that cause
maximum damage are those that are in the same range as the film thickness. In
boundary lubricated contacts the influence of particle size distributions and surface
hardness on abrasive wear was studied [27]. One interesting observation is the
influence of abrasive wear on scuffing. Under some circumstances the large plastic
deformation due to abrasive action can induce scuffing at relatively low loads.
Many worn surfaces of engineering components show grooving in the sliding
direction showing abrasive action. While deeper grooves are due to abrasive action
finer grooves may arise from ploughing action. The erosion of materials by
impingement of high velocity particles also involves abrasive action.
3.6 Fatigue wear
3.6.1 Phenomena
Fracture and fatigue are well known concepts in material science. It is useful to
state these concepts in a simple manner. When a notched bar with a crack length ar
is subjected to tensile stress the stress intensity factor K~ may be expressed as
K~ = Bo(gar176 where B is the geometric factor and o is the applied stress

3.6. FA TI G UE WEAR 87
When a critical stress intensity factor K~c is reached the material undergoes
spontaneous fracture. Thus for a given crack length a stress for fracture is defined.
At lower stresses, applied cyclically, the crack grows incrementally finally leading
to fracture. This domain is referred to as fatigue. In the case of fatigue the material
can undergo large number of stress cycles before failure.
Fatigue wear can occur between surfaces due to cyclic stresses induced at the
asperity and sub-surface levels. Such stresses when repeatedly applied can finally
result in the detachment of a wear particle that may be termed a micro fracture
event. Fatigue wear is more extensively studied in rolling and rolling/sliding
contacts under lubricated conditions and is treated in chapter 8. In dry sliding
contacts when severe wear is involved transfer wear is likely to dominate. The
removal of the transferred material will involve cyclic stressing and fracture but it
may be assumed transfer is rate governing. When mild wear is involved the
possible oxide removal by fatigue process can compete with the milder transfer
process.
3.7 O x i d a t i v e w e a r
3.7.1 Phenomena
Metallic surfaces are covered with oxide films with thickness in the range of 10 to
100 A. Such films have a protective role in preventing metal transfer as discussed
under the section on adhesive wear. The protection from oxide films depends on
their adherence to the substrate and their ability to deform without cracking at the
asperity stresses involved. Also when locally worn the exposed areas should
oxidise and reform the film. This can be considered an ideal situation providing
maximum protection against wear. Under severe conditions high surface
temperatures lead to further oxidation and thicker films. Such thicker films are
expected to wear and reform. This wear mode is referred to as oxidative wear and
is confined to the formation and removal of oxide films. This wear mode may also
be referred to as film wear as proposed by Archard [25]. In normal air under dry
conditions oxygen availability is assured and it is assumed that oxidation is simply
governed by the reaction rate. This assumption may be questioned because in the
highly stressed asperity contact the available oxygen on each surface can limit the
extent of reaction. In the case of lubricated contacts oxygen availability is related to
the oxygen solubility in the lubricant and its diffusion in the confined monolayers
that exist in the highly stressed asperity contacts. The modelling based on kinetics
is of importance not only for oxidative wear but also for chemical wear in which
reactions with chemical additives are involved. Oxidative wear can also be treated
as chemical wear. In some cases there can be deleterious corrosive wear. One
example is the IC engine in which corrosive wear can occur due to acidic
components derived from sulphur components in the fuel.
3.7.2 Oxidative wear modelling
First of all the film wear model of Archard [25] is considered. The wear coefficient
defined in this equation is then derived on the basis of a kinetic treatment of the
oxidation process. The final equation so derived differs somewhat from that of
Quinn [28] as explained below.
Consider the earlier wear model of adhesive wear. During asperity encounter let a
film of thickness ~j be removed instead of a hemispherical particle. In such a case
the wear volume removed at one contact v d over a sliding distance 'd' will be
/[d 2
vd- ~ ~ (3.20a)
Following similar argument as per adhesive wear it can be shown
v
Wear r a t e V r = - =
Kf (3.20b)
l dH
In the above equation Kf is the wear coefficient that is equal to the inverse of the
number of encounters needed to form the critical film thickness ~. It is possible to
derive Kf on the basis of chemical kinetics. If t is the total time needed to form a
film of critical thickness and tr is the time during one encounter
tc
Kf--~
t
The time tr will be d/v where v is the sliding velocity
On the basis of static oxidation tests the oxidation is considered to follow parabolic
law. On this basis the increase in film thickness expressed in terms of mass change
per unit area Am, expressed in kg/m 2, will be related to the oxidation time t as
3.7. OXIDATIVE WEAR 89
Am 2 =kp t (3.21a)
where kp is the rate constant (kg 2 m -4s-z)
On the basis of kinetics kp = Apexp RT (3.21b)
where
mp = Arrhenius constant, kg 2 m -4s 1

Qp = activation energy, J/mole
Tc = contact temperature in ~
R = gas constant, J/mole-~
Expressing mass change in terms of density O (kg/m3) of the oxide and taking f as
the fraction of the volume that is oxide it can be shown from the above equations
Kf = (3.21c)
~292f2v
Substituting for Kf in Eq. 3.20b
Wear rate V =
Apexpl /RT~ W (3.22)
l ~p2f2v H
Quinn modelled oxidative wear assuming
Wear rate Vr =--V= K~W which is the expression based on adhesive wear in
l H
which a hemispherical particle is assumed." His treatment of K in terms of parabolic
law is the same as used for Kf here. The present author considers it is more
appropriate to model on the basis of film wear as done above. The final expression
due to Quinn is obtained by multiplying Eq. (3.20) with d/~.
For an oxidation process treating Qp, Ap, ~j P, and H as constants and for
simplicity assuming f = l , it can be shown
w
V~ = a 0 ~ e x p
/ (3.23)
V I cJ
where ao and a2 are constants
The importance of this model is to relate Kr to the physical process of oxidation. It

should be noted that the oxidation in tribocontacts is different from static oxidation.
The rubbing surfaces can have activation energies lower than in static conditions.
The assumed parabolic law may or may not be operative. Another issue is the
estimation of contact temperatures and the uncertainties involved in such
estimations. The significant difficulties involved in reconciling theory with
experiment were presented by Quinn [29], who addressed some of the issues raised
here. Chemical reactions with antiwear additives can be treated by a similar
approach and subject to similar uncertainties.
The oxidative wear model considered above assumes a critical film thickness for
oxide removal. This is only an assumption and the exact mechanism of oxide
removal need to be understood. The removal mechanism can be fatigue in which
case the oxide failure will be related to the oxide structure and composition. These
are variables and will not be uniform with changing temperatures. Molgaard [30]
has provided a useful discussion of the complexities of oxidative wear. He
considered that the mechanical characteristics of the oxide layer are more
important than the oxidation itself in the wear process. Also the concept of critical
film thickness may only be applicable beyond a particular thickness. Mild wear at
low temperatures is governed by tenacious thin films and the present model may
not be applicable for such cases.
One possible approach to wear modelling is by designed experiments. A

polynomial expression can be developed keeping in view the assumed model. If the
model fits then the mechanism may be justified in the range of conditions studied.
Further justification is necessary by surface analysis. An example may be cited on
the basis of short duration dry wear experiments conducted in a reciprocating tester
by Ravi Sankar et al [31]. The objective was to study the influence of load and
temperature on the wear rate and whether it involves oxidative wear mechanism.
The tests were conducted with cast iron ring piece sliding on a disc cut from a
honed liner. Three different loads of 30, 50 and 70 N and three different bulk
3. 7. OXIDATIVE WEAR 91
temperatures of 50, 100 and 150 ~ were selected and a 32 design adopted. Sliding
speed was fixed at 50 Hz with a stroke of l mm.
If it is assumed that Eq. (3.23) is applicable and v is constant it can be written as
logV~ = a o + allogW + ~a 2
(3.24)
L
where a0, a~ and a 2 are constants. The constant a~ is included to accommodate
possible exponent on W in the empirical relation. The relationships found in this
case were
For ring
858.75
logWr - - 2 7 . 6 9 + 1.051ogW- ~ (3.25a)
For disk
750.35
logV r = - 3 0 . 5 7 + 1.351ogW- (3.25b)
The asperity temperature rise was estimated based on section 3.3 and was added to
bulk temperature to obtain Tc . In view of the low speeds involved the asperity
temperature rise was low and less than 10~ The role of asperity temperatures can
only be established when tested over a wide range. For the test conditions used, the
above relationships show that exponential dependence on contact temperature is
justified. It is possible that an oxidative wear mechanism is operative here. Flake
like removal observed on the liner surface is shown in Fig. 3.6. The surface shown
is that observed at a load of 50 N after 25 minutes of running. No analysis of
surfaces and wear particles was carried out in this case to prove oxidation is the
governing mechanism. The work however shows the importance of approaching
wear empirically, and then only to see whether a particular model can explain the
data. If the present data cannot be reconciled with the model, other polynomial
expressions need to be sought which may or may not fit known models. Wear
processes are complex and it is necessary to analyse wear data objectively without
a preconceived model. The observed data may partially, and in rare cases fully,
justify a known model. These issues will be elaborated upon in later parts of the
book.
Fig. 3.6. Worn liner surface showing flake removal.
3.8 Wear map approach
In engineering design the required material properties can be measured by standard

tests and directly used. No such methodology exists for wear prediction. Every
situation has to be considered case by case and there is a proliferation of test
methods. Also the confidence with which predictions can be made is limited. At
the fundamental level more sophisticated investigations are revealing information
about material transformations, type of oxides, stress distributions and their
interactions. Asperity and bulk temperatures play an important role in all these.
Thus deeper knowledge is available about specific wear mechanisms. Translating
this knowledge into predictive models is the gap in wear research and is very
difficult to fill. The unsatisfactory situation regarding wear modelling has been
discussed in detail by Ming and Ludema [32]. One possible way is to study over a
wide range of operating conditions and map the wear behaviour as a function of
operating conditions. Lim et al [33] have studied wear processes with this approach
for low carbon steel in a pin-on-disk tester. They divided the sliding wear into
zones of ultra mild wear, plasticity dominated wear, mild oxidational wear, severe
oxidational wear, melt wear, and seizure. Plasticity dominated wear was modelled
considering Archard's model and the model due to Suh and Saka [34]. Model of
Suh and Saka is based on the concept of sub-surface plastic deformation leading to
3.8. WEAR MAP APPROACH 93
fatigue and delamination of wear flakes. This model was not discussed in this
chapter. This form of wear typically occurs when there is significant plastic flow at
the interface. This wear process also can be reconciled with a wear coefficient as is
normally expressed in the Archard's equation. Behaviour in each zone was
explained by an available model and approximate equations were provided. The
wear map obtained by them is shown in Fig. 3.7 and is a typical example of the
approach. Real systems will mainly operate in the mild wear regimes.
VELOCITY v
10 -'t 10 -~ t tO
io I
M Mild I #ear (ea S T E ~ L F-------
i S Severe, r-MOde Nap /
-Disk CpnFIgur~clon
we~ URE i
III
Ill
IX: -1 i S
D 10 ............ +_=__
Det(xm~natlo n
Go
(.h i
!
J I~i ~i_
III
bO
._1
weor " t
i
!
eg
0
z i0-' Severe-o
i L.
i DelaMin. -l;ion
.d-oxido, tional
WeO,M"
.~(~r
-, i.............. 1........
tO -2 tO~ 10 ~
NORMALISED VELOCITY "q"
Fig. 3.7. The wear-mode map for unlubricated sliding of mild steel. The various lines
separate the regions of dominance. The dark region represents the range in which transition
from mild to severe mode occurs. (From Ref. [33] reproduced partially by permission of
Elsevier)
In this figure load is in Newtons and sliding velocity is in rn/s. The wear rate
defined here is the normalised wear rate obtained by dividing the earlier defined
wear rate by the geometric area of the pin. Normalised velocity ~ is defined as
(velocity x pin radius)/(thermal diffusivity) while normalised pressure F is defined
as load/(pin area x hardness). Various wear mechanisms observed were due to
complex interactions between stress, temperature, and oxidation. The asperity and
bulk temperatures were calculated theoretically and had dominating influence on
wear mechanisms. These in turn are a function of sliding velocity and normal
stress. Such a map can provide the limits of safe operation for a given material
from wear point of view. It also provides insight into the wear mechanisms
involved. Similar maps are being developed for other materials.
The above approach may be considered semi-empirical and is likely to be

developed further in future. Development of such maps is time consuming even
with one material pair and one test configuration. The wear maps are expected to
change with the test configuration. For example if the tests are conducted in a
reciprocating mode or in rolling/sliding contact the results will be different. There
can also be variations if the area contact is changed to a concentrated line or point
contact. These issues coupled with the need to study a large number of available
and potential materials makes the exercise a daunting task. It is also to be noted
that when lubricants are used the mechanisms change and different wear maps for
such situations are needed. From a practical point of view the range of operating
conditions can be narrowed to those expected in service. This will lead to partial
wear maps that can be useful. Also in situations of low wear it is necessary to
distinguish small variations in wear rates. This is particularly true of lubricated
wear. As an example consider the wear coefficient with a given lubricant is 10 -~ .
If this wear rate has to be compared with that of another lubricant, it would be
necessary to be able to distinguish variation in wear coefficient of the order of
10 -9 . This level of distinction is not easy to achieve in experiments. In such cases
what is necessary is a detailed mapping in the range of interest with the necessary
precision levels. The wear map approach will be further considered in later parts of
the book.
References
1. K. H. Zum Gahr, Microstructure and Wear of Materials, Tribology series 10,Elsevier,

Amsterdam, 1987.
2. R. G. Bayer, Mechanical Wear Prediction and Prevention, Marcel Dekker, New York,
1994.
3. E. Rabinowicz, Friction and Wear of Materials, John Wiley, New York, 1965.
4. K. L. Johnson, Contact Mechanics, Cambridge University Press, Cambridge, 1985.
5. G. M. L. Gladwell, Contact Problems in the Classical Theory of Elasticity, Sijhoff and
Noordhoff, Amsterdam, 1980.
REFERENCES 95
6. J. B. Mann, T. N. Farris and S.Chandrasekar, Effects of friction on contact of transverse

ground surfaces, J. Trib.. ASME, 116 (1994) 430.
7. S. C. Lee and N. Ren, The sub-surface stress field created by three-dimensional rough
bodies in contact with traction, Trib.Trans., STLE, 37 (1994)615.
8. X. Tian, F. E. Kennedy, Maximum and average flash temperatures in sliding contacts, J.
Trib. ASME Trans., 116 (1994) 167.
9. N. P. Suh, Tribo Physics, Prentice Hall, New Jersey, 1986, 406.
10. O. S. Dinc, C. M. Ettles, S. J. Calabrese and H. A. Scarton, The measurement of surface
temperature in dry or lubricated sliding, J. Trib., ASME, 115 (1993) 78.
11. X. Tian and F. E. Kennedy, Contact surface temperature models for finite bodies in dry
and lubricated sliding, J. Tribo. ASME Trans., 115 (1993) 411.
12. M. Kerridge, Metal transfer and wear process, Proc. Phys. Soc. 68B (1955) 400.
13. T. Sasada, The role of adhesion in wear of metals, J. Japan Society of Lubrication
Engineering, 24 (1979) 700.
14. J. F. Archard, Contact and rubbing of flat surfaces, J. Appl. Phys., 24 (1953) 981.
15. E. P. Finkin, An explanation of the wear of metals, Wear, 35 (1975) 239.
16. T. Kayaba and K. Kato, The adhesive transfer of slip-tongue and wedge, ASLE Trans.,
24 (1981) 164.
17. D. A. Rigney, Comments on the sliding wear of metals, Trib. Int., 30 (1997) 377.
18. M. Godet, The third body approach: A mechanical view of wear, Wear, 100 (1984) 437.
19. Y. C. Chiou and K. Kato, Wear mode of micro-cutting in dry sliding friction between
steel pairs (Part 1): Effect of attack angle of the specimen,, J. Japan Society of
Lubrication Engineering, 9 (1988) 11.
20. N. C. Welsh, Phil. Trans. R. Soc., Ser. A 257, 31 (1965).
21. M. M. Khruschev and M. A. Babichev, Research on Wear of Metals, National
Engineering Laboratory, East Kiibridge, chapter 8, 1966, NEL Translation 893.
22. G. Sundararajan, The'differential effect of the hardness of metallic materials on their
erosion and abrasion resistance, Wear, 162-164 (1993) 773.
23. G. Sundararajan, A new model for two-body abrasive wear based on the Iocalisation
of plastic deformation, Wear, 117 (1986) 1.
24. K. Kato, Abrasive wear of materials, Trib. Int., 30 (1997) 333.
25. J. F. Archard, Wear theory and mechanisms, in M. B. Peterson and W. O. Winer (eds.),
Wear Control Handbook, ASME, New York, 1980, 35-80.
26. K. H. Zum Gahr, Microstructure and Wear of Materials, Tribology series 10, Elsevier,
Amsterdam, 1987, 132-148.
27. S. Odi-Oweri and B. J. Roylance, Lubricated three-body abrasive w e a r - contaminant
condition versus bounding surface material hardness, Trib. Int., 20 (1987) 32.
28. T. F. J. Quinn, The effect of hot-spot temperatures on the unlubricated wear of steel,
ASLE Trans., 10 (1967) 158.
29. T. F. J. Quinn, Oxidational wear modelling: I, Wear, 153 (1992) 179.
30. J. Molgaard, A discussion of oxidation, oxide thickness and oxide transfer in wear,
Wear, 40 (1976) 277.
31. P. Ravi Sankar, Rajesh Kumar and A. Sethuramiah, Dry wear of grey cast iron, 2000
AIMETA Int. Trib. Conf., Sep 2000, L'Aquila, Italy, 127.
32. H. C. Meng and K. C. Ludema, Wear models and predictive equations: their form and
content, Wear, 181-183 (1995) 443.
33. S. C. Lira, M. F. Ashby and J. H. Brunton, Wear-rate transitions and their relationships
to wear mechanisms, Acta metall., 35 (1987) 1343.
34. N. P. Suh and N. Saka, The delamination theory of wear, Wear, 44 (1977) 135.
Nomenclature
a contact radius in point contact
a~ asperity contact radius
ar crack length
ao, al, a2 constants
A* ratio of real area to geometric area of contact
Ag geometric area of contact
Ap Arrhenius constant
Ar real area of contact
b half contact width in line contact
B geometric factor
d average asperity contact diameter
d depth of penetration of cone
E effective elastic modulus of the two bodies
El, E 2 elastic modulii of the two materials 1 and 2
f fraction of the volume that is oxide
F normalised load
H hardness
k/ multiplication factor to obtain maximum shear stress
kp rate constant for parabolic growth
kl,2 thermal conductivity of material 1 or 2
K adhesive wear coefficient, non-dimensional
Kab abrasive wear coefficient, non-dimensional
NOMENCLA T (/RE 97
Kf film wear coefficient equal to inverse of cycles needed to form

critical film thickness
K I stress intensity factor
K Ic critical stress intensity factor
I sliding distance, m
I* ratio of half width of asperity contact to half width of geometric
contact
la half width of asperity contact
lg half width of geometric contact
p(r) pressure distribution in circular contact
p(x) pressure distribution in line contact
Po maximum pressure
Pm mean pressure
P load
P load per unit length in line contact
P~ Peclet number based on geometric contact
Pel, Pe2 Peclet number for body 1 or 2 as applicable
q heat flux
qa heat flux on the basis of real area
Qp activation energy
R equivalent radius of two bodies in contact
R molar gas constant
Rl, R2 radii of two bodies in contact
t time needed to form a film of critical thickness
tc time during one encounter of asperities
Tc contact temperature in absolute units
V sliding velocity
V normalised velocity
Vd wear volume removed at one contact over a sliding distance d
V wear volume over a sliding distance l
Vd wear volume generated at all asperity contacts over a sliding
distance d
Vr wear rate, volume/sliding distance
W load
98 CHAPTER 3 . DRY WEAR MECHANISMS AND MODELLING
Greek letters
fractional metal contact area within real area

AAg - AA~
= as in Fig. 3.4
AA g
line contact deformation
Am mass change per unit area
A0~ asperity temperature rise in contact
A0~, A0 asperity temperature rise at the surfaces with all heat flowing to
body 1 or 2
A 0 at asperity temperature rise with thermal interaction
A0at 1 , A0at2 asperity temperature rise with thermal interaction with all heat
flowing to body 1 or 2
A0 g temperature rise in the geometric contact area
V Poisson ratio
V1~ V2 Poisson ratios of bodies 1 and 2
0 semi-included angle of the cone
Oa asperity temperature
Ob bulk temperature
P average density of oxide
0 applied stress
Zma~ maximum shear stress
critical film thickness
99
4. Boundary lubrication mechanisms

metallic materials
4.1 Introduction
Lubricated wear occurs at the asperity contacts. In dry contacts the real area is
involved in the wear process. In lubricated contacts two different situations arise.
In one case the molecular layers adsorbed on the surfaces resist penetration
allowing metallic contact only over a small fraction of the real area. This results in
substantial reduction in wear. In another situation, which arises under severe
operating conditions, adsorbed films cannot survive. In such cases chemical
additives are used to generate wear resistant films. This class of additives is
referred to as antiwear (AW) additives. Another class of more reactive chemical
additives, referred to as extreme pressure (EP) additives, is used to increase the
load carrying capacity.
Conventionally boundary lubrication refers to adsorbed films. For the purpose of

this chapter the boundary lubrication regime is treated in a broader sense. It is
defined as the regime in which molecular layers and/or reacted films provide
effective lubrication. Effectiveness refers to reduced wear as well as increased load
carrying capacity. The industrial applications of these additives were considered in
the second chapter. This chapter is divided into two major sections dealing with
adsorbed films and chemical reaction films. Firstly the conceptual physical model
for the action of adsorbed films is considered. This is followed by a consolidated
treatment of monolayers, and dynamic adsorption followed by a consideration of
modem developments. Modem developments include a brief coverage of
multilayers and nanotribology. The section on reaction films deals with the main
classes of additives and their reactions. This is followed by a consideration of
recent developments related to fundamental mechanisms, multi-component
additives, and related aspects. Some issues related to the limitations of the available
models are raised during the discussion. The final section deals briefly with the
commonly used surface analytical techniques.
100 CHAPTER 4. BOUNDARY L I./BRICA HON MECHANISMS
The aim of this chapter is to provide adequate background for the future chapters
dealing with lubricated wear in terms of theory and practice. Some of the recent
developments may not be relevant for this purpose but may be of importance in the
future. The literature on boundary lubrication is very vast and the coverage is
limited to meet the central objective of the book. Detailed consideration of
advanced chemistry is not within the scope of this chapter and where necessary
only the main reaction paths are mentioned.
4.2 Adsorbed layers
4.2.1 Mechanism of action
It is appropriate to start with the conceptual model of lubrication mechanism due to

Bowden and Tabor [1]. In this model the lubricant molecules are considered
physically adsorbed on the surfaces. Polar molecules are adsorbed more strongly in
comparison to hydrocarbon molecules. The model is usually presented with the
polar heads attached to the surface and the hydrocarbon chains projecting vertically
upwards. Assume a monolayer of molecules on each surface. When two flat
surfaces covered with monolayers are brought into contact under load as shown in
Fig. 4.1 the hydrocarbon chains interact preventing direct surface contact. If the
surfaces are now slid relative to each other the sliding ideally occurs in the
lubricant layers with no metallic contact. In such a situation there will be no wear.
Friction also will be low as the shear occurs in the lubricant layers. It may be noted
that vertically oriented molecules are 20-30 A in length and the diagram highly
exaggerates the length.
Load
Polar heads
r i
f ~ Sliding
'i
__ Methylene group
i interaction=
Fig. 4.1. Interacting boundary layers with flat surfaces

4.2. ADSORBEDLAYERS 101
The real surfaces are rough and the contact situation for such a case is illustrated in
Fig. 4.2a. With reference to the asperity dimensions the molecules will be mere
dots. When the surfaces are loaded the load is transmitted to the asperities via the
lubricant films. For illustration the boundary layer is shown as a thin continuous
film. The stress is assumed to be the same as in the case of dry contacts. Thus the
films are subject to high pressures of the order of the metal hardness and the real
area A r is the same as in the case of dry contact. During sliding the shearing
occurs in the lubricant layers and ideally no wear should occur. In reality the films
are highly stressed and develop defects where metal contact occurs leading to wear.
The metal contact is characterised by fractional film defect, a which is defined as
the fraction of real area that is metallic. The word 'metallic' needs to be qualified.
In real situation the contacts involve oxide as well as metal. In this simple model
the contacts are assumed to be metallic. The contact situation with defects is
illustrated in Fig. 4.2b that shows a magnified view of contact at the asperity. The
flattened asperity is shown having a molecular level roughness and the metal
contact occurs at some of the peaks. The overall friction now arises from metallic
contact as well as the shear in the film. If shear strength of the film is s z and the
shear strength of the metallic junctions is Sm it can be shown that friction force F is
F = A t [ a s m + ( 1 - a)s t ] (4.1a)
where A r is the real area of contact. If now overall friction coefficient is defined
byf
F
f =~ : ct f m + (1 -- a ) Z (4.1 b)
W
where
W = load
fm -- friction coefficient for metallic junctions
f, - friction coefficient for boundary film
In principle for a measured value of f , the fractional film defect can be calculated
if fm and fl are known. For low value of ct, which is the normal zone of interest,
variations i n f w i l l be too small to be measured. Also fl and fm can have variations
depending on the operating conditions. Hence the above equation is inadequate to
determine a quantitatively.
102 CHAPTER 4. BOUNDARY L UBRICA TION MECHANISMS
The above model shows that the efficacy of a boundary layer is governed by the
film defect a . The metallic contact is shown to occur at some fine protrusions for
illustration. The exact spots where metal contacts occur are not clearly defined in
the literature. The fractional metal contact area is normally taken as the fractional
area of defects in the adsorbed film. The penetration through the film is related to
the tenacity with which the molecules are attached to the surface. This tenacity, or
the strength of adsorption, is governed by the nature of the lubricant as well as the
material. The actual orientation of the molecules on the surfaces during rubbing
process can be very different from the vertical orientation observed in preformed
monolayers. In many cases the molecules may lie fiat on the surface and the
interaction can be over the whole molecular chain during sliding. Another
interesting factor is that the friction coefficient of the adsorbed molecules is nearly
constant over a wide range of asperity stresses. Such varying stresses can be
obtained at the asperities by changing from soft to hard metals. This can be
explained if it is assumed that the shear stress increases linearly with pressure as
originally suggested by Bridgrnan [2].
Load
Asperity |
layer contact
(a)
contacts
(b)
Fig. 4.2. (a) Interacting boundary layers with real surfaces and (b) Detail of metal contact at
asperity due to film defect.
4. 2. ADSORBED LA YEtL~ 103
The adsorption of the molecules may be physical or chemical. Adsorption

involving chemical interaction is referred to as chemisorption and involves
chemical bonding with the surface. Several studies were conducted with
monolayers to elucidate the detailed mechanism of boundary lubrication. Most of
these studies were conducted with preformed monolayers under low speed and low
load conditions. The sliding speeds involved were usually less than 10 mm/sec
while the loads involved were less than 100g. In these studies the actual orientation
of the layers on the surface during rubbing was not known. Generally accepted
ideas related to boundary lubrication, and available experimental evidence is
presented in the following sub-sections. The presentation aims to provide a
consolidated picture of boundary lubrication. Some issues that are controversial in
nature are also raised during discussion.
4.2.2 Monolayers as boundary lubricants
Monomolecular films can be deposited by known techniques like transfer of

Langmuir-Blodgett films, retraction from melts, and deposition from solutions.
Initial pioneering studies were due to Hardy. Later investigations of significance
were conducted by Frewing, Bowden and Tabor, Greenhill, Cameron, Zisman, and
others. A detailed coverage of the work by all these authors is available in a book
[3] published in 1969 that appraised the world literature on boundary lubrication.
The detailed structure of a monolayer of stearic acid from [4] is shown in Fig. 4.3.
The stearic acid molecules are vertically aligned with their polar heads anchored to
the surface. The chain has a length of 24.2 A and a cross sectional area of 18.4 A.
This ideal structure may be somewhat modified depending on the nature of
surfaces and the techniques used to deposit the monolayer. Such a film can be
regarded as solid like. These structures occur with many polar compounds that
include alcohols, amines, and fatty acids. Polar molecules are those molecules in
which there is a charge separation. This leads to stronger adsorption via the
positive polar head of the molecule. Thus the (OH), (NH2), and (COOH) groups in
the above molecules are responsible for the stronger adsorption. The adsorption
strength depends not only on the polar group but also on the cohesion between
molecules via methylene group interactions.
4.2.2.1 Chain length effects
An effective condensed film can be formed only when chain length is adequate.
This has been put in evidence by two different studies. In one study Levine and
Zisman [5] deposited monolayers of fatty acids, amines, and alcohols with different
104 CHAP TER 4. B O UNDA R Y L UBI~CA 110N MECHANISMS
chain lengths on stainless steel and glass surfaces. They then measured the contact
angle of methylene iodide on the monolayer. Contact angle is a measure of the
surface energy. When a surface forms a good condensed film the surface energy is
low and the contact angle is high. These measurements showed that the contact
angle increased with the chain length and reached a value of ~_ 69 ~ for a carbon
chain length of N - 15. The contact angle remained nearly constant for N values in
the range of 15 to 26. Below N = 15 there were minor differences between the
different compounds with regard to coverage.
61 ., 2
0-- C)- Corboxyl
A~V AI ky I Chain
-=- -o- Methylene
e- o- Methyl
Fig. 4.3. Schematic representation of stearic acid monolayer.

(Reproduced from Ref. [4] p204 with permission of Dr. K. C. Ludema.)
Another approach to investigation of the influence of chain length is to study the

friction and durability of monolayers. Levine and Zisman [6] have studied the
durability of fatty acid monolayers in slow speed sliding. The number of passes
needed to obtain sudden transition in friction was taken as the criterion for
durability. These studies showed that the long chain fatty acids are more effective
and support the findings in the previous paragraph.
Monolayers with branched chains have been found to be less effective. This is
attributed to the side chains that decrease intermolecular cohesion. One example is
the study conducted with stearic and isostearic acid dissolved in liquid paraffm [7].
Stearic acid showed lower friction in comparison to the isomer when studied as a
4.2. A D S O t ~ E D LA YERS 105
function of acid concentration in a ball-on-flat steel contact. Though monolayers

were not studied it is reasonable to attribute the variation to the nature of the
adsorbed acid. Studies by several other authors also support the above findings.
4.2.2.2 Adsorption Vs chemisorption
Physical adsorption is reversible and the molecules are expected to desorb at the
melting temperature of the polar compound. Bowden et al [8] studied the influence
of temperature on lubrication failure. In these tests 1% lauric acid solution in liquid
paraffin was used as the lubricant. In the slow speed test machine the temperature
was progressively increased till failure. The lubrication was assumed to be
governed by the lauric acid, which adsorbs preferentially on the surface. Several
metals were studied in these experiments. The interesting observation was that with
unreactive metals the failure temperature ranged from 20~ to 50~ while for the
reactive metals it ranged from 90~ to 100~ The melting point of lauric acid is
49~ and in the case of unreactive metals the failure is governed by the melting
point of lauric acid. In the case of reactive metals the failure is attributable to the
melting of the chemisorbed soap. The metals that reacted were zinc, cadmium, and
copper while the unreactive metals were aluminium, chromium, iron, and silver.
Further research in the fifties indicated that even with reactive metals the presence
of oxides and humidity are necessary for soap formation [9,10]. Another related
aspect is the role of fresh metal surfaces in chemisorption. Earlier work due to
Tingle [ 11] suggested that fleshly cut metal surfaces do not promote chemisorption
of fatty acids. On the other hand the work of Mori et al [ 12] showed that organic
acids adsorbed strongly on freshly cut aluminium surface under high vacuum
conditions. At this stage the contradictions are only pointed out and further
consideration will be given to these issues later in the chapter.
4.2.3 Dynamic adsorption and real systems

Studies on preformed monolayers clearly demonstrated their effectiveness in
boundary lubrication. In real systems polar compounds are normally used as
additives in base fluids. The polar molecules compete with the carrier fluid for
adsorption on the surface. A simplified approach to the problem is to consider that
the polar molecules dominate adsorption. Assuming Langmuir adsorption it can be
shown that the extent of surface coverage is governed by the following equation
= K (4.2)
c(1-q,)
where
K = equilibrium constant
(P = fractional coverage
c = concentration of the polar substance in the base fluid
The equilibrium constant is based on the rate constants for adsorption and
desorption at the surface. The relationship shows the dynamic nature of adsorption
at the surface.
The fundamental thermodynamic equation that governs adsorption is
AG o = AH o _ T AS ~ (4.3)
where
AG ~ = free energy change, J/mol
AH ~ = heat of adsorption, J/mol
AS ~ = entropy change, J/mol
T = absolute temperature, ~
Also
AG o = - R T ( l n K) (4.4)
Combining Eqns. 4.2 to 4.4 it can be shown
AH o AS o ~p
In c =
I
RT R ]El
+ In
i " ~a
(4.5)
The above has been verified indirectly. The experimental approach involved
measurement of failure temperature as a function of additive concentration in the
carrier fluid. It was assumed that AH ~ and AS ~ were independent of temperature.
Also failure was considered to occur at a particular level of surface coverage (pc
for all additives. Defining Tr as the failure temperature Eq. (4.5) can be recast as
AH ~
In c = ~ + constant (4.6)
RL
Thus In c and 1/Tr are linearly related at failure. Reasonable linear relationships
were found for different fatty acids and soaps by Frewing [ 13] using a low speed
reciprocating tester. Crew and Cameron [14] also found reasonable linear
relationships for lauric acid dissolved in different normal alkanes. The implicit
assumption here is that the fractional metal contact area is equivalent to (1- r
based on dynamic adsorption. The above equations are as derived in Ref. [13].
In reality the adsorption occurs for both the carrier fluid and the additive. Dorinson
and Ludema [15] developed a more complex relationship considering co-
adsorption and free energy changes for both solvent and solute. There is also some
evidence that improved boundary lubrication occurs when the chain length of the
additive and the solvent are matched [ 16].
Another approach is to measure directly the heat of adsorption and relate the
measured value to tribological performance. One important contribution in this
area is due to Groszek [17]. He measured the small heat effects due to adsorption
by the flow microcalorimeter technique. In this method additive was injected into
the flowing solvent and its adsorption on porous adsorbent was measured by the
increase in temperature. The heat of adsorption was with reference to the selected
flow conditions and the amount of additive injected. The surface coverage
estimated in these experiments was less than a monolayer. The important aspect of
these studies is the influence of adsorbent besides the known influence of the
additive. For example heat of adsorption of stearic acid was found to be 50.8, 13.0,
13.4, and 45.7 cal/g with alumina, silica, carbon, and iron powders respectively.
These results were obtained with n-heptane as the solvent. Wear studies with a 4-
Ball machine with mineral oil as the carrier fluid showed that wear was reasonably
correlated to the heat of adsorption of the additives from benzene on iron powder.
Another study in this direction by Hironaka et al [18] showed that the heat of
adsorption of stearic acid was much higher on FeS and Fe304 as compared to
Fe203. High heats of adsorption were very likely due to chemisorption. From a
fundamental point of view it is obvious that in real situations where surface oxides
with different compositions and morphology exist heats of adsorption cannot be
assigned any specific value. Also when chemisorption is involved the desorption
will be less important and the mechanism may be simply governed by the soap or
other chemisorbed films. Hence the author considers that development of
quantitative relationships between wear and heat of adsorption is difficult.
However it may be accepfed that directionally higher heat of adsorption is
beneficial to boundary lubrication.
108 CHAPTER 4. BOUNI)ARY L UBRICA TION MECHANISMS
The models considered above rely on the overall adsorption/desorption on the

surface. The contact conditions at the asperities are not invoked. To effectively
model the behaviour, asperity temperature and the contact time have to be taken
into account. Rowe [19] has proposed a model for wear based on fractional film
defect. Firstly the mean time of stay of a molecule at a surface site, t r c a r l be
expressed as
t r = to eE/RTs (4.7)
where
E = heat of adsorption, J/mol
Ts = temperature of the surface, ~
to = fundamental time of vibration of the molecule, s
R = molar gas constant
The probability of desorption of a molecule depends on the time available for it

during asperity contact. This time is related to the sliding velocity in the contact.
With these considerations Kingsbury [20] proposed the following expression for a
the fractional defect
a = 1 - exp - (4.8a)
The time t x is the time needed to move one molecular diameter and is expressed
as
x
tx - - (4.8b)
H
where
= diameter associated with an adsorbed molecule, m
= sliding velocity, m/s
Combining Eqs. 4.7, 4.8a, and 4.8b the fractional film defect can be expressed as
x -E~T~]
ot 1 ~ exp ~
u-ot
~ e (4.9)
For small values of a , less than 0.01, the equation may be simplified to
x -E/RT~
~ = ~ e
(4.10)
ut0
For a given molecular species x and t o may be taken as fixed. From the boundary
mechanism point of view the calculation of fractional film defect is based on
dynamic adsorption coupled with the time needed to move a distance of a
molecular diameter. The model thus approaches the real situation and is a step
forward in modelling. The wear model based on a shall be considered in chapter
5. The model considered is for a single molecular species. In principle the idea can
be extended to multiple additive system by modelling the more complex co-
adsorption of the molecules on the surface.
Another idea indirectly related to adsorption is the concept of critical failure

temperature. Blok [21] proposed that mineral oils fail at a critical temperature of
around 150~ leading to scuffing. Applied scientists and engineers have
intensively investigated the validity of this concept. The aspect of lubricant failure
will be considered in later chapters.
There is a problem in reconciling the work with monolayers, dynamic adsorption

studies, and the concept 6f critical temperature. For example, long chain paraffin
with a melting point of say 40~ should fail to lubricate at this temperature when
applied as a monolayer. However such long chain molecules can lubricate
effectively to at least 100~ in practical situations. This may be because in normal
situations the lubrication is governed by adsorption/desorption while in the case of
a monolayer readsorption will be restricted. Polar impurities can also affect the
boundary lubrication of hydrocarbons.
4.2.4 Recent investigations of adsorbed films
Recent developments can be divided into two categories. The first category of
studies involves better understanding of mechanisms with more sophisticated
analytical tools. The studies under ultra high vacuum to elucidate mechanisms form
a part of this category. In these studies the tribological contacts involved are
described as 'macro' or 'micro'. The distinction between micro and macro contacts
is not clearly defined. Even when the contact is macro the asperity contacts are
micro in nature and boundary lubrication is governed by the micro contacts. Hence
this category is treated as macro/micro contacts. The second category of studies
110 ('HAPTER 4. BOUNDARY L UBRICA TION MECHANISMS
involves elucidation of mechanisms at atomic level. Major investigations are being

carried out in this area that is now known as nanotribolgy. The purpose of this
section is to highlight the major findings in these categories.
4.2.4.1 Macro/micro level investigations
4.2.4.1.1 The role of oxides
Hu et al [22] have studied the tribochemical reaction of stearic acid on copper

surface. Surface enhanced Raman spectroscopy was used to identify the reaction
products. Chemisorption was observed when stearic acid was adsorbed in air
atmosphere and not in argon atmosphere. This was attributed to the presence of
cuprous oxide in air. Cuprous oxide was also detected by Raman spectroscopy.
Tribological studies were conducted in a pin-on-disk machine. The layers formed
were analysed and copper stearate was detected after running for 400 meters at a
load of 15N. The complex reaction route involves initial chemisorption forming a
unidentate that eventually gets converted to cupric stearate. These detailed studies
were made possible by Raman spectroscopy that can detect low frequency
vibrations effectively.
Another interesting study in this direction is an electrochemical study of the

interaction between fatty acid and copper oxide by Su [23]. Commercial wire
drawing emulsion in water was used for the studies. The variations in current
density in an electrochemical cell formed the basis of investigation. Current density
increases when oxidation occurs on the scratched copper surface. When stearic acid
was injected over copper oxide current density increased only when there is
chemisorption. With this technique the author successfully demonstrated that
chemisorption occurs with cuprous oxide and not with cupric oxide. It was also
observed that a minimum oxide thickness of 60 A is necessary for effective
chemisorption. The author also observed that the fatty acid might be ineffective in
wire drawing as formation of the required thickness of cuprous oxide is unlikely.
The oxide thickness and its nature were found by electron spectroscopy for
chemical analysis (ESCA) and Auger electron spectroscopy (AES). These
techniques are considered at the end of the chapter.
The above studies and the earlier studies referred to in the sub-section 4.2.2.2
establish the importance of oxides in chemisorption. Nature of oxide and its
thickness are also important. Detailed studies of a similar nature with different
metallic oxides can be valuable in selecting polar additives. Such studies have to be
coupled with an understanding of the oxides formed in real situations.
4.2.4.1.2 Triboemission
Triboemission consists of exoelectrons, positive ions, and photons. Major interest

is in exoelectron emission that occurs during rubbing of surfaces. Such emission
can activate the surface and can lead to chemisorption at the surfaces. Significant
amount of investigation is being done in this area. The most recent work with
diamond pin rubbing against alumina, sapphire, and aluminium [24] showed that
major emissions occur with alumina and sapphire. The tests were conducted under
a vacuum of 10 -g torr. Aluminium surface was less active. Several other research
papers can be cited in this area [25,26,27]. Nakayama et al [26] observed that under
oxygen pressure of 3 x 10 -2 Pa significant electron emission could occur with
aluminum surface. This is attributed to the chemoemission that is a consequence of
surface oxidation during cutting experiment. With a series of metals they found the
chemoemission varied with the ease of formation of the oxide characterised by the
free energy of formation. In general the electron emission is favoured by oxidised
surfaces. With fresh metal surfaces the emission seems to depend on the extent of
plastic straining and fracture. However electron emission can also be a
consequence of surface interaction with surrounding species as observed above. It
thus appears chemisorption under cutting conditions may or may not occur
depending on the operating conditions and environment. This may be the reason
for different observations regarding chemisorption of polar molecules in the earlier
literature under cutting conditions [11,12]. Exoemission is also of importance in
chemical reaction film formation and shall be considered later.
4.2.4.1.3 Ultrahigh vacuum studies
Ultrahigh vacuum can provide an environment where clean surfaces without

contaminants can be studied. Friction studies can be conducted in such
environment. Lubricant molecules can be introduced into such system and their
influence on friction studied as a function of coverage. This leads to precise
understanding of the lubrication mechanisms involved. Such studies are being now
conducted at an advanced level. Several papers have been published by McFadden
and Gellman and [28] is a typical example of this work. While ultrahigh vacuum
studies were done in the past the present studies were conducted with
instrumentation that can measure adsorption coverage precisely and also ensure
initial uncontaminated surfaces. The important observation with pure copper
surfaces is that lubrication is ineffective when surface coverage is less than one
monolayer. A schematic diagram showing the influence of surface coverage on
friction from [28] is given in Fig. 4.4. The studies have so far been mainly
conducted with lower molecular weight alcohols. The static friction approaches
112 ('HAP'IER 4. BOUNDARY L UBRI('A TION ME('HANISMS
limiting values at a coverage of 8-10 monolayers. Further studies are likely to

clarify fundamental mechanisms involved in boundary lubrication. It may be noted
that the tests were conducted at a low temperature of 100~ while the normal force
applied ranged from 10-80 mN. The sliding speeds ranged from 0.5-200 I.tm/s. The
static friction coefficients observed even with several monolayers is higher than
0.25 for the alcohols studied. The clean copper surfaces showed friction coefficient
in the range of 6.0-7.0. Such high values are expected with very clean surfaces due
to junction growth as discussed in chapter 1.
~- stick slip -4,. sliding ---~
.~_.atorns an~l_~ -.,--molecules

molecules
C
o l
. , , . U s clean
u
~ sajrfoces
Nk=,
8
C
O l
0
u
~
la.
- I " r ,' ''1 r " "" " I
0.1 I 10
Adsorbote Coverage ( M L )
Fig. 4.4. Schematic diagram showing variation of friction coefficient between Cu (III)
surfaces with surface coverage. Dashed lines show standard deviation.
(Reproduced from Ref. [28].)
4.2. ADSORBED LA YERS 113
4.2.4.1.4 Multilayers in boundary lubrication
In a sliding or rolling/sliding contact lubricant molecules enter the asperity contact.

Several layers of molecules can enter the contact zone. These multlayers can get
squeezed out leaving only the tenacious monolayer effective in boundary
lubrication. It is also possible that multilayers survive leading to a different form of
boundary lubrication. To study such phenomena it is necessary to measure the film
thickness as well as the tribological response of the layers. Multilayers involved
will be of the order of 100-200 A. With normal smooth surfaces the asperity
dimensions are typically 0.1 gm (1000 A) and are not suitable to study at the
required nanolevel separations. One major approach to such studies has been the
surface force apparatus (SFA). In this apparatus molecularly smooth cleaved mica
surfaces are used for study. The molecular separations can be measured to an
accuracy of 0.1 nm while the force can be measured to a sensitivity of 10 -8 N. The
mica surfaces are curved and the crossed cylinder contact leads to a circular
Hertzian contact whose dimensions depend on the normal load. Friction force in
sliding can also be measured. The detailed description of the apparatus is available
in the book by Israelachvili [29]. Israelachvili and others have conducted several
studies with this apparatus and two representative papers in this area may be cited
[30,31]. These studies showed that at separations approximately less than ten
molecular layers the fluid behaviour could not be handled by the usual continuum
theories. Such layers can have viscosity much higher than the bulk fluid viscosity
and during sliding can exhibit solid or liquid like behaviour. With shearing the
molecular structures can undergo transitions between solid and liquid states leading
in some cases to stick-slip phenomena. It may now be argued that boundary layer
cannot be treated as a monolayer. However the knowledge base is built around very
smooth surfaces with low loads and sliding speeds in the gm/s range. Hence at this
stage it is difficult to make any definitive statement with regard to the behaviour of
boundary films in normal engineering situations.
Another type of study involves the measurement of film thickness to nanometer

level in rolling contact [32]. These studies showed that hexadecane film thickness
could be reconciled with the existing theory for EHD. When fatty acids at a small
concentration of 0.1% by weight were introduced as additives the film thickness
increased by 2-3 times at the lower speeds. At higher speeds exceeding 0.1 m/s the
film thickness corresponds to that of hexadecane alone. Another factor brought into
focus was the influence of water. When water was present the film thickness at
lower speeds increased by 7-8 times and was attributed to the formation of metal
carboxylate in the presence of water. These studies also demonstrate that near
surface effects leading to viscosity changes can occur with adsorbed species. These
studies are yet to be coupled with the studies under sliding conditions in the
previous paragraph.
4.2.4.2 Nano level investigations
In SFA a macro contact area of 10-30 Jam2 is involved and the interaction forces
are measurable with the apparatus. On the other hand the forces involved in atomic
interactions are of the order of 10 -9 to 10 -~ N. A far more sensitive apparatus, the
atomic force microscope (AFM), has been developed for these studies and is
described in detail in the literature [33]. In this apparatus a fine tip approaching the
surface can study atomic interactions. The apparatus can also be adapted to
measure friction force as well as surface topography. This is achieved by the
scanning mode in which the lower surface moves relative to the tip. The control of
displacement to within 0.1 A and the ability to measure very small forces with fine
cantilevers are the major features of this apparatus. Development of fine tips with
nano scale radii is another important requirement for these studies.
AFM has opened up major investigations in nanotribology. The developments

involved are reviewed in some of the recent papers [34,35]. Detailed consideration
of this area is outside the scope of the present book. Only major findings are briefly
summarised below
1. The adhesive forces involved during approach and retraction normally

show hysterisis for solid surfaces. This means that adhesion phenomena
are normally irreversible. When lubricant molecules are applied to the
surfaces the adhesive forces reduce significantly [36]. In this work it was
shown that adhesive forces decreased substantially when stearic acid was
applied to an aluminium surface covered with natural oxide.
2. Friction at the atomic scale can be several times lower than for macro
contacts [37]. The possible reasons cited include ploughing and loose
debris effects in macrocontacts. No explanation seems to be available why
static friction itself is high for macrocontacts as compared to nanocontacts.
3. Friction cannot be explained quantitatively and several approaches are
being considered. With specific reference to boundary illms Israelichvili
[38] observed that hysterisis observed in normal approach and retraction
mode may be correlated with friction observed in the contact. This
reference provides state of the art coverage of molecular films in tribology.
4. Capillary condensation of water between the tip and sample can lead to
strong adhesive force due to meniscus formation. This can cause high
friction at the start of sliding and is referred to as stiction. Stiction is of
4. 2. A D S O R B E D L A Y E R S 115
relevance at the head-disk interface in computers. This area has been well
investigated and led to the development of perfluoro polyether lubricant
coatings grafted to the disk surface. Nanotribological studies have been
useful in this development [39,40].
Other investigations being conducted in this area include different coatings, and
characterisation of wear and roughness at the nano level. This research is finding
application in the computer industry and in micro elecro mechanical systems
(MEMS). The applicability of nano level studies to macro contacts is not clear.
One approach is to conduct nano level studies on real surfaces and see their
applicability to practice. A recent example is the surface obtained in the
macroscopic reciprocating tester using molybdenum based additive. This surface
was then studied in a force microscope observing both topographical changes and
friction [41] along the surface. This study showed the detailed local variations on
the surface. Such studies can be an additional tool for understanding the
mechanisms involved. In yet another study a more rugged version of force
microscope was used to characterise the influence of friction modifiers using actual
wet plate clutch materials [42]. The authors found that the traction behaviour
observed can be correlated with practice despite large difference in the range of
sliding velocities used in the two cases.
4.3 Boundary lubrication with reaction films
4.3.1 Role of chemical additives
Polar molecules considered in the previous sections have temperature limitation of

about 150~ beyond which they are ineffective. Many real systems exceed these
temperatures and have to be protected by other means. Chemical additives that
react with the surfaces and form protective films have been found to be effective
for this purpose. Such additives are designated as extreme pressure (EP) additives.
A tribological system may operate under varying conditions and the purpose of the
EP additive is to act only when needed. Higher temperatures leading to lubricant
failure and seizure can be a consequence of high loads, high speeds, or a
combination of both. In practical terms the role of additive amounts to increasing
the operating severity of the system without failure. Any excessive reaction of the
additive will lead to corrosive wear that is undesirable. Hence it is necessary to
tailor the additive to limit its reaction to a desirable level when needed.
116 (THAP IER 4. BOUNDARY L UBRICA lION MECHANISMS
Another class of chemical additives reacts with the surfaces forming reaction films
under milder operating conditions. These additives are called antiwear (AW)
additives. As their name suggests, these additives form films that reduce the wear
in comparison to a system without these additives. These additives are usually less
active than EP additives. In many AW additives the reactive elements are the same
as in EP additives. The structure of these compounds is different from those of the
EP additives and this difference leads to lower reactivity. These additives are
industrially important and used extensively in engine oils, hydraulic systems, and
gear oils.
Many real additive systems can have additives that provide both EP and antiwear
function. This may be also achieved through a multifunctional additive that may
incorporate active elements, say sulphur and phosphorous, in the same molecule.
The distinction between AW and EP additives is not clear-cut when additive
combinations are used, as the reaction films formed are far more complex.
The purpose of this part of the chapter is to provide information related to the
nature of additives and their mechanism of action. The literature available in this
area is voluminous and the presentation will be limited to the generally accepted
ideas in a consolidated manner. This will be followed by a consideration of the
more recent studies and their implications. Detailed consideration is given to wear
mechanisms and modeling in the later chapters.
4.3.2 Nature of additives and their reaction mechanisms
The chemical reaction between surface and additive will first involve
chemisorption of the additive on the metal surface. This will be followed by
decomposition of the molecule leaving the reactive elements attached to the metal
surface. Further reaction leading to formation of a reacted film of a given thickness
involves diffusion of the active elements into the surface coupled with migration of
metallic ions towards the surface. The ensemble of these reactions can be
reasonably studied under static conditions with approaches well known in physical
chemistry. The reactivity under static conditions is then related to the tribological
performance usually determined with standard test machines. This approach suffers
from two deficiencies. The first is the known fact that films formed in dynamic
conditions involving wear and reformation will be very different from the static
films that involve no rubbing. The second problem is that the performance as
judged by a given standard machine cannot be generalised and is at best
representative of the performance under the chosen conditions. These aspects were
broadly considered in section 2.5.3. These issues need serious consideration and
4.3. BOUNDARY LUBRICATION WITH REACT"ION FILMS 117
shall be dealt with in the relevant chapters. As detailed consideration of test

machines and test procedures shall be taken up later the performance tests used are
not treated in detail here. The main approach involved in testing is only mentioned.
These issues are raised at the outset so that the reader is aware of the limitations
within which EP and AW additive mechanisms are normally studied. The additives
are categorised into sulphur, phosphorous, and chlorine compounds so that their
mechanisms can be explored. This is followed by a consideration of the mixed
systems and multifunctional additives.
4.3.2.1 Sulphur compounds
Large numbers of organic sulphur compounds are used which include elementary
sulphur, sulphurised fats, monosulphides, and disulphides. Forbes and co-workers
proposed that the reactivity of sulphur compounds isrelated to the bond energy of
the C-S bond [43,44] in the additive molecule. Lower bond energy results in easier
reaction with the surface and results in better EP action. Dorinson and Ludema [4]
listed the typical sulphur compounds and their structures. This information is
reproduced in Table 4.1. Bond dissociation energies for C-S bond are also given in
the same table. In the case of elemental sulphur the bond energy refers to the S-S
bond. Disulphides in general have lower bond energies than monosulphides and
hence are considered to be better EP additives. But these statements have to be
qualified in terms of the nature of alkyl and aryl groups involved which have a
substantial influence on the bond strength. Thus it may be seen that phenyl methyl
sulfide having a bond energy of 335 kJ/mole is much less reactive than benzyl
ethyl sulphide with a bond energy of 222 kJ/mole. Similar arguments apply to
disulphides and it is known that dibenzyl disul]~hide reacts more easily than
diphenyl disulphide. Dibenzyl disulphide is considered an EP additive while
diphenyl disulphide is considered an AW additive. In 4-Ball tests EP properties are
characterised by the weld load of balls. Faster reaction results in better protection
of the surface and the welding load will be higher. Short duration tests are
conducted with a fresh set of balls at each load. Other methods used to characterise
EP properties are mean hertz load (MHL), and seizure delay load. Forbes et al [44]
reported that bond dissociatioh energy and EP properties are reasonably correlated
based on tests with a series of sulphur compounds.
Sakurai et al [45] characterised the EP properties by a different approach. In this

approach firstly chemical reactivity was characterised by static reaction over a hot
wire. The film thickness was obtained on the basis of change in electrical resistance
of the hot wire. The film grows with time offering a barrier to further reaction. The
film thickness growth rate follows a parabolic law for the majority of the chemical
additives studied. The relation may be expressed as
Ar 2 =kt (4.11)
where
Ar = film thickness, m
k = rate constant, m2s-t
t = time, s
Table 4.1. Bond dissociation energies in various organosulphur compounds

(Reproduced from Ref. [4] p263 with permission of Dr. K. C. Ludema.)
9 ,, , ,,, ,,, ,,,= ,, , , ,,,,, ,,, , , - _
Bond Structural t ~ e " Dissociation energy, kJ/mole
5 --- $ Elemental sulphur 117-147
RSS- $$R Dimethyl tetrasulphide 151

R$ - - $ R Disulphides 280-289
R - - SR Alkyl monosuiphides 293-306
i SO - C$H7 -- $CH 3 sec-Alkyl monosulphides 280
t - C/brig- $CH 3 t-Alkyl monosulphides 272
CH 3 CH z C H - $CH 3 Vinyl methyl sulphides 218
C6 H .$ $ -- C H3 Phenyl methyl sulphide 251
C6H 5CH2$ - - CH 3 Benzyl methyl sulphide 213
R - - SH Aikyl mercaptans 289-310
i SO-- C3H?-$H sec-Alkyl mercaptans 297
t- C4H9 --$H t-Alkyl mercaptans 289
CH$CH 9 CH-SH Vinyl mercapatans 218
. . - $CH 3 Phenyl methyl sulphide 335
C..6HsCH2--$C 2 H 5 Benzyl ethyl sulphide 222

4.3. BOUNDARY LUBRICATION WITH RF~CTION FILMS 119
By conducting the studies at different temperatures the rate constant can be

obtained at different temperatures. Load carrying capacity was obtained on the
basis of mean Hertz load (MHL). Fig. 4.5 from [45] shows reasonable correlation
between relative reactivity and load carrying capacity. The relative reactivity is the
ratio of k values for additive solution and white oil obtained at 400~ It may be
noted that the reactions were studied with sulphur as well as chlorine compounds.
When mixed additives were used the sulphur additive dominated the effect. Some
earlier investigations also involved reaction studies with metallic powders and
relating the global reaction to the reactivity of the additives. In these studies the
nature of decomposition products was also analysed by gas chromotography and
mass spectrometry and possible reaction routes postulated.
I Sulfur type EP agents

(12) ,
.a. x I/
u 0
0
11), x(13)
I
,,J,
I
uo ~3: 3 -
I
I
I
I
I
>'Z I
ira.
/1111
o / x
A
/A
~o.u2 - 0 17) Chlorine t y p e
o~
I1~r EP agents
.=I~ (4)
~ 1 ~ I
1.2 1.4
Chemical R e a c t i v i t y ,
( Log K )Ep I(Log K)whit e oil
Fig. 4.5. Correlation between chemical reactivity and load-carrying capacity of oils
containing EP agents. (A) sulphur compounds. (.) chlorine compounds. (x) mixed sulphur-
chlorine compounds.
(Reproduced from Ref. [45] by permission of STLE)
120 ('HAPTER 4. BOI~INDARY L UBRICA ]'ION MECHANISMS
The reaction films generated on the surfaces have been analysed with increasing
levels of sophistication. Early detailed work of Godfrey [46] showed that the films
formed with sulphur compounds are composed of both sulphides and oxides. He
also showed that preformed sulphide films had lower load carrying capacity in
comparison to mixed films formed in air atmosphere. The relative proportion of
oxides and sulphides in the film depends on the operating conditions. Also
different types of oxides can be involved in the film. Later work by several authors
[47,48,49] confirms the presence of oxides in surface films. It is also established
that oxides formed increase the load carrying capacity of the sulphur compounds.
Thus the reaction films formed are due to simultaneous action of sulphur and
oxygen competing for the surface. The hot wire reactions of the previous paragraph
should now be qualified. Even for the static reaction the films are ill-defined
combinations of sulphides and oxides, though the overall increase in film thickness
could be expressed by the typical parabolic law of corrosion reactions.
Additives with lower levels of reactivity are unsuitable as EP additives. Some of

them can be used as antiwear additives. The major class of AW additives used are
the phosphorous compounds which are considered next.
4.3.2.2 Phosphorous compounds
The phosphorous compounds used are mainly phosphate and phosphite esters.
These compounds are manly used as antiwear additives. The antiwear function was
evaluated by the scar diameter obtained at 15 kg load in a 4-Ball machine in
several earlier papers [50,51 ]. Based on earlier work there is reasonable evidence
that the mechanism of action of these additives involves hydrolysis of the esters
followed by reaction with the surfaces. The final wear resistant films consist of
phosphate or phosphite films together with oxides. It is also considered that
phosphites provide better antiwear protection at higher loads in comparison to the
corresponding phosphates. Tricresyl phosphate is the most common phosphorous
compound used in this class. In many industrial formulations both sulphur and
phosphorous compounds are used together and in such cases it is difficult to define
the nature of films formed. In phosphates and phosphites a large number of
structures with partial or complete replacement of hydrogen are possible. The alkyl
and aryl groups also can be chosen as desired. The typical structures for fully
substituted phosphate and phosphite esters are (RO)3P=O and (RO)3P
respectively. The reactivity in relation to antiwear behaviour is more difficult to
define. All that can be said is that the effective additives provide optimum level of
reaction to sustain an antiwear film. This is unlike the case with EP additives where
higher reactivity gives better EP action.
4.39 BOUNDARY L U B ~ C A "lION WITH REACTION FILMS 121
4.3.2.3 Chlorine compounds
Chlorine compounds can also be used as EP additives. Their use in industry is

declining due to environmental considerations and the possibility of rusting of
ferrous materials via iron chlorides. Some operations in which they still find
application include severe cutting operations using neat cutting fluids, and stainless
steel wire drawing. The compounds used are alkyl and aryl chlorides. Correlation
of EP activity with C-C1 bond strength was not found to be reliable. On the other
hand reactivity on hot wire as shown in Fig. 4.5 correlated well with EP activity.
The overall mechanism involves decomposition of the chlorine compound and
formation of metallic chloride.
4.3.2.4 Multifunctional additive systems
The individual additives from different classes can be mixed together to obtain
desired antiwear and EP performance. Mixing of two boundary lubrication
additives can alter the functioning of the individual additives both of which now
compete for the surface. This can result in synergism or antagonism of the
additives. These issues were briefly considered in chapter 2. Two systems can be
envisaged. One is the isolated system containing only the EP and antiwear
additives in a base oil. Studies with such systems reported in the earlier literature
are difficult to analyse because different criteria were adopted by the various
authors for evaluation. Some used the standard tests, while others used
modifications of these tests. Some authors found that chlorine and sulphur
additives have a synergistic effect with regard to antiwear and EP functions
[52,53]. With regard to combinations of sulphur and phosphorous additives it is
generally considered that they act synergistically [54].
Additives can be synthesised with two or more active elements in the same
molecule. Thus compounds containing chlorine-sulphur, chlorine-phosphorous,
sulphur-phosphorous are also available. By tailoring the functional groups carefully
required performance equivalent to or better than individual additives can be
achieved. In view of the complexity of the functioning of these additives
experience in practice has a role to play in selection.
Another important class of additives is the zinc dithiophosphates that are

extensively used in engine oils. In view of their importance the mechanism of their
action is being studied over a long period and continues to be pursued even now.
These additives shall be considered under recent developments.
Many industrial lubricant systems are complex and may contain several other
additives like antioxidants, detergents, antirust compounds, and friction modifiers
as considered in the second chapter. These additives in turn will influence the EP
and AW function of the additives. Further complication arises from the fact that
each of these additives can have different structures depending on the
requirements. It appears at present the problems are overcome on the basis of
empirical information. A rough indication of the influence of various additives may
be seen in Table 4.2 reproduced from [55]. A negative sign indicates antagonism
while a positive sign indicates synergism. Exclusory effect refers to one additive
excluding the effect of the other while complimentary effect refers to the
reinforcing action of each other. Graded response refers to the case in which
mixtures of additives of a given function are added to provide effective response
under different operating conditions. For example a mixture of phosphorous
compounds may be used to obtain good AW performance under varied operating
conditions. This table is based on the response of the two component systems. With
multicomponent systems the situation can be more complex.
Table 4.2. Main additive interactions

(Reproduced from Ref. [55] by permission of Leaf Coppin Publishing Ltd. )
Additive pair "Main types of interaction

Antioxidant-antioxi dant Complementary response +
AW/EP-AW/EP Exclusory effects at surface (EP/EP)
Complementary response (EP/AW, AW/AW) +
Graded response (EP/AW, AW/AW) +
EP-friction modifier Exclusory effect at surface
EP-rust inhibitor Exclusory effect at surface
Dispersant/ZDDP Complex formation in liquid
Complex formation at surface +
Detergent/ZDDP Solubilization of AW film
Direct reaction in liquid p h a s e ....... 7
4.3.3 Recent investigations of reaction films
Recent developments include mechanistic studies at a fundamental level and

characterisation of reaction films with sophisticated analytical tools. Such analysis
is also being applied to study additive-additive interactions and to understand
mechanisms of action with newer classes of additives. Sophisticated chemistry is
involved in these studies, which is not discussed here. The coverage only serves to
4. 3. BOUNDARY L UBRICA TION WITH REA C TION H L M S 123
highlight these studies and points to some missing links from the application point
of view.
4.3.3.1 Mechanistic studies
Tribochemical reactions are very complex and a specialised book on tribochemistry

was published in 1984 by Heincke [56]. The complexity arises from the fact that
mechanical activation of the surfaces can lead to a host of processes. These include
lattice and grain boundary effects, catalytic activity of metal surfaces, exoelectron
emission, and thermal excitation. Several interesting examples of dramatic effects
on reaction with mechanical stress have been cited in this book. While directional
influences are well documented very little can be said with regard to specific
application of these ideas. While exoelectron emission is known [24-27] its role in
tribochemistry has been modelled more precisely by Kajdas [57]. His essential
argument is that the low energy electrons generated from the surface due to
rubbing action interact with the additive molecules generating negative ions. The
positive spots on the surface (generated due to the removal of electrons) chemisorb
negative ions. The organic part of the chemisorbed molecule can undergo bond
scission leaving behind the inorganic layer. The earlier model due to Forbes [43]
considered only thermal activation as the source of electron transfer and covalent
bonding. Based on [57] the chemisorption mechanism of monosulphide may be
explained as below:
Surface rubbing generating low energy electrons
Interaction of the monosulphide with electrons as
R-S -R+e --+ R S - + R ~
Chemisorption of RS- on the positively charged metal surface
In the case of disulphide the initial interaction leads to RS- and RS ~ RS ~ in turn
can react with a low energy electron to generate another RS-. Thus increased
chemisorption occurs with the disulphide. It is considered that the chemisorbed
species decomposes eventually leading to iron sulphide. The fate of organic
radicals depends on their nature and environment. The above example is a simple
124 CHAPTER 4. BOUNDARY L UBRICA lION MECHANISMS
case to illustrate the idea and can be extended to other classes of additives.
Reactions can also occur with base fluids and oxygen as discussed in the same
reference. Furey et al [59] have utilised the concept to generate friction polymer
that can be effective in lubrication. Monomers dissolved in base fluid can
polymerise due to reactions initiated by exoelectrons.
One missing link is the mechanism of film growth due to post chemisorption
reactions. This involves diffusion of metal ions and active elements. As it is the
reaction film that is involved in EP and antiwear action, classification of efficacy
only in terms of the initial reactivity is inadequate. The earlier work due to Sakurai
[45] where overall reactivity was characterised in terms of film growth is in the
right direction. These remarks pertain to static reactions. In dynamic situations the
film composition itself is a variable and more difficult to model. The question is
what kind of fundamental investigations will help in modelling the real situations.
This question will be taken up in later chapters.
4.3.3.2 Characterisation of reaction films
The reaction films formed with chemical additives are complex in nature. These
films are being analysed with modem surface analytical tools and there is better
understanding of the films now. Only one example of the reactions involved with
zinc dithiophosphate (ZDTP) is considered here. The typical structure of ZDTP is
RO S s% oR
P P
RO S Zn S OR
where R represents alkyl/aryl group
Both alkyl and aryl groups can be incorporated in the same molecule providing
flexibility with regard to additive response. This class of additives is widely used in
engine oils and they are referred to as multifunctional additives. They have
antioxidant, antiwear, and a level of EP action. Wilmert et al [59,60] elucidated the
nature of films formed in a cam and tappet rig after careful work. The film consists
of inorganic amorphous phosphate deposited over the surface. The reaction
mechanism postulated involves initial oxidation of the dithiophosphates. The
reaction products act as precursors for the formation of the inorganic film. The
phosphate chains interact with zinc complexes and the zinc cations get
incorporated in the polymer. Detailed analysis of the films is possible through
modem tools like secondary ion mass spectroscopy, nuclear magnetic resonance,
4.3. BOUNDARY LUBRICATION WIIH 16~'ACIlON FILMS 125
Raman spectroscopy, and advanced infrared techniques. To understand reaction

mechanisms leading to the final film it is necessary to understand in detail the
interactions in the liquid phase also. It is also of interest to note that the authors
could also clarify the influence of overbased detergent on the phosphate structure.
Such studies are being conducted with other additive systems and in future the
detailed film compositions for several combinations may be available [61,62]. The
discussion here is only indicative and by no means exhaustive. The references cited
are interesting examples of this vast area. Several studies are also being conducted
with regard to synergism and antagonism of mixed additive systems where again
the sophisticated analytical tools are useful [63,64]. Such interaction studies are of
use in formulating oils with the least adverse effects. Yet another interesting
approach to investigations of the mechanisms involves tribological studies under
ultra high vacuum and analyzing the films in-situ as they form [65].
New classes of additives are being synthesized in large numbers and their
performance based on standard tests is being reported. Dithiophosphates where Zn
is replaced by other metals, and organic molecules containing N, S, C1, and P
combinations [66,67] are examples of such additives. Yet another class of additives
being studied is the boron compounds [68]. The future of these additives in terms
of commercial use is yet to be ascertained.
4.4 Surface analysis

Understanding of surface interactions has been made possible by modem analytical
tools. These analytical techniques are evolving continuously. This section is limited
to a brief consideration of three commonly used techniques in the following
paragraphs. Besides surface analysis the modifications that occur in the additives
and base fluids are also important. The commonly used techniques include infrared
(IR) spectroscopy, mass spectrometry, and nuclear magnetic resonance (NMR) for
lubricant analysis. IR techniques are also being applied to study the composition of
surface layers by modification of the conventional techniques. Some publications
available may be consulted for detailed information [69,70,71].
Electron probe microanalysis (EPMA) is a commonly used technique for elemental

analysis. This technique is usually coupled with scanning electron microscopy. In
this technique the electrons that impinge on the surface generate the characteristic
X-rays of the elements present. The instrument then analyses the energy intensity
at the wavelengths that are characteristic to the elements. The technique involves
penetration depths of the order of one micron. Semi-quantitative analysis of the
concentration of the elements is possible by this technique. The technique is less
126 CHAPTER 4. BOUNDARY LUBRICATION MECHANISMS
reliable for elements below carbon and does not provide information on the nature
of bonding between the elements.
The X-ray photoelectron spectroscopy (XPS) is another technique being

increasingly used in tribology. This technique was earlier known as electron
spectroscopy for chemical analysis (ESCA). In this technique the X-rays that
impinge on the surface emit electrons at different energy levels characteristic of
each element. The instrument provides output giving a plot of binding energy and
corresponding intensity. While binding energies are specific to each element, any
shift in the binding energy is characterised by the nature of chemical bonding
involved. This enables analysis of the nature of compounds formed from the
known information of the shifts in reference compounds. Tribological surfaces may
have compounds that are not well defined and in many cases clear identification is
difficult. The typical depth involved is 100 A and hence the analysis is
representative of the top layers of the surface.
Another technique used for surface analysis of the topmost layers is the Auger
electron spectroscopy (AES). The technique is based on the impingement of
electrons with specific energy level on the surface. This leads to interaction of
electrons in different shells and release of an Auger electron from the M shell. The
analysis is based on the energy levels and their intensity. The analysis is limited to
few atomic layers and so is representative of the topmost surface layers. The
technique is well suited to detect lighter elements. Both XPS and AES are capable
of detecting all elements above helium. The AES technique is normally not used
for detecting the chemical bonding involved.
In tribological research XPS and AES are also used for depth profiling. In depth
profiling the elemental composition is obtained as a function of depth. This is
achieved by removing surface layers by ion bombardment at a known rate and
analysing the surface at different stages. This information is of importance to
understand the variations of composition as a function of depth. The techniques
discussed above involve high vacuum and are not suited to analyse organic layers
on the surface. For such an analysis modem IR techniques involving multiple
reflection are well suited. Certain aspects of surface interaction can be
characterised better by Raman spectroscopy which was used in some tribological
investigations.
REFERENCES 127
References
1. F.P. Bowden and D. Tabor, The Friction and Lubrication of Solids, Oxford University
Press, Part I, 1950.
2. P. W. Bridgman, Rev. Mod. Phys. 18 (1946) 1.
3. F. F. Ling, E. E. Klaus and R. S. Fein (eds.) Boundary Lubrication - An Appraisal of
World Literature, ASME, New York, 1969.
series 9, Elsevier, (1985).
5. O. Levine and W. A. Zisman, Physical properties of monolayers adsorbed at the solid-air
interface-I. Friction and wettability of aliphatic polar compounds and effects of
halogenation, J. Phys., Chem. 61 (1957) 1068.
6. O. Levine and W. A. Zisman, Physical properties of monolayers adsorbed at the solid-air
interface-II. Mechanical durability of aliphatic polar compounds and effect of
halogenation, J. Phys., Chem., 61 (1957) 1188.
7. S. Jahanmir and W. Belter, An adsorption model for friction in boundary lubrication,
ASLE Trans., 29 (1986) 423.
8. F. P. Bowden, J. N. Gregory and D. Tabor, Lubrication of metal surfaces with fatty acids,
Nature, 156 (1945) 97.
9. H. A. Smith, The adsorption of nanodecanoic acid on metal surfaces, J. Phys. Chem., 58
(1954) 449.
10. W. Hirst and J. Lancaster, Effect of water on the interaction between stearic acid and
free powders, Trans. Faraday Soc., 47 (1951)315.
11. E.D. Tingle, The importance of surface oxide films in the friction and lubrication of
metals, Trans. Faraday Soc., 326 (1950) 97.
12. S. Mori, M. Sujinoya and Y. Tamai, Chemisorption of organic compounds on a clean
aluminium surface prepared by cutting under high vacuum, ASLE Trans., 25 (1982)
261.
13. I. J. Frewing, The heat of adsorption of long-chain compounds and their effect on
boundary lubrication, Proc. Roy. Soc. London, Series A, 182 (1944) 270.
14. W. J. S. Crew and A. Cameron, Thermodynamics of boundary lubrication and scuffing,
Proc. Roy. Soc. London, Series A, 327 (1972) 47.
series 9, Elsevier, 1985, 229-230.
16. T. C. Askwith, A. Cameron and R. F. Crouch, Chain length of additives in relation to
lubricants in thin films and boundary lubrication, Proc. Roy. Soc. London, Series A,
291 (1966) 500.
17. A. J. Groszek, Heat of preferential adsorption of surfactants on porous solids and its
relation to wear of sliding steel surfaces, ASLE Trans., 5 (1962) 105.
128 ( "HA P 1ER 4. B 0 l~]NDA R Y L r(]BRJ('A l I O N M E C H A N I S M S
18. S. Hironaka, Y. Yahagi and T. Sakurai, Heat of adsorption and anti-wear properties of
some surface active substances, Bull. Japan Petrol. Inst., 17 (1975) 201.
19. C. N. Rowe, Some aspects of the heat of adsorption in the function of a boundary
lubricant, ASLE Trans., 9 (1966) 101.
20. E. P. Kingsbury, Some aspects of thermal desorption of a boundary lubricant, J. Appl.
Phys., 29 (1958) 888.
21. H. Blok, The flash temperature concept, Wear, 6 (1963) 483.
22. Z. Hu, S. M. Hsu and P. S. Wang, Tribochemical reaction of stearic acid on copper
surface studied by surface enhanced Raman spectroscopy, Trib. Trans., STLE, 35
(1992)417.
23. Y. Su, Electrochemical study of the interaction between fatty acid and oxidized copper,
Trib. Int., 30 (1997) 423.
24. G. J. Molina, M. J. Furey, A. L. Ritter and C. Kajdas, Triboemission from alumina,
single crystal sapphire, and aluminium, Wear, 249 (2001) 214.
25. J. Ferrante, Exoelectron emission from a clean annealed magnesium single crystal
during oxygen adsorption, ASLE Trans., 20 (1977) 328.
26. K. Nakayama, J. A. Leiva and Y. Enomoto, Chemi-emission of electrons from metal
surfaces in the cutting process due to metal/gas interactions, Trib. Int., 28 (1995) 507.
27. K. Nakayama and H. Hashimoto, Triboemission, tribochemical reaction, and friction
and wear in ceramics under various n-butane gas pressures, Trib. Int., 29 (1996) 385.
28. C. F. McFadden and A. J. Gellman, Metallic friction: the effect of molecular adsorbates,
Surface Science, 409 (1998) 171.
29. J. N. Israelachvili, lntermolecular and Surface Forces, Second edition, Academic Press
Limited, London, 1992.
30. V Depalma and N. Tillman, Friction and wear of self-assembled trichlorosilane
monolayer films on silicon, Langmuir, 5 (1989) 868.
31. G. Luengo, J. Israelachvili and S. Granick, Generalised effects in confined fluids: new
friction map for boundary lubrication, Wear, 200 (1996) 328.
32. M. Ratoi, V. Anghel, C. Bovington and H. A. Spikes, Mechanisms of oiliness additives
Trib. Int., 33 (2000) 241.
33. D. A. Bonnell (ed.), Scanning Tunneling Microscopy and Spectroscopy-Theory
Techniques and Applications, VCH Publishers, New York, 1991.
34. B. Bhushan, J. N. Israelachvili, U. Landman, Nanotribology: friction, wear and
lubrication at the atomic scale, Nature, 374 (1995) 607.
35. B. Bhushan, Nanoscale tribolophysics and tribomechanics, Wear, 225-229 (1999) 465.
36. N. A. Burnham, D. D. Dominguez, R. L. Mowery and R. J. Colton, Probing the surface
forces of monolayer films with an atomic-force microscope, Phys. Rev. Lett., 64
(1990) 1931.
37. B. Bhushan and A. V. Kulkarni, Effect of normal load on microscale friction
measurements, Thin Solid Films, 278 (1995) 49.
REFERENCES 129
38. A. D. Burman and J. N. Israelachvili, Microtribology and microrheology of molecularly

thin liquid films, in B. Bhushan (ed.), The Modern Tribology Handbook, CRC Press,
New York, 2000, Section 16.
39. N. Binggeli, R. Christoph, H. E. Hinterman and C. M. Mate, Nanotribology at the solid-
liquid interface under controlled conditions, in: B. N. J. Persson and E. Tosati (eds.),
Physics of Sliding Friction, Kluwer Publishers, Netherlands, 1995, 415.
40. V. N. Koinkar and B. Bhushan, Micro/nano scale studies of boundary layers of liquid
lubricants for magnetic disks, J. Appl. Phys., 79 (1996) 8071.
41. K. T. Miklozic, J. Graham and H. A. Spikes, Characterization of boundary films by
atomic force microscopy, 2000 AIMETA Int. Tribology Conf., Sep. 2000, L'Aquila,
Italy, 737.
42. C. G. Slough, H. Ohtani, M. P. Everson and D. J. Melotik, The effect of friction
modifiers on the low-speed friction characteristics of automatic transmission fluids
observed with scanning force microscopy, SAE paper 981099, (1998) 420.
43. E. S. Forbes, The load carrying action of oragano-s.ulphur compounds - A review,
Wear, 15 (1970) 87
44. E. S. Forbes and A. J. D. Reid, Liquid phase adsorption / reaction studies of organo-
sulphur compounds and their load carrying mechanism, ASLE Trans., 16 (1973) 50.
45. T. Sakurai and K. Sato, Study of corrosivity and correlation between chemical reactivity
and load carrying capacity of oils containing extreme pressure agents, ASLE Trans., 9
(1966) 77.
46. D. Godfrey, Chemical changes in the steel surfaces during extreme pressure lubrication,
ASLE Trans., 5 (1962) 57.
47. R. O. Bjerk, Oxygen as an extreme pressure agent, ASLE Trans. 16 (1973) 97-106.
48. T. Sakai, T. Murakami and Y. Yamamoto, Optimum composition of sulphur and
oxygen of surface film formed in sliding contact, Proc. JSLE Int. Trib. Conf. July
1985, Vol. 3, Tokyo, Elsevier.
49. M. Tomaru, S. Hironaka and T. Sakurai, Effect of some oxygen on the load carrying
capacity of some additives, Wear 41 (1977) 117.
50. E. S. Forbes and H. B. Silver, The effect of chemical structure on the load carrying
properties of organophosphorous compounds, J. Inst. Petroleum 56 (1970) 90.
51. H. E. Bieber, E. E. Klaus and E. J. Tewkbury, A study of tricresyl phosphate as an
additive for boundary lubrication, ASLE Trans., 11 (1968) 155.
52. A. Dorinson, The additive action of some organic chlorides and sulfides in the Four-
Ball lubricant test, ASLE Trans., 16 (1973) 22.
53. R. W. Mould, H. B. Silver and R. J. Syret, Investigations of the activity of cutting oil
additives III. Oils containing both organo-chlorine and organo-sulfur compounds,
Wear, 26, (1973) 27.
54. K. Kubo, Y. Shimakawa and M. Kibukawa, Study on the load carrying mechanism of
sulphur-phosphorous type lubricants, Proc. JSLE Int. Trib. Conf. July 1985, Vol. 3,
Tokyo, Elsevier.
55. H. A. Spikes, Additive-additive interaction in lubrication, Lub. Sci., 2 (1989) 3.
56. G. Heinicke, Tribochemistry, Verlag, Munich, 1984.
57. C. Kajdas, Importance of anionic reactive intermediates for lubricant component
reactions with friction surfaces, Lub. Sci., 6 (1994) 203.
58. M. J. Furey, C. Kajdas, T. C. Ward and J. W. Hellgeth, Thermal and catalytic effects on
tribopolymerisation as a mechanism of boundary lubrication, Wear, 136 (1990) 85.
59. P. A. Willermet, D. P. Dailey, R. O. Carter III, P. J. Schmitz and W. Zhu, Mechanism
of formation of antiwear films from zinc dialkyldithiophosphates, Trib. Int., 28 (1995)
177.
60. P. A. Willermet, R. O. Carter III and E. N. Boulos, Lubricant derived tribochemical
films- An infrared spectroscopic study, Trib. Int., 25 (1992) 371.
61. J. Sch~Sfer, P. Rehbein, U. Stolz, D. L/Shr and K.-H Zum Gahr, Formation of
tribochemical films and white layers on self-mated bearing steel surfaces in boundary
lubricated sliding contact, Wear, 248 (2001) 7.
62. S. Dizdar, Wear transition of a lubricated sliding steel contact as a function of surface
texture anisotropy and formation of boundary layers, Wear, 237 (2000) 205.
63. S. S. V. Ramakumar, N. Aggarwal, A. Madhsudhana Rao, A. S. Sarpal, S. P. Srivastava
and A. K. Bhatnagar, Studies on additive-additive interactions: Effects of dispersant
and antioxidant additives on the synergistic combination of overbased sulphonate and
ZDTP, Lub. Sci., 7 (1994) 25.
64. S. Plaza, G. Celichowski and L. Margielewski, Load-carrying synergism of binary
additive systems: dibenzyl disulphide and halogenated hydrocarbons, Trib. Int., 32
(1999) 315.
65. J. M. Martin, Th. Le Mogne, M. Boehm and C. Grossiord, Tribochemistry in the
analytical UHV tribometer, Trib. Int. 32 (1999) 617.
66. C. Guo-Xu, X. Ren-Gen and D. Jun-Xiu, The tribological mechanism of action of
oxovanadium (IV)- dithiphosphate, Lub. Sci., 9 (1997) 307.
67. A. Bhattacharya, T. Singh, A. P. Singh, R. Singh and V. K. Verma, Tribological studies
of some N, S, and CI containing extreme pressure additives, Lub. Sci., 7 (1994) 61.
68. W. Liu, Q. Xue, X. Zhang and H. Wang, Effect of molecular structure of organic
borates on their friction and wear properties, Lub. Sci. 6 (1993) 41.
69. L. J. Bellamy, Infrared Spectroscopy of Complex Molecules, Chapman & Hall, 1975.
70. J. M. Hollas, Modern Spectroscopy, Wiley, New York, 1987.
71. J. Ferrante, Practical applications of surface analytic tools in tribology, ASLE Trans., 38
(1982) 223.
NOMEN( TLA T URE 131
Nomenclature
h r
real area of contact
C concentration of the polar substance in the base fluid
E heat of adsorption
f overall coefficient of friction
fm friction coefficient for metallic junctions
f, friction coefficient for boundary film
F friction force
k rate constant in chemical reaction
K equilibrium constant
N carbon chain number
Sm shear strength of the metallic junctions
Sl shear strength of the film
t time
tr mean time of stay of a molecule at a surface site
tx time needed to move one molecular diameter
to fundamental time of vibration of the molecule
T absolute temperature
To failure temperature
T~ temperature of the surface
W normal load
u sliding velocity
x diameter associated with an adsorbed molecule
Greek letters
Gt fractional film defect

Ar film thickness in EP reaction
AG ~ free energy change
AH ~ heat of adsorption
AS ~ entropy change
(P fractional coverage of adsorbed molecules
133
5. Lubricated wear of metallic materials

Theory and practice
5.1 Introduction
Lubricated wear occurs under mixed and boundary lubrication regimes and
reduction of wear in these contacts has obvious economic advantage. The present
chapter deals with boundary lubricated contacts only. Mixed lubrication, in which
there is partial load support by EHD or hydrodynamic films will be considered in
chapter 8 dealing with fatigue and wear. Lubricated contacts are far more common
industrially and wear control is achieved normally through suitable additives. As
discussed in chapters 2 and 4 wear controlling additives cannot be viewed in
isolation from the rest of the additives in the formulation. Discussion with regard to
wear here assumes that any formulation is already proven to be acceptable with
regard to other perfomaance requirements. The industrial needs in the area can be
broadly categorised as follows:
1. Capability to predict at the laboratory level whether a new formulation

results in lower wear in an existing system.
2. Ensuring performance when design changes are affected. Design changes
usually involve more severe operating conditions.
3. Ensuring wear performance with new designs and materials for which
available experience is inadequate. For example the needs of a ceramic
bearing can be very different from a metallic bearing.
4. In all cases ensure reliability. From a lubrication point of view this means
there should be no catastrophic transitions to high wear or scuffing.
The first section deals with lubricated wear in boundary contacts that are based on
adsorption. The next section deals with the modelling of chemical wear. These
models are viewed in terms of their practical applicability. Such a consideration
identifies the research needs to strengthen the theory-practice interface. The third
section briefly considers the problem of running-in. The empirical modelling of
wear is considered in the next section with an example. The importance of such
modelling has been brought out in the discussion. The problem of lubricant failure
is addressed in the final section. The industrial needs broadly categorised above
form the background for the present and subsequent chapters.
134 CHAPTER 5. L UBRICA TED WEAR OF METALLIC" MATERIALS
5.2 Wear modelling- adsorbed layers
5.2.1 Experimental variables
In this modelling only adsorbed boundary layers are involved. Models can only be
validated by experiment and at the outset it is necessary to consider the role of
variables in experimentation. As in the case of dry wear, wear rate refers to a quasi
steady state situation. Any wear test goes through a stage of running-in until
surfaces adjust to each other. Eventually a steady state is reached in the wear
process. The running-in process is of industrial importance and will be dealt with
later. Completion of the running-in stage even in laboratory machines is a function
of the operating conditions and surface topography. In wear modelling it is
necessary to ascertain that the running-in is complete and the wear rate does
represent steady state.
The second issue in wear modelling is the estimation of contact temperatures.

There is no common procedure in this regard. In some cases overall temperatures
are considered while in other cases asperity temperatures are used. The methods
adapted to calculate temperature rise again differ. An acceptable common
procedure is needed at least for laboratory machines. Specialists in this area have to
come together to evolve such a procedure.
The third issue is the possible interference of EHD or hydrodynamic effects. It is

known that in many test configurations such effects can occur particularly at higher
speeds. Wear may not be uniform over the surface and at microscopic level there
can be convergent wedges leading to hydrodynamic action. This aspect is difficult
to model and can lead to uncertainties in wear modelling.
The fourth issue is the nature of the test configuration selected. As pointed out in
chapter 3 wear depends not only on the operating conditions but also on the test
configuration. Care has to be exercised in generalising wear models obtained from
one test configuration only. The various test configurations are considered in
chapter 7.
5.2.2 Adsorption based models
The basic model available is that proposed by Rowe [ 1]. The model proposed by
Stolarski [2] may be considered as a modification of this model. The problem
treated f'trst is for a single hydrocarbon lubricant. This system is less complex than
5. 2. WEAR MODELLING - ADSORBED LAYERS 135
an additive system where there is competitive adsorption between polar molecules

and the hydrocarbon. First let the wear rate in the lubricated condition be
formulated as follows:
where Kb is the boundary wear coefficient
Unlike dry wear Kb is a function of operating conditions. Rowe considered that for
adhesive wear the wear depends on the extent of metal contact through the
boundary film. Fractional film defect ct can be expressed by Eq. 4.10 considered in
the previous chapter. Thus in boundary lubrication actual metal contact area is
expressed as aAr. Since all asperity contacts do not wear a factor km was
introduced. Rowe defined km as a true dimensionless constant specific to the
rubbing surfaces and considered to be independent of any surface contaminant or
lubricant. On this basis wear rate
Vl:km~
The real area has been modified to consider the growth in contact area. This
modification is derivable from Eqn. 1.4 in chapter 1 which considered the growth in
contact area. The final equation is
v
--~--=km(1 + 3 f z a (5.1c)
Expressing a as per Eq. 4.10 in the previous chapter
V = kmxT Wexp(-E/ (5.2)

t toU I~ ~,RL)
where 7 = (1 + 3 f 2 ) ~
The value of 7 is close to unity since friction coefficient is usually less than 0.15 in
lubricated contacts. The surface temperature rise was estimated with the low speed
equation assuming heat flow to both the surfaces follows stationary source theory.
136 CHAPTER 5. LUBRICA TED WEAR OF METALLIC MATERIALS
The temperature rise was estimated on the basis of the real area. The contact
dimension was based on one equivalent circular area despite the fact that asperity
spots are distributed over the worn area. This will overestimate the temperature rise
at the asperities as can be seen from section 3.2.
The experiments were conducted in a pin-on-disk configuration with hemispherical

copper pin sliding against a steel disk with n-hexadecane as the lubricant. The
maximum load applied was 10 kg while maximum bulk temperature was 66~
The maximum sliding speed used was 0.2 m/s. Taking (kmx/toH) as constant for a
given system
- ~7 exp
l u
where C' is a constant
Experiments were conducted to verify the above by varying one variable at a time
and plotting the relationships. On the basis of slopes obtained E was estimated as
11,700 cal/mole, which is a reasonable value, based on the available literature. The
correlations were inadequate when variations in wear rate were studied as functions
of sliding velocity and temperature separately. The variation with load showed
better correlation. The intercept obtained when wear rate (V/l) is plotted against
(7/u)exp(-11,700/RTr on log-log plot was used to estimate C'. 'x' was obtained
from the available literature. Determination of to .from fundamental considerations
presented problems and two possible to values of 2.0x10 -13 and 2.8x10 -~2 were
estimated from the available theoretical approaches. The latter value was justified
on the grounds that it leads to akm value of 0.23. This is close to the value of 1/3 in
Archard's wear equation (3.15d) in which all junctions lead to wear. This amounts
to an assumption that every adhesive junction leads to a wear particle.
Stolarski [2] proposed a more detailed model for lubricated wear. The author
considered both partial EHD and boundary conditions. He used a different
approach to the boundary lubricated wear and separated the real area (A~) into
plastic (Ap) and elastic (A~) areas giving
A r = Ap + A e
It was considered that adhesive wear should be based on Ap. The model was tested
with ethylene glycol and n-hexadecane as lubricants in a pin-on-disk machine. The
5.2. WEAR M O D E L L I N G - A D S O R B E D L A Y E R S 137
rotating low carbon steel disk was slid against a hemispherical brass pin. The
variables were tested one at a time. The reported correlation between theory and
experiment was good. The author stated the limit of applicability of the equations
was up to a sliding speed of 0.1 m/s for which no reason was mentioned. The
fractional film defect a was obtained by the same procedure as used by Rowe but
using the double exponential expression without simplification. It may be
mentioned that one series of experiments conducted by Stolarski [3] showed that
when sliding speed was changed over a wide range for a fixed sliding distance the
experimental wear volume was nearly constant. The experiments clearly fit
Archard's law with a constant wear coefficient. One of his figures illustrating the
speed influence on wear for brass-steel combination using hexadecane is given in
Fig. 5.1. The validation of the model is inadequate.
Experiment
/
. . . . o. . . . Theory
/o
8O ..... /
/
/
/
70 /
/
o /
/
60 - /
/
/
/
~ 50 -- /
r- /
E /
/
d 40 - /
E /
/
/
0 30
> /
/
/
d 20 . /
/
/
/
f
10 "
0 .................
10 20 30 40 50 60 70 80
SLiding velocity, cm/s
Fig. 5.1. Wear volume Vs sliding distance for a constant sliding distance of 1000 m at 50 N
load. From Ref [3].
Mention may be made of another model for boundary lubrication developed at

IBM mainly for business machines and described by Bayer [4]. The approach is
based on the concept of zero wear. The number of cycles the system can survive
with zero wear is a function of the maximum shear stress in the contact and the
nature of the lubricant. Zero wear is defined as a situation in which the wear is
138 CHAPTER 5. L UBRICA TED WEAR OF METALLIC MATERIALS
confined to the asperity dimension. The model had been successful for the lightly
loaded slow moving contacts involved in such machines. It is difficult to extend the
model to normal wear situations.
5.2.3 Assessment of the adhesive wear model
A detailed fundamental analysis of the model due to Rowe was carried out by
Beerbower [5]. He considered that the heat of adsorption for steel and copper will
be different. Another interesting point raised by him is the probability of contact.
The film defect considered by Rowe was with reference to one surface. If the film
defects for both the surfaces are ct~ and az then the probability of contact will be
a~a2 which will be much lower than either of the individual values. In chapter 4 the
problems involved in assigning a specific value to E have been discussed. It can
vary significantly depending on the nature of oxides and metals. Even with pure
hydrocarbons oxidised polar compounds can form and this can lead to shifts in the
heat of adsorption. The present author considers E to be a variable particularly for
real systems.
Another issue to be considered is the modification done to basic Archard's

equation. The equation as expressed (and ignoring 3') takes wear rate as
proportional to real metal contact area, (aA~). No consideration is given to the size
of the transfer particle. As discussed in section 3.4 the adhesive wear particles
through oxide films will have on average a diameter of (ot~ where d is the
asperity diameter and a is the film defect. It is to be emphasised that this smaller
wear particle will result from a sliding distance of d. The idea can be now applied
to a boundary lubricant by replacing the oxide film with a boundary film with the
same conclusion. There is no oxide now and all the contacts at the defects are metal
to metal. With this concept the wear rate will be
The value of 1/3 in Eq. (3.17) is not used here as km in the above equation
approximates this value. On this basis for a given wear rate the ot values will be
much higher. As an example in a test at 8 kg load [1] the wear rate was 2x10 -8
cm3/cm. The value of a as per Rowe's equation will be 6.7x 10.5 for this case. From
the above equation a will be 1.6x10 -3. This approach is a suggestion based on
physical reasoning. The particles in real situation will have a distribution. So
consideration on the basis of a alone will be approximate. While the suggestion
here is made regarding boundary lubrication the influence of the size factor is
5.2. WEAR MODELLING - ADSORBED LAYERS 139
general and should be considered whenever removal is on a scale lower than the
asperity contact dimension. One additional aspect is the assumption that a
hemispherical particle is involved in dry and lubricated contacts. This is a
convenient way to assess size effects. Even if particles are not hemispherical the
size effect persists though the equation will be modified.
Plastic contact at asperities will be significant for soft materials like copper and
brass used in these tests. It is logical to expect adhesive wear to be more likely with
plastic contacts and so the separation of asperity junctions into plastic and elastic
contacts by Stolarski appears logical. But even in plastic contacts the oxide must
affect the metal to metal contact. This influence was not taken into account by
Rowe or Stolarski. At the other extreme, suppose all contacts are elastic and there
is no metal contact through the oxide films. In such a case the wear will be
governed by the oxide removal through a fatigue process. The oxides are usually
thin in lubricated contacts and the type of equations used for oxidative wear as
discussed in chapter 3 are not applicable. Lubricant affects the wear process by
reducing the extent of oxide contact. Lubricants also reduce frictional traction that
increases the fatigue life. For the intermediate situation the overall wear will be
governed by the extent of metal contact in all the junctions. In case the metal
contact levels are very low the oxide removal rate is expected to govern the wear
rate.
Another question is the actual mechanism operating within the contact zone. The
adsorption model assumes that vacant sites on the surfaces lead to metal/oxide
contacts. Another possibility is to consider that the contacts arise due to local
rupturing of the shearing molecular layer. The film itself has defects, which in turn
influence the local contact. The contact through boundary layers is expected to be a
function of the shear rate and temperature within the film. The modelling in this
case depends on the complex rheological behaviour of the film, which cannot be
predicted at this stage. The rheological behaviour of monolayers and multi-layers
in nano contacts is being modelled at very low speeds and loads as mentioned in
the previous chapter. Any extension to high shear rates will be useful for practice.
This aspect is also of importance with regard to lubricant failure as will be
considered later in the chapter.
Another related problem is the adsorption/desorption phenomenon in the high

pressure contact zone. Unlike the open system where the molecules can adsorb or
desorb from or to the bulk fluid, the monolayers in contact are highly confined.
Adsorption and desorption can occur between the two asperities in which case the
probability of defects will be reduced.
140 CHAPTER 5. L UBRICA FED WEAR OF METALLIC MATERIALS
A more complex relation can be set up when a polar additive is added to the
hydrocarbon base fluid as reported by Rowe [6]. The surface is now covered with
the molecules of base fluid as well as additive. The additive concentration on the
overall surface, 0, is obtained by a relation based on the Langmuir relation given
by Eq. 4.2 in chapter 4. The surface defects for base fluid and the additive covered
zones can be obtained separately on the basis of this equation. It is then assumed
equilibrium establishes itself in the contact zone and the equilibrium constant is
related to the change in heats of adsorption of the base fluid and the additive, and
the overall change in entropy of the system. Only the final equation for the case
where molecular size of base fluid and additive is the same is given below.
exp - E/RTs ( 11 -TIv) .

c= exp(AS~ C c + -7- b to
where
C = additive concentration, mole fraction
w
= wear rate x x/1 + 3 f z and subscripts b and c refer to the base
fluid and additive containing oil at concentration C

AE = difference in heat of adsorption of additive and base fluid
AS ~ = Overall entropy change
to = ratio of fundamental time of vibration of additive and base fluid
molecules
The SI units with regard to the above parameters are given in chapter 4. Rowe used
CGS units in his work. Linear plot between (V/I)'r and the square bracketed
portion can be drawn and from the slope and intercept AE and AS ~ can be obtained
indirectly. Limited testing of the model was done by finding wear rates at different
concentrations of octadecanol, stearic acid, and n-octadecylamine in hexadecane. It
is difficult to comment on this model except to say that it will be subject to the
same problems as in the earlier case. The authors found that wear rates decreased
substantially with additives. Also for a given additive the wear rate decreased with
increasing concentration. As discussed in chapter 4 lubricant failure temperature
increases with the additive concentration as well as the heat of adsorption. These
findings taken together show higher concentrations of the additive are
advantageous from the wear and scuffing point of view as expected.
5.2. W E A R M O D E L L I N G - A D S O R B E D LA YEI~'~" 141
The influence of test configuration has already been discussed at the outset of the
section. These aspects are general and are relevant to testing with antiwear
additives as well. Any evaluation of scuffing properties is also influenced by the
test configuration.
5.2.4 Application of the model
Effective quantitative use of the model is not possible due to the uncertainties
discussed above. The model is inadequate even with a simple hydrocarbon system
in a limited operating range. The model was considered in some detail to appreciate
how fundamental approach was attempted. The interaction of variables on the wear
behaviour is complex and cannot be modelled on the basis of physical chemistry
considerations alone. The way out is to develop empirical relations as will be
considered later. Such relations suffer from their restrictive nature and should be
utilised to develop insight into the mechanisms involved. This in turn can lead to
better scientific modelling. Qualitative considerations based on the above model
are useful. The influence of heat of adsorption based on this model, as well as
earlier observations, is clear. The obvious choice is to use additives with higher
heats of adsorption wherever possible. In the rubbing contact the value of heat of
adsorption is variable due to modifications in the surface and the lubricant. But it
can be argued that it is an advantage to start with an additive having higher heat of
adsorption. The large variation in wear rate with sliding velocity predicted by
theory is doubtful. The increased wear rate with temperature is generally observed
but it is unlikely to be exponential.
The nature of fats and oils used for industrial purposes is varied and they cannot be
defined like pure compounds in fundamental studies. Their use is increasing
because of environmental reasons as discussed in chapter 2. There is already a
large amount of testing activity on potential materials for several applications that
include hydraulic fluids and gear oils. At the other end there is increased use of
synthetic esters due to their superior performance in comparison to mineral oils.
Some of the esters also have better biodegradability, which is an additional
advantage. Most of the testing activity is empirical because formal models
discussed above are inadequate.
Several fat based materials are conventionally used in metal working operations
mainly as stable and metastable emulsions. In metal cutting, wear of the work piece
is to be maximised while at the same time minimising the tool wear and pick-up. In
metal forming like rolling roll wear and pick-up is to be minimised while the
reduction per pass is to be maximised. Another important requirement is the
142 CHAPTER 5. L UBRICA 7ED WEAR OF METALLI(" MATERIALS
surface integrity of the work in terms of brightness, surface roughness, and

freedom from stains. In many of these operations the temperatures are significantly
higher in the contact zone despite water cooling and can easily exceed the melting
points of the soaps. It may be mentioned that some fatty materials and synthetics
have been successful even in the hot rolling of steel. Conventional ideas can only
predict lubrication failure in such cases. This area will be considered in chapter 7
but will be confined to performance evaluation.
5.3 Chemical wear
Chemical wear may be defined as the wear observed with chemical additives that
are intended to react chemically with the surface. In chapter 4 the mechanisms of
action of additives were considered. While these studies provide insight into the
overall behaviour, wear modelling is only possible through an understanding of the
rate of formation and the removal process. To focus attention, only steady state
wear is considered assuming boundary lubrication conditions. The issues related to
scuffing and protection with EP additives will be considered in the final section.
The next part deals with the wear mechanisms. The possibility of modelling with
formal equations is limited and is considered where appropriate.
5.3.1 Wear mechanisms and modelling
Consider two asperities coming into contact. Several layers of molecules consisting
of the additive and base fluid get dragged into the contact and it is assumed that by
the time the maximum contact pressure is reached monolayer surface coverage of
the additive remains on both surfaces. As the contact moves two competing
mechanisms operate. One is the rubbing process resulting in wear, while the other
is the chemical reaction of the adsorbed additive at the higher temperature and
stress conditions in the asperity. Chemical interaction can be thermally activated. It
can also be due to exoelectron emission. If the layers provide adequate protection
against wear the asperities separate with some degree of reaction on their surfaces.
Oxidation can occur due to dissolved oxygen diffusing to the surface and
competing with the additive molecules for surface reaction. As the asperities move
out of contact the temperature decreases. Additional reaction is now possible due to
the easy access of the additive molecules and oxygen as the asperity cools.
Depending on the bulk temperature there can be further reaction at the surface.
This reaction will continue till the asperity again gets engaged in the contact zone.
The chemical interaction zones are illustrated in Fig. 5.2. The surfaces are covered
with the lubricant. The reactions occur at the asperity level and at the bulk
5. 3. CHEMICAL WEAR 143
temperature on the wear track. The processes repeat with each cycle resulting in the
growth of film thickness. As per conventional approach the film detaches when a
critical thickness is reached.
LOAD LOAD
J- I
(a) (b)
~ DISC
EARTRACK
PIN R
(c)
Fig. 5.2. Chemical reaction zones represented by R. (a) Reaction at high asperity
temperature (b) Reaction in the asperity cooling zone and (c) Bulk reaction on the wear
track.
The generally accepted view of chemical wear [7] may be represented

schematically as shown in Fig. 5.3. The wear rate is plotted as a function of
temperature for two additives for a given load and speed condition. Chemical wear,
governed by the reaction films is expected to occur when the film thickness is
adequate, leading to wear within the film itself. In this zone, to the right of minima,
the wear is governed by the chemical reaction rate. Higher reactivity leads to
increased chemical reaction resulting in increased chemical wear. At any given
reactivity the removal and formation rates are balanced and a particular reaction
rate is established. Under conditions where the film formation is inadequate,
adhesive wear is considered to occur in portions of the contact leading to higher

wear. This zone is to the left of the minima. In this zone the wear decreases with
temperature as less metal contact occurs with increasing film formation. At an
optimum temperature the minimum thickness is just enough to sustain chemical
wear. At this point the wear is minimum. Different additives will have different
relationships as shown for additives A and B.
d /
3:
Temperature
Fig. 5.3. Wear rate Vs temperature for additives A and B.
5.3.2 Limitations of the available model
One possible way to model the chemical wear process is to invoke the oxidative
wear model of Quinn discussed in chapter 3. The concept can be applied to the wear
of chemically reacted films. Film wear modelling has two components. One is the
simple expression based on wear coefficient and the other is the estimation of wear
coefficient based on kinetic treatment. These were discussed in section 3.7.2. For
clarity in this sub-section the film wear model refers to the first component while
the kinetic model refers to the second component. In the present case oxidation and
additive reaction occur together at the surface. The chemical reaction will be a
function of the additive concentration. The oxygen availability will be a function of
oxygen solubility. Also oxidation reaction will be a function of additive
concentration since oxygen competes with the additive for surface reaction. For a
given additive concentration it is assumed that the overall film growth is governed
by a parabolic law. This approach is similar to that used for oxidative wear
5.3. CHEMICAL WEAR 145
modelling. Following the approach given in section 3.7.2 and assuming steady
wear, the film wear rate may be expressed by Eq. (3.20b) as follows:
It may be recalled that Kf is the probability of wear and is equal to the inverse of the
number of cycles needed to form a film of given thickness.
Since many real situations involve elastic rather than plastic contacts H needs to be
replaced by the average elastic stress at the asperities. For the present argument H is
retained as the flow pressure. Rowe [7] has tabulated wear coefficients for several
chemical additives based on 4-Ball tests. The overall wear coefficient K/ obtained at
50~ refers to the value as obtained from
where K / is equal to Kf(~/d)
The K / values for good antiwear agents range from 0.1-10.0xl0 8. Consider for the
present purpose a typical value of 108. Considering an asperity contact diameter of
10 ~m and a critical film thickness of 1.0 ~tm the value of Kf will be 10"7. In other
words on average the film thickness grows to 1.0 lam in 107 cycles. If the parabolic
law is recast in terms of cycles
Ar 2 = k nn (5.4)
where n is the number of cycles, Ar is film thickness, and k, is a dimensional

constant
It can be shown that per cycle growth at n = 107 will be as low as 5x10 8 I.tm. This
growth per cycle amounts to 5x10 -4 A. Even at n = 1000 the growth rate per cycle
amounts to 5x10 -2 A. Even if a lower critical removal thickness of 0.1 l.tm and
higher Kfof 10.6 are assumed, the growth rate per cycle amounts to a value less than
an angstrom. These growth rates are physically inconceivable and it is not possible
to apply kinetic model based on parabolic law. The mechanism involved has to be
treated differently.
The above argument is with regard to additive systems used in practice and which
provide low wear. Nakayama et al [8] modelled wear with a copper-steel system
146 CHAPTER 5. L UBRICA TED WEAR OFMETALLIC MATERIALS
using elementary sulphur as the additive. The pin-on-disk machine was operated
with sulphur concentrations ranging from 0.01 to 0.05% by weight and the wear of
the stationary copper pin was monitored. Their results followed the behaviour
schematically shown in Fig. 5.3. The film wear to the right of minima was
modelled kinetically as per the oxidative wear model. It may be recalled that for a
constant critical film thickness In(V//) is proportional to (-Qp/RT~). When In(V//) is
plotted against 1/Tr the value of Qp can be estimated from the slope of the line.
This is a common procedure, to obtain the activation energy. While reasonable
relations were observed by the authors for each concentration, Qp values had a
wide range of 8.9 to 18.2 Kcal/mole. Since Qp is expected to be constant from the
postulated model such large variations are a problem especially because wear rate
is exponentially related to Qp. Similar problems exist with oxidative wear.
Problems like possible variations of critical film thickness, variation in activation
energy and Arrhenius constant, and the accuracy of the temperature estimates by
existing theories is not discussed further. One parameter affects the other and the
problem is difficult to resolve. But one question that arises is the extent to which
such a model can be used. The values of Kf calculated from the available
information in this paper range 102-10 .4 if a critical film thickness of 1.0 ~tm is
assumed. Such high wear rates will be unacceptable for lubricated contacts. This
example is cited to show that in the case of high film wear approximate modelling
is possible. The concepts involved in the model have relevance for situations
involving EP action. In EP action the possibility of scuffing is controlled by fast
reaction with active additives.
5.3.3 Alternative proposal for wear mechanism
As discussed above the wear mechanism cannot be reconciled with the growth and
removal of critical film thickness. One possibility for low Kf in lubricated systems
is the influence of adsorption on the reacted film. This adsorbed layer consisting of
additive and base fluid molecules will reduce contact over the film area just as in
the case of boundary lubrication with adsorbed layers. In such a case the film
contact area is reduced and the wear will be lower. On the other hand many
antiwear additives are known to function effectively above 100~ Adsorption
influences at such temperatures are expected to be lower. In some cases there can
be strong adhering films deposited over the reacted film. The reacted film here
refers to the metal additive reaction. The deposited films can arise from
polymerisation or other reactions within the lubricant system. The other aspect is
the growth of reaction film. It is possible that progressive reaction and slow
building up develops a uniform film with very few defects and pores. Such a film
offers a very strong barrier to reaction restricting film thickness to very low values.
5. 3. CHEMICAL WEAR 147
Removal of such films is likely to be governed by fatigue. Fatigue cycles for

removal will be related to the normal and tangential stresses. Defect free thin films
are expected to survive a large number of cycles. In thicker films cracks can
propagate from the defects leading to earlier failure. Chemical reaction will be
restricted to the small failed zones and may be considered as a repair action that re-
establishes the uniform film. The role of adhering boundary layers is an additional
variable which has to be taken into account. But eventually the reaction films wear
and in so far as their wear is concerned, their mechanical response to cyclic
stressing is of importance.
5.3.4 Assessment of the proposed mechanism
The first issue is whether an assumption can be made that a strong barrier to
reaction can occur with virtually no further growth of the film. As stated by
Molgaard [9] the parabolic oxidation law is applicable when film thickness is
higher than 0.1 I.tm. At lower thickness the oxidation mechanisms are far more
complex. One possibility is that they follow a logarithmic law [10] as expressed
below:
Ar = k~log(att + 1) (5.5)
where 1%is a rate constant and a~ is a constant.
Such an equation predicts a much lower growth rate for the film as a function of
time. Assuming such relations are possible with the additive system the negligible
growth rate of the film is directionally explicable.
A second issue is the evidence with regard to film thickness. Investigation reported
at the laboratory level with zinc dithiophosphate by Vipper et al [11 ] is of interest
in this regard. They studied the influence of copper naphthanate concentration on
the antiwear and EP action of dithiophosphate and observed that an increased
percentage of the naphthanate improved the wear performance. The antiwear films
were analysed and the depth of the chemically reacted film referred to as SS by the
authors was inversely related to the wear. Thus the thinner the reacted film, the
better the wear reducing capability of the film. At the minimum wear rates
observed the films involved were of the order of 50 nm while at the higher wear
rates the films were in the range of 100-300 nm. The SS layer is the zone that
involved diffusion of oxygen and sulphur. They explained the mechanism as being
due to the formation of a complex with the copper compound and reduction of
surface oxidation resulting in thinner films. The authors also took into account the
physically decomposed layer (PD) and the hardened zone due to carbon diffusion
called B layer and considered the SS layer to be of primary importance. The wear
rate at the low SS levels was about 0.1xl0 -3 mm3/km while at the high SS values
the wear rates ranged from 0.5 to 0.7x10 3 mm3/km. Reacted films on engine liner
surfaces analysed by Becker and Ludema [12] also showed reacted films of around
0.1 lam. This film is attributable to the action of zinc dithiophosphte additive and
oxygen in the engine oil. From this evidence it is reasonable to postulate that very
thin films are involved in good antiwear action. While oxygen is involved the
nature of oxides and their role is not well studied in antiwear films. Ludema [13]
considers that Fe304 is the desirable oxide at least in the ring-liner tribology. A
typical example of low SS thickness with copper napthanate at 0.05% copper
concentration and dithiophosphate additive DF-11 is illustrated in Fig. 5.4. The SS
layer is about 20 nm. This diagram is based on depth profiling based on AES
analysis. The thickness is based on the sputtering rate and the time involved to
reduce the concentration of active elements as shown in the figure.
PD SS
ts Fe
4
I
.i/"
0 i .........................
20 30 40 50 GO
Sputtering time, rain
Fig. 5.4. Depth profile of reaction film with zinc dithiophosphate. (Reproduced from Ref.
[ 11] by permission of Leaf Coppin Publishing Ltd.).
The above assessment is satisfying with regard to a workable mechanism for

antiwear additives. At least with one class of important additives there is support for
the postulate that thin films offering strong barrier to reaction provide good
antiwear action. But it is known that glassy deposited films are involved with
5.3. CHFA4ICALWEAR 149
dithiophosphates. As their role in the wear process is not clarified, it amounts to an

assumption that overall wear rate is influenced by the extent of sulphur and oxygen
diffusion into the metal. The argument may be extended to the general class of
antiwear additives many of which do not form deposited films. The equilibrium
film thickness will be related to the nature of additives involved in a given system.
But it is expected that for a given system the film thickness that offers a maximum
barrier to reaction corresponds to best antiwear action. The proposed approach is
apparently similar to the conventional film wear concept shown in Fig. 5.3. But the
important difference is that in the proposed mechanism wear is not governed by
growth and removal of a film with critical thickness. It is governed by cyclic
stressing and removal of the film which is a fatigue process. On this basis the
fatigue removal of the reaction film is the rate determining step. The role of
chemical reaction is one of re-establishing the thin reaction film at the worn zones.
The overall film thickness involved is very low and is usually less than 0.1 pm. The
response of the system to the operating conditions will be different and is no longer
predictable on the basis of the conventional model. If there are thin deposited films
they may offer a further barrier to reaction.
5.3.5 F u n d a m e n t a l considerations
The proposed mechanism is logical but based on limited data from literature.
Fundamental investigations are necessary to establish the mechanism. One key
issue is the barrier nature of the films. Direct evidence with regard to this aspect
should be sought. The other aspect of importance is the study of wear particles.
These issues are discussed below.
5.3.5.1 Nature of barrier films
It is necessary to develop direct evidence of the barrier nature of a film on the wear
scar. This will be referred to as post wear film. An example of such a study in EP
lubrication by Sethuramiah et al [14] may be cited. When tests were conducted by
step-load procedure in a 4-Ball machine with diphenyl disulphide (DPDS) and
elementary sulphur the failure load was higher for DPDS in comparison to sulphur.
The detailed methodology is given in the cited reference. Normal testing involving
fresh sets of steel balls at each load gives a higher failure load for sulphur and is
attributed to its higher reactivity. It was postulated that in step-load tests the
lowering of surface roughness with the less reactive additive provided partial
hydrodynamic load support. Such smooth surfaces can be a result of lower
reaction. The point of interest here is a comparison of the scars obtained in step-
load tests with both the additives under similar conditions. This unpublished work
150 CHAPTER 5. L UBRI('A TED WEAR OF METALLIC MATERIALS
is cited here. The sulphur content on the wear scars obtained at 100 kg was first
assessed by EPMA. The surfaces were then reacted in static conditions under argon
atmosphere with 0.294% sulphur solution under identical conditions for 30 minutes
at 180~ The increase in sulphur content with the sulphur scar was 87% in
comparison to 26% for DPDS scar. It is of interest here to note that when the
reaction is conducted in normal air atmosphere the sulphur content decreased by
22% for the DPDS scar while it increased by 51% for the sulphur scar. This is an
example of competitive reaction of sulphur and oxygen depending on the nature of
surface films. This means films formed with DPDS offer a stronger barrier to
reaction. Better techniques like AES can now be used to study the reactivity of post
wear films more precisely. Modem electrochemical techniques may also be useful
for this purpose. Thus direct investigation of the barrier nature of the films is
feasible and is worth pursuing. Such studies can be more easily conducted on
laboratory specimens, but some reference surfaces from real systems will be useful
for comparison.
The mechanical characterisation of the post wear films is hardly attempted so far.
Extensive research is being conducted on the adherence and durability of coatings
by scratch tests and other techniques. Adaptation of these techniques to study
antiwear films will be valuable. Such characterisation will form a useful link
between the chemical and mechanical nature of the films.
5.3.5.2 Analysis of wear particles
A wear particle is directly related to the wear process. Wear particles are analysed
for their morphology and size range in oil condition monitoring. Oil condition
monitoring refers to the assessment of equipment condition by periodically
examining wear debris. Such an analysis detects the onset of malfunctioning by
observing the changes in particle size distribution and their nature [15]. Special
techniques like ferrography [16] are used to separate the particles size wise.
Elemental analysis supplements the study of particles. For example a sudden
increase in copper content can mean impending failure of the copper based
bearings. Detailed analysis of the wear particles for their structure and composition
is what is of importance in wear mechanisms unlike the global composition studies
conducted in condition monitoring. Such studies are conducted in great detail on
worn surfaces but there are only few studies reported on wear particles in
lubricated contacts. On the other hand detailed analysis of wear particles is quite
extensively reported for dry wear. One interesting study [ 17] pertains to the study
of wear fragments obtained with zinc dithiophosphate additive in a steel-cast iron
system. The flakes consisted of amorphous regions in which very small crystalline
5. 3. CHFA4ICAL WEAR 151
iron sulphide particles were observed. The amorphous region contained zinc, iron,
phosphorous, and oxygen but no detectable sulphur. The amorphous region was
likely to be glassy phosphate containing iron and zinc. Such studies open up
interesting possibilities. For example is the iron sulphide obtained due to partial
contact through glassy films? Does the absence of oxygen in crystalline material
support the current hypothesis [ 18] that iron oxides are 'digested' in developing
glassy films? Such broadening of knowledge is possible by a detailed analysis of
wear particles. Basic investigations of this kind coupled with surface analysis will
clarify wear mechanisms much better. The available detailed information on
surface films is not effectively linked to the removal process involved. These links
can be established through wear particle analysis. The problems involved in
developing this kind of technology should not be underestimated. Wear particles
may be clustered, structures may not be uniform, and wear particles may be mixed
from the initial nmning-in to later film wear. The analytical tools needed for
detailed study may not be generally available and detailed analysis can be
undertaken only by specialists. With the available technologies in condition
monitoring, including on-line monitoring, it should be feasible to provide
representative samples for a given condition. Also, as is the situation in any
technology, there will be limits within which one has to work. Even with this
scenario a lot of progress that is relevant should be feasible. If such attempts are
not made trial and error approaches that are time consuming and limited in scope
have to continue.
5.3.6 Practical aspects
The proposed mechanism lays emphasis on the characterisation of post wear films
and wear particles. Besides helping in eventual modelling such characterisation is
of direct practical relevance. Comparison of films in the laboratory tests and real
systems will be useful in establishing the operating conditions in a machine that
correlates with practice. In many cases it is impractical to simulate the real systems
in the laboratory. This is because there are a large number of tribological systems
operating over a wide range of operating conditions. To make a laboratory rig more
widely applicable the zones of applicability to real systems should be identified.
This identification is possible through the analysis of post wear films and wear
particles as suggested above. The idea is to simulate the wear process instead of the
test rig. Becker and Ludema [12] studied ring-liner tribolgy in a laboratory
reciprocating test. It is of interest here to note that the validity of the simulation test
was based on a comparison of wear particles and surfaces obtained in engine tests
and the simulator. Morphology was the main aspect considered. As stated by them
the operating conditions of an engine are not simulated and they relied on the
152 CHAPTER 5. LUBRI('A TEl) WEAR Ot=METALLI(" MATERIALS
simulation of the wear process. Detailed analysis of wear particles and post wear
films as suggested above will strengthen such simulations.
The complexity of chemical reactions is not unique to tribology. Many chemical

reactions in real systems cannot be modelled with laboratory information alone.
Catalytic reactions, for example, in hydrocracking are very complex. Industrial
realisation of such processes involves different levels of empiricism. Solutions
emerge because of the large scale R&D effort at different scales. This effort is
justified because of the significant economic benefits accruing from such
technologies. The suggestions made to broaden the understanding of the additive
action are worth pursuing. The extent to which investigations can be conducted
depends on economic considerations. To start with, investigations with engine oils
can be economically justified. These oils are consumed in large quantities and
improvement of engine life is an important consideration.
5.4 Running-in
Running-in refers to the adjustment of newly assembled components under
controlled conditions. Freshly assembled surfaces are not well matched and can
have misalignment at the micro level. This results in non-uniform loading with
some zones highly stressed in comparison to others. If the design load is applied
right in the beginning there can be large plastic deformation at some patches. Such
patches can result in large scale adhesion leading to scuffing and failure of the
components. To avoid this situation the main strategy employed is to run the
system at moderate operating conditions to sta~. with, and progressively adjust the
surfaces to each other. The severity on the components is gradually increased to the
design levels. If the running-in is incomplete there is a possibility of failure in
service. While well run-in components do not fail there is always a possibility of a
small percentage of failures. This is because of the statistical nature of micro level
misalignments with some odd components unable to run-in effectively with the
prescribed procedure.
The procedures adopted for running-in are empirical and based on experience.
Running-in is of importance in bearings, gears, engines, and other components.
Major interest is in the ring-liner contact of engines where contact conditions are
severe. Criteria to assess the completion of the running-in are usually based on the
final expected roughness reached. The characterisation of roughness itself can have
varied levels of sophistication, particularly with reference to engine liners. Some
limits also apply with regard to the quality of the surface based on microscopic
observation. These criteria as suggested in some research papers [19,20] as well as
5. 4. R U N N I N G - I N 153
manuals of engine manufacturers are empirical and based on experimental

observations. One different and interesting approach reported by Joseph and
Raman [21 ] is based on determining the variations in compression pressure during
operation. The authors showed that maximum compression pressure is reached
when running-in is complete. The pressures were measured by a piezo electric
sensor. The running-in done in the industry basically completes the initial and
difficult phase. The completion of running-in is gradual and may take easily more
than hundred hours in an engine.
The industry is interested in reducing the time needed for running-in. They are also
interested in well defined criteria to decide the completion of running-in. Yet
another need is the procedure for running-in. Empirical methods have their
problems. When a design is changed and the operating conditions are more severe
the criteria have to be re-established by trial and error, which is a time consuming
process. This also applies to situations where surface coatings or materials are
changed. The available answers from tribologists are limited in the area. Some
partial answers will be attempted in the relevant parts of the text.
The general nature of the running-in process followed by a steady state is illustrated
in Fig. 5.5. The overall process may be described by the following equation [22]
9----(9 0 -Vs)e -bt -t-9 s (5.6)
V0
i
3: " ~ " ~ " " ~ _ ~ . . . . a . _ w _
Time (hours)
Fig. 5.5. General nature of running-in behaviour. The two curves represent two different 'b'
values given in the equation.
154 CHAP TER 5. L UBRI CA TED WEAR 0 F M E TAL L IC MA TERIA L S
where "Q,'Q0,'Q~ represent wear rate at time t, wear rate at t = 0, and steady state
wear rate respectively. The rate is expressed as wear volume per unit time.
The two curves represent two different values of the exponents b involved. The
running-in in these curves is satisfactory in the sense that the process gradually
changes to steady state. If the running-in stage is ineffective there can be large
fluctuations in this zone with the associated risk of scuffing. One theoretical model
is available which looks at the process only in terms of change in roughness [23].
As running-in is a complex process with several processes of adjustment, it is
better to concentrate on the overall model as expressed by the above equation.
5.4.1 Modelling running-in and steady state wear
The complexity of wear mechanisms has been discussed earlier. Empirical

modelling is a practical way to study the influence of various parameters on the
wear rate. It is firstly necessary to quantify the running-in and steady state wear
precisely. Once this procedure is available the influence of operating variables can
be studied systematically by designed experiments. Rajesh Kumar et al [24] have
recently reported such a methodology. The concepts and the final results are
discussed here. Numerical procedures used are given in detail in the reference and
not given here. The procedure starts with the reasonable assumption that the
running-in and steady state wear follow the exponential relation given in Eq. 5.6.
Integration of this equation with the boundary condition V = V~ at t = oo and V = 0
at t = 0 we obtain
V = a(1 - e-bt)+'Qst (5.7)
where
a= and V represents wear volume at time t

b
In the selected reciprocating test wear volume was obtained as a function of time.
The value of b was obtained by an iterative procedure such that l1- R I is less than
lxl 0 -4 where R is the coefficient of determination.
The methodology leads to a proper statistically based determination of the running-

in and steady state wear rates.
The experiments were conducted in a reciprocating tester with an EN 31 steel ball

sliding against a flat of the same steel. Tests were conducted at a constant frequency
5. 4. R U N N I N G - I N 155
of 50 Hz and a stroke of 1.0 mm. Commercial engine oil with a viscosity of 129.9
cSt at 40~ and 13.3 cSt at 100~ was used. The oil contains zinc dithiophosphate
as antiwear additive and the zinc and phosphorous contents were 742 and 1890 ppm
respectively. Sulphur content was not determined. The variables selected were
Load 20N, 40N, 60N

Roughness (Rq)0.35 ~tm, 0.55~tm, 0.75 ~tm
Bulk temperature 50~ 100~ 150~
The roughness refers to that of the disk. The experimental design consisted of one-
third fraction of 33 factorial design. Each test was of eight hours duration with ten
steps. At each step the wear scar on the ball was measured from which the wear
volume was obtained with known equation that takes elastic recovery into account.
The disk wear was determined after the final stage by measuring the scar
dimensions. The surface temperature rise was estimated on the basis of geometric
area only and was added to the bulk temperature to obtain the contact temperature.
Estimation on the basis of asperity temperatures was not done in this case.
Firstly data from each experiment were fitted to a regressed curve as discussed
earlier. The nature of fit observed is given for one case in Fig. 5.6. From this
equation the necessary wear parameters were calculated. The next step was to
obtain the empirical equation that fitted all the experiments with suitable exponents
on the variables by statistical methods. The final equation obtained for steady state
wear of the ball is given below:
ws -- 1.3 7x 10-4 p 0.74R q0.98 Tc-O.79 (5.8)
0.0003
0.00025
ID 0.0002
---e-- Experimental
0.00015
O ------ Theoretical
~" 0.0001
~: 0.00005
0 =IF ~ T I I
0 2 4 6 8 10
Time (hours)
Fig. 5.6. Comparison of experimental wear volume with the theoretical (regressed) curve
obtained at 20 N load, 0.35 lain roughness and 150~
where
W$ = steady state wear rate, mm3/hour
P = load, N
Rq = rms roughness, gm
Tc = contact temperature, ~
This equation shows that the steady state wear was clearly influenced by roughness,
and temperature as well as load. It is of interest to see that the exponent of
temperature is negative. Thus wear rate decreases with temperature in this case.
Steady wear rate also decreases as the initial roughness decreases. The strong
influence of initial roughness was unexpected as the final scar roughness was nearly
the same for all three initial roughness values. It is normally considered that initial
roughness mainly affects the running-in part of the wear only. In fact the
investigation done with regard to roughness had a practical aim of assessing
whether initial roughness influenced steady state wear. This is of importance in
engines with regard to the life of the liner. The empirical relation shows the system
specific wear behaviour and the need for such modelling. None of the effects can be
predicted by the existing theories and the response to wear is specific to the system.
While the empirical relationships are of relevance to practice, study of post wear
films and wear particles under different operating conditions is necessary to
strengthen the theory-practice interface.
Another matter of importance is to decide when the running-in is complete. A

possible practical criterion is to consider that running-in is complete when 95% of
wear rate at a given point equals the steady state rate. The running-in period was
obtained on this basis and was related to the variables considered. Similarly the
initial wear rate was also related to the variables involved. These relationships are
not given here. It is of interest to note that the running-in time varied from 0.65 to
4.42 hours depending on the operating conditions within a total run of eight hours.
In repeat tests it was observed that the repeatability of initial wear rate is poorer in
comparison to the steady state wear rate. This may be attributed to the variability
involved in the initial wear of the point contact. Thus arbitrary criteria used to
determine steady state wear in laboratory machines are unacceptable. It may be
argued that it is impractical to conduct long duration tests. But then one has to at
least keep in mind that significant errors are possible in short duration tests. Such
realisation will help in ameliorating the present procedures as will be discussed in
chapter 7. It may be noted that in the present situation the running-in involves
basically a change over from point contact to area contact leading to steady state. In
real systems running-in refers to micro level adjustments. The present situation is
5. 4. R U N N I N G - I N 157
treated as a running-in process in the sense that the evolution of wear depends on
the initial contact conditions that include load, temperature and roughness. The
methodology developed here is general and is applicable to wear study in any
machine and test geometry. The empirical relationship amounts to a wear map. The
observed relationships can be represented graphically where the influence of
parameters can be effectively visualised. An unpublished example utilising Surfer
7.0 software is given in Fig. 5.7. Wear rate as a function of roughness and
temperature at a load of 20 N is shown in (a), while (b) shows the wear behaviour in
terms of intensity ranges. The wear rates are given in mm3/hr. For comparison
purposes the overall wear coefficient K / will be useful. The K / values ranged from
7.19xl 0 -9 to 3.9xl 0 -8 in these experiments. These relatively low wear coefficients
are typical of antiwear action. It may be observed that distinction at such low wear
rates is effective through the adapted procedure. Limited analysis of films indicated
their thickness ranged from 0.06 to 0.12 lam that is again typical for this additive.
No detailed film analysis was carried out. The empirical relations can also lead to
more realistic approaches to wear mechanisms. For example the influence of
temperature on steady wear rate may be reconciled with the possibility of more
protective glassy films as temperature is increased. The wear of the underlying
reaction film will now depend on its composition and the extent of contact through
the glassy film. Further progress is possible only by detailed film studies as
discussed earlier. The conventional model would predict increasing wear with
exponential temperature dependence. The analysis was not done in terms of
absolute temperature, but it is clear from the relationships that the reality of wear in
the present system is altogether different. This is unlike the case for the empirical
dry wear model discussed in section 3.7.2. In this case wear rate depended
exponentially on the absolute temperature.
5.5 Failure of boundary lubrication

Boundary lubricant layers will fail when conditions are severe enough. This results
in significant oxide/metal contact leading to large scale adhesion. Such a transition
leads to what is commonly referred to as scuffing. From a practical point of view
scuffing may be defined as a situation resulting in unacceptable surface damage and
constitutes a failure of the component. In some cases the scuffing may be severe
enough to lead to seizure. It is convenient to start the consideration of scuffing with
a mineral base oil and then go on to a consideration of the influence of additives.
(a)
1,oi/ , /,
]40]/ /
~ot ~ /
10ft / ~ --
0.4 0.5 0.6 0.7

Roughness (Rq)
Fig. 5.7. 3-D representation of wear behaviour: (a) Wear rates as a function of roughness
and temperature at 20 N load and (b) Wear behaviour shown in intensity ranges.
5.5. FAILURE OF BOUNDARY LUBRICATION 159
5.5.1 Scuffing with mineral oils
In boundary lubrication the failure of a mineral lubricant will result in a large scale
contact on the thin oxide films at the asperities. The oxide thickness was that
obtained under lubricated conditions with relatively low shear stresses and limited
contact through films. Also the oxygen availability was limited to the amount
dissolved in the lubricant. When lubrication fails there is high friction and
correspondingly high shear stress leading to the removal of oxide layer. While
oxidation should increase d tae to higher temperatures, it cannot keep pace with the
removal rate leading to melallic contact and scuffing. Hence the hypothesis that
lubrication failure leads to scuffing is reasonable with mineral oils. It is common to
refer to 'lubrication failure' as 'lubricant failure' and this terminology is adopted
from now on. The prediction of lubricant failure is at present difficult. One
possibility is that failure occurs when film defect reaches a critical value. It may be
recalled from chapter 3 that this criterion was used to model failure temperature of
fatty acids and other polar compounds as a function of concentration. However the
actual failure temperatures observed were generally low in these tests. For example
the failure temperature observed by Frewing [25] for 1% stearic acid in white oil
was as low as 60~ Grew and Cameron [26] observed scuffing temperatures
ranging between 50 and 160~ for hexadecylamine in n-hexadecane for different
concentrations and loads. On the other hand large number of researchers have
reported that mineral oils alone without additives have failure temperatures around
150~ [27,28]. As pure hydrocarbons should fail at lower temperatures it is
considered that higher failure temperatures are due to sulphur and oxygen
containing impurities in the base oil. There is also clear evidence of the role of
polar impurities in fuels. One important problem faced in aircraft engines was the
failure of piston pumps observed with hydrotreated aviation turbine fuels briefly
considered in chapter 2. The fuel pumps are lubricated by the fuel itself and the
impurities act as boundary lubricants. Their removal through hydrotreatment
created the tribological problem of high wear and in some cases seizure. It was
difficult to understand the relative importance of different impurities which
included polycyclic aromatics, oxygenated compounds, and sulphur compounds
[29]. The present solutions include change of piston pump materials as well as
incorporation of ppm level of additives. The additive approach is limited to
military aircraft. Similar problems exist with low sulphur (<0.05%) diesel fuels
with the fuel pumps prone to high wear and scuffing problems [30]. In all such
cases test methods at laboratory level become necessary. Such procedures, which
are empirical, take a long time to develop. Validation is attempted by a comparison
between the test parameters in the developed rig and fuel pump. The laboratory test
procedures are being developed on a trial and error basis [31 ].
The criterion for mineral oil failure first proposed by Blok [32] and used with
modifications to date is the idea of critical failure temperature. He considered
lubricant failure occurs at around 150~ which is the sum of the bulk and flash
temperatures. The flash temperature considered was the maximum temperature. For
rolling/sliding contacts the equation proposed by Winer [33] in SI units is
1.11fwlu,-u2l 1
Tf = 1/2 1/2 t/2
(5.9)
(~1 U1 "4" f12U2 ) lo
where
Tf = flash temperature rise ~
f = friction coefficient
(0 = instantaneous width of band contact
W = instantaneous load in conjunction
l = instantaneous contact length
U I,U 2 ---- instantaneous velocities of surfaces 1 and 2 tangential to the
conjunction zone and perpendicular to the band length

fl ~,2 = thermal contact coefficients of 1 and 2 with fl = ~/k 9 c where k,
p and c are the thermal conductivity, density and specific heat
This equation considers the maximum temperature rise based on Hertzian contact
dimensions. The equation follows the same methodology as used to calculate the
temperature rise over the geometric area with the difference that the heat is now
flowing into the two surfaces which are both moving relative to the source. Moving
source theory is applied to both the surfaces assuming Peclet numbers exceed 5.0
for both the surfaces which is normally the case for gears. This failure criterion
ignores the effect of viscosity. Instantaneous values are considered to accommodate
the general case of line contacts with variable curvatures.
The main interest of Blok was the scuffing of gears that involve rolling/sliding
contacts. Such contacts involve EHD lubrication and any failure should involve the
failure of the EHD film resulting in the boundary regime and eventual scuffing.
Hence viscosity which determines the film thickness must be an additional
parameter to be taken into account. Other issues involved are the estimation of bulk
temperatures and friction coefficient. A large effort in this area has finally led to
gear scuffing criteria that are easy to use. These are discussed by Enrrichello [35]
and briefly considered here. The contact temperature is expressed as
Tc = Tb + Tf
5.5. FAILURE OF BOUNDARY LUBRI('A TION 161
where Tb and "If refer to the bulk and flash temperatures
The flash temperature rise is obtained on the basis of Eq. (5.9) taking into account
the load sharing between gears. The friction coefficient is obtained empirically on
the basis of surface roughness. The bulk temperature that is applicable is again
obtained empirically on the basis of inlet temperature at a selected speed.
The recommended failure temperatures in ~ are as follows
T s = 146 + 59 ln(V40) for mineral oils without anti-scuff additives and (5.10a)
"Is = 245 + 59/n(V4o) for the oils containing anti-scuff additives (5.lOb)
At a fundamental level the issue to be resolved is the failure criterion for thin films
taking into account the roughness effects. For this purpose effective modelling of
mixed lubrication is necessary. The complexity involved in mixed lubrication shall
be considered in chapter 8. Some models are available which consider that scuffing
is essentially related to EHD film failure [35,36]. The concept is that when film
thickness is reduced to a critical value the asperity contact through such films leads
to failure. These models are not considered here in detail. As boundary lubrication
effects are not known such models based on EHD film thickness alone are not
complete. It may be argued that film thinning is a necessary but not sufficient
criterion for scuffing. The model has been successful when mineral oils of different
viscosities were tested in FZG gear rig [36]. The success can be due to the fact that
the critical temperature for thinning are higher than the adsorption related failure
temperatures of mineral oils. As stated earlier the mineral oils have typical failure
temperature of about 150~ under boundary lubrication conditions. The failure
temperatures ranged from 120~ to as high as 420~ depending on the oil viscosity
in the gear tests. With some exceptions these temperatures are higher than
boundary failure temperature. It is difficult to separate the relative influence of film
thinning and asperity contact. Understanding shear failure of both boundary and
thin EHD films without asperity contact can form a useful basis for such a study.
This can be done only with surfaces of nano level roughness. Such studies on film
failure are not available.
Scuffing is considered rather simply as that process which results in unacceptable

surface damage. Scuffing has been defined by the Institution of Mechanical
Engineers [37] as "gross damage characterised by the formation of local welds
between surfaces". OECD [38] defined it as "localised damage caused by the
occurrence of solid-phase welding between scuffing surfaces, without local
melting". Ludema [39] considered scuffing to be due to roughening of surfaces by

plastic flow whether or not there is material loss or transfer. The possibility of local
welding at asperity level is always there due to adhesion and is invoked in
modelling adhesive wear. Gross damage leading to scuffing can only occur when
adhesive growth in contact area propagates to a level that amounts to gross transfer
of material. In some cases there can be an increase in the weight of one of the
surfaces due to material transfer when scuffing occurs. The exact mechanism by
which a transition occurs from local (asperity) level to gross damage is not well
understood and continues to be an area of investigation. In dry scuffing tests a
recent carefully conducted work [40] suggests that scuffing of aluminium alloys is
due to sub-surface plastic failure. Yet another work conducted in vacuum in dry
conditions [41] and low temperatures considered that there is a critical growth of
the real area beyond which scuffing occurs. Relevance of these studies to
lubricated contacts is difficult to predict at this stage. Different definitions arise as
there is no clearly accepted model for scuffing. At present the practical way out is
to accept qualitatively that scuffing leading to gross damage does occur beyond a
level of severity. Experimentally based semi-empirical models offer the best
guidelines available to predict scuffing.
5.5.2 Control of scuffing- EP additives
The role of EP additives and the generally accepted mechanisms have been already
covered in chapter 4. When mineral oils fail to lubricate and there is a tendency to
scuff the EP additives take over. At elevated temperature they react swiftly forming
a reaction film. This reaction film prevents metal-metal contact and hence avoids
scuffing. This is schematically illustrated in Fig. 5.8 in which the arrows indicate
scuffing load. This can be observed in any machine that can be loaded to the level
of EP film failure. Increasing temperature here is a consequence of increasing
severity of operating conditions. The influence of polar compounds which can
provide some additional protection in comparison to mineral oil is also illustrated in
the diagram. It may be seen that the EP additive becomes active at a threshold
temperature when it reacts effectively on the surface. This means the additive will
interact only when necessary and thus avoid unnecessary reaction and wear. It is of
interest to consider what happens beyond initial film formation. As the temperature
increases due to severity the reaction rate should increase causing increased film
wear as discussed in the section on chemical wear. The assumption here is that
unlike the case with antiwear additives the strong reaction leads to the conventional
film wear where relatively thick films wear out and the overall reaction follows the
parabolic law. It is also possible that films may change and offer an increased
5.5. FAILURE OF BOUNDARY LUBRICATION 163
z iiiii
[] 0,4
i-
0,3 !......
r-l
i----
z
0,2
t.~
L
Ua
rl
U 0.1
:
....
/
Tr
0 .....................................................................................................................
TEMPERATURE
Fig. 5.8. EP action of additive (III) with threshold reaction temperature T, as compared with
base oil (I) and fatty oil additive (II). Arrow represents scuffing.
barrier to reaction reducing the wear rate. Finally the severity can reach such a level
that EP films fail. The failure is normally attributed to an imbalance between the
removal rate and the reaction rate. Neither of them can be modelled with any
certainty. As increased temperature should increase the reaction rate exponentially
it is difficult to prove such a hypothesis on the basis of reaction rate in one direction
alone. Also there should be an adequate reason to explain the change over to a high
removal rate. The possibility is that at high temperatures, significant film
modification mechanisms set in, altering the film composition at the surface and
sub-surface level. This can cause a weakening of the film leading to a removal rate
that is higher than the rate of formation. Formal approaches based on solid state
chemistry can clarify these mechanisms. If the film does not fail by this mechanism
it may eventually fail by melting. Another possibility is the softening of the
underlying metal due to high temperatures leading to film collapse [14]. The
possibility of film melting is difficult to envisage for sulphide-oxide films as iron
sulphide has a melting temperature in excess of 1000~ The additive concentration
is another factor of importance. It is known that load carrying capacity increases
with concentration up to a point beyond which the influence becomes negligible.
The net reaction rate reduces with lower concentrations and can lead to lower
failure load. Another complicating factor is the nature and extent of oxidation that
changes the film composition. The iron oxides that form are highly complex and
can be a combination of several oxides. While it is known that oxides are a
necessary component for load carrying capacity their role has not been studied in
164 CHAPTER 5. L UBRI('A TED WEAR OF METALLIC MATERIALS
detail. A recent study shows that dissolved oxygen concentration can influence
significantly the load bearing capacity of dibenzyl and diphenyl disulphides [42].
The mechanisms are complex and it is difficult to establish any guidelines in this
area.
One important qualitative aspect has to be considered in relation to EP action. This

is related to the rate at which severity is increased. When the rate is low the films
will have a particular composition. When the rate is high the films are ill
conditioned and will have different composition. Scuffing condition for the two
cases will be different. This translates into different severity levels for failure and is
not brought out in the schematic diagram. These considerations are of importance in
EP evaluation and will be discussed in chapter 7.
The above discussion shows that EP action of an additive cannot be modelled

effectively. Progress in understanding is possible only through a detailed study of
worn films up to a stage prior to scuffing in the labojratory machines. Reaction rate
studies on the post wear films can provide realistic estimates of kinetic parameters.
These studies can also clarify how the EP films evolve in progressive running and
their role in failure.
Antiwear additives can also provide an increased load carrying capacity as

compared to base oils alone. These additives are slow acting in comparison to EP
additives. Some of these additives are called mild EP additives. Wherever necessary
a combination of EP and antiwear additives is used. Their mechanism can only be
discussed in general terms as one of combined EP and antiwear action.
Scuffing control is also possible through material modifications. One example is

sulphide-nitride coatings obtained by simultaneous diffusion of nitrogen and
sulphur. Another example is piston rings for which a range of coatings are available
that include chrome plating, molybdenum coating, and gas nitriding. One important
role of piston ring coatings is to control scuffing during running-in in addition to
wear control. Detailed consideration of tribological coatings is available in several
sources. A recent review [43] may be consulted for information.
5.5.3 Practical aspects
The practical formulations in industry for EP and antiwear action can be

combinations of one or more additives. The basic technology involved is to have
the required EP activity while at the same time controlling the wear. Industrial
systems operate under varied conditions. EP action will be needed only when there
5.5. FAILURE OF BOUNDARY LUBRICA IION 165
is an excursion to severe operating conditions. For the rest of the time it is the low
wear that is important. Any excessive action of the EP additive can lead to large
wear that is unacceptable. This balancing is achieved by controlling the activity
levels of the additives involved. From a consideration of the antiwear mechanism
proposed earlier, effective films offer a strong barrier to chemical reaction. Such
films can interfere with quick reaction of EP additives when needed. The author
has not come across detailed mechanistic studies of this aspect in the literature.
One example of a complex formulation is the lubricant used for hypoid rear axles
that involve a high degree of sliding. Strong EP additives control the high speed
and shock load scuffing tendency. Another need is to control the ridging and
rippling under low speed and high torque condition due to plastic deformation.
Phosphorous additives are used to control this problem. It is also to be noted that
EP additives cannot be used in some systems. For example EP additives cannot be
used in engine oils due to their strong corrosive action on copper based bearing
materials and other components. Only the slow acting dithiophosphate has been
found to be a suitable additive. The scuffing control during running-in is hence
more difficult and the variety of surface coatings used assist in the control of
scuffing. The possible control due to the slow acting dithiophosphate cannot be
relied upon.
The modelling possibility of EP action is limited. The present approach is to

evaluate scuffing in laboratory machines and this is considered in chapter 7. Such
empirical testing has its limitations and EP action can only be ensured through final
evaluation in the real system. It is also to be realised that the additives generate
reaction products during use and their influence on the overall system also needs
careful consideration. Environmental issues related to disposal of such products is
also becoming more important.
References
1. C. N. Rowe, Some aspects of the heat of adsorption in the function of a boundary

lubricant, ASLE Trans., 9 (1966) 101.
2. T. A. Stolarski, A system for wear prediction in lubricated sliding contacts, Lub. Sci., 8
(1996) 315.
3. T. A. Stolarski, Adhesive wear of lubricated contacts, Trib. Int., 12 (1979) 169.
4. R. G. Bayer, Prediction of wear in a sliding system, Wear 11 (1968) 319.
5. A. Beerbower, A critical survey of mathematical models for boundary lubrication, ASLE
Trans., 14 (1971) 90.
6. C. N. Rowe, Role of additive adsorption in the mitigation of wear, ASLE Trans., 13

(1970) 179.
7. C. N. Rowe, Lubricated wear, in M. B. Peterson and W. O. Winer (eds.), Wear Control
Handbook, ASME, New York, 1980, 143-160.
8. K. Nakayama and T. Sakurai, The effect of surface temperature on chemical wear, Wear,
29 (1974) 373.
9. J. Molgaard, A discussion of oxidation, oxide thickness, and oxide transfer in wear,
Wear, 40 (1976) 277.
10. S. A. Bradford, Fundamentals of corrosion in gases, ASM Handbook, Vol. 13, 1987, 61-
76.
11. A. B. Vipper, A. K. Karaulov and O. A. Mischuk, New data on the mechanism of
antiwear action of zinc dithiophosphates in lubricating oils, Lub. Sci., 7 (1994) 93.
12. E. P. Becker and K. C. Ludema, A qualitative model of cylinder bore wear, Wear, 225-
229 (1999) 387.
13. G. C. Barber and K. C. Ludema, The break-in stage of cylinder-liner wear: A correlation
between fired engines and laboratory simulator, Wear, 118 (1987) 57.
14. A. Sethuramiah, H. Okabe and T. Sakurai, Critical temperatures in EP lubrication,
Wear, 26 (1973) 187.
15. M. Lukas and D. P. Anderson, Laboratory used oil analysis methods, Lub. Eng., (1998)
31.
16. W. W. Seifert and V. C. Westcott, A method for the study of wear particles in
lubricating oil, Wear, 21 (1972) 22.
17. M. Hallouis, M. Belin and'J. M. Martin, The role of sulphur in ZDDP induced reaction
films formed in the presence of ZDDP: Contribution of electron spectroscopic imaging
technique, Lub. Sci., 2 (1990) 337.
18. J. M. Martin, C. Grossiord, Th. Le. Mogne, S. Bec and A. Tonck, The two-layer
structure of Zndtp tribofilms: Part I: AES, XPS and XANES analyses, Trib. Int., 34
(2001) 523.
19. A. V. Sreenath and N. Raman, Running-in wear of compression ignition engine: Factors
influencing the conformance between cylinder liner and piston ring, Wear, 38 (1976)
271.
20. P. Pawlus, A study of the fuctional properties of honed cylinder surfaces during
running-in, Wear, 176 (1994) 247.
21. K. C. Joseph and N. Raman, Fractal characterisation of running-in behaviour of an IC
engine piston ring and cylinder liner combination, in Harprasad (ed.), Proc. Second
Int. Conf. On Industrial Tribology, Dec 1999, Hyderabad, India, 247.
22. M. Zheng, A. H. Naeim, B. Walter and G. John, Break-in liner wear and piston
assembly friction in a spark ignition engine, Trib. Trans., STLE, 41 (1998) 497.
23. Y. Z. Hu and K. Tonder, Application of a dynamic system model for running-in, Proc.
Int. Conf. On Wear of Materials, ASME, 1991, 201.
REFERENCES 167
24. R. Kumar, B. Prakash and A. Sethuramiah, A systematic methodology to characterise

the running-in and steady state wear process, Wear, 252 (2002) 445.
25. I. J. Frewing, The heat of adsorption of long-chain compounds and their effect on
boundary lubrication, Proc. Roy. Soc. London, Series A, 182 (1944) 270.
26. W. J. S. Crew and A. Cameron, Thermodynamics of boundary lubrication and scuffing,
Proc. Roy. Soc. London, Series A, 327 (1972) 47.
27. R. M. Matveevsky, The critical temperature of oils with point and line contact
machines, Journal of Basic Engineering, ASME, 89 (1965) 754.
28. E. F. Leach and B. W. Kelly, Temperature, the key to lubricant capacity, ASLE Trans.,
8(1965)271.
29. J. Appeldoorn and W. G. Dukek, Lubricity ofjet fuels, SAE paper 660712 (1966) 428.
30. D. Wei and H. A. Spikes, The lubricity ofjet fuels, Wear, 111 (1986) 217.
31. D. Cooper, Laboratory screening tests for low sulphur diesel fuel lubricity, Lub. Sci., 7
(1995) 133.
32. H. Blok, Theoretical study of temperature rise at surfaces of actual contact under
boundary lubrication conditions, Proc. Inst. Mech. Engrs., London, 2(1937) 471.
33. W. O. Winer and H. S. Cheng, Film thickness, contact stresses and surface
temperatures, in M. B. Peterson and W. O. Winer (eds.), Wear Control Handbook,
ASME, 1980, 121-139.
34. R. Enrrichello, Friction, lubrication and wear of gears, in P. Blau (ed.), Friction
Lubrication and Wear Technology, ASM International, 1992, 535-545.
35. S. C. Lee and H. S. Cheng, Correlation of scuffing experiments with EHL analysis of
rough surfaces, J. Trib., ASME, 113 (1991) 318.
36. J. Castro and J. Seabra, Scuffing and lubricant breakdown in FZG gears Part I.
Analytical and experimental approach, Wear, 215 (1998) 104.
37. Memorandum on Definitions and, Symbols and Units, Proc. I. Mech. E., 4, 1957.
38. Glossary of Terms and Definitions in the Field of Friction, Wear and Lubrication,
OECD, Research Group on Wear of Materials, Paris, 1969, 3.
39. K. C. Ludema, A review of scuffing and running-in of lubricated surfaces, with
asperities and oxides in perspective, Wear, 100 (1984) 315.
40. T. Sheiretov, H. Yoon and C. Cusano, Scuffing under dry sliding conditions-Part II:
Theoretical studies, Trib. Trans., STLE, 41 (1998) 447.
41. Q. Quyang and K. Okada, A study on the quantitative description of the seizure
behaviour of steels at low twmperature in a vacuum, Trib. Trans., STLE, 41 (1998)
301.
42. T. Murukami and H. Sakamoto, Effect of dissolved oxygen on lubricating performance
of oils containing organic sulfides, Trib. Int., 32 (1999) 359.
43. K. Holmberg, A. Mat-thews, and H. Ronkainen, Coating tribology-contact mechanics
and surface design, Trib. Int., 31 (1998) 107.
Nomenclature
a an integration constant
al constant
b non-linear coefficient having inverse relation with running-in
period
/~I ~f12 thermal contact coefficients of I and 2 with/5' = ~/k 9 c where k,
p and c are the thermal conductivity, density and specific heat
& elastic part of real area
A~ plastic part of real area
A~ total real area
C' constant
C additive concentration, mole fraction
E heat of adsorption
H hardness
k~ rate constant in logarthimc growth law
km wear coefficient for metal contact as defined in the text
k. rate constant in parabolic growth law adapted to number of cycles
K~ overall wear coefficient in chemical wear
boundary wear coefficient in adsorption based models
Kf film wear coefficient (inverse of number of cycles needed to form
critical thickness)
l sliding distance
l instantaneous length in line contact
n number of cycles
P load
Qp activation energy
R coefficient of determination
Rq rms roughness
t time
to fundamental time of vibration of a molecule
t; ratio of fundamental time of vibration of additive and base fluid
molecules
u sliding velocity
bulk temperature
NOMENC LA T( IRE 169
L critical contact temperature

TT flash temperature rise
Z threshold reaction temperature

temperature of the surface
V wear volume
9 wear rate expressed as wear volume per unit time at time t
"Vo, wear rate at time zero and steady state
UI,U2 instantaneous velocities of surfaces 1 and 2 tangential to the
conjunction
W load
W instantaneous load in conjunction
Ws steady state wear rate expressed as wear volume per unit time
X diameter of area associated with an adsorbed molecule
Greek Letters
fractional contact area as applicable

instantaneous width of band contact
7 =(1 + 3 f 2 1 ~
AE difference in heat of adsorption of additive and base fluid
Ar film thickness
AS ~ overall entropy change
171
6. Wear of non-metallic materials
6.1 Introduction
The use of non-metallic materials in tribological applications is of growing

importance. The major classes of materials used are polymers and ceramics and the
main types used have been covered in chapter 1. Another class of materials being
developed is metal matrix composites in which the metal matrix is modified by
incorporating fibres and/or particles. The investigations being done in these areas
are very extensive and several specialized books are available, some of which were
cited in chapter 1. Both dry and lubricated wear of polymers and ceramics are
covered in this chapter. Coverage of the material in one chapter is necessarily
limited. The scope of the chapter is confined to a broad appreciation of the
mechanisms involved and the issues related to applications. Relatively more
emphasis is laid on lubricated wear.
The first section considers the main wear modes of commonly used polymeric
materials and composites in dry wear. The second section deals with the lubricated
wear of polymers. This section includes the application aspects and observations on
the lubricated wear. The observations are meant to comment on the missing links
using the available information. The dry and the lubricated wear of ceramics form
the third and fourth sections. The lubricated wear of ceramics again includes sub-
sections dealing with the application and observations on lubricated wear.
Sequencing of the sub-sections is different for polymers and ceramics. Abrasive
wear is excluded from the coverage.
6.2 Dry wear of polymers
Polymer tribology has been covered at an introductory level in section 1.5.2. To

recall the main polymers used are high density polyethylene (HDPE), polyamide
(Nylon 6-6), polyoxymethylene (Acetal), and polytetrafluoroethylene (PTFE). The
repetitive units from which these polymers are made are shown in Fig. 6.1. In this
figure the repetitive unit is the same for normal low density polyethylene (LDPE)
and HDPE. The major difference between these two polymers is with regard to the
molecular chain structure. In HDPE the molecular chains are linearly ordered while
t
172 CHAP TER 6. WEAR OF NON-ME TAL L IC MA TERIA L S
in the case of LDPE the chains have a degree of random orientation. The
polyetheretherketone (PEEK) and polyimide are also included in this figure. These
two polymers are used for high temperature applications. In particular PEEK is
gaining importance for such applications. Wear of pure polymers is considered first
followed by polymer composites.
Polymer Repetetive Unit Melting point

f. . . . . -1
i HI H
I
i H
I /
I__ C w C ~ l - - l C u ~130eC
Polyethylene I I I ,
i N H '. H
L.. . . . . -J
r. . . . --1
I H I H
Polyoxymethylene Iiw CI I - - /C__ 0 ~175~
( acetal ) ~ 0 -- I
~_.J
H 0 0
I I I
Polyhexomethylene - - N - - ( C H 2 1 6 - N -- C - 1 C H 2 ) 6 - C ~-- 260eC
o d i p o m i d e (nylon 6-6) I
H
Polytet ro~ I u o r o e t h y l e n e L_ C C ~'--C --C --

r --~ 330-C
(PTFE) IlL_. FI. . . FI / I FI FI
0
n
c , ~ o - -
Polyet herether Ketone . _ ~ ~ ] ~ =~ 3 5 0 "
(PEEK) 0
o ~o
Polyirnlde --N~c
"Co C-" 0
-Q --400%
Structures of C o m m o n Thermoplastics.
Fig. 6.1. Polymer structures.
6.2.1 W e a r o f p u r e p o l y m e r s
The main mechanism postulated for polymer wear may be called the transfer
mechanism. This mechanism refers to the situation where the counter face is a hard
surface in comparison to the polymer. Usually polymer applications involve a hard
6.2. DRY WEAR OF POLYMERS 173
metallic counter face and so the majority of studies were done with smooth metal
counter faces and in some cases against glass surfaces. Polymer-polymer tribology
has also been studied, though to a lesser extent. The transfer mechanism involves
the following sequence:
1. As sliding starts polymer material is transferred to the counter face.

2. With further sliding the transferred material gets oriented in the direction
of sliding. Surface roughness helps in anchoring the transferred layer.
3. The transferred material is removed after further sliding as loose debris.
The debris particles can also agglomerate resulting in large clustered
particles.
4. There can be back transfer to the pin that can influence the above
processes.
The main difference between the various polymers is in the nature of transfer.
Polymers with smooth molecular profiles tend to have low friction and thin transfer
films as discussed in section 1.5.2. Thus both HDPE and PTFE transfer thin films
onto the counter face. The wear of pure PTFE is high because the films are weakly
attached to the counter face. In comparison the wear of HDPE is much lower.
Nylon and other branched chain polymers involve lumpy transfer and relatively
high friction. Their wear resistance depends on the strength of adhesion to the
counter face. For example the wear of LDPE is much higher than nylon due to its
poor adherence. The above observations are directional and generally accepted. At
a fundamental level many questions remain. The detailed crystalline structures of
the polymers are firstly very different. For example PTFE has a banded structure
with alternating crystalline and amorphous regions. HDPE has a spherulite
structure. Such aspects have to be taken into account for a fundamental
understanding of transfer. The removal process also needs to be explained
effectively from basic considerations. Fundamental studies in these areas are
continuing and the interested readers can consult the available literature [ 1,2,3,4].
There is evidence that HDPE can undergo fatigue wear [5] beyond a certain
number of stress cycles. Thus other wear modes besides transfer are also possible.
Some work reported on PTFE wear in a three pin-on-disk machine by Agarwal et
al [6] showed structural changes. These tests were conducted at speeds of 0.73 and
1.47 m/s at two different loads of 44 and 74 N. The wear particles showed higher
crystallinity as compared to the bulk polymer based on X-ray diffraction. On the
other hand the overall pin surface showed lower crystallinity as assessed by
multiple internal reflection (MIR) spectroscopy. The original polymer had a
crystallinity of 85% and the increase in crystallinity of wear particles was up to
174 CI-IAPIER 6. WEAR OF NON-METALLIC MATERIALS
5%. The reduction in crystallinity at the surface was up to 8%. On the pin surface
in some localised zones the removal process was probably due to fatigue.
Increases in crystallinity of the order of 30% in wear particles were found for ultra
high molecular weight polyethylene (UHMPE) by Marcus et al [7] in sliding tests
with stainless steel using distilled water as the lubricant. Several other examples of
structural changes at the surface and subsurface are available with different
polymers. These variations complicate the understanding of wear mechanisms.
Wear rate for polymers is defined in terms of the specific wear rate volume/(sliding
distance x Load) and has the SI units of m3/Nm. Attention may be paid to the units
as sometimes the wear volume is expressed in mm 3. This approach is adopted
because hardness is ill defined for viscoelastic materials. A dimensionless wear
coefficient, commonly used for metals, is rarely used. Data on wear rate for
polymers is recorded by several authors with different machines and operating
conditions. Such data can be used for rough screening of candidate materials. Some
examples will be considered in the next section.
Quantitative wear models are limited for polymers. Jain and Bahadur [8] modelled
wear on the basis of fatigue. Fatigue was related to the cyclic tensile stresses acting
at the asperities. Fatigue is one of the mechanisms, and such modelling can be
effective only when fatigue wear is dominant. There is evidence for subsurface
fracture and removal of thick flakes observed by some researchers. Modelling in
such cases has been attempted by fracture mechanics approaches [9]. A more
recent effort to model wear is based on dimensional analysis [ 10].
The roughness influence of the counter face has been studied by several
researchers. The wear is minimum at an optimum roughness below and above
which the wear tends to increase. Such behaviour is also observed with regard to
friction. Czichos [11] and several others proposed that the higher wear and friction
with increasing roughness is due to the increased abrasive wear of the polymer.
The increasing friction and wear below an optimum roughness is more difficult to
explain. The reason advanced is that with very smooth surfaces the transferred
layers are more easily removed. This is because of the lack of mechanical
anchoring provided by the roughness. The optimum roughness is in the range of
0.2 to 0.4 ~m. When the roughness is optimum, the roughness is just covered by
the transferred polymer with minimum possibility for abrasion. With metal-metal
contacts also when roughness is very low, a sudden increase in friction and damage
can occur. Unlike the case with polymers this effect is due to strong adhesion. The
effect was experimentally demonstrated by Hirst and Hollander [ 12] for boundary
lubricated steel surfaces. Strong adhesion occurred when the surface roughness cr
6. 2. D R Y WEAR OF POLYMERS 175
was very low with typical values of less than 0.02 ktm. This effect is probably due
to the increased interatomic forces between the very smooth surfaces.
In polymer friction, unlike in metallic contacts, the roughness of the hard counter
face is unlikely to change. High peaks that can be approximately characterised by
10- point height (or other measurements) may persist and continue to abrade the
polymer. 10-point height refers to the distance between the average height of the
five highest asperities and the five lowest valleys in a selected sampling length.
The sharpness of asperities, as determined by slopes is also of importance since
sharper asperities will cause more abrasion. In the case of polymer composites the
influence of fillers on roughness also has to be taken into account.
Polymer-polymer friction has been investigated by Czichos [11]. He found that

work of adhesion was reasonably correlated to friction for several polymer pairs.
As work of adhesion can be found more accurately for polymers this is direct
evidence of the role of adhesion on friction. With metallic surfaces it is difficult to
define adhesion as discussed in chapter 1 and such correlations are not possible.
Most of the research in dry polymer wear was conducted with sliding contacts.
More attention is now being paid to other situations that involve rolling/sliding and
impact [13,14]. Significant attention is also being paid to abrasive wear of
polymers. One important material that is emerging for low wear applications in this
area is PEEK.
6.2.2 Wear of polymer composites
Polymer composites may be defined as those materials in which particles and/or

fibres of different types are incorporated in the polymer matrix. Such fillers serve
two purposes. Firstly the fillers can increase the strength properties. Many fillers
also improve the tribological properties. In some cases polymer-polymer
composites can also be used. One example is the blending of PTFE with PEEK to
reduce friction. The composites may be classified into particulate and fibre
composites. Fibre composites consist of reinforcing fibres. These composites are
receiving increased attention. The normally used fibres include carbon, glass, and
aramide. The composites can have randomly oriented short fibres or directionally
oriented long fibres. The performance of such composites depends on their
orientation with respect to the sliding direction. The stress distributions within the
fibres and at the fibre-polymer matrix are very important in the wear behaviour of
these composites. The developments are available in the literature and some
selected references may be cited [ 15,16,17,18].
176 CHAPTER 6. WEAR OF NON-METALLIC MATERIALS
The most extensively used composites are PTFE composites and emphasis will be
placed on these materials. A wide variety of fillers are used that include bronze,
graphite, MoS2, lead sulphide, copper sulphide, and copper oxides. The particles
can have a wide range of sizes and morphologies. In many cases more than one
filler is used with the desire to control friction as well as wear. A wide variety of
materials are commercially available. The fundamental question is what filler
should give the best performance. Unfortunately the action mechanisms of fillers
are not well understood. The possible mechanisms for wear reduction postulated
may be classified as follows:
1. Development of filler enriched layers on the polymer as well as on the

counter face. The fillers in the enriched layers support part of the load.
2. The polymer transfer mechanism is inhibited by the particles at the surface
and in the bulk leading to low wear of the polymer matrix.
3. Inhibition of subsurface crack growth that prevents thick material transfer.
Blanchet and Kennedy [19] have provided good evidence for the inhibition of
subsurface crack growth for filled PTFE at higher sliding speeds. While there is no
doubt that these mechanisms are important their relative importance depends on the
operating conditions. Another important issue is why some fillers are more
effective than the others. One interesting observation based on surface analysis is
that the fillers were found to be effective when they react chemically with the
counter face [20,21 ]. It is argued that such a reaction will result in strong adhesion
and provide a wear resistant surface film. In the overall wear process several other
aspects have to be taken into account. It may be considered that chemical reaction
is one of the requirements for wear reduction. The number of fillers and their
combinations are endless and the idea of chemical reactivity can help in choosing
the fillers.
Zhao and Bahadur [22] recently attempted to generalise the concept of chemical
reactivity on the basis of free energy change involved in a reaction. It is known that
a chemical reaction is feasible when the free energy change is negative. The
possibility of reaction increases as the negative free energy increases. The authors
have evaluated the available experimental work, and showed that when chemical
reactions occurred the free energy change was negative. When reaction did not
occur and the filler was ineffective, the possible chemical reaction between filler
and metal had a positive free energy change. This idea strictly applies to
equilibrium reactions. Nevertheless this is a reasonable approach and provides a
criterion for choosing fillers. It is worthwhile to pursue this idea with a detailed
6.2. DRY WEAR OF POLYMEtLS" 177
analysis of the reaction kinetics. Such solid state reactions are also of importance in
EP lubrication as considered in the previous chapter.
The friction and wear behaviour of composite materials as well as pure polymers
varies in a complex manner with operating conditions. The viscoelastic behaviour
is a function of strain rate and temperature. Theoretical prediction of friction is
available for rubber at low sliding speeds [23] and is not considered here. It is
difficult to make theoretical predictions on this basis for polymers particularly
under severe operating conditions. Reliance has to be placed on the experimental
observations. The wear behaviour can best be appreciated from the available
literature. One typical example of such a study by Holmberg and Wickstrom [24] is
considered here. They have reported the friction and wear behaviour of 22 different
materials that are commercially available. Only wear tests are considered here and
the wear behaviour is reproduced in Fig. 6.2. The wear was measured in terms of
the depth of pin wear in 1000 h. The wear tests were conducted with a multi-pin-
on-cylinder tester that measured wear of 12 different pins simultaneously at a
sliding speed of 0.1 m/s. The polymers were grouped into five classes that included
both pure and filled materials. The polymer groups tested were polyamides, PTFE,
polyethylenes, polyesters, and polyacetals. The polyethylene used was UHMWPE
with a molecular weight of more than 106. The composites used were commercial
materials. The tests were conducted at ambient temperature as well as at a low
temperature of-35~ From this figure the large influence of fillers can be clearly
seen. Filled PTFE and polyethylene stand out as low wear materials. This fact
coupled with their low friction makes them an important group of materials for
tribological applications. The wear of pure PTFE can easily be a thousand times
higher than the filled materials and so rarely used in the pure form. It is also seen
that the low temperature wear behaviour can be very different from the
performance at ambient temperature. The difference between the individual
polymers in each group is also evident. For example polyamide 66 (nylon 66)
showed higher wear than nylon 6. The filler materials studied here are those that
are commonly used. Some oil-impregnated polymers were also tested and showed
low wear.
The purpose of this example is to show the typical wear behaviour of a wide range
of polymers and their composites at a given set of operating conditions. Such
studies are directed towards selection of polymers and composites based on their
friction and wear behaviour. While available knowledge on mechanisms is helpful,
it is not adequate to select materials for specific applications. Several such studies
are reported in the literature and the author has selected one example where a large
number of materials were evaluated.
178 CHAPTER 6. WEAR OF NON-METALLIC MA TER1ALS
6.3 Lubricated wear of polymers
The interest in this section is confined to the influence of conventional liquid

lubricants on wear. Solid lubricants like graphite are sometimes used in composites
and their influence on wear is not separately considered here. Polymer wear with
conventional lubricants has not received much attention. One area that has been
extensively studied is the lubrication of UHMWPE used in artificial human joints
with water, saline, and serum. Two excellent reviews by Unsworth [25] and
Dowson [26] cover the tribological and clinical aspects of these joints. The present
chapter does not cover this area but will consider some aspects of this lubrication in
chapter 8. The first part of this section considers examples of studies on lubricated
wear. Based on this information some observations only could be made regarding
mechanisms. It may be observed that the mechanistic studies with liquid lubricants
are limited. The next part of the section deals with practical aspects of lubricated
bearings.
)
' 9
)0
| ,
20
/ -
30! ,
4t'
i. POtYA.mES 1. P A 6
2. PA 66
3 PA 6 cost
4. PA 6 2 .__.M
S.PA 6 6 - M o S 2 j ~ ] - ~
6 RA 66*corbon : 1
fibre O T:+22~
7. PA 66.cortX:m r J ~I=-3s'c
8 PA66cost+MoS~
9. PA cost~Oil ;~1
~zPOLYT~IRA '0. RT~E n o t measured
FLUOROETHYLENES
11. RIFE+ grophite
12. PT FE. glass
13. PTFE + bronz~
!3- POLYETHYLENES 16.UHMWPE
15. REX
16.UHMWPE*MoS 2
18 PBTP ///////ZI
_19. PBTP +gloss ~,; > ? / / A
5. POLYACETALS 20. POM H
21 POM Co.
22. POM Co..int. lub
L 9 1 , l 9 i 9
0 10 20 30 40
WEAR, AH (ram/1000h)
Fig. 6.2. Wear for polymer materials at normal room temperatures and low ambient
temperature (p = 5 MPa, ~ = 0. l ms-I). (Reproduced from Ref. [24]).
6. 3. L UBRICA TED WEAR OF POL YMEIL'~" 179
6.3.1 Experimental observations of lubricated wear
Bramham et al [27] have reported on the influence of contaminants on aircraft

plastic bearing liners. The PTFE linings consisting of aramid fabric with different
kinds of resins were the materials tested. The tests were conducted in line and area
contact. Some liner materials also incorporated glass fibres. The counter face
material was stainless steel with a roughness of 0.05 gm. In line contact they found
water was particularly deleterious at the high stress level of 60 MPa with regard to
wear, while the wear with a hydraulic mineral oil was slightly lower than the dry
case. They also found that at lower stress levels the wear rates for both fluids are
lower than the dry case. The friction coefficients were always lower than the dry
case. There was a time effect also with water and the large increase was observed
after several hours of operation in line contact. The comparative wear rates with
area contact were lower with the lubricants.
Dickens et al [28] have studied the speed effects on polymer wear with polyphenyl
oxide (PPO), PTFE, and PEEK. The authors used a series of polydimethyl siloxane
fluids of different viscosities as lubricants. The important observation was that in
all cases partial hydrodynamic effects occurred and increased with speed. They
persisted even at the low sliding speed of 10.3 m/s. The hydrodynamic effects were
higher in the line contact. The friction coefficients ranged from 0.01 to 0.1 for PPO
and PEEK while they ranged from 0.01 to 0.05 for PTFE. In general friction and
wear decreased with sliding speed for all materials. The variations in friction in
area contact were lower and though wear clearly decreased with sliding speed they
could not be related to the friction variations in a systematic manner. For area
contact the friction coefficients ranged from 0.07 to 0.1 for PPO and PEEK while
for PTFE they ranged from 0.02 to 0.04.The reduction in the lubricated wear rate
was 3-4 orders of magnitude compared to the dry wear at the highest sliding speed
of 1.0 m/s. The importance of this work is the clear evidence of hydrodynamic
effects. Such effects arise from macro as well as micro deformations. Hence
increased hydrodynamic effects are to be expected in line contacts as observed
experimentally. The issues related to hydrodynamic effects on wear will be
discussed in chapter 8. It may be noted that the term 'hydrodynamic effect' is used
in a general sense and covers both hydrodynamic as well as elastohydrodynamic
effects.
Another example of lubricated wear studies by Sethuramiah et al [29] may be cited.

In this work the friction and wear of pure and graphite filled PTFE were studied in
a three pin-on-disk machine. Graphited PTFE was a commercial product
containing 15% graphite. The polymer pins were slid against a mild steel disk. The
180 CHAPIER 6. WEAR OF NON-METALLIC MATERIALS
roughness, Ra was 0.2 pm in the radial direction. The tests were conducted in three
steps of 30-minute duration and the friction and wear in the final step formed the
basis for comparison. The speed range was 0.73-1.47 m/s while the load range was
44 to 164 N. The corresponding contact pressures ranged from 0.44 to 1.64 MPa.
The lubricant used was base oil with a viscosity of 46.1 cSt at 40~ One ml of the
oil was spread on the disk surface at the start of each step of 30-minute duration.
Between the steps the disk was cleaned before applying fresh lubricant. The wear
rates observed in these tests are summarized in Fig 6.3. The range of friction
coefficients observed for lubricated PTFE are shown in brackets.
Firstly focusing on the behaviour of pure PTFE it may be seen that wear rate in
lubricated contact is two to three orders of magnitude lower than in dry contact.
The specific wear rate of dry PTFE ranged from 3.8 to 6.0xl 0 -13 m3/Nm. Speed has
a strong influence on lubricated wear. The wear rate was lower at the lower speed
of 1.05 m/s as compared to the higher speed of 2.09 m/s. The range o f f values for
low speed condition was 0.04 to 0.06. For the high speed condition the range was
0.07 to 0.12. At higher speed the increased temperature will reduce the film
thickness and increase boundary contact and wear. Thus in the case of PTFE it may
be argued that wear is inversely related to the hydrodynamic effects.
Graphited PTFE (GP) showed a complex behaviour. The overall wear rates for this
material in the dry case were clearly lower than that of pure PTFE, which can be
attributed to the influence of graphite layers. The wear rates were however an order
of magnitude lower at 2.09 m/s as compared to the wear at 1.09 m/s. The friction
coefficients were comparable to those of dry PTFE. In lubricated conditions GP
showed a range of friction coefficients similar to lubricated PTFE at the two speeds
considered. But the wear rates had a narrow range for both speeds and were much
lower than for the lubricated PTFE case. This may be because the influence of
increased contact at higher speed is offset by the lower wear rate at the contacts.
This observation is based on the fact that dry wear at the high speed condition is an
order of magnitude lower than at the lower speed as stated earlier. The lubricated
wear rates were comparable to the high speed dry wear of GP at the speed of 2.09
m/s. Wear mechanisms involved were not investigated in this case.
Another interesting example in this area is the work reported by Mens and de Gee
[30]. They studied the friction and wear properties of several polymer blends,
which in some cases included glass fibres also. The tests were conducted in air and
water environments. Water may increase or decrease wear rates depending on the
polymer. For example the wear rate of PEEK was 14.8x10 -~5 in the dry condition
while it reduced to 1.4x10 ~s m3/Nm in water. On the other hand with
6. 3. L UBRICA TED WEAR OF POLYMERS 181
polyetherimide the wear rate increased by three times in water. The influence of
polymer blending and glass fibres is again different for different polymers. The
tests were conducted at a nominal pressure of 5 MPa for a duration of 20 hours at
room temperature. The sliding speed was 0.25 m/s. Thefvalues ranged from 0.2 to
0.78 for dry conditions while they ranged from 0.08 to 0.40 in water. Thus with
some exceptions most of the tests in water did not show significant hydrodynamic
effects and friction reduction in water may be attributed to the boundary effects.
This work based on long duration tests is of practical relevance when polymers are
to be used for water lubricated bearings or in humid atmospheres.
10-~2
,,,,Dry P T F'E
Load 9 44-74N
I~speed. 0.73 and 1.46
E
10-~3
E
oL Lubncoted PTFE
Load-44 ~
o
~ lo-14 -Graphited PTFE Dry Speed 2.09 m/s
Load 144-16&N (0.07-0.12)
._. Speed: 1.0 S mls
u X~~ ~ Lubricated PTFE
tU Load: 44-164N (0s
m Speed" 1.05 m/s
~(f15 Graphited PTFE

Grophited PTFE Dry p.~ Lubricated
[~ Lo a d : &4 -134 N ~/j~Load -.74-194 N
Speed: 2.09 m/s Speed- 1.05,
2.09 m/s
Fig, 6.3, Histogram of wear behaviour on a log scale.

(Reproduced from Ref. [29] by permission of Leaf Coppin Publishing Ltd)
Another aspect of lubricated wear of composites is the issue of filler enrichment on

the surface. This has been observed in dry sliding conditions as discussed earlier.
When a lubricant is used it can interfere in the transfer process and it is necessary
to investigate this problem. A brief study of this aspect has been recently conducted
by Choudhary [31 ]. He studied two bronze and graphite filled PTFE materials with
the following composition:
182 CHAPTER 6. WFAR OF NON-METALLIC MATERIALS
Composite D 25% bronze + 5% graphite + 70% PTFE

Composite G 55% bronze + 5% graphite + 40% PTFE
The wear tests were conducted in a 2 pin-on-disk machine in dry and lubricated
conditions. The tests were conducted for a duration of 90 minutes at a sliding speed
of 0.71 m/s. The comparative wear data obtained are shown in Table 6.1 that
includes the Cu/F ratio on the pin surface. This ratio was based on the relative
intensity of Cu and F peaks observed by EPMA over the total pin area. Copper
intensity was taken as representative of the bronze content. On the basis of the
Cu/F ratio, bronze enrichment was observed on the worn pin surfaces. It is
interesting to observe that the enrichment was significantly higher for lubricated
contact as compared to the dry contact with one exception for composite D. As
may be seen the lubricant decreased the wear rate in all these cases. No
investigation was done to understand why enrichment was higher for lubricated
cases. Such enrichment is expected to have a positive effect on lubricated wear.
Examination of steel counter faces showed transfer films in all cases.
Table 6.1. Specific wear rate and corresponding copper to fluorine ratio for composites D
and G. (Reproduced from Ref. [31])
Material pTFEiComposites G PTFE composite D

Test Dry Lubricated Dry Lubricated
condition
Load (N) Swr Cu/F Swr C~F Swr Cu/F Swr Cu/F
(m3/Nm) (m3/Nm) (m3/Nm) (m3/Nm)
xlO -15 xlO -Is
. . . .
xlO-~s xl0 -~5
318 13.70 .,,
2.70 4.56 * 54.91 0.66 9.15 0.41
.
530
. . . .
28.77 2.44 i2.33 5.95 102.50 0.24 25.60 0.96
Swr = Specific wear rate

* Fluorine could not be detected in this case
Cu/F ratio of virgin surface for G ~ 0.27 and for D < 0.10
6.3.2 Observations on lubricated w e a r m e c h a n i s m s
Detailed studies on mechanisms related to liquid lubricants are not available in the
literature. As such only some observations are made here on the basis of
experimental studies discussed above. This discussion also points to the need for
further investigation in specific areas.
6. 3 L UBRICA TED WEAR OF POL Y M E ~ 183
Firstly there seems to be no doubt that even in nominally parallel area contacts
partial hydrodynamic effects persist as observed above [28,29]. Asperity
deformation is likely to be one of the important factors in promoting hydrodynamic
effects. There can also be local wedge effects in contact due to non-uniform wear
that are more difficult to analyse. Clear theoretical analysis of these aspects is
necessary. Such an analysis is available for artificial joints with a metal-UHMWPE
combination lubricated with water and other fluids [32]. It can form a useful basis
for analysis of hydrodynamic effects. The problem can only be understood by
specially designed experiments. Most of the investigations were conducted to study
wear and friction, and hydrodynamic effects are at best assessed in a qualitative
manner. While low friction coefficients do indicate hydrodynamic effects, detailed
analysis should involve estimation of film thickness and the extent of boundary
contact through such films.
Polymer wear can be attributed to a degree of polymer-counter face contacts

through boundary films. These contacts will include those to the metallic surface as
well as to the transferred polymer layers. If lubricant can weaken the polymer
adhesion to the counter face the removal rate will be higher. On the other hand, the
intervening hydrodynamic and boundary films will reduce transfer to the counter
face. It is proposed that the final wear is the result of these competing mechanisms.
This aspect is of importance and merits detailed investigation. For example, low
wear in a given lubricated system may be only because of reduced transfer though
the adhesion to the surface itself may be weaker. Hence there is a need to
understand the adhesion strength of polymers besides the hydrodynamic effects.
With polymer composites the influence of these effects will be more complex as
filler particles are also involved in the boundary lubrication process. They also
influence the elastic properties of the near surface material. Hence the analysis may
have to be first confined to pure polymers.
One case has been cited where significant enrichment of bronze was observed
when bronze and graphite filled PTFE was slid against steel. This enrichment was
higher for lubricated contacts than the dry contacts. This aspect may be significant
for lubricated wear of composites and deserves further investigation. Observation
of counter faces revealed transfer films, which is also a necessary condition for
effective wear control. In some cases visible transfer may not be observed with
some fluids as noted in the earlier literature by Evans [33]. His systematic work
with several fluids and polymers also showed the importance of solubility
parameters in the wear process. Directionally, deleterious effects are higher with
those fluids in which the polymer is more soluble. Hence solubility of the polymer
in the selected fluid should be taken into account.
The data with regard to strength of adsorption of different fluids on polymers is not
available. It is generally expected that on low energy surfaces the adsorption of
fluids will be weaker as compared to metals. However the adsorption effects on
polymers may be important if these surface become active due to processes like
molecular scission. Adsorption effects will be significant for the metallic counter
face. With composites containing metallic fillers strong adsorption can occur on
their surfaces. No detailed studies are available on these issues. Development of
knowledge in this area can be of importance in the selection of fluids. The polymer
surfaces can also be modified chemically. One interesting example of such
modification is the grafting of acrylic acid groups on polyethylene surfaces [34].
This study dealt with the influence of surface modification on polymer-polymer
friction. Such modifications may be of relevance for effective lubrication of
polymers if adsorption strength can be increased.
6.3.3 Application aspects

Dry and lubricated polymer beatings are extensively used for light duty
applications and are described in the literature [35]. Several such bearings are
commercially available with a wide variety of materials and designs. Current
interest is in heavy-duty applications. While there has been extensive investigation
of materials on laboratory machines for severe conditions, reported information in
real systems is limited. Three examples of the reported work are considered here.
The first two examples refer to lubricated bearings and the third example considers
a high temperature dry bearing application.
Zhang et al [36] have recently demonstrated the advantage of oil lubrication of

PTFE based bearings. The bearing consisted of PTFE and Pb composite that is
impregnated into a porous bronze layer. The porous bronze layer in turn was
anchored to the steel backing through a copper layer. The load carrying capacity of
the dry bearing as determined by Chinese standard procedure was 8.82 MPa x m/s.
This is the pv (pressure x velocity) value for failure. The same bearing was
operated in lubricated conditions. The lubricants studied were engine oil,
triethanolamine, and glycerol. When the lubricants were used the load carrying
capacity increased by 14 times and the bearings ran cooler and showed much lower
wear. Among the three lubricants engine oil showed the lowest wear. Details of the
lubrication system were not given but the low friction coefficients with a maximum
value of 0.005 clearly indicate strong hydrodynamic effects. Such effects should be
mainly due to the usual hydrodynamic action of the designed bearing. Film transfer
and lead enrichment on the polymer surfaces were observed. It is not possible to
comment on how this transfer was taking place.
6.3. LUBRICATED WEAR OF POLYMERS 185
The above example is an interesting case for switching to lubrication of polymer

bearings. The author stated that since water is known to be deleterious it was
necessary to make sure lubricants work satisfactorily. This is a general concern of
industry. This aspect can be easily checked in laboratory machines if such
information is not already available in literature. Special attention is needed when
corrosive or humid environments are involved.
The second example is the replacement of conventional babbit material for thrust
pad application. Hydrogenerator thrust pads normally operate at pressures of 3-5
MPa. Applications with higher pressures also exist. When babbit pads are used the
friction and wear during start-up is high in the boundary regime. To avoid this
problem, the rotor is jacked up with a separate hydrostatic system. This can be
overcome if the babbit is replaced by another material with low friction. Pure PTFE
pads have been developed find used in Russia but the design details are not clear.
They are claimed to operate up to 11 MPa with low friction and wear which is an
additional advantage. Recent development in this direction in the U.K. has also
been reported [37]. The design consisted of pure PTFE pads joined to steel backing
via a bronze wire mesh. Initial trials with these pads have been successful. In all
these cases turbine oil was used as the lubricant.
Choudhary et al [38] have adopted a different approach to develop thrust pads.

Firstly they have selected bronze + graphite composite pads that have lower wear
and similar friction as compared to pure PTFE. Such pads also have higher strength
and better heat transfer. On the basis of pin-on-disk tests in dry and lubricated
conditions PTFE with 55% bronze and 5% graphite was selected. The turbine oil
used was ISO 57 grade with a viscosity of 6 x 10.3 Ns/m 2 at 40~ Simulation tests
were then conducted in a scaled down thrust bearing rig. The pads were directly
adhered to the steel backing by a selected adhesive. The pad surface was parallel to
the runner surface. As this new technique may have problems of debonding, the
strength of adhesion was studied in cyclic bending tests. There was no debonding
for more than 105 cycles with a high shear stress of 9.7 MPa at the interface. This
provided adequate confidence in selecting this approach. The thrust bearing tests
were conducted for 200 hours with 50 start and stop cycles. The results were
compared with those of babbit pads. The performance of the composite pads was
superior to babbit both with regard to friction and wear. The applied pressure was
0.32 MPa. Partial hydrodynamic effects occurred even when the rig was operated
at the lowest speed of 1.54 m/s. The problem is similar to the laboratory
evaluations where hydrodynamic effects persist in nominally parallel contacts as
discussed in the previous sub-section. The full details of the development work are
given in the cited reference.
186 CHAPTER 6. WEAR (91=NON-METALLIC MATERIALS
The above work again shows the advantages of polymer bearings in lubricated
contacts. This is a case where the lubricant is pre-determined. There can be other
applications where a choice of lubricants is possible. Basic investigations of the
boundary effects and transfer mechanisms will be valuable in these applications.
Another interesting example of bearing development for dry conditions has been
reported by Marx and Junghans [39]. They have studied PEEK based materials
with a well instrumented dry bearing tester. They found that the bearing material
with PEEK + 10% carbon fibre + 10% PTFE + 10% graphite showed the best
performance amongst the tested materials. They emphasised the importance of tests
in actual bearings. Several aspects like heat transfer cannot be simulated in small
scale tests. The selected bearing can operate effectively up to 230~ This example
shows the potential of specialised polymer bearings. It may be observed that some
situations demand application under dry conditions where lubrication circuitry is
avoided. But there can be many situations where a switch over to lubricated
bearings will be of advantage.
6.4. D r y w e a r o f c e r a m i c s
The manufacture of ceramics involves basically two steps that involve green
shaping and densification. The raw ceramic in powder form is first shaped by
pressing, extrusion, or other-methods. The densification is done by hot pressing,
isostatic pressing, or other specialised processes like reaction bonding. Sintering
aids are added where required to control the microstructure and improve the
toughness. With the complex processing involved it is not easy to control the
structure and porosity of ceramics. Also different processing routes will result in
different properties. Ceramic materials used in tribological applications have been
introduced in section 1.5.1. Wear mechanisms and modelling are first considered.
The present approach based on wear maps is then presented. The properties of
typical ceramics are included in this coverage. It is considered that in a limited
coverage this is the best way to obtain an overview of the dry wear.
6.4.1 Wear mechanisms and modelling
Wear mechanisms involved with ceramics are complex. The mechanisms involve
tribochemical interactions, abrasion, microfracture, large scale fracture, and
fatigue. Additional problems include adhering transfer layers whose influence on
wear is ill defined. The detailed equations are not considered in this section but
they may have partial success in modelling within a narrow range of conditions.
6. 4. DRY WEAR OF CERAMICS 187
The present consensus seems to emphasise the wear map approach and the
development of semi-empirical equations for wear modelling. Wear map approach
will be considered in the next sub-section.
One dominant factor in ceramic wear is the asperity level fracture. The maximum
tensile stress at the asperity will be the governing stress. This value, gm~x,is k/po in
which td is evaluated from Eq. 3.6. Stress intensity factor K~ has been defined in
section 3.6.1 and is
K, : Bo(na,) ~ (6. la)
From this equation it can be shown that severe wear occurs when
namax ~-~ c >

- K IC (6.1b)
where B is the geometric factor, a~ is the typical crack dimension, O'max is the
maximum tensile stress, and K ic is the fracture toughness expressed in Nm 3/2. In
other words it is assumed that fracture leading to wear occurs when the stress
intensity factor exceeds the fracture toughness of the material. Expressing k j in
terms of Eqn. 3.6 and assuming typical v value of 0.25 for ceramics it can be
shown for a circular contact
max = P 0 (1 + 1 0 f ) / 6 (6.2)
where po is the maximum Hertzian contact stress at the asperity.
Combining the Eqs. 6.1 b and 6.2 it can be shown that for mild wear
(1 + 10f)Po ~ / K ,c -< 6/(B'ct-~) (6.3)
The fight hand side of the equation is a constant designated as Cm. The left hand
side is referred to as the severity index SCm.This concept and its modifications have
been extensively used in modelling ceramic wear. Firstly it is necessary to define
a,:. Adachi [40] and Hsu and Shen [41] have taken mean grain size as the typical
value. The modification to the S~m involves thermal effects, which can intensify
crack propagation due to thermal shock. Such effects are evaluated on the basis of
contact temperature rise.
While the transition to severe wear can be defined as above this does not provide a
means of modelling wear as a function of severity index. It is however reasonable
to expect that the wear will increase with severity index. The wear volume is
considered to be proportional to the real contact area that may be estimated from
Eq. 1.7. Another factor to be considered is the possibility of fracture at the level of
geometric contact. Such a situation can arise when stress levels are high enough to
promote fracture in the gross contact area. Under such circumstances the crack
length is taken to be the radius of geometric contact. Such wear has been referred
to as 'ultra severe wear' in [41 ].
6.4.2 C e r a m i c w e a r m a p s
Definitive equations to predict ceramic wear over a wide range of operating

conditions are not available. The current approach is to develop wear maps, which
firstly relate wear rates to operating conditions. The next step is to model the wear
rates in terms of semi-empirical equations centred round the severity index. A
detailed account of the wear map approach to ceramic wear has been recently
published by Hsu and Shen [41]. These authors studied alumina, Yttrium doped
zirconia, silicon carbide, and silicon nitride over a wide range of operating
conditions in dry as well as lubricated conditions. Table 6.2 provides information
on the properties of the ceramics used. In this table Y-TZP refers to the yttrium
doped tetragonal zirconia. The large variation between different ceramics with
regard to toughness and thermal properties may be noted. The value of the severity
index, and the associated wear are strongly influenced by these properties. The
authors used a modified 4-Ball configuration in which the ball slid on three flat
surfaces. The load range was 2 to 360 N while the speeds ranged from 1.9 to 570
mm/s. The load influence was studied by step-wise loading at a given speed. At
each step the test was run for 5 minutes. Thus wear evaluation was based on very
short duration tests. The lubricants studied were purified paraffin oil as well as
water. Three dimensional wear maps were constructed for all conditions as a
function of load and sliding speed. The various wear mechanisms observed in
different zones were also identified. These wear mechanism maps for dry
conditions are reproduced in Fig. 6.4. They have identified wear modes that
include micro-abrasion and different forms of fracture. There are also major
differences in the response of each material to stress and sliding speed. Such maps
provide useful guidance with regard to the zones of safe operation.
The wear rates mentioned here are in mm3/s. It will be useful to relate the wear rate
to known parameters. The mild, severe, and ultra severe wear can be expressed by
three different equations. It is more convenient to model the wear in all modes by a
single equation. The search for correlation under dry conditions indicated that wear
volume can be reasonably correlated to
(~ max ( T~
where T* is the temperature rise in the nominal contact area, T o is ambient

temperature, N is the normal load, and l is the sliding distance. The ratio of the two
temperatures is considered to be directionally related to the influence of thermal
shock on fracture.
It may be noted that the studies were confined to ceramics sliding on themselves.
The efforts in this direction are continuing. The authors also discussed the
possibilities and limitations of such approaches. The issues involved include long-
term effects like fatigue and possible tribochemical interactions even under
nominally dry conditions. Also the question whether wear rates observed in step-
load tests are representative of the steady state need to be addressed.
Table 6.2. Material properties of the ceramics in wear map studies. (Reproduced from Ref.
[41 ] by permission of CRC Press)
Description A1203 Y-TZP SiC Si3N4

Process Sintered Sintered Sintered and HIW
post-HIP
Sintering aid/Impurities w AI203, AI Mg, Fe,
Y203 AI, W
Density, g/cm 3 3.9 6.05 >3.17 3.25
Phase t-ZrO2
Average grain size, m 2-15 ~ 1.0 3-8 0.3-2
Elastic modulus, GPa 372 220 430 310
Poisson's ratio 0.22 0.28 0.16 0.28
Hardness, GPa (20~ 16 + 0.8 13 + 0.7 31 + 1.6 24 + 1.2
(500~ 8 + 0.4 4 + 0.2 18 + 0.9 17 + 0.9
(1000~ 4 + 0.2 3 + 0.2 10 + 0.5 13 + 0.7
Fracture toughness, MPa.m t/2 4.5 8.5 3.2 5.4
Compressive strength, GPa 2.6 1.9 2.5 3.0
Specific heat, J/g~ 0.88 0.4 0.95 0.65
Thermal conductivity, W/m~ 35.6 1.8 110 33.0
Thermal expansion, 1/~ 7.1x10 6 1.0xl0 6 4.1x10 6 3.5x10 "6
a HIP - hot-isostatically-pressed
l/1 Thermal-shock -
,-- 100 -
E
E
Micro- ~
V]
a brasion~~
in duced intragra-
nular fracture -"EE 100 FI - 8--1()6
\
mn~/s
T~~-~o~-
v
1-10~, ~ induced intragr-
(U
0 fracture ~f r10a Lc t ~ u ~r' ~e 0 Onulor
1 c ~
0. 10
- Localized ~/ 4 ~n J ~ \ lntrogranukar
mic ro*oc ture ~';--10" rnma/s 9- IDef~- I inlet- .~ fraclure
"U Defor- |'/ J maSonJ granular
motion 168 ~'--166 JMicro-I fracture
1 I ~ I
l abrasi~ I I
1
0 1 2 3 0 1 2 3
Mean Hertzian pressure (GPo) Mean Hertzian pressure (GPo)
(a) A I 2 0 3 (b) Y - T Z P
1000 j ' --
J -'lO~mm3/s ol
Thermal -shock-
.,;,7 ~ i n d u c e d introgr- E 100 - ~ induced intragr-
!-- E
/ --ha ~ a c t u re ~10-~ "~~lar fracture
"O
qU
QJ
Q.
lntragranular
01
t/I
~caru~
tnicro-~ fracture
10 ~kocGlized fracture m 10
! micro-
Jfra c tu re in tragranular .C
_
fracture! intc~rgranular
fracture
q3 J Micro fracture .~-
I abrasian Micro,-/
1 L J ,, 1 --
abrasion I I
0 1 2 3 0 1 2 3
Mean Hertzian pressure (GPo) Mean Hertzian pressure (GPa)
(c) SiC (d) Si 3 N4
Fig. 6.4. Wear mechanism maps under dry sliding conditions. (Reproduced from Ref. [41]
by permission of CRC Press)
Adachi [40] on the other hand attempted to relate wear volume directly to the
severity indices with and without thermal effects in dry sliding. The tests were
conducted in a pin-on-disk machine with hemispherical pins. Alumina, silicon
carbide, and silicon nitride were the ceramics tested. Sliding speeds ranged from
0.01 to 2.3 m/s in most cases. In one case speeds up to 52 m/s were employed. The
load range used was 2 to 100 N. Scatter was significant though a rough correlation
could be seen. Also the temperature rise calculated in this case was based on the
asperity rise while in [41 ] the overall temperature rise was considered. The specific
wear rates in the mild regime can vary between 10.9 to 10.6 mm3/Nm. Usually for
metals in dry sliding the minimum wear rates observed are nearly 10.6 and so there
is a scope to use ceramics in dry contacts provided one stays in the low wear zone.
On the other hand in severe wear ceramics can show wear rates as high as 102.
The empirical correlations only provide a global picture. For example in the mild
wear region the relative contributions due to different mechanisms is not known
but the wear rate is correlated nevertheless to the empirical parameter selected.
Such relation can be specific to the machine and operating conditions selected. In
the test configurations used in the above examples as wear occurs, the initial point
contact changes to area contact. Thus the overall stress levels decrease with
increasing wear. If continuous Hertzian stress acts as in the case of rolling/sliding
the nature of the empirical relations may change. Development of large databases
may take more time. In the meanwhile the obvious methodology for selection is to
collect at least partial data for the operating range involved. The situation is akin to
the dry and lubricated wear of metallic materials discussed in earlier chapters. An
additional complication involved is the sensitivity of ceramics to wear. Even minor
changes can affect the wear response and this creates difficulties in generalising
data. Several other papers on ceramic wear maps are available in the literature with
varying nomenclature [42,43,44]. In the present section a consideration is given
only to the two recent and comprehensive approaches to wear maps. It is
considered that this is adequate to meet the objective of this chapter. More recently
attempts have also been made to construct ceramic-metal wear maps [45].
6.5 L u b r i c a t e d w e a r of cera m i cs
Tribological application of ceramics can involve dry as well as lubricated

conditions. For moderate temperature applications fluid lubricants with or without
additives offer a means of controlling ceramic wear and friction. The lubricant
behaviour is specific to the ceramic and cannot be generalised. For high
temperature applications fluid lubricants cannot be used and new approaches are
necessary. Moderate applications can involve ceramic-ceramic as well as metal-
ceramic combinations. In the present section lubricant-ceramic interactions at
moderate temperatures are first considered. This is followed by applications at
moderate and high temperatures. The word 'moderate' usually represents
applications not exceeding 250~ Beyond 150~ synthetic fluids have to be used
for lubrication. The types of synthetic fluids have been covered in chapter 2.
6.5.1 Ceramic-lubricant interactions
In general terms [46] the adsorption of hydrocarbons is weak on ceramic surfaces.

With polar molecules like fatty acids the adsorption is strong on oxide ceramics.
This is similar to the situation with metallic oxides. On the other hand their
adsorption is weak on non-oxide ceramics like SiC. Water has a significant
influence on the tribochemical reactions. It is known to interact with most of the
ceramics resulting in the formation of oxides and hydroxides of variable
composition. As an example alumina can foma different types of hydroxides on the
surface [47]. In the case of silicon nitride significant surface interaction with the
formation of hydrous silicon oxides is observed. This can also lead to surface
smoothening resulting in hydrodynamic effects [48]. The effect of such chemical
interactions on wear is variable and depends on the operating conditions. The
detailed wear maps generated in presence of water and cited earlier [41 ] showed,
that the influence of water is variable and depends on the tribochemical reactions
as well as its influence on crack growth. In this connection an important review of
work till the end of the eighties by Lancaster [49] is of importance. He observed
that besides tribochemical reaction, water could affect crack growth. It can also
modify the aggregation of the wear debris. The aggregated wear debris in some
cases can act as a wear resistant third body. Disruption of such layers can increase
the wear rate. The overall influence will be a combination of the three effects
involved.
An interesting attempt has been made to distinguish the relative proportion of

chemical and mechanical wear with silicon nitride sliding on itself in water [50].
The chemical reaction generates ammonia, which is soluble in water. One mole of
the nitride results in four moles of NH3. By measuring the ammonia concentration
the equivalent chemical wear can be calculated. The total wear can then be
separated into 'chemical' and 'mechanical'. When wear rates were low the
proportion of chemical wear was more significant while mechanical wear
predominated when wear rates were high. This idea can be put to practical use.
Many industrial lubricants will contain dissolved water. Determination of the ratio
of total silicon to that present as nitride in the wear particles can be a useful guide
to assess the extent of tribochemical interaction with water. In real systems
attention has to be paid to the possible ingress of silica particles, which can affect
the ratio.
One point of interest here is to know how the 'mechanical' wear is influenced by
water. This influence is complex and cannot be predicted theoretically. With regard
to humidity one possible approach can be to first obtain steady wear situation in
6.5. L UBRICA TED WEAR O k C E I ~ M I ( ' S 193
dry air. Then humidity may be introduced at the required level and the change in
wear rate can be obtained. The changes can be reliable only when they differ
adequately from the mean value for the dry case. One interesting example [51] of
such an approach from another area of tribology may be mentioned. To study the
influence of friction modifiers steady friction was first obtained. The friction
modifier was then introduced into the system and the change in friction noted. The
usual procedure of evaluation with and without friction modifier involves a large
number of tests, as statistical variations have to be accounted for. The author
considers this concept of differential change should be more actively explored in
tribology.
The role of base fluids on ceramic tribology has also been studied though less
extensively than with water. Hydrocarbon fluids in many cases were found
effective in reducing friction and wear of ceramics. This improvement is related to
the chemical interaction with the surfaces though the adsorption is expected to be
weak. One interesting example is the study on friction and wear of silicon nitride
sliding on itself by Jahanmir et al [52]. The study was conducted with pins of
hemispherical cap of 3.0 mm radius. The tests were conducted at a low sliding
speed of 1.0 mm/s at a load of 9.8 N. Clear evidence of tribochemical reactions was
observed with water as well as humid air as significant amorphous silica was
detected in the wear track. The tests with hexadecane showed an order of
magnitude reduction in wear compared to water. There was clear evidence of
carbon formation as the reaction product on the wear track showed amorphous
silica as well as carbon. The reaction to form silica was attributed to dissolved
water present in the hydrocarbon. A satisfactory explanation for chemical
interaction of the hydrocarbon was not available. The authors considered
decomposition is possibly due to the catalytic activity of the ceramic. This work is
also a pointer to the significant role that dissolved water can play in tribochemical
reactions.
Klaus et al 53] studied the lubricated wear of silicon nitride with various lubricants
in a 4- Ball machine in the machine load range of 10-40 kg and a speed of 600 rpm.
Steel-steel, steel-ceramic and ceramic-ceramic combinations were studied. The
lower balls were replaced by ceramic flats. It is of interest to note that a steel-
ceramic combination was vastly superior to a steel-steel system with the lubricants
studied. On the other hand a ceramic-ceramic couple showed comparable
performance to a steel-ceramic couple in some cases while in other cases it showed
catastrophic wear. The authors directionally discussed the efficacy in terms of
friction polymer formation at the high temperatures involved. The lubricants
studied included mineral oil, polyglycol, polyol ester, engine oil, and tricresyl
phosphate (TCP) and tributyl phosphate. They also observed superior performance
with 1% TCP in mineral oil as compared to neat TCP. The comparisons were based
on short duration tests of 30 minutes. The synergistic effect of a metal-ceramic
couple is interesting and deserves further investigation.
The detailed information for lubrication with purified paraffin oil for four different
ceramics has been plotted in the earlier cited reference [41 ]. The paraffin oil that is
a representative hydrocarbon lubricant was found beneficial for all the four
ceramics studied. In all cases paraffin was found to be useful in reducing wear. In
particular the speed effects were significantly lower as compared to the dry case.
This was attributed to the low interfacial temperatures in the presence of lubricant.
The liquid paraffin was considered to be inert and hence the possible chemical
reactions were not taken into account. On the other hand Jahanmir et al [52] found
carbon formation with hexadecane at very moderate sliding conditions as discussed
earlier. As the conditions in the wear map studies were more severe, carbon
formation should be expected in this case also with liquid paraffin. Information in
this regard is not available and cannot be commented upon. Detailed wear maps
with other fluids for different ceramics is not available. It is reasonable to expect a
positive influence with lubricants over a wide range.
The above studies are based on conventional lubricants and additives. These are all
well studied with metallic contacts and a similar base may eventually develop for
ceramics. But the question is whether there can be an all-together different
chemistry suitable for ceramics. This is likely to be addressed in the future.
6.5.2 Moderate temperature applications

The practical application of ceramics normally involves replacement of existing
components with ceramic-ceramic or metal-ceramic combinations. The metal-
ceramic combination is also referred to as a hybrid combination in the literature. In
many situations the lubricant is pre-decided due to the other components in the
system. In some cases a new formulation is possible if the component is
exclusively lubricated. For example one can envisage a new formulation for a
ceramic wire-drawing die. On the other hand if components like piston rings or the
cam follower in an engine are to be replaced, they have to function with the
existing lubricant that has to satisfy the requirements of the total system. The
problem is demanding and is considered below with an example.
Winn et al [54,55] reported on the lubricated wear of ceramics. They studied steel-
ceramic combinations in a 3 pin-on-disk machine. Domed pins were used and the
6. 5. L UBRICA TED WEAR OF CERAMICS 195
initial Hertzian stress applied on each pin was 596 MPa. The tests were conducted
at different speeds of 0.06, 0.25, and 1.0 m/s. One lubricant used was the mineral
based fully formulated SAE 30 grade engine oil. The second lubricant studied was
a synthetic lubricant with a similar additive package. Alumina and silicon nitride
were investigated in the tests. Initial work indicated alumina had relatively higher
wear under these conditions and the effort was focussed on the nitride ceramic.
Tests with durations as long as six months were conducted to study long term wear.
This is a unique feature of these studies. Nominal temperature of the pins was
maintained at 100~ The wear rates ranged from 4 x 10-t2 to 10it nma3/Nm and
were lower than for the steel-steel system. Surface polishing due to tribochemical
reactions was put in evidence. Surface studies confirmed dithiophosphate based
films. Partial hydrodynamic effects were observed at the high speed of 1.0 m/s.
With ester based lubricants higher initial wear was observed followed by a lower
steady state wear. These studies were with reference to ceramic pins sliding on a
steel disk. Some tests were also conducted with steel pin sliding against ceramic,
which behaved somewhat differently. A possible explanation was attempted. The
pin wear volume was obtained neglecting elastic recovery. This study provided
adequate confidence for some engine applications like cam followers. If the tests
were conducted for short duration of an hour or so, as is the usual practice, the
conclusions could be different and less reliable. The study clearly shows that
running-in can persist over several kilometres. Earlier observations made by the
author in section 5.4.1 with regard to running-in are pertinent in this regard. Similar
studies may be necessary to evaluate possible ceramic replacements for valve seats,
piston rings, and liners.
One comment regarding wear maps is important from a practical point of view.
The present approaches define mild wear as less than 10.7 mm3/Nmand concentrate
on the speed-load transitions to severe wear. In lubricated wear, as seen above, the
interest is in very low wear rates which are orders of magnitude lower than this
value. Also the ability to distinguish small differences in wear rates is important.
There is a clear need for the generation of wear maps that provide expanded
information in the mild wear region. Development of semi-empirical equations to
relate wear rates to operating parameters will also be valuable. The situation is
similar to the case of antiwear additives discussed in the previous chapter.
Another area where extensive investigations are being carried out is the
development of ceramic rolling element bearings. Silicon nitride is found useful for
this purpose. The development of quality ceramics has undergone positive changes
and it is now possible to have materials with assured performance. This is unlike
the case two decades ago when batch-to-batch variations were common. The
196 ('HAPFER 6. WEAR OF NON-METALLIC MATERIALS
bearings have the advantage of high stiffness, lower weight, and high temperature
capability. Extensive fatigue tests continue to be conducted with different
configurations [56]. Direct bearing tests are also being undertaken. Most of the
studies were conducted with lubricants and showed that their performance is better
than metallic bearings in many cases. They are already being used in some
applications. Future applications are expected for gas turbine bearings with shaft
speeds up to 75,000 rpm.
Yet another area in which ceramics are being increasingly used is the artificial
human joint. Ceramic-ceramic as well as metal-ceramic combinations are being
used in such applications. Alumina is the main ceramic for these applications. Such
developments were possible by controlling roughness to nano levels and ensuring
significant hydrodynamic effects in the contact [26,32].
6.5.3 High temperature applications

The use of ceramics without liquid lubricants at high temperatures is a desirable
goal. If such a goal is achieved, engines for example, can be operated without
lubricant. Engine cooling can be eliminated, increasing the thermal efficiency
significantly. Another advantage is lower mass, which again contributes to higher
efficiency. The major problems to be overcome are the friction and wear, both of
which are high for the current ceramics at high temperatures.
Attempts to lubricate ceramics at temperatures beyond 800~ were based on the

use of solid lubricants like graphite [57] incorporated in the ceramics. The more
recent effort has concentrated on the idea of developing lubricious oxides on the
surface. Some metals can form homologous series of non-stoichiometric oxides
that can have desirable friction and/or wear properties. The metals with such
properties are Ti, V, Mo, and W. Several ceramic composites with TiC and TiN
have been studied for high temperature applications. They show an advantage with
regard to wear as compared to ceramics alone, due to the formation of several
titanium oxides on the surfaces. Inclusion of Mo in the composite further enhances
the wear reduction. The mechanisms involved are very complex and detailed
surface studies are being conducted as reported by Woydt et al [58]. This paper
also provides a good insight into the concept 6f lubricious oxides.
Two other approaches for high temperature lubrication are being developed. Lauer
et al [59] have used the idea of carbon lubrication. The idea is based on the
catalytic dehydrogenation of ethylene by a nickel catalyst. They have used several
nickel containing composites that included metals and ceramics and demonstrated
6.5. LUBRICA TED WEAR OF CERAMICS 197
low friction at 500~ due to carbon film formation in ethylene gas. The second
approach is vapour phase lubrication. In this approach additive is introduced in the
vapour phase with controlled surface reaction [60]. Tricresyl phosphate was tried
for such lubrication with good results. The future of the ideas mentioned here is not
yet clear.
6.5.4 Observations on lubricated wear of ceramics
Ceramics are prone to tribochemical reactions, and the influence of water vapour
and lubricants has been discussed above. The complexity involved in film
formation is similar to the case of metallic contacts. The difference is that metal-
metal systems were studied over a longer period of time and the available
experience has established the practice. Even be're the problems persist when the
operating severity is increased. Also the evaluation methods are semi-empirical and
there is a need for a more scientific approach. These issues were discussed at some
length in the previous chapter. Systematic development of wear maps was
proposed to improve the present situation. Similar considerations apply to ceramics
intended for application in moderate conditions. Consideration of possible
mechanisms and empirical relations should follow from such maps.
The issue of steady state wear is again of importance. From the information in the
previous sections wear determinations were based on 5-minute tests in one case. In
another case studies of several months were involved to establish the steady state.
As these issues were already discussed in detail in the previous chapter they are not
further considered here.
Fundamental studies are conducted more from the point of view of developing
understanding at the conditions chosen by the researcher. The interest is normally
confined to a model that describes the interactions and development of a possible
wear equation. Scientific curiosity demands such an approach. This results in an
excellent ability to understand mechanisms in a narrow range. Due to the large
number of variables in real systems, application of the vast amount of available
literature is limited. The need of the hour is to shift focus to the gaps of knowledge
in practice. The previous chapter addressed this problem in section 5.3. Similar
considerations apply to ceramics. Ceramics are more prone to local fatigue and
fracture. The influence of base fluids and additives on crack growth will be an
additional consideration in developing basic understanding. Such studies however
are useful mainly in moderate temperature applications. With regard to the high
temperature applications no suggestions can be made as the available knowledge is
inadequate.
198 ('HAP TER 6. WEAR OF NON-METALLIC MATERIALS
References
1. C. M. Pooley and D. Tabor, Friction and molecular structure: The behaviour of some
thermoplastics, Proc. Roy. Soc. Series A, 329 (1972) 251-274.
2. K. Tanaka and Y. Uchiyama, Friction, wear, and surface melting of crystalline polymers,
Proc. Wear of Materials, ASME, 1977, 499-530.
3. V. A. Smurugov, A. I. Senatrev, V. G. Savkin, V. V. Biran and A. I. Sviridyonok, On
PTFE transfer and thermoactivation mechanism of wear, Wear, 158 (1992) 61.
4. S. K. Biswas and K. Vijayan, Friction and wear of PTFE - a review, Wear, 158 (1992)
193.
5. J. K. Atkinson, K. J. Brown, and D. Dowson, The wear of high molecular weight
polyethylene, Part I: The wear of isotropic polyethylene against dry steel in
unidirectional motion, J. Lub. Tech.. ASME, 100 (1978) 208.
6. V. R. Agarwal, U. T. S. Pillai and A. Sethuramiah, New observations on PTFE wear
mechanism, Proc. Wear of Materials, ASME, 1989, 501.
7. K. Marcus and C. Allen, The sliding wear of ultrahigh molecular weight polyethylene in
an aqueous environment, Wear, 178 (1994) 17..
8. V. K. Jain and S. Bahadur, Development of a wear equation for polymer-metal sliding in
terms of the fatigue and topography of the sliding surfaces, Wear, 60 (1980) 237.
9. N. P. Suh, Tribo Physics, Prentice Hall, New Jersey, 1986, chapter 6.
10. N. Viswanath and D. G. Bellow, Development of an equation for the wear of polymers,
Wear, 181-183 (1995) 42.
11. H. Czichos, Influence of adhesive and abrasive mechanisms on the tribological
behaviour of thermoplastic polymers, Wear, 88 (1983) 27.
12. W. Hirst and A. E. Hollander, Surface finish and damage in sliding contact, Proc. Roy.
Soc. A. 337 (1974) 379.
13. H. Bohm, S. Betz, and A. Ball, The wear resistance of polymers, Trib. Int., 23 (1990)
399.
14. T. A. Stolarski, Tribology of polyetheretherketone, Wear, 158 (1992) 71.
15. D. Konur and F. L. Matthews, Effect of the properties of the constituents on the fatigue
performance of composites: a review, Composites, 20 (4), (1989) 317.
16. A. K. EI-Senussi and J. P. H. Webber, Critical strain energy release rate during
delamination of carbon fibre reinforced plastic laminates, Composites, 20 (3), (1989)
245.
17. K. Friedrich, Z. Lu, and A. M. Hager, Overview of polymer composites for friction and
wear application, Theor. Appl. Fracture Mechanics, 19 (1993) 1.
18. K. Friedrich, Z. Lu and A. M. Hager, Recent advances in polymer composites'
tribology, Wear, 190 (1995) 139.
19. T. A. Blanchet and F. E. Kennedy, Sliding wear mechanism of polytetrafluoroethylene
(PTFE) and PTFE composites, Wear, 153 (1992) 229.
REFERENCES 199
20. S. Bahadur and D. Gong, The transfer and wear of nylon and CuS-nylon composites:
filler proportion and counter face characteristics, Wear, 162-164 (1993) 394.
21. S. Bahadur and A. Kapoor, The effect of ZnF2, ZnS, and PbS fillers on the tribological
behaviour of nylon 11, Wear, 155 (1992) 49.
22. Q. Zhao and S. Bahadur, The mechanism of filler action and criterion of filler selection
for reducing wear, Wear, 225-229 (1999) 660.
23. K. A. Grosch, The relation between the friction and visco-elastic properties of rubber,
Proc. Roy. Soc., London, Series A, 274 (1963) 21.
24. K. Holmberg and G. Wickstrom, Friction and wear tests of polymers, Wear, 115 (1987)
95.
25. A. Unswoth, Recent developments in the tribology of artificial joints, Trib. Int., 28
(1995) 485.
26. D. Dowson, New joints for the millennium: wear control in total replacement hip joints,
Proc. Instn. Mech. Engrs., Part H, Journal of Engineering in Medicine, 215 (2001)
335.
27. R. W. Bramham, R. B. King, and J. K. Lancaster, The wear of PTFE-Containing dry
bearing liners contaminated by fluids, ASLE Trans., 24 (1981) 479.
28. P. M. Dickens, J. L. Sullivan, and J. K. Lancaster, Speed effects on the dry and
lubricated wear of polymers~ Wear, 12 (1986) 273.
29. A. Sethuramiah, K. L. Awasthy, Braham Prakash, and P. K. Mahapatra, Lubricated
wear of PTFE and graphited PTFE, Lub. Sci., 3 (1991) 181.
30. J. M. M. Mens and A. W. J. de Gee, Friction and wear behaviour of 18 polymers in
contact with steel in environments of air and water, Proc. Wear of Materials, ASME,
(1991) 563.
31. T. R. Choudhary, Tribological investigation of heavy duty polymer based thrust pads,
PhD thesis, IIT Delhi, 2001.
32. D. Jalali-Vahid, M. Jagatia, Z. M. Jin, and D. Dowson, Prediction of lubricating film
thickness in a ball-in-socket model with a soft lining representing human natural and
artificial joints, Proc. Instn Mech. Engrs., Part H, Journal of Engineering in Medicine,
215 (2001) 363.
33. D. C. Evans, Polymer-fluid interactions in relation to wear, in D. Dowson, M. Godet,
and C. M. Taylor (eds.), Proc. 3rd Leeds-Lyon Symposium on Tribology, Instn Mech.
Engrs. Publication, 1978, 47-59.
34. L. Lavielle, Polymer-polymer friction: relation to adhesion, Wear, 151 (1991) 63.
35. V. A. Bely, A. I. Sviridenok, M. I. Petrokovets, and V. A. Savkin, Friction and Wear in
Polymer-Based Materials, (Translated from the Russian), Pergamon Press, 1982,
chapter 9.
36. Z. Zhang, W. Shen, W. Liu, Q. Xue, and T. Li, Tribological properties of
polytetrafluoroethylene based composite in different lubricant media, Wear, 196
(1996) 164.
37. J. E. L. Simmons, R. T. Knox, and W. O. Moss, The development of PTFE

(polytetrafluoroethylene)-faced hydrodynamic thrust bearings for hydrogenerator
application in United Kingdom, Proc. Instn Mech. Engrs., Part J, Journal of
Engineering Tribology, 212 (1998) 345.
38. T. R. Choudhary, A. Sethuramiah, O. Prakash, and G. V. Rao, Development of
polytetrafluoroethylene composite lining for a hydrogenerator thrust pad application,
Proc. lnstn Mech. Engrs., Part J, Journal of Engineering Tribology, 214 (2000) 375.
39. S. Marx and R. Junghans, Friction and wear of highly stressed thermoplastic bearings
under dry sliding conditions, Wear, 193 (1996) 253.
40. K. Adachi, K. Kato, and N. Chen, Wear map of ceramics, Wear, 203-204, (1997) 291.
41. S. M. Hsu and M. C. Shen, Wear Maps, in Bharat Bhushan (ed.), Modern Tribology
Handbook, Vol 1, CRC Press, New York, 2001, chapter 9.
42. K. Hokkirigawa, Wear map of ceramics, Wear of Matrials, ASME, 1991,353.
43. Y. S. Wang, S. M. Hsu, and R. G. Munro, Ceramic wear maps: alumina, Lub. Eng.,
(1991)63.
44. A. Blomberg, M. Olsson, and S. Hogmark, Wear mechanisms and tribo mapping of
A1203 and SiC in dry sliding, Wear, 171 (1994) 77.
45. J. R. Gomes, A. S. Miranda, J. M. Vieira, and R. F. Silva, Sliding speed-temperature
transition maps for SiaN4/iron alloy couples, Wear, 250 (2001) 293.
46. P. Studt, Boundary lubrication: adsorption of oil additives on steel and ceramic surfaces
and its influence on friction and wear, Trib. Int., 22 (1989) 111.
47. R. S. Gates, E. E. Klaus, and S. M. Hsu, Tribochemical mechanism of alumina with
water, Trib. Trans., STLE, 32 (1989), 357.
48. T. Saito, Y. Imada, and F. Honda, An analytical observation of the tribochemical
reaction of silicon nitride sliding with low friction in aqueous solutions, Wear 205
(1997) 153.
49. J. K. Lancaster, A review of the influence of environmental humidity and water on
friction, lubrication, and wear, Trib. Int., 23 (1990) 371.
50. T. Saito, T. Hosoe, and F. Honda, Chemical wear of sintered Si3N4, hBN and Si3N4-
hBN composites by water lubrication, Wear, 247 (200 I) 223.
51. S. G. Arabyan, I. A. Holomonov, A. K. Karaulov, and A. B. Vipper, An investigation of
the effectiveness of antifriction additives in motor oils by laboratory methods and
engine tests, Lub. Sci., 5 (1993) 241.
52. S. Jahanmir and T. E. Fischer, Friction and wear of silicon nitride lubricated in humid
air, water, hexadecane, and hexadecane + 0.5% stearic acid, Trib. Trans., STLE, 31
(1987) 31.
53. E. E. Klaus, J. L. Duda. and W. T. Wu, Lubricated wear of silicon nitride, Lub. Eng.,
47 (1990) 679.
REFERENCES 201
54. A. J. Winn, D. Dowson, and J. C. Bell, The lubricated'wear of ceramics, Part 1: The
wear and friction of silicon nitride, alumina and steel in the presence of a mineral oil
based lubricant, Trib.Int., 28 (1995) 383.
55. A. J. Winn, D. Dowson, and J. C. Bell, The lubricated wear of ceramics, Part I: The
wear and friction of silicon nitride, alumina and steel in the presence of a mineral oil
based lubricant, Trib.Int., 28 (1995) 383.
56. M. Hadfield and T. A. Stolarski, The effect of the test machine on the failure mode in
lubricated rolling contact of silicon nitride, Trib. Int., 28 (1995) 377.
57. A. Gangopadhyay and S. Jahanmir, Friction and wear characteristics of silicon nitride-
graphite and alumina-graphite composites, Trib. Trans., STLE, 34 (1991) 257.
58. M. Woydt, A. Skopp, I. Dorfel, and K. Witke, Wear engineering oxides/antiwear
oxides, Trib. Trans., STLE, 42 (1999) 21
59. J. L. Lauer and B. G. Bunting, High temperature solid lubrication by catalytically
generated carbon, Trib. Trans, STLE, 31 (1987) 339.
60. E. E. Gramham and E. E. Klaus, Lubrication from vapour phase at high temperature,
ASLE Trans., 29 (1986) 229.
Nomenclature
ar typical crack dimension
B geometric factor
Cm constant= 6/(Bx/-~)
k/ multiplication factor to obtain ~m~
KI stress intensity factor
Kic fracture toughness
l sliding distance
N normal load
P0 maximum Hertzian contact stress at the asperity
Scm severity index
T" temperature rise in the nominal contact area
To ambient temperature
Greek Letters
v Poisson ratio
(~'max maximum tensile stress
203
7. Tribological evaluation methodologies
7.1 Introduction
The role of lubricants in controlling wear and scuffing has been discussed in
chapter 5. As discussed it is not possible to predict performance from fundamental
considerations alone. The ideal approach is to evaluate lubricants in real systems.
Such direct evaluations are done in some cases. Prescribed engine tests to evaluate
wear and fuel efficiency is an example of such an approach. Such testing is
expensive and time consuming. Also industrial lubricants are used in a variety of
applications and operating conditions and it is impractical to use real systems for
evaluation. Experimental evaluation in tribological rigs is the present available
approach to assess performance. Several standard rigs from ASTM, IP, DIN, and
other bodies are available for testing. In addition several non-standard in-house
techniques are also used that are aimed at improving the evaluation capability. The
performance level required is specified for a given lubricant. All such methods
have their possibilities and limitations. Critical evaluation of the test methods is
attempted in this chapter with the background available from chapter 5. All the
available test methods will not be covered. The emphasis will be on the
methodology centred round well known tests.
At the outset it is necessary to realise that performance considerations are limited

here to lubricant-metal interactions. In some cases the performance problems are
related to other factors like ingress of abrasives, starvation of the lubricant, and
quality control of the materials. Careful analysis of a tribological problem is
necessary to assess whether the problem is due to the lubricant or otherwise.
The first section of the chapter considers different contact geometries used in
tribological testing and their implications. The second section deals with the
evaluation of antiwear additives while the third section deals with EP evaluation.
The final section deals briefly with the performance evaluation of metal working
lubricants. Issues of boundary lubrication in metal working are also considered in
the final section and a new approach is suggested. Other aspects of metal working
are covered very briefly. The major objective of this chapter is to develop a more
logical framework for tribological evaluation with smaller test figs.
204 CHAPTER 7. TRIBOLOGICAL EVAL UA TION METHODOLOGIES
7.2 Test configurations

The common test geometries used to study wear include pin-on-flat, 4-Ball, ring-
on-flat, pin and V-block, and rolling/sliding disk contact. The geometries involved
are sketched in Fig. 7.1. Other geometries like crossed cylinders are also used in
some cases. Disk machines which run with line contact have an essentially constant
Hertzian stress throughout the test for a given normal load. This geometry can
simulate gear contact conditions and has been widely used to evaluate scuffing
with mineral oils as discussed in chapter 5. The geometry is also used to study
fatigue wear extensively. The other machines are normally operated in
unidirectional sliding with one surface stationary. In some test machines
reciprocating sliding is adopted with one specimen oscillating against a fixed
specimen with a selected frequency and stroke length as considered in the later part
of the chapter. The 4-Ball machine has a unique tetrahedral geometry. In the pin-
on-disk configurations the pin can be flat or hemispherical. In some cases pin-on-
cylinder geometry can also be used and the geometry is similar to that of a block-
on-cylinder with the block replaced by the smaller pin. The disk can rotate in a
horizontal or vertical plane. Debris accumulation is less likely when the disk
rotates in a vertical plane.
The testers used for lubricant evaluations are mainly 4-Ball, ring-on-block, and pin
and V-Block configurations. Reciprocating testers with varied types of specimens
oscillating against a flat are also used. The normal force is applied by lever loading,
pneumatic loading or hydraulic loading. The friction force is measured by
transducers of different kinds that respond to the frictional force acting on the
stationary specimen. Commercial machines adopted in standards are well designed
ensuring uniform normal loading and accurate measurement of frictional force. The
stresses acting in line contact at the start of the test can be calculated on the basis of
equations given in section 5.2. In the case of the 4-Ball machine the load acting on
one ball will be 0.408 L where L is the machine load. The steel balls are of 12.7
ram. The initial elastic contact diameter can be expressed in terms of machine load
L as follows:
d H = 0.0873(L) '/3 (7.1)
where d H is Hertzian diameter in mm and L is machine load in kg
In the case of V-Block the load acting at the contact in terms of machine load L in
kg depends on the geometry. In the standard machine as per ASTM D 3233 with a
7.2. TEST CONFIGURATIONS 205
journal radius of 6.35 mm and V-151ocKs with 96 ~ angle the load acting at the four
line contacts is Lcos42 ~ The initial stresses at the four line contacts can be
obtained from the equations given in section 3.2.
J
(a) (b)
....~...~A pp h ed load
/T~$1 ring Direction o!

LFrictlon ~orc~ ol
hne of contocl
(c) (d)
vt = v2 Pure rolling
V1 # v2 nollmg/~ldlng
(e)
Fig. 7.1. Triobological test machine configurations. (a) Pin-on-disk (b) 4-Ball (c) Pin and
V-Block (d) Block-on-ring (e) Disk-disk.
The above machines do not simulate the conditions in the real systems. The
possibilities and limitations of such tests will be considered in the later parts of the
chapter. Some rigs are also available that completely or partially simulate a sub-
assembly. Rating wear in an actual vane pump used in hydraulic systems may be
considered a full simulation test. Several available gear tests may be considered
partial simulation tests as the gear test is conducted with a specific type of gear
combination.
The determination of wear is usually done in these tests by measuring scar

dimensions and converting them to wear volume. In many laboratory rigs the wear
observed is small and not detectable by wei'ght loss. For example typical wear
volumes in a 4-Ball wear test can range between 10 .3 to 10 .4 mm 3. The
206 CHAPTER 7 TRIBOLOGICAL EVALUA IION METHODOLOGIES
corresponding weight loss is much less than a milligram and cannot be measured.
In some cases the scars may not be uniform and approximation by a suitable
programme will be necessary. One example of such a situation is the reciprocating
tester. The lower stationary surface will have higher wear at the reversal points as
compared to the centre.
7.3 Wear evaluation
The evaluation of lubricated wear can in principle be conducted with any test
geometry and operating conditions. For the purpose of comparison as well as
laying down wear limits standard tests at prescribed conditions are necessary.
Several standard tests are available. Development of standards involves a large
effort that is reflected in the thoroughness of the test procedures. Test repeatability
and reproducibility are also specified. Normally standard tests are conducted at one
prescribed set of conditions. Users in turn specify the wear limits based on their
requirements. Many test rigs can operate over a wide range of operating conditions
and not limited to the prescribed conditions. The purpose of this section is to
understand how wear evaluations can be improved. The standard rigs are
considered a means to this end. The development of the ideas will be centred round
the 4-Ball wear test. This rig is chosen because of its wide spread use. A
generalisation is then attempted taking into account other test rigs.
7.3.1 4-Ball rig evaluation
The basic geometry and stresses involved have been considered in the first section.
The ASTM D 4 1 7 2 - 94 describes the standard procedure in detail. The procedure
consists of a rotating top ball against the three fixed lower balls immersed in the
lubricant. The steel balls are 12.7 mm in diameter and are made of AISI 52100
steel. The oil temperature is controlled by a temperature controller. Pneumatic
loading is adapted to obtain good precision. Tests are conducted at a temperature of
75 + 2~ at a speed of 1200 + 60 rpm for a duration of 60 + 1 minutes. The tests
are usually conducted at 15 + 0.2 or 40 + 0.2 kg load. The average wear scar of the
stationary balls forms the basis for wear comparison. The repeatability is specified
for these conditions. The tests can also be conducted at any other condition.
Firstly for any comparison of wear behaviour it is necessary to obtain wear rates.
For this purpose it is necessary to obtain the wear volume for a given scar diameter
and also to obtain the wear coefficient.
7. 3. WEAR E VA L UA TI O N 207
The wear volume V of a stationary ball is
V = [1.55xl 0 -2 d 3 - 1.03xl 0 -5 L]d (7.2)
where d is the wear scar diameter in mm and L is the machine load in kg.
The above equation takes into account the elastic recovery of the ball. Thus the
actual wear volume is less than what is obtained on the basis of final scar only. It is
expressed in a convenient form in terms of the scar diameter and machine load.
The wear coefficient can be obtained by the following equation
K - [VH/23.3(rpm)tx0.408L] (7.3)
where H is the hardness in kg/mm 2, and t is the test duration in minutes
The above equations are based on the available literature [1 ] and are derivable from
fundamental considerations [2]. The wear coefficient is based on Archard's
approach assuming real area is proportional to hardness. The denominator is the
product of sliding distance and the load acting on the ball. The influence of elastic
recovery is higher for smaller scar dimensions though this is not readily apparent
from the final form of the equation given above. At this stage it is useful to
calculate wear volume and wear coefficient based on one hour tests for 40 kg
machine load for randomly selected wear scars. Wear coefficient is based on final
scar dimension. Hardness is taken as 720 kg/mm 2. The data is given in Table 7.1.
Table 7.1
Wear coefficient calculation for some scar diameters
,. ,,, ,, . , .., ... ,. . . . . , ,
Wear scar dia, mm Wear vol, mm 3x l0 3 Wear coefficient x l 0'~'

0.336 0.0591 0.155
0.44 0.400 1.052
0.54 1.100 2.693
0.61, , ..., ,,
1.895
. ,. ,,, ,, ,,., ,,,
4.984,,
This table shows that wear coefficients can vary by about 2 to 8 times for a
variation of the order of 0.1 mm in wear scar. As wear rate expressed in terms of
wear volume per unit sliding distance is proportional to the wear coefficient, wear
rates have similar variations. These variations are higher when the scar diameters
208 CHAPTER 7. TRIBOLOGICAL EI'AL UA TION METHODOLOGIES
are smaller. The wear rates may also be expressed on the basis of running time
using the above equation.
7.3.1.1 A s s e s s m e n t of w e a r
Firstly it may be observed that as per the standard, repeatability of the wear scar in
these tests is 0.12 mm with a 95% confidence limit. Such variations when
translated into wear volumes will amount to large variations in wear rates as
discussed above. The variations are most likely due to the running-in effects. A
typical schematic diagram for two repeat tests 1 and 2 is shown in Fig. 7.2. As
wear volume is of importance the plots are made on the basis of cumulative wear
volume as a function of time with 15 minute intervals. The lower ball pot is not
disturbed and at each stage the wear volume is obtained by average scar diameter
and the operation continued until the end of the test. Wear rate in terms of wear
volume per unit time considering only the final wear volumes are obtained from
the slopes of the D 1 and D2 lines. It is evident that significant variation can occur
on this basis. If it is assumed that running-in is completed in 30 minutes the slope
of the regressed line between 30 and 90 minutes gives the steady state wear rate.
The regressed lines are shown as RL1 and RL2 for the two tests. For clarity the
regressed line is shown as a part of the overall curve. The wear rates will be lower
now as compared to the case based only on the final wear volume. Better
repeatability is also expected by such a procedure as the variations due to running-
in effects are reduced. The possibilities and limitations of such a procedure will be
considered below.
This procedure needs to be assessed from a basic point of view. The running-in and
steady state wear have been modelled and empirical equations were developed as
given in section 5.4.1. The tests were conducted in a reciprocating tester (RT). The
running-in time was taken as the point at which 95% of the wear rate equals the
steady state wear rate. Such detailed studies to determine wear rate are the best
possible approach with load, temperature, and sliding speed as variables. This will
be very useful in mapping wear behaviour in the 4-Ball machine. Programmes
developed on this basis will be of utility to the lubricant formulators and users. The
importance of such mapping was discussed in chapter 5. One additive can be better
or worse than the other depending on the operating conditions. Surely distinction at
one or two conditions is inadequate for a formulator to make an effective judgment.
Such investigations are possible through cooperative effort. One interesting
example of such an effort is the study of transitions in lubrication regimes under
different operating conditions by the International Research Group (IRG) on wear
[3]. Pending such approaches, at least the wear rate should be based on regression
7. 3. WEAR E VA L UA ?'ION 209
lines as described above. Even here the problem is to know the starting point for
regression. Running-in effects will depend on the speed, load, temperature, and the
additives in the formulation. Usually visual observation of the curves can indicate
the steady state portion of the curve that can be regressed. An example to
determine wear rate based on regression of points between 30 and 120 minutes in a
4-Ball machine based on [4] is shown in Fig. 7.3. The figure is based on the
tabulated data given in Table 7.2. The tests were conducted at 40 kg and 1200 rpm
and 75~ The purpose of the tests conducted under different conditions was to
distinguish between two running-in oils for industry. The wear volume obtained
was for separate tests with different time intervals. In each test the variability of the
initial wear influences the wear volume. The correlation was hence not very good
with correlation coefficients of 0.935 and 0.883 for oils 1 and 2 respectively. The
variability observed with oil 1 was lower and hence the regression was better.
While better statistical analysis can be done, the inherent variability due to separate
tests should be taken into account. This can be eliminated if the wear scar is
measured at different time intervals in the same test. Many machines now are
equipped to measure wear scar without dismantling the lower pot assembly. With
such an approach the regression should give a much better assessment of wear rate.
This involves an assumption that the start and stop influences are small. This is a
reasonable assumption with conformal wear scars. Comparison of wear rates
between additives then becomes more meaningful. Also the total testing time will
....-4
j ~73.. j / /
f J /
t j /
f /
Test
iI ///
__ i i 9 i i .
0 15 30 45 60 75 90
Time (minutes)
Fig. 7.2. Schematic behaviour of two repeat tests 1 and 2 in 4-Ball tester. RL refers to
regression line while D refers to the line drawn from origin to the final wear volume.
210 CHAPTER 7. 1RIBOLOGICAL EVAL UA TION METHODOLOGIES
Table 7.2
Wear data for two oils in 4-Ball machine
,,,,,, ,,,, ,,
Time (hrs) Wear volume (oil i)'"imm ~) Wear volume ioil2)(mm3i ....
Series 1 Series 2
0.25 2.11E-04 1.27E-04
0.5 2.50E-04 1.98E-04
0.75 2.58E-04 2.77E-04
1.0 2.97E-04 4.71E-04
1.25 2.97E-04 3.39E-04
1.5 3.39E-04 3.95E-04
1.75 4.14E-04 5.51E-04
2.0 5.04E-04 7.78E-04
9.00E-04
8.00E-04
7.00E-04 y = 0.0003x + 3E-05
R 2 = 0.7796
6.00E-04
5.00E-04 Series1
- - - I - Series2
4.00E-04
- Linear (Series2)
3.00E-04
Linear (Series1)
y = 0.0002x + 0.0001
2.00E-04
R 2 = 0.8751
1.00E-04
0.00E+00 ..... [ " ! . . . . . . . t "' i --
0 0.5 1 1.5 2 2.5
Fig. 7.3. Regression of wear data in a 4-Ball tester for oils 1 and 2. X-axis is time in hours
and Y-axis is cumulative wear volume in mm 3.
be reduced compared to separate tests. Separate tests for different time intervals
were conducted in the cited example due to lack of provision for periodic
measurement. On the basis of available correlation the wear rates for oils 1 and 2
were 2.0x10 -4 and 3x10 -4 mm3/hour respectively. A better method is to utilise the
running-in equation and perform non-linear regression as reported in 5.4.1. The
detailed methodology is given in the corresponding reference [5]. Such regression
is possible only if there is continuous increase in wear volume with time. The
7.3. WEAR EVALUATION 211
procedure could thus be adapted only for oil 1. The regressed equation obtained by
this procedure is
V = 1.83x10 -4 (1 - e -'4"36t ) 4 - 1 . 3 lxl0 -4 t (7.4)
On this basis the results are as follows:
Running-in period 1.09 hours

Steady state wear rate 1.31x 10 ~ mm3/hour
Correlation coefficient 0.962
Provided the wear scars are measured at each stage in a continuous test and
adequate numbers of points are available, this method can be used to classify wear
rates accurately. It may be recalled that completion of the running-in is defined as
the point at which 95% of the value equals steady state wear rate.
The lower variability in the case of oil 1 suggests that this formulation is probably
more effective in the running-in stage. Similar tests were conducted under different
operating conditions with short duration tests with different time intervals. In all
cases oil 1 showed much less random variation with time as compared to oil 2. The
suggested simple linear regression does not ensure that the running-in influence is
eliminated. In practical terms the effect will be substantially reduced and to that
extent provides better information on wear rate. On the basis of the selected
regression zone the comparison between additives will be more meaningful. Based
on this approach it is possible to evolve a standard procedure for wear rate
determination. Such a procedure can define the minimum number of tests to be
done and the nature of regression to be done. The best approach will be to use the
non-linear regression as a standard procedure. While issues of standardisation
depend on the agreement between the concerned organisations, the approaches may
be adapted by researchers and formulators who are interested in distinguishing
between different formulations more effectively.
Large number of wear tests are reported with 4-Ball machines at different
conditions. The major interest in such studies is to distinguish between different
additives used in formulations. While the nature of additives is getting
sophisticated the approaches to wear testing in many cases simply depend on the
final wear scar diameter in a given test. The inadequacy of such an approach has
been discussed above. A recent paper [6] discussed the wear behaviour on the basis
of wear volume and wear coefficients. The wear volume was that obtained at the
212 CHAPTtiR 7. lt,UBOLOGI('AL EVALIM l l O N ME1HODOLOGIES
end of the test. Another paper [7] used the concept of 'delta wear' to characterise
the wear behaviour. Delta wear was defined as the difference between observed
scar diameter and the Hertzian diameter at the selected load. The idea is perhaps to
account for the elastic recovery. On the basis of Eqn.7.2 delta wear is not
proportional to wear volume and it is better to calculate the wear volume directly
by this equation. An interesting paper [8] considered step load tests but
characterised the wear behaviour on the basis of the step between 30 and 60
minutes only using the concept of delta wear as above. One more example of step
load tests [9] is a comparison of wear scar with progressive loading up to thirty
minutes only with 5 minutes steps. Influence of hexoyl borate was compared on
this basis. There is an obvious need for uniform procedure to determine wear rates
based on regression. While investigations are reported at few different conditions
the author has not come across any detailed wear mapping for antiwear additives.
The wear tests serve two purposes. One is the specification requirement imposed
for 'pass' under defined test conditions. Such tests specify a maximum allowable
scar diameter. The more important utility is in screening formulations. The
confidence levels in screening are important and the first step is to have an
effective procedure to clearly distinguish in a given machine. It is hoped the
suggested approach including wear mapping will be useful for this purpose. The
issue of relating such information to practice will be considered in the later parts of
the chapter.
7.3.2 Other laboratory rigs for wear evaluation
Several other test geometries can be used for evaluating wear. These include the
ring-on-block test based on ASTM D 2714, as well as reciprocating rigs of
different kinds. The Falex V-Block tester is also used for wear studies as per
ASTM D 2670. In this machine a ratchet mechanism advances to maintain constant
load on the pin. The number of ratchet teeth that advance during the testing period
indicates the wear. Disk machines with a given slide/roll ratio can also be used for
wear evaluation. Disk machine tests have the advantage of constant stress during
operation. The problem of contact stress variations has already been considered in
the first section. Most of these tests use prescribed steel specimens with given
hardness and roughness. Very little information is available regarding correlations
if any between different test machines and the choice is difficult. One interesting
paper [10] evaluated the EP and antiwear properties of different metal based
dithiophosphates as well as different types of zinc dithiophosphates with 4-Ball and
FZG test rigs. It was observed that performance of different zinc salts depends not
only on their chemical structure but also on the stress and nature of contact
7. 3. WEAR EVA L UA TI O N 213
geometry in a given test machine. Such differences are expected because the gear
rig operates with line contact and high Hertzian stress whereas the wear as well as
EP tests in 4-Ball machines will have variable contact stress. In the antiwear tests
the stresses in the contact zone as the scar develops will be lower than in FZG
tester. Also the sliding speeds and temperatures involved will be different in both
cases. This example shows the problem involved in selecting any particular wear
test. If there is mapping of wear available at different operating conditions for
various test machines an explanation is possible.
One additional factor to be taken into account in laboratory testing is the possibility
of partial hydrodynamic effects that may develop in the contact. This has to be
assessed by observing the changes in wear volume in conjunction with friction
coefficient. A large decrease in friction with time is indicative of hydrodynamic
effects. This will also be reflected in negligible change in wear volume with
different time intervals and the test is no longer representative of the antiwear film
alone. Theoretical evaluation of EHD influence is difficult. With the low pressures
involved in the worn scar the problem will fall in the regime of rigid-variable
viscosity with the added complication of ill-defined pressure distribution. The
practical approach can be to select conditions that minimise such effects by
reduced speeds or higher temperatures. Electrical contact resistance between the
rubbing surfaces can be measured with a small applied potential in the milli volt
range. A short indicates a metal contact while high resistance indicates separation
by oil film. Suitable electronic circuitry can be utilised so that voltage output can
be interpreted as a function of metal contact. Such approach is difficult to apply for
additives as the antiwear films themselves can have high resistance. An attempt by
Sethuramiah et al [11] to study possible hydrodynamic effects in a ball-on-disk
machine using tricresyl phosphate as an additive has been reported. Though metal
contact influence was accommodated by an empirical approach the influence of
hydrodynamic effects were far less than theoretically anticipated. Thus the
assumed hydrodynamic effects may as well be due to high resistance of antiwear
films. The author hence considers the variation of friction coefficient is a simpler
and better criterion for assessment of any hydrodynamic effects in the contact.
Most of the work being reported in the literature on wear does not take into account
the surface temperature rise in contact. For example with a 4-Ball machine running
at 1200 rpm under a load of 40 kg the average surface temperature rise will be
15~ for an assumed friction coefficient of 0.1 and a scar diameter of 0.5 mm. The
asperity temperature rise can also be estimated for an assumed asperity contact
dimension. Such an analysis will help in a better assessment of the additive.
214 CHAPTER 7. TRIBOLOGICAL EVALUA TION METHODOLOGIES
The metallurgical aspects are also important in testing. The wear tests are done
with a given steel combination. The 4-Ball machine uses AISI steel balls with
Rockwell C hardness of 64-66. On the other hand in the case of the block-on-ring
test the ring is made of SAE 4620 steel with a Rockwell C hardness of 58-63. The
block is made of SAE 01 tool steel with a hardness of 27-33 HRC. Thus the steel
combinations are different for the two machines. The wear behaviour for a given
additive is expected to be different in the two cases. The question that arises is
whether relative rating of additives will be same in the two machines. This question
cannot be answered unless a necessary investigation is done. It is obviously
preferable to have the same material combination as in the intended application.
While standard tests can be conducted with standard materials for specification
purposes, research investigations should simulate the required material
combination whenever possible. In some machines like 4-Ball it is impractical to
change materials. On the other hand in many reciprocating machines, extensively
used to study ring-liner tribology, actual ring and liner pieces are adapted for
testing. In many cases the purpose is in fact to evaluate material influence on wear.
Also many machines are now available in which several contact geometries can be
adapted. It may be mentioned that a larger number of test rigs are available for dry
testing with greater degree of simulation. Blau [ 12] has listed the current available
test rigs for dry friction conditions. The rigs are used for diverse purposes from
floor friction to brake performance. On the other hand most of the studies reported
on antiwear additives are confined to 4-Ball, ring-on-block and reciprocating
testers. In all test machines there is a need to assess wear rate more effectively. The
considerations involved are similar to those discussed for 4-Ball machine. The
above discussion is limited to the issues related to better wear evaluation and the
need to map performance as a function of operating conditions. The approaches
still remain empirical. The practical utility of such tests has already been discussed.
The nature of empirical relationships observed can, to an extent, provide an insight
into the mechanisms involved as discussed in earlier chapters. The issues involved
in basic modelling are elaborated in chapter 5 and hence not considered here.
The wear in real systems is complex and depends on many factors. There can be
several starts and stops, operating conditions may vary over a wide range, and there
can be some ingress of abrasive particles despite filtration. Another important
factor is the deterioration of the base oil as well as additives with use, which again
influences wear. In some formulations like engine oils the complex interactions
between several additives further complicate additive action. Despite this scenario
formulations with better antiwear capability will give relatively longer component
life and their development is important. Two additional aspects arise with regard to
development of formulations. The formulations may have to be additionally
7. 4 LOAD CARRYING CAPACITY OF L U B R I C A N I S 215
evaluated with artificial ageing to assess the real use situation. Only few
investigations are available with used oils. The other issue is the response of the
additive to the operating environment. From this point of view a formulation
should have the least possible variation in wear rate in the required operating range.
Wear determination in real systems and their comparison with laboratory rig tests
is useful. But wear determination in many real systems is possible only after a long
duration when measurable wear has occurred. These issues will be considered in
chapter 9.
7.4 Load carrying capacity of lubricants

In the present context the load carrying capacity of a lubricant refers to the EP
additive in the formulation. The purpose of EP additives and their action
mechanisms have already been elaborated in earlier chapters. The present section
deals with their evaluation. The test configurations used are 4-Ball, ring-on-block,
and journal rotating against V- blocks. The standard test methods are listed below
in Table 7.3. In addition FZG gear rig as per DIN standard is also employed to
study the load carrying capacity and is adapted by ASTM and IP. As stated in the
standards it is for the users to decide the issues related to correlation with practice
and the acceptable levels of performance required. This clause is similar to the
situation with the testing of antiwear additives.
Table 7.3
Test methods for EP properties
ASTM D 2782 Measurement of EP properties of lubricating fluids (Timken method)

ASTM D 2783 Measurement of EP properties of lubricating fluids (4-Ball method)
ASTM D 3233 Measurement of EP properties of lubricating fluids (Falex pin and Vee
block methods)
ASTM D 5182 Load-carrying capacity tests for oils- FZG machine
Detailed descriptions of these methods are available in the relevant standards. The
present section only discusses the main issues involved in EP testing considering
the test methods as a means. The load carrying capacity, which is also called EP
property, is normally the load at which a lubricant fails under prescribed
conditions. The first three machines initially involve point or line contact but
during test the geometry changes to that of a conformal contact. For convenience
216 CHAPTER 7. 7RIBOLOGICAL EI,'AL IM TION METHODOLOGIES
these are grouped under conformal contact machines. The gear and disk machines
are classified under line contact machines.
7.4.1 Conformal contact machines
The prescribed conditions vary significantly between these test machines. Firstly in
the three machines the nature of steels used are different. The sliding velocities are
also different for the three machines. For 4-Ball testing the sliding speed is 0.683
m/s while it is 2.06 m/s for the Timken machine and 0.096 m/s for the V- Block
machine. If it is assumed that the temperature rise in contact is related to lubricant
failure directionally the failure load for the three machines should be
V- Block machine > 4-Ball machine> Tirnken machine
The test method prescribed for 4-Ball machine is rather complex. Firstly the test
load at each stage is prescribed in such a manner that the variations are
approximately logarithmic. At each stage the tests are conducted with a fresh set of
steel balls and the test run for a short duration of 10 seconds. The average wear
scar is measured at each stage. The testing is continued till welding is observed.
Weld point is one parameter for rating. Another parameter is called the load-wear
index calculated on the basis of wear scars for ten stages below welding. It is
obtained by
Z LdH/ds
where d H is the Hertzian diameter, d s is the scar diameter, and L is
10
the machine load
The load-wear index is indicative of the wear behaviour under EP conditions. A

higher value indicates lower wear.
Both Timken and Falex tests prescribe a run-in procedure. After the run-in the
loads are increased in the prescribed fashion till scuffing is observed. Each test
shall be done with fresh surfaces. The Falex test incorporates another procedure
also, in which the loading is increased in steps to failure with the same surfaces.
All the machines have provision for determining frictional torque although this
determination is not mandatory for assessing EP properties. The machines are well
designed so that loading is uniform. Material specifications as well as roughness
range are rigorously followed. The Timken machine does not involve welding. The
failure criterion is based on the nature and extent of scoring observed as per laid
7. 4 LOAD CARRYING CAPACITY OF LUBRICANTS 217
down procedure. OK load is defined as the load which is one step below the failure
load. OK load may be taken as the limiting load without failure. In the Falex tester
the failure is considered to occur when the pin breaks or when the load cannot be
maintained.
The relative ratings will depend on the nature of the machine and operating
conditions. From the available information in ASTM D 3233 the results from Falex
machines correlate reasonably with 4-Ball tester.
7.4.2 Influence of operating parameters
From a fundamental point of view the EP film failure should depend on the
procedure adapted. These effects are now considered with two examples. The first
case is the work on a 4-Ball tester. The other is an example of scuffing tests in a
reciprocating tester.
7.4.2.1 4-Ball tester
This example is based on the work of Sethuramiah et al [13]. 4-Ball tests were
conducted for one minute at a speed of 0.54 m/s with a fresh set of steel balls. In
each test weld load was obtained. From the measured torques with suitable springs
the friction coefficient was calculated. Two types of behaviour were observed in
these tests. In one case the welding occurred with a sudden transition from a
relatively low friction. In another case the friction increased to a higher level
followed by eventual failure with a transition from higher friction. These are
termed Tr~ and Tr2 transitions. The behaviour depends on the nature of the additive
and its concentration. It may be noted that the additives which survived Trt
transition showed high friction without failure for 2 to 3 stages before the weld
load was reached. The stages used in these tests were 20 kg load increments. The
behaviour observed is illustrated in Fig. 7.4. It was argued that the Tr~ was related
to lubricant failure temperature while Trz was related to the EP film failure. The
behaviour up to Tr~ is termed the X stage while behaviour beyond the Tr~ is termed
the Y stage. The failure temperatures were calculated with the available friction
coefficient and an assumed contact area based on the Hertzian contact dimension.
Better estimations of temperature are possible now but the previous reported values
may be used for comparison. The temperatures at Tr~ ranged from 300 to 360~
The temperature at Tr2 transitions ranged from 720 to 800~ The relatively high
lubricant failure temperatures were considered to be due to the oxidised lubricant
products present on the surface. The Tr2 transitions were attributed to the EP film
failure. The tests were conducted with 0.294% of sulphur concentration for all
additives. Diallyl monosulphide, dibenzyl monosulphide, and diphenyl disulphide

that are less reactive directly fail beyond Tr~ indicating their inability to react fast
enough to sustain EP action. At a higher concentration diphenyl disulphide became
more effective showing both types of transitions. The failure loads for different
additives were different and the Tr2 failure temperatures were based on friction
coefficient before failure. These temperatures were treated as 'critical failure
temperatures' in the paper.
Failure temperatures alone do not offer an explanation of the mechanisms. The

possible mechanisms for failure in EP conditions have been considered in section
5.5.2. The high removal rates can be due to compositional changes in the film near
Tr2 transition. In the same paper it was shown that with sulphur (0.29%) and
diphenyl disulphide (1.48%) large amount of sulphur was observed on the surface
just before failure at Tr2. Oxygen was not evaluated. Oxygen availability for
reaction depends on the solubility as well as its consumption due to oxidation of
the fluid. These effects depend on the temperature in the contact and can change
the sulphur to oxygen ratio unfavourably. Such films with changed composition
can have high wear rate. The reaction rate is inadequate to cope with the situation
and leads to failure. This is only a tentative explanation. Direct failure at Tr~ on the
other hand can be mainly due to the inability of less reactive additives to react fast
enough to cope with the transition when the total load shifts to EP films. Yet
another complication is the possibility of metallurgical changes at the high
temperatures involved near T~'2. Any softening near the surface can adversely affect
the EP action. Limited study.by the authors indicated that for elementary sulphur
and diphenyl disulphide there-is softening below the surface as evidenced by micro
hardness measurements as a function of depth at one stage below welding. The
reduction was nearly half the original hardness in the immediate vicinity of the
surface. It may be noted that in the cited paper only softening was considered as an
explanation for failure. The idea was that the harder films would not be able to
adjust with the easier deformation of the substrate leading to film failure.
Some studies conducted with a step load procedure were also reported in the same
paper. These studies were conducted with an initial running-in followed by one
minute tests at progressively increasing loads till failure. These tests were
revealing. Less active additives showed higher weld loads in comparison to
standard tests. The friction coefficients were low ranging from 0.03 to 0.07
indicating partial EHD effects. The wear scars were much smoother. Thus the load
carrying capacity of diphenyl disulphide at 1.48% concentration increased to 380
kg as compared to the value of 240 kg in one minute tests. Hexachloroethane with
a low friction coefficient of 0.03-0.04 showed a weld load of 400 kg which is much
7. 4 LOAD CARRYING CAPACITY OF L UBRICANIS 219
higher than the value of 190 kg obtained in the one minute tests. The increase in
weld loads in step-load tests was attributed to the hydrodynamic effects. The issue
of barrier films in such cases was discussed in chapter 5.
1.Start
2.Seizure
,, 3~Weld
X Stage Trl Trl Transition
point
TWine (.)
(*)
Y Stage 3~r2
1.2.3 as In (a)
Tr2 Tranmltlon point
1 x st,o,, beyand Trl
rima (.}
Fig. 7.4. (a)Zrl and (b)Tr2 transitions observed in one minute tests in 4-Ball machine.
(Reproduced from Ref. [ 13].
The purpose of the above example is to demonstrate the variability involved in

testing based on operational conditions and to consider the failure mechanism to
the extent possible. The above methods involve varying scar diameters in a test.
Thus the apparent pressures based on geometric area keep varying. However as
areas increase with load the pressure variations may not be large and typically for a
4-ball machine the pressures may range from 150-250 kg/mm 2. In standard tests the
initial contact pressures will be high and the contact may be elasto-plastic rather
than elastic. For example with a machine load of 400 kg the elastic contact
220 CHAPTER 7. TRIBOLOGICAL EVALUA IION METHODOLOGIES"
diameter based on Hertz theory will be 0.642 mm. The pressure acting on the ball
contact will be 473 kg/mm 2, which is significantly higher than the yield pressure.
These initial conditions in tests are also expected to play a role in determining the
nature of films that develop.
7.4.2.2 Reciprocating tester
It is of interest to consider the influence of rate of loading on scuffing. Gondal et al

[ 14] conducted scuffing tests with different formulations in a SRV tester. This is a
reciprocating tester, which was also used in the experiments conducted to develop
wear equations discussed in chapters 3 and 5. Recently ASTM has adapted this
machine to evaluate friction and wear properties of EP oils as per D 6425. A sketch
of the machine is given in Fig. 7.5. The lubricants studied are given in Table 7.4.
The tests were conducted at 10 Hz frequency with a stroke length of 2.0 mm. The
load was increased at 14 N/s for slow rate tests while it increased at 42 N/s for
rapid tests. Area contact did not produce scuffing and the tests were conducted with
machined liner surfaces with chrome plated and cast iron tings. The initial contact
was thus line contact. The scuffing data obtained are given in Table 7.5.
",/
,/
J/
/,,
/,,
/
/,
//
/I
T
Fig. 7.5. SRV Reciprocating tester: (1) test cell; (2) loading system; (3) horizontal
movement (reciprocating); (4) heating system; (5) Piezo measurement system; (6) ball
holder; (7) disc specimen fixture; (8) ball specimen; (9) flat disc specimen.
T a b l e 7.4
Lubricants studied
,., ,,. , ,. , . . , ,,,,.
Oil Description Viscosity cSt at 40 and Mo, ppm Zn, ppm

100~
A Base oil 64.2 7.78 0 0
A1 A + 1% MoDTP l -do- -do- 86 113
A2 A + 1% MoDTP 2 -do- -do- 380 0
B Engine oil 129.4 13.94 0 742
B1 B + 1% MoDTP 1 -do- -do- 86 885
B2
. . . . . . . . .
B + 1% MoDTP 2 -do- -do- 380 742
Table 7.5
Scuffing tests
ul, ,. , .
S. No Contact Type of loading 0il No. of tests Scuffn~ load, N

1 L3 Slow A 1 450
2 L3 Slow A1 1 X
3 L3 Slow A2 1 X
4 L3 Slow B 1 X
5 L4 slow A 1 X
6 L3 Rapid A 5 400
7 L4 Rapid A 2 X
8 L3 Rapid B 1 X
9 L4 Rapid B 1 X
10 L3 Rapid A1 3 700
11 L4 Rapid A1 1 X
12 L3 Rapid A2 3 900
13 L4 Rapid A2 1 X
14 L3 Rapid B1 1 X
15 L3 Rapid B2 1 X
X no scuffling
L3 cast-iron ring on flat, line contact
L4 chrome-plated ring on flat, line contact
(Tables 7.4 and 7.5 reproduced from Ref. [14] by permission of Leaf Coppin Publishing
Ltd.)
222 CHAPTER 7. TRIBOLOGICAL EVAL UA HON METHODOLOGIES
The data shows that base oil alone scuffs with CI ring at 450N in a slow test while
it scuffs at 400 N in a rapid test. With additive treated oils scuffing was observed
only in the case of rapid test. Base oil with 1% MoDTP 1 (A1) showed failure at
700N while base oil with 1% MoDTP 2 (A2) showed scuffing at 900 N. Engine oil
alone (B) and with MoDTP addition (B1, B2) did not scuff within the load range of
1200 N. With chrome plated rings no scuffing was observed with any lubricant
both in slow and rapid tests. This example shows that scuffing is very much
influenced by rate of loading as well as materials.
The above examples clearly bring out the fact that scuffing is influenced by the rate
of loading, nature of lubricant, material combination, and test procedure. The work
also points to the need for understanding film failure mechanisms by studying their
detailed composition at different stages up to welding as suggested in chapter 5. If
we are equipped with this information it is easier to compare various machines and
to arrive at more objective test procedures. It is also important to record friction
coefficient values. This will provide an insight into the partial hydrodynamic
effects. Such effects may be more important when there is a run-in done before the
test. Failure mechanisms have a direct relevance to the additive formulation. For
example if the film failure is related to the composition changes of the film the
additive chemistry has to minimise these variations. In the related area of antiwear
additives effort is being made to improve performance of dithiophosphates. Based
on the available understanding of the mechanisms reduced surface oxidation is
desirable. Hence there are attempts to use antioxidants like copper naphthanates in
the formulation as cited in 5.3.2. Such insights are possible when failure
mechanisms are understood.
7.4.3 Line contact testers
Another important test used for EP evaluation is the FZG gear tester. This rig is
mainly used to evaluate gear oils. This is a back-to-back gear tester in which load
application is by a clutch mechanism. Each increasing load is called a 'stage' and it
is common to specify load carrying capacity by stage number. The normal force on
the teeth varies from 99 N to 15826 N from first to twelfth stage. The
corresponding Hertzian contact pressures range from 0.146 GPa to 1.841 GPa. The
test gears used are spur gears. The pinion has 16 teeth while the gear wheel has 24
teeth. The procedures to assess scuffing under the prescribed operating conditions
are rather elaborate and described in the standard. Other gear rigs available include
the IAE rig specified in IP and the Ryder gear rig used to evaluate gas turbine
lubricants. Industry in general relies more on the FZG test for the specification of
gear oils. The advantage of the gear rig is that contact stresses remain constant for a
given applied load. Also it simulates at least the spur gears widely used in industry.
The possible complications involved with varying stresses discussed in the
previous section are avoided. The test is run at each stage for 15 minutes until
failure. From a tribological point of view this means that films are progressively
conditioned with each stage and will be more uniform. Failure of such films can be
due to an imbalance between formation and removal rates or other reasons as
discussed earlier. Attention may have to be paid to the partial EHD effects that may
arise in gear rigs. If there is smoothing of surfaces due to chemical additives this
factor can additionally contribute to such effects. In contrast the standard 4- Ball
test involves increasing loads on flesh surfaces and running for a short duration of
10 s. The films are ill conditioned in such a case and the situation is closer to the
control of fast growing metal contact area. The problem is similar with the Timken
machine but since each test is run for 15 minutes there will be conditioning of the
film in the later stage. The author has no clear information whether failure is
basically governed by the initial severity of contact in this case. With Falex, in one
procedure the situation is similar to the 4-Ball except for the test duration of one
minute in each case and run-in involved. In the second Falex procedure where
increasing loads are applied on the same surfaces, film conditioning occurs as in
FZG rig though to a lesser degree. The author considers that these qualitative
considerations are important in selecting a suitable test rig. It will be of interest to
see whether the correlations between the rigs is possible by adapting different run-
in and progressive loading procedures.
In the case of rear axle oils there is a need to test the oils in selected rear-axles.
Such tests involve evaluation under high-speed shock loading conditions as well as
low speed high torque conditions. This is because the usual EP tests cannot
differentiate these properties effectively. The desired rigs are not easily available
and the testing costs in the specialised laboratories are high. So an attempt was
made by the author and his colleagues to distinguish the lubricants with a disk
machine. The disks were run with opposite peripheral velocities to make the test
conditions severe. Opposite peripheral velocities also ensure boundary conditions.
Hence partial EHD effects are eliminated. Intermittent loading was applied with an
eccentric mechanism. The initially conditioned run-in disks were immediately
loaded to a high load. If no scuffing was observed the test was conducted at a
higher load with a new pair of conditioned disks. Such testing could correlate well
with the shock load test results. This was based on the testing of reference gear oils
with known shock load behaviour in rear-axle rigs. The maximum load applied in
the disk machine was 200 kg. It is of interest to note that if the loading was done
gradually on the same disks scuffing did not occur within the capacity of the
machine and distinction was not possible. This again showed the influence of film
224 ('HAP IER 7. TI~JBOLOC.I('AL Et'AL UA TION METHODOLOGIES
conditioning on EP action. The research carried out in this area as well as the
nature of rear-axle tests with which correlation was attempted is available in
publications [15,16]. It is necessary to resort to such trial and error procedures in
the absence of detailed knowledge of mechanisms.
7.4.4 Practical considerations
The present approach is to specify a particular limit for EP property in a given

machine. This is based on experience available with different fluids and their EP
properties in a given machine. Neat drilling fluid containing EP additives for
example, will have to meet a particular load carrying capacity in a 4-Ball machine.
A drilling test under standard conditions may also be specified. Specification for a
mild EP formulation for a reduction gear may have to typically meet minimum
fourth stage pass in FZG rig. A 4-Ball test may or may not be specified for
performance. Many organisations may have their own in-house rigs for evaluation
though they may not be used for specification purposes. Such approaches are
satisfactory within limits. Also reasonable extrapolations are valid provided the
additive chemistry is essentially the same. Problems arise when new kinds of
additives are to be evaluated or when there is any significant change in operating
conditions or materials. It is clear that the whole approach to selection of additives
is empirical with some criteria based on experience in test machines. Attempts to
generalise failure criteria of mineral oils on the basis of critical temperatures was
discussed in chapter 5. While there is no doubt that temperature is an important
consideration in EP failure it is not a sufficient condition. A clear understanding of
the mechanism of film failure is necessary as discussed earlier. The present
scenario is unlikely to change if mechanistic studies are not pursued. It is desirable
to conduct a detailed thermal analysis of the problem and to arrive at an agreed
methodology for temperature evaluation in the test machines. The important
influence of temperature on EP action can be then be clarified. The present
methods assess EP action only on the basis of failure load.
7.5 Metal working lubrication
Metalworking is a vast area and this brief section is intended to look into the
aspects of boundary lubrication only. Metal working operations involve major
plastic deformation coupled with high temperatures. The processes may be divided
into metal removal and metal forming processes. The metal removal processes can
be further divided into cutting and finishing processes. Metal forming processes
basically involve shaping of metal with large plastic deformation. The various
7. 5 METAL WORKING L UBRI('A 770N 225
metal forming operations include rolling, wire drawing, deep drawing, forging, and
extrusion.
Lubricants are used to good effect in many metal removal processes. In such
operations one of the important roles of the lubricant is removal of the large
amount of heat generated. Oil in water emulsions are commonly used in such
operations. The oil part can have boundary additives of different kinds including
fat based materials. The cutting temperatures within the tool can typically range
from 500 to 700~ in a lathe machine. The access of the lubricant into the cutting
zone is difficult and is probably through capillary action. Also how an organic
material can function under the high temperatures is not clear. In practice it is
known that boundary additives are effective and used extensively. It is easy to
simulate cutting, grinding and finishing operations for evaluating different
formulations. In severe operations that involve high temperatures, and limited
access, as in drilling of hard steels neat cutting oils with EP additives are used.
Simulated tests for such cases involve measurement of drill torque and the number
of holes that can be effectively drilled. An interesting example of such studies with
vegetable based cutting oil may be cited [ 17]. Several other issues like corrosion of
the work piece, toxicity, and related issues have to be additionally taken into
account. 4-Ball EP performance level may also be specified in some cases.
7.5.1 Metal forming operations

Many metal forming operations involve mixed lubrication conditions with partial
hydrodynamic effects. The simplest test that can be used to study a lubricant is a
ring compression test. Compression of a lubricant coated surface under plastic
deformation changes the ring dimensions depending on the interfacial friction. A
complex test can involve a laboratory rolling mill to simulate high speed cold
rolling. Such a rig can be very expensive. Several levels of simulation are possible
between the two extremes considered. Rigs being used include small scale low
speed rolling, wire drawing, cup drawing, forging, and extrusion. The problem with
the smaller rigs in many cases is that they cannot simulate the large-scale industrial
operations and the available theoretical base is not adequate to extrapolate the
results to real conditions. An example may be cited from the important area of cold
rolling. While film thickness for cold rolling conditions can be predicted for
smooth surfaces the modelling of friction presents major problems. A recent
attempt to predict friction based on empirical modelling [ 18] is indicative of the
complexities involved. The problem of friction is more complex in deep drawing
where different zones will have different friction coefficients. The importance of
interfacial friction arises in the FEM analysis of the deformation processes
necessary for design. The boundary conditions to be defined necessarily involve

interfacial friction.
The effect of boundary lubrication on roughness and surface integrity of the work
is not governed by friction alone. The nature of surface interaction with the
botmdary lubricant is obviously important. As many tests normally used involve
mixed lubrication it is difficult to isolate boundary effects. A fundamental approach
should involve the capability to investigate boundary effects under plastic
deformation with a wide range of sliding speeds. Such mapping will considerably
aid in understanding boundary lubricant effects that can be integrated into the
overall assessment of the lubricant. A new approach to study the boundary
lubrication problem in plastic deformation developed by Banerjee et al [19] is
described below. The discussion here is conf'med to the main concepts involved.
7.5.1.1 Evaluation of boundary effects
Investigation of boundary effects in plastic deformation at a laboratory level needs

a rig that can operate over a wide range of loads and sliding speeds. A simple and
effective idea used for this purpose was an oblique plastic impact. A steel ball
impacts an oblique plane of the work piece B. The ball indents the work piece with
plastic deformation and also slides along the plane. The initial velocity component
along the plane reduces due to friction as the ball slides. Finally the ball rebounds
with the final velocity components and makes a second impact on another plate H.
The onset of impact and rebound are illustrated in Fig. 7.6. The rig is illustrated in
Fig. 7.7. By modelling the flight from rebound to second impact as a projectile
motion and measuring the time and distance the velocity along the plane at the
point of rebound can be calculated. It can be shown that the average friction
coefficient at the interface during sliding is
f = (VHI - VH2 ) / (V I + V 2 ) (7.5)
where
V 1 and V2 = Velocity components perpendicular to the plane at onset
and rebound
VHI and VH2 = Velocity components along the plane at onset and
rebound
The velocity components are also shown in Fig. 7.6. The detailed theoretical
consideration is available in the reference. From the available geometry it is
7.5 METAL WORKING L UBRICA TION 227
obvious that a wide range of impact loads and sliding velocities are possible by
simply altering the impact height and oblique angle.
V V
Impoct onset Forces & Velocities

ot the boll
(a) (b)
' /vHx
Rebound
(c)
Fig. 7.6. Impact and rebound components of the ball. (a) Velocity components at onset. (b)
Forces and velocities for the ball. (c) Rebound velocity components. (Reproduced from
Ref. 19 by permission of STLE)
The lubricant is applied to the surface of the low carbon steel piece. The nominal
thickness of this layer was 1.0 ~xn to minimise possible hydrodynamic effects
during sliding. It was assumed that the contact essentially operated in boundary
lubrication regime. This assumption is supported by the work of Imado et al [20]
who have observed significant metal contact when traction fluids were studied
under plastic impact with sliding. On the other hand when the oblique impact
involved elastic contact there is evidence of full separation as reported by Jacobson
228 CHAPTER 7. TRIBOLOGI('AL EVAL UA lION METHODOLOGIIs
[21]. The steel surface had R~ roughness of 0.25 l.tm. Different lubricants were
compared with this technique. Two oblique plane angles of 30 and 45 ~ were used.
The fall heights were varied between 0.85 to 2.20 m resulting in an overall
variation of 2.04 to 4.64 m/s in the initial sliding velocities. An EN 31 steel ball
with a diameter of 0.167 m was used for the impact tests discussed below.
The lubricant efficacy was evaluated on the basis of morphology of the crater
surface. The surfaces were mainly observed with an optical microscope. Some
craters were also examined by SEM. Typical difference between effective and
ineffective lubricants is the extent of surface flow caused by asperity interaction.
A-Solid block H-Front plate

B-Specimen l - C h a r g e amplifier
C-Stand J-Dual channel recorder
D-CIQmp K-Steel ball
E-Solenoi d coil L-Connec~in 9 wire
F, G-Accelerometer
o
I
I
I J
u ,
~ .-~
Fig. 7.7. Test rig assembly. (Reproduced from Ref. [ 19] by permission of STLE)
Fig. 7.8a shows the observed crater surfaces for dry, n-hexadecane, stearic acid,
and oleic acid at x 50 magnification. The surfaces were observed at the centre of
the crater except for n-hexadecane. With n-hexadecane the surface observed was
slightly away from the centre and shows the border between impacted and original
suface. The test conditions involved 1.5 m fall height and an oblique angle of 30 ~
The fatty acids were used at a concentration of 5% in hexadecane. In the case of
dry impact as well as hexadecane alone, there is significant plastic flow as well as
evidence of material pull out leaving shallow pits. The flow is such that the original
grooves in the surface are covered in many zones. Similar observation may be
7.5 METAL WORKING L UBRICA lION 229
~ -Crater boundary
Sliding Direction Sliding Direction
(i) (ii)
" . .,, .. ~'~
; '
Sliding Direction ~ Sliding Direction
(iii) (iv)
Fig. 7.8a Optical photomicrographs of impacted craters: (i) dry contact (ii) n-hexadecane
(iii) stearic acid and (iv) oleic acid. (Reproduced from Ref.[ 19] by permission of STLE)
230 CHAPTER 7. TRIBOLOGICAL EVALUA TIONMETHODOLOGIES
made with hexadecane alone though there was some reduction in the scale of
deformation. With oleic acid the clear pattern of original striations can be seen. In
this case the plastic deformation does not extend to the valleys and also there is no
evidence of metal pull out. This is interpreted as due to effective boundary
lubrication that prevents asperity level adhesion and prevents the large scale flow
observed with hexadecane and in dry sliding. The behaviour of stearic acid is not
as effective as oleic acid. In some zones flow similar to dry case can be seen. Such
zones may arise due to partial failure of the film at some contacts. In Fig. 7.8b two
SEM micrographs are included for typical surfaces. These micrographs at high
magnification are given for dry and oleic acid impact. The differences in surfaces
are more obvious in these photomicrographs. Palm oil used neat also showed very
effective boundary action and the observed surface is not included in the figure. It
is reasonable to consider that the large variations in crater morphology with
different lubricants is due to boundary effects. This additionally supports the
assumption that boundary lubrication was involved in the contact. More detailed
consideration with additional lubricants is available in the cited reference [ 19].
The friction coefficient has been found useful mainly in deciding the effective
operation zone for a lubricant based on transition. The friction coefficient was
related to P(VHave )0.5 where P is the normal load and VH.ve is the average of the
velocities along the sliding plane at onset and rebound conditions. The transition
point in friction determines the operating parameters below which the lubricant is
effective.
The above study clearly brings out the possibility of evaluation of boundary effects
by the simple plastic impact technique. Various studies at different temperatures
and wider operating conditions are possible. Some work was also conducted with
regard to the influence of polar compounds on surface hardness. It is of interest to
observe that at identical impact conditions the crater depths were different for
different lubricants. Surface hardness was inversely related to crater depth. There is
thus directional evidence for the influence of polar molecules on strength
properties. Investigators in the past have reported such influences on single crystals
[22]. Plastic deformation with some level of orientation in the direction of sliding
may promote such effects. It is also possible to study the surface reactions by FTIR
or other techniques. In the present case metal transfer to the ball was not
investigated. Such studies can be useful to assess the lubricant failure.
Investigations done with regard to lubricant-metal interactions under plastic
deformation at a fundamental level are limited and the author considers that there is
a need to strengthen this area.
7.5 METAL WORKING LUBRICATION 231
Sliding Direction~
(i)
Sliding Dircction~
(ii)
Fig. 7.8b. SEM photomicrographs of impacted craters: (i) dry contact and
(ii) oleic acid. (Reproduced from Ref. [19] by permission of STLE)
232 ('HAP TER 7. 1RIBOLOGI('AL Et'AL lIA l I O N METHOI)OLOGIES
7.5.2 Other aspects
The above consideration of metalworking lubricants is limited and only an

approach to study boundary aspects was given. The lubricants in practice are
varied and complex. The available literature mentions the lubricants in general
terms and it is difficult to give a detailed account of the products involved. The
available information is given in [23]. The complexity can be appreciated with one
example of the cold rolling of steel. In cold rolling of steel meta-stable emulsions
are common. They should separate fast enough in the roll bite to provide
essentially an oil layer. This 'plate out' characteristic has to be optimised for
different operations and this itself involves difficult technology. Besides boundary
additives several other additives in smaller quantities like brighteners, corrosion
inhibitors are also added. Even in cold rolling the interfacial temperatures can
range from 500 to 600~ The role of lubricant is probably related to its survival in
the short contact time. The decomposed lubricant on the surface should be easily
removable without staining. The roughness and brightness of the finished products
is important and is related to the initial roughness of the roll and work as well as
operating conditions in a complex manner. In hot rolling of steel also lubricants are
used and include all sorts of materials starting from fats to solid lubricants. Several
other ideas used in metal fomaing include coating of the sheet with soft metals as
well as 'conversion coatings' that are obtained by phosphating. Such coatings are
used in conjunction with solid or liquid lubricants. Existing practices and
formulations will have their limits of applicability. For example in the high speed
rolling of steel beyond 12 m/s the existing lubricant technologies were inadequate
and new approaches became necessary. Synthetics and semi-synthetics with
recirculation is one of the effective solutions available today. Ability to screen
boundary performance in such cases becomes important and the approach
suggested above can be useful in screening at high sliding velocities.
The oblique plastic impact technique involves very short contact duration of the
order of 10.6 seconds. At this stage it is not known whether the technique can be
applied to other evaluations. With more severe operating conditions the technique
may be adaptable to evaluate the shock loading capability of rear-axle lubricants.
References
1. C. N. Rowe, Lubricated wear, in M. B. Peterson and W. O. Winer (eds.), Wear Control

2. M. B. Peterson, Design considerations for effective wear control, in M. B. Peterson and
W. O. Winer (eds.), Wear Control Handbook, ASME, 1980, 451-456.
t~FEREN(TES 233
3. G. Salomon, Failure criteria in thin film lubrication-the IRG programme, Wear, 36

(1976) 1.
4. P. Shrivastava, Wear characteristics of running-in oils, M. Tech thesis, liT Delhi, 1997.
5. R. Kumar, B. Prakash and A. Sethuramiah, A systematic methodology to characterise
running-in and steady state wear processes, Wear, 252 (2002) 445.
6. R. W. Hein, Evaluation of bismuth naphthanate as an EP additive, Lub. Eng., 56 (11),
(2000) 45.
7. D. E. Weller and J. M. Perez, A study of the effect of chemical structure on friction and
wear: Part 1- Synthetic ester based fluids, Lub. Eng., 56 (11), (2000) 39.
8. J. M. Perez, D. E. Weller and J. L. Duda, Sequential Four-Ball study of some lubricating
oils, Lub. Eng., 55 (9), (1999) 28.
9. Z. S. Hu, J. X. Dong and G. X. Chen, Synthesis and tribology of aluminium
hexoxylborate as an antiwear additive in lubricating oil, Lub. Sci., 12 (1999) 79.
10. M. Born, J. C. Hipeaux, P. Marchand and G. Parc, The relationship between chemical
structure and effectiveness of some metallic dialkyi and diaryl dithiophosphates in
different lubricated mechanisms, Lub. Sci., 4 (1992) 93.
11. A. Sethuramiah, V. P. Chawla and C. Prakash, A new approach to the study of the
antiwear behaviour of additives utilising metal contact circuit, Wear, 86 (1983) 219.
12. P. J. Blau, Friction and Wear Transitions of Materials, Noyes Publications, New Jersey,
1989, 33.
13. A. Sethuramiah, H. Okabe and T. Sakurai, Critical temperatures in EP lubrication,
Wear, 26 (1973) 187.
14. A. K. Gondal, B. Prakash and A. Sethuramiah, Studies on the tribological behaviour of
two oil-soluble molybdenum compounds under reciprocating sliding conditions, Lub.
Sci., 5 (1993) 337.
15. V. K. Jain, V. P. Sharma and A. Sethuramiah, Application of disc machines for testing
the shock loading capability of hypoid and industrial gear oil, Trib. Int., 19 (2), (1986).
16. A. Sethuramiah and V. K. Jain, Evaluation of rear axle extreme pressure (EP) lubricants
by a disc machine, Wear, 52 (1979) 49.
17. W. Belluco and L. De Chiffre, Testing of vegetable-based cutting fluids by hole making
operations, Lub. Eng., 57(1), (2001) 12.
18. H. R. Le and M. P. F. Sutcliffe, A semi-empirical friction model for cold metal rolling,
Trib. Trans., STLE, 44 (2001) 284.
19. R. K. Banerjee, C. R. Jagga and A. Sethuramiah, Friction and surface morphological
studies in plastic deformation under boundary lubrication condition, Trib. Trans.,
STLE, 44 (2001) 233.
20. K. lmado, H. Miyagawa, A. Miura, N. Ueyama, and H. Fujio, Behavior of traction oils
under impact loads, Trib. Trans., STLE, 37, (1994) 378.
21. B. O. Jacobson, Rheology and Elastohydrodynamic Lubrication, Tribolgy series, 19,
Elsevier, 1991, chapter 15.
234 REf+EREN(,ES
22. M. Ciflan and E. Saibel, Rebinder effect and wear, Wear, 56 (1979) 69.
23. J. P. Byers (ed.), Metal Working Fluids, Marcel Dekker, New York, 1994.
Nomenclature
d ball scar diameter
dH Hertzian diameter
H hardness
K wear coefficient
L machine load in 4-Ball machine
P normal load
t test duration
Trl, Tr2 temperatures at which transition in friction takes place
V wear volume
VH1,VH2 velocity components along the oblique plane at onset and rebound
VHave average of the velocities along the plane, (VHI + VH2)/2
Vl, V2 velocity components normal to the plane at onset and rebound
235
8. Fatigue and wear in mixed lubrication
8.1 Introduction
Mixed lubrication is receiving increasing attention which is focussed mainly on

EHD contacts. The theory of mixed lubrication is considered in the first section.
The author has attempted to explain the main concepts based on physical
reasoning. While theoretical developments are advanced, experimental verification
of the ideas is limited. Experimental aspects are considered in the second section.
The practical utilisation of the modem theory is illustrated with the example of
artificial joints and included as part of the second section.
Fatigue leading to pitting is an important consideration in many EHD contacts.

Pitting in rolling/sliding contacts is considered in the third section. A brief
coverage is first made of the interacting parameters involved in pitting. This is
followed by a practical consideration of pitting in rolling element bearings and
gears.
The lubricant-metal interaction effects on pitting are considered in the fourth

section. This is a research area and the available understanding based on laboratory
evaluations is presented in the first part. The next part is devoted to some pertinent
observations on the mechanisms. These observations bring out the importance of
competition between crack growth and wear.
The fifth section deals with the problem of wear in rolling/sliding contacts. This
section is treated as an extension of boundary lubrication dealt with in earlier
chapters. The contact models are first considered followed by an example of
experimental work.
Contact fatigue and mixed lubrication are specialised areas with voluminous
literature. The author has'presented a concise treatment of the main ideas. It is
hoped that such coverage will provide sufficient insight to appreciate the role of
lubricants in pitting and wear. Detailed theoretical approaches are available in the
cited references.
236 ('HAP TER 8. FATIGUE AND WEAR IN MIXED L UBRICA TION
8.2 Theory of mixed lubrication
Mixed lubrication may be defined as a situation in which the load is shared by the
asperities and the fluid film. The conventional approach is considered first. This is
followed by a discussion of the recent developments in mixed lubrication.
8.2.1 Conventional approach

Mixed lubrication regime may be defined in terms of the film thickness to the
composite ratio called A ratio as considered in chapter 1. The other way of
visualising it is in terms of friction coefficient. As considered in chapter 1 the film
thickness in a journal bearing is a function of the Sommerfeld number,
(rl N/p)(r/c) 2 . For a given bearing with a fixed r and c it follows film thickness
will be related to (q N/p). This term for convenience is now termed S. If the
friction coefficient f is now plotted as a function of S the relationship obtained is
schematically shown in Fig. 8.1.
IA B
,f
C
~N/Ip(S)
Fig. 8.I. Schematic relation between S andffor a hydrodynamic bearing.
This is normally referred to as the Stribeck curve. In this curve the behaviour to the
right of the minimum follows the hydrodynamic regime with no asperity contact.
The film thickness increases with S and the viscous friction increases. To the left of
the minimum the film thickness is inadequate for complete separation and the load
is increasingly shared by asperities as S decreases. Finally at B the total load is
supported by the asperity contacts a n d f remains nearly constant from B to A. The
region C to B is designated as the mixed lubrication regime. Typicalfvalues in the
boundary regime are 0.08 to 0.1, while typical values in hydrodynamic regime are
less than 0.01.
8. 2. THEORY OF MIXED L UBRICA l l O N 237
In usual EHD contacts operating in Elastic-Viscous regime the film thickness

varies with (rl u) ~ as can be seen from Eq. 1.28. The load dependence is small and
may be ignored. Several modified Stribeck curves have been drawn for EHD
contacts to visualise the regimes. The traction varies in a complex manner in EHD
contacts due to the non-Newtonian behaviour of the lubricant. Hence the variations
in friction coefficient will be specific to the lubricant used. The author considers it
better to define the various regimes on the basis of the A ratio. The concept is
common to both hydrodynamic and EHD contacts Recalling from chapter 1 the
regimes based on A ratio are:
Hydrodynamic/EHD > 3.0
Mixed 0.5 to 3.0
Boundary < 0.5
Mixed lubrication is more important in EHD contacts as film thickness in such

contacts is usually small. In steady running the desirable goal is to have minimum
possible asperity interaction.
8.2.2 Recent developments
The above approach to mixed lubrication based on the A ratio needs the value of
film thickness. In the simple approach the film thickness is calculated on the basis
of smooth surfaces. When film thickness is of the same order as roughness, the
roughness will surely affect the flow of the fluid in the contact. Early efforts in this
direction involved the solution of Reynolds' equation modified for the roughness
effect [1,2,3]. The average flow model due to Patir and Cheng [3] was able to
quantify the roughness effects on the basis of orientation. Their model predicted
highest film thickness for transverse roughness as compared to isotropic and
longitudinal roughness. Transverse roughness refers to the case in which the
roughness is predominantly oriented in the direction perpendicular to the entraining
direction. Real roughness is made up of several peaks and valleys. In reference [3]
the degree of orientation was quantified in terms of the ratio of autocorrelation
functions in the two directions. While asperity contacts can occur in this model
their influence was considered only in terms of change in flow factors. The
influence of deformation on roughness was not considered.
From the important beginnings cited above, the study of the roughness effect has
advanced conceptually and mathematically. Firstly there was a realisation of the
238 ('HAPIF?,R 8. t;A TIGIfE AND WEAR IN MIXED L HBRICA TION
fact that as asperities approach each other there will be local pressure fluctuations
as well as deformation. These effects in turn generate separating film at the
asperities. This domain of lubrication is now referred to as micro EHD (MEHD)
lubrication. MEHD may be viewed as a positive response of the asperities to
prevent direct contact. The fluid flow at the asperity level helps the overall macro
flow in the contact and global fluctuations in film thickness are reduced. The
pressure fluctuations at the asperity are, however, large and they may have a
negative effect in terms of fatigue. The solution to such problem involves advanced
fluid dynamics. Several papers with increasing levels of sophistication are now
being published in this area some of which may be cited [4,5,6,7]. Most of the
solutions so far attempted consider sinusoidal roughness that moves in the contact
zone or is stationary with reference to the moving rigid smooth surface. Recently
some models considered three-dimensional roughness [8]. Spikes [9] presented a
useful overview of mixed lubrication where theoretical and experimental issues are
discussed.
The influence of asperity deformation was theoretically evaluated by Kweh et al

[4] and is shown here as an example. The authors investigated the influence of a
sinusoidal roughness in a heavily loaded contact and illustrated in Fig. 8.2. The
maximum contact pressure between the disks was 1.06 GPa with a mean sliding
velocity of 24.87 m/s. One smooth surface rotated against the transverse roughness.
The point of interest is that film thickness calculated at different operating
temperatures, curves a, b, and c, are nearly fiat. This means the sinusoidal
roughness shown along the X-axis is completely flattened. The amplitude of the
roughness in this case was 0.25 lam. The other aspect that can be seen is the
significant pressure fluctuations due to asperities as compared to the smooth
Hertzian pressure variation marked d. The values obtained theoretically depend on
the various assumptions made, but the present example serves the purpose of
visualising the strong influence of MEHD lubrication.
The above studies were mainly confined to the situations where there are no
asperity contacts. The film thickness in MEHD does not depend only on the
pressure fluctuation and deformation. It also depends on the shear behaviour of the
film. Theoretical modelling of the shear behaviour depends on the rheological
model selected as well as the temperature involved. The film thickness involved in
MEHD at the asperities is at the nano level when conditions become severe. This
thickness is similar to the dimensions of typical boundary films. Thus it is difficult
to separate the two effects. This also has a bearing on rheological models. The
available models are based on bulk fluid flow and their applicability to very thin
films can be questioned. Such problems are very difficult to resolve. These issues
8.2. THEORY OF MIXED LUBRICATION 239
will also have a bearing on the scuffing models being developed and discussed in
chapter 5.
16
a,
nl ~2 v
{L
0
LU
LU Z
IX
:3 U
Vl
u') I
U.I
IX
~L I
..J
----JO
-25 "-20 -"15 "-10 -S 0 5 10 15 20
x/a x 10-1
Fig. 8.2. Longitudinal centreline pressure distributions and film thickness profiles for d =
0.25 l.tm. (a) 140~ (b) 80~ (c) 50~ (d) d = 0, 140~ (Reproduced from Ref. [4] by
permission of ASME)
The asperities eventually make contacts due to their inability to sustain full MEHD
effect. Some asperities will sustain the MEHD effect while some others will make
contact. Such problems are again being modelled by considering load sharing
between asperities and fluid films [8]. There are theoretical attempts to re-define A
ratio on this basis.
Another issue of relevance is the composite roughness. When two rough surfaces
are in relative motion their contact is simulated by an equivalent surface moving
against a rigid fiat surface as considered in chapter 1. Such an approach is
commonly used in modelling surface contact. In some cases there can be asperity
level conformity between the surfaces with peaks of one surface matching with the
valleys of the other surface. In such a case the composite roughness cannot be
defined on this basis. This aspect will be considered in the next section with an
example.
240 CHAPTER 8. FA TIGI~:E AND WEAR IN MIXED LUBRICATION
8.3 Experimental considerations
8.3.1 Film thickness
The significant advances in theory need to be backed by experimental verification.

First consider MEHD without asperity contact. Any verification of the theory
demands an ability to measure nano level film thickness. Film thickness
measurements are possible by electrical and optical techniques. The commonly
used electrical techniques are based on capacitance and eddy current. Such
techniques will have a resolution of the order of a micrometer and are unsuitable
for measuring very thin films. Capacitance techniques were used in the past to
measure macro EHD film thickness and served the useful purpose of verifying the
theoretical predictions. Eddy current techniques are more common for measuring
thicker hydrodynamic films. An effective optical technique by light interference
was first developed by Gohar and Cameron [10] and is now widely used by several
researchers to measure EHD film thickness. This technique can be applied only
when one of the contacting surfaces is transparent. A glass or sapphire disk is used
with a semi-reflective chromium coating facing the rotating steel ball. When light
is shown through such a contact there is reflection from the coated surface and the
steel ball. The reflected beams are out of phase and the resulting interference
fringes form the basis for film thickness measurement. The minimum thickness that
can be measured by such a technique is around 100 nm. A modified technique with
a silica spacer that can measure film thickness as low as 5 nm has been developed
and reported by Johnston et al [11]. The composite roughness involved in the
contact was 11 nm and the authors assumed that this roughness may be effectively
flattened due to micro EHD effects and film thickness measurements of 5 nm are
meaningful. With this technique influence of additives on the film thickness is also
being studied at the nano level [12] and it is hoped that the influence of boundary
and MEHD films may eventually be clarified.
The experimental work of Kaneta et al [13] exemplifies the roughness influence

with single bump as well as transverse and longitudinal patterns. They used the
conventional interferometry techniques to measure film thickness. These studies
directionally support the existing theoretical ideas. For example, their results for a
transverse asperity showed minimum deformation in pure rolling as compared to
rolling/sliding contact. This is in accord with the present theoretical understanding.
The minimum thickness that can be measured with such a technique will be about
100 nm and so verification cannot be extended to zones involving nano level
8.3. EXPERIMENTAL CONSIDERA T"IONS 241
thickness. The influence of real surface roughness is yet to be clarified

experimentally.
8.3.2 Asperity contact
Asperity contact through the films is of practical importance. Precise identification

of MEHD film failure is an experimentally difficult task. The problem of asperity
contact has been studied over a long period of time through observations of friction
and metal contact as a function of operating conditions. A better insight into the
problem can be achieved when friction and metal contact are simultaneously
observed. As it is not easy at present to estimate micro effects for real surfaces the
observations are generally based on A ratio calculated on the basis of smooth
surface film thickness. The metal contact circuit was first reported by Furey [14]
and later used with and without modifications by several others. The main concept
involved is illustrated irl the circuit given in Fig. 8.3.
Battery
it
" if
R3
i,~
c
L -0
my Re.corder
Fig. 8.3. Metal contact circuit
A small DC potential of 10-20 mV is applied across the contacting surfaces by

suitably selecting the R~/R2 ratio. In the lubricated contact the lower disk D rotates
against the stationary ball. The disk is insulated from the shaft while the ball is
insulated from the holder. The leads from the rotating and stationary contact are
connected to a milli volt recorder via RC circuit. When an oil film separates the
242 CHAPTER 8. FATIGUE AND WEAR IN MIXED LUBRICATION
sliding contact, the full applied potential will be the output. On the other hand
when there is asperity contact there is shorting and the potential is nearly zero. The
RC circuit modulates these variations and an average voltage is recorded with time.
The actual contact situation can be seen by connecting a CRO across the contact. In
such a case the fast changes from zero to full potential can be seen. The main
problem of the circuit is that shorting can occur with one or more contacts. Thus
the circuit does not provide information on the actual number of contacts. But
statistically the probability of contact increases when more asperities tend to
contact. The average output increases with decreasing contact. The nature of the
circuit used is such that the variation is non-linear. The usual approach is to relate
the milli volt output to the contact percentage or separation percentage. The overall
response of the circuit depends on the resistance and capacitance values selected. A
modified circuit [ 15,16] can be used to obtain linear response instead of a non-
linear response. In the modified circuit square pulses are generated at high
frequency. A comparator allows the pulses to pass through a gate only when a
metal contact is detected. The pulses are then integrated over a selected short time
interval and displayed as output voltage. The metal contact circuit is now usually
referred to as the electrical contact resistance (ECR) technique.
One recent example of the use of ECR may be cited. Lugt et al [17] studied the
influence of running-in in mixed lubrication in a disk machine. They studied the
influence of running-in under pure rolling and with a small sliding component. The
completion of running-in was based on the selected threshold value for electrical
contact resistance. The running-in was first completed in pure rolling. When the
operation was then shifted to rolling/sliding condition the surfaces had to be run-in
again to achieve completion. To ensure completion of running-in for both
conditions, the process had to be repeated by shifting from one condition to the
other. This work showed that completion of running-in achieved at one condition
might not be adequate for another condition.
The detection of metal contact depends on the low resistance assumed at the
contacting asperities in comparison to the very high resistance of the oil film. This
assumption is normally valid. But cases can arise where coherent oxides and/or
reaction films are involved at the surface. In such cases there can be high resistance
due to the films that may be interpreted as the oil film by the circuit. Also when
nano level films are involved they may conduct electricity due to the tunnelling
effect. In such cases though a separation exists between the two asperities it will be
interpreted as a contact. Such considerations may become important if ECR is to
applied for detecting MEHD film failure. Any fine-tuning of the circuit will depend
on the resistance of the now conductive nano films.
8.3. EXPERIMENTALCONSIDERATIONS 243
8.3.3 A s p e r i t y l e v e l c o n f o r m i t y
As introduced in the sub-section 8.2.2 asperity level conformity refers to a situation

in which there is matching at the asperity level. Such conformity may extend
throughout the surface or may be localised and has been reported in the literature
[18,19]. A detailed investigation of this aspect was conducted by Tyagi and
Sethuramiah [20,21]. The tests were conducted in a disk machine with a cast iron
and aluminium alloy disk pair with two different lubricants. The disks rotated in
opposite directions at 2000 and 2250 rpm resulting in a relative sliding speed of 10
m/s. The upper aluminium disk rotated at the lower speed of 2000 rpm. One
example of conformity obtained with diester is shown in Fig. 8.4. The surfaces
obtained involved four step load tests of 5 minutes each followed by prolonged
running for 10 minutes at 294.30 N. The surfaces clearly show the peaks of one
surface matching with the valleys of the other. The roughness was measured
leaving one mm from the edge. The measurements were done in the transverse
direction. To represent contact condition the upper disk profile was inverted. The
matching occurred throughout the contact and similar patterns were observed all
along the surface.
Lower Surface
Fig. 8.4. A case of conforming surfaces. (Reproduced from Ref. [20])
The influence of roughness in such a contact cannot be treated in terms of the usual
composite roughness cr * . It may be recalled from chapter 1 that a * = x/ a, 2 + a 22
where or, and o-2 refer to the roughness of the two surfaces.
The probability of contact for a given film thickness will now be far lower than that
based on cr ~. The effective roughness will be composed of the difference in
heights at each point. This roughness profile, called composite profile was
generated on the basis of digitised values of the filtered roughness. An example of
the individual profiles and the composite profile are given in Fig. 8.5 for the diester
lubricant called oil B. The rms value of the composite profile, rmsz, was 0.22 lam as
compared to that based on random contact, rmsi,2, of 1.35 ktm. The value of 1.35
pm is what was obtained on the basis of the equation above. The cross-correlation
value of 0.96 mentioned in the figure needs an explanation. The roughness profiles
were measured with nominal relocation. This may not exactly correspond to the
contact situation. Cross-correlation was done by slight shifting of one profile
SCALE:
HORIZONTAL: 1-25mm/Ol V.
VERTICAL . 557,um/OlV.
-3.51 4.04
.8
1.357
1.914
2.417
3.028
3.585
4.142 - ~
4.699 J 9 J
- 3.51 4.04
{o)
SCALE :
HORIZONTAL : 1.37mrn]DIV. SCA LE :
VERTICAL 9557AJmIDLY. HORIZONTAL-.S3rnm/DIV.
- 4.2 4.04 VERTICAL:.571Jum ]DIV.
1.357 !.371 -
4.699[- ~ , 4.79700001. . . . .
-4.2 4.04 -1.84 1.37
(b) (c)
Fig. 8.5. (a) Filtered R-profile, CI disc step-load test, 294.30 N, oil B; (b) Filtered R-profile,
AI disc step-load test, 294.30 N, oil B; (c) Composite profile of (a) and (b), cross-
correlation coefficient = 0.96, r.m.s.l,2 - 1.35 Jam, r.m.S.z- 0.22 ~tm.
(Reproduced from Ref. [20])
8.3. EXPERIMENTALCONSIDEt~TIONS 245
relative to the other in the computer. The relative location of profiles at which the
cross-correlation function reached a maximum value was obtained. This position
was taken to be the actual contact between the surfaces. The maximum value of the
function was 0.96 in the given example.
The importance of conformity can now be appreciated. For example with an

assumed film thickness of 1 ~tm the normally calculated A value will be 0.74
suggesting significant metal contact. But the A value based on composite profile
will be as high as 4.55 suggesting negligible contact. A limited attempt was made
to study the onset of metal contact based on the A value for the composite profile.
The values for onset of contact obtained for two cases were 2.11 and 2.26.
Normally the onset of metal contact is expected for A value around 3.0. The
observed values were lower, and detailed investigation was not conducted in this
regard. It may be observed that the film thickness calculation was based on smooth
surfaces. It is not known whether there was any increase in film thickness due to
the transverse roughness generated. The details including observations on friction
coefficients are given in the cited references.
The mechanism of development of such conformity is not clear and may involve
micro abrasion by third bodies generated in the contact. But if it is achieved it is
expected to reduce metal contact significantly and the influence can be more
important than MEHD effects. Promotion of such effects by chemical additives or
other means can be practically useful. Also it may be worthwhile to re-examine the
concept of equivalent composite roughness, even if there is no matching. Surfaces
can have random or oriented roughness and it is necessary to define the conditions
in which the normally used concept of equivalent roughness against a smooth rigid
body is valid.
8.3.4 Advances in practice
Improved mixed lubrication is a desirable goal to be achieved in practice. The

general aspects are well known and used extensively to improve practice. The
modem theoretical developments are not yet widely used in practice. Some of the
theoretical concepts have been utilised in the EHD analysis of artificial joints. One
interesting example is the ceramic-polymer joint that is briefly considered here. For
the present purpose the artificial joint is simply considered a ball and socket joint
from the lubrication point of view. Relevant references cited in chapter 6 provide
detailed information about these joints. The problem is to have minimal contact
between asperities when lubricated by the synovial fluid. There are two basic types
of material combinations being used for such joints. One type of joint involves a
246 (-'HAP TER 8. FA FIGUE AND WEAR IN MIXED L UBRICA HOAr
metal or ceramic ball sliding against UHMWPE. The polymer is attached to a

metallic backing. Polymer being an elastic material, significant elastic deformation
at the asperity level is expected. The solution of such a problem needs
incorporation of the time dependent variation of asperity heights into the Reynolds'
equation. The theoretical evaluation [22] showed that considerable MEHD support
is available in such contacts when calculations were based on transverse sinusoidal
roughness. This translates into increased life of the joint as asperity contact is
reduced. This is an interesting example of the use of MEHD lubrication in practice.
Another approach to reduce asperity contact is by making the roughness very low.
This approach is now being adopted for metal-metal, metal-ceramic and ceramic-
ceramic joints of more recent origin. The film thickness calculation is direct by
using the applicable Elastic-Isoviscous equation [23]. The ball and socket joint
operates with small clearances. For such conditions the elastic effects are high
enough but the influence of the pressure on viscosity is negligible. Hence the
pressure effect on viscosity is neglected. The calculations at the applicable
conditions of the joint give typical film thickness values of 20-50 nm. Mixed
lubrication in such joints is being achieved by using very smooth surfaces. The
roughness levels achieved for these surfaces are 5-10 nm which are feasible with
the modem technology available. Such a system can also be a useful tool to study
nano tribology with engineering stresses. This example makes use of the
conventional theory but is cited as an example where very smooth surfaces can be
used to reduce the asperity contact. Besides artificial joints surfaces with nano level
roughness find application in magnetic storage devices and micro electro
mechanical systems. Most of the engineering systems, however, continue to
operate with conventionally manufactured surfaces with roughness ranging from
0.05 to 1.0 ~tm.
8.4 Pitting in rolling/sliding contacts

In rolling/sliding contacts the repetitive stresses induce fatigue which ultimately
leads to pitting. Pitting results in material removal in local areas leading to
malfunction of the components. The pits are deep and visible to the naked eye. The
term 'spalling' is also used to describe pitting of the catastrophic kind. In some
cases there can be micro pitting that may heal itself. In this section the general term
'pitting' is used to describe fatigue induced failure. At this stage it is necessary to
distinguish between fatigue wear and pitting. Fatigue wear involves material
removal at the asperity scale and is expected to be governed by the cyclic stresses
at the asperities. On the other hand pitting involves deeper cracks and visible pit
8. 4. PITTING IN ROLLING SLIDING CONTACTS 247
formation. The section first examines why pitting occurs. This is followed by a
practical consideration of rolling and rolling/sliding contacts.
8.4.1 Role of contact stresses
The line and point contacts are subject to high stresses even when they operate
without asperity contact. First consider pure rolling. The maximum Hertzian stress
Po in such contacts can exceed the yield pressure of the material. As stated in
chapter 3 the maximum shear stress is 0.30po for line contact at a depth of 0.78b
while it is 0.3 l a for a circular contact. These maximum shear stresses act on an
oblique plane. There can be local plastic deformation at these depths if the shear
stress exceeds 1.67Y. When inclusions are available at these zones, cracks nucleate
at these points. The cracks grow with cyclic stressing leading to pits on the surface.
When asperity contacts occur there are large pressure fluctuations superimposed on
the Hertzian distribution. These local stresses affect the stress distribution in the
near surface zone. Below a certain depth the distributions follow the macro
distributions obtained on the basis of the smooth surface. The pitting problem now
becomes more involved due to the interaction of the stresses at two levels.
In rolling/sliding contacts there is frictional traction at the asperities as well as in

the bulk film. The major influence of friction is to induce tensile stresses at the
surface. As considered in chapter 3 in a line contact the maximum tensile stress
acts at the trailing edge and has a value of 2fp0 where p0 is the maximum Hertzian
stress. The tensile stresses act at the asperity level as well as the macro Hertzian
contact level and will be higher for the asperities as compared to the macro contact.
The influence of tensile stresses is mainly at the surface as these stresses fall
rapidly with depth. These stresses lead to the development of cracks at the surface
level. Cracks will also be initiated due to shear stresses as discussed in the previous
paragraph. With regard to shear stresses an additional factor is to be considered
when frictional traction is involved. When friction is involved the point of
maximum shear stress moves upwards thereby influencing sub-surface stress
distributions. Also the value of maximum shear stress increases with friction. The
response of the material to fatigue depends on the complex interaction of the
stresses involved making it difficult to predict the fatigue life.
The stress influence on fatigue was considered in terms of a line contact. Similar
considerations are applicable to point contacts. With regard to point contacts one
difference may be noted. In point contacts tensile stress is involved even in
frictionless contact. The maximum tensile stress for this case can be obtained from
Eq. 3.6. For typical steel surfaces the value would be 0.133p0 and this stress can
induce tensile cracks.
248 CHAP TER 8. FA 7"1G [:E AND WEAR 1N MIXED L (/BRI CA TION
It is hoped that the above qualitative consideration will help the reader to
appreciate the factors governing the fatigue in rolling/sliding contacts. The asperity
contact is now well known to be detrimental to fatigue life. Quantitative theoretical
predictions are still not adequate and the present practice involves semi-empirical
approaches. The approaches available for rolling element bearings and gears are
considered below.
8.4.2 Rolling element bearings

The main approach to fatigue life of these bearings is based on the Lundberg-
Palmgren theory [24]. The original equation considered that life was based on the
value of maximum sub-surface orthogonal shear stress. Orthogonal shear stress
acts parallel to the surface. The maximum orthogonal shear stresses will be
somewhat lower than the maximum shear stress considered above and acts at a
lower depth. The final equation for bearing life was later expressed in terms of
bearing loads and is the most widely used for life estimation. Following the
terminology of ISO 281-1990 the relationship is
L,0 = (C/P) n, L c,0 (8.1)
Lc~o is the life under a known dynamic load of C. Lc~0 refers to the number of
cycles to failure with 90% probability of survival and is equal to 106 cycles. Thus C
represents the dynamic load that results in a life of 106 cycles. The value of n~ is 3
for ball bearings and 10/3 for roller bearings. P is the equivalent applied load and is
calculated on the basis of the given procedure taking into account the radial and
axial loads. L10 is the fatigue life obtained at the applied load P.
The life of a bearing is affected by the nature of the steel as well as operating
conditions. The operating conditions that affect life include temperature, load
distribution, and lubrication. It is also necessary to introduce a factor if life has to
be estimated for any failure probability other than 90%. This is empirically
accommodated by multiplicative factors as given in the standard. Life with any
failure probability can now be expressed as
L,a = ala2a3Llo (8.2)
Where a~, a2, and a3 are the adjustment factors for reliability, material properties,
and operating conditions. The approaches are empirical and effective selection of
adjustment factors is an ongoing debate. Additional factors proposed include those
8. 4. PITTING IN ROLLING SLIDING CONTA('TS 249
due to dust particles. Any particles that enter the contact will intensify stress effects
at local level and so can affect fatigue life.
Fatigue life determination is statistical in nature. Cumulative failures have to be

plotted as a function of failure cycles. From such plots and assuming Weibull
distribution the life with a given probability of survival is determined. The Weibull
distribution function [25] is accepted as the best way to characterise the relation
between cumulative failures and life. A large number of failure tests have to be
completed to justify the use of Weibull distribution and evaluations are
experimentally intensive. Also there is no reason to suppose that the influence of
various factors is multiplicative. One factor can influence the other. The more
fundamental approach to the problem is based on failure initiating stress proposed
by Ioannides et al [26].
Lubricant influence based on A ratio has been investigated by Tallian [27] and
Skurka [28]. These studies showed clear improvement of fatigue life with film
thickness. The observations of Tallian showed that life increased by 1.5 times as A
increased from 1.0 to 3.0. Skurka found a steeper rise by more than three times for
the same variation in A. Skurka found that life did not increase with A beyond 3.0.
However progressive increase was observed by Tallian up to a A value of 10.0.
Guidance regarding the a3 factor mentioned above is based on these studies. The
lubricants studied were mineral oils without additives.
8.4.3 Gears
Available guidance for pitting life of gears is based on experimental studies and
may be considered as a conventional model. As discussed by Dudley [29] the
pitting life can be calculated from the following equation
Lb/L a = (Wta/Wtb )"-" (8.3)
La is the number of cycles to failure at a given tooth load of Wt~ while Lb is the
number of cycles to failure at any other tooth load of Wtb. L~ is the known value at
the corresponding tooth load. The value of n2 is taken to be 3.2 for the boundary
regime, 5.3 for mixed regime and 8.4 for full film regime. The values of the
exponents once again show the strong influence of A ratio on pitting life. Pitting is
usually found to occur near the pitch line while the scuffing problems are
encountered in the zones of maximum sliding velocity.
It may be appreciated that the prediction of pitting is semi-empirical and can only
be obtained with reliable known information on pitting life at a given condition.
Also the data are based on case hardened steels and the exponents can be different
for different materials. Fundamental approaches based on stress analysis on the
lines proposed for rolling element bearings are now being attempted.
8.5 Lubricant-metal interaction effects in fatigue
It is now well recognised that asperity contact through EHD films reduces the
fatigue life. As considered above this effect is accommodated in real systems
through an adjustment factor for life. There is no adjustment factor to account for
chemical additives. This is because the effects are ill understood and this is a
research area. Many rolling/sliding contacts use additives. For example gears with
A ratios less than 2.0 normally use mild EP additives in the oil to protect surfaces
from scuffing. A cam and tappet contact in an engine will be lubricated with engine
oil containing several additives. It is hence necessary to know whether chemical
additives and base oils affect fatigue life. The present section first considers the
main findings in laboratory rolling and rolling/sliding testers. The mechanisms are
not well understood and the second part discusses the mechanisms from a different
perspective.
8.5.1 Laboratory studies
Laboratory simulations to study the lubricant and additive influence have been
conducted with disk machines and 4-Ball machines over a long period of time. In
fact the early experimental verification of EHD was with disk machines. The tests
can be conducted under controlled conditions with proper material control. Also
different slide/roll ratios can be attained in the same machine. The present
consideration will be mainly confined to the influence of lubricants and additives
on fatigue.
In rolling/sliding contacts the slide/roll ratio is defined as
SRR = 21u, - u z l / ( u , +u2) (8.4)
Where u~ and uz are the surface velocities of the two rotating disks. Relative sliding
velocity is the difference in the surface velocities of the two disks. The average
rolling speed is equal to (u~ + u2)/2. When the surface velocities are equal, and in
the same sense, the situation is one of pure rolling and the relative sliding velocity
8.5. L UBRICANT-METAL INTERA CTION 1;5FFF~CISIN FA TIG UE 251
is zero. Normally tests are conducted with surface velocities in the same sense with
varying sliding velocities. In some cases the tests may be conducted with opposite
peripheral velocities. In such a case, the relative sliding velocity will be the sum of
the surface velocities.
Scott [30] and Rounds [31] did extensive work with a rolling 4-Ball tester with
high stress levels. In the rolling tester the lower balls are free to rotate thus
providing rolling contact. The fatigue failure is assessed on the basis of the
vibration level. The machine is cut off when vibration exceeds the pre-selected
value. The failure time is relatively low in view of the high stress imposed. Scott
reported on the deleterious effects of water content on fatigue. Rounds studied
different mineral oils. He also studied the influence of additives on fatigue life. He
found that with naphthenic base oil, oleic diamine, oleic alcohol, and oleic acid
increased fatigue life at concentrations up to 2%. With higher concentrations the
fatigue life decreased. The maximum increase of about four times was observed
with diamine followed by oleic alcohol and oleic acid. Similar effects were
observed with zinc dialkyl dithiophosphate and sulfurised terpine oil. Chlorinated
paraffin reduced fatigue life significantly while TCP had no effect. He also found
differences in fatigue life between napthenic and paraffinic oil. Napthenic oils
showed higher fatigue life. Steel obtained by different processing routes also
affected fatigue life. The influence of additives varied significantly with the nature
of refining of the naphthenic oils. Another interesting observation is with regard to
pre-formed films formed by heating in additive solutions. A large increase in
fatigue life was observed with such pre-treated balls in some cases. The pre-
treatment effect depends strongly on the time and temperature used.
A very extensive investigation involving several hundred tests was conducted on

behalf of ASME in early seventies and reported in three publications [32,33,34].
The objective of the programme was to understand the influence of material and
lubricant chemistry on fatigue life. The tests were conducted in a disk machine at
two different SRR values of 3.3% and 30%, which were called slip percentage.
Several types of steels and lubricants were used for studies. The contact stresses
varied from 1.11 to 4.12 GPa. Large scale regression analysis was done. The data
had significant scatter and the regression lines were the best possible fit of this
data. The following observations can be made from this work:
With paraffinic mineral oils the chemical additives, which showed a

beneficial effect in the work of Rounds, decreased the fatigue life in these
tests. The base oil used by Rounds was napthenic and this could be one of
the reasons for the variation. Oleic acid was also found to reduce fatigue
252 CHAP TER 8. FA ITGUE AND WEAR IN MIXED L UBRI('A TION
life significantly. These observations demonstrate that influence of

additives is specific to the stress levels, base fluids, and other parameters.
ZDTP additive improved fatigue life with increased stress eventually
approaching the life for the mineral oil.
,, Fatigue life with low SRR was significantly higher than with high SRR.
This may be expected as increased sliding will result in higher
temperatures in the contact and reduced film thickness.
. Commercial pentaerythritol ester with an additive package was similar in
performance to the mineral oil at high stress levels.
There are many variables considered including flash temperature, inlet viscosity,
pressure coefficient of viscosity, surface finish, and slip ratio. Such a large
regression analysis leads to some peculiarities. For example the regression equation
resulted in a negative coefficient for viscosity. In other words fatigue life decreased
with an increase in viscosity. Several of these issues were discussed and are
available with the cited papers. The importance of this work is that generalisations
are difficult with regard to additive influence. A directional argument provided in
Part III [34] suggested that the additive and base fluid influence is related to a
combination of corrosion and protective film formation.
More recently Wang et al [35] conducted rolling fatigue tests in a 4-Ball machine
with several different base fluids. They studied eight mineral and synthetic fluids.
Experiments were conducted at different stresses that ranged from 8.1 to 9.2 GPa.
An interesting feature of the work is the determination of the pressure coefficient
of viscosity, (z, by experimentation. The main outcome of these studies is that the
fatigue life is related to A provided proper ot value is used in the film thickness
calculation. The oils included mineral oils, polybutylene, polyglycols, and a polyol
ester. It thus appears that the A ratio is the primary variable irrespective of the wide
variations in the nature of base fluids. It may be observed that the approach used
was to calculate film thickness at a lower temperature of 23~ and a higher
temperature of 80~ and explain the trends from the ability to form EHD film.
Film thickness in each case at the operating conditions was not calculated. The
authors stated that the race temperature varied from 60~ to 90~ Accurate
analysis is only possible when film thickness is calculated at the operating
temperatures in each case. Amongst the individual classes the curves relating stress
to failure cycles had somewhat different slopes. Two such curves for different
fluids with similar viscosity at 100~ are reproduced in Fig. 8.6. The stress-life
curve is a familiar representation in the area of fatigue and fracture mechanics. As
stress decreases, life increases. When the cumulative number of failures is plotted
against number of cycles at a given stress, the L~0 and Ls0 lives can be obtained
8.5. LUBRICANT-METAL INI'ERA('HON E1;FE('IS IN FA H(;~/E 253
from such curves. L~0 represents 90% probability of survival and will be lower than
the Lso value. The authors found that ZDTP improved the fatigue life of polyol
ester and attributed this to film formation that reduced asperity stresses. These
variations in slope show that comparisons on the basis of one stress level alone are
inadequate.
~,~oU ~ .~
~s~oo ~ p A.c~- , t
~'~
~oo~
Stress Cycles (x|.E6}

(a}
.~ 91oo
~ aToo - p.~, G S,L~O
om o.1
(b)
Fig. 8.6. S-N curves with three mineral and synthetic oils with viscosity of 9 cSt at
temperature 100~ S.N.-350 (mineral), TMP-05 (polyoi ester) and P.A.G.-9 (polyalkyl
glycol): (a) 10% life L~o, (b) 50% life Lso. (Reproduced from Ref. [35]).
The influence of base fluids can change with temperature. Some fluids can be more
reactive inducing crack growth. However the first step is to evaluate the A ratio as
accurately as possible. Then the chemical effects, if any, can be separated. The
main problem is the determination of ct for which many laboratories are not
equipped.
8.5.2 O b s e r v a t i o n s on m e c h a n i s m s
The influence of chemical additives is variable. Invoking the possibility of

corrosion and film formation as competing mechanisms is to be probed further.
Most of the chemical additives tried have a significant beneficial effect on wear in
normal boundary lubrication. These include conformal as well as concentrated
contacts. Chemical interactions are involved at sliding asperities and the
mechanisms involved were discussed in earlier chapters. If increased corrosion
decreases fatigue life, it is also expected to increase chemical wear. In reality the
additives reduce wear rates by orders of magnitude as compared to the base fluid.
254 ( ' H A P T E R 8. t:A 71GUI ( A N D WEAR I N M I X E D LUBRICA l I O N
In effect chemical reaction is being invoked simultaneously to explain low wear

and low fatigue life. A better explanation is necessary which is attempted below.
Surface cracks are detrimental to fatigue life. These cracks develop at asperity
contacts due to tensile stresses and the near surface shear stresses. The near surface
shear stress distribution will have a sub-maximum which is higher than the sub-
surface maximum value. The influence will be more important with rougher
surfaces. Many surface cracks develop, some of which only propagate to critical
depth after several stress cycles. The crack grows with each stress cycle in a non-
linear manner leading to a pit that may be considered as a local fracture event.
These ideas do not take into account the wear process. When there is wear two
effects can occur. One is the removal of some of the initiated cracks as they form
and the other is the reduction in the depth of the partly grown cracks. Any
reduction in crack depth would mean more cycles to pit formation increasing the
failure life. Also partial removal of cracks at the initiation stage will also reduce the
possibility of development of critical sized cracks. In other words wear contributes
to increased fatigue life. Fan et al considered the competition between crack
propagation and wear [36]. The emphasis in this paper was on the influence of
hardness on such competition. This paper supports the qualitative argument
presented here. This argument may be extended now to the role of additives as
discussed below.
As wear occurs in sliding contacts it is best to consider a rolling/sliding contact as

an example. First consider that an additive is effective and the wear rate is low.
Also assume that careful experiments at the given conditions revealed that the
fatigue life decreased with the additive in comparison to the base fluid. This is
exemplified by the large scale work in rolling/sliding contacts considered in 8.5.1.
How can these effects be reconciled? When wear rate is low as with antiwear
additives the crack removal rate will be low. Also the crack depths will be less
affected. This leads to higher probability of failure. Thus directionally low wear
and low fatigue life can be reconciled. Some other additive effects will tend to
prolong the fatigue life. The stresses may be more uniformly distributed reducing
asperity normal stresses. The tangential stresses also will be reduced due to lower
friction. Provided such effects are secondary, the fatigue life should decrease when
low wear is involved. This argument may be considered as the mechanical aspect.
It does not take into account the influence of the additive itself on crack growth.
The reaction film formation is a rate process. The cracks grow as the film builds.
How the additive affects these initial processes is not known and is complex.
Unlike the steady state wear in which quasi equilibrium may be envisaged, the
8.5. LUBRICANT-METAL INTERACIION EH'EC1S IN k'A TIGUE 255
initial running-in process is more random. While crack generation can be high at
the high spots there is also high wear at these spots and film formation. But pitting
life is orders of magnitude higher than the running-in period. It may be reasonable
to consider crack growth on the basis of the steady wear zone. In the steady wear
zone one can anticipate cases of high crack growth due to chemical interaction.
With good antiwear additives strong corrosive reaction does not occur since they
form effective barrier films as discussed in chapter 5. Any major corrosive
interaction should simultaneously increase the wear and cannot be reconciled with
the low wear rates observed with these additives.
Validation of the arguments given so far can, in principle, be done in disk machines
by observing the wear rate and crack growth rate with different additives. As crack
growth is non-linear, and crack depth itself varies due to wear, the problem is very
difficult to handle. Indirect evidence can be gathered if wear rates and fatigue life
are obtained over a range of sliding speeds in the boundary regime. Boundary
regime will avoid the uncertainties involved in mixed lubrication and it will be
easier to interpret the additive effects. It is expected that low wear rates should
result in low fatigue life for a given additive if the influence on stresses and friction
are secondary. Chemical additives increase fatigue life if 'secondary' effects
become 'primary'. As the interest in practice is to obtain increased fatigue life, the
'secondary' influences should become 'primary' influences. In other words
positive additive effects on fatigue are possible when crack growth reduces
substantially due to low friction and low asperity stresses. For example the reported
observation [34] that ZDTP is relatively more effective at higher stress can be due
to the nature of film. The film formed at the higher Hertzian stresses may reduce
the asperity stresses and friction more effectively. Such an effect in some cases
may be related to increased film thickness.
The above argument is valid for conditions in which wear occurs due to relative
sliding. In nominally pure rolling wear at the asperity level is expected to be
negligible. Also tangential stresses at the asperities will be lower. Though normal
stresses at asperities are high the net crack growth will be lower due to the lower
tangential stresses. Thus in pure rolling the failure life can be substantially higher.
Chemical additives can still react forming films. As there is no wear the influence
of chemical additives on fatigue life should be attributed to their influence on crack
growth. In the 4-Ball tests considered earlier the stresses used were high and
ranged 7.0-9.0 GPa. While such high stresses have the advantage of reducing the
failure time, these stresses are unrealistic as there is significant plastic deformation
involved in these tests. Chemical reactions can be significant under such high
stresses. Two possible additive effects can be envisaged. If reaction film formation
is fast enough this will hinder the initial processes of crack generation and growth.
In such a case the fatigue life should increase. If the film formation is slow the
fatigue life may not be affected. The work due to Rounds [31] showed that in some
cases pre-formed reaction films increase fatigue life several fold. It can be argued
that these effective films increase life because they hinder the crack growth until
they get removed from the surface. This is not possible if the film is ineffective or
the additive has corroded the surface during pre-treatment. The influence of such
pre-treatments is not known for rolling/sliding contacts. Such studies may provide
additional insight into the fatigue mechanisms.
The two effects considered above are mechanical in the sense that the influence of
reaction film is only considered in terms of crack growth probability. The third
possibility is the influence of the chemical additive on the crack growth itself. This
possibility was already considered for rolling/sliding contacts. In the present case
the added complication is the short duration of tests in which steady state cannot be
assumed. Also reaction film composition will be different for situations involving
wear and no-wear. It is hence necessary to apply realistic stress levels to link to
practice. More meaningful evaluation for rolling contacts is possible only when
stress levels are lower and correspond to the levels used in practice. The
evaluations considered in 8.4.2 are usually conducted at realistic stress levels and
should be adopted for any meaningful evaluation.
Several factors influence fatigue and the above argument highlights one important
aspect that should be taken into account. Several recent experimental and
theoretical investigations are being conducted to understand the role of asperity
stresses on fatigue. One recent interesting experimental investigation may be cited
[37] which provides a state of art consideration of the problem.
8.6 Wear in rolling/sliding contacts
Wear occurs due to relative sliding between surfaces. Pitting life has been
extensively studied because of its effect on failure. The wear problems in these
contacts are less widely studied. Once the pitting problem is overcome, the
component life will be determined on the basis of allowable wear in the system. In
mixed lubrication the wear modelling can be approached in terms of the extent of
asperity contact through the films. The wear due to the asperity contact is being
treated in terms of the available models of boundary lubrication. Limitations of
such modelling have already been discussed in the earlier chapters. The theoretical
modelling and experimental observations are considered below.
8.6. WEAR IN ROLLING ,'SLIDING CONTACTS 257
8.6.1 Modelling of contact in mixed lubrication
The contact model of rough surface on a flat due to Greenwood and Williamson
has been considered in section 1.3. Firstly consider dry contact. It is usual to solve
for the contact parameters in terms of standardised variables. The equations can
now be expressed in terms of the standardised variables as follows:
n = qA, Fo (h) (8.5)

A~ = rtl3rlA. oF, (h) (8.6)
L = --4E ]3~ (8.7)

3
where
= standardized separation given by d/o
= standard deviation of height distribution

co
Fn (h) = ~(s-h) ~ ~p(s)ds

h
~(s) = standardized surface height probability density function. For
Gaussian surface, it is given by ( l / - ~ ) e x p ( - s 2 / 2 )
The contacts made at the asperities can be plastic or elastic. When the extent of
deformation at an asperity is such that a maximum Hertzian pressure of 0.6H is
reached the contact is considered plastic. If 5p is the necessary deformation to reach
this Hertzian pressure, critical standardised deformation 5p/~ may be designated as
o~. All asperities whose deformation is equal to or above this value are considered
plastic contacts. This permits separation of elastic and plastic parts of contact as
follows:
oo
Number of plastic contacts, np = flA, ~q~(s)d(s) (8.8)

h+to c
Number of elastic contacts, n~ = n - n P (8.9)

oo
Real area of plastic contacts, A ---/I;13qA n a ff(S - h) q~(s)ds (8.10)

h+to c
Real area of elastic contacts, Are = A r - Arp (8.11)
where
o3r = (rt 0.6H/2E) 213
The above approach is approximate. In reality there will be a combination of

elastic, elastic-plastic, and plastic contacts with varying pressures at different
contacts.
These equations have been solved by Rajesh Kumar [38] with a computer
programme. An area contact between two steel surfaces with the following
parameters may be used to illustrate the results. The parameters represent the
equivalent roughness parameters against a smooth rigid disk.
Contact parameters:
c" = 0.21 ~tm, 13" -- 225 l.tm, 1] = 800/mm 2, E = 113.19 GPa, H = 6.68 GPa, A, =
0.1870 mm 2, L = 60 N.
Results:
ne = 132, np= 3, Are = 2.80 x 10-z mm 2, A n, = 0.18 x 10 -2 mm 2
The extent of plastic contact in this case is low. It will be of interest to calculate the
plasticity index as explained in chapter 1. This calculates to be 0.52. As this value
is less than 0.6 the contact may be treated as essentially elastic.
The contact problems can be treated in terms of equivalent roughness against a

smooth rigid plane only if contacts can be treated as random. As discussed earlier
in this chapter when there is conformity at the asperity level the problem has to be
treated differently.
Surface roughness in some cases is non-Gaussian. Such problems can be solved by

using the appropriate distribution function generated mathematically [39]. A
different approach based on the slope of the bearing area curve has been attempted
by Raj esh Kumar [38].
The above approaches assume dispersed contact and that stresses at one asperity do
not influence the distribution at the other asperities. Also these models are
concerned with normal stresses only. How the load distribution varies as the
asperity contacts shift does not appear to have been modelled. All that we have at
present is a statistical distribution model where the average pressure and the
8.6. WEAR IN ROLLING /SLIDING CONTACTS" 259
number of asperities in contact are calculated. Such models may be reasonable for
steady wear situations. But when dealing with scuffing and running-in, a few large
contacts are likely to play a critical role. No effective models are available to treat
the problem from this point of view.
The contact in mixed lubrication can be approached in terms of the film thickness.
The model considers that any asperity height greater than the film thickness will
make a contact. In effect the separation d is replaced by the film thickness h. From
the present theories of mixed lubrication in concentrated contacts the problem of
contact is more complex and has already been discussed in section 8.2. But a
conventional model based on asperity contact ignoring MEHD effects may be
considered as the worst possible situation in which maximum contact occurs.
The problem of wear in boundary lubrication has been approached in terms of

asperity contact as discussed in chapter 5. Stolarski [40] modelled wear by
separating elastic and plastic contact and assigning separate wear coefficients for
elastic and plastic contacts. The elastic contacts were considered to wear by
fatigue. For plastic contacts the wear coefficient was obtained theoretically. The
limitations of such models were also covered in chapter 5.
In mixed lubrication the load is shared by the EHD film and asperity contact. This
load sharing affects the film thickness because the EHD film does not carry the
total load. Johnson et al [41 ] proposed the following approximate equation to relate
the fluid pressure to the total pressure in such contacts.
Pf/Pt =(ho/~)6.3 (8.12)
where p f is the fluid pressure, pt is the total pressure, ho is the film thickness
based on smooth film theory and h is the film thickness to obtain film pressure pf.
This equation shows that the EHD film is stiff and the variation in thickness is very
small with variations in fluid pressure. The solution for asperity load sharing can be
obtained approximately by neglecting the variation in film thickness. The more
accurate procedure would account for the small variations in film thickness. The
next issue is the asperity contact model. The complete analysis involves
determination of elastic, plastic, and elastic-plastic contacts. Simplifications may
be attempted by considering the contact conditions as purely elastic or plastic.
These considerations resulted in several models of load sharing in EHD contacts
[42,43,44].
260 CHAPTER 8. FA HGUE AND WEAR I N M I X E D LUBRICATION
8.6.2 Experimental investigations

The major problem with all these models is the lack of experimental verification.
Ideally one has to obtain the asperity contact load and film thickness
simultaneously. At present this is not possible. Also in view of the present
theoretical developments the asperity deformation due to EHD pressure itself has
to be taken into account. The above models neglect this influence and the
implications are not known. As wear in mixed lubrication is of importance study of
wear with different A ratios would be useful. Such studies are limited. The work
reported by Sastry et al [45] is considered below.
Experiments were conducted in a disk machine with mineral oil at different A

ratios. The variations were achieved by changing the oil temperature and hence the
oil viscosity. EN 31 and low carbon steel disks formed the test pair. The two disks
rotated with 10% slip, which introduced relative sliding in the contact. The wear of
the mild steel disk was measured as a function of time. The tests were conducted
for a duration of eight hours. The relationship between cumulative wear and time
was regressed and the expected wear rate between 20 and 21 hours was estimated.
This wear rate was called the Wear Factor (WF) and defined as (wear in
milligrams/sliding distance (m) x load (N)). The relationship observed is given in
Fig.8.7.
The load considered here is the total load. Large influence of A ratio on WF can be
seen. The following equation describes the relationship.
WF = 8.38x10-4e -3.78A (8.13)
The correlation coefficient was found to be 0.8, which means the relationship is
approximate. Using the models of Tallian [42] for plastic and elastic-plastic
contacts the asperity loads were estimated. When WF was plotted in terms of
asperity loads, similar curves were obtained but the correlation coefficients were
poorer than above.
Empirical relationships of the above nature for different lubricants may be useful
for comparison. But any detailed modelling needs an understanding of the wear
mechanisms. Adhesive and/or fatigue wear may be involved in such contacts when
the lubricant has no chemical additive. Also any modelling of asperity contact in
mixed lubrication has a level of uncertainty. Karmakar et al [46] attempted to
model wear in rolling/sliding contact under boundary lubrication where the total
load is supported by the asperities. It was assumed that the tensile stresses at the
8.6. WEAR IN ROLLING SLIDING CONTACTS 261
asperities governed the fatigue. Modelling of wear needs information with regard
to depth of damage and the number of cycles needed for damage. This information
was obtained by a new approach by Karmakar et al [47]. In this experimental
method cyclic stressing was done on a fiat mild steel surface with an EN 31 steel
ball. After stressing to different levels the response of the prestressed material to
one pass sliding was observed. It was interesting to observe that material removal
by fatigue occurred in one pass sliding only when the prestressing exceeded a
given number of cycles. The information generated with regard to removal depth,
number of cycles, and the calculated tensile stresses formed the basis in modelling
wear in rolling/sliding contacts in [46]. The experimental wear rates observed in
rolling/sliding contact with an EN 31 and mild steel disk pair showed good
correlation with the developed fatigue wear model. Fatigue mechanism was
justified in this case on the basis of surface observations as discussed in the cited
references. The stress analysis done in this work was approximate and further
progress in theoretical modelling is possible. The model is capable of being
extended to mixed lubrication if asperity load sharing can be accurately predicted.
Depending on the operating conditions and materials the wear mechanism may be
governed by adhesion instead of fatigue. In such a case the modelling has to be
different.
2.6
2.5
--. 2.0
Z
E
"-1.S
o o
' o
la.
~ ~
3:
1.O-
o o
O.S-
o o
1 , i0 00. 1 . ~ 0 1 _ i
00 0.~. 0.8 1.2 1.6
x -------~-
Fig. 8.7. Influence of A on wear factor. (Reproduced from Ref. [46] by permission of
ASME).
262 CHAPTER 8. FA HGUE AND WEAR IN MIXED LUBRICATION
Wear modelling has to be based on the extent of asperity contact and the wear
mechanisms operating at the asperities. The issues involved are similar to those
already discussed in earlier chapters. One additional factor has to be considered in
mixed lubrication. These contacts are subject to high EHD pressures in the whole
contact area. Such pressures may additionally influence the bulk reaction in the
contact area when chemical additives are involved. As definitive wear models are
not available empirical relationships between wear rate and A may be developed
for different lubricants and additives as discussed above. Such relationships would
be useful to compare different formulations. These relationships coupled with
surface analysis can also provide a reasonable insight into the mechanisms
involved.
References
1. H. Christensen, A theory of mixed lubrication, Proc. lnstn. Mech. Engrs., Tribology

Group, 186 (1972) 421.
2. H. Christensen and K. Tonder, The hydrodynamic lubrication of rough bearing surfaces
of finite width, J. Lub. Tech., ASME, 93 (1971) 324.
3. N. Patir and H. S. Cheng, An average flow model for determining effects of three-
dimensional roughness on partial hydrodynamic lubrication, J. Lub. Tech., ASME,
100 (1978) 12.
4. C. C. Kweh, H. P. Evans, and R. W. Snidle, Micro-elastohydrodynamic lubrication of an
elliptical contact with transverse and three-dimensional sinusoidal roughness, J. Trib.,
ASME, 111 (1989) 577.
5. C. H. Venner and W. E. ten Napel, Surface roughness effects in an EHL line contact, J.
Trib., ASME, 114 (1992) 616.
6. L. Chang, A deterministic model for line contact partial elastohydrodynamic lubrication,
Trib. Int., 28 (1995) 75.
7. N. Fang, L. Chang, and G. L. Johnston, Some insights into micro-EHL pressures, J.
Trib., ASME, 121 (1999) 473.
8. X. Jiang, D. Y. Hua, H. S. Cheng, X. Ai, and Si C. Lee, A mixed elastohydrodynamic
lubrication model with asperity contact, J. Trib., ASME, 121 (1999) 481.
9. H. A. Spikes, Mixed lubrication-an overview, Lub. Sci., 9 (1997) 221.
10. R. Gohar and A. Cameron, The mapping of elastohydrodynamic contacts, ASLE Trans.,
10(1967)215.
11. G. J. Johnston, R. Wayte, and H. A. Spikes, The measurement and study of very thin
lubricant films in concentrated contacts, Trib. Trans., STLE, 34 (1991) 187.
12. H. A. Spikes, Film-forming additives-Direct and indirect ways to reduce friction, Lub.
Sci., 14 (2002) 147.
REFERENC'ES 263
13. M. Kaneta, T. Sakai, and H. Nishikawa, Effects of surface roughness on point contact
EHL, Trib. Trans., STLE, 36 (1993) 605.
14. M. J. Furey, Metallic contact and friction between sliding surfaces, ASLE Trans., 4
(1961) 1.
15. H. Czichos, W. Grimmer, and H. U. Mittmann, Rapid measuring techniques for
electrical contact resistance applied to lubricant additive studies, Wear, 40 (1976) 265.
16. A. Sethuramiah, V. P. Chawla and C. Prakash, A new approach to the study of the
antiwear behaviour of additives utilizing a metal contact circuit, Wear, 86 (1983) 219.
17. P. M. Lugt, R. W. M. Severt, J. FogelstrOm, and J. H. Tripp, Influence of surface
topography on friction, film breakdown and running-in in the mixed lubrication
regime, Proc. Instn Mech. Engrs., Part J, J. Trib., 215 (2001) 519.
18. R. C. Coy and A. Dyson, A rig to simulate the kinematics of the contact between cam
and finger follower, Lub. Eng., ASLE, 39 (3) (1983) 143.
19. M. Ram and A. Sethuramiah, Study of lubrication mechanism of two-stroke engine oils
using a disc machine, Trib. Int., 17 (1984) 73.
20. M. Ram Tyagi and A. Sethuramiah, Asperity level conformity in partial EHL, Part I: its
charecterization, Wear, 197 (1996) 89.
21. M. Ram Tyagi and A. Sethuramiah, Asperity level conformity in partial EHL, Part II:
its influence in lubrication, Wear 197 (1996) 98.
22. D. Dowson and Z. M. Jin, Microelastohydrodynamic lubrication of low-elastic-modulus
solids on rigid substrates, J. Phys. D: Appl. Phys. 25 (1992) A116.
23. D. Dowson, New joints for the millennium: wear control in total replacement hip joints,
Proc. Instn. Mech. Engrs., Part H, Journal of Engineering in Medicine, 215 (2001)
335.
24. G. Lundberg and A. Palmgren, Dynamic capacity of roller bearings, Acta Polytech.
Mech. Engng. Ser. 2, 96 (4) (1952).
25. W. Weibull, A statistical representation of fatigue failure in solids, Acta Polytech.
Mech. Engng., Ser. 1, 49 (9) 1949.
26. E. Ioannides and T. Harris, A new fatigue life model for rolling bearings, J. Trib.,
ASME, 107 (1985) 367.
27. T. E. Tallian, On competing failure modes in rolling contact, ASLE Trans., 10 (1967)
418.
28. J. C. Skurka, Elastohydrodynamic lubrication of roller bearings, J. Lub. Tech., ASME,
92 (1970) 281.
29. D. W. Dudley, Gear Wear, in M. B. Peterson and W. O. Winer (eds.), Wear Control
30. D. Scott, Study of the effect of lubricant on pitting failure of balls, Proc. Conf. on
Lubrication and Wear, Instn. Mech. Engrs, 1957, 463.
31. F. G. Rounds, Some effects of additives on rolling contact fatigue, ASLE Trans., 10
(1967) 243.
32. M. A. H. Howes, S. Bhattacharyya, F. C. Bock, and N. M. Parikh, Chemical effects of

lubrication in contact fatigue, Part I: The test program, Data, and metallurgical
observations, J. Lub. Tech., ASME, 98 (1976) 286.
33. S. Bhattacharyya, F. C. Bock, M. A. H. Howes, and N. M. Parikh, Chemical effects of
lubrication in fatigue, Part II: The statistical analysis, summary, and conclusions, J.
Lub. Tech., ASME, 98 (1976) 299.
34. W. E. Littmann, B. W. Kelley, W. J. Anderson, R. S. Fein, E. E. Klaus, L. B. Sibley,
and W. O. Winer, Chemical effects of lubrication in contact fatigue, Part III: Load-life
exponent, life scatter, and overall analysis, J. Lub. Tech., ASME, 98 (1976) 308.
35. Y. Wang, J. E. Fernandez, and D. G. Cuervo, Rolling-contact fatigue lives of steel AISI
52100 balls with eight mineral and synthetic lubricants, Wear, 196 (1996) 110.
36. H. Fan, L. M. Keer, W. Cheng, and H. S. Cheng, Competition between fatigue crack
propagation and wear, J. Trib., ASME, 115 (1993) 141.
37. D. N61ias, M. L. Dumont, F. Champiot, A. Vincent, D. Girodin, R. Foug6res, and L.
Flamand, Role of iclusions, surface roughness and operating conditions on rolling
contact fatigue, J. Trib., ASME, 121 (1999) 240.
38. Rajesh Kumar, Investigation into the running-in and steady state wear processes, PhD
thesis, liT Delhi, 2002.
39. C.A. Kotwal and B.Bhushan, Contact analysis of non-Gaussian surfaces for minimum
static and kinetic friction and wear, Trib. Trans., STLE, 39 (1996) 890.
40. T. A. Stolarski, A system for wear prediction in lubricated sliding contacts, Lub. Sci., 8
(1996) 315.
41. K. L. Johnson, J. A. Greenwood, and S. Y. Poon, A simple theory of asperity contact in
elastohydrodynamic lubrication, Wear, 19 (1972) 91.
42. T. E. Tallian, The theory of partial elastohydrodynamic contacts, Wear, 21 (1972) 49.
43. R. A. Thompson, W. Bocchi, A model for asperity load sharing in lubricated contacts,
ASLE Trans., 15 (1972) 67.
44. Y. Tsao and K. N. Tong, A model for mixed lubrication, ASLE Trans., 18 (1975) 90.
45. V. R. K. Sastry, D. V. Singh, and A. Sethuramiah, A study of wear mechanisms under
partial elastohydrodynamic conditions, Proc. Wear of Materials, ASME, 1987, 245.
46. S. Karmakar, U. R. K. Rao, and A. Sethuramiah, An approach towards fatigue wear
modelling, Wear, 198 (1996) 242.
47. S. Karmakar, U. R. K. Rao, and A. Sethuramiah, Characterisation of sliding wear in
dynamically stressed material, Wear, B 162 (1993), 1081.
Nomenclature
al, a2, a3 adjustment factors for reliability, material properties, and operating
conditions
NOMENCLATURE 265
An nominal area of contact

Ar real area of contact
Are real area of elastic contacts
A rp real area of plastic contacts
C radial clearence
C load at which the life of a rolling contact bearing is 106 cycles
E effective modulus of the two bodies
Ez, E2 elastic modulii of the two bodies, 1 and 2
Fn (h) = if(s-h) n q~(s)ds

h
h standardized separation given by d/G
h film thickness to obtain film pressure of p f
ho film thickness based on smooth film theory
H hardness
L load
La known number of cycles to failure at a given tooth load Wt~
Lb number of cycles to failure at any other tooth load Wtb
Lclo life with 90% probability under load C and is equal to 106 cycles
tna life for the selected probability with load P
Llo life with 90% probability obtained at the applied load P
Lso life with 50% probability
N revolutions per second
n total number of asperity contacts
n! exponent
n2 exponent
ne number of elastic contacts
no number of plastic contacts
P pressure
Pf fluid pressure
P equvalent applied load in rolling contact bearing
r journal radius
rmsz rms value of the composite profile with conformity
rnqs1,2 rms value of the composite profile based on random contact
RI, R2 radii of the bodies 1 and 2
SRR slide/roll ratio
U sliding velocity
U 1, U 2 surface velocities of the two disks
WF wear factor, dimensional
Wt~ gear tooth load with known life L~
Wtb gear tooth load at which life Lb is to be found
Y yield strength
Greek letters
(/, pressure coefficient of viscosity

radius of curvature of asperity
~p the necessary deformation to reach the Hertzian pressure of 0.6H
q absolute viscosity
q asperity density
standardized surface height probability density function. For a
Gaussian surface, it is given by (1/~/~n)exp(-sZ/2)
A Ratio of film thickness to composite roughness
V1 ~ V 2 Poisson ratios of the two bodies 1 and 2
O" standard deviation of height distribution
O'* composite roughness
O" 1 ~ 0" 2 rms roughness of the two surfaces 1 and 2
COc critical standardised deformation given by 5p/~
267
9. Wear in real systems and laboratory rigs
9.1 Introduction
The available knowledge regarding lubricated wear has been discussed in earlier
chapters. Several possibilities and limitations exist with regard to the available
knowledge. Linking this information to real situations is the purpose of this
chapter. As it is not possible to cover all components the chapter mainly considers
two different contacts. The selected examples are the piston ring-liner, and the cam
follower in engines.
The first section deals with the complexity of wear in real systems and the
importance of full-scale tests. This is followed by a consideration of available wear
models and practical aspects of running-in. The next section deals with the problem
of wear measurement. Newer approaches to measure wear more accurately will be
the main consideration of this section. The fourth section deals with strategies for
laboratory evaluations in a consolidated manner. This discussion will rely on the
material in previous chapters with due consideration of linkage to practice. Thus
the vision of the chapter is limited to the gains that can be achieved in the short-
term. The long-term needs in the area have already been covered in the earlier
chapters. Some repetition from previous chapters was necessary to provide a
consolidated presentation.
9.2 The complexity of wear in real systems
Tribological components are sub-assemblies in the overall system. Wear in the

components depends on the nature of the lubricant and the extent of boundary
contact. In this discussion wear refers to steady state wear. Steady state may be
considered as the stable wear after prolonged running. The system operates under
varied conditions, which translate into varying wear rates. In many cases there are
dust particles in the system despite filtration. These particles additionally
contribute to wear. In time the lubricant deteriorates and this is an additional
variable. Abrasive wear was not considered in detail in this book. It is very difficult
to assess the role of abrasive particles in real systems. During investigation of
industrial problems the author has found several cases of high wear due to abrasive
268 ('HAPTER 9. WEAR IN RJ{AL SYSTEMS AND LABORA TORY lUGS
particles. The components include turbocharger bearings in locomotive engines and

liners of tractor engines operating in dusty atmospheres. In some cases like engines
the environment of combustion products also affects wear.
The interest in practice is to minimise component wear and prolong the life. The
component wear behaviour may be illustrated by the curve given in Fig. 9.1. The
component first goes through a running-in process with high wear rate. This is
followed by steady low wear rate. As the wear depth increases beyond a point the
wear rate increases significantly and the component has to be replaced. This is
related to the increased probability of irregular contact as tolerance limits are
approached. Usually the condition also results in increased levels of vibration.
When failure rate is plotted in terms of running time a similar curve is obtained that
is commonly known as the bathtub curve. The wear levels that can be tolerated
depend on the operating conditions and precision needed. For example the
performance of a high speed rolling element bearing will be very sensitive to minor
levels of surface pitting. Such damage will be tolerable in a low speed operation.
Besides prolonged life there can be other reasons for controlling wear. For example
the wear in a ring-liner (RL) zone contributes to blow-by and hence emissions from
an engine. Wear can also modify contact stresses in comparison to the designed
values. The other problem of wear is its non-uniformity. In such cases the life is
governed by the zone with the maximum wear depth and not by the overall wear of
the component.
The interest in lubricated wear from the point of view of industry was broadly
categorised in the beginning of chapter 5. In all cases the important need is to
assess wear in real systems. However the wear rates can vary over a range as the
component is subject to coupling effects with the rest of the system. As an example
it is known that in an automobile the engine liner wear is substantially affected by
the number of start-and-stop cycles. One good way to simulate the real situation is
by testing the vehicle on test stands where desired operating conditions are
imposed. Such tests with prescribed driving cycles are used in the automobile
industry. These evaluations can be used to study fuel consumption, oil
consumption, and emissions from an engine. The tests can also be done on the
engine alone on a test bed. These bench tests will be called sub-assembly tests to
distinguish them from the smaller laboratory rig tests. Such tests under prescribed
conditions are in extensive use for evaluating lubricants as discussed in chapter 2.
Unlike other performance parameters wear evaluation can be done only with very
long duration tests of the order of 1000 h for liners. This is because the usual
gauging methods used to evaluate wear depth are not sensitive enough. This
9.2. THE COMPLEXITY OF WEAR IN REAL SYSTFA4S 269
translates into long test durations. Ring wear and cam follower wear can be
measured with shorter duration tests.
T
%
Ib i I
I I
I j
,
II
', llI
...... I , J
Running time (hours)
Fig. 9.1. Wear rate variation with running time: (I) Running-in zone, (II) Steady-state zone,
and (III) Run-out zone.
Two aspects are of interest with regard to wear tests. The first issue is whether sub-
assembly tests will be meaningful. As discussed the wear in real systems will be
affected by coupling effects and the environment of the sub-assembly. But in so far
as the lubricant-metal interactions go it may be argued that a separate engine test is
meaningful for evaluation. Additional variations that occur in real systems are
perturbations around this wear behaviour. So distinctions can be made in engine
tests with regard to lubricant formulations. The second aspect of importance is the
test duration. It is an expensive proposition to conduct very long duration tests.
Any effective techniques to measure small amounts of wear will reduce the testing
time considerably and much quicker direct evaluations are possible. These
developments will be considered later in the chapter.
9.3 Wear in real systems-importance and modelling

This section deals mainly with ring-liner and cam follower wear in engines. These
examples are selected because of their industrial importance. Also these are the
contacts for which a level of wear modelling is available. First the lubrication of
the contacts is considered, followed by a discussion of the importance of wear and
models. A brief consideration is then given to other components. The final part
considers the running-in in practice and may be treated as an extension of running-
in previously considered in chapter 5.
270 CHAPTER 9. WEAR IN REAL SYSTFAYS AND LABORA TORY RIGS
9.3.1 Lubrication of contacts
In IC engines lubricated wear is of importance in the ring-liner (RL) and cam

follower (CF) contacts. Wear in main bearings is also a concern when the film
thickness is low. One useful book available on engine and vehicle technology [ 1]
may be consulted for detailed understanding of the various components. A
specialised book [2] is also available dealing with tribology of the engine
components.
In the engine the piston reciprocates between the top dead centre (TDC) and the
bottom dead centre (BDC). Piston rings are used for sealing the flow of gases from
the combustion chamber. Normally two rings are used for this purpose that are
designated as compression rings. In addition an oil control ring is located below the
compression rings. This ring scrapes away the excess oil from the surface into the
crankcase. A thin oil film on which the rings travel achieves the lubrication
between the tings and liner.
The basic lubrication mechanism is hydrodynamic with varying pressures and

speeds acting between the ring and the liner as the piston travels. This results in
varying film thickness along the stroke. Near the dead centres the film thickness is
inadequate to separate the rings from the liner resulting in mixed and boundary
lubrication. Minimum thickness is observed few degrees beyond TDC due to
squeeze film effect. As the top ring position will be below TDC the maximum wear
occurs near the top ring reversal point (TRR). This is illustrated in Fig. 9.2. Though
there is reversal at the bottom dead centre also the wear is low due to lower
temperatures and loads acting in this zone. In the 4-stroke engine the film thickness
also varies between different strokes. Lowest thickness is obtained during the
power stroke. The theoretical estimation of film thickness is an active research area
for more than 50 years. It is only now that reasonable film thickness predictions are
available. The major considerations not only involve the varying gas pressures and
temperatures but also the fact that oil flow is restricted due to the influence of one
ring on the other. As oil flow continuity is to be maintained the net oil flow
decreases and it is now accepted that the lubrication occurs under starved
conditions. The major issues involved in this lubrication were reviewed by Ruddy
et al [3] in the early eighties and are still pertinent today. Recent attempts are being
made to estimate oil consumption, blow-by, and wear with comprehensive
simulation of the ring pack [4].
The cam and follower mechanism operates the engine valves. Various
configurations are possible [1]. These mechanisms include pivoted valve train
9.3. WEAR IN REAL SYSTEA4S-IMPORTANCE AND MODEl.LING 271
systems with finger followers as well as direct acting cam and tappet contacts. The
lift is achieved by the lobed part of the non-circular cam. The contact involves high
Hertzian stresses and the lubrication is achieved by an EHD mechanism. During a
cycle of operation the stresses and sliding speeds vary significantly. A good
approach to film thickness calculation is now available based on EHD analysis
incorporating squeeze film effects [5].
The main bearings are subject to fluctuating loads and are a special case of
hydrodynamic analysis. The mobility method [6] is used for this analysis. More
effective methods are also being researched now [7]. Advanced techniques are
becoming necessary due to higher loadings on the bearings.
9.3.2 Importance of wear
IC engines are becoming more compact with increasing specific loads. This
situation has led to severe operating conditions. Such operating conditions result in
higher temperatures and lower film thickness. Further reduction in film thickness is
due to the lower viscosity oils being used to decrease hydrodynamic losses. Lower
film thickness increases the probability of asperity contact leading to increased
wear and associated problems. At the same time increased reliability and longer
life is expected from modem engines. Stringent requirements on emissions and oil
consumption in the first ten thousand mile of operation also demand low wear rates
t.~ U
0 ,k~u!l J~::~uqA:~ u o u o ! ; ! $ O d 0
m
Fig. 9.2. Wear depth variation with position on cylinder liner: (TDC) Top dead centre,
(BDC) Bottom dead centre, and (TRR) Top ring reversal point.
272 CHAPTER 9. WEAR IN REAL SYSIEMS AND LABORA TORY RIGS
in the ring-liner zone as wear intensifies blow-by and oil consumption. Thus wear
control is of major importance in the present context. Three possible strategies are
possible to control wear.
Wear can be controlled by better chemical additives. As discussed in chapter 4 a

large number of new additives are being developed and tested. However as far as
engine oils go the main antiwear additive is still ZDTP. Varied structures of ZDTP
with different alkyl/aryl groups are possible and this is where modifications are
being affected. It is not clear whether any major breakthrough is possible in future.
There may be other possibilities when other additives in the formulation provide a
synergistic effect. Some new types of friction modifiers may be developed that act
synergistically with ZDTP in wear control.
The second approach is the use of wear resistant coatings/materials. Piston ring
coatings are already common to provide scuff resistance particularly during
running-in. It is expected that the modem technologies available for surface
coatings will at least provide partial solutions to the wear problem. Some attempts
in the past to use suspended graphite or molybdenum disulphide as solid lubricants
were not successful. This is probably due to the inability of these lubricants to
displace the adsorbed oil layers on the surface [8]. Any new ideas that allow the
particles to coat the surfaces by suitable pre-treatment will be useful. The earlier
interest in solid lubricants was friction reduction but any effective coating can
reduce wear as well. The possible use of ceramics in some components is another
possibility being explored as discussed in chapter 6.
The third possibility is to take advantage of the theoretical ideas available in mixed
lubrication. As discussed in chapter 8 modified roughness patterns can assist in
forming thicker films. Recently Priest and Taylor [9] discussed these possibilities.
They have given examples of surface modification for cam and follower contacts
as well as micro grooving to improve film thickness in main bearings. One problem
in applying the theoretical ideas is the significant modification that occurs in
roughness due to the running-in. In effect what is needed is the required pattem
beyond running-in. Hence, implementing ideas in practice is more involved.
Another aspect of roughness is the possibility of asperity level conformity as
discussed in section 8.3.3. Coy [10] has given an example of such conformity in a
cam and pivoted follower contact. There is no detailed information on this aspect
with different cam and follower mechanisms. Investigation of this aspect should be
useful. Such conformity can significantly reduce asperity contact for a given
thickness and is important in the present context.
9.3. WEAR IN REAL SYSTEMS-IMPOR TAN('E AND MODI:),LLING 273
The engineering solution of the problem is likely to be a combination of the above

approaches. The current film thicknesses observed in modem engines are already
low. As reported in [9] the minimum film thickness observed in the ring-liner zone
can be as low as 0.2 ktm for modem gasoline engines. The cam and follower
operates with minimum thickness of 0.1 jam. Even in main bearings that rely on
hydrodynamic lubrication the minimum thickness is around 1.0 I.tm.
9.3.3 Modelling of wear
The available methods to calculate film thickness provide a good estimation of film
thickness. Wear modelling on the other hand lags behind and is empirical. As
discussed in earlier chapters wear modelling cannot be done on the basis of
fundamental considerations even in controlled laboratory experiments. The present
approach to modelling is considered below.
The initial approach to wear in ring-liner contact was based on a limiting film
thickness [ 11 ]. In this modelling wear was considered to occur only in those zones
where film thickness was below this level. Wear in these zones was modelled on
the basis of Archard's equation assuming a constant wear coefficient. The present
wear model as discussed by Priest and Taylor [9] is based on a modification of the
above idea. In this model a constant boundary wear factor kb is assumed for A ratio
< 0.5. Wear volume for the boundary zone is expressed as
V=kbW/ (9.1)
where V is the wear volume, kb is the wear factor, W is the load, and l is the sliding
distance
It may be noted that the wear coefficient has the units of specific wear rate. It was
obtained from laboratory tests the details of which were not given. The terms
'boundary wear' and 'boundary lubrication' are used in a broader sense throughout
this chapter and in reality refer to chemical wear in most cases.
For the mixed lubrication zone the wear coefficient is assumed to be proportional
to A. The linear relation is obtained by taking kb is zero for A - 4.0 and the
maximum value at A = 0.5. The wear factor at any A value kb(A) is obtained from
this relation. Wear volume and hence the wear depth can be estimated from this
equation at different liner locations. The ring wear can also be obtained by
integrating the changing wear volume in each stroke. It is claimed that this
approach reasonably predicts the experimentally observed wear patterns.
274 CHAPTER 9. WEAR IN REAL SYSTEMS AND LABORA TORY RIGS
There are two problems with such modelling. The first problem is the value of kb in
boundary lubrication. There is no doubt that kb varies with temperature. Both
fundamental considerations and empirical modelling in a reciprocating rig reported
in section 5.4.1 show that the wear will be influenced by temperature. Over a
temperature range of 50 ~ to 150~ the wear rates for engine oil decreased by about
three times. The material combination was an EN 31 ball sliding against an EN 31
flat. As fundamental modelling is not possible the possibility of empirical
modelling may be considered. This involves the recognition that 1%is a function of
temperature and is proportional to Tea where 'a' is a negative exponent and Tr is the
contact temperature in ~ The value of 'a' has to be derived empirically from the
experimental data. It may be considered that the temperature influence is the same
for both kb and kb(A). Implicit assumption here is that wear coefficient is
independent of load and speed, which may be acceptable. Surface temperature rise
in the contact due to friction is not known and its influence also needs a theoretical
analysis. Here again the relations may be tried on the basis of liner temperature
only as a first approximation. It may be also noted that the reaction film is a
'mixed' film since the film on the moving rings with varying temperature is
changing and interacting with the liner film. This is unlike a laboratory experiment
where film behaviour can be characterised at one operating temperature. Under
such a situation empirical consideration of the temperature influence appears to be
the only possibility.
The kb value should be derivable from the experimentally observed wear at the
TDC. The values of kb observed for different formulations will be very useful
reference data for laboratory rig tests.
Another problem with the available model is the assumption that wear reduces
linearly with A. The experimental work reported in section 8.6.2 showed
significant non-linearity in wear under partial EHD conditions. It appears this
assumption has to be experimentally verified for the ring-liner contact that involves
hydrodynamic lubrication. Probably evaluations in a motored engine using only
mineral oil may be useful for this purpose. As film thickness can be calculated
along the stroke the variations in liner wear can be related to the A ratio. The nature
of the relationship can then be used with confidence.
Cam and follower contacts operate in the EHD regime and the cyclic variations in
film thickness can be reasonably predicted. In this case again wear is modelled by
the same procedure as in the case of ring-liner contact. Reasonable prediction of
the wear pattern is observed in this case also as reported by Bell and Colgan [ 12].
They used a boundary wear factor value of l x l 0 -17 m3/mN based on earlier
9.3. WEAR IN REAL SYSTEMS-IMPORTANCE AND MODELLING 275
laboratory tests. The problems involved with this modelling are similar to those
discussed for ring-liner contacts. It may be possible in this case to model wear as a
function of A with a disc machine with appropriate material combination. The
appropriate method has to be researched. This will enable firming up at least one
aspect of the wear model. In so far as the influence of temperature on kb goes, it
may have to be accommodated empirically.
Some engine tests are available as given in Table 2.2 of chapter 2 to specify wear
limits. Such specifications already exercise a degree of control on wear. Some of
these tests have the potential of being adapted for comparative wear studies.
The above discussion shows that fundamental research should now focus on real
systems. The gaps involved are major and the possible strategies were discussed in
chapter 5. In the short-term, improvement in the existing methods should be sought
as discussed above. It is also necessary to explore modem techniques like FFM.
For example scratch tests on the worn surfaces may provide an indication of the
wear resistance of the films. Such information will be valuable in realistic wear
modelling. Successful use of tribology for industrial problems is of importance for
its widespread acceptance and should be actively pursued.
9.3.4 Wear in other components
The author has so far considered two typical contacts with regard to wear
modelling. The contacts considered are industrially important and there is work
reported on wear modelling. In many other systems like gears, hydraulic systems,
roll wear in cold and hot rolling the modelling is not even at the level considered in
the above contacts. It is expected that the existing models for ring-liner contact can
be extended to reciprocating contacts leading to a reasonable prediction capability.
In many of these contacts the problem of wear is known and selection of lubricants
is based on experience. For example a synthetic lubricant may perform better than
emulsions in high speed cold rolling and will be the preferred lubricant subject to
cost considerations. In high-speed gears there is more interest in preventing
scuffing. This is evidenced by large amount of effort to model scuffing. More
recently there have been models published with regard to gear wear [13]. These
models take into account the changing contact stresses due to wear and integrate
the wear in each cycle incrementally. However the wear prediction is simply based
on a constant wear coefficient. The value of wear coefficient and its variation is the
crux of the problem in wear modelling as discussed above. Some other wear
equations relate wear life to operating conditions empirically. One example for life
of cutting tool [14] may be cited. The relationship is expressed as
276 CHAPTER 9. WEAR IN I?JZ,AL SYSI't,~S AND LABORA TORY RIGS
T t v lna t l/robl/p = constant (9.2)
where T t is tool life, t is feed rate, v is the cutting speed, and b is depth of cut and n,
m, p are the exponents that satisfy the equation.
Such equations are general in nature and provide some prediction capabilities as a
function of operating conditions for a given system. It may be noted there is no
explicit parameter for the lubricant. Any variations due to lubricant cannot be
predicted from this equation and can only be assessed from experiments.
Equations to predict wear quantitatively are not well developed. But the efforts to
reduce wear with different strategies will lead to lower wear rates in future. This is
of importance irrespective of the capability to model wear precisely. The parallel
development needed is the ability to classify wear in laboratory rigs in the same
order as observed in the real contact.
9.3.5 Running-in in practice

Running-in of the components has to be done in the real system and its
effectiveness has to be characterised. The major research reported in this area is
with regard to ring-liner contacts. The nature of honed surface roughness for a liner
is given in Fig 9.3. The plateaux are smoother zones obtained in the two-stage
honing process that first involves rough honing with a cross hatched pattern. This is
followed by fine grinding of the rough surface resulting in smoother plateaux.
Different levels of finer roughness at the plateaux can be obtained by varying the
manufacturing process. Smoother plateaux were considered to be closer to the run-
in surfaces and hence need lower time for running-in. On the other hand smoother
surface plateaux are more prone to scuffing during increased loading [15]. Some
laboratory studies recently reported also seem to support the idea [16]. Thus some
optimum roughness is to be sought for the plateaux. Several approaches to
characterise the initial surface roughness have been proposed. Completion of the
running-in has been characterised by the expected changes in these parameters
[17]. The important issue is to know what is the best procedure to adopt for the
quickest running-in of a given engine. There are no effective answers to this
question and the available procedures are empirical as stated in chapter 5. From the
point of view of practice another aspect to be investigated is the influence of
running-in and initial roughness on steady state wear. This has implications to the
liner life. Laboratory studies reported in chapter 5 indicated that initial roughness
influences steady state wear though final roughness may be same during the steady
9.3. WEAR IN REAL SYSTFA4S-IMPOR TANCE AND MODELLING 277
state. The material combination in this case involved EN 31 steels. This may be
related to the mechanical properties of the films that develop slowly with time of
running. Thus steady state wear is also a parameter to be taken into account in
selecting initial roughness and the running-in procedure.
One practical approach is the use of surface coatings on rings and in some cases on
the liner. Chromium and molybdenum coatings are widely used for piston rings.
Gas nitriding is also being used for surface modification. Even with such coatings
the effective running-in procedure has to be empirically developed though the time
needed may be shorter.
Scale" horizontal, ! div =. 250 ~m; vertical, 1 div = 5 ~m ..

.-..- ~-.t-r---4----].l~,~,t ,,...,.~,..- ...t.,..-,. 4,. . . . ~- - - -I-- -4---...~.--- 8.- - - g,-- J - - - -~-- -.a
.... i.. L._ i--- i___ ;---i----

Ra -- 0.77 [am, Rzd = 5.83 [.tm, Rt-- 8.72 [.tm, Lt = 4.8 om, Lc = 0.8 ~m, La = 4.0 ~m
Fig. 9.3 An example of honed surface roughness profile (P) Plateaued zone (V) Valleys.
In the past fine abrasives were used to assist in the quicker running-in by removing
asperities faster. Another approach tried was the soluble chromium salts that
eventually formed oxides that acted as fine abrasive. Such procedures have the
problem of using a separate oil during running-in as well as the complication of the
abrasive particles that remain in the system. While some commercial oils are
available for the purpose the extent of their application is not known. One
additional possibility is to develop protective oxide films on the surface during
final stages of manufacture. The author has not come across any research in this
direction.
The associated problem is to assess surfaces and coatings in the laboratory. Large
amount of work has been reported in this area with a wide variety of test
configurations [18,19]. They have similar problems in interpretation as have been
discussed for antiwear and EP additives in chapter 7. Here again mapping is
necessary over a wider range of operating conditions.
Running-in is also of importance for other components like gears and bearings.
While some guidelines are reported the procedures are at best semi-empirical.
278 CHAPTER 9. WEAR IN REAL SYSTEMS AND LABOtL4 TORY RIGS
Running-in can change friction coefficient and contact temperature and these
parameters can be used as indirect indicators of the running-in process. Blau [20]
has given an elaborate consideration of these possibilities that are not considered
here.
The problem of load carrying capacity has already been considered in earlier
chapters. Current laboratory tests and their limitations were already discussed. The
guidelines for practice mainly relate to contact temperatures and are mainly
applicable to gears. For many other systems experience seems to be the only
answer despite some guidance from laboratory EP tests.
9.4 The problem of wear measurement
Any assessment of wear by conventional techniques like gauging of liner wear

have limitations. There can easily be an error of + 1.0 ~m in such measurements.
Hence for assessment of wear several hundred hours of testing is needed as wear in
many engines will be of the order of 1.0 ~m per hundred hours in the zone of
highest wear. Any comparisons with laboratory wear tests become difficult, as it is
expensive to run such long duration tests. This is also true with many other systems
like gears in which the wear rates are low. Accurate wear measurements in short
duration tests will help in quicker evaluation of lubricants. It will assist in faster
comparison of wear between laboratory and large-scale tests. There is growing
interest in such evaluations and the three main reported approaches are considered
below. The first two approaches are related to the liner wear measurement and are
discussed under this category. The third approach based on roughness trace
analysis is separately considered as it is being used for measuring very low wear.
All evaluations depend on the comparison of roughness parameters but differ in the
methodology and scale of observation.
9.4.1 Removable insert method
Henein et aI [21 ] have reported an interesting new approach to wear measurement.

They have inserted a wear probe in the liner, which was removable. The probe was
machined carefully so that it was flush with the liner. The probe was so located that
the top ring was centred on it at TDC position. The top quarter of the probe does
not contact the ring and was used as a reference surface. The probe was
manufactured so that its position in the cylinder liner was reproduced accurately
whenever it was reinstalled. An area of 2.2 x 3.5 mm was selected for scanning.
The 3-D and 2-D roughness of the worn surfaces was obtained by a laser stylus
9. 4. THE PROBLF_,M OF WEAR M F ~ S U R E M E N T 279
meter. The system with the associated software was also equipped to measure the
total volume above a given reference. The difference in volume after due
correction for any change in reference level gave the wear volume of the scanned
portion. The nominal wear depth was obtained on the basis of the scanned area.
The method was used to study wear in five unequal steps with a total time of 3 h.
The important finding was that major wear occurred in the first hour for the
selected operating conditions of the engine. The average wear rate in the first one
hour was 12 times that during second and third hours. Several roughness
parameters obtained before and after wear were also compared.
The above method should be applicable for long duration tests as well depending
on the selected reference level. The method is surely of value but the limitation is
the availability of the techniques used. As the authors' pointed out this can be a
valuable tool to quickly assess the running-in process for different conditions and
to ensure that the initial running-in is effectively completed. In principle the
method should be applicable for wear evaluation in any reciprocating machine.
9.4.2 Bearing area method
The liner is usually honed and the valleys may typically range from 4-10 pm. Such
honing is considered advantageous for lubricant retention. The roughness trace of a
typical honed surface was illustrated in Fig. 9.3.
Since the bearing area curve was not considered earlier it is explained here. In a 2-
D profile trace several parallel lines may be drawn. Each line cuts through solid
and empty space. The percentage of the length that passes through solid at each
depth can be plotted as a function of depth. This is normally called a bearing area
curve. This is illustrated in Fig. 9.4. This curve will now be referred to as a beating
length curve though it is conventionally called bearing area curve. The percentage
of bearing length increases with depth.
The engine wear in short duration tests of the order of 100 to 200 h is low and will
occur within the honed roughness. The idea that has been used for estimation is to
have an internal reference that will be the same for worn and the unworn liners.
The approach used by Rajesh Kumar et al [22] considered the 90% point of the
bearing length curves to be the same before and after wear. The bearing length
curve aider wear was superimposed on the original curve with common 90% point.
The area between the two curves provided the worn area over the assessment
length. At each measurement zone the volume was obtained by considering the
area to be representative over a circumferential distance corresponding to one
280 CHAPTER 9. WEAR IN R t ~ L SYSI'EMS AND LABORA TORY RIGS
degree. The wear depth was obtained on the basis of difference at 0% bearing
length with some modification. The procedure to find the worn area is illustrated in
Fig. 9.5.
The above procedure was used to assess wear in 200 h endurance tests. The tests
were conducted on a two-cylinder, four-stroke, air-cooled tractor engine of 34 BHP
(25 kW). Effective comparison between liners subjected to two different running-in
procedures was possible by this procedure. The radial wear depth ranged from 0.7
to about 3.4 ~m depending on the distance from TDC. The air-cooled engine did
not show the expected uniform decrease along the stroke length. The possibilities
and limitations of the method were discussed in the cited reference. The procedure
is based on nominal relocation at different distances and 2-D profiles. Estimation
based on 3-D profilometry and better relocation will improve the utility of the
technique.
---I 0
.. [
50 100"1.
Assessment l e n g t h , Lo m m Bearing length as
9/, of assessment
length La
(a) (b)
Fig. 9.4. Schematic diagram of roughness profile with corresponding bearing length curve
(a) Roughness profile (b) Bearing length curve.
9.4.3 Precision techniques for wear measurement
As discussed above any wear evaluation needs a reference. For honed surfaces with
deep valleys advantage can be taken of the fact that the deeper part of the surface
can be used as a reference. For surfaces without honing, and low roughness wear
can be measured with reference to some deeper unchanged part like a valley. Such
measurements will be meaningful for very small depths only. One technique used
9. 4. THE PROBLFA4 OF WEAR MEASURFA/IENT 281
(a)
.E
s
d)
s
'Z
90% Point,
New
(b)
0 50 100% E
Bearing length as % of
assessment length La
EL
QJ 90% Point-"
s
(c)
Wear
depth I 9
50 100%
assessment length L a---II~
123 ~ ~i-- =-'- ~-"
*!
I "i-I l
i- il~J
u 30 70 90%
assessment length L a
Fig. 9.5. Methodology for wear estimation using bearing length curve: (a) and (b) Bearing
length curves of new and used liners, and(c) Wear estimation after matching 90% point.
involved AFM techniques [23]. The advantage of this technique is the generation
of 3-D topography before and after wear. Matching can be done on the basis of
relatively deeper unaffected portions. Point to point height variations can be
compared. The wear volume can also be obtained by subtracting bearing volumes
obtained before and after test. Another approach reported [24] involved accurate
relocation of the surface to the extent possible by specially developed methods
used for precision machine tools. The roughness measurements were done on a
normal stylus instrument. The relocation was further improved by software
techniques. Some techniques based on image processing have also been reported
[25]. All these methods lead to good assessment of the small wear that occurs on
the surface as well as its distribution. Tests are also possible by using a fine scratch
as a reference. The change in scratch depth itself can provide a rough indication of
relative wear depth. A recent paper used this technique to assess seal wear [26].
The authors were able to put in evidence the variations in wear depth due to
thermoelastic deformations in a carbon graphite seal. It is expected that wear
measurement techniques will advance significantly in the near future. The problem
that has to be resolved is the scale of observation that is meaningful for specific
applications.
9.5 Strategies for laboratory evaluation
Sub-assembly tests are the best way to evaluate lubricated wear. In principle
comparison between formulations as well as adequacy for new systems is possible
by these tests. The techniques being developed now for measurement as discussed
above will facilitate such evaluations. But even with reduction in test time using
such tests on a routine basis to evaluate formulations is not a practical solution.
This is because reasonable test duration is necessary for comparisons. Strategies
are obviously necessary to make laboratory rig evaluations more meaningful and
the present section addresses these problems. The emphasis is on ideas that provide
improvement in the short-term. Some aspects considered in earlier chapters are
repeated to provide consolidated treatment. The issues related to rig selection and
operating conditions, lubrication regime, and wear mapping are first considered.
The final part considers the linkage of laboratory tests to practice. In this section
'lubricant' and.'formulation' are used interchangeably. Antiwear additives are only
referred to as additives unless otherwise qualified.
9.5.1 Selection of test rig and operating conditions
Several rigs are available for testing. The rig selection should preferably be based
on the nature of contact. Thus it is good to use a reciprocating tester with area
contact to study ring-liner wear. For gears as well as cam and followers disk
machines are the obvious choice. The test material combination should be the same
as the real system. These considerations are elementary but often ignored in
lubricant evaluation.
Any test rig that may or may not simulate the stresses and materials will still be
capable of ranking lubricants. The problem with such ranking is that influence of
contact stresses and materials are changed. Correlations between different rigs over
a wide range of operating conditions is necessary for this purpose which is not
available. So the best strategy is to at least simulate what can be simulated.
9.5. STRATEGIES FOR LABORA TORY EVALUATION 283
As has been emphasised in chapter 7 any ranking based on one test condition is
totally inadequate. It is essential to develop wear maps on the lines suggested in
chapter 5. The operating range should preferably cover the conditions in practice.
Some operating conditions like velocity cannot be easily simulated. For example
the average sliding speed in a real ring-liner contact can easily exceed several m/s
while many reciprocating testers operate at an average velocity of less than 1 m/s.
In this specific case low speeds are meaningful with regard to wear near the TDC
where speeds are low. The temperature range that has a strong influence on wear
can be easily simulated. The required stress ranges can also be normally simulated.
All area contacts present problems with regard to aligmnent of the test pieces.
Careful design is necessary to ensure alignment. Even with good design it may be
necessary to do pre-running to ensure alignment. With concentrated contact like a
ball-on-disk the problem can be avoided. As the ball wears the contact changes to
area contact and the contact can simulate area contact. The area, and hence the
apparent pressure, keeps varying as the test progresses. The widely used 4-Ball
wear tester is an example of varying area contact. Material simulation is possible
with a hemispherical contact sliding against a disk. The 4-Ball machine also can
accommodate different materials and configurations though the changes are more
difficult.
New and innovative ideas are possible for laboratory rigs. The tests need not be
confined to known standard test rigs. The advantage with standard rigs is the
known levels of repeatability and the levels of precision reached in manufacture.
Also these rigs are widely available and comparisons between laboratories are
easier. One possible innovation is the careful modification of the existing test
procedure to suit requirements. A modified test procedure say, in a 4-Ball EP
tester, may correlate with an FZG rig in EP performance. The difference in EP
action between these rigs was discussed in section 7.4.3. Such considerations can
be used to develop correlations between figs. The idea here is not to emphasise the
correlations between rigs per se. The important possibility of wide variation in
conditions should be better exploited in the existing rigs.
The need for new rigs is particularly relevant when the operating conditions in real
system cannot be simulated in the existing rigs. Some examples discussed in
chapter 5 were with regard to a hypoid rear-axle and high-speed cold rolling. In
many such cases involving high severity, reliability is more important though better
lubrication is expected to reduce the wear.
284 CHAPTER 9. WEAR IN REAL SYSIEA/IS AND LABORA TORY RIGS"
9.5.2 Lubrication regime for testing
The regime to be used should be boundary lubrication. Any partial hydrodynamic

effects complicate the interpretation of wear data. In wear tests particularly at high
speeds some hydrodynamic effects can occur. Such effects can be indirectly
assessed on the basis of friction coefficient as well as variation in cumulative wear
with time as discussed in chapter 7. Real components operate in rrdxed and
boundary conditions depending on the nature of contact and operating conditions.
But in so far as distinction between the additives is concemed, the wear rates
should depend on the boundary contact zone. This is valid provided the mixed
lubrication conditions are similar for the different additives. This assumption is
reasonable if tests in the sub-system are conducted under the same conditions. One
complication that can arise even in such testing is the influence of the additive on
roughness. Suppose one additive has a significant smoothening effect on the
surface. In such a case the mixed lubrication effects change and this can influence
comparison of wear rates. These effects may or may not be significant and need to
be assessed on the basis of the final roughness of worn surfaces. Another
complication is related to asperity stresses. In mixed lubrication the load is
distributed between asperity contacts and the fluid film and the contacts will be
essentially elastic. If the total load is supported by boundary contact some asperity
contacts can reach plastic deformation. The effect will not be significant if the
contacts are essentially elastic in both cases. Usually in laboratory wear testing
where area contacts are involved the contacts are expected to be mainly elastic in
long duration tests.
The above remarks apply to machines that work with area contact. In the case of
disk machines mixed lubrication will be common because of significant EHD
effects. One way of assuring boundary regime is to work with A ratio less than 0.5.
Such a requirement will restrict the choice of operating conditions. This problem
can be overcome if asperity contact levels in mixed lubrication can be modelled.
This is difficult as discussed in chapter 8. One possible way is to develop empirical
relationships between A and wear rate as discussed in section 8.6.2. Here again the
methodology has to be researched. In cases where there is a level of asperity
conformity, composite roughness has to be evaluated differently as discussed in
section 8.3.3.
9.5.3 Development of wear maps

9.5. STRATEGIES FOR LABORA TORY Et'AL UA lION 285
The first step is to define wear rate. Theoretically the run-in process continues over
a long period of time and steady wear rate is what is obtained after infinite time.
Modelling wear rate on this basis was discussed in section 5.4.1. From the
available computer programmes steady wear rate can be obtained. In practice the
change in wear rate becomes negligible beyond an adequate running time. The
practical situation with regard to 4-Ball evaluation was discussed in chapter 7. The
suggested approach was to evaluate wear over a relatively long duration with step-
wise determination of cumulative wear volume. The information can be used to
obtain wear rate by regression analysis. Though this is by definition not steady
state wear rate, it is expected this value is meaningful for comparison purposes.
Quantitative change between any two lubricants will be different for the two cases.
But it is reasonable to assume that ranking in relatively long duration tests is
adequate from a practical point of view. The actual test duration needed depends on
the operating conditions. The best approach can be to decide the running time on
the basis of selected total sliding distance. As discussed in chapter 7 the author
considers the present classification of wear with short duration of 30 minutes or so
to be inadequate and the testing strategy must change. It may be appreciated that
'long duration' is only qualitatively expressed and has to be based on exploratory
testing work. A quantitative approach may be possible on the basis of rate of
change in wear rate with time.
The necessity for wear maps has been discussed in chapter 5. Such mapping as
shown in Fig. 5.7 has clear relevance to practice. The question with regard to
practice is the extent of mapping that is necessary. For this purpose it is necessary
to appreciate the time requirements. Ideally in a given tester the variables to be
considered are the load, temperature, sliding speed, and surface roughness. For
lubricant evaluation it is possible to fix the initial roughness of the test pieces,
which leaves only three variables. If the variables are tested at three levels each, a
full factorial design will involve 27 test combinations. Assuming each test is
repeated at least once the total number of experiments will be 54. With each test
run lasting at least for three hours the total test time for evaluation of one lubricant
will be about 200 h. The testing can be reduced for example by selecting one-third
fractional design that will involve a total testing time of nearly 70 h. Such partial
designs will be less accurate as compared to full designs. Once the data is available
useful empirical relations can be developed as discussed in chapter 5.
The wear maps discussed above are illustrative only. The variables may be tested at
more than three levels in which case the mapping will take more time. In some
cases different material combinations and roughness effects may be important
variables needed for study. It may be appreciated that specific guidelines cannot be
evolved for each case and the investigator has to make his choices. The author
considers that the first step in screening can be based on a study of the influence of
temperature only. In this approach the wear rate is studied with a fixed load and
speed at different temperatures. If possible the load and speed conditions should
correspond to practice. The temperature range selected must include the operating
temperatures in the real system. Even in this case development of a relationship can
mean at least 20-25 h of experimentation for each additive assuming five different
temperatures are selected and some repeat tests are done. The total contact
temperature that includes the surface temperature rise should be the governing
temperature for evaluation. The surface temperature rise calculation may be based
on the geometric contact area as discussed in chapter 1. While asperity
temperatures can also be estimated as discussed in chapter 3 they are less certain as
asperity contact sizes are not known. In effect the overall contact temperature may
be treated as a thermal parameter that reasonably describes the overall influence on
wear. The speed and load effects are not known and it cannot be assumed that their
influence will be related to the change in contact temperature alone. This initial
screening will be helpful in identifying the candidate additives. The selected
additives may then be studied with more extensive mapping.
The wear maps have the advantage of providing information under different
conditions. As an example a formulation for gear oils may operate under different
conditions of load and speed. The ideal candidate will be the formulation with
minimum wear but also least possible variation over the range of operating
conditions. Assessment of formulations from this point of view is not possible
without mapping. In some cases the formulators already know from experience
several aspects of the performance in real systems. With such experience
development of maps can be restricted to the required zones only where experience
is not available. The wear map approach quantifies performance unlike experience
that is generally qualitative.
9.5.4 Linkage to practice
The discussion so far has been related to the best approach to compare
formulations in a given test machine. It is important to know how they are related
to the wear rates in the large-scale tests. Here the comparison is discussed only in
terms of sub-assembly tests for reasons already stated. The wear in a sub-assembly
depends on the extent of asperity contact through mixed films and the applicable
wear coefficient in the boundary contact. Any prediction of A ratio needs
estimation of film thickness as a first step. As discussed earlier the film thickness
can be effectively estimated for complex geometries involving ring-liner and cam
9.5. STRATEGIES FOR LABORA TORY EVAL UA HON 287
follower in engines. These developments are capable of being extended to many

industrial situations involving reciprocating area contacts like compressors. Also
for EHD contacts film thickness can be estimated with EHD theory. Thus as a
starting point it is assumed that the information regarding A values is available.
The two areas of uncertainty are now the influence of A on wear rate and the
applicable boundary wear coefficient. The wear coefficient may be expressed in
units of specific wear rate or as a non-dimensional coefficient as convenient. The
problems involved are already discussed in section 9.3.3. The role of laboratory
evaluations will now be discussed in terms of bridging the gaps. The extent to
which laboratory evaluations can be used in choosing alternate formulations will be
considered next.
The ideal approach is first considered. To obtain information on laboratory

machines wear mapping has to be done under boundary conditions over a range of
operating conditions that will include the operating conditions of the sub-assembly.
These tests may be conducted with area contact or line contact as applicable. The
empirical relations developed will provide the required boundary wear coefficients
at different operating conditions. Such information can be generated for various
additive formulations. The same formulations may then be evaluated with known
variations in A ratios. Varying A ratios in EHD can be generated in disk machines.
But when partial hydrodynamic effects are involved the influence of A ratio can be
different and suitable rigs may have to be devised. For example a parallel thrust
pad with varying levels of hydrostatic lubrication can be used for such a purpose.
A comparison of the wear maps in the two cases should be capable of providing
relationships combining mixed lubrication as well as boundary effects. Such
development is not straightforward and needs substantial research. Also
comparison has to be done with the wear information generated in the sub-
assembly to substantiate the models. The effort needed can only be justified in
terms of the cost~enefit ratio. The recent interest in engine wear may justify such
an effort. The obvious advantage of the suggested approach is the generation of
information to directly predict wear rate in the real operating systems.
Partial implementation of the above ideas may be considered as the next best
alternative. The development of wear maps under boundary conditions alone can
be very useful in ranking formulations. Mixed lubrication effects can decrease the
asperity contact as well as the stresses at the asperities. Both these effects can only
reduce the boundary wear. Though real wear cannot be predicted in terms of these
effects the relative wear is expected to be governed by the basic boundary wear
behaviour of additives. It is hence reasonable to rank formulations on the basis of
wear behaviour in boundary contact conditions. Even here some problems have to
288 CHAPTER 9. WEAR IN REAL SYSTEA4S AND LABORA TORY RIGS
be envisaged regarding development of wear maps and the corresponding wear

equations. While contact stress and temperature can be simulated there can be
difficulties in simulating speed effects. Here speed effect refers to the influence of
speed on boundary lubrication. Any extrapolation of the empirical equation to high
speed conditions needs a judgment based on observed wear behaviour as a function
of speed. For example if speed effects on wear coefficient are small it is reasonable
to ignore this effect. Other problems that can arise include the inability to simulate
contact conditions or materials. These issues were already discussed and are not
considered here. Yet another practical problem can be the inability to avoid mixed
lubrication under certain operating conditions. The relationships in such cases have
to be restricted to the cases involving boundary lubrication. Despite these
limitations the information generated will provide the ability to rank different
formulations with regard to wear under different operating conditions.
The relative rating done in the laboratory has to be correlated with practice. This
requires determination of wear rate in the sub-assembly test. The duration of such
test depends on the accuracy with which wear can be assessed. As discussed earlier
some of the newer approaches will be helpful in this regard. Also any comparison
on the basis of steady state wear requires relatively longer durations. The sub-
assembly tests can be conducted under several coriditions. In the first instance the
obvious choice is to run the tests under the mos.t, severe conditions expected in
operation. The interest will be on the zone of maximum wear. If wear data for
different formulations is available then comparison0f the data is possible with the
laboratory tests. Such a comparison will provide adequate confidence for
laboratory evaluation. Identical wear rates are not to be expected as in the sub-
assembly tests there can be mixed lubrication. For comparison all tests in sub-
assembly have to be done with same conditions so that mixed lubrication effects
are the same for all the additives. Though it is desirable to do sub-assembly tests
under different operating conditions comparisons on the basis of the most severe
conditions may be adequate.
The above approaches are no doubt time consuming. But as has been stated earlier
the extent of wear mapping to be done has to be decided by the investigator and
some simplifications are possible. It needs to be emphasised that wear mapping is
not for the purpose of comparison alone. The maps amount to engineering
modelling of boundary lubrication. They provide realistic information on all
aspects of boundary lubrication that include influence of operating parameters,
possible wear transitions, and running-in behaviour. When transitions occur in
wear behaviour, empirical equations may have to be obtained for different zones.
9.5. STRATEGIES FOR LABORA TORY EVALUATION 289
The long-range research that can be conducted has been discussed in chapter 5 and
is not considered here.
The comparisons are between fresh formulations, which may not undergo major
deterioration in the laboratory tests. On the other hand the long duration test in the
sub-assembly may involve lubricant deterioration, ~hich can affect wear rates. To
test this aspect some tests should be conducted on the oil samples periodically
drawn from the sub-assembly test. This will clarify any adverse effects due to
deterioration. Environment also should be kept in mind. For example the
deterioration in an engine test will be in the presence of combustion products
unlike the normal atmosphere in the laboratory test. The influence of acidic
components generated in the combustion process on additives, as well as their
direct interaction with materials is of importance. In diesel engines soot also plays
a role in engine wear. Some examples of research in these areas may be cited [27,
28,29]. Correlations in such cases will depend on the extent of environmental
influence and have to be carefully judged.
The progress in this direction eventually depends on the acceptance and use of the
proposed methodology. As discussed in earlier chapters no clear demarcation
between additives is possible with short duration tests. These tests have a history of
more than five decades. The author was not successful in ordering this large
information as it was not possible to define steady state wear rate satisfactorily due
to reasons already discussed in chapter 7. Also different criteria were used to obtain
wear rate and many comparisons between additives were at the laboratory level
only. Information on correlation with practice is very limited. Even though
mapping is initially expensive, in the longer run the advantages far outweigh the
initial costs.
A question that arises is the extent to which "wear map approach can help in
evaluating novel formulations. The general feeling in the industry with testing in
other areas like oxidation stability and detergency is that correlations are
meaningful provided they have similar chemistry. This aspect is vague to define
and may be considered to mean additives of similar class with variations in
structure. Thus all ZDTP additives with different alkyl/aryl groups may be
considered to have similar chemistry. The author considers that comparison on the
basis of wear maps may be capable of relaxing this restriction and laboratory
screening will be more meaningful even when all together different chemistry is
involved.
290 CHAPTER 9. WEAR IN REAL SYSTEMS" AND LABORA TORY RIGS
Another relevant question concerns the user industry. The question is how to
choose between different formulations. Large-scale users have facilities to do the
relevant testing and also in some cases define their own requirements. They also
influence the development of standards. With small scale users the choices are
difficult. While independent laboratories can provide answers these may be
expensive to come by. Such users may like to consider the idea of wear mapping.
This should provide a better method to choose in comparison to short duration tests
normally done.
The approaches suggested are logical but need to be tested. Verification from the
available literature is not possible as comparisons based on wear map approach are
not available. The author considers the suggested direction should be pursued by
those involved in boundary lubrication research and practice. It is hoped the
methodology will be used, and further developed in future.
References
1. H. Heizler, Vehicle and Engine Technology, Arnold, London, 1999.

2. C. M. Taylor (ed.), Engine Tribology, Tribology series, 26, 1993.
3. B. L. Ruddy, D. Dowson, and P. N. Economou, A review of studies of piston ring
lubrication, in Tribology of Reciprocating Engines, paper V (i), Proc. 9th Leeds-Lyon
Symposium on Tribology, 1982.
4. S. D. Gulwadi, Analysis of tribological performance of a piston ring pack, Trib. Trans.,
STLE, 43 (2000) 151.
5. C. M. Taylor, Fluid film lubrication in automobile valve trains, Proc. lnstn. Mech.
Engrs., J. Eng. Trib. Part J, 208 (1994) 221.
6. J. F. Booker, Dynamically loaded journal bearings: mobility method of solution, J. Basic
Eng., ASME, D 187 (1965) 537.
7. C. M. Taylor, Automobile engine tribology-design considerations for efficiency and
durability, Wear 221 (1), (1998) 1.
8. V. R. K. Sastry, S. D. Phatak, and A. Sethuramiah, Experimental study of the influence
of liquid lubricants on burnished MoS2 films, Wear, 86 (1983) 213.
9. M. Priest and C. M. Taylor, Automobile engine tribology-approaching the surface, Wear,
241 (2000) 193.
10. R. C. Coy, Practical applications of lubrication models in engines, New Directions in
Tribology, MEP (1997) 197.
11. L. L. Ting, Lubricated piston ring and cylinder bore wear, in M. B. Peterson and W. O.
Wirier (eds.), Wear Control Handbook, ASME, 1980.
REFERENCES 291
12. J. C. Bell and T. Colgan, Pivoted-follower valve train wear: Lub. Eng., STLE, 47 (2),
(1991) 114.
13. A. Flodin and S. Andersson, Simulation of mild wear in helical gears, Wear, 241 (2000)
123.
14. M. C. Shaw, Metal Cutting Principles, Clarendon Press, Oxford, 1984.
15. P. Pawlus, A study on the functional properties of honed cylinder surfaces during
running-in, Wear, 176 (1994) 247.
16. Y. R. Jeng, Impact of plateaued surfaces on tribological performance, Trib. Trans.,
STLE, 39 (1996) 354.
17. J. Michalski and P. Pawlus, Description of honed surface topography, Int. J. Mach.
Tools, 34 (2), (1994) 199.
18. B. J. Taylor and T. S. Eyre, A review of piston ring and cylinder liner materials, Trib.
Int., April (1979) 79.
19. S. M. Hill, S. E. Hartfield-Wunch, and S. C. Tung, Bench wear testing of common
gasoline engine cylinder bore surface/piston ring combinations, Trib. Trans., STLE, 39
(1996) 929.
20. P. J. Blau, Friction and Wear Transitions of Materials, Noyes Publications, New Jersey,
1989, 337-351.
21. A. H. Naeim, M. Zheng, H. Shengqiang, B. Walter, and G. John, In situ wear measuring
technique in engine cylinders, Trib. Trans., STLE, 41 (1998) 579.
22. R. Kumar, S. Kumar, B. Prakash, and A. Sethuramiah, Assessment of engine liner wear
from bearing area curves, Wear, 239 (2000) 282.
23. R. Gahlin and S. Jacobson, A novel method to map and quantify wear on a micro-scale,
Wear 222 (1998) 93.
24. K. Sasajima, K. Naoi, and T. Tsukuda, A software-based relocation tehnique for surface
asperity profiles and its application to calculate volume changes in running-in wear,
Wear, 240 (2000) 152.
25. E. Decenciere and D. Jeulin, Morphological decomposition of the surface topography of
an internal combustion engine cylinder to characterize wear, Wear, 249 (2001) 482.
26. P. J. Guichelar, M. W. Williams, C. W. Omo, and D. Wilde, Experimental verification
of thermoelastc mechanical seal face deflection by local wear measurements, Lub.
Eng., 56 (7), (2000) 26.
27. Y. Yahagi, Y. Nagasawa, S. Hotta, and Y. Mizutani, Corrosive wear of cast iron under
reciprocating lubrication, SAE 861599 (1986).
28. C. Kim, C. Passut, and D. M. Zang, Relationships among oil composition combustion
generated soot, and diesel engine valve train wear, SAE 922199 (1992).
29. J. J. Truhan, C. B. Covington, and L. M. Wood, The classification of lubricating oil
contaminants and their effect on wear in diesel engines as measured by surface layer
activation, SAE 952558 (1995).
292 CHAPTER 9. WEAR IN IU:~AL SYSTFAtS AND LABORA "IORY RIGS
Nomenclature
b depth of cut
kb wear factor, dimensional
l sliding distance
m exponent
n exponent
P exponent
t feed rate
Tt tool life
V wear volume
v cutting speed
W load, N
293
Index
Abrasive wear chain length effects, 103

modelling, 84 dynamic adsorption, 105
phenomena, 83 role of oxides, 110
three-body, 86 ultrahigh vacuum studies, 111
Adhesive wear adsorbed multilayers, 113
adhesion, 78 Boundary lubrication with reacted films
modelling, 79 additives and mechanisms, 116
phenomena, 78 characterisation of reaction films,
Adsorbed monolayers 124
See boundary lubrication chlorine compounds, 121
Adsorption mechanistic studies, 123
adsorption Vs chemisorption, 105 multifunctional additive systems,
dynamic adsorption, 105 121
fractional coverage, 106 phosphorous compounds, 120
heat of adsorption, 107, 108, 138, 140 sulphur compounds, 117
Antiwear additives, 116 triboemission, 111
Asperity contact Ceramics
See contact of surfaces dry friction, 17
Atomic force microscopy, 114 dry wear of ceramics, 186
Base oils, 38, 45 wear maps, 188
technology of refining, 39 wear mechanisms and modelling,
dewaxing, 40 186
hydrocracking, 40, 41,42 lubricated wear of ceramics, 191
hydrofinishing, 39, 41 ceramic-lubricant interactions, 192
physicochemical properties, 44 high temperature applications, 196
solvent treatment 39 moderate temperature applications,
typical structures, 39 194
Bearing observations, 197
journal bearing, 24, 25, 236 Chemical wear- See under lubricated
slider bearing, 24, 25 wear
Boundary lubrication with adsorption Contact of surfaces, 5
adsorbed monolayers, 103 implications of contact model, 9
adsorption Vs chemisorption, 105 Contact stresses, 68
294 INI)EX
frictional traction, 70 turbine oils, 51, 52

sub-surface stresses, 70 Lubricants, synthetic
Crude distillation, 35, 37 See synthetic lubricants
Dry wear Lubricated wear of ceramics
abrasive, 83 ceramic-lubricant interactions, 192
adhesive, 78 high temperature applications, 196
ceramics, 186 moderate temperature applications,
wear maps, 188 194
fatigue, 86 observations, 197
oxidative, 88 Lubricated wear of metallic materials
pure polymers, 171 adsorption based models, 134
polymer composites, 175 application of the model, 141
Elastohydrodynamic lubrication, 27 assessment of the adhesive wear
Engine wear model, 138
cam and follower, 274 chemical wear, 142, 143
ring-liner, 270, 273,276, 278, 279 alternative wear mechanism, 146
Environmental issues assessment of the proposed
biodegradability, 60, 61 mechanism, 147
re-refining, 62 fundamental considerations, 149
toxicity, 60, 61 analysis of wear particles, 150
water endangering number, 61 nature of barrier films, 149
Extreme pressure additives, 115 limitations of the available model,
Fatigue life 144
gears, 249 practical aspects, 151, 164
rolling element bearings, 248 experimental variables, 134
Fatigue wear, 86, 261 Lubricated wear of polymers, 178
Friction Mixed lubrication
junction growth, 3, 112 conventional approach 236
laws of friction, 1, 9 experimental considerations, 240
non-metals, 17 advances in practice, 245
ploughing, 5, 84, 85 asperity contact, 241
stick-slip, 5, 113 asperity level conformity, 245
Hydrodynamic lubrication, 23 film thickness, 240
Lambda ratio, 29, 236 lambda ratio, 29, 236
Lubricants (petroleum based) lubricant-metal interactions in fatigue,
formulation methodology, 45 250
nature of additives, 42 observations on mechanisms, 253
performance issues, 51 pitting in rolling/sliding contacts, 246
tribological implications, 55, 60 contact stresses, 247
specifications gears, fatigue life, 249
automotive, 46 rolling element bearing, fatigue
industrial, 49 life, 248
gear oils, 51 recent developments, 237
hydraulic oils, 50 wear in rolling/sliding contacts, 256
metal working fluids, 49 experimental investigations, 260
INDEX 295
modelling of contact, 257 load carrying capacity of lubricants,

Monolayers, 103 215
Multilayers, 113 conformal contact machines, 216
Nanotribology, 114 influence of operating parameters,
Oxidative wear modelling, 88 217
designed experiment, 90 test configurations, 204
Pitting wear evaluation, 206
See Mixed lubrication assessment of wear, 208
Polymers Tribology in perspective, 1-34
dry wear of polymers, 171 Wear
lubricated wear, 178 See under respective heads for
polymer friction, 18, 175 detailed entries
structures, 172 abrasive, 83
Roughness adhesive, 78
See Surface roughness ceramics
Running-in dry, 186
general nature, 153 lubricated, 19 l
modelling, 154 fatigue, 86
running-in in practice, 276 lubricated wear, metallic materials,
Scuffing 133-167
control with EP additives, 162 oxidatve, 87
failure of boundary lubrication, 157 polymers
failure temperature, 161 dry, 71
flash temperature, 160 lubricated, 178
influence of operating parameters, 217 fatigue, 86
Stress intensity factor, 87, 187 Wear in real systems
Surface analysis, 125 complexity of wear, 267
Surface force apparatus, 113 importance and modelling, 269
Surface roughness lubrication of contact, 270
asperity level conformity, 243 modelling of wear, 273
equivalent roughness, 9 running-in in practice, 276
roughness parameters, 6, I 0 wear measurement, 278
Surface temperature rise bearing area method, 279
asperity temperatures, 71 precision techniques, 280
with thermal interaction, 75 removable insert method, 278
without thermal interaction, 72 strategies for laboratory evaluation,
theoretical calculation, 11 282
Synthetic lubricants development of wear maps, 284
applications, 57 linkage to practice, 286
structures, 59 lubrication regime, 284
types, 56 selection of test rig, 282
Tribological evaluation methodologies
boundary effects in plastic
deformation, 226

(Tribology Series 42) A. Sethuramiah (Eds.) - Lubricated Wear - Science and Technology-Academic Press

Uploaded by

Copyright:

Available Formats

(Tribology Series 42) A. Sethuramiah (Eds.) - Lubricated Wear - Science and Technology-Academic Press

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Tribology Series 42) A. Sethuramiah (Eds.) - Lubricated Wear - Science and Technology-Academic Press

Uploaded by

Copyright:

Available Formats

TRIBOLOGY SERIES

Vol. 6 Friction and Wear of Polymers (Bartenev and Lavrentev)

The developments in practice will depend on the effective dialogue between

* The role of chemical reactions in wear and scuffing is examined from a

Tribology may be defined as the science and technology of surfaces in relative

1.2 Laws of friction and their explanation

where p is the normal pressure, a is a constant, s i is the shear strength of the

The above theory cannot be quantified as ~: cannot be defined for a tribological

1.2.1 Other aspects of friction

Besides adhesion and plastic deformation, it is recognised that ploughing which

1.3 Contact of surfaces

The average curvature of summits is nearly same as that obtained for

2. The density of summits (asperities) per unit area rl is approximately

n = rlA n I ~0(z)dz (1.6)

A, = rt[3rlA. I (z - d)q0(zldz (1.7)

L = 47 E ~0.5 rlA. (z - d) 3/2q~(z)dz (1.8)

,p(z) = Gaussian height distribution function =

[~*: 13'131 (1.10a)

1.3.1 Implications of the contact model

The deformation of an asperity contact will become plastic beyond a critical

1.4 Surface temperatures in sliding contact

The frictional heating in tribological contacts has a significant influence on the

1.4.1 Theoretical calculation

Cp = specific heat (J/Kg~

v = sliding velocity (m/s)

0946i )= 0946/q,lk,,) (1.14)

Taking k ~ = k l it can be shown

A0= 0.946( q l / 0 + k*)-' (1.15)

Another approach frequently used to estimate temperature rise is to assume the

Both the approaches above are equivalent.

Theoretically for semi-infinite bodies with 0~ bulk temperature the surface

Contact temperature ( Tc ) = Bulk temperature ( Tb ) + A0 (1.18)

The above approach is applicable to lubricated contacts where bulk temperatures of

AO~ = 0.946 e (1.19a)

This equation is the same as for stationary source multiplied by a coefficient E. E

Another appro0ch suggested by Greenwood [ 15] involves an interpolation method

The above procedure can be illustrated by the following example. Consider a

Thermal conductivity (k) -- 43 W/m~

Fig. 1.5. 'y' as a function of Peclet number.

For the square contact l = [,]1"81x10-5 = 2.127x10-3m

q = ~ _ f L v_ 0.4x100x5.11 =1.12x107 W/m 2

A01 =0.75 Pe =18.71

For body II receiving heat from stationary source Eq.(1.12) is applicable. So

Now temperature rise A0 is obtained from

1.5 Friction of non-metals

Friction of some ceramics is sensitive to the environment. Silicon nitride is an

1.5.2 Polymer friction

linked polymers, which decompose at high temperatures unlike thermoplastics that

In a broader sense, lubrication may be defined as the complete or partial separation

hydrodynamic load support. This regime of lubrication is referred to as boundary

In this equation rl is constant provided shear stress is proportional to shear strain.

Viscosity is expressed in absolute units (N.s/m 2) in the above equation. Many

Viscosity varies significantly with temperature. The temperature dependence can

where, A and b ~ are constants for the given fluid

The important equation used in terms of cSt (Walther equation) is expressed as

loglog(v + c ' ) : a ' - b' (logT) (1.23)

1.6.1.1 Viscosity index

Viscosity index is widely used to characterise lubricants as will be seen in chapter

V I : [{(antilogN)-1}/ 0.00715] + 100 (1.24b)